1 12 CANCER 1 12.1 Epidemiology 2 12.1.1 Methodological

1
12
CANCER
2
12.1
Epidemiology
3
12.1.1
Methodological considerations
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
The majority of epidemiological studies of mobile phone use and risk of cancer have used a casecontrol design, which means that cases of the studied disease are identified over a certain time period in a given
population, and controls, i.e. individuals free from the disease, are randomly selected from the same study
population during the same time period. The purpose of the controls is to provide information about the
exposure distribution in the population from which the cases came. The preferable control selection method is
population based; i.e. a random selection from a population register that covers the population from which the
cases were identified. When the study base is unavailable for random sampling, other methods for control
selection is used, e.g. hospital controls, random digit dialling, friends or neighbourhood controls. Hospital
controls are usually other patients randomly selected at the same hospital, with diseases assumed to be unrelated
to the studied exposure. They often have higher participation rates than population controls, but it is impossible
to know whether the patients with other disease have the same exposure distribution as the general population
from which the cases were identified, as it is not a random sample of the population who become hospital
patients. Random dialling of phone numbers until a suitable control agrees to participate (“random digit
dialling”) attempts to achieve a population based control sampling, but the actual number of eligible potential
controls is unknown, and adequate participation rates cannot be calculated. Friend or neighbourhood controls
are also associated with the problem of not knowing whether they are representative of the study base. Selection
bias may also occur if not all eligible cases and controls can be included in the study, e.g. when individuals’
active participation is required and not everyone agrees or can be reached, or if cases have died or are too ill to
participate. Non-participation need not necessarily lead to selection bias, even if participation rates differ
between cases and controls, but bias occurs if the likelihood of participation is related to both the exposure and
to case-control status. Selection bias may also occur as the result of internal missing data, i.e. if cases and
controls differ in the completeness of answers to questions in interviews or questionnaires. Few studies have
evaluated potential selection bias caused by non-participation. A validation study (Vrijheid et al., 2009b) from
the Interphone multicentre study found that regular mobile phone use was less prevalent among nonparticipants, both among cases and controls, compared to participants (56% mobile phone users among nonparticipating controls compared to 69% among participants, and 50% mobile phone users among nonparticipating cases compared to 66% among participants). With an overall higher participation rate among cases
than among controls, it was estimated that selection bias led to an underestimation of risk estimates by around
10%.
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Exposure information in case-control studies is collected retrospectively, after identification of cases
and controls, often by means of questionnaires or interviews, but sometimes through independent sources such
as registers. All but one of the case-control studies of mobile phone use have collected retrospective selfreported information, with the advantage that detailed information about the exposure can be collected. The
disadvantage is, however, that it is vulnerable to recall bias. If cases believe that their disease was caused by
mobile phone use they may overestimate their previous phone use, especially if they are asked to recall habits
many years prior to their disease. Controls on the other hand, may be less thorough when reporting about past
exposures, and have not given past mobile phone use much thought until being faced with these questions in an
interview or questionnaire. For both cases and controls it is very difficult to remember mobile phone use habits
many years earlier, and misclassification will inevitably occur. If this is entirely independent of the disease, i.e.
non-differential, it will in general lead to a dilution of the risk estimates towards unity, but if the
misclassification is systematic it may also lead to bias away from the null. A validation study within Interphone
reported considerable random misclassification of mobile phone use when asking healthy volunteers to report
their mobile phone use 6 months earlier (Vrijheid et al., 2006). The study also found systematic
misclassification; light users tended to underestimate, and heavy users tended to overestimate their amount of
phone use. Validation studies have also reported that recall of number of calls was slightly better than recall of
duration of calls (Inyang et al., 2009; Vrijheid et al., 2006). If, on the other hand, cases tend to overestimate or
controls underestimate exposure, this will lead to an overestimation of the risk estimates or even spurious
associations. A validation study from the Interphone study compared self-reported information about duration of
mobile phone use and number of calls to operator recorded information as far back as approximately 4 years
prior to diagnosis (Vrijheid et al., 2009a). The study found little difference overall in recall between cases and
controls, but observed that cases tended to overestimate their mobile phone use more the further back in time
they were reporting about, which was not observed among controls, indicating the potential for overestimation
of risk estimates.
1
57
58
59
60
61
62
63
64
65
66
67
68
Cohort studies generally determine exposure status at baseline, and subsequently follow the cohort
members for occurrence of the studied disease (or diseases). Provided that mechanisms are in place for followup of the cohort members, e.g. in health data registers, selection bias is usually not a problem, although
comparability between exposed and unexposed must be ensured, e.g. through control of confounding. For
cancer outcomes that are usually rare, the cohort needs to be very large, which often lead to collection of less
detailed exposure information than in case-control studies. If exposure information is too crude it may hamper
the ability to detect effects if limited to a small subgroup with specific exposure characteristics. Some of the
cohort studies have only used register based exposure information, with few details about mobile phone use
characteristics, while others have used self-reported information about mobile phone use which is subject to
non-differential exposure misclassification as in the case-control studies. An important difference compared to
case-control studies is that when exposure information is collected before the occurrence of the disease recall
bias will not affect results.
69
70
71
72
73
74
75
76
77
78
79
80
81
A cross-sectional design should never be used for cancer outcomes, but is common in studies of
“soft” outcomes like various types of symptoms, headaches, sleep disturbances, behavioural problems, and
similar outcomes. A cross-sectional design has major methodological limitations. It includes prevalent cases,
and may therefore investigate survival or prognosis rather than disease occurrence. Most importantly, it cannot
determine the time sequence of events; i.e. whether the exposure preceded the disease or the disease was already
present when exposure begun. It is possible that the disease itself affects the exposure, e.g. a person who
perceives him/herself as electrically hypersensitive may avoid situations with known exposures like mobile
phone use, and may therefore appear as less exposed than the general population. Parents with a child who has
health problems may find it convenient to provide the child with a mobile phone for easy contact, which would
make the child appear as more exposed than other children. Non-participation in a cross-sectional study may
lead to selection bias; the probability of participation may be related to both the studied outcome and the
exposure. In addition, exposure information is always collected retrospectively, often by self-reports, and may
be affected by the disease, i.e. recall bias.
82
12.1.2
83
84
85
86
87
88
89
90
91
92
93
At the time of the previous WHO review, only a handful of epidemiological studies were available,
mostly focusing on occupational exposures and cancer mortality, and no study on mobile phone use. To date,
over 30 studies of mobile phone use and brain tumour risk have been conducted, some of them reported in
multiple publications, or with several follow-ups, or elaborations of exposure assessment and study designs.
Most of them have used a case-control design and only few were cohort studies. In addition, a large number of
studies present results for other tumour types. Only one study so far has addressed tumour risk in children. The
literature search covering the period January 1, 1992 – December 31, 2013 identified 617 potentially relevant
papers and an additional five studies were identified through hand search. After exclusion of irrelevant studies,
commentaries, review articles, and duplicate reports, 69 articles were fully reviewed. One study on survival was
excluded (Hardell & Carlberg, 2013), and one study did not provide enough information for assessment of the
inclusion criteria, and is briefly described, but not tabulated. The remaining articles are reviewed in full below.
94
95
96
97
98
99
The available evidence is presented below in a subsection for brain tumours, which includes brain
tumours overall, glioma, meningioma, acoustic neuroma, and brain tumours in children, and another section
presents the studies of salivary gland tumours. The last section discusses the studies of other tumour types. For
adult brain tumours, the studies are in addition divided into cohort studies and case-control studies, and within
each of these sections studies are presented together when published by the same research group, in the same
geographical area, or as part of a consortium. Otherwise the studies are presented according to publication year.
Mobile phone use
100
12.1.2.1
101
Cohort studies
102
103
104
105
106
107
108
Only three cohort studies have so far been conducted that are focused on RF exposure from mobile
phone use, of which one is essentially uninformative because of small numbers and short follow-up. The two
other cohort studies have published one or several updates with longer follow-up time than in the originally
published study. In this section, all updates are presented as the early publications provide essential information
for the evaluation of potential non-differential exposure misclassification in the analyses of longer follow-up
periods, and for the evaluation of potential random variation. Details of the results are shown in Table
12.1.2.1.1.
Brain tumours
2
109
110
111
112
113
114
115
An early study from the US (Dreyer, Loughlin & Rothman, 1999) identified a cohort of mobile phone
users (n=285 561) through two mobile phone operators and compared cause-specific mortality among hand-held
mobile phone users with those who used non-handheld mobile phones, i.e. mobile phones where the antenna is
not situated on the handset (e.g. car phones, bag-phones). Information about mortality was obtained from the
National Death Index. The follow-up of the cohort was blocked by a lawsuit after only one year, and therefore
mortality data were available only for 1994. In total, only 6 deaths from brain cancer were identified. [The short
follow-up and small numbers makes the study largely uninformative.]
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
A Danish cohort study identified all mobile phone subscriptions through the Danish mobile phone
operators from the start of mobile phone services in 1982 until the end of 1995 (Johansen et al., 2001), in total
723 421 subscriptions. The subscriptions were linked to persons through the Danish population registry. The
cohort was restricted to private subscribers because corporate subscriptions (in total 200 507) could not be
linked to an individual. Other reasons for exclusion were e.g. failure of the linkage to the population registry and
duplicates. The final cohort included 420 095 mobile phone subscribers, and in the first publication from this
cohort, subscribers were followed from first subscription until the end of 1996, emigration or death, whichever
occurred first. The average follow-up was 3.1 years (range 0–15 years), and 85% of the cohort were male, 58%
were younger than 50 years at the time of the first subscription, and only 6% were 60 years or older. Information
was available on year of the first subscription and type of system (analogue or digital), but no information on
amount of phone use. Cancer incidence in the cohort was ascertained by linkage to the Danish Cancer Registry.
Standardised incidence ratios (SIR) were calculated comparing cancer incidence in the mobile phone subscriber
cohort with national rates allowing for age, sex and calendar period. In total, 154 cases of nervous system
tumours were identified in the subscriber cohort during the study period, resulting in an overall SIR of 1.0 (95%
CI 0.8–1.1). Results did not differ between analogue and digital phone users, and did not change with duration
since start of subscription; the SIR for ≥5 years since first subscription was 1.0 (95% CI 0.7–1.6). For specific
types of brain tumours analyses were only made for ever having had a subscription; for glioma SIR was 0.94
(95% CI 0.72–1.20), with no substantial differences between various locations (SIR for the temporal lobe was
0.86; 95% CI 0.42–1.54), for meningioma 0.86 (95% CI 0.49–1.40), and for nerve sheath tumours (the majority
of which is acoustic neuroma) 0.64 (95% CI 0.26–1.32). [The results in this publication are only informative for
short-term mobile phone use; 69% of the cohort started their subscription 1995 or 1996. This first analysis of the
cohort is mainly of interest as it demonstrates that short-term mobile phone use did not lead to a higher
incidence of brain tumours in the cohort than in the general Danish population, which means that updates of the
cohort for analyses of longer-term use are unlikely to be affected by misclassification from new short-term
mobile phone users in the Danish population during additional follow-up because the incidence rate in this
group would be similar to that of the non-exposed population.]
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
The Danish cohort study was updated by Schüz et al. (2006c) by extending the follow-up with an
additional 7 years, through December 31, 2002. In the updated analyses, members of the subscriber cohort were
excluded from the calculations of the national incidence rates used for estimation of SIR, to reduce potential
non-differential exposure misclassification. As in the first follow-up, adjustment was made for age, sex and
calendar period. In total, 580 tumours of the brain and nervous system were identified during the entire study
period, including also the 154 cases identified in the first follow-up. The SIR for ever being a mobile phone
subscriber was estimated at 0.96 (95% CI 0.87–1.05) for men, and 1.03 (95% CI 0.82–1.26) for women. Risk
did not increase with increasing duration of subscription, there was even a reduced risk estimate for ≥10 years
since first subscription (SIR=0.66; 95% 0.44–0.95, based on 28 exposed cases). For specific tumour types
related to ever having had a subscription, SIR for glioma was 1.01 (95% CI 0.89–1.14), with small differences
between specific locations (e.g. SIR for temporal lobe tumours was 1.21; 95% CI 0.91–1.58 and for parietal lobe
0.58; 95% CI 0.36–0.89), for meningioma SIR was 0.86 (95% CI 0.67–1.09), and for cranial nerve sheath
tumours 0.73 (95% CI 0.50–1.03). [The results do not provide evidence in support of the hypothesis that mobile
phone use increases the risk of brain and nervous system tumours, or specific types of brain tumours, but only a
small proportion of the cohort had used mobile phones at least 10 years, and the reduced risk in this category
may be a result of chance. The results are more informative for intermediate-term use; 42% of the person-years
of follow-up refer to at least 5 years since first subscription. The results for smoking related cancers (discussed
below under “Other cancers”) indicate that, at least among men, the subscriber cohort is likely to have a
healthier lifestyle than the general Danish population, and the authors show that the early subscribers have
higher socioeconomic status. Glioma is not, however, associated with lifestyle factors such as smoking or
alcohol consumption, and if anything, the incidence is higher among persons with higher socioeconomic status.]
163
164
165
A third follow-up of acoustic neuroma incidence in the subscriber cohort from 1998 until the end of
2006 was focused on long-term use (Schüz et al., 2011). Information was also collected on marital status and
various indicators of socioeconomic status through matching of cohort participants aged ≥30 years with another
3
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
existing nationwide cohort created in 1990. After linkage of the two cohorts 2 883 665 persons with 22 884 931
person-years of follow-up were available for analyses. Incident cases of acoustic neuroma were identified from
a clinical data base at the main acoustic neuroma treatment centre in Denmark and from the nationwide cancer
registry. Information was also collected about the size of the tumour and tumour laterality. Analyses were made
with log-linear Poisson regression models, adjusted for calendar period (1998–2001, 2002–2006), age, highest
attained educational level, disposable income, and marital status. A dichotomous exposure categorisation was
used, with the cut-point at ≥11 years, chosen to assure that no one in the unexposed group could have as long
period of mobile phone subscription as the exposed group. In total, 404 acoustic neuroma cases in men and 402
cases in women were identified during the study period. Long-term mobile phone use was uncommon among
women, and no case of acoustic neuroma with long-term use was identified, to be compared to 1.6 expected.
Among men, 15 cases were long-term users, resulting in an incidence rate ratio (IRR) of 0.87 (95% CI 0.52–
1.46), with adjustment for all potential confounding variables, and essentially the same when adjustment was
limited to age and period. The size and spread of the acoustic neuroma tumours did not differ between long-term
mobile phone subscribers and others. From another cohort, the authors had collected information on preferences
of side of mobile phone use in the Danish population; 53% preferred the right side, 35% the left, and 13% had
no preference. In long-term mobile phone subscribers 47% of acoustic neuroma tumours were located on the
right side, while the corresponding proportion among others was 48%. There was no evidence that the
proportion of acoustic neuromas located on the right side had increased over time.
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
A third follow-up was also made of central nervous system tumour incidence in the Danish cohort
from 1990 until the end of 2007, using the same approach as for acoustic neuroma to collect information on
indicators of socioeconomic status for confounding control (Frei et al., 2011). Analyses were also made of the
incidence of all cancers combined and of smoking related cancers. In total, 358 403 persons contributed exposed
person-years to the study. A one-year latency period was used in the calculations of person-years. Time since
first subscription was categorized into 1–4, 5–9, ≥10 years. The upper category was further divided into 10–12,
and ≥13 years when numbers allowed. Analyses were made with log-linear Poisson regression models, adjusted
for age, calendar period (1990–1995, 1996–2002, 2003–2006), highest attained educational level, and
disposable income. In total, 122 302 cancers among men, and 133 713 cancers among women had occurred
during the follow-up period. Corresponding numbers for nervous system tumours were 5111 men, and 5618
women. A slightly reduced overall cancer incidence among men in the subscriber cohort was observed, probably
attributable to a lower incidence of smoking related cancers. For women, the corresponding relative risk
estimates were close to unity. For central nervous system tumours, all relative risk estimates were close to unity
for both men and women, regardless of time since first use. Also for specific types of brain tumours, relative
risk estimates were close to unity, e.g. for glioma, being a mobile phone subscriber ≥10 years was associated
with an IRR of 1.04 (95% CI 0.85–1.26) in men and 1.04 (95% CI 0.56–1.95) in women. For ≥13 years the IRR
was 0.98 (95% CI 0.70–1.36) in men.
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
[Overall, the series of results from the Danish cohort study provide no support for the hypothesis that
mobile phone use increases the risk of glioma, meningioma, or acoustic neuroma, either for short-, intermediate, or long-term use, the longest ≥13 years. The important strengths of the Danish cohort study are the prospective
registration of mobile phone start year, independent of the disease, which prevents recall bias, and the
completeness of follow-up through the population registry and high quality cancer registry, which prevent
selection bias. Limitations are the inability to identify users of corporate subscriptions, and lack of information
about amount of phone use. In the first analysis of the cohort, comparisons were made with national cancer
incidence rates, and the cohort was included also in the comparison group. The effect on the risk estimates
introduced by this non-differential exposure misclassification is, however, minor, as mobile phone use was still
very uncommon in the Danish population at this time, and a potentially increased incidence among mobile
phone users would not have any substantive effect on the national incidence rates. In the updates of the analyses
the subscriber cohort was excluded from the calculation of national incidence rates. Inability to identify
corporate subscribers mainly reduces the statistical power, but have little effect on the risk estimates as they
constitute less than 5% of the Danish population aged ≥18 years in 1996 (www.dsk.dk). It is also unlikely that
someone would have a subscription for someone else, without being a mobile phone user him/herself, as this
was in the beginning of the era of mobile phone use when it was very expensive to make mobile phone calls.
Thus, the effect of the non-differential exposure misclassification introduced from using subscriptions to
identify mobile phone use is likely to be minimal. Control of confounding was made to the same extent as in
most case-control studies, which make analyses of time since first subscription comparable to analyses of time
since first use in the case-control studies. Without information about amount of mobile phone use, the study is
unlikely to be able to detect an increased risk if it is restricted to a small subgroup of heavy mobile phone users.]
4
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
A prospective cohort study was conducted in the UK by Benson and co-workers, as part of the
Million Women Study (Benson et al., 2013a). Between 1996 and 2001, 1.3 million women were recruited to the
cohort through the national breast cancer screening programme. The cohort is contacted with questionnaires
regularly, and in 1999–2005, 65% of the originally recruited women answered baseline questions on mobile
phone use: how often (never, less than once a day, every day) and how many years they had used a mobile
phone. In 2009, new and more detailed questions on mobile phone use were asked, but these were only used in
the current study for sensitivity analyses. Mobile phone use was reported by 34% of the women who answered
the questionnaire in 1999 and 79% reporting in 2005. In 2005, 32% had used a mobile phone at least 5 years. In
total, 791 710 women free of cancer at baseline were followed for cancer occurrence from the time they
answered the 1999–2005 questionnaire until the end of 2009, with an average follow-up time of 7 years. Cox
regression models were used for analyses, and control of confounding was made from age, area based
socioeconomic status, geographical region, height, BMI, smoking, alcohol, strenuous exercise, and menopausal
hormone therapy. Sensitivity analyses were performed to avoid bias from prodromal symptoms by excluding
the first three years of follow-up in the analyses of nervous system tumours. In addition, a sensitivity analysis
excluded women who answered the questionnaire 1999–2000 to reduce potential exposure misclassification, as
many of these women may have changed their mobile phone use during the follow-up period. In total, 51 680
incident invasive cancers and 562 non-invasive intracranial central nervous system (CNS) tumours were
identified, of which 1261 were intracranial CNS tumours, including 571 glioma, 251 meningioma, 110 pituitary
tumours, and 96 acoustic neuromas. No risk increase was found for glioma; >10 years of mobile phone use was
associated with a RR of 0.78 (95% CI 0.55–1.10). The corresponding result for meningioma was 1.10 (95% CI
0.66–1.84). For pituitary tumours a raised risk was observed for ever having used a mobile phone (RR=1.52;
95% CI 0.99–2.33), but with no dose-response pattern with duration of use. For acoustic neuroma, the observed
risk estimate increased with increasing mobile phone duration from >5 years of phone use, and for >10 years of
use the RR was 2.46 (95% CI 1.07–5.64). Risk estimates did not vary with amount of use for any of the tumour
types; for glioma the RR for daily use was 0.8 (95% CI 0.56–1.14), and for acoustic neuroma 1.37 (0.61–3.07).
The sensitivity analyses did not give materially different results. [The strength of this large cohort study is the
prospective design with individual information on amount of mobile phone use, which prevents the major
sources of bias identified in the case-control studies, i.e. recall bias and selection bias, and reduces nondifferential exposure misclassification, identified as a limitation in the Danish cohort study. Adjustment for a
large number of potential confounding factors is another strength. Available information on amount of phone
use did not allow identification of the heaviest users. Sensitivity analyses did not indicate dilution of risk
estimates by changed habits of mobile phone use during follow-up. The statistical power was good for glioma
and meningioma, but limited for acoustic neuroma, as shown by the wide confidence intervals. Finding an
increased risk of acoustic neuroma already after 5 years of mobile phone use seems unlikely to reflect causality,
given the slow growing nature of this tumour type (Thomsen & Tos, 1990). There is a possibility that mobile
phone use leads to an earlier detection of an acoustic neuroma, by making the person aware of a unilateral
hearing loss, which may be an alternative explanation for the observation of an increased risk after a short
induction period. The study found a reduced risk for lung cancer (discussed below), which indicates that the
mobile phone users have a healthier lifestyle. As discussed above, glioma incidence has not been shown to be
related to smoking or other lifestyle factors despite considerable research efforts, and none of the known risk
factors would explain a reduced glioma risk associated with mobile phone use.]
263
264
265
266
267
268
269
An updated follow-up of the Million Women cohort to 2011 was later presented in a letter to the
editor (Benson et al., 2013b), including 1727 intracranial CNS tumour, of which 875 were glioma, 397
meningioma, and 126 acoustic neuroma. Analyses were conducted in the same way as in the first study
discussed above. For glioma, the RR associated with ≥10 years of mobile phone use was 0.77 (95% CI 0.62–
0.96), for meningioma 1.08 (95% CI 0.78–1.49), and for acoustic neuroma 1.17 (95% 0.60–2.27). [From this
update it is clear that the statistical power for acoustic neuroma is limited, although somewhat better in the
update, while results for glioma and meningioma seem to be quite stable.]
270
5
Table 12.1.1. Cohort studies of mobile phone use and brain tumours
Country
No. individuals
Time
period
Source
Outcome
Exposure
No. exp Relative risk/SIR
cases
(95% CI)
Comments
Reference
154
1.0 (0.8-1.1)
135
0.95 (0.79-1.12)
Johansen et
al. (2001)
19
1.03 (0.62-1.61)
Standardized
incidence ratios, by
age and sex.
<1 year
43
0.8 (0.6-1.1)
1-4 years
87
1.1 (0.9-1.3)
≥5 years
24
1.0 (0.7-1.6)
Age range
Denmark
1982-1995
≥18
Denmark
1982-2002
≥18
Denmark
1990-2007
≥30
420 095 mobile
phone subscribers
Brain and
Ever subscriber
nervous system Ever subscriber, men
Comparison group:
national incidence
rates
420 095 mobile
phone subscribers
Ever subscriber, women
Time since first subscription
Brain and
Ever subscriber, men
nervous system Ever subscriber, women
The whole Danish
population
358 403 mobile
phone subscribers
All Danes born in
Denmark in 1925
or later and alive in
1990, 3.21 million
persons included
in the CANULI
cohort
491
0.96 (0.87-1.05)
89
1.03 (0.82-1.26)
51
0.90 (0.67-1.18)
1-4 years
266
1.03 (0.91-1.17)
5-9 years
235
0.96 (0.84-1.09)
≥10 years
28
0.66 (0.44-0.95)
Time since first subscription
<1 year
Glioma
Ever subscriber
257
1.01 (0.89-1.14)
Meningioma
Ever subscriber
68
0.86 (0.67-1.09)
Nerve sheet
Ever subscriber
tumours, cranial
nerves
(acoustic
neuroma)
32
0.73 (0.50-1.03)
714
1.02 (0.94-1.10)
132
1.02 (0.86-1.22)
1-4 years
180
1.07 (0.92-1.24)
5-9 years
258
0.95 (0.83-1.08)
≥10 years
276
1.06 (0.94-1.20)
10-12 years
187
1.08 (0.93-1.25)
≥13 years
89
1.03 (0.83-1.27)
1-4 years
34
0.97 (0.69-1.36)
5-9 years
58
1.05 (0.81-1.37)
≥10 years
40
1.03 (0.75-1.40)
10-12 years
34
1.05 (0.75-1.47)
≥13 years
6
0.91 (0.41-2.04)
Brain and
Ever subscriber, men
nervous system Ever subscriber, women
Time since first subscription
Men
Time since first subscription
Women
All subscribers
included in
calculations of
national incidence
rates, but constitute a
small proportion of the
population.
Standardized
incidence ratios, by
age and sex.
Schüz et al.
(2006c)
The subscriber cohort
was not included in
calculations of
national incidence
rates.
~200 000 corporate
subscribers included
in comparison group,
constitute less than
5% of unexposed.
Log linear Poisson
regression models,
adjusted for age,
calendar period,
highest attained
education,
disposable income.
Frei et al.
(2011)
Exposed were
subscribers between
1987-1995. Nonsubscribers
comparison group.
Corporate subscribers
could not be
identified, constitute a
small proportion of the
unexposed
population.
6
Glioma
Men
Ever subscriber
324
1.08 (0.96-1.22)
1-4 years
85
1.20 (0.96-1.50)
5-9 years
122
1.05 (0.87-1.26)
≥10 years
117
1.04 (0.85-1.26)
10-12 years
80
1.06 (0.85-1.34)
≥13 years
37
0.98 (0.70-1.36)
32
0.98 (0.69-1.40)
1-4 years
8
0.87 (0.43-1.75)
5-9 years
14
1.02 (0.60-1.72)
≥10 years
10
1.04 (0.56-1.95)
50
0.78 (0.58-1.05)
1-4 years
15
0.92 (0.55-1.56)
5-9 years
14
0.56 (0.33-0.96)
≥10 years
21
0.90 (0.57-1.42)
30
1.02 (0.71-1.47)
1-4 years
9
1.08 (0.56-2.09)
5-9 years
13
1.04 (0.60-1.79)
≥10 years
8
0.93 (0.46-1.87)
Time since first subscription
Women
Ever subscriber
Time since first subscription
Meningioma
Men
Ever subscriber
Time since first subscription
Women
Ever subscriber
Time since first subscription
Denmark
1998-2006
≥30
2 883 665 Danes
included in the
CANULI cohort
Acoustic
neuroma
Mobile phone subscription
≥11 years
Men
No
389
Yes
15
Women
No
Yes
402
0
1.0
0.87 (0.52-1.46)
Log linear Poisson
regression models,
adjusted for age,
calendar period,
highest attained
education,
disposable income,
marital status.
Schüz et al.
(2011)
No exposed case,
Exposed were
1.6 expected
subscribers ≥11
years. Nonsubscribers and
subscribers <11 years
were comparison
group. Corporate
subscribers could not
be identified,
constitute a small
proportion of the
unexposed
population.
7
UK
1999-2005
– followed
through
2009
Mean age
60.3 (SD
5.1)
791 710 women
Intracranial
participating in the CNS tumours
UK Million Women
Study, who
answered a baseline questionnaire
1999-2005
Ever mobile phone user
No
507
1.0
Yes
754
1.01 (0.90-1.14)
<5 years
203
1.00 (0.84-1.20)
5-9 years
406
1.02 (0.89-1.17)
≥10 years
103
1.02 (0.81-1.27)
<Daily
664
1.02 (0.90-1.15)
Daily
90
1.00 (0.80-1.26)
Duration of use
Frequency of use
Glioma
Ever mobile phone user
No
237
1.0
Yes
334
0.91 (0.76-1.08)
<5 years
89
0.93 (0.71-1.21)
5-9 years
185
0.92 (0.75-1.13)
≥10 years
40
0.78 (0.55-1.10)
<Daily
298
0.92 (0.77-1.10)
Daily
36
0.80 (0.56-1.14)
Duration of use
Prospectively
Benson et
collected self-reported al. (2013a)
information on mobile
phone use.
Definition of “user” did
not require a minimal
amount of use.
Cox proportional
hazards models
adjusted for age, area
based socioeconomic
status, geographical
region, height, BMI,
smoking, alcohol,
strenuous exercise,
menopausal hormone
therapy.
Frequency of use
Meningioma
Ever mobile phone user
No
102
1.0
Yes
149
1.05 (0.81-1.38)
<5 years
41
0.88 (0.60-1.31)
5-9 years
82
1.21 (0.89-1.65)
≥10 years
20
1.10 (0.66-1.84)
<Daily
130
1.05 (0.80-1.37)
Daily
19
1.11 (0.67-1.85)
No
29
1.0
Yes
67
1.44 (0.91-2.28)
<5 years
19
1.00 (0.54-1.82)
5-9 years
38
1.80 (1.08-3.03)
≥10 years
8
2.46 (1.07-5.64)
<Daily
59
1.45 (0.91-2.31)
Daily
8
1.37 (0.61-3.07)
Duration of use
Frequency of use
Acoustic
neuroma
Ever mobile phone user
Duration of use
Frequency of use
8
UK
1999-2005
– followed
through
2011
Same as above:
Intracranial
CNS tumours
791 710 women
participating in the 1727 cases
UK Million Women
Study, who
answered a baseline questionnaire
1999-2005
Glioma
875 cases
Ever mobile phone user
Yes
0.94 (0.85-1.04)
Duration of use
<5 years
0.99 (0.83-1.17)
5-9 years
0.93 (0.82-1.06)
≥10 years
0.90 (0.77-1.05)
Same study as above Benson et
updated with an
al. (2013b)
additional two years
follow-up.
Numbers of exposed
cases not given.
Ever mobile phone user
Yes
0.86 (0.75-0.99)
Duration of use
<5 years
0.96 (0.75-1.23)
5-9 years
0.86 (0.72-1.02)
≥10 years
0.77 (0.62-0.96)
Meningioma
Ever mobile phone user
397 cases
Yes
1.01 (0.82-1.25)
Duration of use
<5 years
0.90 (0.63-1.28)
5-9 years
1.04 (0.80-1.34)
≥10 years
1.08 (0.78-1.49)
Acoustic
neuroma
Ever mobile phone user
126 cases
Duration of use
Yes
1.19 (0.81-1.75)
<5 years
0.94 (0.53-1.66)
5-9 years
1.46 (0.94-2.27)
≥10 years
1.17 (0.60-2.27)
271
272
Case-control studies
273
274
275
276
277
278
279
280
281
282
Altogether 28 case-control studies of mobile phone use and brain tumours have been conducted.
Several of these studies have also been included in pooled analyses. Some of the Interphone studies have not
been published separately and were only included in a pooled analysis. Most of the studies have included
several types of brain tumours: glioma, meningioma and acoustic neuroma, and some have also provided
analyses of all types of brain tumours combined. Some studies have focused on a single type of brain tumour,
e.g. glioma, or acoustic neuroma. To facilitate comparability between studies and between tumour types, Table
12.1.2.1.2 below presents brain tumour results focused on ever mobile phone use and categorized according to
time since first use, with results within each study stratified on specific tumour types, while Table 12.1.2.1.3
presents the corresponding results for measures of amount of mobile phone use, i.e. cumulative call time and
cumulative number of calls.
283
Hardell studies
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
Hardell and co-workers conducted a case-control study of mobile phone use and brain tumours in
Sweden (Hardell et al., 1999), with case recruitment from regional cancer registries in two geographic regions
during 1994–96 (Uppsala/Örebro) and 1995–96 (Stockholm). Only patients who were still alive when recruited
to the study were included. In total, 270 brain tumour cases were identified, of whom 37 were excluded because
they were too ill to participate, and 8 cases because they were not primary brain tumours. Finally, 209 cases
participated in the study. The participation rate is reported to be 90%, but according to conventional standards,
cases who were too ill, as well as deceased cases, should have been included in the denominator. Thus, the
participation rate is likely to be much lower. Two controls per case were selected from the population register
matched to cases on sex, age, and geographic region, and 425 (91%) agreed to participate. Information about
mobile phone use was collected through a postal questionnaire complemented with a telephone interview. Of the
cases, 37% had used a mobile phone (defined as at least 8 h in total, with a one year latency period) and the
corresponding proportion among the controls was 38%. Hours of phone use was categorized according to the
median value among the controls (224 h for analogue and 88 h for digital phones). Conditional logistic
regression was used to estimate ORs. There were no associations between mobile phone use and all brain
tumours combined (OR=0.98; 95% CI 0.69–1.41), malignant brain tumours (OR=0.98; 95% CI 0.63–1.50),
meningioma (OR=1.05; 95% CI 0.49–2.27), or acoustic neuroma (OR=0.78; 95% CI 0.14–4.20), nor were there
any associations with analogue or digital phone use considered separately, whether for 1, 5 or 10 year latency
9
301
302
303
304
305
306
307
308
309
periods, and no dose-response or significant laterality effects were seen. Subsequent reanalysis of the same
laterality data (side of phone use versus side of tumour occurrence) showed an association of borderline
significance between ipsilateral mobile phone use and temporal, temporoparietal and occipital tumours
combined (Hardell et al., 2001), but also a risk reduction at other locations. [Despite apparently high
participation rates, only about one third of the total number of brain tumour cases appears to have been included
in the study (Ahlbom & Feychting, 1999). Some, but not all, of the discrepancy is likely to be explained by the
poor prognosis of the disease. No rationale is given for grouping tumours with temporal, temporoparietal and
occipital location. Increased risk estimates at some locations and reduced risk estimates at other locations
suggest that random variation or bias may have affected results in these subgroup analyses.]
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
A second case-control study of brain tumours was conducted by Hardell and co-workers (2002), with
case recruitment during 1997–2000 in four Swedish geographic regions. After exclusion of patients with
erroneous diagnoses (e.g. metastasis, wrong date of diagnosis), 2253 eligible cases were identified. Only cases
alive at the time of recruitment were included, and 1429 answered the questionnaire (63% of eligible). One
control per case matched to cases on age, sex and geographic region was selected from population registers
(n=1617). In total, 1429 controls participated (91%). Information on exposure to cellular and cordless phones
was collected through mailed questionnaires and was completed over the phone. Analogue phones had been
used by 17% of cases and 15% of controls, with a one year latency period, but with no requirement for a lowest
amount of use, which is a difference compared to the first study where a minimum of 8 h was required. Digital
phones had been used by 30% of both cases and controls. Hours of phone use was categorized according to the
median value among the controls (85 h for analogue and 55 h for digital phones). Conditional logistic regression
was used to estimate ORs, and included 1303 complete pairs (58% of cases and 81% of controls). No additional
potential confounders were included. Ever use of an analogue mobile phone was associated with an OR of 1.3
(95% CI 1.02–1.6) for all brain tumour types combined, a result driven by acoustic neuroma (OR=3.5; 95% CI
1.8–6.8); no associations were found for malignant brain tumours (OR=1.1; 95% CI 0.8–1.6), or benign brain
tumours other than acoustic neuroma (for meningioma OR=1.1; 95% 0.7–1.5). All results for digital phone use
and use of cordless phones were close to unity. No consistent dose-response patterns were observed with
cumulative hours of phone use. Sub-analyses with different latency periods showed no coherent patterns for any
tumour types. ORs for acoustic neuroma were of the same magnitude for short-term use (1–5 years) as for
intermediate and long-term use (>10 years). Results according to location of the tumour were also driven by
acoustic neuroma (defined as occurring in the temporal area, although acoustic neuromas are actually located on
the vestibular portion of the eighth cranial nerve). [The authors report a higher response rate, but
unconventionally exclude from the denominator cases who were too ill, deceased, whose physician did not
allow contact, or had no known address. Overall, the proportion of mobile phone users in the second study
seems to be unexpectedly low. It is noteworthy that the reported prevalence of mobile phone use had not
increased much between the first and second Swedish study; the increase in the proportion of users among
controls is at the most 6%. According to the Swedish Post and Telecom Agency, the number of mobile phone
subscriptions divided by the total Swedish population size (including also age groups with no mobile phone use)
increased from 28% to 71% between 1996 and 2000 (PTS, 2011). Although some persons may have had
multiple subscriptions, these cannot explain the entire difference in the increase in proportion of users. In
addition, the second study had no requirement on amount of use to be defined as a mobile phone user which the
first study had, reflected by the higher median hours of use in the first study compared to the second. Therefore,
an even higher increase in the proportion of mobile phone users would have been expected. A self-administered
paper questionnaire has the disadvantage that it immediately reveals all questions to the participant, and with no
clear definition of mobile phone use in terms of amount required, participants with small amounts of phone use
may be more inclined to answer “no” when they see the complicated follow-up questions that need to be filled
out after a “yes” answer. It is also likely that there is a difference between cases and controls. The long delay
between diagnosis and case recruitment tends to lead to loss of high grade tumours. A substantially increased
risk for acoustic neuroma already after less than 5 years of phone use is unlikely to reflect causality as this is a
slow growing tumour likely to have been present several years before detection (Thomsen & Tos, 1990)].
350
351
352
353
354
355
In further analyses of malignant brain tumours in the same material (Hardell, Hansson Mild &
Carlberg, 2002), increased risks were reported for ipsilateral use of mobile phones, although with no coherent
pattern with latency periods or amount of use. Furthermore, reduced risks for contralateral use were also
observed. For example, the risk for malignant brain tumours associated with ipsilateral use of an analogue
mobile phone was 1.85 (95% CI 1.16–2.96) whereas the risk for contralateral use was 0.62 (95% CI 0.35–1.11).
A similar pattern was found for use of digital phones. These analyses were adjusted for socio-economic status.
356
357
In a third paper on the same material, the authors report results where the matching had been
resolved, using unconditional logistic regression to estimate ORs (Hardell, Hansson Mild & Carlberg, 2003).
10
358
359
360
361
362
363
364
The exposure definition had also changed, compared to the original paper, defining the unexposed group as not
using any type of mobile or cordless phone. Generally, results were similar to the two previous papers, except
that the reduced risks for contralateral use had disappeared, and there appeared to be a relation with phone use
latency, which was not seen in the original analyses. Socioeconomic status was added as a potential confounder,
but matching variables were only partially controlled for (age and sex, but not geographical region). [As results
of matched analyses with the new exposure definition were not presented or discussed, there is no possibility to
assess the impact on the results from resolving the matching.]
365
366
367
368
369
370
371
372
373
374
375
376
377
378
A number of additional re-analyses of these data have also been published. One publication stratified
analyses on age at diagnosis (Hardell et al., 2004b), and found the highest ORs in the youngest age group (20–
29 years). The OR for use of an analogue phone with >5 years latency was 8.17 (95% CI 0.94–71) and for
cordless phones 4.3 (95% CI 1.22–15) for all brain tumours combined. Another publication stratified data on
population density (Hardell, Carlberg & Hansson Mild, 2005b), as mobile phone base stations are more densely
situated in urban areas, which gives lower average output power levels during phone calls in urban compared
with rural areas (Lönn et al., 2004b), and therefore RF exposure during a mobile phone call is hypothesized to
be higher in rural than in urban areas. The re-analysis of the case-control data found risk estimates that were
highest in rural areas for use of a digital mobile phone at least 5 years prior to diagnosis, with an OR for use in
rural areas of 3.2 (95% CI 1.2–8.4), compared to 0.9 (95% CI 0.6–1.4) in urban areas, but the results were based
on very small numbers. In addition, risk estimates for cordless phones were also higher in rural areas. [If RF
fields were causally related to brain tumour risk one would not expect the risk associated with cordless phone
use to be higher in rural areas than urban as the cordless phone base station is placed inside the home and
exposure levels are therefore independent of population density.]
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
A third case-control study of mobile phone use and brain tumours was performed by Hardell and coworkers, published in two papers, benign and malignant brain tumours separately (Hardell, Carlberg & Hansson
Mild, 2005a; 2006c). Incident cases of brain tumours aged 20–80 years diagnosed between July 1, 2000 and
December 31, 2003 were identified through the regional cancer registers covering the geographical regions of
Uppsala/Örebro and Linköping in Sweden. In total, 1097 eligible patients with primary brain tumours were
identified, and 729 participated in the study (66% of eligible). The participation rate is reported to be 88% for
malignant and 89% for benign tumours. [As in the previous studies, participation rates are unconventionally
calculated; i.e. omitting a large proportion of cases from the denominator (e.g. deceased cases). The
participation rate is likely to be lower for malignant brain tumours than for benign because of the poorer
prognosis (91% of deceased cases had a malignant tumour), and is probably below 60% for malignant tumours].
One control per case was randomly selected from the population registry matched to cases on age in 5-year age
groups, in total 820 controls, of whom 692 participated (84%). Information on exposure to cellular and cordless
phones was collected through mailed questionnaires and was completed over the phone. Analogue phones had
been used by 18% of cases and 11% of controls, with a one year latency period, defining use as any mobile
phone use, regardless of total amount of use. Digital phones had been used by 57% of cases and 50% of
controls. Hours of phone use was categorized according to the median value among the controls (80 h for
analogue and 64 h for digital phones). Unconditional logistic regression was used to estimate ORs, ignoring the
matching, with adjustment for age, sex, socioeconomic status and year of diagnosis; all controls were used in the
analyses of all specific types of tumours.
398
399
400
401
402
403
404
405
406
407
408
For all malignant tumours combined, an odds ratio (OR) of 2.6 (95% CI 1.5–4.3) was reported for
any use of an analogue mobile phone at least one year prior to diagnosis. The corresponding result for digital
phones was 1.9 (95% CI 1.3–2.7) and for cordless phones 2.1 (95% CI 1.4–3.0). For phone use starting 1–5
years prior to diagnosis a risk estimate for malignant tumours of 1.6 (95% CI 1.1–2.4) was reported for digital
phone use, and 1.8 (95% CI 1.2–2.8) for cordless phones. No case and 3 controls had used an analogue phone
with such short latency. With >10 years latency period the risk estimates were 3.5 (95% CI 2.0–6.4), 3.6 (95%
CI 1.7–7.5) and 2.9 (95% CI 1.6–5.2) for analogue, digital and cordless phones, respectively. Risk estimates
were higher for high grade astrocytoma. In analyses of the total number of hours of phone use dichotomized at
the median hours among controls, risk estimates were generally higher in the highest exposure category; >80 h
of analogue phone use was associated with an OR of 4.0 (95% CI 2.2–7.3), digital phone use >64 h with an OR
of 2.4 (95% CI 1.6–3.7), and cordless phone use >243 h with an OR of 2.4 (95% CI 1.5–3.6).
409
410
411
412
For meningioma, any use of an analogue mobile phone at least one year prior to diagnosis was
associated with an OR of 1.7 (95% CI 0.97–3.0), and for >10 years of use the OR was 2.1 (95% CI 1.1–4.3). For
digital and cordless phones the ORs were identical for any use at least 1 year before diagnosis, OR=1.3 (0.9–
1.9), while >10 of use of a cordless phone was associated with an OR of 1.9 (0.97–3.6). An amount of use above
11
413
414
median was associated with an OR of 2.2 (95% CI 1.1–4.3) for analogue phone use, 1.7 (95% CI 1.1–2.6) for
digital phone use, and 1.7 (95% CI 1.1–2.6) for cordless phone use.
415
416
417
418
419
420
421
422
423
For acoustic neuroma any analogue phone use was associated with an OR of 4.2 (95% CI 1.8–10),
digital phones 2.0 (95% CI 1.05–3.8) and cordless phones 1.5 (0.8–2.9). Risk estimates were raised for all
latency periods of use of analogue phones; >1–5 years was associated with an OR of 9.9 (95% CI 1.4–69), for
5–10 years the OR was 5.1 (95% CI 1.9–14) and for >10 years 2.9 (95% CI 0.9–8.0). Few cases had started to
use a digital or cordless phone more than 10 years prior to diagnosis. Risk estimates for shorter latency periods
of digital or cordless phone use were generally raised but not significantly; a significantly raised risk estimate
was only found for digital phone use starting >5–10 years prior to diagnosis (OR= 2.7; 95% CI 1.3–5.7).
Amount of use above median total hours were associated with ORs 6.0 (95% CI 2.2–17), 2.5 (95% CI 1.2–5.2),
and 1.7 (95% CI 0.8–3.5), for analogue, digital, and cordless phone use, respectively.
424
425
426
427
428
429
Risk estimates for different anatomical locations of the tumour did not differ much for malignant
tumours; they were slightly higher for frontal lobe tumours than for tumours in the temporal lobe or at other
locations. For benign tumours (excluding acoustic neuroma), odds ratios were higher for temporal lobe tumours;
ORs over 2 were observed for tumours in the temporal lobe already within 5 years of use of digital and cordless
phones, and for >10 years of phone use ORs were above 5 for all three types of phones (all cases in this
category had meningioma).
430
431
432
433
Results did also not vary much according to laterality of phone use in relation to laterality of the
tumour; e.g. for malignant tumours increased risks were found for both ipsilateral and contralateral phone use in
all analyses except for digital phone use where the OR for contralateral use was not statistically significantly
raised. The same pattern was seen for acoustic neuroma.
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
[Compared with the earlier studies by Hardell and co-workers, reported risk estimates in the third
study are considerably higher in all exposure categories, also for very short durations of exposure. For example
for malignant brain tumours, the OR for digital phone use with a latency period of 1–5 years was 1.6 (95% CI
1.1–2.4; 100 exposed cases) whereas in the second study, with case recruitment during 1997–2000, the
corresponding result was 1.08 (95% CI 0.81–1.43; 100 exposed cases, latency period 1–6 years). In the first
study (case recruitment 1994–1996) an OR of 0.98 (95% CI 0.63–1.50; 53 exposed cases) was reported for >1
year of mobile phone use. This pattern is similar also for cumulative hours of use categorized according to the
median hours among controls. The third study reports dose-response patterns for both malignant and benign
tumours, whereas no consistent dose-response patterns were found in the first two studies. For analogue phones
the median cumulative hours of use among controls was 224 h in the earliest study, 85 h in the second and 80
hours in the third. The corresponding numbers for digital phones were 88 h, 55 h, and 64 h. The first study
required a minimum amount of 8 cumulative hours of use to be regarded as a mobile phone user, while the
second and third studies had no such requirements. This may explain the higher amount of use in the first study
despite the early observation period. The low prevalence of mobile phone use in the third study is surprising,
especially when considering that no minimum amount of use was required; only 51% of the controls were
mobile phone users (56% of cases). The Swedish Post and Telecom Agency (PTS) reported that 87% of a
random sample of the Swedish population in the age range 16–75 years were mobile phone users in 2002, and
90% in 2003 (PTS, 2003). The questionnaire used was the same as in the second study discussed above, and
required the participants’ ability to distinguish between NMT450, NMT900, and GSM phones. For each of these
three types of phones separately, the participants were asked to state the average minutes per day that they had
used the mobile phone encompassing their entire history of mobile phone use, i.e. one estimate that should cover
up to around 15 years of mobile phone use, with no place to report changes in these habits. This makes estimates
of cumulative hours of phone use likely to be highly uncertain, as the amount of mobile phone use has changed
dramatically since mobile phones became available. Similarly, the proportion of time that a hands free device
had been used was also only asked once, with only one short line to state the years, and no place to report
changes. The side of the head where the phone was held were only referring to present time, i.e. when the
questionnaire was filled out, with no room to report changes. Recall bias is likely in such unspecific exposure
measures. All participants were contacted by telephone “to verify exposures and get additional detailed
information” by an interviewer that was blind to case-control status. The authors do not report whether they
recorded how often the interviewer by accident found out the case-control status during the interview, which
seems likely to be common considering that interviews were made with cases who had been diagnosed with a
life threatening disease, or a disease that may have affected their mobile phone habits through unilateral hearing
loss. And of course, cases and controls are not blind to their own disease status when they answer the
questionnaire. Thus, participants and personnel in this type of epidemiological study can never be blinded in the
same way as in an experimental study.]
12
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
Hardell and co-workers conducted a study of deceased cases of malignant brain tumours (Hardell,
Carlberg & Hansson Mild, 2010), who would have been included in the second and third of their previous
studies (Hardell et al., 2002; Hardell, Carlberg & Hansson Mild, 2006c) had they not died before being
contacted for inclusion in these studies. The study included 535 deceased cases of malignant brain tumours
diagnosed between 1997 and 2002 aged 20–80 years (cases occurring in 2003 were excluded because the Cause
of Death registry was not updated longer than to 2002 when this study started). Deceased controls were selected
from the Cause of Death register, matched to cases on year of death, sex, age and geographic region. Two
controls per case were selected; one who had died from another cancer diagnosis, and one who had died from
another major chronic disease such as cardiovascular disease, neurologic disease, lung disease, gastrointestinal
disease, infection, and diabetes. No relative could be identified for 64 cases, and no control fulfilled the
matching criteria for 7 cases. A questionnaire with questions about the mobile phone use history of the deceased
case/control was sent to a close relative. Relatives were contacted between November 2006 and August 2008.
Finally, 346 relatives to cases (65% of all deceased, 75% of contacted), 343 relatives to controls who died from
other cancers (74%) and 276 relatives to controls who died from other diseases (60%) participated in the study.
Results were very similar to the results from the pooling of the two original case-control studies (discussed
below), except that no raised risks were found for use of cordless phones. [A severe limitation of the study is the
reliance on the ability of close relatives to report correctly about mobile phone use for distant time periods for a
relative who died from a malignant brain tumour several years earlier. The authors discuss that the use of
controls who had died from other cancer types or some other malignant disease would offset recall bias when
relatives report on the subjects’ past mobile phone use. It seems unlikely, however, that relatives of persons who
have died from other cancers or from other diseases such as cardiovascular diseases would believe that mobile
phone use had caused the disease that their relative died from, as is likely to be the case for relatives of many
cases of malignant brain tumours, especially as headlines about mobile phone use causing brain tumours
appeared several times in the media in connection to the publication of the earlier studies by the Hardell group.
An indication of the difficulty for relatives to report correctly about amount of mobile phone use 10 to 20 years
earlier for their deceased relative is seen in the difference in the median cumulative hours of use reported by
controls in the original studies, compared with that reported by relatives to controls in this study. The controls in
the two original studies reported a median of 85 and 80 h for analogue phone use, 55 and 64 h for digital phone
use, and 195 and 243 h for cordless phone use. The relatives of deceased controls reported 149 h for analogue
phone use, 183 h for digital phones use, and 548 h for cordless phone use, even though the time period covered
should be the same. Considering these severe limitations, the results of this study are not tabulated.]
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
A fourth study of mobile phone use and brain tumours was conducted by Hardell and co-workers,
published in three separate papers (Carlberg et al., 2013; Hardell et al., 2013a; b). Incident cases of brain
tumours aged 18–75 years diagnosed during the period 2007 to 2009 were identified through the six regional
cancer registers in Sweden (in one region from 2008). Only histologically confirmed cases were included. In
total, 1314 eligible patients with malignant brain tumours and 1010 with benign brain tumours were identified.
593 of the patients with malignant tumours (45% of eligible), and 814 (81%) with benign tumours participated
in the study. The main reason for non-participation among patients with malignant tumours was that the case
had died (the study included only living persons), while the main reason for cases with benign tumours was
refusal. Of the participating cases with benign tumours, 709 were diagnosed with meningioma and brain
tumours with acoustic neuroma. Other types of benign tumours were rare and were not analysed further. One
control per case was randomly selected from the Swedish population registry, matched to the case on age, sex,
and geographic region of residence, in total 1601 controls of whom 1368 (85%) participated; the number of
selected controls was determined after having excluded cases who could not be approached because they had
died, the physician did not give permission or similar reasons. Results for acoustic neuroma were only presented
as pooled with two of the previous studies conducted by the same group, where the new cases constitute a small
part of the data (73 out of 316 cases) [these results are discussed below when pooled results are described].
Information about exposure to cellular and cordless phones was collected in the same way as in previous
studies, i.e. through mailed questionnaires and completed over the phone. Any type of mobile phone had been
used by 92% of cases with a malignant brain tumour, 84% with meningioma, and 89% of controls, with a one
year latency period, defining use as any amount of mobile phone use and at least 39 h cumulative use of a
wireless phone (mobile or cordless phone). Time since first mobile phone use was categorized into <1-5, >5-10,
>10-15, >15-20, >20-25, and >25 years [the maximum time possible for use of a handheld mobile phone is 23
years. Thus, for time periods longer than 23 years only car phones and so called bag phones were available, on
which the antenna was situated far from the body with considerably lower RF exposure to the head. The first
handheld mobile phones were available on the market in Sweden from 1987]. Separate analyses were made for
use of analogue, digital 2G, digital UMTS, and any type of mobile phones, cordless phones, any type of digital
phone (including mobile and cordless phones), and any type of wireless phones. Unconditional logistic
regression was used to estimate ORs, ignoring the matching, with adjustment for age, sex, socioeconomic status
13
528
529
and year of diagnosis [the matching variable residential geographic region was ignored]; all controls were used
in the analyses of all specific types of tumours.
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
For all malignant brain tumours combined, an OR of 1.6 (95% CI 0.99–2.7) was reported for use of
any type of mobile phone at least one year prior to diagnosis. For mobile phone use starting 1–5 years prior to
diagnosis the OR was 1.8 (95% CI 1.0–3.4), and the ORs were lower, and non-significant, for starting mobile
phone use during the three times periods between >5 and ≤20 years before diagnosis. The highest OR was
observed for >25 years since start of mobile phone use (OR=2.9; 95% CI 1.4–5.8) [this category does not
include any use of handheld phones]. Similar patterns were reported for analogue and digital mobile phone use
separately (only a small number of persons had used UMTS phones longer than 5 years), and also for use of
cordless phones, although for cordless phones the highest risk estimate was found in the category >15–20 years
since first use (very few persons had used cordless phones longer). For any type of digital phone (digital mobile
or cordless phones) results were very similar to results for cordless phones [it seems like all cordless phones
were classified as digital, although the first cordless phones were analogue; digital cordless phones became
available on the market late 1992. Thus, the analysis of “digital type” of phones is likely to include also
analogue phones as no digital phones were available early enough to be classified into the category >20–25
years since first use. Furthermore, it seems unlikely that respondents would be able to distinguish between
analogue and digital cordless phones.] The total number of hours of mobile phone use was categorized into
quartiles according to the distribution of any type of wireless phone use among the controls (>39–405, 406–
1091, 1092–2376, >2376 h). For any type of mobile phone use, ORs were above unity for all categories of
cumulative hours of use, but with no consistent dose-response pattern over the first three quartiles (ORs going
from 1.4, 1.7, 1.5), and with an OR of 2.8 (95% CI 1.6–4.8) in the highest quartile (which included 25% of
mobile phone users among cases and 13% among controls). ORs were similar for cordless phone use, although
the ORs in the third and fourth quartiles were slightly higher (2.1 and 3.1). To evaluate potential recall bias,
meningioma cases were used as controls in an analysis of malignant brain tumours in relation to time since first
use, with results that were very similar to the original analysis. [Surprisingly, meningioma cases were not used
as controls in analyses of cumulative hours of use, where recall bias is of greater concern].
554
555
556
557
558
For meningioma, any use of a mobile phone at least one year prior to diagnosis was associated with
an OR of 1.0 (95% CI 0.7–1.4). ORs were close to unity for all types of phones analysed separately, or together,
and for any latency periods. ORs for cumulative hours of mobile phone use were close to or below unity in all
quartiles except in the highest, where ORs were raised for use of analogue and digital phones, as well as for
cordless phones, the latter being 1.8 (95% CI 1.2–2.8).
559
560
561
562
563
564
565
566
567
568
For malignant brain tumours, the OR for ipsilateral use of any type of mobile phone was slightly
higher than for contralateral use (OR=1.7; 95% CI 1.01–2.9 and 1.4; 95% CI 0.8–2.5, respectively). Similar
findings were seen for cordless phone use. For meningioma the corresponding results did not differ according to
laterality of phone use. Analyses restricted to tumours occurring at temporal, temporofrontal, tempoparietal, or
temporooccipital locations showed higher risk estimates for malignant brain tumours, but again with no
consistent dose response pattern. The highest OR was observed for >25 years since first use (OR=4.8; 95% CI
1.7–14) [only bag phones and car phones], and the next to highest was in the category 1–5 years since first use
(OR=3.0; 95% CI 1.2–7.6). A similar pattern was observed for cordless phones. [No analyses were restricted
only to the temporal lobe, and results for other lobes were not shown]. For meningioma, the corresponding
analyses showed results that were similar to the overall findings.
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
[Effect estimates reported in this fourth study from the Hardell group are lower than those reported in
the third study, for both malignant brain tumours and meningioma. The highest ORs were observed for time
periods when handheld mobile phones were not yet available on the market, and for short term use (>1–5 years
since first use). The prevalence of mobile phone use among controls is considerably higher than in previous
studies by the same group, almost 90%, and more in line with other surveys. The Swedish Post and Telecom
Agency reported a prevalence of 94% 2007 and 2008, and 95% 2009 (a few percent higher if counting “ever
use”) (PTS, 2011). The unexposed group is very small, only 22 cases (3.7%) of malignant brain tumours and
107 controls (7.8%), which affects all analyses. It is questionable whether there were unexposed subjects in all
categories of age and sex. In previous studies, the unexposed group was defined as “no wireless phone use”,
while the fourth study also included persons with in total ≤39 h of wireless phone use into the unexposed group
(3rd percentile). The reason for choosing this particular cutpoint is unclear, as it does not correspond to a
workweek in Sweden (which the authors claim), the legal working time in Sweden is 40 h/week, and it still
leaves a very small unexposed group. Furthermore, results for the temporal lobe are not reported separately,
only together with overlapping lobes, which is not how analyses have been performed in any previous study.
Unfortunately, the distribution of hours of phone use is only shown for all wireless phones combined (including
14
584
585
586
587
588
589
590
591
592
593
also cordless phones), which makes comparisons with previous studies difficult. The questionnaire used is the
same as in the second and third studies (discussed in connection to the third study), with the additional
requirement that participants need to be able to distinguish also UMTS/3G phones from the three older types.
The participant was asked to estimate the average minutes per day that he/she had used each of the four different
mobile phone types (one estimate per phone type), which could now encompass a history of mobile phone use
that could be over 20 years. This makes estimates of cumulative hours of phone use likely to be highly
uncertain, as the amount of mobile phone use has changed dramatically since mobile phones became available,
especially during the last years of the fourth study. The questions on the side of the head where the phone was
held were also only referring to present time, i.e. when the questionnaire was filled out, with no room to report
changes. Recall bias is likely in such unspecific exposure measures.]
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
Hardell and co-workers have published several articles where the data from the second and third
studies (Hardell et al., 2002; Hardell, Carlberg & Hansson Mild, 2005a; 2006c) were pooled, mostly reporting
overlapping analyses in the various publications (Hansson Mild, Hardell & Carlberg, 2007; Hardell, Carlberg &
Hansson Mild, 2006a; b; Hardell et al., 2006; Hardell & Carlberg, 2009; Hardell, Carlberg & Hansson Mild,
2011b). The same study design was used in the two studies to allow pooling of the data, but formal tests of
heterogeneity between the results from the two studies were not made. In the pooled analysis the OR of
malignant brain tumours for less than 5 years since first digital mobile phone use was 1.2 (95% CI 0.96–1.5),
and were similar also for analogue and cordless phones. There were statistically significant increased risk
estimates for 5–10 and >10 years latency periods for both digital and cordless phones, for analogue phones only
for >10 years since first use. Malignant brain tumours was associated with an OR of 1.1 (95% CI 0.8–1.6) for
contralateral analogue phone use, and 2.1 (95% CI 1.5–2.9) for ipsilateral use. For meningioma, significantly
increased risk estimates were reported for analogue phones with a 10 year latency and for cordless phones after
5–10 years. For acoustic neuroma all types of phones were associated with significantly increased risk estimates
after less than 5 years since first use; e.g. for analogue phone use the OR was 2.3 (95% CI 1.2–4.1), and
increased to over threefold with >5–10 and >10 years latency. Cumulative lifetime hours of use were presented
with the cut point at the median hours among controls, showing a clear dose-response pattern only for analogue
phone use and acoustic neuroma risk. In addition, analyses were also made with cut-points at 1000 and 2000 h
of use for malignant brain tumours and 500 and 1000 h for benign [no rationale for the choice of cut points was
presented]. Increased odds ratios were reported for all exposure categories for malignant tumours and for
acoustic neuroma, with highest risk estimates in the highest exposure category (OR=5.9; 95% CI 2.5–14 for
malignant tumours and OR=5.1; 95% CI 1.9–14) for acoustic neuroma). One paper presented results with hours
of use categorized according to tertiles (Hardell et al., 2006), and another paper added analyses of hours of use
as a continuous variable (Hansson Mild, Hardell & Carlberg, 2007), which essentially did not change the overall
impression of the results. Several papers presented analyses stratified according to age at first use (Hardell,
Carlberg & Hansson Mild, 2006a; b; Hardell & Carlberg, 2009). Highest risk estimates were found among
persons who started to use a mobile phone before 20 years of age; for astrocytoma in this age group the OR was
5.2 (95% CI 2.2–12) and for acoustic neuroma 5.0 (95% CI 1.5–16) for ever use of a mobile phone >1 year
prior to diagnosis. In a letter to the editor (Hardell, Carlberg & Hansson Mild, 2011b), pooled analyses were
presented using different age ranges, including one similar to that used in the Interphone study. In addition,
analyses was made where cordless phone users were included in the unexposed group, in the same way as in
Interphone, as well as analyses restricted to tumours in the temporal lobe [none of the previous pooled analyses
had presented results according to tumour location]. From these results, it is evident that slightly stronger risk
estimates were found in the youngest (<30 years at diagnosis) and oldest (>59 years) age-groups, although it is
unclear whether the differences between age-groups were statistically significant. Adding users of cordless
phones to the unexposed category lowered the risk estimates only marginally. Risk estimates for tumours in the
temporal lobe were not higher than in other, less exposed areas of the brain.
630
631
632
633
For acoustic neuroma, an additional pooled study was published in 2013 (Hardell et al., 2013b),
which included 73 new cases and 1368 new controls identified during 2007–2009 in addition to all the cases and
controls used for the pooled study discussed above. The total number of cases and controls were 316 and 3530,
respectively. Results were essentially similar to the previous pooled analysis discussed above.
634
635
636
637
638
639
640
[Some of the implausible results in the individual studies disappeared in the pooled analyses. The
increased risk for malignant brain tumours after a very short latency period in the third study was lowered by the
lack of association in the 2002 study. The observed reduced risk estimate for malignant brain tumours associated
with contralateral mobile phone use in the 2002 study (OR=0.62; 95% CI 0.35–1.11 for analogue phone use)
was offset by the increased risk for both ipsi- and contralateral phone use in the third study (OR=2.6; 95% CI
1.3–5.4 for contralateral analogue phone use and OR=3.2; 95% CI 1.6–6.2 for ipsilateral use). The difference
between the results in the two studies for contralateral phone use is statistically significant. The pooled analyses
15
641
642
643
644
645
have the same limitations as discussed above for the individual studies. A pooled analysis was also conducted
including the data from the study of deceased cases and controls (Hardell, Carlberg & Hansson Mild, 2010), but
these additional data have severe problems with potential exposure misclassification, as discussed above, and
therefore the pooled results that include data from deceased cases and a selection of deceased controls (Hardell,
Carlberg & Hansson Mild, 2011a) are not further discussed here.]
646
Early US studies
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
Muscat et al. conducted two case-control studies in the US, one of malignant brain tumours (Muscat
et al., 2000), the other of acoustic neuroma (Muscat et al., 2002), using the same ascertainment and data
collection procedures. Eligible patients were identified from five medical centres, and were diagnosed with a
malignant brain tumour 1994–1998, in the ages 18 to 80 years, or acoustic neuroma 1997–1999, with no upper
age limit. Controls were selected from the same hospitals frequency matched on age, sex, race and month of
admission, with a variety of benign conditions, except in two centres where controls were recruited primarily
from patients with malignant conditions, excluding leukaemia and lymphoma. In total, 571 cases with malignant
brain tumours were contacted of whom 469 participated. An additional 55 eligible cases were never approached,
which gives a participation rate of 75%. Among controls 422 participated (90%). Participation rates were not
reported for acoustic neuroma, in total 90 patients and 86 controls participated. Information about mobile phone
use was obtained by structured personal interviews (proxies were interviewed for 9% of brain tumour cases and
1% of controls, and one acoustic neuroma case). The interview asked about years and amount of mobile phone
use for each phone the respondent had ever used. Information about the preferred side of the head referred to
current use, i.e. at the time of the interview. Regular mobile phone use, defined as having a mobile phone
subscription, was reported by 14% of brain tumour cases and 18% of their controls. For acoustic neuroma the
corresponding numbers were 20% of cases and 27% of controls. Analyses were made with unconditional
logistic regression models, adjusted for age, education, sex, race, study centre, occupation, date of interview,
and for brain tumours also proxy subject. No raised risks of malignant brain tumours were seen for regular use,
frequency of use, or duration of use, or for site or histologic subtype of brain cancer. The OR for ≥4 years of
regular use was 0.7 (95% CI 0.4–1.4; 17 exposed cases), and for the highest category of cumulative hours of use
(>480 h) the OR was 0.7 (95% CI 0.3–1.4; 14 exposed cases). The OR for a malignant brain tumour in the
temporal lobe was 0.9 (95% CI 0.5–1.7). A non-significantly raised OR was seen for neuroepitheliomatous
tumours (OR=2.1; 95% CI 0.9–4.7; 14 exposed cases). There was no trend in risk of acoustic neuroma in
relation to cumulative measures of phone use, and no significant relation between side of phone use and side of
tumour. The OR for 3–6 years of mobile phone use was 1.7 (95% CI 0.5–5.1), while it was 0.5 (95% CI 0.2–
1.3) for 1–2 years. [The studies are limited by the short duration of mobile phone use among the majority of
subjects, and by the hospital-based identification of cases and selection of controls from other patient groups at
the hospitals, as discussed above.]
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
In another case-control study from the US (Inskip et al., 2001), Inskip and colleagues identified
patients with intracranial tumours of the nervous system from five hospitals acting as regional referral centres
for diagnosis and treatment of brain tumours. Eligible patients were diagnosed between June 1994 and August
1998, 18 years or older. Controls were admitted to the same hospitals as the cases with non-malignant
conditions frequency matched to cases on age, sex, race, hospital, and proximity of residence to hospital. The
most common diseases among controls were injuries and disorders of the circulatory, musculoskeletal,
digestive, and nervous systems. In total, 782 cases (92% of eligible) and 799 (86%) controls participated in the
study. Information about mobile phone use and covariates were obtained through personal interviews. The
interview asked about years of mobile phone use, and average amount of use over the whole period. Information
about the preferred side of the head referred to current use, i.e. at the time of the interview. Most of the cases
were interviewed within three weeks after diagnosis. Proxies were interviewed for 16% of patients with glioma,
8% with meningioma, 3% with acoustic neuroma and 3% of controls. Conditional logistic regression was used
for analyses; apart from matching variables adjustment was made for date of interview, type of respondent,
education, household income, type of health coverage, marital status, religion, radiotherapy to the head or neck,
and handedness. Participating cases included 486 glioma, 197 meningioma, and 96 acoustic neuroma. Of
controls, 29% reported having used a mobile phone at least five times. Regular mobile phone use, defined as at
least two times per week, was not associated with any of the tumour types; e.g. the OR for glioma was 0.8 (95%
CI 0.6–1.2). Also, results showed no consistent link between average daily use, duration of use, or cumulative
use of mobile phones (mainly analogue) and risk of brain tumour overall or according to histological subtype or
anatomic site or side of use, e.g. ≥60 minutes of daily use was associated with an OR for glioma of 0.7 (95% CI
0.3–1.7; 12 exposed cases), and duration of use ≥5 years with OR=0.6 (95% CI 0.3–1.4; 11 exposed cases).
Corresponding results for meningioma was 0.5 (95% CI 0.1–2.2; 5 exposed cases) and 0.7 (95% CI 0.2–2.4; 6
exposed cases), and for acoustic neuroma OR=0.3 (95% CI 0.0–2.7; 1 exposed case) and 1.9 (95% CI 0.6–5.9; 5
16
698
699
700
701
702
exposed cases). For glioma in the temporal lobe the OR=0.8 (95% CI 0.5–1.4; 41 exposed cases). A nonsignificantly reduced OR was seen for neuroepitheliomatous tumours (OR=0.5; 95% CI 0.1–2.0; 8 exposed
cases). [This study suffers from the same limitations as the studies by Muscat et al., described above, i.e.
hospital-based design and few subjects that have used mobile phones for a longer time period, thus results are
only of interest for evaluation of potential effects of short-term mobile phone use.]
703
Interphone studies
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
The Interphone study is an international collaborative case-control study conducted in 16 study
centres from 13 countries (Australia, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan, New
Zealand, Norway, Sweden, and the United Kingdom) using the same common core study protocol (Cardis et al.,
2007; Interphone Study Group, 2010; 2011). Eligible patients were diagnosed between 2000 and 2004 (the
study period varied between centres from 2 to 4 years), and were in the age range 30–59 years at diagnosis.
Several study centres have published national or pooled regional analyses (Christensen et al., 2004; Christensen
et al., 2005; Hepworth et al., 2006; Hours et al., 2007; Klaeboe, Blaasaas & Tynes, 2007; Lahkola et al., 2007;
Lahkola et al., 2008; Lönn et al., 2004a; 2005; Schlehofer et al., 2007; Schoemaker et al., 2005; Schüz et al.,
2006a; Takebayashi et al., 2006; Takebayashi et al., 2008), most often using wider age ranges. Ascertainment of
cases was population based in all centres except Japan and France, where it was hospital-based. Cases were
ascertained in close collaboration with neurosurgery, oncology, neurology, and otorhinolaryngology clinics for
rapid recruitment, and completeness was checked through secondary sources, e.g. cancer registries. Diagnoses
were either histologically confirmed or based on unequivocal imaging. In total, 4301 glioma, 3115 meningioma,
and 1361 acoustic neuroma were identified in the age range 30–59 years, and interviews were completed with
2765 glioma (64%; range across centres 36–92%), 2425 meningioma (78%; range 56–92%), and 1121 acoustic
neuroma cases (82%; range 70–100%).
720
721
722
723
724
725
726
727
728
729
730
731
732
Controls were randomly selected from population registries (one Canadian centre, Denmark, Finland,
Germany, Israel, Italy, Norway, Sweden), electoral lists (Australia, one Canadian centre, part of France, New
Zealand), general practitioners’ lists (both UK centres), or random digit dialling (one Canadian centre, part of
France, Japan). Controls were individually matched to cases in two Canadian centres, France, Israel, Japan, New
Zealand, and UK North, while the other 9 centres used frequency matching. Matching variables were birth year
(within 5-year categories), sex and study region; in Israel also ethnic origin. In total 14 354 controls were
identified in the age range 30–59 years, and 7658 participated in the interview (53%; range across centres 42–
74%). In the overall pooled international analyses a post hoc matching was applied in centres with frequency
matched controls, which, together with the more restricted age range, led to the exclusion of a large number of
subjects from the international analyses. The numbers of cases and controls included in the analyses were
additionally reduced because of restraints from the matching; the final numbers were 2708 glioma cases with
2972 matched controls, 2409 meningioma cases with 2662 matched controls, and 1105 acoustic neuroma cases
with 2145 matched controls.
733
734
735
736
737
738
739
740
741
742
743
744
745
A detailed history of mobile phone use and potential confounding factors was collected through
structured personal interviews using a computer assisted questionnaire. Proxies were interviewed for
participants who had died or were too ill (13% of glioma, 2% of meningioma, 0.3% of acoustic neuroma cases,
and 1% of controls). Participants were asked if they had used a mobile phone regularly, defined as on average at
least once per week over a period of at least 6 months, and regular users were further asked about number and
duration of calls, and type of phone (aided by show-cards) for each mobile phone used, or when changes of
habits had occurred. Information about the preferred side of the head referred to current use, i.e. at the time of
the interview. Regular mobile phone use was reported by 64% of glioma controls, 56% of meningioma controls
and 61% of acoustic neuroma controls. All exposure variables except time since first use were censored at one
year before the reference date (date of diagnosis for the case, and corresponding date for the matched controls).
Cumulative number of calls and hours of use were categorized at the deciles of the distribution among the
controls. Stratified analyses of cumulative hours of use were made according to time since first use (1–4, 5–9,
10+ years). In these analyses, deciles were collapsed into 1 st, 2–5th, 6–7th, 8–9th, and the 10th decile.
746
747
748
749
750
751
752
Analyses were made with conditional logistic regression models, adjusted for education. Further
adjustment for other potential confounding variables did not change risk estimates appreciably, and were not
included in the final analyses. Formal tests of heterogeneity between study centres were conducted; no evidence
of heterogeneity in effects across study centres was found. Generally, risk estimates were reduced below unity;
regular mobile phone use ≥1 year was associated with an OR of 0.81 (95% CI 0.70–0.94) for glioma, 0.79 (95%
CI 0.68–0.91) for meningioma, and 0.85 (95% CI 0.69–1.04) for acoustic neuroma. For acoustic neuroma a
latency period of ≥5 years was also applied in secondary analyses, which gave an OR of 0.95 (95% CI 0.77–
17
753
754
755
756
1.17). Risk estimates did not increase with increasing time since first use of mobile phones; ≥10 years since first
use was associated with OR 0.98 (95% CI 0.76–1.26) for glioma, 0.83 (95% CI 0.61–1.14) for meningioma, and
0.76 (95% CI 0.52–1.11) for acoustic neuroma with a 1 year latency and 0.83 (95% CI 0.58–1.19) with 5 years
latency.
757
758
759
760
761
762
For cumulative number of calls there were no increasing trends in risk estimates over the 10 exposure
categories; most risk estimates were below unity and in the 10 th decile the OR was 0.96 (95% CI 0.71–1.31) for
glioma, 0.80 (95% CI 0.55–1.17) for meningioma, 0.93 (95% CI 0.61–1.41) for acoustic neuroma with 1 year
latency and 1.55 (95% CI 0.84–2.86) with 5 years (the OR in the 9th decile with 5 year latency was 0.62; 95% CI
0.34–1.12). The only significantly increased risk estimate for cumulative number of calls was in the second
decile for acoustic neuroma with 5 years latency (OR=2.32; 95% CI 1.39–3.87).
763
764
765
766
767
768
769
770
771
772
773
For cumulative hours of use the pattern was very similar, with the exception that there were raised
ORs in the 10th decile; for glioma 1.40 (95% CI 1.03–1.89), for meningioma 1.15 (95% CI 0.81–1.62), and for
acoustic neuroma 1.32 (95% CI 0.88–1.97) with 1 year latency and 2.79 (95% CI 1.51–5.16) with 5 years, but
again no upward trend in the 9 first deciles. For all three tumour types the 9th decile had among the lowest ORs;
0.71 (95% CI 0.53–0.96) for glioma, 0.76 (95% CI 0.54–1.08) for meningioma, 0.48 (95% CI 0.30–0.78) for
acoustic neuroma with 1 year latency, and 0.60 (95% CI 0.34–1.06) with 5 years. The highest decile (≥1640 h of
use) included subjects who reported implausible number of hours of use, more common among cases than
controls, e.g. 10 glioma cases and no controls reported ≥12 h of mobile phone use/day. Excluding subjects who
reported >5 hours of use per day resulted in lower risk estimates; ORs for the 10 th decile of cumulative hours of
use were reduced to 1.27 (95% CI 0.92–1.75) for glioma, 1.02 (95% CI 0.70–1.48) for meningioma, and 1.16
(95% CI 0.75–1.80) for acoustic neuroma.
774
775
776
777
778
779
When analyses of cumulative hours of use were stratified according to time since first use, the risk
estimates in the category ≥1640 h was highest for short-term users (1–4 years prior to diagnosis) for glioma
(OR= 3.77; 95% CI 1.25–11.4) and meningioma (OR= 4.80; 95% CI 1.49–15.4), but not for acoustic neuroma
(OR= 0.63; 95% CI 0.14–2.80). For long-term users (≥10 years) in the 10th decile, ORs were 1.34 (0.90–2.01)
for glioma, 0.95 (0.56–1.63) for meningioma, and 1.93 (95% CI 1.10–3.38) for acoustic neuroma (the OR in the
second highest category of hours of use was 0.39; 95% CI 0.20–0.74 for acoustic neuroma).
780
781
782
783
Analyses were also made according to anatomic location of the tumour. Regular use of mobile
phones were associated with an OR of 0.86 (95% CI 0.66–1.13) for glioma in the temporal lobe, while the
corresponding result for meningioma was 0.55 (95% CI 0.36–0.82). In the top decile of cumulative hours of use
the OR for a tumour in the temporal lobe was 1.87 (1.09–3.22) for glioma and 0.94 (0.31–2.86) for meningioma.
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
Analyses according to laterality of phone use in relation to laterality of the tumour showed higher risk
estimates for ipsilateral mobile phone use than for contralateral in virtually all exposure categories for glioma
and meningioma; 14 out of 15 analyses showed a ratio of ipsi- to contralateral use above 1. For time since first
use the highest ratio of ipsi- to contralateral use was in the category 1–1.9 years of use for glioma and 2–4 years
for meningioma. For cumulative number of calls the ratio of ipsi- to contralateral use increased with increasing
numbers of calls; however, risk estimates were considerably reduced for contralateral use. In the top decile of
cumulative number of calls (≥27 000 calls) the OR for ipsilateral use was 1.51 (95% CI 0.91–2.51) for glioma
and 1.01 (95% CI 0.56–1.82) for meningioma, while the corresponding results for contralateral use was 0.61
(95% CI 0.32–1.18) and 0.66 (95% CI 0.30–1.46), respectively. For cumulative hours of use, the highest ratio of
ipsi- to contralateral use was found in the lowest exposure decile (<5 h). In the top decile (≥1640 h) the OR for
ipsilateral use was 1.96 (95% CI 1.22–3.16) for glioma and 1.45 (95% CI 0.80–2.61) for meningioma. The
corresponding results for contralateral use were 1.25 (95% CI 0.64–2.42) and 0.62 (95% CI 0.31–1.25),
respectively. [The higher risk estimates for ipsilateral use than contralateral in almost all exposure categories,
also with very short durations and small amounts of use, and considerable risk reductions for contralateral use,
indicate that recall bias may have affected cases’ retrospective report of the preferred side of phone use.]
799
800
801
802
803
804
805
806
For acoustic neuroma, risk estimates for contralateral mobile phone use were generally higher than
for ipsilateral use, with the exception of the highest exposure categories. [Higher risk estimates for contralateral
use are expected because the most common early symptom of acoustic neuroma is unilateral hearing loss and/or
tinnitus, which can appear several years before the disease is eventually diagnosed. In Interphone 79% of
acoustic neuroma cases and 25% of controls reported hearing problems. The majority of hearing problems
occurred more than one year prior to diagnosis; 25% of cases had hearing problems that started more than 5
years before diagnosis. Thus, many cases are likely to have changed the preferred side of phone use, or were
affected by unilateral hearing loss already before becoming a mobile phone user, because of hearing problems
18
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
caused by the tumour, which would lead to apparently higher risk estimates for contralateral use.] In the primary
analyses with 1 year latency, the OR for regular mobile phone use was 0.77 (95% CI 0.59–1.02) for ipsilateral
use and 0.92 (95% CI 0.70–1.22) for contralateral use. In the highest category of time since first use, ≥10 years,
the OR was 1.18 (95% CI 0.69–2.04) for ipsilateral use and 0.69 (95% CI 0.33–1.42) for contralateral use. Also
for cumulative hours of use and cumulative number of calls the ORs were generally higher for contralateral use,
with the exception of the top exposure decile, where ORs were raised for ipsilateral use, and considerably
reduced for contralateral use, e.g. for ≥1640 h of use the ipsilateral OR was 2.33 (95% CI 1.23–4.40) and the
contralateral OR=0.72 (95% CI 0.34–1.53). [It is noteworthy that the question asked about side of phone use
refers to current habits: “When you use a mobile phone, do you generally use it on the right or left side of your
head?” with answer alternatives left, right and both sides. One would have expected that also cases who started
to use a mobile phone ≥10 years before diagnosis or who had used a mobile phone ≥1640 h would have been
affected by hearing problems prior to their acoustic neuroma diagnosis, causing a change of the side of phone
use, which would have resulted in a reduced risk estimate for ipsilateral use and a raised estimate for
contralateral use also among long-term users. The observed pattern of results clearly indicates an effect of
reverse causation for short periods of use and small amounts (early tumour symptoms causing a change of the
preferred side of use). Results for the longest period of use and top decile of amount of use are compatible with
either recall bias or that cases with long term and/or heavy use are more prone to report the preferred side before
any hearing problems had occurred than cases with shorter periods or smaller amounts of use.]
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
[The strengths of the Interphone study are rapid case ascertainment, population-based controls (for
the great majority of centres), large sample size, data collection through structured computer assisted personal
interviews with well-trained interviewers to minimize potential interviewer bias, and careful validation and
quantification of potential sources of bias. As in all case-control studies relying on retrospectively collected selfreported information, exposure assessment has limitations, and exposure misclassification is likely to be
substantial, and possibly affected by recall bias. Although the validation study found little difference between
cases and controls in reporting of overall mobile phone use (Vrijheid et al., 2009a), there was a tendency among
cases for an increasing overestimation of mobile phone use the further back in time the reporting referred to,
which was not seen among controls. The validation study covered at the most four to six years of phone use,
while the exposure estimation referred to well over 10 years and therefore the extent of exposure
misclassification for the most distant exposure periods is unknown. Additional evidence of potential recall bias
is the reports of implausible amounts of mobile phone use, more common among cases than controls. The shape
of the exposure-response pattern for cumulative hours of use adds further indication of an effect from potential
recall bias, with no raised risk estimates in the first 9 deciles of exposure, and a raised risk only in the 10 th
decile, and with the lowest risk estimates observed in the 9th decile; for acoustic neuroma even an indication of a
downward trend through the first 9 categories. Another limitation is the low response rates, especially among
controls, which could have introduced selection bias. A validation study based on a non-responder questionnaire
estimated that selection bias may have led to a downward bias of the risk estimates for regular use by
approximately 10% for the most plausible scenarios (Vrijheid et al., 2009b). Most of the observed odds ratios in
the Interphone study were below unity, and selection bias is a plausible explanation for part of this observation,
but is unlikely to fully explain the risk reductions (Vrijheid et al., 2009b). Response rates varied considerably
between study centres, between 42% and 74% among controls. Reduced risk estimates were, however, found
also in centres with high response rates, and there was no correlation between level of risk reduction and
participation rates.]
849
850
851
852
853
854
855
856
857
858
859
860
861
862
Appendix 2 of the main Interphone study on glioma and meningioma (Interphone Study Group, 2010)
presents analyses restricted to regular mobile phone users, i.e. where non-regular users are excluded and the
lowest exposure categories are used as reference. These analyses were made post hoc in an attempt to adjust for
potential selection bias caused by greater non-participation among controls, which may have led to a downward
bias in the risk estimates. The analyses are based on the assumption that selection bias is the only reason for the
reduced risk estimates observed, and that participation is not dependent on amount of mobile phone use. If these
assumptions are incorrect, however, new bias may have been introduced, as are discussed in detail in Appendix
2 of the paper. In these analyses, an increased risk of glioma was observed already within 5 years since start of
mobile phone use (OR=1.68; 95% CI 1.16–2.41). An alternative explanation for the strongly reduced risk
estimates for glioma in the lowest exposure categories discussed in the paper is prodromal symptoms from the
tumour making not yet diagnosed glioma cases less likely to take on a new habit like mobile phone use
immediately prior to diagnosis. Restriction to regular users will in this situation introduce upward bias in the
risk estimates. [An increased risk of glioma after less than 5 years of mobile phone use is incompatible with
glioma incidence trends.]
19
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
Lahkola and co-workers conducted pooled analyses of glioma and meningioma using data from five
North European Interphone study centres (Lahkola et al., 2007; 2008), based on a wider age-range (18–69 years)
and larger number of controls than in the main Interphone study. These analyses included approximately three
times as many controls, and 45% more cases than was included from these centres in the main Interphone study.
All five studies were population-based, and the analyses included in total 1496 glioma cases with 3134 controls
frequency matched on age, sex, and geographic region, and 1204 meningioma cases with 2945 controls.
Participation rates were 60% among glioma cases (range 37–81% between countries), 74% among meningioma
cases (range 71–90%), and 50% among controls (range 42–69%). Another difference compared to the main
Interphone study was that an induction period of one year was used also in the definition of regular use, i.e.
regular use was defined as at least one call per week during a period of at least six months disregarding the 12
months immediately prior to diagnosis. Analyses were made with conditional regression analysis, conditioning
on the variables used for frequency matching. Analyses of heterogeneity in results between study centres were
conducted and no evidence of heterogeneity was found. Control of confounding from educational level, family
history of glioma, previous radiation therapy to the head and neck region, neurofibromatosis or tuberous
sclerosis did not appreciably change the results, and were not included in the final model. Exposure cutpoints for
amount of phone use were determined by the distribution among the controls. Regular mobile phone use was
associated with an OR of 0.78 (95% CI 0.68–0.91) for glioma and 0.76 (95% CI 0.65–0.89) for meningioma. No
increased risk was observed in the highest category of time since first use. The OR in the highest decile of
cumulative hours of use (>1475 h) was 1.13 (95% CI: 0.86–1.48) for glioma and 1.13 (95% CI 0.82–1.57) for
meningioma. Results did not differ between men and women or between age groups.
883
884
885
886
887
888
889
890
891
892
For glioma, analyses according to laterality of phone use in relation to laterality of the tumour showed
generally higher ORs for ipsilateral use than for contralateral use in all exposure categories. For ipsilateral use
ORs were around 1.1 for short- and intermediate-term use, and for ≥10 years since start of mobile phone use the
OR was 1.39 (95% CI 1.01–1.92). For contralateral use, ORs for short and intermediate-term use were around
0.7, while for long-term use the OR was 0.98 (0.71–1.37). [It is noteworthy that laterality specific ORs for ≥10
years since first use were higher than the corresponding overall OR (which was 0.95) for both ipsi- and
contralateral use, indicating that missing information about phone use laterality may have been more prevalent
among controls.] For meningioma, risk estimates were also generally higher for ipsilateral use than for
contralateral use, e.g. for ipsilateral regular use the OR was 0.81 (95% CI (0.66–0.99), and for contralateral
regular use it was 0.67 (95% CI 0.54–0.83).
893
894
No systematic differences between results for use of analogue and digital phones were found, neither
for glioma nor meningioma.
895
896
897
898
899
900
No analyses were presented according to lobe of the tumour. Lobe-specific results were only
presented in a separate publication based on the Swedish Interphone data (Lönn et al., 2005), where no
differences were found between lobes. [This study has the same strengths and limitations as the main Interphone
study. An additional strength is that the study was made in countries with an early introduction and widespread
use of mobile phones in the general population, and with a large number of long term users of analogue mobile
phones which have higher output levels than later generations of mobile phones.]
901
902
903
904
905
906
907
908
909
910
911
912
913
914
Schoemaker and colleagues presented results from pooled analyses of acoustic neuroma based on
data from six North European Interphone study centres in five countries (Schoemaker et al., 2005), adding also
the UK-North study centre to the studies included by Lahkola et al. described above (Lahkola et al., 2007;
2008). As in the studies by Lahkola et al., a wider age-range and larger number of controls was included
compared to the main Interphone study. Participation rates were 83% among cases (range between centres 69–
91%) and 51% among controls (range 42–69%). In total, 676 cases and 3546 controls were included in the
analyses. Analyses were made as in the studies by Lahkola et al. described above. No evidence of heterogeneity
in results between study centres was found. Regular mobile phone use was associated with an OR of 0.9 (95%
CI 0.7–1.1), and the risk estimate did not increase with increasing number of years since first use, nor with
amount of use assessed as number of calls or number of hours of use. ORs for contralateral mobile phone use
were slightly higher than for ipsilateral use in all exposure categories except the highest. The highest risk
estimate was observed for a total duration of mobile phone use of 10 years or more (OR= 1.8; 95% CI 1.1–3.1),
while the OR for at least 10 years since first ipsilateral use was 1.3 (95% CI 0.8–2.0). No differences were found
between results for use of analogue phones and digital phones.
915
916
917
Cardis and co-workers conducted a pooled analysis of case-control data on glioma and meningioma
from 5 Interphone countries (Australia, Canada, France, Israel, and New Zealand) (Cardis et al., 2011a),
including 808 (62%) glioma and 842 (70%) meningioma cases. All countries except Australia had individually
20
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
matched controls, but for this study controls were post-hoc matched to cases on age, sex, region of residence,
and in Israel also country of birth, taking into consideration the timing of the interview. All controls were used
for both glioma and meningioma, resulting in 1–19 controls for glioma and 1–23 controls for meningioma cases.
The control participation rate was 53%. The tumour location was determined from MRI and CT scans by
neuroradiologists who also estimated the tumour centre on a three-dimensional 1 cm x 1 cm x 1 cm grid. Scans
were available for 59% of glioma and 51% of meningioma cases; for the remaining cases radiology reports were
used to construct a tumour volume and a computer algorithm developed to estimate the tumour centre. Controls
were given a fictive tumour location corresponding to that of their matched case. Information on mobile phone
use and potential confounders was collected as described for the main Interphone study, and in addition RF
exposure was assessed through estimates of the total RF dose at the tumour location. RF exposure during mobile
phone use is highly localized, and reaches only a few centimetres into the brain (see Chapter 2), thus, estimating
the RF dose at the tumour location would reduce exposure misclassification considerably. The total cumulative
specific energy (TCSE) at the tumour location was estimated from the participants’ reported mobile phone use
history and was based on network characteristics, frequency bands, communication systems, self-reported
information about calling time, laterality of phone use, hands free devices, type of phones used, and frequency
of use in urban and rural locations (Cardis et al., 2011b). Only the self-reported cumulative calling time and the
tumour location were significant predictors of total RF dose (predicting 43% and 13% of the variability,
respectively). The information needed to calculate TCSE was not available for all participants, and to allow
comparability with the TCSE results, analyses of regular mobile phone use and cumulative call time were made
for three different subgroups of cases and their matched controls; 1) all available cases (808 glioma, 842
meningioma), 2) restricted to cases with information about TCSE with tumour centre estimated by
neuroradiologists or from radiology reports (551 glioma, 675 meningioma), and 3) restricted only to cases with
the tumour centre estimated by neuroradiologists from MRI and CT scans (326 glioma, 343 meningioma).
Cumulative call time and TCSE were categorized into quintiles according to the distribution among regular user
controls. For glioma, the highest quintile of cumulative call time (≥735 h) was associated with an OR of 1.17
(95% CI 0.88–1.56) when all cases were included, while the ORs in the more restricted subgroups were 1.25
(95% 0.88–1.77) and 1.72 (95% CI 1.07–2.77) in subgroup 2 and 3, respectively [indicating that the subgroups
with information about tumour location were not entirely representative of all eligible cases]. Results of
analyses of TCSE showed an OR in the highest quintile of 1.35 (95% CI 0.96–1.90) for glioma in subgroup 2,
and 1.66 (95% CI 1.03–2.67) in subgroup 3. In a sensitivity analysis where information on tumour laterality was
not included in the estimation of TCSE, to reduce potential recall bias, the OR was 1.23 (95% CI 0.89–1.72) in
subgroup 2 (this analysis was not made for subgroup 3). For meningioma, ORs in the highest quintile of
cumulative call time were closer to unity, but were also higher in subgroup 2 and 3 than when all cases were
included. Analyses of TCSE showed an OR in the highest exposure quintile of 0.90 (95% CI 0.66–1.24) in
subgroup 2, and 1.01 (95% CI 0.63–1.62) in subgroup 3. In sensitivity analyses were tumour laterality was not
included in estimation of TCSE the OR was 0.84 (95% CI 0.62–1.15).
954
955
956
957
958
959
960
Analyses of TCSE were also made according to exposure time windows, divided into <3 years, 3–6
years, and ≥7 years before diagnosis, using the same exposure cutpoints as in the overall analysis. This analysis
was not made for cumulative call time. In the time window <3 years most ORs were below unity for both
glioma and meningioma. For glioma, all ORs for the time window ≥7 years before diagnosis were above unity,
with an OR in the highest exposure category of 1.91 (95% CI 1.05–3.47). For meningioma, the OR for the
highest exposure category in the ≥7 year time window was 2.01 (95% CI 1.03–3.93) while ORs in the other
exposure categories were at or below unity.
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
[The assessment of the total RF dose at the tumour location was made to reduce non-differential
exposure misclassification which would have been expected to result in stronger associations compared to
results from using only self-reported call time, should there be a true risk increase associated with RF exposure.
ORs based on TCSE were, however, essentially identical to ORs based solely on self-reported cumulative call
time. Thus, it seems either that the estimated TCSE did not result in less exposure misclassification, or that RF
exposure is not related to tumour risk, which is also noted by the authors. The estimated TCSE is heavily
influenced by self-reported cumulative call time, and other factors seem to provide only minor contributions to
the variability of the TCSE. Thus, recall bias is a concern also in these analyses. The analyses of exposure time
windows are not comparable to analyses previously presented in the overall Interphone study, as they include all
subjects in all three exposure time windows, and in addition use different time categories (Interphone used 1–4,
5–9, ≥10 years). Without a comparable analysis of cumulative call time, it is not possible to assess whether
estimates of TCSE give different results, although it seems likely that an analysis of “ever regular use” for the
time window ≥7 years, disregarding RF dose and restricted to the same subgroup as used in the time window
analysis, would also have resulted in a significant risk increase. It is possible that the risk increase observed in
this analysis is simply caused by the restriction of the analyses to a non-representative subgroup of cases. It is
21
976
977
noteworthy that the quintiles used in the analyses of TCSE do not correspond to 20% controls in each category;
e.g. for glioma it varies between 18% and 24%.]
978
979
980
Data from other national Interphone studies are essentially included in the main Interphone analysis,
and results are well in line with those presented in the main study. These studies are therefore not presented in
detail here.
981
Other brain tumour studies
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
A register-based case-control study was conducted in Finland by Auvinen and co-workers (2002). All
patients diagnosed with brain tumours in 1996 aged 20 to 69 years were ascertained from the national Cancer
Registry and 5 age- and sex-matched controls per case were selected from the national population registry. In
total, 198 glioma and 129 meningioma cases were identified. Four controls were excluded because of a previous
brain tumour diagnosis. Information about start- and end date of mobile phone subscriptions, and type of
network, was obtained from the mobile phone network providers. The average duration of subscription was 2–3
years for analogue phones and less than 1 year for digital. There was no information available about the
frequency or duration of calls or about use of cell phones provided by an employer. Conditional logistic
regression was used for analyses. Adjustment for place of residence, occupation, and socioeconomic status did
not change the results. The OR for brain tumours combined with ever-subscription was 1.3 (95% CI 0.9–1.8);
for glioma 1.5 (95% CI 1.0–2.4), and meningioma 1.1 (95% CI 0.5–2.4). Analogue phone use gave an OR of 2.1
(95% CI 1.3–3.4) for glioma and for digital phone use OR=1.0 (95% CI 0.5–2.0). An increased risk of glioma
was found already after 1-2 years duration of subscription to an analogue phone (OR=2.4; 95% CI 1.2–5.1).
Also for meningioma, the OR for analogue phone use was raised, although confidence intervals were wide
(OR=1.5; 95% CI 0.6–3.5), with the highest OR for <1 year subscription (OR=2.3; 95% CI 0.6–9.2). Acoustic
neuroma was not analysed separately. Exposed and unexposed glioma cases did not differ in terms of location of
the tumour (lobe and laterality) or histologic type. [The strength of the exposure assessment method is that recall
bias is avoided. Considering the low prevalence of mobile phone use in the population, misclassification of the
exposure from lack of information about corporate subscriptions is likely to have a minimal effect on the risk
estimates. Limitations are the lack of information about amount of phone use, and the short duration of mobile
phone use in the population at the time of the study. An increased risk of glioma and meningioma after only up
to 2 years of exposure seems unlikely to be real, and chance is an alternative explanation.]
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
A hospital-based case-control study of cerebral glioma was conducted in six districts in Greece with
case recruitment between June 1, 2005 and May 31, 2007 (Gousias et al., 2009). Cases were ascertained from
the 7 hospitals in the study area, and were in the age range 22–82 years. Two controls per case were recruited
among patients with cervical myelopathy or disk herniation that were treated at the neurosurgical department of
one of the hospitals, matched to cases on sex, age and district of residence. The case-control study included the
first 41 out of a total of 56 cases, and 82 controls. No information was given about non-participation, thus
participation rates cannot be calculated. Exposure information was collected through interviews. Mobile phone
use was measured crudely as “minute-years” where the respondent was asked to multiply the number of minutes
talking on a mobile phone per day with the number of years of use. The exposure was used as a continuous
variable (minute-years) in the logistic regression analysis, which did not take the matching variables into
consideration, but was adjusted for alcohol consumption, smoking, and severe head trauma. No association with
mobile phone use was found (OR=1.00; 95% CI 0.99–1.01). [The study is limited by the hospital-based design
with further limitation by possible non-comparability of populations in the catchment areas between the
hospitals used for recruitment of cases and the hospital used for controls. In addition, exposure assessment was
very crude, and numbers of subjects small.]
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
Spinelli and colleagues conducted a pilot hospital-based case-control study of brain tumours in two
districts in France, were eligible cases were 18 years or older and diagnosed with a primary malignant brain
tumour during 2005 (Spinelli et al., 2010). One control per case, matched on age and sex, was selected among
patients treated at the neurosurgery department for diseases unrelated to cancer at the same hospital as the case.
Exposure information was collected through a preliminary self-administered questionnaire and face-to-face
interviews at the hospitals. Unconditional logistic regression was used for analyses, adjusted for age and sex. In
total 122 (75%) cases and 122 controls (participation rate not given) participated in the study. Cumulative
mobile phone use was estimated through the number of hours of subscription per month times the number of
years of use, and categorized into four categories [the rationale for choice of cutpoints was not given].
Malignant brain tumours were not associated with mobile phone use, the OR in the highest exposure category
≥36 hour-years was 1.07 (95% CI 0.41-2.82). [The exposure definition is unclear; hours of subscription might
22
1030
1031
mean hours included in the contract or actual hours used, which could make a large difference. Lack of
information about control participation rate, and the hospital based study design are other limitations.]
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
Corona and co-workers performed a hospital-based case-control study of acoustic neuroma in two
municipalities in Bahia, Brazil (Corona et al., 2012). The study was conducted between 2006 and 2010, and 85
patients ≥18 years diagnosed between 2000 and 2010 were identified in collaboration with otorhinolaryngology
and neurology clinics. Three control patients per case were selected from the same outpatient clinic as cases.
Controls were the first three patients who had sought care after the case for unilateral hearing loss or tinnitus,
and where a MRI scan excluded a tumour diagnosis, in total 181 controls. The reference date for controls was
the date of their MRI examination. Exposure information was collected through personal interviews or over the
telephone, with 44 cases (52%) and 104 controls (57%). The main reason for non-participation was inability to
locate the person because of outdated registry information, although refusal was more common among controls
than cases. Proxy respondents were used for 4 cases and 2 controls. Analyses were made with unconditional
logistic regression and included as covariates variables with p≤0.20 in the univariate analyses (which variables
are not stated). Regular use (defined as at least once per week over a period of at least 6 months) was associated
with an OR for acoustic neuroma of 1.38 (95% CI 0.61–3.14) compared with non-use/irregular use. For duration
of use ≥6 years the OR was 1.81 (95% CI 0.73–4.47), with the cutoff at the median, with non-users as
comparison. In the highest category of amount of use the OR was 1.15 (95% CI 0.33–4.08), reason for choice of
cutpoints not given. The OR for ipsilateral use was higher than for contralateral, the latter being considerably
below unity. Slightly higher risk estimates were reported for analogue phone use than for digital phones,
although none of the results were statistically significant. [The study is limited by the hospital-based design.
Another limitation is the poor participation rates among both cases and controls, caused by a long delay between
disease occurrence and recruitment, and the small number of subjects included. It is unclear how laterality of
phone use among controls was included in the analyses.]
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
A hospital-based case-control study of acoustic neuroma was conducted by Han and colleagues in
Pittsburg, USA (Han et al., 2012). Cases were patients who underwent Gamma Knife surgery at the University
of Pittsburgh Medical Center between 1997 and 2007. In total, 822 were identified (age-range not given), and
patients with neurofibromastosis Type 2 or living outside the North American continent were excluded from the
study. In total 712 cases were contacted for participation in the study. One control per case, matched on age and
sex, was identified from the outpatient degenerative spinal disorders service at the same medical centre, among
patients evaluated for cervical or lumbar spine disorders. Information about exposure was obtained through a
self-administered questionnaire, received by cases through mail, and answered by 420 patients (59%), either by
mail or over the telephone, whereas controls filled out the questionnaire while waiting for their outpatient
appointment. Information about number of contacted controls is lacking, thus, the participation rate cannot be
calculated for controls. Analyses included complete matched pairs, in total 343 case-control pairs, and were
made with conditional logistic regression, adjusting for race, education, cigarette smoking, alcohol consumption,
occupational exposure to noise, history of hay fever or diabetes, and family history of cancer. Regular mobile
phone use, defined as at least once per week during at least six months, was associated with an OR of 0.95 (95%
CI 0.58–1.58), and the OR for more than 10 years of cell phone use was 1.29 (95% CI 0.69–2.43), based on 92
exposed cases. Corresponding results for cordless phones were close to unity. The study did not investigate
amount of phone use. [The study is limited by the hospital-based design, the low participation rate among cases,
lack of information on control participation, and different procedures for data collection among cases and
controls, which may affect comparability of the information. A strength is the large sample size, with a large
number of long-term users, as well as comprehensive information on potential confounders. All included cases
went through stereotactic radiosurgery, thus, they are unlikely to be patients with early detected small tumours.]
1074
Studies with uncertainties related to inclusion criteria
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
Warren and colleagues conducted a hospital based case-control study in the US primarily focused on
intratemporal facial nerve (IFN) tumours in relation to mobile phone use (Warren et al., 2003). Cases were
identified from the fiscal database at the authors’ academic, tertiary-care medical centre. The study included 51
18 cases of IFN. Controls were selected from the same database matched on age, sex and ethnicity, from three
patient groups: 1. rhinosinusitis, 2. dysphonia or gastroesophageal disease, 3. acoustic neuroma (as an
alternative case group with a tumour location assumed to be less exposed than the IFN tumours). In total 51
cases of acoustic neuroma and 141 other control patients were included. Risk factor information was collected
by telephone interview. Unconditional logistic regression was used for analysis, with no adjustment for potential
confounders. Analyses of acoustic neuroma risk in relation to mobile phone use were also made. For acoustic
neuroma ever use of a mobile phone was associated with an OR of 1.2 (95% CI 0.6-2.2), and use at least once
per week with an OR of 1.0 (95% CI 0.4-2.2). Corresponding results for cordless phone use was 0.8 (95% CI
23
1086
1087
1088
1089
0.4-1.6) and 1.0 (95% CI 0.5-2.0). [Response rates are not stated and it is not clear how many subjects were
excluded for various reasons. Acoustic neuroma patients selected were matched to IFN cases, and it is not clear
if other acoustic neuroma cases would have been eligible if the study had been focused on acoustic neuroma.
The study is given little weight in the overall assessment, and is not included in the table].
Table 12.1.2. Case-control studies of mobile phone use and brain tumours – Time since first use
Country
No. cases/controls
Time period
Source of controls
Exposure
No. exp Relative risk
cases
(95% CI)
Comments
Reference
In total at least 8 h mobile
phone use was required to be
defined as mobile phone user.
Hardell et al.
(1999)
Age range
Sweden
Brain tumours
Mobile phone use
1994-1996
209/425 matched on
sex, age, geographic
region.
Population registry
>1 year
78
0.98 (0.69-1.41)
>5 years
34
0.83 (0.49-1.42)
>10 years
16
1.20 (0.56-2.59)
Malignant tumours
Time since first use
136 cases
>1 year
Meningioma
Time since first use
46 cases
>1 year
Acoustic neuroma
Time since first use
13 cases
>1 year
Sweden
Brain tumours
1997-2000
1429/1470 matched
on age, sex and
region
Mobile phone use,
analogue
20-80
20-80
Population registry
Matched analyses, no additional
adjustment.
53
0.98 (0.63-1.50)
16
1.05 (0.49-2.27)
5
0.78 (0.14-4.20)
>1 year
188
1.3 (1.02-1.6)
>5 years
120
1.4 (1.04-1.8)
>10 years
46
1.8 (1.1-2.9)
>1 year
224
1.0 (0.8-1.2)
>5 years
33
0.9 (0.6-1.5)
>1 year
238
1.0 (0.8-1.2)
>5 years
102
1.3 (0.99-1.8)
Mobile phone use, digital
Cordless phone use
Malignant brain
tumours
Mobile phone use,
analogue
588 cases
>1 year
79
1.13 (0.82-1.57)
>1-6 years
36
1.09 (0.68-1.75)
>6 years
43
1.17 (0.75-1.81)
>1 year
112
1.13 (0.86-1.48)
>1-6 years
100
1.08 (0.81-1.43)
12
1.71 (0.67-4.34)
Mobile phone use, digital
>6 years
Exposure categories for no. of
years overlap.
Definition of “user” did not
require a minimal amount of
use.
Hardell et al.
(2002; 2002)
Exposure categories for no. of
years overlap in most analyses
(except for acoustic neuroma,
where non-overlapping
estimates could be calculated
from the tables).
Matched analyses, no additional
adjustment.
High reported participation
rates, but not calculated
according to conventional
standards for cases.
No rationale given for choice of
6 years as cutpoint for analyses
of malignant tumours.
Cordless phone use
>1 year
Meningioma
611 cases
104
1.13 (0.85-1.50)
>1-6 years
68
0.97 (0.69-1.36)
>6 years
36
1.56 (0.92-2.63)
60
1.1 (0.7-1.5)
78
0.8 (0.6-1.03)
90
0.9 (0.6-1.1)
Mobile phone use,
analogue
>1 year
Mobile phone use, digital
>1 year
Cordless phone use
>1 year
24
Acoustic neuroma
159 cases
Mobile phone use,
analogue
>1 year
38
3.5 (1.8-6.8)
>1-5 year
12
3.0 (1.0-9.3)
>5-10 years
19
3.8 (1.4-10.2)
>10 years
7
3.5 (0.7-16.8)
>1 year
23
1.2 (0.7-2.2)
>1-5 years
21
1.2 (0.6-2.2)
>5 years
2
2.0 (0.2-22.1)
11
1.8 (0.7-5.0)
Mobile phone use, digital
Cordless phone use
>1 year
Sweden
2000-2003
20-80
Malignant brain
tumours
Mobile phone use,
analogue
359/692 matched on
age
>1 year
>1-5 years
0
-
Population registry
>5-10 years
20
1.8 (0.9-3.5)
>10 years
48
3.5 (2.0-6.4)
>1 year
198
1.9 (1.3-2.7)
>1-5 years
100
1.6 (1.1-2.4)
>5-10 years
79
2.2 (1.4-3.4)
>10 years
19
3.6 (1.7-7.5)
68
2.6 (1.5-4.3)
Mobile phone use, digital
Definition of “user” did not
require a minimal amount of
use.
Hardell et al.
(2005a; 2006c)
Unmatched analyses, adjusted
for age, sex, SES, year of
diagnosis.
High reported participation
rates, but not calculated
according to conventional
standards for cases.
Cordless phone use
>1 year
Meningioma
305/692
171
2.1 (1.4-3.0)
>1-5 years
83
1.8 (1.2-2.8)
>5-10 years
58
2.1 (1.3-3.4)
>10 years
30
2.9 (1.6-5.2)
>1 year
35
1.7 (0.97-3.0)
>1-5 years
1
1.2 (0.1-12)
>5-10 years
14
1.4 (0.7-2.8)
>10 years
20
2.1 (1.1-4.3)
Mobile phone use,
analogue
Mobile phone use, digital
>1 year
151
1.3 (0.9-1.9)
>1-5 years
96
1.2 (0.8-1.8)
>5-10 years
47
1.4 (0.9-2.3)
>10 years
8
1.5 (0.6-3.9)
Cordless phone use
>1 year
140
1.3 (0.9-1.9)
>1-5 years
81
1.3 (0.9-1.9)
>5-10 years
40
1.1 (0.7-1.8)
>10 years
19
1.9 (0.97-3.6)
25
Acoustic neuroma
84/692
Mobile phone use,
analogue
>1 year
20
4.2 (1.8-10)
>1-5 years
2
9.9 (1.4-69)
>5-10 years
11
5.1 (1.9-14)
>10 years
7
2.6 (0.9-8.0)
>1 year
53
2.0 (1.05-3.8)
>1-5 years
29
1.7 (0.9-3.5)
>5-10 years
23
2.7 (1.3-5.7)
>10 years
1
0.8 (0.1-6.7)
>1 year
45
1.5 (0.8-2.9)
>1-5 years
31
1.8 (0.9-3.6)
>5-10 years
13
1.4 (0.6-3.2)
>10 years
1
0.3 (0.03-2.2)
Mobile phone use, digital
Cordless phone use
Sweden
2007-2009
18-75
Malignant brain
tumours
Mobile phone use,
analogue
593/1368 matched on >1 year
age, sex, and
>1-5 years
geographic region of
>5-10 years
residence
>10-15 years
144
1.8 (1.04-3.3)
0
-
2
0.6 (0.1-3.1)
25
1.4 (0.7-3.0)
>15-20 years
39
1.4 (0.7-2.7)
>20-25 years
48
2.1 (1.1-4.0)
>25 years
30
3.3 (1.6-6.9)
Mobile phone use, digital
546
1.6 (0.996-2.7)
>1-5 years
42
1.8 (1.01-3.4)
>5-10 years
213
1.6 (0.97-2.7)
>10-15 years
187
1.3 (0.8-2.2)
>15-20 years
104
2.1 (1.2-3.6)
>20-25 years
0
-
>25 years
0
-
>1 year
Definition of “user” did not
Hardell et al.
require a minimal amount of
(2013a; b)
mobile phone use, but in total at
least 39 h of any type of
wireless phone use.
Unmatched analyses, adjusted
for age, sex, SES, year of
diagnosis.
Participation rate among
malignant brain tumour cases
45%, benign 81%, controls
85%.
Results for acoustic neuroma
only reported pooled with
previous material.
Cordless phone use
>1 year
461
1.7 (1.1-2.9)
>1-5 years
102
2.0 (1.1-3.4)
>5-10 years
188
1.6 (0.95-2.7)
>10-15 years
108
1.6 (0.9-2.8)
>15-20 years
57
2.1 (1.2-3.8)
>20-25 years
6
1.5 (0.5-4.6)
>25 years
0
-
26
Meningioma
709/1368
Mobile phone use,
analogue
>1 year
>1-5 years
Carlberg et al.
(2013)
108
0.9 (0.6-1.5)
0
-
>5-10 years
3
0.5 (0.1-2.1)
>10-15 years
21
0.8 (0.4-1.6)
>15-20 years
39
1.1 (0.6-1.9)
>20-25 years
29
0.9 (0.5-1.5)
>25 years
16
1.3 (0.6-2.8)
Mobile phone use, digital
>1 year
593
1.0 (0.7-1.4)
>1-5 years
70
1.1 (0.7-1.7)
>5-10 years
236
0.9 (0.7-1.4)
>10-15 years
212
1.0 (0.7-1.5)
>15-20 years
75
1.0 (0.6-1.5)
>20-25 years
0
-
>25 years
0
-
Cordless phone use
US
1994-1998
18-80
>1 year
522
1.1 (0.8-1.5)
>1-5 years
109
1.0 (0.7-1.5)
>5-10 years
217
1.0 (0.7-1.5)
>10-15 years
128
1.1 (0.8-1.7)
>15-20 years
61
1.2 (0.7-1.8)
>20-25 years
7
1.3 (0.5-3.4)
>25 years
0
-
Malignant brain
tumours
Regular mobile phone use
Never
403
469/422 matched on
age, sex, race and
month of admission
Ever
66
0.8 (0.6-1.1)
Hospital-based
1 year
21
0.7 (0.4-1.3)
2-3 years
28
1.1 (0.6-2.0)
≥4 years
17
0.7 (0.4-1.4)
1.0
No. of years
1997-1999
Acoustic neuroma
No. of years
18+
90/86
0
72
1.0
1-2
7
0.5 (0.2-1.3)
3-6
11
1.7 (0.5-5.1)
Regular defined as having a
mobile phone subscription.
Muscat et al.
(2000)
Unmatched analyses adjusted
for age, education, sex, race,
study centre, occupation, date
of interview, proxy subject.
Muscat et al.
(2002)
27
US
Glioma
Regular mobile phone use
1994-1998
486
No use
18+
798 controls
frequency matched on
age, sex, race,
hospital, proximity of
residence to hospital
Ever
<0.5 years
24
0.6 (0.3-1.1)
Hospital-based
0.5-<3 years
31
0.9 (0.5-1.6)
≥3 years
30
0.9 (0.5-1.5)
285
85
1.0
0.8 (0.6-1.2)
Duration of regular use
Never or rarely
398
1.0
Meningioma
Regular mobile phone use 130
1.0
197
No use
0.8 (0.4-1.3)
32
Regular use defined as ≥2
calls/week.
Inskip et al.
(2001)
Conditional logistic regression
adjusted for matching variables
and date of interview, type of
respondent, education,
household income, type of
health coverage, marital status,
religion, radiotherapy to the
head or neck, handedness.
Ever
Duration of regular use
165
Never or rarely
1.0
6
0.5 (0.2-1.4)
<0.5 years
12
0.8 (0.4-1.9)
0.5-<3 years
14
1.1 (0.5-2.5)
≥3 years
Acoustic neuroma
Regular mobile phone use
96
No use
56
1.0
Ever
22
1.0 (0.5-1.9)
74
1.0
Duration of regular use
Never or rarely
<0.5 years
4
0.3 (0.1-1.3)
0.5-<3 years
8
1.8 (0.7-4.5)
10
1.4 (0.6-3.4)
≥3 years
Interphone
Glioma
Regular mobile phone use
13 countries
2708/2972 matched
on birth year, sex,
geographic region
Never
1042
1.0
Ever
1666
0.81 (0.70-0.94)
1-1.9 years
156
0.62 (0.46-0.81)
2-4 years
644
0.84 (0.70-1.00)
5-9 years
614
0.81 (0.60-0.97)
≥10 years
252
0.98 (0.76-1.26)
2000-2004
30-59
Population-based
Time since first regular
use
Meningioma
Regular mobile phone use
2409/2662 matched
on birth year, sex,
geographic region
Never
1147
1.0
Ever
1262
0.79 (0.68-0.91)
1-1.9
178
0.90 (0.68-1.18)
2-4 years
557
0.77 (0.65-0.92)
5-9 years
417
0.76 (0.63-0.93)
≥10 years
110
0.83 (0.61-1.14)
Population-based
Interphone
study group
(2010)
Conditional logistic regression
analyses adjusted for education.
Other variables did not change
risk estimates (sociodemographic factors,
occupational EMF exposure and
ionizing radiation, medical
history, medical ionizing and
non-ionizing radiation exposure
and smoking).
No evidence of heterogeneity in
results between study centres
was found.
Time since first regular
use
Acoustic neuroma
Regular mobile phone use
1105/2145 matched
on birth year, sex,
geographic region
Never
462
1.0
Ever
643
0.85 (0.69-1.04)
1-1.9
63
0.73 (0.49-1.09)
2-4 years
276
0.87 (0.69-1.10)
5-9 years
236
0.90 (0.69-1.16)
≥10 years
68
0.76 (0.52-1.11)
Population-based
Regular use defined as at least
one call per week over a period
of at least six months.
Interphone
study group
(2011)
Time since first regular
use
28
Nordic-UK
Interphone
centres
2000-2004
18-69
Glioma
Regular mobile phone use
1496/3134 frequency
matched on birth
year, sex, geographic
region
Never
629
1.0
Ever
867
0.78 (0.68-0.91)
1.5-4 years
384
0.77 (0.65-0.92)
5-9 years
342
0.75 (0.62-0.90)
≥10 years
143
0.95 (0.74-1.23)
Population-based
Time since first regular
use
Meningioma
Regular mobile phone use
1204/2945 frequency
matched on birth
year, sex, geographic
region
Never
631
1.0
Ever
573
0.76 (0.65-0.89)
1.5-4 years
286
0.72 (0.60-0.86)
5-9 years
214
0.78 (0.64-0.96)
≥10 years
73
0.91 (0.67-1.25)
Population-based
Time since first regular
use
Acoustic neuroma
Regular mobile phone use
676/3546 frequency
matched on birth
year, sex, geographic
region
Never
316
1.0
Ever
360
0.9 (0.7-1.1)
1.5-4 years
174
0.8 (0.7-1.0)
5-9 years
139
0.9 (0.7-1.2)
≥10 years
47
1.0 (0.7-1.5)
Population-based
Glioma
Subscription
1996
198/989 matched on
age and sex
Ever
Register-based,
subscriptions.
Meningioma
129/643
Register-based,
subscriptions.
Greece
Glioma
2005-2007
41/82 matched on
age, sex, district of
residence
Lahkola et al.
(2007)
Includes a wider age-range than
Interphone core protocol,
frequency matching results in a
larger number of controls.
Glioma and meningioma
analyses includes Nordic
countries and UK South,
acoustic neuroma analyses
includes also UK North.
Conditional regression analysis, Lahkola et al.
(2007; 2008)
conditioning on frequency
matching variables. Adjustment
for educational level, family
history of glioma, radiation
therapy to the head and neck,
neurofibromatosis or tuberous
sclerosis did not change the
results.
No evidence of heterogeneity in
results between study centres
was found.
Schoemaker et
al. (2005)
Time since first regular
use
Finland
20-69
Regular use defined as at least
one call per week over a period
of at least six months.
?
1.5 (1.0-2.4)
<1
?
1.2 (0.5-3.0)
1-2
?
1.6 (0.8-2.9)
>2
?
1.7 (0.9-3.5)
?
1.1 (0.5-2.4)
<1
?
1.5 (0.5-4.6)
1-2
?
1.2 (0.4-3.6)
>2
?
0.8 (0.2-3.5)
No. or years
Subscription
Ever
No. of cases and controls is
only given for analogue and
digital phones separately, may
overlap.
Auvinen et al.
(2002)
Conditional logistic regression
analysis. Adjustment for place
of residence, occupation,
socioeconomic status did not
change the results.
No. or years
Continuous variable
Minute-years
1.00 (0.99-1.01)
Hospital-based
Brazil
Acoustic neuroma
Regular mobile phone use
2006-2010
44/104
Never
10
1.0
18+
Hospital-based
Ever
34
1.38 (0.61-3.14)
Duration of use
No use
9
1.0
<6 years
12
1.14 (0.42-3.08)
≥6 years
23
1.81 (0.73-4.47)
Logistic regression analysis, did Gousias (2009)
not take the matching variables
into consideration, adjusted for
alcohol consumption, smoking,
severe head trauma.
Regular use defined as at least
one call per week over a period
of at least six months.
Corona (2012)
Unconditional logistic regression
analysis, covariates were
variables with p≤0.20 in the
univariate analyses (not stated
which variables).
29
Pittsburg, USA Acoustic neuroma
Regular mobile phone use
1997-2007
343/343 matched on
age and sex
Never
140
1.0
Ever
203
0.95 (0.58-1.58)
Hospital based
Years of mobile phone
use
<10 years
111
0.79 (0.45-1.37)
≥10 years
92
1.29 (0.69-2.43)
Never
76
1.0
Ever
267
0.93 (0.53-1.63)
<10 years
197
0.91 (0.52-1.60)
≥10 years
70
1.07 (0.51-2.24)
age-range not
given
Cordless phone use
Regular use defined as at least
one call per week over a period
of at least six months.
Han et al.
(2012)
Low participation rate among
cases, no information given for
controls.
Conditional logistic regression
analysis adjusted for race,
education, residency, smoking,
alcohol, occupational exposure
to noise, hay fever, diabetes,
family history of cancer.
Years of cordless phone
use
1090
Table 12.1.3. Case-control studies of mobile phone use and brain tumours – Amount of phone use
Country
Diagnosis
Exposure
Sweden
Brain tumours
Mobile phone use
1994-1996
<136 h
38
1.01 (0.64-1.58)
20-80
Not reported for
specific types of
brain tumours
>136 h
40
0.96 (0.61-1.52)
Sweden
Brain tumours
Mobile phone use, analogue
Time period
No. exp Relative risk
cases
(95% CI)
Comments
Reference
In total at least 8 h mobile
phone use was required to be
defined as mobile phone user.
Hardell et al.
(1999)
Amount of use not reported for
specific types of brain tumours.
Hardell et al.
(2002)
Age range
1997-2000
≤85 h
114
1.3 (0.99-1.8)
20-80
>85 h
96
1.2 (0.9-1.6)
≤55 h
165
1.0 (0.8-1.3)
>55 h
130
0.9 (0.7-1.2)
≤183 h
136
0.9 (0.7-1.1)
>183 h
161
1.1 (0.9-1.4)
≤80 h
17
1.4 (0.7-2.8)
>80 h
51
4.0 (2.2-7.3)
≤64 h
79
1.6 (1.1-2.4)
>64 h
119
2.4 (1.6-3.7)
≤243 h
81
1.8 (1.2-2.8)
>243 h
90
2.4 ( 1.5-3.6)
≤80 h
15
1.4 (0.7-2.7)
>80 h
20
2.2 (1.1-4.3)
≤64 h
77
1.1 (0.8-1.7)
>64 h
74
1.7 (1.1-2.6)
≤243 h
59
1.0 (0.7-1.6)
>243 h
81
1.7 (1.1-2.6)
≤80 h
7
3.0 (1.1-8.6)
>80 h
13
6.0 (2.2-17)
Mobile phone use, digital
Definition of “user” did not
require a minimal amount of
use.
Cordless phone use
Sweden
2000-2003
Malignant brain
tumours
20-80
Mobile phone use, analogue
Definition of “user” did not
require a minimal amount of
use.
Hardell et al.
(2005a; 2006c)
Mobile phone use, digital
Cordless phone use
Meningioma
Mobile phone use, analogue
Mobile phone use, digital
Cordless phone use
Acoustic neuroma
Mobile phone use, analogue
Mobile phone use, digital
30
≤64 h
23
1.7 (0.8-3.4)
>64 h
30
2.5 (1.2-5.2)
≤243 h
21
1.4 (0.7-2.9)
>243 h
24
1.7 (0.8-3.5)
>39-405 h
90
1.7 (0.9-3.0)
406-1091 h
22
1.6 (0.8-3.4)
1092-2376 h
18
2.6 (1.2-6.0)
>2376 h
14
7.7 (2.5-24)
>39-405 h
202
1.4 (0.8-2.3)
406-1091 h
138
1.9 (1.1-3.3)
1092-2376 h
84
1.4 (0.8-2.5)
122
3.2 (1.8-5.6)
>39-405 h
164
1.3 (0.8-2.2)
406-1091 h
120
1.7 (1.01-3.0)
1092-2376 h
98
2.1 (1.2-3.7)
>2376 h
79
3.1 (1.8-5.5)
>0-≤8.7
17
1.0 (0.5-2.0)
>8.7-<60
12
0.6 (0.3-1.3)
>60-≤480
19
0.9 (0.5-1.8)
>480
14
0.7 (0.3-1.4)
1-60
9
0.9 (0.3-3.1)
>60
9
0.7 (0.2-2.6)
Cordless phone use
Sweden
2007-2009
Malignant brain
tumours
18-75
Mobile phone use, analogue
Mobile phone use, digital
>2376 h
Definition of “user” did not
Hardell et
require a minimal amount of
al.(2013a)
mobile phone use, but in total at
least 39 h of any type of
wireless phone use.
Categories according to
quartiles of the distribution of
any type of wireless phone use
among controls.
Cordless phone use
US
1994-1998
Malignant brain
tumours
18-80
1997-1999
Acoustic neuroma
18+
US
Glioma
Cumulative hours of use
Cumulative hours of use
<13
26
0.8 (0.4-1.4)
18+
13-100
26
0.7 (0.4-1.3)
>100
32
0.9 (0.5-1.6)
Interphone
Glioma
Muscat et al.
(2002)
Regular use defined as ≥2
calls/week.
Inskip et al.
(2001)
“Regular use” defined as at
least one call per week over a
period of at least six months.
Interphone
study group,
(2010)
Cumulative hours of use
<13
Acoustic neuroma
Muscat et al.
(2000)
Cumulative hours of use
1994-1998
Meningioma
“Regular" defined as having a
mobile phone subscription.
8
0.7 (0.3-1.9)
13-100
13
1.1 (0.5-2.4)
>100
11
0.7 (0.3–1.7)
<13
5
0.7 (0.2-2.3)
13-100
8
1.2 (0.5-3.1)
>100
9
1.4 (0.6-3.5)
Cumulative hours of use
Cumulative hours of use
13 countries
<5h
141
0.70 (0.52-0.94)
2000-2004
5-12.9
145
0.71 (0.53-0.94)
30-59
13-30.9
189
1.05 (0.79-1.38)
31-60.9
144
0.74 (0.55-0.98)
61-114.9
171
0.81 (0.61-1.08)
115-199.9
160
0.73 (0.54-0.98)
200-359.9
158
0.76 (0.57-1.01)
360-734.9
189
0.82 (0.62-1.08)
735-1639.9
159
0.71 (0.53-0.96)
>1640
210
1.40 (1.03-1.89)
<1.5 *100 calls
147
0.74 (0.55–0.99)
1.5-3.4
141
0.71 (0.54–0.95)
3.5-7.4
161
0.76 (0.58–1.00)
Cumulative number of calls
31
Meningioma
7.5-13.9
174
0.90 (0.68–1.20)
14-25.4
180
0.78 (0.59–1.02)
25.5-41.4
156
0.83 (0.62–1.10)
41.5-67.9
163
0.71 (0.53–0.94)
68-127.9
186
0.93 (0.70–1.23)
128-269.9
193
0.96 (0.72–1.28)
>270
165
0.96 (0.71–1.31)
<5h
160
0.90 (0.69-1.18)
5-12.9
142
0.82 (0.61-1.10)
13-30.9
144
0.69 (0.52-0.91)
31-60.9
122
0.69 (0.51-0.94)
61-114.9
129
0.75 (0.55-1.00)
115-199.9
96
0.69 (0.50-0.96)
200-359.9
108
0.71 (0.51-0.98)
360-734.9
123
0.90 (0.66-1.23)
735-1639.9
108
0.76 (0.54-1.08)
>1640
130
1.15 (0.81-1.62)
<1.5 *100 calls
159
0.95 (0.72-1.27)
1.5-3.4
136
0.62 (0.46-0.83)
3.5-7.4
148
0.90 (0.68-1.19)
7.5-13.9
143
0.80 (0.61-1.07)
14-25.4
122
0.60 (0.45-0.81)
25.5-41.4
111
0.81 (0.58-1.13)
41.5-67.9
129
0.79 (0.58-1.09)
68-127.9
134
0.92 (0.67-1.26)
128-269.9
100
0.81 (0.57-1.16)
80
0.80 (0.55-1.17)
Cumulative hours of use
Cumulative number of calls
>270
Acoustic neuroma
Cumulative hours of use
<5h
58
0.77 (0.52-1.15)
5-12.9
63
0.80 (0.54-1.18)
13-30.9
80
1.04 (0.71-1.52)
31-60.9
66
0.95 (0.63-1.42)
61-114.9
74
0.96 (0.66-1.41)
115-199.9
68
0.96 (0.65-1.42)
200-359.9
50
0.60 (0.39-0.91)
360-734.9
58
0.72 (0.48-1.09)
735-1639.9
49
0.48 (0.30-0.78)
>1640
77
1.32 (0.88-1.97)
<1.5 *100 calls
59
0.76 (0.51-1.14)
1.5-3.4
60
0.68 (0.45-1.03)
3.5-7.4
73
1.11 (0.76-1.61)
7.5-13.9
87
1.22 (0.84-1.77)
14-25.4
79
1.11 (0.75-1.64)
25.5-41.4
55
0.64 (0.42-0.98)
41.5-67.9
50
0.74 (0.49-1.12)
68-127.9
62
0.65 (0.43-0.98)
128-269.9
56
0.67 (0.44-1.02)
>270
62
0.93 (0.611.41)
<125 h
368
0.75 (0.64-0.89)
125-503 h
193
0.69 (0.55-0.85)
Shown here are analyses using Interphone
1 year latency for comparability. study group
Paper presents also secondary (2011)
analyses using 5 year latency.
Cumulative number of calls
Nordic-UK
Interphone
centres
2000-2004
Glioma
Cumulative hours of use
Lahkola et al.
(2007)
32
18-69
>503 h
262
0.90 (0.73-1.10)
<2172
352
0.73 (0.62-0.87)
2172-7792
205
0.74 (0.60-0.91)
>7792
265
0.91 (0.74-1.12)
<125 h
278
0.68 (0.57-0.82)
125-514 h
125
0.79 (0.62-1.02)
>514 h
140
0.88 (0.68-1.13)
<2195
285
0.68 (0.57-0.82)
2195-7790
130
0.86 (0.67-1.10)
>7790
128
0.83 (0.64-1.07)
Cumulative number of calls
Meningioma
Cumulative hours of use
Lahkola et al.
(2008)
Cumulative number of calls
Acoustic neuroma
Cumulative hours of use
<116h
168
0.9 (0.7-1.1)
116-534 h
89
0.9 (0.7-1.2)
>534
94
0.9 (0.7-1.2)
Schoemaker et
al. (2005)
Cumulative number of calls
<2149
5 Interphone
study centres:
Australia,
Canada,
France, Israel,
New Zealand
Glioma
173
0.9 (0.7-1.1)
2149-8000
82
0.8 (0.6-1.1)
>8000
99
1.0 (0.7-1.3)
Never
266
1.0
Ever
542
0.92 (0.75-1.13)
Regular mobile phone use
Cumulative call time
<13 h
69
0.88 (0.63-1.24)
2000-2004
13-<61
103
0.93 (0.69-1.25)
30-59
61-<200
110
0.74 (0.55-0.99)
200-<735
123
0.94 (0.71-1.26)
≥735
137
1.17 (0.88-1.56)
Never
196
1.0
Ever
355
0.93 (0.73-1.18)
<13 h
44
0.83 (0.55-1.26)
13-<61
68
0.93 (0.65-1.32)
61-<200
63
0.66 (0.46-0.96)
200-<735
90
1.07 (0.76-1.50)
≥735
90
1.25 (0.88-1.77)
<76.7
67
0.76 (0.53-1.09)
76.7-
68
0.94 (0.66-1.35)
284.1-
60
0.80 (0.54-1.18)
978.9-
57
0.89 (0.61-1.30)
103
1.35 (0.96-1.90)
Never
356
1.0
Ever
486
0.80 (0.66-0.96)
<13 h
102
0.86 (0.64-1.15)
13-<61
101
0.79 (0.60-1.04)
61-<200
97
0.71 (0.53-0.94)
200-<735
89
0.69 (0.51-0.93)
Glioma, subset with
information on total
cumulative specific
energy (TCSE)
Regular mobile phone use
Cumulative call time
TCSE (joules/kg)
≥3123.9
Meningioma
Unconditional logistic regression Cardis et al.
analyses, stratified on age, sex (2011a)
and region, and adjusted for
education and the interaction
between study region and time
period of interview (6-month
intervals).
For 59% of glioma cases the
tumour location was determined
by neuroradiologists from MRI
and CT scans, for 41% it was
estimated from radiological
reports.
Total cumulative specific energy
(TCSE) at tumour location
estimated from self- reported
mobile phone use history,
based on network
characteristics, frequency
bands, communication systems,
calling time, laterality of phone
use, handsfree devices, type of
phones used, and frequency of
use in urban and rural locations.
TCSE was only available for a
subset of individuals, for
comparability, results for
cumulative call time is also
given for this subset.
For TCSE calculated without
consideration of laterality of
phone use the OR in the top
category was 1.23 (0.89-1.72)
for glioma.
Regular mobile phone use
Cumulative call time
33
≥735
Meningioma, subset
with information on
total cumulative
specific energy
(TCSE)
97
1.01 (0.75-1.36)
Regular mobile phone use
Never
294
1.0
Ever
381
0.77 (0.63-0.95)
<13 h
80
0.84 (0.60-1.16)
13-<61
82
0.80 (0.59-1.10)
61-<200
73
0.64 (0.46-0.89)
200-<735
67
0.63 (0.45-0.88)
≥735
79
1.06 (0.75-1.48)
<76.7
103
0.90 (0.67-1.21)
76.7-
71
0.74 (0.53-1.04)
284.1-
56
0.56 (0.39-0.80)
978.9-
62
0.72 (0.51-1.02)
≥3123.9
88
0.90 (0.66-1.24)
Cumulative call time
For 51% of meningioma cases
the tumour location was
determined by neuroradiologists
from MRI and CT scans, for
41% it was estimated from
radiological reports.
TCSE (joules/kg)
1091
1092
Case-case studies
1093
1094
1095
1096
1097
Radiofrequency exposure during mobile phone use is highly localized, and declines rapidly with
distance to the exposure source. The energy absorption reaches only a few centimetres into the brain (see
Chapter 2). This means that tumours in mobile phone users, if caused by the radiofrequency exposure, would be
expected to more often be located closer to the exposure source than tumours in patients who are not mobile
phone users. Three case-case studies have been conducted to test this hypothesis.
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
A case-case study of mobile phone use and the risk of acoustic neuroma was conducted in Japan
(Sato et al., 2011). In total, 1589 cases were identified between 2000 and 2006, from 22 consenting hospitals in
Japan out of 68 contacted. Exposure information was collected through a self-administered postal questionnaire,
answered by 800 cases (51%), of whom 787 had a unilateral tumour with known tumour location. Analyses
were made with reference dates at 1 and 5 years before diagnosis, and were restricted to regular mobile phone
users (at least once per week during >6 months) who used the mobile phone on one preferred side the head, and
had no symptoms of the tumour at the reference date. In analyses with one year latency, 180 cases were
included, and with five years 150 cases. Case-case analyses were made according to a method described by
Inskip and colleagues, where mobile phone use laterality was related to tumour laterality (Inskip et al., 2001).
The overall risk ratio for regular mobile phone use with one year latency was 1.08 (95 % CI 0.93–1.28) and with
5 years latency 1.14 (95% CI 0.96–1.40). Average call duration >20 minutes/day (heavy use) was associated
with a risk ratio of 2.74 (95 % CI 1.18–7.85) with one year latency and 3.08 (95 % CI 1.47–7.41) with five
years. Risk ratios for 10–20 minutes/day was 0.82 (95% CI 0.65–1.19) with one year latency and 0.84 (95% CI
0.62–1.44) with five years. Among heavy users, the tumour diameter tended to be smaller in cases with
ipsilateral use than contralateral use, a pattern that was not seen among patients with smaller amounts of use. [A
case-case design prevents selection bias caused by non-participation among controls, but cannot prevent
selection bias caused by earlier detection of the tumour. As noted by the authors, hearing loss is a common early
symptom of acoustic neuroma and may lead to an earlier detection of tumours among heavy mobile phone users.
This may explain some of the risk increase observed among heavy users, an assumption supported by the
finding of smaller tumours in this group. In addition, recall bias will not be prevented by this type of case-case
design, as the self-reported side of mobile phone use was used in the analyses, but may be less severe than in a
case-control study. The study found some indication of recall bias in the reported side of mobile phone use, and
it is noteworthy that risk estimates of intermediate categories of amount of use were slightly reduced.]
1121
1122
1123
1124
1125
1126
1127
1128
1129
Larjavaara and co-workers conducted a case-case study of glioma based on data from seven
Interphone study centres (Denmark, Finland, Norway, Sweden, UK-South, Italy, and Germany) (Larjavaara et
al., 2011b). In total 873 glioma cases were included. For each patient, neuroradiologists determined the
localization of the tumour mid-point on a three-dimensional 1 cm x 1 cm x 1 cm grid, by scrutinizing
radiological images, blind to cases’ exposure status. Exposure was defined as the shortest distance between the
mid-point of the tumour and the typical location of a mobile phone during a phone call. To avoid recall bias,
self-reported laterality of phone use was not taken into consideration when calculating the distance. Case-case
analyses were made with unconditional logistic regression, using distance between the midpoint of the tumour
and the exposure source as a binary outcome: ≤5 cm and > 5 cm. Analyses were adjusted for age, sex, education
34
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
and country. Exposure was estimated as regular use, cumulative call time (categorized according to tertiles),
laterality and duration of use in years (categorized as in previous studies based on the Nordic-UK Interphone
data). The reference category in all analyses was never regular use. In addition, analyses were performed using a
case-specular method, where the actual location of the tumour (case) was compared to a hypothetical location
(specular control). The hypothetical control location was determined by symmetrically reflecting the location of
the actual tumour site across the midpoint of the axial and coronal planes to obtain a mirror-image location, and
analyses were made with logistic regression conditioned upon matched sets (case and specular). The results
showed that tumours were equally distributed between the right and the left side of the head. No substantial
differences in the mean distance between exposure source and tumour location were observed between different
categories of mobile phone use. The distance was somewhat shorter for non-regular users and for contralateral
users, and longer for the highest category of cumulative hours of use and longest duration of use, although none
of the differences were statistically significant. The ORs for having a tumour located within five centimetres of
the exposure source were below unity for all categories of mobile phone use when compared to non-regular
users. For regular use the OR was 0.80 (95% CI 0.56–1.15), for ≥10 years duration it was 0.85 (95% CI 0.39–
1.86), and for the highest tertile of cumulative hours of use 0.58 (95% CI 0.35–0.96). In the case-specular
analysis most ORs for having a tumour within 5 centimetres of the exposure source were above unity, both
among regular users (OR= 1.19; 95% CI 0.89–1.59) and non-regular users (OR= 1.30; 95% CI 0.95–1.80). [This
case-case analysis minimizes recall bias, because the self-reported side of mobile phone use was not taken into
consideration when calculating the distance from the tumour location to the exposure source. Selection bias
caused by non-participation among controls is eliminated in this analysis, and for glioma it is unlikely that
mobile phone use will affect detection of the tumour. Defining exposed locations as within 5 centimetres from a
hypothetical exposure source may result in some exposure misclassification. The case-specular analysis is
complicated by the non-symmetrical distribution of glioma within the brain, which may explain ORs above
unity for both mobile phone users and for non-users.]
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
In the pooled analysis of case-control data on glioma and meningioma from 5 Interphone countries
made by Cardis and co-workers a case-case analysis was also conducted (Cardis et al., 2011a). In total, 556
glioma and 672 meningioma cases were included. The tumour location was determined from MRI and CT scans
by neuroradiologists who also estimated the tumour centre on a three-dimensional 1 cm x 1 cm x 1 cm grid.
Scans were available for 59% of glioma and 51% of meningioma cases; for the remaining cases radiology
reports were used to construct a tumour volume and a computer algorithm developed to estimate the tumour
centre. The outcome was defined as having a tumour located in the most exposed area of the head (the area
receiving 50% of the absorbed energy), not considering self-reported side of phone use. Controls were cases
with tumour locations outside this area. Mobile phone exposure was estimated as regular use, time since start of
use (categorized as 1–4, 5-9, ≥10 years), and cumulative call time (categorized according to quintiles among
cases with tumours located outside the most exposed area). The reference category in all analyses was never
regular use. Regular mobile phone use was associated with an OR of 1.35 (95% CI 0.64–2.87) for glioma and
0.74 (95% CI 0.49–1.11) for meningioma. A raised OR was observed for glioma in the highest category of time
since first use (OR= 2.80; 95% CI 1.13–6.94), but in the category 5–9 years the OR was considerably reduced
(OR=0.72; 95% CI 0.27–1.90). In the highest category of cumulative call time the OR was close to unity. For
meningioma, the ORs in the highest exposure categories of both time since first use and cumulative call time
were above unity but with wide confidence intervals. [The analyses were based on small numbers of exposed
subjects, and no consistent dose response patterns were observed. For a large proportion of cases the estimated
tumour centre was uncertain, which may have resulted in some exposure misclassification. A smaller proportion
of glioma cases were considered exposed in this study compared to the study by Larjavaara et al. discussed
above (8% compared to 23%), which may have reduced the statistical power and introduced exposure
misclassification considering that 50% of the energy was absorbed at locations that were considered unexposed.]
Table 12.1.4. Case-case studies of mobile phone use and brain tumours
Country
Time period
No. cases
Exposure
No. exp Relative risk
cases
(95% CI)
Comments
Reference
Age range
35
Japan
Acoustic neuroma
One year latency
2000-2006
787 cases
Regular mobile phone use
All ages
Hospital-based
Years since start
180
1.08 (0.93-1.28)
≤5
112
1.06 (0.88-1.31)
5-9
56
1.05 (0.82-1.45)
>10
12
1.62 (0.79-4.77)
≤1
64
1.24 (0.94-1.76)
1-3
69
0.96 (0.79-1.24)
3-5
22
1.21 (0.78-2.26)
>5
25
1.11 (0.77-1.87)
≤3 min
79
1.18 (0.93-1.57)
3-10 min
50
0.89 (0.72-1.21)
10-20 min
28
0.82 (0.65-1.19)
>20 min
23
2.74 (1.18-7.85)
1.0
Sato et al.
(2011)
Number of calls per day
Average daily call duration
7 Interphone
study centres:
Denmark,
Finland,
Germany,
Italy, Norway,
Sweden, UK
South
Glioma
Regular mobile phone use
873 cases
Never
91
Population based
Ever
107
Cumulative call time
0.001-46 h
33
47-339 h
38
2000-2004
>339 h
30
18-69
Duration of use
0.80 (0.56-1.15)
Tumour location determined by
neuroradiologists from MRI and
CT scans.
Larjavaara, et
al. (2011b)
Odds ratio of a tumour location
within 5 cm of the exposure
0.82 (0.51-1.31) source, not considering self0.97 (0.60-1.56) reported side of phone use.
0.58 (0.35-0.96) Unconditional logistic regression
adjusted for age, sex,
education, and country.
0.85 (0.57-1.25) 23% of cases located within 5
1.5-4 years
65
5-9 years
30
≥10 years
10
0.71 (0.43-1.18) cm of exposure source.
0.85 (0.39-1.86)
Ipsilateral
51
0.80 (0.52-1.22)
Contralateral
37
0.77 (0.47-1.24)
Laterality
5 Interphone
study centres:
Australia,
Canada,
France, Israel,
New Zealand
Glioma
Regular mobile phone use
556 cases
Never
14
Ever
30
Time since first regular use
1-4 years
12
2000-2004
5-9 years
7
30-59
≥10 years
11
Cumulative call time
<39 h
6
39-<220
4
220-<520
5
520-<1147
10
≥1147
5
For 59% of glioma cases the
Cardis et al.
tumour
location
was
determined
(2011a)
1.0
by neuroradiologists from MRI
1.35 (0.64-2.87) and CT scans, for 41% it was
estimated from radiological
1.37 (0.59-3.19) reports.
0.72 (0.27-1.90) Odds ratio of a tumour location
in the area receiving 50% of the
2.80 (1.13-6.94) absorbed energy, not
considering self-reported side of
1.19 (0.40-3.51) phone use.
0.93 (0.27-3.14) Unconditional logistic regression
analyses, stratified on age, sex
1.38 (0.42-4.53) and region and adjusted for
2.55 (0.94-6.91) education and timing of
0.99 (0.30-3.27) interview.
8% of glioma and 20% of
meningioma were located in the
most exposed area.
36
Meningioma
Regular mobile phone use
672 cases
Never
66
1-4 years
38
For 51% of meningioma cases
the tumour location was
determined by neuroradiologists
0.74 (0.49-1.11) from MRI and CT scans, for
41% it was estimated from
0.67 (0.41-1.07) radiological reports.
Ever
69
5-9 years
22
0.75 (0.42-1.34)
≥10 years
9
1.34 (0.55-3.25)
<39 h
16
0.55 (0.29-1.02)
39-<220
21
0.93 (0.51-1.68)
220-<520
9
0.52 (0.23-1.14)
520-<1147
10
0.67 (0.30-1.48)
≥1147
13
1.41 (0.66-3.02)
Time since first regular use
1.0
Cumulative call time
1176
1177
Brain tumours in children
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
Currently, there is only one study available on mobile phone use and cancer risk in children. Aydin and
colleagues conducted a population-based case-control study of brain tumours among children aged 7–19 years
in Denmark, Norway, Sweden and Switzerland (Aydin et al., 2011). Eligible cases were diagnosed with a
primary brain tumour during 2004–2008. Cases were ascertained through pediatric, oncology, and neurosurgery
departments and/or from population-based cancer registries. All diagnoses were either histologically confirmed
or based on unequivocal diagnostic imaging. Two controls per case were randomly selected from population
registries, matched on age, sex, and geographic region. In total, 352 (83%) cases and 646 (71%) controls
participated in the study. Information about mobile phone use history and potential confounding factors were
collected through personal interviews with the child and at least one parent using a computer assisted
questionnaire. Regular use of a mobile phone was defined as at least one call per week over period of at least six
months. Exposure within six months of the diagnosis (or corresponding date for the matched controls) was not
taken into consideration. Analyses were made with conditional logistic regression, and apart from the matching
variables evaluation of confounding from a large number of potential risk factors were made, but as this did not
appreciably change the results they were not kept in the final model. For about one third of the participants
exposure information was available from mobile phone operators. Cumulative hours of phone use and number
of calls were categorized according to the median and 75 th percentile of the distribution among the controls.
For self-reported regular mobile phone use an OR of 1.36 (95% CI 0.92–2.02) was found. Risk estimates
did not increase with increasing time since first use; the OR for >5 years since first use was 1.26 (95% CI 0.70–
2.28), and were not higher for ipsilateral mobile phone use than contralateral, or for tumours in the temporal
lobe compared to other locations. The ORs increased with increasing cumulative hours of use, but none of the
effect estimates were statistically significant; in the upper quartile the OR was 1.55 (95% CI 0.86–2.82). No
consistent trend was seen for cumulative number of calls; the OR in the highest quartile was 1.42 (95% CI 0.79–
2.53). In the small subset with operator traffic data, an increased OR was observed in the upper quartile of time
since first subscription (OR=2.15; 95% CI 1.07–4.29). No consistent trend was observed with operator-recorded
cumulative duration of calls, the OR in the upper quartile was 1.38 (95% CI 0.53–3.61). A projected incidence
trend was calculated based on the assumption of an OR of 2.15 after three years of regular mobile phone use and
was found to be incompatible with the observed incidence trend among Swedish children 7–19 years old
between 1990 and 2008. [The strengths of the study are rapid case ascertainment, population-based controls, and
data collection through structured computer-assisted personal interviews with well-trained interviewers.
Limitations are the use of retrospectively collected self-reported information about mobile phone use and
limited statistical power. The findings of non-significant effect estimates slightly above unity in most analyses
do not provide evidence against random variation as the explanation because the analyses are not independent;
all analyses include the same participants. If random variation affects the distribution between unexposed and
exposed, this will affect all analyses of specific indices of amount or duration of phone use. Use of operatorrecorded traffic data is a strength, and is likely to reduce misclassification of exposure, but this information was
only available for a small non-random subset of the participants, which prevents causal interpretations,
especially in the light of incompatible incidence trends.]
37
1216
Table 12.1.5. Case-control studies of mobile phone use in children and brain tumours
Country
No. cases/controls
Time period
Source of controls
Exposure
No. exp Relative risk
cases
(95% CI)
Comments
Reference
Regular use defined as at least
one call per week over a period
of at least six months.
Aydin et al.
(2011)
Age range
Denmark,
Norway,
Sweden,
Switzerland
2004-2008
Brain tumours
Regular mobile phone use
352/646 matched on
age, sex, geographic
region.
Population registry
Never
158
1.0
Ever
194
1.36 (0.92-2.02)
≤3.3 years
95
1.35 (0.89-2.04)
3.3-5.0 years
53
1.47 (0.87-2.49)
>5.0 years
46
1.26 (0.70-2.28)
≤35
94
1.33 (0.89-2.01)
36-144
48
1.44 (0.85-2.44)
>144
49
1.55 (0.86-2.82)
≤936
94
1.34 (0.89-2.02)
937-2638
50
1.47 (0.86-2.51)
>2638
47
1.42 (0.79-2.53)
7-19
Time since first use
Cumulative hours of use
Conditional logistic regression
analyses. No additional
adjustment in final model, a
large set of confounders
assessed.
Additional analyses were made
of a small subset with operator
recorded exposure information,
but was not randomly selected.
Cumulative no. calls
1217
1218
12.1.2.2
1219
1220
In total, six studies of salivary gland tumour risk in relation to mobile phone use have been
conducted, of which one was a cohort study and five case-control studies.
1221
1222
1223
1224
1225
1226
1227
1228
A Danish cohort study (Johansen et al., 2001) included 420 095 mobile phone subscribers. The cohort
includes only private subscribers; the approximately 200 000 corporate subscriptions could not be linked to a
person. A detailed description of the study design is given in the section about brain tumours (Page x). Briefly,
the cohort was followed from first subscription through 1996. Average follow-up time was 3.5 years. Cancer
incidence in the cohort was ascertained by linkage to the Danish Cancer Registry. Standardised incidence ratios
(SIR) were calculated comparing cancer incidence in mobile phone subscribers with national rates allowing for
sex, age and calendar period. In total, 7 cases of salivary gland tumours were identified in the subscriber cohort
during the study period, all males, resulting in a SIR of 0.78 (95% CI 0.31–1.60) for men.
1229
1230
1231
1232
1233
1234
1235
The Danish cohort study was updated by Schüz et al. (2006c) by extending the follow-up with an
additional 7 years, to December 31, 2002. In total, 26 cases of salivary gland tumours were identified during the
entire study period, including also the 7 cases identified in the first follow-up. All 26 cases were males. National
incidence rates used for comparison were calculated after exclusion of the exposed cohort. The SIR for ever
being a mobile phone subscriber was estimated at 0.86 (95% CI 0.56–1.26) for men, and 0.00 (95% CI 0.00–
1.02) for women. No dose-response analyses of time since first subscription were made. The results provide no
evidence in support of the hypothesis that mobile phone use increases the risk of salivary gland cancer.
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
[The study has limited power to analyse such a rare disease as salivary gland cancer. The registerbased design minimizes selection bias, and guarantees that exposure information is collected independent of the
disease, and does not need to rely on subjects’ ability to remember their mobile phone use. Using a subscription
as a proxy for mobile phone use results in non-differential misclassification of the exposure, since the actual
user of the phone is unknown. This error is, however, unlikely to be of any greater magnitude during these early
years of mobile telephony when mobile phone use was quite expensive. Not being able to include corporate
users also introduce exposure misclassification, as they are likely to be among the earliest and heaviest users of
mobile phones, which primarily weakens the statistical power of the study. However, the study covers a period
when mobile phones were still used by a small minority of the population and the resulting exposure
misclassification would have only a minor effect on the risk estimates.]
1246
1247
Auvinen and colleagues performed a case-control study of brain tumours and salivary gland cancer
nested within the Finnish population (Auvinen et al., 2002). Cases of salivary gland tumours in the age range
Salivary gland tumours
38
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
20–69 diagnosed during 1996 were identified through the nationwide Finnish Cancer registry (n=34). Five ageand sex-matched controls per case were randomly selected from the population registry. Information about startand end date of mobile phone subscriptions, and type of network, was obtained from the mobile phone network
providers. No information was collected on amount of phone use. Information was only available on private
subscriptions; corporate mobile phone use could not be assessed. In total, 3 (9%) cases of salivary gland cancer
had had an analogue phone subscription, and 1 (3%) case had had a digital phone subscription. The
corresponding numbers for controls were 15 (9%) and 3 (2%), respectively. Ever having had a mobile phone
subscription was associated with an OR of 1.3 (95% CI 0.4–4.7). All but one case and one control had had a
subscription for a maximum of two years. [The study can only address potential effects of short-term mobile
phone use. Considering the low prevalence of mobile phone use in the population, misclassification of the
exposure from lack of information about corporate subscriptions is likely to have a minimal effect on the risk
estimates.]
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
Hardell and colleagues conducted a case-control study of mobile phone use and the risk of malignant
salivary gland tumours in the age range 21–80 years (Hardell et al., 2004a). Regional cancer registries that
covered the whole of Sweden were used to identify 407 cases, during the period 1994-1999 (in some regions to
2000). At the time of the data collection 114 cases were excluded either because they had died (96), were too ill
to participate (3), or the treating physician did not grant permission to contact the case (15). The participation
proportion among cases was reported to be 91%, but if calculated according to conventional standards it was
66% (i.e. when including deceased, too ill, and physician refusal in the denominator). Four controls per case,
matched on age and sex, were selected from controls used in previous brain tumour studies by the same group,
covering the period 1994–1996 (Hardell et al., 1999) and 1997–2000 (Hardell et al., 2002). The participation
proportion among the previously selected controls was 92%. New controls were selected when the matching
criteria could not be fulfilled, and to cover two geographical regions that were not covered by the previous
studies, in total 357 controls of whom 303 participated (85%). In total, 267 cases and 1053 controls were
included in the analyses. Information on mobile phone use history was collected through a mailed questionnaire,
and all subjects were also contacted by phone to supplement the answers. Of the cases, 12% had used analogue
phones and 17% digital phones. For controls, the proportions were 11% and 16%, respectively. Hours of phone
use was categorized according to the median value among the controls. Unconditional logistic regression was
used to estimate odds ratios, adjusted for age and sex. Adjusting also for geographical region did not change risk
estimates, but was not reported. Risk estimates for ever use of an analogue phone was OR=0.9 (95% CI 0.6–
1.5), for digital phone use OR=1.0 (95% CI 0.7–1.5), and for cordless phone use OR=0.99 (95% CI 0.68–1.43).
Risk estimates did not increase with time since first use or hours of use. The authors report that no consistent
pattern of risk estimates was found according to laterality of phone use, but these data were not shown. In the
analyses by anatomic location, the risk estimate for the submaxillary gland was OR=1.4 (95% CI 0.6–3.5) for
any phone use, for the parotid gland OR=1.0 (95% CI 0.7–1.4), and for other locations OR=1.1 (95% CI 0.5–
2.7). The results do not support the hypothesis that mobile phone use increases the risk of salivary gland
tumours. [The study has limited power to study potential effects of long-term mobile phone use. In the original
studies from which the majority of the controls were drawn, matching of controls to brain tumour cases was
made on age, sex and geographic region.]
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
Lönn and colleagues reported results on parotid gland tumours pooling data from the Swedish and
Danish Interphone case-control studies (Lönn et al., 2006). The studies cover approximately 6.7 million people
(the whole of Denmark and a large part of Sweden). Eligible cases were persons 20–69 years old diagnosed
during 2000–2002 with malignant parotid gland tumours or benign pleomorphic adenoma (the latter only from
two geographical regions in Sweden). In total 71 malignant and 128 benign eligible cases were identified
through the nationwide cancer registries and from clinics treating these patients. There was no overlap of cases
with the study by Hardell et al. described above (Hardell et al., 2004a). Controls were randomly selected from
population registries, in Denmark individually matched to cases on age and sex, in Sweden stratified on age, sex
and geographical region. In total 60 (85%) malignant cases, 112 (88%) benign cases, and 681 (70%) controls
participated in the study. Information about mobile phone use history was collected through standardized
personal interviews that included questions about the year when mobile phone use started, number of phone
calls, hours of phone use, preferred side of the head, and use of hands-free device. Amount of phone use was
categorized according to the 25th and 75th percentiles among the controls. Analyses were conducted using
unconditional logistic regression, adjusted for age, sex, residential area, country and education. Using
conditional logistic regression analyses for the Danish material did not change results materially. Regular
mobile phone use (defined as at least once per week during at least six months) was reported by 60% of
controls. Ever being a regular mobile phone user was not associated with parotid gland tumour risk; OR=0.7
(95% CI 0.4–1.3) for malignant tumours and 0.9 (95% CI 0.5–1.5) for benign tumours. Risk estimates did not
increase with duration of use, time since first use, cumulative call time or cumulative number of calls. Odds
39
1306
1307
1308
1309
1310
1311
1312
1313
ratios for ipsilateral mobile phone use (use on the same side as where the tumour was diagnosed) were increased
in some analyses, but were accompanied by corresponding reduced risks for the contralateral side. The OR for
ipsilateral use that started more than 10 years prior to the diagnosis was 2.6 (95% CI 0.9–7.9) for benign
tumours, while the corresponding OR for contralateral use was 0.3 (95% CI 0.0–2.3). The results do not support
the hypothesis that mobile phone use increases the risk of parotid gland tumours. [It does not seem biologically
plausible that RF exposure would increase the risk of salivary gland tumours on one side of the head and at the
same time protect against tumours on the opposite side. Therefore the findings related to ipsilateral use were
interpreted as information bias.]
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
The Israeli Interphone study (Sadetzki et al., 2008) identified benign and malignant parotid gland
tumours among all Jewish individuals in Israel 18 years or older during 2001 to 2003. In total, 531 cases with
confirmed tumours were identified through periodic review of pathology/cytology reports from all 22
otolaryngology departments in the country. Of the 460 (87%) participating cases, 58 were malignant, and the
benign were distributed as 264 pleomorphic adenoma, 117 Warthin's tumour, and 21 others. Controls were
randomly selected from the population registry, individually matched to cases identified for the Interphone study
which also included glioma, meningioma and acoustic neuroma. Controls originally matched to the other
tumour types were post hoc matched to parotid gland tumour cases on age, sex, interview date and continent of
birth. Through this procedure, up to 7 controls per case was selected. In total, 1929 controls were selected, of
whom 1266 (66%) participated. Information on mobile phone use history was collected through personal
interviews according to the Interphone study protocol, in the same way as in the Swedish/Danish study
described above. Analyses were made using conditional logistic regression, with adjustment for smoking.
Regular mobile phone use (at least once per week over a period of at least six months) was reported by 55% of
controls, and was not associated with parotid gland tumours; the OR for malignant tumours was 1.06 (95% CI
0.54–2.10) and for benign tumours 0.85 (95% CI 0.64–1.12). Overall, the OR was 0.87 (95% CI 0.68–1.13). No
increased risks or trends were observed according to time since start of phone use, duration of use, cumulative
number of calls, or cumulative calling time. However, based on subgroup analyses (regular users only, rural
areas – based on the evidence of greater phone energy output in rural compared to urban areas, because of
higher distances to antennas -, and particularly ipsilateral use) the authors concluded that their results suggest an
association between mobile phone use and parotid gland tumours. For example for ipsilateral mobile phone use,
the odds ratios for above-median use were: 1.58 (95% CI 1.11–2.24) for cumulative number of calls and 1.49
(95% CI 1.05–2.13) for cumulative call time. However, the corresponding results for contralateral use were 0.78
(95% CI 0.51–1.19) and 0.84 (95% CI 0.55–1.28), respectively. [With no raised risk estimates overall, an
increased risk associated with ipsilateral phone use is likely to indicate recall bias when cases report usual side
of phone use prior to diagnosis, as it must be accompanied by a reduced risk in other subgroups, i.e. among
subjects with contralateral phone use or who lack information about side of phone use, considering that more
than half of the subjects reported ipsilateral phone use.]
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
Duan and co-workers (2011) performed a hospital-based case-control study of mobile phone use and
epithelial parotid gland malignancies in Beijing, China. Cases were recruited retrospectively from the authors’
clinic covering the time period 1993 to March 2010. Unmatched controls were recruited from other patients
treated for other diseases at the same clinic during the same time period, including patients with impacted teeth,
maxillofacial trauma, infections, temporomandibular joint disorders, maxillofacial nerve disorders, noncancerous potentially oral and maxillofacial tumours (without salivary gland tumour involvement), salivary
gland infections, congenital cleft lips and palates, or maxillofacial deformities, and excluding patients with
potential malignancies. In total, 221 eligible cases and 2643 eligible controls were identified. Only living
subjects, who agreed to participate, were included in the study leaving 136 (62%) cases, and 2051 (78%)
controls for analyses. Cases were in the age-range 7 to 80 years; no information about the age-range among
controls was given. Information on mobile phone use and potential confounders were collected through personal
or telephone interviews. Amount of phone use was categorized according to the median and third quartile
among the controls. Analyses were performed using unconditional logistic regression with adjustment for sex,
age, residential area, marital status, education, income, and smoking. No adjustment was made for year of
diagnosis. No association was found between regular mobile phone use (at least once per week during at least
six months) and epithelial parotid gland tumours (OR=1.14; 95% CI 0.72–1.81) or mucoepidermoid carcinoma
(OR=1.37; 95% CI 0.63–2.10). However, in analyses of indices of amount of mobile phone use considerable
risk increases were reported. For example, for time since first mobile phone use, significantly increased risk
estimates were observed in all exposure categories, varying from 1.7 to 5.4. [Reported risk increases for amount
of phone use are incompatible with overall risk estimates close to unity. Therefore, these results are impossible
to interpret. Exposure cutpoints do not seem to have been defined as described by the authors. In addition, there
is potential for confounding from date of diagnosis, as mortality is likely to be higher among the cases with
40
1363
1364
malignant tumours, than among controls, and therefore a smaller proportion of cases are likely to have been
included from the early part of the study period when mobile phone use was less common.]
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
Söderqvist and colleagues (2012) conducted a case-control study of mobile phone use and the risk of
malignant salivary gland tumours in parts of Sweden. The regional cancer registries in Uppsala/Örebro and
Linköping were used to identify 88 eligible cases during the period 2000–2003. There was no overlap between
this study and the previous study by the Hardell group. In total, 69 cases agreed to participate. The participation
proportion among cases was reported to be 88%, but if calculated according to conventional standards it was
78% (i.e. when including in the denominator also six deceased cases and four cases for whom the physician did
not permit contact). Four controls per case, matched on county, age and sex, were randomly selected from the
population registry, and 262 agreed to participate (83%). Information about mobile phone use and use of
cordless phones were collected through mailed questionnaires, using the same questionnaire as in the previous
studies by the same group. Cases and controls were in the age-range 22 to 80 years. Analyses were made using
unconditional logistic regression, adjusted for age, sex, year of diagnosis and socioeconomic status. Hours of
phone use was categorized according to the median value among the controls, and in additional analyses,
cutpoints at 1000 h and 2000 h were used. Mobile phone use was reported by 43% of cases and 42% of controls,
and was not associated with malignant parotid gland tumours; OR=0.8 (95% CI 0.4–1.6). Risk did not increase
with increasing amount of use or time since first use. Only 2 cases had used a mobile phone longer than 10
years. No associations were found with cordless phone use. The results do not support the hypothesis that
mobile phone use increases the risk of parotid gland tumours. [The prevalence of mobile phone use was low,
and the study had limited statistical power to study long-term mobile phone use.]
Table 12.1.6. Case-control studies of mobile phone use and salivary gland tumours
Country
Time period
No.
cases/controls
Exposure
Age range
Source of
controls
Finland
Malignant salivary Mobile phone subscription
gland
Never
34/170
Ever
1996
20-69
Population
registry
Sweden
1994-1999
(2000)
21-80
No. exp
cases
Relative risk (95%
CI)
30
1.0
4
1.3 (0.4-4.7)
<1 year
0
0.0 -
1-2 years
3
1.7 (0.4-7.5)
>2 years
1
2.3 (0.2-25.3)
Comments
Reference
Auvinen et al.
(2002)
Time since first subscription
Malignant salivary Mobile phone use, analogue
gland
Ever, time since first use
267/1053
>1 year
31
Population
>5 years
17
registry
>10 years
6
Exposure categories for
no. of years overlap.
Hardell et al.
(2004a)
0.92 (0.58-1.44)
0.78 (0.44-1.38)
0.71 (0.29-1.74)
Cumulative hours of use
≤91 h
16
0.94 (0.52-1.68)
>91 h
15
0.90 (0.49-1.66)
>1 year
45
1.01 (0.68-1.50)
>5 years
8
1.22 (0.54-2.78)
≤64 h
22
0.96 (0.58-1.59)
>64 h
23
1.07 (0.64-1.80)
48
0.99 (0.68-1.43)
Mobile phone use, digital
Ever, time since first use
Cumulative hours of use
Cordless phone use
Ever
41
Sweden,
Denmark,
Interphone
Malignant parotid
gland
Regular mobile phone use
Never
35
1.0
60/681
2000-2002
Ever
25
0.7 (0.4-1.3)
Population
registry
Time since first regular use
<5 years
14
0.7 (0.3-1.3)
5-9 years
8
0.7 (0.3-1.7)
≥10 years
2
0.4 (0.1-2.6)
<30 h
7
0.7 (0.3-1.6)
30-449 h
11
0.7 (0.3-1.4)
5
0.6 (0.2-1.8)
20-69
Lönn et al. (2006)
Cumulative hours of use
≥450 h
Benign
pleomorphic
adenoma
112/321
Population
registry
Regular mobile phone use
Never
35
1.0
Ever
77
0.9 (0.5-1.5)
<5 years
47
1.0 (0.6-1.8)
5-9 years
23
0.8 (0.4-1.5)
≥10 years
7
1.4 (0.5-3.9)
<30 h
20
1.1 (0.6-2.3)
30-449 h
34
0.9 (0.5-1.6)
≥450 h
22
1.0 (0.5-2.1)
Time since first regular use
Cumulative hours of use
Israel,
Interphone
Malignant parotid
gland
Regular mobile phone use
Never
25
1.0
2001-2003
58/194
Ever
33
1.06 (0.54-2.10)
≥18
Population
registry
Time since first regular use
<5 years
21
1.25 (0.58-2.68)
5-9 years
11
0.92 (0.37-2.27)
≥10 years
1
0.47 (0.05-4.51)
18
1.21 (0.58-2.53)
5
0.67 (0.19-2.38)
10
1.22 (0.43-3.48)
Sadetzki et al.
(2008)
Cumulative hours of use
<5479
5480-18 996
≥18 997
Benign parotid
gland
Regular mobile phone use
Never
150
1.0
402/1072
Ever
252
0.85 (0.64-1.12)
Population
registry
Time since first regular use
<5 years
117
0.77 (0.56-1.06)
5-9 years
123
0.95 (0.68-1.32)
≥10 years
12
0.93 (0.44-1.98)
Cumulative hours of use
<266.3
103
0.78 (0.57-1.06)
266.4–1034.9
75
1.05 (0.72-1.53)
≥1035
73
1.08 (0.72-1.62)
42
China
1993-2010
Malignant parotid
gland
7-80 for cases, 136/2051
not given for
Hospital-based
controls
Sweden
2000-2003
22-80
Regular mobile phone use
Never
45
1.0
Ever
91
1.37 (0.64-2.10)
0-6 years
67
1.69 (1.05-2.73
7-8 years
6
4.17 (3.25-5.10)
9-10 years
3
5.36 (4.09-6.63)
>10 years
15
4.13 (3.28-4.99)
Time since first regular use
Malignant salivary Mobile phone use
gland
Ever
69/262
Time since first use
30
0.8 (0.4-1.6)
Population
registry
≤5 years
14
1.0 (0.4-2.2)
6-10 years
14
0.9 (0.4-2.1)
>10 years
2
0.3 (0.1-1.4)
≤66 h
16
0.9 (0.4-1.9)
>66 h
14
0.7 (0.3-1.6)
19
0.6 (0.3–1.2)
Results for extent of
Duan et al. (2011)
phone use (time since first
use, number of calls,
calling time, etc) do not
correspond to the overall
result. Selective nonresponse because of
higher mortality among
cases is likely to produce
an upward bias.
Söderqvist et al.
(2012)
Cumulative hours of use
Cordless phone use
Ever
1383
1384
12.1.2.3
1385
1386
1387
1388
Other tumours that have been studied in relation to mobile phone use are pituitary gland tumours (3
studies), leukaemia (4 studies), Non-Hodgkin’s lymphoma (4 studies), uveal melanoma (3 studies), testicular
cancer (2 studies), and melanoma and other skin cancer (3 studies). In addition, the two cohort studies report
results for several other tumour types.
1389
Pituitary gland tumours
1390
1391
1392
1393
1394
1395
1396
1397
Takebayashi and co-workers included newly identified patients with pituitary adenoma in the age
group 30–69 years, in parts of Tokyo between December 2000 and November 2004, and interviewed 101 (76%)
cases and 161 (49%) controls as part of the Japanese Interphone study (Takebayashi et al., 2008). Controls were
selected through random digit dialling, matched to cases by age, sex, and residence. Exposure information was
collected through personal interviews, as described above for the Interphone study. Analyses were performed
with conditional logistic regression, with adjustment for education and marital status. Regular mobile phone use
was associated with an OR of 0.90 (95% CI 0.50–1.61), and effect estimates did not increase with time since
first use or amount of use.
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
Schoemaker and Swerdlow extended the UK-South Interphone study to also include pituitary gland
tumours in the age group 18–59 years, identified between 2001 and 2005 (Schoemaker & Swerdlow, 2009). In
total, 317 (63%) cases participated. Controls were randomly selected from general practitioners patient lists,
considered to cover approximately 98% of the UK population, and were frequency matched to cases (including
also brain tumour cases) on the sex, age, and health-authority distribution. Of the selected controls, 603 (43%)
participated. Exposure information was collected through personal interviews, as described above for the
Interphone study. Unconditional logistic regression was used for the analyses, with adjustment for adjusted for
sex, age, geographic area, reference date, and Townsend deprivation score (as a measure of socioeconomic
status). Regular mobile phone use was associated with an OR of 0.9 (95% CI 0.7–1.3), and effect estimates did
not increase with time since first use or amount of use.
1408
1409
1410
1411
1412
Benson and colleagues studied several different tumour types in relation to mobile phone use in their
cohort study based on data from the UK Million Women cohort (Benson et al., 2013a). See page x for a
description of the study design. In total, 110 cases of pituitary gland tumours were identified. Ever use of a
mobile phone was associated with a RR of 1.52 (95% CI 0.99–2.33), while for daily mobile phone use the RR
was 1.45 (95% CI 0.68–3.10). At least 10 years duration was associated with a RR of 1.61 (95% CI 0.78–3.35).
Other tumours
43
1413
Leukaemia
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
Two case-control studies of leukaemia have been conducted. Kaufman and colleagues conducted a
hospital-based case-control study of leukaemia with 180 eligible cases 18 years or older diagnosed at one
hospital in Bangkok 1997–2003 (Kaufman, Anderson & Issaragrisil, 2009). All eligible cases agreed to
participate. Controls were identified among other patients at the hospital, e.g. with acute infections, trauma,
appendicitis, matched to cases on age, sex, and residential requirements. A total of 756 controls were eligible,
and all of them participated in the study. Exposure information was collected through interviews. Analyses were
conducted with unconditional logistic regression, but authors state that results from conditional analyses did not
differ from those presented. Adjustment was made for age, sex, income, solvents, pesticides, working with
power lines, residence near power lines. Mobile phones were used by 14% of controls, with a median duration
of 2 years. Mobile phone use was associated with an OR of 1.5 (95% CI 1.0–2.4), and there were no differences
between cases and controls with regard to duration or amount of use [ORs were not presented]. Use of a GSM
phone was associated with an OR of 2.1 (95% CI 1.1–4.0), with similar duration and amounts of use between
cases and controls. A “high risk” group was identified based on the proportion of time that calls were initiated,
antenna was extended, and/or whether metal glasses were ever worn; the OR for this “high risk” group was 1.8
(95% CI 1.1–3.2). [The authors list a number of other factors that they believe are associated with higher
exposure levels, but include only a selected number when constructing their “high risk” group. The study is
limited by the short duration of mobile phone use among the majority of subjects and the hospital-based
identification of cases and selection of controls from other patient groups at the hospitals.]
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
Cooke and colleagues conducted a case-control study of leukaemia (excluding chronic lymphocytic
leukaemia) in South East England with eligible cases aged 18–59 years diagnosed between 2003 and 2007 (to
2009 in two areas) (Cooke, Laing & Swerdlow, 2010). In total, 806 (50%) cases participated. Controls were
selected among non-blood relatives to cases, who did not live with the case. In total 585 (75%) controls agreed
to participate. Exposure information was collected through face-to-face interviews, using a modified version of
the Interphone questions about mobile phone use. Analyses were conducted with unconditional logistic
regression adjusted for age, sex, socio-economic status, area of residence, ethnicity, smoking, interview lag time
and calendar period. Conditional logistic regression did not give materially different results. Amount of phone
use was categorized with cutpoints according to the median and third quartile of the distribution among controls.
Regular mobile phone use was associated with an OR of 1.06 (95% CI 0.76–1.46). The highest category of time
since first use (≥15 years) was associated with an OR of 1.87 (95% CI 0.96–3.63), while ORs in the three lower
categories were close to unity. Amount of phone use estimated as cumulative number of calls or cumulative
hours of use was not consistently associated with leukaemia. None of the results stratified on type of phone use
was statistically significant, but ORs were slightly higher for analogue than digital phone use (e.g. the OR for
≥10 years since first use of an analogue phone was 1.54 (95% CI 0.91–2.62), while the corresponding result for
use of a digital phone was 0.78 (95% CI 0.44–1.38). [A limitation is low case participation rate, which would
have resulted in a loss of statistical power, and the use of non-blood relatives as controls, although this is likely
to have resulted in a higher response rate among controls.]
1450
1451
1452
1453
1454
1455
1456
Two cohort studies that included analyses of leukaemia are available. The Danish cohort study of
mobile phone subscribers identified 84 cases of leukaemia in their first analysis (Johansen et al., 2001), and in
the update with follow-up through 2002 (Vrijheid et al., 2006), 351 cases (318 men and 33 women) were
identified (see page xx for a detailed description of the study). In the most updated analysis, the SIR for ever
being a mobile phone subscriber was 1.00 (95% CI 0.90–1.12) for men and 0.97 (95% CI 0.67–1.36) for
women. There was no trend in risk estimates according to time since first subscription, the SIR for ≥10 years
was 1.08 (95% CI 0.74–1.52) for men and women combined.
1457
1458
1459
1460
1461
Benson and colleagues also included leukaemia as an outcome in their cohort study based on data
from the UK Million Women cohort (Benson et al., 2013a). See page x for a description of the study design. In
total, 860 cases of leukaemia were identified. Ever use of a mobile phone was associated with a RR of 0.91
(95% CI 0.79–1.05), while for daily mobile phone use the RR was 0.88 (95% CI 0.66–1.19). At least 10 years
duration of mobile phone use was associated with a RR of 0.92 (95% CI 0.70–1.21).
1462
Non-Hodgkin lymphoma
1463
1464
1465
1466
Two case-controls studies and two cohort studies have been conducted on non-Hodgkin lymphoma.
Hardell and co-workers conducted a case-control study in four geographical regions in Sweden with case
recruitment from December 1999 to April 2002 (Hardell et al., 2005). Eligible cases were 18–74 years old and
identified through the clinics where these cases were treated and through pathologists. In total 1133 cases with
44
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
non-Hodgkin’s lymphoma were identified, and 910 (80%) cases participated in the study. Controls were
randomly selected from the population registry, frequency matched to cases by age (10-year groups), sex and
geographic region; 1016 (92%) agreed to participate. Exposure information was collected through mailed
questionnaires, completed over the telephone. Analyses were made with unconditional logistic regression,
adjusted for age, sex and year of diagnosis (cases) or enrolment (controls). Hours of phone use was categorized
according to the median among the controls. Ever use of an analogue mobile phone >1 year prior to diagnosis
was associated with an OR of 0.94 (95% CI 0.68–1.30). Effect estimates were close to unity also for use of
digital or cordless phones, and did not increase with longer time since first use or higher amounts of use. In
analyses of subtypes of lymphoma, higher ORs were observed for T-cell lymphoma (53 cases in total), but risk
estimates had wide confidence intervals, and reached significance only for cordless phone use. For analogue and
digital phones no consistent dose-response patterns with amount of use or time since first use were observed,
while for cordless phone use higher ORs were found with longer times since first use, but not according to
amount of use. [Results for T-cell lymphoma were unstable, and finding the highest risk increase for cordless
phones which have the lowest exposure speaks in favour of random variation as the explanation.]
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
Linet and colleagues conducted a case-control study of non-Hodgkin lymphoma in 4 geographic
regions in the USA (Linet et al., 2006). Cases were 20–74 years old, identified through the National Cancer
Institute’s population-based Surveillance, Epidemiology and End Results (SEER) program between July 1998
and June 2000. In total 551 (61%) of the cases participated in the study. Controls were frequency matched to
cases on age, sex, ethnic group and geographic region, identified through random digit dialling when younger
than 65 years, while ≥65 years were randomly selected through Medicare eligibility files, 462 (55%) controls
participated. Exposure information was collected through personal interviews. Analyses were made with
unconditional logistic regression, adjusted for age, geographic area, gender, ethnic group, education, marital
status and family history of any lymphoproliferative malignancy among first-degree relatives. Regular mobile
phone use, defined as >100 times, was associated with an OR of 0.9 (95% CI 0.6–1.4). There were no trends in
risk associated with amount of use or time since first use. In the highest category of time since first use (>8
years) the OR was 1.6 (95% CI 0.7–3.8). Results were similar in separate analyses of the major subcategories of
lymphoma.
1494
1495
1496
1497
The Danish cohort study of mobile phone subscribers reported results for non-Hodgkin lymphoma
only in their first analysis, where 120 cases were identified (Johansen et al., 2001), (see page xx for a detailed
description of the study). The SIR for ever being a mobile phone subscriber was 0.93 (95% CI 0.77–1.13) for
men and 1.04 (95% CI 0.52–1.86) for women. No other analyses of lymphoma were conducted.
1498
1499
1500
1501
1502
1503
Benson and colleagues reported results for non-Hodgkin lymphoma in their cohort study based on
data from the UK Million Women cohort (Benson et al., 2013a). See page x for a description of the study
design. In total, 2058 cases of non-Hodgkin lymphoma were identified. Ever use of a mobile phone was
associated with a RR of 0.97 (95% CI 0.88–1.06), while for daily mobile phone use the RR was 0.94 (95% CI
0.78–1.13). At least 10 years duration of mobile phone use was associated with a RR of 0.99 (95% CI 0.83–
1.17).
1504
Uveal melanoma and other cancers in the eye
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
Stang and colleagues conducted a case-control study of mobile phone use and uveal melanoma (Stang
et al., 2009), as a follow-up of their earlier study of occupational exposure to radio sets/mobile phones (Stang et
al., 2001) (see section of occupational exposure (page x). Eligible cases were aged 20–74 years and diagnosed
between 2002 and 2004 at one hospital; 459 (94%) cases participated in the study. Three control groups were
identified; agreement to participate was obtained from 827 (57%) population based controls who were selected
from censuses matched to cases by age, sex, and region of residence, 187 (57%) sibling controls who were
within 10 years of the case’s age, and 180 (52%) controls among other ophthalmology patients (only during the
first half of the study period). Information about mobile phone use was obtained through computer-assisted
telephone interviews using the Interphone questionnaire. Analyses were made with conditional logistic
regression accounting for the matching factors. Regular mobile phone use, with never use of a mobile phone as
reference, was associated with an OR of 0.7 (95% CI 0.5–1.0) using population based controls, and closer to
unity with sibling and patient controls. ORs when population based controls were used were all below unity,
while ORs with sibling or hospital based controls were slightly higher. There were no consistent associations for
categories of time since first use, or measures of amount of mobile phone use. [As in the Interphone study, nonparticipating population based controls were less likely to be mobile phone users, which may at least partly
explain the reduced risk estimates observed. Representativity of patient or sibling controls is questionable.]
45
1521
1522
1523
1524
1525
The Danish cohort study of mobile phone subscribers identified only 8 cases of eye cancer in their
first analysis (Johansen et al., 2001), and in the update with follow-up through 2002, 44 cases were identified, of
which 38 were men (see page xx for a detailed description of the study) (Vrijheid et al., 2006). The SIR for ever
being a mobile phone subscriber was 0.94 (95% CI 0.66–1.29) for men and 1.10 (95% CI 0.40–2.39) for
women.
1526
1527
1528
1529
1530
Benson and colleagues identified 87 cases of cancer in the eye in their cohort study based on data
from the UK Million Women cohort (Benson et al., 2013a). See page x for a description of the study design.
Ever use of a mobile phone was associated with a RR 1.01 (95% CI 0.64–1.60), while for daily mobile phone
use the RR was 0.75 (95% CI 0.29–1.97). At least 10 years duration of mobile phone use was associated with a
RR of 0.82 (95% CI 0.31–2.19).
1531
Testicular cancer
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
A case-control study of mobile phone use and testicular cancer was conducted by Hardell and
colleagues (Hardell et al., 2007). Eligible cases were identified through the Swedish cancer registry, were 20–75
years old and diagnosed in Sweden between 1993 and 1997. One control per case was randomly selected from
the population registry, matched on age. Exposure information was collected through a postal questionnaire,
supplemented by telephone if necessary, with participation of 888 (84%) cases and 870 (89%) controls.
Analyses were made with unconditional logistic regression adjusted for age, year of diagnosis and
cryptorchidism. Hours of phone use was categorized according to the median among the controls. Use of an
analogue phone >1 year was associated with an OR of 1.0 (95% CI 0.8–1.3). Results for digital or cordless
phones were essentially identical. Risk estimates did not increase with time since start of mobile phone use or
cumulative hours of use, but the numbers of subjects with long-term use was small.
1542
1543
1544
1545
1546
1547
The Danish cohort study of mobile phone subscribers identified 187 cases with testicular cancer in
their first analysis (Johansen et al., 2001), and in the update with follow-up through 2002, 522 cases were
identified (see page xx for a detailed description of the study) (Vrijheid et al., 2006). The risk estimate in early
study was close to unity, and with the updated follow-up the SIR for ever being a mobile phone subscriber was
1.05 (95% CI 0.96–1.15). The majority of persons in the cohort had only used mobile phone during a short
period.
1548
Malignant melanoma and other skin cancers
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
Hardell and colleagues conducted a Swedish nationwide case-control study of malignant melanoma
in the head and neck region (Hardell et al., 2011). Eligible cases were 20–77 years old, diagnosed during 2000
and 2003. In total 347 cases (82%) participated in the study. One control per case was randomly selected from
the Swedish population registry, matched on age and sex. The analyses included also controls matched to cases
with skin cancers at other locations, and in total 1184 controls (80%) participated. Information about mobile
phone use and other risk factors was collected through mailed questionnaires. Analyses were made with
unconditional logistic regression, adjusted for age, sex and year of diagnosis. Contrary to previous studies,
analogue and digital phones were not analysed separately; results were presented for mobile phone use, cordless
phones use, and the two combined (wireless phone use). Cumulative call time was categorized according to the
median value among controls. The categories 0–1000, 1001–2000, and >2000 h were also used. Furthermore,
analyses were made according to laterality of phone use in relation to tumour side, according to age at start of
phone use, interaction analyses with known risk factors for malignant melanoma. In total, a very large number
of subgroups were analysed [likely a least 500], although not all results are shown in tables. For mobile phone
use, no associations were observed overall, or according to time since first use. The OR for >1 year of mobile
phone use was 1.0 (95% CI 0.7–1.3), and for >10 years it was essentially identical, and did not change with
cumulative hours of use; >128 h of mobile phone use was associated with an OR of 1.0 (95% CI 0.7–1.4). For
cordless phones, most risk estimates were also close to unity, with the exception of >365 h of use with >1–5
years latency, where the OR was 1.6 (95% CI 0.9–2.6). In corresponding categories with longer latencies results
were close to unity. Results for wireless phone use (mobile and cordless phones combined) were similar to the
results for cordless phones. Results using the categorization of hours of use with the highest >2000 h showed no
significant associations overall (not reported in tables). Subgroup analyses according to location of the tumour
showed an OR of 2.1 (95% CI 1.1–3.8) for tumours located in the temporal area, cheek or ear for >365 h of
cordless phone use with >1–5 years latency; no associations for longer latencies or overall. Results for the
corresponding category for tumours at other locations were also slightly raised for cordless and wireless phone
use, but were not statistically significant, while again, all other categories had results close to or below unity. In
what appears to be a post hoc analysis, mobile phone use >365 h (the median for cordless phone use) was
46
1575
1576
1577
1578
1579
1580
1581
1582
1583
associated with an OR of 2.1 (95% CI 0.7–6.1) for tumours with location in the temporal area, cheek or ear with
>1–5 years latency, based on 5 exposed cases and 19 controls. [This means that the OR for the category 129–
365 h with corresponding latency and tumour location must be considerably reduced, as the category >128 h
consisted in total of 7 cases and 55 controls.] Analyses according to age at first use of mobile or cordless phones
indicated raised risks in the youngest category (<20 years), but were based on very small numbers and were not
statistically significant. No interactions between wireless phone use and known risk factors for malignant
melanoma were found. [Of the large number of analyses performed in this study, almost all showed no
associations, regardless of time since first use, amount of use or tumour location, and the few raised ORs
showed no consistent does-response pattern with amount of use or time since first use.]
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
The Danish cohort study of mobile phone subscribers studied skin cancer among persons ≥30 years
after linkage of the cohort to another nationwide cohort where information about indicators on socioeconomic
status was available (Poulsen et al., 2013), similar to the latest updates of brain tumours described above (Frei et
al., 2011). In total, 4675 cases of basal cell carcinoma, 540 cases of squamous cell carcinoma, and 802 cases of
melanoma were identified among the subscribers during the follow-up 1990 through 2007. Analyses were made
with log-linear Poisson regression models, stratified by sex and adjusted for age and calendar period, and in
addition education and income. Analyses were made for the different cancer types overall, and more detailed
analyses for tumour locations at head and neck, and torso and legs, respectively. In addition, the ratios of the
IRRs for these two locations were calculated. The longest latency period investigated was ≥13 years since first
subscription. The results showed no evidence of an increased risk of any type of skin cancer located in the head
and neck, or torso and legs among the early mobile phone subscribers. For example, for melanoma overall, the
IRR was 0.94 (95% CI 0.87–1.01), For men the IRR for melanoma of the head and neck was 1.05 (95% CI
0.80–1.37), and did not vary with time since first subscription. The corresponding result for women was
IRR=0.76 (95% CI 0.34–1.72), but was based on small numbers. Results for other types of skin cancers were
based on larger number of subjects, and indicated no raised risks.
1599
1600
1601
1602
1603
Benson and colleagues identified 2116 cases of melanoma in their cohort study based on data from
the UK Million Women cohort (Benson et al., 2013a). See page x for a description of the study design. Ever use
of a mobile phone was associated with a RR 1.06 (95% CI 0.96–1.16), and for daily mobile phone use the RR
was 1.06 (95% CI 0.89–1.26). At least 10 years duration of mobile phone use was associated with a RR of 1.09
(95% CI 0.92–1.29).
1604
Other cancer types
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
Both of the available cohort studies (Benson et al., 2013a; Johansen et al., 2001; Vrijheid et al., 2006)
included analyses of several other cancer types. The longest updated publication from the Danish cohort study
of mobile phone subscribers (Vrijheid et al., 2006) report results for all smoking related sites combined and
separately (i.e. lung, larynx, bladder, buccal cavity/pharynx, oesophagus, liver, cervix uteri, stomach, kidney,
pancreas) and breast, prostate, other cancers, unspecified cancers, apart from the cancers already discussed
above. The study found a reduced risk for all cancers combined among men (SIR=0.93; 95% CI 0.92–0.95;
11 802 observed cases), while results for women were close to unity (SIR=1.03; 95% CI 0.99–1.07; 2447
observed cases). The risk reduction among men was mainly confined to smoking related cancer sites, the SIR
for all smoking related sites combined was 0.88 (95% CI 0.86–0.91), and for male lung cancer 0.82 (95% CI
0.78–0.87). For women, the risk of smoking related cancers was instead increased, with SIR=1.11 (95% CI
1.02–1.21). Also cervical cancer was associated with an increased risk estimate, SIR=1.30 (95% 1.08–1.54).
Risk estimates for other cancer sites were close to unity, including also the most common gender specific cancer
sites prostate cancer (1001 cases) and female breast cancer (711 cases). [Mobile phone use in the early period
was more common among persons with higher socioeconomic status, persons who also smoke less, which is a
likely explanation for the risk reduction observed for smoking related cancer sites among men (analyses were
not adjusted for socioeconomic status). The risk increase for these cancer sites among women and for cervical
cancer may indicate that the earliest female mobile phone users had a different lifestyle compared to the general
female population.]
The UK cohort study based on the Million Women cohort collected information about mobile phone
use from the women in the cohort between 1999 and 2005 (Benson et al., 2013a), and cancer incidence was
followed through 2009 (see page x for a detailed description of the study design). The study report results for all
invasive neoplasms combined, and separately for cancers of the thyroid, other head and neck, oesophagus,
stomach, colon, rectum, pancreas, lung, breast, endometrium, ovary, kidney, bladder, and multiple myeloma. A
reduced risk estimate was observed for all cancers combined (RR= 0.97; 95% CI 0.95–0.99), and for lung
cancer (RR= 0.89; 95% CI 0.84–0.95) and stomach cancer (RR= 0.84 95% CI 0.70–1.00). No significantly
47
1631
1632
1633
raised risk estimates were observed; the highest was for pituitary gland tumours discussed above [adjustment
was made for indicators of socioeconomic status and smoking, but residual confounding from smoking cannot
be ruled out, which also an observed risk reduction for stroke may indicate.]
1634
Studies with uncertainties related to inclusion criteria
1635
1636
1637
1638
1639
1640
1641
1642
1643
Warren and colleagues conducted a hospital based case-control study in the US primarily focused on
intratemporal facial nerve (IFN) tumours in relation to mobile phone use (Warren et al., 2003). The study is
described in more detail on page x. The study included 18 cases of IFN. Controls used for analyses were other
patients with rhinosinusitis, dysphonia or gastroesophageal disease, in total 141 control patients. Risk factor
information was collected by telephone interview. Ever use of a mobile phone was associated with an OR of 0.6
(95% CI 0.2-1.9), and use at least once per week with an OR of 0.4 (95% CI 0.1-2.1). Corresponding results for
cordless phone use was 0.3 (95% CI 0.1-0.9) and 0.6 (95% CI 0.2-1.7). [Response rates are not stated and it is
not clear how many subjects were excluded for various reasons. The study is given little weight in the overall
assessment, and is not included in the table].
Table 12.1.7. Studies of mobile phone use and other tumours
Country
Design
Time period
No. subjects
cases/controls
Age range
Exposure
No. exp
cases
Relative risk (95%
CI)
Comments
Reference
Takebayashi et al.
(2008)
Source of
controls
Pituitary gland tumours
Japan
Case-control
Regular mobile phone use
2000-2004
101/161
Never
39
1.0
30-69
Random digit
dialling, matched
on age, sex, and
residence.
Ever
62
0.90 (0.50-1.61)
Conditional logistic
regression, adjusted for
education and marital
status.
<2.4 years
14
0.86 (0.39-1.88)
Extended Japanese
Interphone study.
2.4- years
13
0.75 (0.31-1.81)
4.5- years
22
1.64 (0.74-3.66)
7.2- years
13
0.75 (0.31-1.82)
<32 h
15
1.00 (0.46-2.16)
32- h
14
0.97 (0.40-2.32)
160- h
12
0.72 (0.31-1.70)
620- h
21
1.33 (0.58-3.09)
Time since first regular use
Cumulative hours of use
UK
Case-control
Regular mobile phone use
2001-2005
317/603
Never
116
1.0
18-59
General
practitioners
patient lists,
covers 98% of
population,
frequency
matched on sex,
age, and healthauthority
distribution
Ever
175
0.9 (0.7-1.3)
1.5-4 years
89
1.0 (0.7-1.5)
5-9 years
62
0.8 (0.51.2)
10-17 years
24
1.0 (0.51.9)
<113 h
79
0.9 (0.6-1.3)
113-596 h
44
1.1 (0.7-1.8)
>596 h
51
1.1 (0.7-1.7)
<2203
72
0.8 (0.6-1.2)
2203-8300
45
1.1 (0.7-1.8)
>8300
57
1.2 (0.7-1.9)
Time since first regular use
Unconditional logistic
regression adjusted for
sex, age, geographic
area, reference date, and
Townsend deprivation
score (as a measure of
SES).
Schoemaker &
Swerlow (2009)
Extended UK Interphone
study.
Cumulative hours of use
Cumulative no. calls
48
UK
Cohort study
1999-2005 –
followed
through 2009
791 710 women
participating in the
UK Million
Mean age 60.3 Women Study,
who answered a
(SD 5.1)
base-line
questionnaire
1999-2005
Ever mobile phone user
No
33
1.0
Yes
77
1.52 (0.99-2.33)
<5 years
29
2.31 (1.31-4.06)
5-9 years
30
1.08 (0.64-1.82)
≥10 years
11
1.61 (0.78-3.35)
<Daily
68
1.53 (0.99-2.36)
Daily
9
1.45 (0.68-3.10)
Duration of use
110 cases of
pituitary gland
tumours
Frequency of use
Thailand
Case-control
Ever mobile phone user
1997-2003
180/756
No
145
≥18 years
Hospital-based
matched on age,
sex, residential
requirements
Yes
35
UK
Case-control
Regular mobile phone use
2003-2009
806/585
Never
132
1.0
18-59
Non-blood
Ever
relatives not living Time since first regular use
with the case
0.5-4 years
674
1.06 (0.76-1.46)
Prospectively collected
self-reported information
on mobile phone use.
Benson et al.
(2013a)
Definition of “user” did not
require a minimal amount
of use.
Cox proportional hazards
models adjusted for age,
area based SES,
geographical region,
height, BMI, smoking,
alcohol, strenuous
exercise, menopausal
hormone therapy.
Leukaemia
1.0
1.5 (1.0-2.4)
195
0.98 (0.68-1.42)
5-9 years
307
1.07 (0.75-1.53)
10-14 years
111
0.98 (0.63-1.51)
50
1.87 (0.96-3.63)
<284 h
283
0.97 (0.69-1.36)
284-1156 h
160
1.14 (0.76-1.71)
>1156 h
176
1.19 (0.79-1.80)
<5350
292
1.01 (0.71-1.42)
5350-16062
166
1.13 (0.76-1.69)
>16062
160
1.03 (0.68-1.56)
318
1.00 (0.90-1.12)
33
0.97 (0.67-1.36)
≥15
Unconditional logistic
regression, adjusted for
age, sex, income,
solvents, pesticides,
working with power lines,
residence near power
lines. Duration or amount
of use did not differ
between cases and
controls but were not
reported.
Kaufman et al.
(2009)
Unconditional logistic
regression adjusted for
age, sex, socio-economic
status, area of residence,
ethnicity, smoking,
interview lag time and
calendar period.
Cooke et al.
(2010)
Modified version of the
Interphone questions
about mobile phone use.
Cumulative hours of use
Cumulative no. calls
Denmark
Cohort study
Ever subscriber, men
1982-2002
420 095 mobile
phone
subscribers
Ever subscriber, women
≥18
Time since first subscription
<1 year
Comparison
group: the whole 1-4 years
Danish population 5-9 years
who were not
≥10 years
subscribers
33
1.09 (0.75-1.52)
151
1.05 (0.90-1.24)
135
0.92 (0.77-1.08)
32
1.08 (0.74-1.52)
Standardized incidence
ratios, by age and sex.
Vrijheid et al.
(2006)
~200 000 corporate
subscribers included in
comparison group,
constitute less than 5% of
unexposed.
49
UK
Cohort study
1999-2005 –
followed
through 2009
791,710 women
participating in the
UK Million
Mean age 60.3 Women Study,
who answered a
(SD 5.1)
base-line
questionnaire
1999-2005
Ever mobile phone user
No
382
1.0
Yes
478
0.91 (0.79-1.05)
Daily use
53
0.88 (0.66-1.19)
Duration of use ≥10 years
67
0.92 (0.70-1.21)
Prospectively collected
self-reported information
on mobile phone use.
Benson et al.
(2013a)
Definition of “user” did not
require a minimal amount
of use.
Cox proportional hazards
models adjusted for age,
area based SES,
geographical region,
height, BMI, smoking,
alcohol, strenuous
exercise, menopausal
hormone therapy.
860 cases of
leukaemia
Non-Hodgkin lymphoma
Sweden
Case-control
Mobile phone use, analogue
Exposure categories for
no. of years overlap.
1999-2002
910/1016
Ever, time since first use
18-74
Population
registry,
frequency
matched to cases
by age, sex and
geographic region
>1 year
141
0.94 (0.68-1.30)
>5 years
131
1.02 (0.73-1.43)
>10 years
74
0.96 (0.65-1.42)
≤198 h
61
0.82 (0.56-1.22)
>198 h
80
1.08 (0.73-1.59)
>1 year
422
1.02 (0.81-1.28)
>5 years
133
0.92 (0.66-1.27)
7
1.13 (0.38-3.35)
≤91 h
223
0.98 (0.76-1.26)
>91 h
199
1.09 (0.82-1.45)
313
1.01 (0.80-1.28)
Hardell et al.
(2005)
Unconditional logistic
regression, adjusted for
age, sex and year of
diagnosis or enrolment.
Cumulative hours of use
Mobile phone use, digital
Ever, time since first use
>10 years
Cumulative hours of use
Cordless phone use
Ever
USA
Case-control
Lifetime mobile phone use
1998-2000
551/462
Never
234
1.0
20-74
Random digit
dialling (<65 yrs),
Medicare
eligibility files (≥65
yrs) frequency
matched on age,
sex, ethnic group
geographic region
Ever
317
1.0 (0.7-1.3)
1-2 years
26
0.9 (0.5-1.9)
3-5 years
32
0.5 (0.3-0.9)
6-8 years
15
1.4 (0.5-3.6)
>8 years
20
1.6 (0.7-3.8)
≤78 h
35
0.8 (0.4-1.4)
79-208 h
23
0.8 (0.4-1.5)
≥209 h
35
1.1 (0.6-2.1)
Total years of use
Unconditional logistic
regression, adjusted for
age, geographic area,
gender, ethnic group,
education, marital status
and family history of any
lymphoproliferative
malignancy among firstdegree relatives.
Linet et al. (2006)
Standardized incidence
ratios, by age and sex.
Johansen et al.
(2001)
Cumulative hours of use
Denmark
Cohort study
Ever subscriber, men
1982-1995
420 095 mobile
phone
subscribers
Ever subscriber, women
>18
Comparison
group: national
incidence rates
109
0.93 (0.77-1.13)
11
1.04 (0.52-1.86)
All subscribers included in
calculations of national
incidence rates, but
constitute a small
proportion of the
population.
50
UK
Cohort study
1999-2005 –
followed
through 2009
791 710 women
participating in the
UK Million
Mean age 60.3 Women Study,
who answered a
(SD 5.1)
base-line
questionnaire
1999-2005
Ever mobile phone user
No
874
Yes
1.0
1184
0.97 (0.88-1.06)
Daily use
134
0.94 (0.78-1.13)
Duration of use ≥10 years
176
0.99 (0.83-1.17)
Prospectively collected
self-reported information
on mobile phone use.
Benson et al.
(2013a)
Definition of “user” did not
require a minimal amount
of use.
Cox proportional hazards
models adjusted for age,
area based SES,
geographical region,
height, BMI, smoking,
alcohol, strenuous
exercise, menopausal
hormone therapy.
2058 cases of
non-Hodgkin
lymphoma
Uveal melanoma and other cancers in the eye
Germany
Case-control
2002-2004
459/827, 187, 180 Never
20
1.0
20-74
3 control groups:
1. population
based controls
selected from
censuses
matched by age,
sex, region of
residence
Sporadic
44
0.9 (0.7-1.3)
Regular
36
0.7 (0.5-1.0)
≤4 years
19
0.8 (0.5-1.2)
5-9 years
14
0.6 (0.4-1.0)
≥10 years
3
0.6 (0.3-1.4)
2. sibling controls
who were within
10 years of the
case’s age
Cumulative hours of use
15
0.6 (0.4-1.0)
8
0.9 (0.5-1.5)
12
0.8 (0.5-1.3)
≤1176
14
0.8 (0.5-1.2)
>1176-4350
10
0.6 (0.3-1.0)
>4350
11
0.8 (0.5-1.3)
38
0.94 (0.66-1.29)
6
1.10 (0.40-2.39)
3. controls among
ophthalmology
patients (only
during the first
half of the study
period)
Mobile phone use
Time since first regular use
≤44 h, regular use
>44-195 h, regular use
>195 h, regular use
Cohort study
Ever subscriber, men
1982-2002
420 095 mobile
phone
subscribers
Ever subscriber, women
Cohort study
1999-2005 –
followed
through 2009
791 710 women
participating in the
UK Million
Mean age 60.3 Women Study,
who answered a
(SD 5.1)
base-line
questionnaire
1999-2005
87 cases of eye
cancer
Conditional logistic
regression accounting for
the matching factors.
Results shown are when
population based controls
are used.
Standardized incidence
ratios, by age and sex.
Vrijheid et al.
(2006)
~200 000 corporate
subscribers included in
comparison group,
constitute less than 5% of
unexposed.
Comparison
group: the whole
Danish population
who were not
subscribers
UK
Stang et al.
(2009)
Cumulative no. calls
Denmark
≥18
Interphone questionnaire
was used to collect
exposure information.
Ever mobile phone user
No
35
1.0
Yes
52
1.01 (0.64-1.60)
Daily use
5
0.75 (0.29-1.97)
Duration of use ≥10 years
5
0.82 (0.31-2.19)
Prospectively collected
self-reported information
on mobile phone use.
Benson et al.
(2013a)
Definition of “user” did not
require a minimal amount
of use.
Cox proportional hazards
models adjusted for age,
area based SES,
geographical region,
height, BMI, smoking,
alcohol, strenuous
exercise, menopausal
hormone therapy.
Testicular cancer
51
Sweden
Case-control
Mobile phone use, analogue
1999-2002
888/870
Never any type of wireless
20-75
Population
Ever
registry, matched Time since first use
on age
>1-5 year
515
1.0
175
1.0 (0.8-1.3)
99
0.9 (0.6-1.2)
>5-10 years
62
1.2 (0.8-1.8)
>10 years
14
1.5 (0.6-3.7)
≤160 h
110
1.3 (0.9-1.7)
>160 h
65
0.7 (0.5-1.01)
164
1.1 (0.8-1.5)
154
1.1 (0.8-1.4)
10
2.8 (0.8-11)
≤182 h
96
1.3 (0.9-1.8)
>91 h
68
0.9 (0.6-1.3)
Ever
174
1.0 (0.8-1.4)
Ever subscriber, men
187
1.05 (0.96-1.15)
Unconditional logistic
regression adjusted for
age, year of diagnosis
and cryptorchidism.
Hardell et al.
(2007)
Standardized incidence
ratios, by age.
Vrijheid et al.
(2006)
Cumulative hours of use
Mobile phone use, digital
Ever
Time since first use
>1-5 year
>5-10 years
Cumulative hours of use
Cordless phone use
Denmark
Cohort study
1982-2002
357 553 male
mobile phone
subscribers
≥18
Corporate subscribers
included in comparison
group, constitute less than
5% of unexposed.
Comparison
group: the whole
male Danish
population who
were not
subscribers
Malignant melanoma and other skin cancers
Sweden
Case-control
Mobile phone use, any type
2000-2003
347/1184
Population
registry, matched
on age and sex.
Analyses included
also controls
matched to cases
with skin cancers
at other locations
Never any type of wireless
20-77
Malignant
melanoma in
the head and
neck
Ever
?
1.0
223
1.0 (0.7-1.3)
>1-5 year
85
0.9 (0.7-1.3)
>5-10 years
81
1.0 (0.7-1.5)
>10 years
57
1.0 (0.71.5)
≤128 h
120
1.0 (0.7-1.4)
>160 h
103
1.0 (0.7-1.4)
138
0.9 (0.6-1.2)
>1-5 year
80
1.1 (0.7-1.5)
>5-10 years
41
0.8 (0.5-1.2)
>10 years
17
0.6 (0.4-1.1)
≤365 h
64
0.9 (0.6-1.2)
>365 h
74
0.9 (0.6-1.4)
Unconditional logistic
regression adjusted for
age, year of diagnosis.
Hardell et al.
(2011)
Time since first use
Cumulative hours of use
Cordless phone use
Ever
Time since first use
Cumulative hours of use
52
Denmark
358 403 mobile
phone
subscribers
1990-2007
≥30
All Danes born in
Denmark in 1925
or later and alive
in 1990, 3.21
million persons
included in the
CANULI cohort
Basal cell carcinoma
Head and neck
Ever subscriber, men
1725
0.98 (0.93-1.03)
275
0.93 (0.82-1.05)
1-4 years
349
1.01 (0.91-1.13)
5-9 years
647
0.96 (0.89-1.04)
10-12 years
455
0.96 (0.87-1.05)
≥13 years
274
1.02 (0.90-1.15)
1-4 years
67
1.02 (0.80-1.30)
5-9 years
99
0.78 (0.64-0.95)
10-12 years
87
1.02 (0.83-1.26)
≥13 years
22
1.20 (0.79, 1.82)
234
1.01 (0.88-1.16)
13
0.85 (0.49-1.47)
1-4 years
34
0.86 (0.61-1.21)
5-9 years
84
1.01 (0.81-1.26)
10-12 years
79
1.17 (0.93-1.48)
≥13 years
37
0.91 (0.66-1.27)
65
1.05 (0.80-1.37)
6
0.76 (0.34-1.72)
1-4 years
16
1.16 (0.69-1.94)
5-9 years
22
1.01 (0.65-1.57)
10-12 years
16
0.92 (0.55-1.54)
≥13 years
11
1.20 (0.65-2.22)
Ever subscriber, women
Time since first subscription
Men
Time since first subscription
Log linear Poisson
Poulsen et al.
regression models,
(2013)
adjusted for age, calendar
year, education, income.
Exposed were
subscribers between
1987-1995. Nonsubscribers during the
same period comparison
group. Corporate
subscribers could not be
identified, constitute a
small proportion of the
unexposed population.
Women
Squamous cell carcinoma
Head and neck
Ever subscriber, men
Ever subscriber, women
Time since first subscription
Men
Malignant melanoma
Head and neck
Ever subscriber, men
Ever subscriber, women
Time since first subscription
Men
UK
Cohort study
1999-2005 –
followed
through 2009
791 710 women
participating in the
UK Million
Mean age 60.3 Women Study,
who answered a
(SD 5.1)
base-line
questionnaire
1999-2005
Ever mobile phone user
No
780
Yes
1.0
1336
1.06 (0.96-1.16)
Daily use
160
1.06 (0.89-1.26)
Duration of use ≥10 years
191
1.09 (0.92-1.29)
2116 cases of
melanoma
Prospectively collected
self-reported information
on mobile phone use.
Benson et al.
(2013a)
Definition of “user” did not
require a minimal amount
of use.
Cox proportional hazards
models adjusted for age,
area based SES,
geographical region,
height, BMI, smoking,
alcohol, strenuous
exercise, menopausal
hormone therapy.
1644
1645
12.1.3
Occupational RF exposure and cancer risk
1646
12.1.3.1
Introduction
1647
1648
1649
Information on cancer risks in relation to occupational RF exposure comes from three types of
epidemiological study designs: cohort studies, investigating a wide range of cancer (and non-cancer) outcomes
in groups with potential RF exposure followed up over time; case-control studies of specific cancer sites,
53
1650
1651
1652
1653
1654
1655
1656
1657
investigating history of occupational RF among other exposures; and analyses of routinely collected datasets on
cancer incidence or mortality, in which risks of cancer have been assessed in relation to job title. The most
extensive literature addresses brain tumours and leukaemia. Most studies published before 1993 and therefore
reviewed in the previous WHO EHC report (WHO, 1993b) included only a few exposed cases each. Two cohort
studies with somewhat larger numbers are noted: Milham (1988) followed up a large cohort of male amateur
radio operators in the USA and observed significantly raised mortality from certain types of leukaemia and
Robinette et al. (1980) found no difference in cancer mortality between US naval personnel occupationally
exposed to radar and those who were not. The following sections focus on studies published after 1992.
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
General shortcomings of published studies investigating occupational RF exposure in relation to
cancer risk are that exposures have been assessed retrospectively, and are often only based on likelihood of
exposure and exposure intensity rather than actual measurements. Many studies investigate particular
occupational groups, where RF exposure may come with the job, such as policemen, but there is no information
available on actual individual exposures. Such studies often lack a comparable unexposed control group and
compare incidence or mortality rates in the study population against the general population, but employed group
of subjects are often in better health than the general population overall (‘healthy worker effect’) or are
dissimilar in terms of socioeconomic status or lifestyle factors. The retrospective nature of the studies has the
consequence that information on other risk factors that could confound the association under investigation is
often not available. Most cohort studies had included small numbers of cancer cases by type of cancer, therefore
reducing the ability of the study to detect associations with risk. We discuss here results from cohort studies and
case-control studies separately, with results for main cancer types being summarised in Tables 1-7.
1670
1671
1672
1673
1674
1675
The literature search was focused to capture environmental and occupational sources of
radiofrequency electromagnetic fields and neoplasms. We identified 598 articles, 541 of which were excluded
because they were not relevant. Among the 57 articles that were judged as potentially eligible, 34 were excluded
because they were case series, reviews or letters to the editor with no new data and the remaining 23 were
included. We manually retrieved a further 31 articles through examination of cited papers and expert reports,
totalling 54 included articles in sections 12.1.3 and 12.1.4.
1676
12.1.3.2
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
Two populations of male electrical utility workers, one in Québec, Canada (Hydro-Québec) and one
in France (Électricité de France), were followed up for malignancies for the period 1970–1988 and 1978–1989,
respectively (Armstrong et al., 1994). Nested-case-control analyses of the 2679 cases of cancer identified during
follow-up were conducted in relation to pulsed EMF (PEMFs). Controls were chosen at random from risk sets
of cohort members surviving to the date of diagnosis of the case, and matched on cohort and year of birth. Four
controls were chosen for cases with haematological or brain cancers or melanoma, whereas one control was
chosen for other cancer sites. Exposures to pulsed electromagnetic fields were assessed through a job-exposure
matrix estimated separately for the two cohorts from measurement of 829 French and 466 Canadian workers
made in 1991–1992. Each of these workers wore a positron meter for a week monitoring the proportion of time
during which the electric field was more than 200 V/m in the 5–20 MHz range, measured in parts per billion.
Analyses were based on conditional logistic regression with case-control sets and further adjustment for
socioeconomic status, cumulative exposure to chemicals, and total duration of employment. No association was
observed with exposure to pulsed electromagnetic fields for leukaemia, other haematopoietic cancers, brain
cancer or melanoma, or at other sites, except for a significant excess of lung cancer. There was evidence for a
dose-response relationship with increasing cumulative exposure, calculated as the proportion of time greater
than 100 parts per billion (OR=1.27, 95% CI: 0.96–1.68 for ≥median and OR=3.11, 95% CI: 1.60–6.04 for
≥90th percentile. The association was limited to workers at Hydro-Québec and not present among the lesser
exposed workers at Électricité de France. The median value corresponded to 21.3% at Hydro-Québec and 2% at
Électricité de France, thus exposure levels were higher at the former. Adjustment for crude indicators of
smoking and other factors left the results little changed. The relative risk for ≥90th percentile was greatest for
exposures more than 20 years before diagnosis. [This finding could suggest causality or could be due to chance
because the many cancer sites analysed could have resulted in spurious significant associations. No corrections
for multiple comparisons were made.]
1700
1701
1702
1703
1704
Cantor et al. investigated mortality records for 1984–1989 from 24 states in the USA including
information on occupation and industry in order to identify workplace exposures potentially related to female
breast cancer risk (Cantor et al., 1995). Using a case-control approach and after excluding housewives, 33 509
female subjects who had died from breast cancer and 117 794 controls were randomly selected among noncancer deaths by strata of age and ethnicity. A job exposure matrix was developed on the basis of professional
Cohort studies of occupationally exposed populations
54
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
judgment of an industrial hygienist, supported by the general occupational hygiene literature and two databases,
the NIOSH’s Job Exposure Matrix and the Integrated Management Information System of the Occupational
Safety and Health Administration. This was used to estimate the probability and level of 31 workplace
exposures, categorized as 0–4 and 0–3, respectively, with 0 being non-exposed. Analyses were conducted
separately for women of white and black ethnicity and were conducted for each of the 31 exposures, including
RF exposure. Among females of white ethnicity the odds ratio of a breast cancer death in relation to RF
exposure was raised for exposure probability group 3 (OR=1.15, 95% CI: 1.1–1.2), but not for group 4
(OR=0.99, 95% CI: 0.8–1.2), after adjusting for socioeconomic status and age. Odds ratios by exposure level
were significantly raised for category 1 and 3 but not for category 2. Among black females, odds ratios were
significantly raised in exposure probability categories 1 and 3, but not 2 and 4, and in exposure level categories
1 and 3. [This study did not have any information on any of the well-established risk factors for breast cancer. In
the absence of information on which particular jobs were associated with RF exposure, it is impossible to
evaluate whether they would attract women with a particular risk profile regarding to breast cancer, such as, for
example, jobs involving night shifts being more common among women without children, in the light of
nulliparity increasing breast cancer risk. The study was large so that associations of moderate size can reach
statistical significance, and many analyses were conducted increasing the probability of observing statistically
significant associations. No adjustment was made for multiple comparisons. Breast cancer was weakly
associated with certain categories of likelihood and level of RF exposure, but without a clear dose-response. It is
therefore likely that these findings were by chance or due to uncontrolled confounding.]
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
Cancer risks in relation to RF, ELF and light at night were investigated in a cohort of 2619
Norwegian female radio and telegraph operators, certified between 1920 and 1980, working at sea (Tynes et al.,
1996). Such subjects had potential exposure to light at night, RF (405 kHz–25 MHz) and to ELF (50 Hz). Light
exposure at night is thought to decrease melatonin production by the pineal gland and it is hypothesised that this
might increase susceptibility to sex-hormone-related cancers. Information on incident cancer was obtained by
linking to the Cancer Registry of Norway for the period 1961–1991. The observed number of cancers was
compared against those expected in the Norwegian female population standardised for age and calendar year.
There were 140 observed cancers over the follow-up time, with a SIR of 1.2 (95% CI: 1.0–1.4). Among those,
50 were of breast cancer, with a significantly raised SIR of 1.5 (95% CI: 1.1–2.0) and 12 of uterine cancer with
a SIR of 1.9 (95% CI: 1.0–3.2). Relative risks were not appreciably raised for leukaemia (SIR=1.1, 95% CI:
0.1–4.1), lymphoma (SIR=1.3, 95% CI: 0.4–2.9) or brain cancer (1.0, 95% CI 0.3–2.3). In a nested case-control
analysis of 50 breast cancer cases with four to seven matched controls drawn from the cohort, detailed job
histories were collected from the Norwegian Seamen Registry. Controls were selected among certified radio and
telegraph operators alive at the time of diagnosis and matched on year of birth. Duration of employment and
shift work was not associated with breast cancer risk diagnosed under age 50, but there were statistically
significant exposure-response relationships for both when restricted to ages at diagnosis of 50 years and over.
When shift work was adjusted for duration of employment and vice versa, the effect of shift work was stronger
than that of duration of employment, and tests for trend were no longer significant.
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
Szmigielski (1996) examined cancer incidence in the whole population of military career personnel in
Poland over the period 1971–1985 and provided an update in 2001 extending the follow-up period with five
years to 1990 (Szmigielski, Sobiczewska & Kubacki, 2001; 2006). Only the second paper is discussed here,
because it is based on a longer time period and because it includes important information needed to interpret the
original study. The studied population was that of all career personnel in the Polish army in the age range of 25–
59 years during active service. The total population size is not disclosed but the population size averaged
124 500 subjects per year. Cancer diagnoses were ascertained from military hospital records and the central
military medical board, but it is unclear whether subjects who leave the services remain to be followed up for
cancer. Radiofrequency and microwave exposure assessment was based on military safety, service personnel
and health department records for all subjects, and for those who developed cancer also on records from hospital
and central medical board data. For the period 1985–1990, case ascertainment was done prospectively, whereas
for 1971–1985 this was conducted retrospectively. For the period 1985–1990, additional information on RF was
also obtained from questionnaires given to personnel who had been newly diagnosed with a neoplasm. The
researchers assessed all service posts with RF/MW exposure for exposure levels in this period, allowing
classification into four exposure categories (1–2, 2–6, 6–10, ≥10 W/m2). The annual average number of exposed
personnel was 3860. Annual cancer incidence rates in unexposed and exposed populations were calculated by
dividing the number of cancers by the population size in that year. The ratio of average annual rates between
exposed and unexposed personnel was then taken as the relative risk. Results were provided overall, in 10-year
age categories and for 12 cancer sites. The authors reported 138 neoplasms in exposed and 2355 in unexposed
personnel, with a relative risk of 1.83 (p<0.01). Raised relative risks were reported for several categories
including haematological and lymphatic cancers, cancers of the skin, nervous system, oesophagus and stomach,
55
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
colon and rectum. Relative risks were greater than 3-fold for various subcategories of lymphatic cancer and
leukaemia, and were raised in all subgroups of age. It can be derived that leukaemia risk overall was more than
6-fold increased. About 85% of service men were deemed to have average exposure levels not exceeding
2 W/m2, with the remainder around 2–6 W/m2; higher intensities were recorded incidentally. Fields were mostly
pulse-modulated RF/MW at 150–3500 MHz. Cancer morbidity rates were over two-fold greater in subjects with
average levels of 6–10 or >10 W/m2 compared with those 1–2 W/m2. [The description of the study population
remains unclear. It is, for example, not stated whether women were included and whether there was adjustment
in the results for sex. Statistical methods employed are poorly described and unconventional for cohort studies.
Cancer occurrence appears to have been assessed cross-sectionally by year, instead of subjects having been
followed-up longitudinally. The calculated relative risks were not weighted for changes in population sizes over
years, and did not take into account difference in age- or sex- distribution (or changes over time in these
distributions) between the two groups. Using additional sources for exposure assessment such as hospital
records and supplementary questionnaires for diagnosed cases only gives substantial potential for bias in the
direction of an overestimation of risk because there is scope for extra sources of RF exposure to be added to
cases, but not to cohort members who did not develop cancer. The observation of a significantly raised risk for
all cancer sites overall supports the presence of such bias. The likely biases in this study, the unsatisfactory
design and statistical methods, and the unusually high observed relative risks have the consequence that this
study can be given little weight in the overall assessment.]
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
A cohort of 138 905 male workers employed in any of five electrical utility companies in the USA
between 1950 and 1986 was followed up for mortality (Savitz et al., 1997). Following the finding by Armstrong
et al. (1994) of an excess risk of lung cancer in relation to pulsed electromagnetic fields, the researchers
examined lung cancer deaths in this cohort. There were 1692 deaths from lung cancer during follow-up ending
in December 1988. Follow-up was based on the National Death Index from 1979; the earlier period was covered
through multiple sources. Job titles were grouped into 28 categories for assignment of exposure. In addition to
exposure to 60 Hz magnetic fields, which is not in the remit of this report, exposure to pulsed EMFs was
estimated by assuming that workers with similar work activities as in the Hydro-Québec component of the study
by Armstrong et al. would have the same scores. Units were derived as the proportion of workers exceeding the
100 parts per billion (ppb) threshold (see the description of Armstrong et al. (1994) above) and the duration of
work in each job. Analyses were adjusted for age, calendar time, ethnicity, social class, active vs. inactive work
status, and years of exposure to occupational lung carcinogens. Data on tobacco smoking were, however, not
available. No significantly raised rate ratios were observed for any of the categories of years in exposed
occupations, in 10 year age groups, with an exposed job defined as one in which more than 25% of the
measurements exceeded the 100 ppb time threshold. Cumulative exposure was derived as the proportion of
workers in the job group exceeding the 100 ppb time threshold and the duration of work in each job. For total
exposure, rate ratios were between 1.3 and 1.4 in the above background categories, with weak evidence for a
dose-response. These associations were similar when analysed by time since exposure. [No formal statistical
tests for trend were reported. The evaluation of pulsed EMFs was tenuous because the positron meter used in the
study by Armstrong et al. (1994) had been found to respond to much higher frequencies than intended
(including mobile phone frequencies), and because the current study assumed exposure levels to be the same as
in that study. The uncertainties in exposure assessment and the weak associations observed either suggests that
there is an association with pulsed EMF but errors in exposure assessment have attenuated the rate ratios or that
there is no association, with residual association due to confounding (e.g. cigarette smoking) or biases.]
1805
1806
1807
1808
1809
1810
1811
1812
1813
A small cohort of 481 female and 201 male plastic-ware workers exposed to RF EMF generated by
dielectric heat sealers in Italy was followed up for mortality over 1962–1992 (Lagorio et al., 1997). Study
subjects were derived from a single manufacturing plant and could be categorised into RF-sealer operators,
other labourers, and white-collar workers. There were 15 deaths among females, and one among males, during
the study period, ascertained through the Registry Office of the municipalities of residence and death. The SMR
for all cancer mortality among female RF-sealer operators was slightly elevated (SMR=2.0, 95% CI: 0.7–4.3),
but with large imprecision due to small numbers of deaths. [Analyses were conducted by specific cause of death
with only single observed deaths in individual categories, which could therefore not be meaningful. There was
also one death among all males overall, too few to conduct analyses.]
1814
1815
1816
1817
1818
1819
Finkelstein et al. retrospectively assessed cancer incidence in a cohort of 22 197 police officers in
Ontario, Canada, due to concerns about health effects from police radar (Finkelstein, 1998). The cohort was
assembled from lists of officers and retirees employed since 1970, or a later date if no earlier records were
available, from 83 participating Ontario police departments. There was no information on individual exposure to
radar. Information on cancer and mortality was obtained by probabilistic linkage to the Ontario Cancer Registry
and the Ontario Mortality Database. Follow-up started on the date of employment or at which complete cohort
56
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
identification was possible at the specific department, whichever was later, and ended at the end of 1995.
Among male police officers, there was no increased risk of cancer overall (age and calendar-year adjusted
SIR=0.90, 95% CI: 0.83–0.98) compared with the population of Ontario, or for cancer overall excluding lung
cancer (SIR=0.96, 95% CI: 0.88–1.05), for which the SIR was particularly low (SIR for lung cancer=0.66, 95%
CI: 0.52–0.82). Risk was elevated for melanoma (SIR=1.45, 95% CI: 1.10–1.88), and non-significantly so for
testicular cancer (SIR=1.30, 95% CI: 0.89–1.84) and prostate cancer (SIR=1.16, 95% CI: 0.93–1.43). There
were 16 cases of incident brain tumours, with no increased risk compared with the population (SIR=0.84, 95%
CI: 0.48–1.36). When follow-up was evaluated from start of employment for all officers, the SIR for testicular
cancer was borderline statistically significant (SIR=1.33, 95% CI: 1.0–1.74). Among female officers, only 11
tumours were observed (SIR=0.77) and no further analyses were reported. [The lack of exposure assessment and
of an internal unexposed control group necessitates the use of general population rates to derive estimates of
relative risk in this study. Comparison of cancer rates in this occupational group against the general population
rates is, however, problematic, in particular for cancers which are strongly associated with lifestyle, because the
group of interest might be different with regard to lifestyle and other confounding factors. The strong reduction
in risk of lung cancer rate suggests lower cigarette smoking prevalence and associated with this, a higher
socioeconomic profile of police officers compared with the general population. Both testicular cancer and
melanoma risks are positively associated with socioeconomic status, which could have contributed to the
findings related to these neoplasms. The raised risk of melanoma might be due to increased sun light exposure
due to holidays or certain characteristics of their profession. Analyses of follow-up from start of employment
during years that employment records were not complete might have introduced bias due to those not on the
records being different in relation to cancer risk.]
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
A large cohort study of employees of Motorola, a manufacturer of wireless communication products,
was followed up for mortality in the United States (Morgan et al., 2000). Workers’ exposures occurred at
frequencies of 30, 150 and 450 MHz for two-way radios (from 1960s), 800 MHz for mobile phones and higher
frequencies for microwave communications. The cohort comprised 195 775 employees who had had at least 6
months of cumulative employment, contributing 2.7 million person-years over a period of 20 years (1976–
1996). Information on employee’s job code and start and end dates of employment were extracted from the
company’s electronic personnel files, and on vital status was derived from the US Social Security
Administration’s Master Mortality File and the US National Death Index. No data existed from personal
exposure monitoring, but relative exposures were classified into ‘background’, ‘low’, ‘moderate’ and ‘high’
based on business sectors, work site, job code and descriptions, calendar period and expert assessments. Expert
assessments were conducted ‘blinded’ to disease status of the individual. There were 6296 deaths, and mortality
rates compared to those of Arizona, Florida, Illinois and Texas, where most of the company’s facilities were
located, were significantly reduced for all causes (SMR=0.66, 95% CI: 0.64–0.67) and for cancer specifically
(SMR=0.78, 95% CI: 0.75–0.82), after adjusting for age, sex and ethnicity. Likewise, SMRs for individual
cancer types such as lymphatic or haematopoietic cancer and central nervous system cancer was reduced in the
entire cohort as well as in the group of employees considered to have had moderate or high RF exposure.
Further analyses showed that SMRs for these causes did not increase with increased duration of employment. To
reduce bias due to the ‘healthy worker effect’, internal cohort comparisons to estimate relative risks for usual,
peak and cumulative RF exposure scores were conducted with adjustment for age, sex and calendar period of
employment, which showed rate ratios near or below 1.0 for brain cancers, lymphatic and haematopoietic cancer
and non-Hodgkin’s lymphoma. There were no trends of relative risk with exposure duration or with latency
assumptions (5, 10, 20 years).
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
Groves et al. (2002) reported an update of the cohort study by Robinette et al. (1980) of mortality in
40 581 male US Navy veterans of the Korean war who had graduated into the Navy during 1950–1954 and had
potential exposure to high intensity radar. Radar or radio operator and aviation electrician's mate jobs were
classified as ‘low radar exposure potential' (usually well below 10 W/m2) and aviation electronic technician,
electronics technician and fire control technician as ‘high radar exposure potential’ (potential to exceed 1 kW/m2
but usual exposures less than 10 W/m2). The cohort was followed up to 1997 with the use of beneficiary records
of the Department of Veterans and matching against the US Social Security Administration’s Death Master file
and the US National Death Index for 1979–1997. There were 8393 deaths during the follow-up period.
Compared against white males in the USA, standardised mortality ratios for overall mortality were reduced for
the cohort overall (SMR=0.74, 95% CI: 0.73–0.76 for the entire cohort) and for the high (SMR=0.69, 95% CI:
0.67–0.71) and low exposure (SMR=0.80, 95% CI: 0.78–0.82) cohort separately, after controlling for age at
entry and attained age. Likewise, SMRs were reduced in all groups for cancer-specific mortality and for the high
radar exposure potential group for brain cancer (SMR=0.71, 95% CI; 0.51–0.98). There were no significantly
raised mortality rates for specific types of cancer. There was a tendency for deaths from all cancers, lung cancer,
lymphoid malignancies and brain cancer to be less common in the high exposure group than in the low exposed
57
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
group. For leukaemia deaths, the SMR was 1.14 (95% CI: 0.90–1.44) in the high radar exposure potential group,
0.77 (95% CI: 0.58–1.04) in the low exposure potential group and 0.96 (95% CI: 0.80–1.16) in the overall
cohort. When the high and low exposure groups were compared, there was an excess risk of death in the high
exposure group from leukaemia (RR=1.48, 95% CI: 1.01–2.17) and non-lymphocytic leukaemia specifically
(RR=1.82, 95% CI: 1.05–3.14). Relative risks for all leukaemia compared with the low exposure group were
statistically significantly raised for highly exposed aviation electronic technicians (RR=2.60, 95% CI: 1.53–
4.43), but not for highly exposed electronics technicians (RR=1.30, 95% CI: 0.83-2.03) or fire control
technicians (RR=1.04, 95% CI: 0.48-2.27) or low exposure aviation electrician mates (RR=1.12, 95% CI: 0.393.19). [This cohort study had the strengths of being large with long duration of follow-up, and relatively good
exposure assessment, but had the limitation that only mortality and not incidence was investigated, and there
was no information on confounders. Mortality rates were generally considerably lower in this cohort compared
with the general population, a likely manifestation of the selection of physically fit subjects into the navy.
Internal comparisons, however, showed that excess leukaemia mortality was particularly evident in aviation
electronics technicians, but it is not clear why this raised risk was not present in other occupations with similarly
high exposure levels.]
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
In a retrospective cohort study, mortality was compared between 4417 Belgian men who served in
anti-aircraft radar units in Germany at some time during 1963–1994 and 2932 Belgian men who worked in
military units that were not equipped with radar in the same region (Degrave et al., 2009). Modelled exposure
levels for typical job locations ranged between 10–500 V/m. There were 424 deaths in the exposed and 271
deaths in the unexposed group over the period 1968–2004. There was no significant difference in overall
mortality between the two groups, nor a trend with duration of service in exposed jobs, after adjusting for age.
Cause-specific mortality was raised for neoplasms (RR=1.23, 95% CI: 1.03–1.47), and specifically lymphatic
and haematopoietic neoplasms (RR=7.22, 95% CI: 1.09–47.9) and eye, brain and other nervous system
neoplasms (RR=2.71, 95% CI: 0.42–17.49). Comparison against national mortality statistics rather than the
control group showed decreased standardised mortality ratios for cancer, for both the exposed and unexposed
groups. There was a non-significant increase of cancer mortality compared with the control group with length of
service in radar battalions, but cancer mortality was not greater in subjects who had served before shielding of
microwave generators was introduced (late 1970s). Cause of death could only be established from the national
death registry for 71% of exposed and 70% of control subjects, and had to be obtained from reports from
relatives of the deceased for other deaths, and remained unknown for about 10 percent of subjects. [This
incompleteness of cause of death ascertainment leaves open the potential for bias. The radar had an average
radiated power of 1.5 kW and peak power in pulses of about 500 kW, and also emitted ionising radiation, and
there is therefore the possibility that the raised leukaemia incidence was due to the latter.]
1911
12.1.3.3
1912
12.1.3.3.1 Brain and other intracranial tumours
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
Grayson (1996) investigated risk of primary malignant brain tumour in relation to a range of EMF
exposures in a case-control analysis nested within a cohort composed of all male members of the US Air Force
who had completed at least one full year of service between 1 January 1970 and 31 December 1989. Incident
cases were identified through screening of hospital discharge records of individuals who were serving in the US
Air Force during the study period. Four controls per case were selected, exactly matched on year of birth and
ethnicity, and being in the Force at the time of the case’s diagnosis. A job-exposure matrix was developed for
exposure assessment, based on job titles, including allowance for changes in intensity scores over time. Job titles
were classified as having ‘no’, ‘possible’ and ‘probable’ potential for exposure to RF/MW frequencies. The
‘probable’ category was assigned to occupations that had been reported to have been overexposed in the past,
essentially all occupations involved in maintenance and repair of RF/MW transmitters. The ‘possible’ category
was assigned to job categories involving the operation of such transmitters for which excessive exposures had
not been reported. All other job categories were considered unexposed. The analyses included 230 cases and
920 matched controls. Odds ratios were non-significantly raised for ever having had potential exposure to ELF
(OR=1.28, 95% CI: 0.95–1.74) and significantly raised for ever exposure to RF/microwave (OR=1.39, 95% CI:
1.01–1.90) after adjusting for age, ethnicity and senior military rank. There was no clear evidence for dose
response according to categories of product of exposure intensity score and duration in the specific job,
however. No association was reported for ionising radiation exposure, but numbers of exposed subjects were
much lower than for non-ionising radiation. [There was no attempt to identify cases from subjects who had left
the Air Force, which is a drawback especially for brain tumours which may take long to develop.]
Case-control studies
58
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
Berg et al. (2006) investigated the association between risk of glioma or meningioma and
occupational RF field exposure based on data from the German Interphone study. Cases diagnosed aged 30 to
59 years (October 2000–September 2001) or aged 30 to 69 years (October 2001–October 2003) were recruited
from the four neurosurgical clinics in the study regions of Bielefeld, Heidelberg/Mannheim and Mainz. Controls
were selected from compulsory population registers in the three regions matched to cases on age, sex and centre.
Participation rates were 80% for glioma, 88% for meningioma, and 63% for controls. Information on past
exposures was collected by personal computer-assisted interview, which included questions on occupational
activities related to RF/MW fields such as use of industrial heating equipment, use of radar, manufacturing
semiconductor chips or microelectronic devices and work with broadcasting/telecommunication antennae and
masts. Job activities were grouped into ‘no exposure’, ‘not probable’, ‘probable’, ‘high’ RF fields exposure
based on an activity-exposure matrix. Analyses were adjusted for region, sex, socioeconomic status,
urban/rural, age at diagnosis, cigarette smoking status and ionising radiation exposure. Those with highly
probable exposure had an odds ratio of glioma of 1.22 (95% CI: 0.69–2.15) overall and among those, subjects
with more than 10 years of exposure had an odds ratio of 1.39 (0.67–2.88). For meningioma, these odds ratios
were 1.34 (95% CI: 0.61–2.96) and 1.55 (95% CI: 0.52–4.62), respectively, the latter based on 6 exposed cases.
There was no association of glioma (OR=1.04, 95% CI: 0.68–1.61) or meningioma risk (OR=1.12, 95% CI:
0.66–1.87) in relation to the combined groups of probable and high potential RF/MW exposure vs. none or notprobable exposure.
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
Karipidis et al. (2007a) conducted a case-control study of glioma in relation to occupational exposure
to ionising and non-ionising radiation in Melbourne, Australia. Histologically confirmed cases first diagnosed
1987–1991 at ages 20–70 years were recruited from 14 Melbourne hospitals, with controls selected from
electoral roll in the study area matched by age, sex and postcode of residence. Exposure assessment was based
on a detailed job history obtained from each subject by self-administered questionnaire followed by face-to-face
interview, and included self-reported RF and ionising radiation exposure. This information was evaluated
against a Finnish job exposure matrix (FINJEM) and job histories were also evaluated by expert review to
evaluate the potential for exposure. The study included 416 cases and 422 controls, accounting for 66% of cases
and 65% of controls who were eligible and with whom contact was made. For 44% of cases and 2% of controls
a next of kin was interviewed instead of the subject due to death or disability. Only eighteen cases and 17
controls had been occupationally exposed to RF, jobs with the highest numbers being plastic product workers,
telephone installation crew, line men and cable jointers, physiotherapists and woodworking occupations. There
was no trend of odds ratios in relation to cumulative exposure to RF, adjusted for sex, age and years of
education, with tertile of exposure (OR=0.57, 95% CI: 0.16–1.96, OR=1.80, 95% CI: 0.53–6.13, and OR=0.89,
95% CI: 0.28–2.81 for exposed tertiles 1, 2 and 3 versus no exposure, respectively). Odds ratios of risk in
relation to number of years of exposure were not significantly raised for any category, whether exposures were
based on self-report or on expert assessment. [Participation rates were not particularly high in this study, leaving
open the possibility for selection bias and additionally, for a high proportion of cases a proxy was interviewed
which could have under or overestimated reported exposures.]
1969
1970
1971
1972
Spinelli et al. (2010), in their case-control study of glioma (discussed in Section 12.1.2.1), also
investigated risk in relation to occupational sources of RF, among many other residential and occupational
exposures, but observed no association with microwave frequency field exposure, based on 7 and 4 exposed
cases and controls respectively (OR=1.20, 95% CI: 0.30–4.77).
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
A case-control study in Gironde, France, of brain and other intracranial tumours in relation to EMF
exposure was reported by Baldi et al. (2011). The study included 105 gliomas, 67 meningiomas and 23
neurinomas. The authors state these comprised cases diagnosed between May 1999 and April 2001 in the
Gironde area, but the source of case ascertainment and its completeness is not stated. Controls were randomly
selected from the local electoral rolls, with two controls per case individually matched on age, sex and area.
Participation rates were 70% for cases and 69% for controls. Job histories were collected by personal interview
for every job lasting more than six months, which was assessed on type of EMF, exposure duration and
exposure probability. Furthermore, in a second step, expert judgment was compared against a Swedish Job
Exposure Matrix, prioritising expert judgment in case of disagreement. Only 7 cases and 16 controls were
occupationally exposed to RF, with a relative risk of 1.50 (95% CI: 0.48–4.70) for all intracranial tumours. For
glioma specifically, the relative risk was 1.44 (95% CI: 0.50–4.13), with numbers being too small for other
tumour types.
59
1985
Studies with uncertainties related to inclusion criteria
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
A case-control study in the USA (Warren et al., 2003) examined risks of intratemporal facial nerve
tumours in relation to mobile phone use and occupational RF exposures (previously discussed in Section
12.1.2.1). The study was small and was not population-based. It included 51 cases of acoustic neuroma and 18
of intratemporal facial nerve tumours, with cases obtained from the fiscal database at a tertiary care centre.
Controls were selected from the same database matched on age, sex and ethnicity, and were distributed in three
groups: those with (1) rhinosinusitis (2) dysphonia or gastroesophageal disease (3) acoustic neuroma, the latter
as an intermediate-exposure comparison group. Risk factor information was collected by telephone interview.
No associations were observed with occupational exposures to RF. [Response rates are not stated and it is
therefore not clear how many subjects were excluded for various reasons. The study is given little weight in the
overall assessment, and is not included in the table.]
1996
12.1.3.3.2 Ocular melanoma
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
Holly et al. conducted a case-control study to investigate risk factors for intraocular (uveal)
melanoma (Holly et al., 1996). They recruited patients diagnosed with uveal melanoma at the Ocular Oncology
Unit at the University of California San Francisco, or patients referred to the centre for treatment between
January 1978 and February 1987. White males between 20–74 years of age at diagnosis and resident in 11
specified states were considered eligible for the study. A total of 221 men were interviewed, corresponding to a
participation rate of 93%. All cases were histopathologically confirmed. Two male white controls per case were
recruited matched on geographic area and age group by random digit dialing methods, with a participation rate
of 74%. The questionnaire was administered by telephone by trained interviewers and included demographic
characteristics, occupational history and information about chemical exposures. Information about six
occupations that subjects had held the longest was coded to the Alphabetical Index of Industries and
Occupations and analyses were conducted according to groups of similar occupations. Information on ever
having worked with or having had regular exposure to chemicals, defined as at least 3 hours a week for at least 6
months, was also obtained. Analyses were conducted by occupational group, chemical exposures, and ever
exposure to microwaves or radar (excluding home microwave ovens). It is not stated how information about
microwaves or radar had been obtained but it can be implied it was added to the questions about chemical
exposures. There were 21 cases and 22 controls who had ever had exposure to microwave or radar,
corresponding to an odds ratio of 2.1 (95% CI: 1.1–4.0). Analyses were adjusted for age, number of large naevi,
eye colour, and tanning or burning response to half hour sun exposure in the summer. [The fact that this odds
ratio related to ‘ever’ exposure, with no further data to investigate dose-response, the small numbers of subjects
and the lack of information about how this information was collected do not give confidence in the results.]
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
Stang et al. (2001) examined risk of ocular melanoma in relation to occupational exposures to various
sources of electromagnetic fields in Germany. The 118 cases were obtained from two sources: 37 from a
population-based study (84% participation rate) and 81 from a single hospital (88% participation rate). In total,
475 controls were included: 327 population-based controls chosen at random from a population register (48%
participation rate) and 148 hospital-based controls (79% participation rate) with ocular disease, matched on sex,
age and region of residence. Participants were interviewed about a variety of lifestyle and medical factors,
occupational history and sources of EMF. They reported a raised risk of ocular melanoma in subjects with selfreported exposure for at least 6 months for several hours per day to radio sets (OR=3.3, 95% CI: 1.2–9.2), which
was stronger for the population-based than for the hospital-based study (OR=4.6, 95% CI: 1.0–19.2 and 1.8,
95% CI: 0.3–14.3, respectively). There was no relation to duration of exposure, however, and risk was not raised
for radar exposure (OR=0.4, 95% CI: 0.0–2.6). [The study was small, and combined subjects from two different
study designs. The choice of controls with ocular disease is a potential concern if such disease were associated
with EMF exposures. While population-based controls are generally considered methodological superior to
hospital-based controls, participation rates among population-based controls were low, which could have
introduced selection bias. It is not reported whether the study hypotheses were revealed to participants at the
time of invitation to the study, which could have resulted in biased participation, either by weakening or by
artificially increasing the association.]
2034
2035
2036
2037
2038
2039
A case-control study conducted in nine European countries investigated risk of ocular melanoma in
relation to occupational exposure to EMF (Behrens et al., 2010). In total, 293 incident cases of ocular melanoma
and 3198 controls aged 35–69 years, with controls being population-based or hospital-based depending on the
country, were interviewed about past jobs involving EMF including RF. Cases were diagnosed between 1994–
1997 with ascertainment being nationwide in Denmark and Latvia, from certain administrative areas in France,
Germany, Italy and Sweden, from hospital recruitment areas in Spain and Portugal, and from a single eye clinic
60
2040
2041
2042
2043
2044
2045
2046
2047
in the United Kingdom. Controls were matched by region, sex and 5-year age group. Participation rates were
82% for cases, 62% for population-based controls and 86% for hospital-based controls. Classification of
exposure to radar units was based on assumptions as published by Baumgardt-Elms et al. (2002), which used
expert opinion and measurements from unpublished expert reports. Analyses were adjusted for country and 5year age groups, sex, eye colour, skin tone, number of episodes of eye burns, alcohol consumption and
educational level. None of the female subjects had been exposed to radar, and among men, there was no
association of risk with potential exposure to radar units. [The number of exposed cases and the effect estimates
were not reported.]
2048
12.1.3.3.3 Testicular cancer
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
Hardell et al. (1998) conducted a case-control study of risk factors for testicular cancer in the Middle
and Northern parts of Sweden. Cases aged 30–75 years incident during 1989–1992 reported to the Swedish
Cancer Registry were recruited into the study. Two controls per case were selected from the Swedish population
register matched on year of birth. The study included 148 cases and 314 controls, with participation rates of
91% and 87%, respectively. Self-administered questionnaires were used to obtain information on a range of
occupations and occupational exposures. Non-significantly raised risks were observed for amateur radio
operators (OR=2.2, 95% CI: 0.7–6.6), work with radar equipment (OR=2.0, 95% CI: 0.3–14.2) and work as
engineer in the electronics and telecommunications industry (OR=2.3, 95% CI: 0.87–6.7). [These relative risks
were based on very small numbers of exposed subjects and the study did therefore lack statistical power to
detect an effect if there were indeed one.]
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
A German population-based case-control study of testicular cancer investigated risk in relation to
occupational exposures from proximity to mobile phones, radio and radar, ascertained by personal interview
(Baumgardt-Elms et al., 2002). The study area comprised five German geographic regions covering a population
of about 1.5 million male residents at ages 15–69 years. In total, 269 cases of testicular cancer incident during
1995 and 1997 were recruited through an active reporting system of clinical and pathological departments in the
study regions. In total, 797 age- and region-matched controls were randomly selected from population registers
within strata of region and five-year age group. Participation rates were 76% and 57%, respectively. Data were
collected on five categories of EMF exposure, including radiofrequency emitters and radar units in the RF
spectrum. A weighted score was calculated taking into account the duration of self-reported exposure in years
and distance to the source (≥30 m, 10–29 m, 0–9 m). Relative risk was close to unity in men who reported ever
working near radiofrequency transmitters (OR=0.9, 95% CI: 0.60–1.24) based on 50 exposed cases and 166
exposed controls. It was neither raised or reduced in those who reported to have ever worked near radars
(OR=1.0, 95% CI: 0.60–1.75), based on 22 exposed cases and 58 controls, and was non-statistically
significantly reduced in those who were assessed by experts to have had radar exposure based on their
occupational history (OR=0.4, 95% CI: 0.1–1.2). Analyses were adjusted for age and region. Analyses of risk by
tertiles of a score weighted by duration and distance did not show significantly raised associations.
[Participation rates of cases and controls were relatively low which could have affected the findings either
upwards or downwards.]
2077
12.1.3.3.3 Other malignancies
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
La Vecchia et al. (1990) conducted a hospital-based case-control study in Greater Milan, Italy, of
occupational exposures in relation to bladder cancer risk. The study included 263 cases with histologically
confirmed bladder cancer ascertained from teaching and general hospitals in the study region and 287 controls
were recruited for a wide range of acute, non-neoplastic and non-urinary tract diseases in the same hospitals. It
was reported that less than 3% of cases and controls refused to be interviewed. The study was conducted
between 1985 and 1988. Subjects were interviewed with a structured questionnaire including a wide range of
questions on sociodemographic, lifestyle and occupational exposure characteristics, including radar exposure.
Analyses were adjusted for age, sex and cigarette smoking. The study reported no association with radar, but
numbers of exposed subjects and relative risks were not reported. [The lack of information on exposed numbers
gives this study little weight. It was unclear whether this study included only newly diagnosed patients during
the study period.]
2089
2090
2091
2092
2093
Karipidis et al. (2007b) conducted a population-based case-control study of non-Hodgkin’s
lymphoma (NHL) in Australia. The study consisted of 694 NHL cases diagnosed in two regions in Australia
during 2000–2001 ascertained from the New South Wales Central Cancer Registry, and 694 population-based
controls matched by age, sex and region of residence selected from electoral roll. The participation rate was
61% for controls, and it was stated that 85% of approached cases participated, but it was not stated how many
61
2094
2095
2096
2097
2098
2099
2100
2101
cases could not be approached for example because they had died before their existence was notified to the
research team. A detailed occupational history was obtained by a mailed self-administered lifetime calendar and
an extensive (60-minute) telephone interview. Exposure to ionising, UV, RF, ELF radiation was assessed using
a Finnish job-exposure matrix. Analyses were adjusted for the matching variables and ethnicity. There was a
suggestion that odds ratios increased in relation to a measure of cumulative exposure to RF (W/m2-years, the
product of level and duration of exposure), but this was based on very small numbers and none of the odds ratios
or the trend test reached statistical significance. In the highest category of 175–900 W/m2-years, the odds ratio
was 3.15 (95% CI: 0.63–15.87), based on 6 exposed cases.
2102
62
Table 12.1.8. Studies of haematological or lymphopoietic cancer incidence and mortality in people occupationally exposed to RF EMF
Country
Time period
Canada and France
1970–1988 (Quebec)
1978–1989 (France)
Study design
and population
Exposure source
and assessment
Outcome and subgroups
Nested casecontrol of
electrical utility
workers
Pulsed EMF, >200V/m
in 5–20 MHz range
Incidence of leukaemia
Cumulative exposure (OR)
<median
≥median
≥90th percentile
57/201
38/173
9/28
1.00
0.69 (0.40–1.17)
0.80 (0.19–3.36)
Incidence of haematological
cancer
Cumulative exposure (OR)
<median
≥median
≥90th percentile
167/672
135/521
28/101
1.00
0.90 (0.65–1.25)
0.96 (0.48–1.90)
Incidence of Hodgkin's disease
Cumulative exposure (OR)
<median
≥median
≥90th percentile
43/169
24/97
4/12
1.00
0.90 (0.43–1.89)
1.33 (0.23–7.68)
Incidence of Non-Hodgkin’s
lymphoma
Cumulative exposure (OR)
<median
≥median
≥90th percentile
54/252
56/183
13/40
1.00
1.41 (0.83–2.38)
1.80 (0.62–5.25)
Incidence of multiple myeloma
Cumulative exposure (OR)
<median
≥median
≥90th percentile
13/50
17/68
2/21
1.00
0.84 (0.30–2.34)
0.20 (0.03–1.39)
Job-exposure matrix
based on 100 person302 cases, 1193 weeks of
controls, males
measurements from
only
exposure meters worn
by workers to derive
Adult
estimates of shortduration PEMFs or
high-frequency
transient fields
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
63
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
Comments
Reference
Controls
Armstrong et
selected at
al. (1994)
random from
the cases’
risk set and
matched by
utility and
year of birth.
Analyses
adjusted for
case-control
set and
socioeconom
ic status,
exposure to
chemicals,
duration of
employment.
Norway
1961–1991
Cohort of
women certified
as radio and
telegraph
operators 19201980
Radio and telegraph
operators on merchant
ships with potential
exposure to RF fields
Incidence (SIR)
Leukaemia
Lymphoma
Exposure of personnel
to RF/MW
Incidence of
haematological/lymphatic cancer
Ever exposed (RR)
No
Yes
211
36
1.00
5.33 (p=0.01)
Incidence of leukaemia
Ever exposed (RR)
No
Yes
19
88
1.00
6.55 (p<0.01)
Mortality from leukaemia (SMR)
1
5.0 (1.27–27.9)
2
5
1.1 (0.1–4.1)
1.3 (0.4–2.9)
2619 females
Adult
Poland
1971–1990
Annual rates in
all career
personnel in
Polish army
Military safety, service
personnel/health
124500 subjects department records,
(3860 exposed) for cases also hospital
records and
25-59 y
questionnaire
Italy
1962–1992
Cohort of Italian RF field exposure
plastic factory
through work in
workers
dielectric heat sealing
department
481 females
Adult
Canada
1964–1985
Unclear, reports that
>10 W/m2 frequently
exceeded
Cohort of police Potential exposure to
officers in
traffic radar
Ontario, Canada Not assessed
22197 subjects,
both sexes
Adult
Incidence (SIR)
Leukaemia
12
Hodgkin’s disease
8
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
64
Adjusted for Tynes et al.
age, time
(1996)
since
certification,
calendar
year, age at
first
childbirth.
Also
exposed to
light at night.
CrossSzmigielski
sectional
et al. (1996;
study. For all 2001; 2006)
cancers
combined,
RR=1.83,
(p=0.01).
Multiple
sources to
identify
exposure
among
cases, but
not among
non-cases.
Sex, age,
calendar
periodspecific
SMRs.
Lagorio et al.
(1997)
Analyses for Finkelstein
(1998)
0.60 (0.31–1.05) males only,
standardised
for age and
calendar
0.84 (0.36–1.66) year.
USA
1976–1996
Cohort of US
Motorola
employees
95775 subjects,
both sexes
(24621 exposed)
Adult
Moderate or high peak
RF exposures through
design, manufacture
and testing of wireless
devices
Job title with
expert
assessment of
usual exposures
Mortality from
lymphatic/haematopoetic cancer
External comparison (SMR)
All cohort
Moderate/high exposure
193
20
Internal comparison, cumulative
exposure (RR)
No exposure
<median
≥median
148
21
34
1.00
0.74 (0.39–1.28)
0.67 (0.40–1.05)
Mortality from leukaemia
External comparison (SMR)
All cohort
Moderate/high exposure
79
11
0.8 (0.6–1.0)
0.77 (0.38–1.38)
Internal comparison, cumulative
exposure (RR)
No exposure
<median
≥median
66
8
13
1.00
0.64 (0.26–1.33)
0.57 (0.28–1.04)
Mortality from Hodgkin's Disease
External comparison (SMR)
All cohort
Moderate/high exposure
19
3
1.1 (0.7–1.8)
1.11 (0.23–3.24)
Internal comparison, cumulative
exposure (RR)
No exposure
<median
≥median
70
7
14
1.00
0.52 (0.12–1.50)
0.59 (0.22–1.33)
Mortality from non-Hodgkin’s
lymphoma
Internal comparison, cumulative
exposure (RR)
No exposure
<median
≥median
12
3
4
1.00
0.99 (0.21–3.29)
0.95 (0.25–2.84)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
65
Adjusted for Morgan et al.
age, sex and (2000)
ethnicity for
0.77 (0.67–0.89) SMR and
0.54 (0.33–0.83) age, sex and
period of hire
for RR
United States
1979–1997,
Cohort of radar
technicians in
US Navy
Occupations with high
or low radar exposure
potential
40581 males
(20021 high
exposure
potential)
Job title, expert
assessment on
potential for high
exposure, type and
power of radar units
Adult
Mortality from leukaemia
External comparison (SMR)
All cohort
Low exposure
High exposure
Internal comparison, high vs. low
group (RR)
Mortality from lymphoma or
multiple myeloma cancer
External comparison (SMR)
All cohort
Low exposure
High exposure
Internal comparison, high vs. low
group (RR)
Australia
2000–2001
Populationbased casecontrol study
694 cases, 694
controls
20-74 year
Occupational
exposure to RF
Detailed questionnaire
and interview and jobexposure matrix
Incident Non-Hodgkin’s
lymphoma
Cumulative exposure (product of
level and duration) (OR)
Unexposed
>0–<25
25–<175
175–900
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
66
113
44
69
0.96 (0.80-1.16)
0.77 (0.58-1.04)
1.14 (0.90-1.44)
69/44
1.48 (1.01-2.17)
182
91
91
0.91 (0.79-1.06)
0.94 (0.77-1.16)
0.89 (0.72-1.09)
91/91
0.91 (0.68-1.22)
Adjusted for
age at entry
and attained
age.
Groves et al.
(2002)
Response
Karipidis et
rate: 66%
al. (2007b)
(cases), 65%
(controls).
678/685
4/4
6/3
6/2
1.00
1.08 (0.27–4.35)
1.89 (0.46–7.69)
3.15 (0.63–15.9)
P trend=0.09
Adjusted for
matching
variables
(age, sex,
region) and
ethnicity.
Belgium
1968–2004
Cohort of military
personnel
serving 19631994 in 2
battalions
operating
complex radar
devices
compared with 3
battalions not
operating radar
High radar exposure
Characterisation of
exposure levels at
site, no individual
assessment
Mortality from lymphatic and
haematopoietic cancer (RR)
Exposure group
Control batallion
Radar batallion
7349 males
(4417 exposed)
Adult
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
67
1
11
Cause of
Degrave et
death
al. (2009)
retrieved for
1.00
71% in radar
7.22 (1.09–47.9) group and
70% in
control
group.
Table 12.1.9. Studies of brain and other nervous system cancer incidence and mortality in people occupationally exposed to RF EMF
Country
Time period
Norway
1961––1991
Study design
and population
Exposure source
and assessment
Outcome and subgroups
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
Comments
Reference
Cohort of
women certified
as radio and
telegraph
operators 19201980
Radio and telegraph
operators on merchant
ships with potential
exposure to RF
Incidence of brain cancer (SIR)
5
Age, time
since
certification,
calendar
year, age at
first
childbirth.
Tynes et al.
(1996)
1.0 (0.3–2.3)
2619 females
Also
exposed to
light at night.
Adult
United States, 1970–1989, Nested casecontrol study of
male personnel
in US Air Force
230 cases, 920
controls, males
Adult
Canada and France
1970–1988 (Quebec)
1978–1989 (France)–
Nested casecontrol study
within
populations of
electrical utility
Potential exposure to
RF/MW
Incident brain tumour
Job title-timeexposure matrix, from
personnel records,
measurements in
radio rooms of 3 ships
Ever exposed (OR)
No
Yes
136/639
94/281
1.00
1.39 (1.01–1.90)
Product of RF/MW exposure
score and duration in months for
each occupation (OR)
None
2–48
49–127
125–235
236–610
136/639
15/62
29/71
25/68
25/80
1.00
1.26 (0.71–2.24)
1.50 (0.90–2.52)
1.26 (0.71–2.22)
1.51 (0.90–2.51)
Pulsed EMF, >200V/m
in 5-20 MHz range
Job-exposure matrix
based on 100 personweeks of
Incident malignant brain cancer
Cumulative exposure (OR)
<median
≥median
≥90th percentile
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
68
49/176
35/149
9/29
1.00
0.84 (0.47–1.50)
1.90 (0.48–7.58)
Controls
Grayson et
matched on al. (1996)
year of birth,
ethnicity and
being in
cohort at
time of
case’s
diagnosis.
Further
covariates
were age,
rank,
ethnicity
Controls
Armstrong et
selected at
al. (1994)
random from
the cases’
risk set and
workers
84 cases, 325
controls, males
Adult
Italy
1962–1992
Cohort of Italian
plastic factory
workers
481 females
Adult
measurements from
exposure meters worn
by workers to derive
estimates of shortduration PEMFs or
high-frequency
transient fields
Incident benign brain tumour
Cumulative exposure (OR)
<median
≥median
≥90th percentile
RF field exposure
through work in
dielectric heat sealing
department
Mortality from brain cancer (SMR) 1
10.0
Sex, age,
calendar
periodspecific
SMRs
Incidence of brain cancer (SIR)
0.84 (0.48–1.36)
Analyses for Finkelstein
males only, (1998)
standardised
for age and
calendar
year
9/50
16/48
1/6
1.00
1.58 (0.52–4.78)
–
Job exposure matrix
based on 1 week
monitoring of
employees in 19911992
matched by
utility and
year of birth.
Analyses
adjusted for
case-control
set and
socioecono
mic status,
exposure to
chemicals,
duration of
employment
Lagorio et al.
(1997)
Canada
1964–(1997)
Canada
1964––1985
Cohort of police Potential exposure to
officers in
traffic radar
Ontario, Canada Not assessed
22197 subjects,
both sexes
16
Adult
USA
1976––1996,
Cohort of US
Motorola
employees
195775 total
(24621
exposed), both
Moderate or high peak
RF exposures through
design, manufacture
and testing of wireless
devices
Mortality for CNS and brain
cancer
External comparison (SMR)
All cohort
moderate/high exposure
Job title with expert
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
69
53
7
0.60 (0.45–0.78)
0.53 (0.21–1.09)
Adjusted for Morgan et al.
age, sex and (2000)
ethnicity for
SMR and
age, sex and
period of hire
sexes
Adult
Poland
1971–1990–
assessment of usual
exposures
Annual rates in
all career
personnel in
Polish army
Exposure of personnel
to RF/MW
Cohort of radar
technicians in
US Navy
Occupations with high
or low radar exposure
potential
40581 males
(20021 high
exposure
potential)
Job title, expert
assessment on
potential for high
exposure, type and
power of radar units
Military safety, service
personnel/health
124500 subjects department records,
(3860 exposed) and for cases also
hospital records and
25-59 year
questionnaire
United States
1979–1997
Adult
Internal comparison,
Cumulative exposure (RR)
no exposure
<median
≥median
Nervous system/brain cancer
Ever exposed (RR)
No
Yes
Mortality from brain cancer
External comparison (SMR)
All cohort
Low exposure
High exposure
Internal comparison,
high vs. low group (RR)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
70
for RR
34
7
10
81
7
1.00
0.97 (0.37–2.16)
0.91 (0.41–1.86)
1.00
2.70 (p<0.01)
88
51
37
0.86 (0.70–1.06)
1.01 (0.77–1.33)
0.71 (0.51–0.98)
37/51
0.65 (0.43–1.01)
CrossSzmigielski
sectional
et al. (1996;
study. For all 2001; 2006)
cancers
combined,
RR=1.83,
(p=0.01).
Multiple
sources to
identify
exposure
among
cases, but
not among
non-cases.
Adjusted for Groves et al.
age at cohort (2002)
entry,
attained age
Germany
2000–2003–, –
Case-control
study in four
neurosurgical
units in three
regions in
Germany
366 glioma, 381
meningioma
cases and 1494
controls
Occupational
exposures to RF/MW
obtained by computerassisted interview
Glioma
Total exposure (OR)
No/not probable
Probable/high
328/653
38/79
1.00
1.04 (0.68–1.61)
Job title, expert
assessment on
potential for high
exposure, type and
power of radar units
Participation Berg et al.
rates: 80% (2006)
(glioma
cases), 88%
(meningioma
cases), 63%
(controls).
Adjusted for
SES,
urban/rural,
exposure to
ionising
radiation,
cigarette
smoking,
age at
diagnosis.
30-69 year
Likelihood of exposure (OR)
No exposure
Not probable
Probable
High
308/607
20/46
16/42
22/37
1.00
0.84 (0.48–1.46)
0.84 (0.46–1.56)
1.22 (0.69–2.15)
Duration of high exposure (OR)
Not highly exposed
<10yrs
≥10yrs
344/695
9/17
13/20
1.00
1.11 (0.48–2.56)
1.39 (0.67–2.88)
Meningioma
Total exposure (OR)
No/not probable
Probable/high
355/714
26/48
1.00
1.12 (0.66–1.87)
Likelihood of exposure (OR)
No exposure
Not probable
Probable
High
340/687
15/27
15/31
11/17
1.00
1.11 (0.57–2.15)
1.01 (0.52–1.93)
1.34 (0.61–2.96)
Duration of high exposure (OR)
Not highly exposed
<10yrs
≥10yrs
370/745
5/9
6/8
1.00
1.14 (0.37–3.48)
1.55 (0.52–4.62)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
71
Melbourne, Australia–
1987–1991–
Populationbased casecontrol study,
cases identified
from 14
hospitals in
Melbourne area,
controls from
electoral roll in
the study area
Occupational
exposure to RF
Detailed job history
obtained from
questionnaire and
interview, job
exposure matrix
(FINJEM), self-report
and expert review
Glioma (OR)
Cumulative exposure (W m-2
years, tertiles)
Unexposed
0<11
11<52
>52
Duration of exposure (yrs)
Self-reported exposure
Unexposed
0 < to 3
3 > to 8
>8
416 cases and
422 controls,
both sexes
20-70 year
Expert assessment
Unexposed
0 < to 3
3 > to 6
>6
Belgium
1968–2004
Cohort of
military
personnel
serving 19631994 in 2
battalions
operating
complex radar
devices, 3
battalions not
operating radar
High radar exposure
Characterisation of
exposure levels at
site, no individual
assessment
Eye, brain, nervous system
cancer mortality (RR)
Exposure group
Control battalion
Radar battalion
Self-reported
occupational RF
exposures collected
by personal interview
Incident brain tumours overall
Ever exposed (OR)
No
Yes
396/404
4/7
8/4
6/6
1.00
0.57 (0.16–1.96)
1.80 (0.53–6.13)
0.89 (0.28–2.81)
P trend=0.91
385/373
9/16
8/18
12/14
1.00
0.53 (0.23–1.21)
0.43 (0.18–1.00)
0.82 (0.37–1.82)
P trend=0.08
381/396
10/9
12/8
11/8
1.00
1.20 (0.48–3.04)
1.65 (0.66–4.17)
1.57 (0.62–4.02)
P trend=0.17
2
8
1.00
2.71 (0.42–
17.49)
Response
Karipidis et
rates: 85%
al. (2007b)
(cases), 61%
(controls)
Adjusted for
sex, age and
years of
education
Cause of
Degrave et
death
al. (2009)
retrieved for
71% in radar
group and
70% in
control group
7349 males
(4417 exposed)
Adult
France
1999–2001
Case-control
study of general
population of
Gironde, south
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
72
148/375
7/22
1.00
1.50 (0.48–4.70)
Participation
rates: 70%
(cases),
69%
Baldi et al.
(2011)
western France,
controls selected
from local
electoral rolls
221 brain
tumours
including 105
glioma, 442
controls
Probability of
exposure to RF and
duration of exposure
for each job assessed
by occupational
hygienists
≥16year
France–
2005
Hospital-based
case-control
study from two
main centres
122 cases, 122
controls
Occupational
exposure to RF/MW
Job history, expert
judgment, job
exposure matrix
(controls).
Incident glioma
Ever exposed (OR)
No
Yes
71/191
7/16
1.00
1.44 (0.50–4.13)
Incident meningioma
Ever exposed (OR)
No
Yes
61/121
0/2
1.00
–
Incident acoustic neuroma
Ever exposed (OR)
No
Yes
31/59
1/5
1.00
0.40 (0.05–3.42)
Glioma
Ever occupationally exposed
(OR)
No
yes
109/112
7/4
1.00
1.20 (0.30–4.77)
Adjusted for
exposure to
pesticides,
cigarette
smoking,
educational
level.
Participation
rates: 72%
(cases),
unknown for
controls.
Adjusted for
sex, age,
education
level and
occupational
exposure
levels.
≥18 year
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
73
Spinelli et al.
(2010)
Table 12.1.10. Studies of breast cancer incidence and mortality in people occupationally exposed to RF EMF
Country
Time period
USA –
1984–1989,
Study design
Exposure source and
and population assessment
Outcome and subgroups
Case-control
study of
mortality in 24
USA states
Mortality from breast cancer
Exposure probability (OR)
White women
0
1
2
3
4
24505/844
84
2107/6939
668/2630
1518/6484
199/699
1.00
1.00 (0.9–1.1)
1.06 (0.97–1.2)
1.15 (1.1–1.2)
0.99 (0.8–1.2)
Black women
0
1
2
3
4
3437/1259
6
28/585
92/409
274/965
35/115
1.00
1.22 (1.0–1.4)
0.93 (0.7–1.2)
1.27 (1.1–1.5)
1.08 (0.7–1.6)
Exposure level (OR)
White women
0
1
2
3
24505/844
84
1183/4260
1940/6758
1369/5734
1.00
1.15 (1.1–1.2)
0.95 (0.9–1.0)
1.14 (1.1–1.2)
Black women
0
1
2
3
3437/1259
6
150/487
243/815
236/772
1.00
1.23 (1.0–1.5)
1.02 (0.8–1.2)
1.34 (1.1–1.5)
Incidence of breast cancer (SIR)
50
1.5 (1.1–2.0)
31 workplace
exposure to RF
Job exposure matrix,
industrial hygienist
33509 cases,
opinion, expert
117794 controls, databases
females only
All ages
Norway
Cohort study of
Radio and telegraph
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
74
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
Comments
Reference
Based on
record
linkage,
adjusted for
age and
socioecono
mic status.
Cantor et al.
(1995)
Standardise
Tynes et al.
1961–1991
women certified operators on merchant
as radio and
ships with potential
telegraph
exposure to RF
operators 1920–
1980
2619 females
Nested case–control
Duration of employment, yrs (OR)
Age at diagnosis <50y
none
3/15
>0–2.6
13/67
>2.6–8.8
13/68
Adult
Age at diagnosis ≥50y
none
>0–3.2
>3.2–14.6
USA
1976–1996
Cohort of US
Motorola
employees
195775 total
(24621
exposed), both
sexes
United States
1979–1997
Moderate or high peak
RF exposures through
design, manufacture
and testing of wireless
devices
Adult
Job title with expert
assessment of usual
exposures
Cohort of radar
technicians in
US Navy
Occupations with high
or low radar exposure
potential
40581 males
(20021 high
exposure
potential)
Job title, expert
assessment on
potential for high
exposure, type and
power of radar units
Adult
Mortality from breast cancer
External comparison (SMR)
All cohort
moderate/high exposure
Mortality from breast cancer
External comparison (SMR)
All cohort
Low exposure
High exposure
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
75
1/19
5/44
150/46
214
9
4
2
2
1.0
0.9 (0.2–3.7)
0.8 (0.2–3.6)
P trend=0.80
1.0
1.9 (0.2–17.9)
5.9 (0.7–47.7)
P trend=0.02
0.8 (0.7–0.9)
0.9 (0.4–1.8)
1.09 (0.41–2.91)
1.13 (0.28–4.54)
1.05 (0.26–4.20)
d for age
(1996)
and
calendar
year. Also
exposed to
light at night.
Tests for
trend in
nested
case–control
analyses no
longer
significant
after
adjusting for
shift work
category
(p=0.29 for
age<50,
p=0.32 for
age≥50y),
adjusted for
age at first
birth and
time since
certification.
Adjusted for Morgan et al.
age, sex and (2000)
ethnicity for
SMR and
age, sex and
period of
hire for RR
Adjusted for Groves et al.
age at
(2002)
cohort entry
and attained
age
Table 12.1.11. Studies of lung cancer incidence and mortality in people occupationally exposed to RF EMF
Country
Time period
Canada and France
1970–1988 (Quebec)
1978–1989 (France) –
Study design
Exposure source and
and population assessment
Outcome and subgroups
Nested casecontrol within
populations of
electrical utility
workers
Incidence of lung cancer
Cumulative exposure (OR)
<median
≥median
≥90th percentile
200/229
308/279
84/63
1.00
1.27 (0.96–1.68)
3.11 (1.60–6.04)
Incidence of lung cancer (SIR)
5
1.2 (0.4–2.7)
508 cases, 508
controls, males
only
Adult
Norway
1961–1991,
Cohort of
women certified
as radio and
telegraph
operators 1920–
1980
Pulsed EMF, >200V/m
in 5–20 MHz range
Job-exposure matrix
based on 100 personweeks of
measurements from
exposure meters worn
by workers to derive
estimates of shortduration PEMFs or
high-frequency
transient fields
Radio and telegraph
operators on merchant
ships with potential
exposure to RF
2619 females
Adult
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
76
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
Comments
Reference
Controls
Armstrong et
selected at
al. (1994)
random from
the cases’
risk set and
matched by
utility and
year of birth.
Analyses
adjusted for
case–
control set
and
socioecono
mic status,
exposure to
chemicals,
duration of
employment
Adjusted for
age and
calendar
year
Tynes et al.
(1996)
USA
1950–1988,
Cohort of
electrical utility
workers
138905 males
Adult
Italy
1962–1992
Pulsed EMF
Job exposure matrix
based on scores at
Hydro-Quebec as in
Armstrong et al (1994)
Cohort of Italian RF field exposure
plastic factory
through work in
workers
dielectric heat sealing
department
481 females
Mortality from lung cancer
Cumulative exposure (units of
'proportion>10ppb-years') (RR)
0–0.70
0.70–2.50
2.50–5.40
5.40–11.90
11.90–41.15
404
323
362
401
196
1.00
1.27 (1.06–1.51)
1.25 (1.04–1.51)
1.32 (1.09–1.60)
1.35 (1.07–1.69)
Mortality from lung cancer (SMR)
1
5.0
Adult
Poland
1971–1990–
Annual rates in
all military
career
personnel
personnel in
Polish army
Exposure of personnel
to RF/MW
Military safety, service
personnel/health
department records,
and for cases also
124500 subjects hospital records and
(3860 exposed) questionnaire
Incidence of larynx/lung cancer
Ever exposed (RR)
No
Yes
724
27
1.00
1.16 (p>0.05)
Incidence of lung cancer (SIR)
77
0.66 (0.52–0.82)
25–59 year
Canada
1964–1985
Cohort of police
officers in
Ontario,
Canada
Potential exposure to
traffic radar
22197 subjects,
both sexes
Adult
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
77
Adjusted for Savitz et al.
age,
(1997)
calendar
time,
ethnicity,
social class,
work status,
and
exposure to
asbestos.
No
information
on cigarette
smoking.
Sex, age,
calendar
period–
specific
SMRs
Lagorio et al.
(1997)
Cross–
sectional
study.
Szmigielski
et al. (1996;
2001; 2006)
Calculated
from data in
paper.
Multiple
sources to
identify
exposure
among
cases, but
not among
non-cases.
Analyses for Finkelstein
males only, (1998)
standardise
d for age
and
calendar
year
USA
1976–1996
Cohort of US
Motorola
employees
195775 total
(24621
exposed), both
sexes
United States
1979–1997
Adult
Job title with expert
assessment of usual
exposures
Cohort of radar
technicians in
US Navy
Occupations with high
or low radar exposure
potential
40581 males
(20021 high
exposure
potential)
Job title, expert
assessment on
potential for high
exposure, type and
power of radar units
Adult
Belgium
1968–2004
Moderate or high peak
RF exposures through
design, manufacture
and testing of wireless
devices
Cohort of
military
personnel
serving 1963–
1994 in 2
battalions
operating
complex radar
devices
compared with 3
battalions not
operating radar
High radar exposure
Characterisation of
exposure levels at site,
no individual
assessment
Mortality from lung cancer (SMR)
All cohort
moderate/high exposure
612
94
0.8 (0.7–0.9)
0.7 (0.6–0.9)
Mortality from trachea, bronchus
and lung cancer
External comparison (SMR)
All cohort
Low exposure
High exposure
897
497
400
0.75 (0.70–0.80)
0.87 (0.79–0.94)
0.64 (0.58–0.70))
Internal comparison
High vs. low group (RR)
400/497
0.73 (0.63–0.83)
Mortality from respiratory and
intrathoracic cancer
Exposure group (RR)
Control battalion
Radar battalion
7349 males
(4417 exposed)
Adult
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
78
28
45
1.0
1.07 (0.66–1.71)
Adjusted for Morgan et al.
age, sex
(2000)
and ethnicity
for SMR and
age, sex
and period
of hire for
RR
Adjusted for Groves et al.
age at
(2002)
cohort entry
and attained
age
Cause of
death
retrieved for
71% in
radar group
and 70% in
control
group
Degrave et
al. (2009)
Table 12.1.12. Studies of uveal melanoma incidence in people occupationally exposed to RF EMF
Country
Time period
US
1978–1987
Study design
Exposure source and
and population assessment
Outcome and subgroups
Hospital-based,
case-control,
study from 1
unit at UCSF,
controls from
random digit
dialing, resident
in 11 states
Incidence of uveal melanoma
Ever exposed to microwave or
radar (OR)
No
Yes
Occupational
microwave and radar
exposure information
collected by interview
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
200/425
21/22
1.0
2.1 (1.1–4.0)
Population-based: 1995–
1997
Hospital-based: 1996–
1998
Populationbased casecontrol study
from active
surveillance and
cancer registry,
hospital-based
at one
department of
Ophthalmology
Populationbased: 37 cases
and 699
controls, 35–69
yr, both sexes
Hospital-based:
81 cases and
148 controls,
Reference
Participation Holly et al.
rates: 93%
(1996)
(cases),
74%
(controls).
Adjusted for
age, number
of large
naevi, eye
colour,
tanning
response,
221 cases, 447
controls, males
only
Germany–
Comments
Occupational exposure
to radio sets or radar
Questions about use of
radio set s or radar
occupationally for at
least a few hours a day
Incidence of uveal melanoma
Radar units
Ever exposed
Both groups
Population–based
Hospital–based
Radiosets
Both groups
Ever exposed
≥5 yrs prior
≥3 years prior
Population–based
Ever exposed
≥5 yrs prior
≥3 years prior
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
79
0.8/4.0 %
2.7/4.6 %
0/2.7 %
0.4 (0.0–2.6)
0.8 (0.0–6.5)
0.0 (0–3.6)
7.6/4.4%
7.6/3.8%
5.9/3.8%
3.3 (1.2–9.2)
3.3 (1.2–9.2)
2.5 (0.8–7.7)
13.5/5.5%
13.5/4.6%
10.8/4.6%
4.6 (1.0–19.2)
4.5 (1.0–17.5)
3.8 (0.7–16.9)
Participation Stang et al.
rates for
(2001)
cases and
controls:
population–
based (84%,
48%),
hospital–
based (88%,
79%).
Adjusted for
age, sex,
region, SES
and eye and
hair colour
35–74 year,
both sexes
9 European countries
1994–1997
Populationbased and
hospital- based
case-control
study,
depending on
country
293 cases,
3198 controls
Hospital–based
Ever exposed
≥5 yrs prior
≥3 years prior
Occupational exposure
to RF
Based on interview,
and for radar on
assumptions as in
(Baumgardt-Elms et
al., 2002), and jobexposure matrices
4.9/2.0%
4.9/2.0%
3.7/2.0%
Incidence of uveal melanoma
Ever occupational radar exposure Not
reported
35–69 year
1.8 (0.3–14.3)
2.0 (0.3–15.8)
1.3 (0.1–11.4)
No association
Participation Behrens et
rates: 82%
al. (2010)
(cases),
62%
(population–
based
controls),
86%
(hospital–
based
controls).
Numbers
and relative
risks not
reported in
the paper.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
80
Table 12.1.13. Studies of cutaneous melanoma incidence in people occupationally exposed to RF EMF
Country
Time period
Canada and France
1970–1988 (Quebec)
1978–1989 (France)
Study design
Exposure source and
and population assessment
Outcome and subgroups
Nested casecontrol within
populations of
electrical utility
workers
Incidence of malignant melanoma
Cumulative exposure (OR)
<median
≥median
26/107
≥90th percentile
23/84
3/20
508 cases, 508
controls, males
only
Adult
Norway
1961–1991
Cohort of
women certified
as radio and
telegraph
operators 1920–
1980
Pulsed EMF, >200V/m
in 5–20 MHz range
Job-exposure matrix
based on 100 personweeks of
measurements from
exposure meters worn
by workers to derive
estimates of short–
duration PEMFs or
high-frequency
transient fields
Radio and telegraph
operators on merchant
ships with potential
exposure to RF
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
Incidence of malignant melanoma 9
(SIR)
2619 females
Adult
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
81
1.00
1.38 (0.66–2.87)
0.31 (0.03–2.82)
0.9 (0.4–1.7)
Comments
Reference
Controls
Armstrong et
selected at
al. (1994)
random from
the cases’
risk set and
matched by
utility and
year of birth.
Analyses
adjusted for
case-control
set and
socioecono
mic status,
exposure to
chemicals,
duration of
employment.
Diagnoses
include
uveal
melanoma.
Age and
calendar
year
stratified.
Tynes et al.
(1996)
Canada
1964–1985
Cohort of police
officers in
Ontario,
Canada
Potential exposure to
traffic radar
Incidence of malignant melanoma 41
(SIR)
Not assessed
22197 subjects,
both sexes
Adult
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
82
1.45 (1.10–1.88)
Analyses for Finkelstein
males only, (1998)
standardised
for age and
calendar
year.
Table 12.1.14. Studies of testicular cancer incidence and mortality in people occupationally exposed to RF EMF
Country
Time period
Sweden–
1989–1992
Study design
Exposure source and
and population assessment
Exposure groups
Case-control
study; cases
identified from
cancer registry
and controls
from population
register
Incidence of testicular cancer
Ever had occupation (OR)
Amateur radio operator
Radar worker
Engineer in
electronics/telecommunication
148 cases, 314
controls, males
only
Potential RF exposure
through work in
telecommunications or
radar or amateur radio
operator
Self-administered
questionnaire about job
history
No. of
Relative risk
exposed
(95% CI)
cases
(/controls)
7/7
2/3
8/9
2.2 (0.7–6.6)
2.0 (0.3–14.2)
2.3 (0.8–6.7)
23
1.30 (0.89–1.84)
Comments
Reference
Adjusted for
case-control
set.
Hardell et al.
(1998)
30–75 year
Canada
1964–1985
Cohort of police Potential exposure to
officers in
traffic radar
Ontario, Canada Not assessed
22197 subjects,
both sexes
(20601 males)
Incidence of testicular cancer
(SIR)
Analyses for Finkelstein
males only, (1998)
standardised
for age and
calendar
year.
Adult
United States
1979–1997
Cohort of radar
technicians in
US Navy
Occupations with high
or low radar exposure
potential
40581 males
(20021 high
exposure
potential)
Job title, expert
assessment on
potential for high
exposure, type and
power of radar units
Adult
Germany–
1995–1997
Populationbased casecontrol study in
Hamburg,
Bremen, Essen,
Occupations involving
RF emitters or radar in
job of at least 6 months
duration
Job titles and job
Mortality from testicular cancer
External comparison (SMR)
All cohort
Low exposure
High exposure
Internal comparison
High vs. low group (RR)
Incidence of testicular cancer
RF emitters
Ever exposed (OR)
No
Yes
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
83
9
4
5
0.53 (0.28–1.02)
0.46 (0.17–1.24)
0.60 (0.25–1.43)
5/4
1.30 (0.35–4.89)
219/631
50/166
1.0
0.9 (0.60–1.24)
Adjusted for Groves et al.
age at
(2002)
cohort entry
and
attained
age.
Participation Baumgardtrates: 76%
Elms et al.
(cases),
(2002)
57%
(controls).
Saarbrucken
and the Federal
State of
Saarland
269 cases, 797
controls
≥15 year
descriptions, expert
assessment for radar
exposure
Weighted measure by duration
and distance from source (tertiles)
(OR)
not exposed
>0–≤6
6–≤15
15–≤102
220/635
19/52
14/54
16/56
1.0
(0.56–1.74)
0.7 (0.38–1.35)
(0.46–1.56)
Radar units
Ever exposed (OR)
No
Yes
247/739
22/58
1.0
1.0 (0.60–1.75)
Weighted measure by duration
and distance from source (tertiles)
(OR)
not exposed
>0–≤45
45–≤135
135–≤2225
251/741
7/15
4/21
7/20
1.0
1.4 (0.55–3.77)
0.5 (0.17–1.55)
0.9 (0.36–2.19)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
84
2103
2104
12.1.4
Environmental RF exposure and cancer
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
Studies of residential RF exposure such as from radio and TV transmission masts in relation to cancer
risk have started to emerge in the late nineties, with only one study published before the previous Environmental
Health Criteria report (WHO, 1993b). This study, in San Francisco in 1992 (Selvin, Schulman & Merrill, 1992)
was focussed on statistical analysis of spatial data and the results are not reported according to standard
epidemiologic practice. Indeed the authors did not even report a relative risk. The source of exposure was a large
TV tower, and the three statistical methods considered in the paper all showed that the pattern of cancer
incidence was essentially random with respect to the tower. Studies published in the nineties have only been of
television and radio towers, but with the introduction of mobile telephony, studies have also started to emerge
with respect to mobile phone base stations from 2004 onwards. Many of the studies around transmitters have
been of ecological design investigating incidence rates of cancer in the vicinity of masts. They are therefore
affected by the limitations of an ecological design, which is mainly that it assesses cancer risks in relation to
exposure at population level rather than at the level of the individual, therefore, it may not follow that individuals
with the disease are the individuals with the high exposure levels (‘ecological fallacy’). Furthermore, there is a
lack of information on other risk factors that are potentially confounding the investigated association. Other
studies have been of case-control design using register data. The use of such register data minimises the
likelihood of selection and recall bias usually associated with case-control studies based on patient contact. The
considerable drawback is, however, the lack of information on relevant confounding factors and other sources of
RF. Some studies have been of cross-sectional design, in which there is no follow-up of individuals and
exposure and disease are assessed at one point in time, which is associated with particular methodological
weaknesses including that the study population does not include people who were exposed but have moved out
of the area. Most early studies of environmental RF exposure and cancer risk have been conducted in response to
local public concerns about the exposure source or perceived cancer clusters near a TV or radio transmitter and
have focussed on occurrence of leukaemia and brain tumours in adults or children. Many studies have applied
very crude exposure assessment based on distance from the transmitter, with some more recent studies providing
improved exposure assessment.
2130
2131
2132
2133
In the following sections we discuss studies of radio and television masts, mobile phone base stations,
and DECT cordless phone base stations. For each of these exposures, where applicable, ecological studies are
discussed separately from cohort or case-control studies because the methodological issues vary between these
study designs. Tables 1–4 summarise the findings for the main cancer types that have been investigated.
2134
12.1.4.1
2135
2136
The evidence for RF exposure from radio and television masts in relation to cancer risk comes from
ecological studies and from case-control or cross-sectional studies.
2137
12.1.4.1.1 Ecological studies
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
Hocking et al. (1996) compared cancer incidence during 1972–1990 in three local government areas
immediately surrounding three towers broadcasting TV and FM radio in northern Sydney to the cancer incidence
in six adjacent areas, estimating power densities from information on commencement of service of each tower,
power and frequency band. The areas were selected because of the similar distance from the towers to the nearest
borders, their residents having a similar upper-middle class socioeconomic status, and their areas being large
enough for there being a decrease in power density. For leukaemia incidence in adults and children combined
they found an age-, sex- and calendar period-adjusted relative risk of 1.24 (95% CI: 1.09–1.40) for the inner
three areas compared with the surrounding areas, based on 1206 cases overall. Their highest relative risk, 1.67
(95% CI: 1.12–2.49), was for the subcategory ‘other leukaemia’, but risk was also significantly raised for
lymphatic leukaemia (RR=1.32, 95% CI: 1.09–1.59). For childhood leukaemia they observed a relative risk of
1.58 (95% CI: 1.07–2.34) for incidence and 2.32 (95% CI: 1.35–4.01) for mortality. Neither for all ages nor for
children were there any risk elevations for brain tumours. Rate ratios for leukaemia or brain tumours among
adults only were not presented, but there appears to be no excess risk among adults. Comparison of childhood
cancer incidence and mortality against rates for New South Wales showed no increased risk of childhood brain
tumours in the inner or outer areas. For leukaemia, however, both incidence (SIR=1.8, 95% CI: 1.2–2.5) and
mortality (SMR=2.4, 95% CI: 1.4–3.7) were raised in the inner area, but not in the outer areas (SIR=1.1, 95%
CI: 0.9–1.4 and SMR=1.0, 95% CI: 0.7–1.4, respectively). The towers broadcasted at frequencies between 63–
533 MHz, with television frequencies transmitting at 100 kW power. Calculated power densities in the vicinity
Radio and television masts
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
85
2
2156
2157
2158
2159
2160
2161
2162
of the towers ranged from 8 µW/cm within 2 km of the centre point of the towers to 0.2 µW/cm2 at 4 km
(approximately the boundary of the inner area), and lower values at longer distances. Some measurements found
actual levels to be five times less than those calculated, however. Because data were not only available at local
government area level, no attempts were made to test incidence or mortality with levels of RF power density.
[Limitations to this study included that no allowance could be made for population movements and that there
was no information available on confounders. Given the wide variation in RF levels between the areas, the
dichotomisation of exposure into two aggregate areas seems somewhat crude.]
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
McKenzie and colleagues re-examined the Sydney results discussed above with respect to all
childhood leukaemia and acute lymphoblastic leukaemia specifically at local government area level over the
same time period (McKenzie, Yin & Morrell, 1998). They investigated 16 local government areas in the vicinity
of the transmitters, conducted field strength measurements in the areas and used a more accurate method for
calculating field strengths in the areas based on distance, effective radiated power and angle of declination.
Areas proximate to the towers that had been previously excluded by Hocking et al. because they had dissimilar
socioeconomic distribution were included as well as further outer areas at comparable distance from the towers
as those in the original ‘outer’ area. They found that distance from the transmitters was a less reliable proxy of
exposure than modelled exposure levels. They examined the possibility of a dose-response relationship with
ALL risk at local government area level and investigated secular changes in age-specific ALL incidence over
three time intervals spanning the study. There was a non-significant positive association between calculated RF
exposure level and age-standardised incidence of total (RR=1.38, 95% CI: 0.99–1.91) and ALL leukaemia
(RR=1.45, 95% CI: 0.96–2.19) after adjusting for area-level socioeconomic status and age. It was found,
however, that this association was entirely due to an excess in Lane Cove, one of the three local government
areas in the original ‘inner area’, whereas another area of similarly high RF exposure level (Willoughby) had an
incidence rate of leukaemia which was comparable to that of New South Wales as a whole. Analyses excluding
Lane Cove showed no positive association between predicted RF exposure and total (RR=0.90, 95% CI: 0.56–
1.44) or ALL leukaemia (RR=0.83, 95% CI: 0.45–1.55). A replication of the original analyses comparing inner
and outer areas showed a relative risk of 1.5 overall and a relative risk around 1.0 after exclusion of Lane Cove
for both total and ALL. Furthermore, a significant proportion of the cases in Lane Cove were diagnosed before
the television transmitters came into 24-hour service, suggesting the original finding might have been due to
chance or some local factor other than RF radiation.
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
Hocking and Gordon (2003)later published a follow-up on their original work investigating survival
from childhood leukaemia in the investigated regions. It was based on 134 patients diagnosed during 1972–1990
and therefore also included in the original study, and a further 26 patients diagnosed during 1990–1993. Survival
rates were investigated in relation to municipality of residence at the time of diagnosis because residence at the
time of death was not available, but the authors established that there was little movement between areas. Among
160 cases of childhood leukaemia, 21 of 36 (58%) who resided in the inner area and 53 of 124 (43%) cases in the
outer area had died before the end of 1993, the censor date of the study. Five-year and ten-year survival rates
were lower in the inner than in the outer areas. There was a significant difference between the two survival
curves for all cases (log-rank test p=0.04). Using Cox’s proportional hazard model to adjust for age and year at
diagnosis and sex, the resulting mortality hazard ratio was 1.8 (95% CI: 1.0–3.0). Analyses restricted to the 123
diagnosed cases of acute lymphatic leukaemia also showed significant difference in survival between the two
areas, with the adjusted mortality hazard ratio being 2.1 (95% CI: 1.1–4.0). [Survival as the outcome instead of
risk of disease has some additional complexities because factors such as effectiveness of treatment also play a
role, but no information was available on received treatments. Because the study included the same patients and
study population as in the original paper the methodological problems discussed for the original study also apply
to this study.]
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
Dolk et al. (1997b) followed up an apparent cluster of leukaemias and lymphomas near a UK radio
and TV transmitter at Sutton Coldfield, West Midlands, England. Field strength measurements in the vicinity of
the transmitter showed a maximum total power density at any one point of 0.013 W/m2 for TV and 0.057 W/m2
for FM radio with considerable variability between different measurement points. Observed numbers of selected
types of cancers during 1974–1986 were compared against those expected based on national incidence rates with
regional adjustment stratified by sex, 5-year age group, year, deprivation index for the areas within a 2 km radius
and within a 10 km radius. Ten bands of increasing distance from the transmitter were defined as the basis of
testing for declining incidence with increasing distance. Relative risk for all cancers was significantly raised
within a 0–2 km radius (SIR=1.09, 95% CI: 1.01–1.17, 703 cases) and within 0–10 km radius (SIR=1.03, 95%
CI: 1.02–1.05, 17 409 cases). Risks of individual types of cancer were raised for adult leukaemia within 2 km
(SIR=1.83, 95% CI: 1.22–2.74) with a significant decline in risk with distance from the transmitter over 10 km
distance (Stone trend test p=0.001). There was no excess of childhood cancer overall and excess of childhood
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
86
2213
2214
2215
2216
leukaemia was not statistically significant (O/E=2/1.1 (<2km); O/E=34/29.7 (<10km)). A significant decline in
relative risk with increasing distance was also observed for bladder cancer and skin cancer. Socioeconomic
deprivation score at census enumeration district level showed greater affluence near to the transmitter than in
areas further away.
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
The Sutton Coldfield Study was followed up by a study of cancer incidence near 20 high-power TV
and frequency modulation (FM) radio transmitters in Great Britain (Dolk et al., 1997a). Through record linkage
with cancer registry data 3305 cases of adult leukaemia, 1540 of skin melanoma and 8307 cases of bladder
cancer were identified diagnosed during 1974–1986 and resident within 0–10 km of a transmitter. There was no
excess risk in the 0–2 km band for adult leukaemia (SIR=0.97, 95% CI: 0.78–1.21) or childhood leukaemia
(SIR=1.12, 95% CI: 0.61–2.06). The highest ratio for all transmitters combined was in the 2–3 km radius
(SIR=1.15). The SIR within 0–10 km radius for all transmitters combined was 1.03 (95% CI: 1.00–1.07) for
adult leukaemia, with some evidence for a decline with distance, despite the absence of a raised risk in the 0–
2 km band. SIRs were 0.90 (95% CI: 0.85–0.94) for skin melanoma and 1.09 (95% CI: 1.06–1.11) for bladder
cancer, with no evidence for a decline with distance. [Therefore, this study only provided very weak support to
the authors’ earlier results.]
2228
2229
2230
2231
2232
2233
The Sutton Coldfield results have also been followed up by another group (Cooper, Hemming &
Saunders, 2001). They used more timely cancer data (1987–1994) to reanalyse cancer incidence around the
transmitter and found considerably weaker results than the original. Mainly, there was no statistically significant
excess risk overall related to the 0–2 km band (SIR=1.32, 95% CI: 0.81–2.05). They did not observe an
increased risk of leukaemia within a 2 km radius of the transmitter and observed a decrease in risk with
increasing distance from the transmitter which was restricted to women.
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
An Italian study was conducted to address concerns of local residents of leukaemia incidence in
children and mortality in adults within a 10 km radius of the Vatican Radio station (Michelozzi et al., 2002). The
radio station consists of numerous short and medium-wave transmitters with transmission powers varying
between 5 and 600 kW and with different frequency ranges. Mortality statistics for 1987–1998 were obtained for
adults from the Lazio Region Geographic Information System. Data on incidence in children for 1987–1999 was
collated from several sources including a register of leukaemia cases in associated hospitals (1989 onwards), a
system of hospital discharges (1996 onwards) and from hospitals directly (1987–1995). The residence of each
case at the time of diagnosis or death was retrieved from the register offices. Population data were available at
census tract level with an average of 253 residents per census tract. There were 40 deaths of leukaemia among
adults and 8 incident childhood leukaemia cases during the study period. In adults of both sexes taken together
the SMR within 2 km of the station was 1.8 (95% CI: 0.3–5.5) based on 2 cases. Stone’s test for trend in rates
over successive 2 km bands around the station gave a p-value of 0.14. The excess risk and the trend were
essentially confined to males. In children the SIR for those living within a 2 km radius was 6.1 (95% CI: 0.4–
27.5) based on one case. The Stone test for trend was reported as p=0.04 when comparing cumulative number of
cases within 2 km bands. [The completeness of case ascertainment from the several sources is unclear and
subjects treated outside the study area would not have been included.]
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
Ha et al. reported a cross-sectional ecology study of cancer incidence in the vicinity of AM radio
transmitters in South Korea (Ha et al., 2003). The study region included 11 ‘high-power’ regions located 2 km or
less from transmitters with 100–1500 kW power and 31 ‘low-power’ regions located 2 km or less with
transmitters with 50 kW power. Four control areas for each high-power area were selected adjacent to high
power areas to obtain expected age-specific incidence rates. Cancer registrations were obtained for 1993–1996
for cases over age 10 years, with census and population registration data for 1995. Population sizes ranged from
3152 to 126 523 persons per area. Rate ratios comparing high to low areas were somewhat raised for all cancers
(RR=1.2, 95% CI: 1.1–1.4), but were not significantly raised for leukaemia, malignant lymphoma, brain cancer
for both sexes or female breast cancer. Sex-specific analyses showed a borderline raised relative risk for brain
cancer in females (RR=2.0, 95% CI: 1.0–13.6) but not males (RR=1.6, 95% CI: 0.8–8.8). The authors also
compared observed rates against expected based on age-specific incidence rates in control areas between regions
with transmitters of different transmitting power (100 kW, 250 kW, 500 kW, 1500 kW) for each of the five
outcomes investigated. SIRs were above 1.0 for all categories except the 500 kW group for all cancers. For
leukaemia, they were significantly raised for the 250 kW and 1500 kW group but not for the 500 kW group
(<1.0). For brain cancer, SIR was raised for the lowest transmission power group but not for the higher ones and
none were raised for female breast cancer. [Therefore, there were several categories showing excess risk but for
none of the outcomes there was a consistent dose-response. Furthermore, the derivation of expected rates from
the 44 control areas of unreported population size might have resulted in unstable estimates if the population size
was relatively small. The major methodological problem with this study is that it investigates incidence rates in
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
87
2269
2270
2271
administrative units which contain a transmitter, and therefore does not even use distance from exposure. Control
areas were at least 2 km away from the transmitters so that it is unclear whether control areas had indeed lower
exposure levels than the areas that were considered exposed.].
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
A further study in South Korea of similar design reported cancer mortality rates over 1994–1995 in
ten areas including AM radio broadcast towers of over 100 kW power, and ten control areas without radio
broadcasting towers (Park, Ha & Im, 2004). It observed statistically significantly raised mortality from cancer in
the exposed areas (direct standardised mortality rate ratio: 1.29, 95% CI: 1.12–1.49). While most cancer-type
specific mortality rate ratios were above 1.0, none of them reached statistical significance, including for
leukaemia (SMR=1.70, 95% CI: 0.84–3.45). However, leukaemia mortality was significantly raised in children
under 15 years (SMR=2.29, 95% CI: 1.05–5.98) and in young adults aged 15–29 years (SMR=2.44, 95% CI:
1.07–5.24). [This study is very similar to that of Ha et al. (2003) and shares the same methodological issues. It is
unclear whether the study areas are identical or overlap. Analyses were based on subjects’ residence at their
place of death, and because the authors state that in Korean tradition most people return to their home town
shortly before they die, there could be considerable exposure misclassification. As with other ecological studies,
control for confounding was not possible.]
2284
Studies with uncertainties related to inclusion criteria
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
Hallberg and Johansson (2002) reported melanoma incidence statistics in relation to FM broadcasting
in four countries in response to Dolk et al’s finding of an inverse association with distance from the Sutton
Coldfield transmitter. They attributed the increasing trends in incidence of melanoma to increased density of FM
transmitters by comparing incidence rates of melanoma over time. They subsequently reported that median
melanoma incidence rates were positively correlated with average number of FM transmitters across
municipalities in Sweden (Hallberg & Johansson, 2005). [Exposure estimates did not include distance to
transmitters. Confounding from UV exposure (e.g. through leisure-time, holidays abroad or sunbed use), an
established risk factor for melanoma, was not taken into consideration. The lack of information on other
contributory factors, and the problem that the associations were measured at population-level means that this
study does not provide meaningful data that can be used to assess a potential impact of RF exposure on
melanoma risk. The study is given little or no weight in the overall assessment.]
2296
12.1.4.1.2 Case-control and cross-sectional studies
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
A case-control study based on an apparent cluster of childhood leukaemia (Maskarinec, Cooper &
Swygert, 1994) was prompted by an observation of an unusually high number of childhood leukaemia cases in
Waiainae, a region in Hawaii. The excess was attributed to the presence of a military installation with lowfrequency radio towers transmitting at 23.4 kHz. Cases were defined as children under age 15 diagnosed with
acute leukaemia between 1979–1990 who had resided in three census tracts in the region. They were ascertained
from records of the local health centre providing primary care to >90% of the population. Four controls were
ascertained from current patients in the health centre, age- and sex matched to cases. There were 12 leukaemia
cases, and the odds ratio for having lived within 2.6 miles of the radio towers, the median value, before diagnosis
was 2.0 (95% CI: 0.06–8.3). Comparison of the 12 cases against incidence rates from the Hawaii Tumor
Registry showed a SIR of 2.09 (95% CI: 1.08–3.65). It was reported that the greatest excess of cases was during
1982–1984, but the authors could not suggest a particularly reason regarding the radio transmitter why this
would be so. [It is unclear how many cases were initially part of the observed cluster, but the study appears to
cover at least part of the same region, and therefore an inflated SIR is likely to be observed. Although
information on other risk factors was collected, it is unclear to what extent analyses have been adjusted for them.
An additional drawback of the case-control analysis is that its small size meant it was unlikely to detect risk
increases.]
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
Ha et al. (2007) followed up their observations with a case-control study of childhood leukaemia and
brain cancer risk in relation to estimated RF exposure from AM radio transmitters (Ha, 2008). Cases with a
leukaemia or brain cancer diagnosis under age 15 during 1993–1999 were recruited from 14 hospitals using the
South Korean Medical Insurance Data System. Controls were recruited from the same hospitals, matched on age,
sex and year of first diagnosis and were diagnosed with respiratory diseases. Exposures were assessed for each
individual child using a prediction program incorporating a geographic information system, and were estimated
as the total RF field from all nearby transmitters and the peak RF field exposure from any of them. This program
was validated against actual measurements for 11 transmitters. A total of 31 AM transmitters with a power of at
least 20 kW were considered. A total of 1928 cases of leukaemia, 956 brain tumour patients and 3082 controls
were recruited. Risk of leukaemia after adjusting for residential location, socioeconomic status and population
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
88
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
density of the community was borderline significantly raised among children resident within 2 km of the nearest
AM transmitter compared with those more than 20 km away (OR=2.15, 95% CI: 1.00–4.67), but significantly
reduced (OR=0.66, 95% CI: 0.44–0.99) for those resident 2–4 km from the transmitter, with no evidence for
trend (p=0.10). For leukaemia risk overall, odds ratios were below 1.0 for increasing levels of predicted total RF
field compared to the first quartile (e.g. OR=0.83, 95% CI: 0.63–1.10 for top quartile), and for the second and
third quartile even statistically significantly so. In analyses separately for lymphocytic and myelocytic
leukaemia, odds ratios were similarly reduced. The authors later also provided risk estimates for peak RF field
(Ha, 2008) showing no increased risk for leukaemia overall, but a raised risk for lymphocytic leukaemia
(OR=1.40, 95% CI: 1.04–1.88) and a reduced risk for myelocytic leukaemia (OR=0.63, 95% CI: 0.41–0.97) in
the highest quartile, corresponding to 608.35 mV/m or greater. Further subdivision of this top quartile into less
or greater than 1012.07 mV/m (95th percentile), did not show further raised risks for lymphocytic leukaemia
(1.45, 95% CI: 1.06–2.00 and 1.24, 95% CI: 0.81–1.91) respectively). No associations of distance or predicted
exposure levels were observed for childhood brain cancer. [This study is methodologically much stronger than
past studies into this topic which had been of ecological design, due to its case-control approach with large
numbers of subjects and individual exposure estimation.]
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
A case-control study investigated exposure to RF EMF in relation to childhood leukaemia risk in an
area of West Germany containing high-power radio and TV broadcast towers (Merzenich et al., 2008). In total,
1959 cases aged 0–14 years diagnosed 1984–2003 were identified from the cancer registry and three controls per
case from population registers matched on age, sex and transmitter area. Exposures to RF fields from AM, FM
and television transmitters were calculated for each individual, and were found to be highly correlated with
measured fields (Schmiedel et al., 2009). Risk was not raised (OR=0.86, 95% CI: 0.67–1.11) when comparing
upper (≥90th, 0.701 V/m) and lower (<90th, 0.504 V/m) quantiles of individual RF exposure for exposure up to
1 year prior to diagnosis. Relative risks for children living within 2 km of the nearest transmitter compared with
those living 10–14 km away were not statistically significantly raised overall (OR=1.04, 95% CI: 0.65–1.67), or
for lymphoid leukaemia (OR=1.31, 95% CI: 0.80–2.15). Analyses for AM and FM transmitters separately did
not affect the results. Exposure assessment was based on address of diagnosis only (or equivalent date for
controls) but a subanalysis of subjects who had not moved address since birth did not materially affect the
results. As this study did not involve patient contact, information on other exposure sources or on confounders
were not available.
2352
Studies with uncertainties related to inclusion criteria
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
Preece et al. (2007) conducted a cross-sectional study of health measures in relation to RF exposure in
three villages following raised concerns about health effects of RF exposure from military antenna systems at
Akrotiri, Cyprus. This study has also been described in section 5.1.3.2. The researchers collected longitudinal
and short-term radiofrequency measurements and administered questionnaires including information on general
wellbeing, specific illnesses, reproductive history and mortality, including from cancer, from approximately
1870 individuals, with 87% response rate. Mortality data were collected from several sources including national
records, cemeteries and reports from family members. The two villages considered exposed had field strengths
with a maximum of 0.30 V/m from the 17.6 MHz military transmissions and up to 1.4 V/m from unspecified
sources, mainly mobile phone frequencies, whereas for the unexposed village these were <0.01 V/m. They
reported no overall differences between villages for cancer or leukaemia, but the authors report that the numbers
of deaths were small. [The study’s cross-sectional design and its lack of comprehensive source of mortality data
make it uncertain whether adequate information on population size and demography could be obtained and
whether all deaths were identified. The study is given little or no weight in the overall assessment.]
2366
12.1.4.2
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
Studies of cancer risks in close proximity to mobile phone masts have, due to mobile telephony being
a relatively new technology, started to emerge from 2004. There are a few studies investigating cancer reports
around individual base stations or in individual towns or villages, and more recently, studies have started to
emerge investigating cancer risk in relation to numerous base stations in wider geographical areas. These studies
have mostly been of ecological design, with the limitations described above. Studies around a single transmitter
are particularly problematic because the population exposed to it is small due to its limited range of transmission,
and investigations of cancer therefore are limited to small numbers of observed cancers of heterogeneous types
diagnosed over a long time interval. Furthermore, such studies have usually been initiated due to the observation
of a cancer cluster or concerns of local residents and conducting a study in the area in which these cases occurred
is likely to show a positive association. Most studies into this topic have had deficiencies in identification of the
study population base as well as outcome ascertainment. They may not have had complete enumeration of all
Mobile phone masts
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
89
2378
2379
2380
2381
2382
2383
individuals in the study population in a well-defined study area and over a well-defined calendar period, and they
may not have been able to ascertain all cases with the outcome of interest in this population, leaving scope for
bias. Because most studies have been of ecological design, they have lacked information on other risk factors for
cancer that might confound a possible association between RF exposure and cancer risk, as well as information
on other sources of RF exposure, and they have not usually taken account of population movements which could
therefore introduce exposure misclassification.
2384
12.1.4.2.1 Studies around single or a few transmitters in small geographical areas
2385
Studies with uncertainties related to inclusion criteria
2386
2387
2388
2389
2390
2391
Several small ecological studies around one or more transmitters in a small geographical area have
been published, that have assessed the correlation between all cancer and distance from the transmitter. The
selection of the study base is ill-defined in these studies, and they do therefore not meet the inclusion criteria for
this report. They are discussed here because some have sometimes attracted sometimes substantial media
attention, but they are given little or no weight in the overall analysis due to the very small numbers observed of
specific cancer-types, and weaknesses in study design.
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
Wolf and Wolf (2004) investigated people living in an area within a 350 m radius of a mobile phone
transmitter for 3–7 years in Netanya, Israel. The exposed group consisted of 622 people identified from one of
the authors’ health clinic, and the control group of 1222 subjects identified from a clinic nearby. It was stated
that this group was closely matched on environmental, workplace and occupational characteristics, but no further
details were given of this. The antenna was installed in 1996 and had a total maximum transmission power at
frequencies of 850 MHz of 1500 watt, and measured and predicted power density was far below 0.53 µW/cm2.
The number of cancer cases in the exposed population between July 1997 and June 1998 was compared against
the number in the control population, incidence rates in the town of Natanya, and national incidence rates. Eight
cases of six different types of cancer were diagnosed in the study period, and this rate (8/622) was several-fold
higher than that of the general population (31/10 000 per year) and in the nearby clinic (2/1222). [The study
design is methodologically weak and rate ratios are based on very small numbers of cases of heterogeneous types
of cancer, and are not age- and sex standardised. The selection of the study population is poorly described, in
particular criteria for decision in respect to residential boundaries for exposed and unexposed groups and
matching, with a high likelihood of selection bias towards the exposed area. Selection of study subjects from
health clinics results in an ill-defined study population base. As with most other studies, there is no information
on confounding factors or other sources of RF fields.]
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
A study in Naila, Germany, investigated cancer incidence in the vicinity of two mobile phone
transmitters (Eger et al., 2004). The first GSM mobile phone mast in Naila became operational in 1993 and a
second was installed in 1997. The ‘inner’ area was defined as being within a 400 meter radius from the cellular
transmission site, and ‘outer’ was defined as being more than 400 meters away. It was calculated that the main
beam hits the ground at 350 meters distance. All four general practice doctors in the town took part in the study
and their team researched the names of patients from the selected streets who had been ill with tumours since
1994. The authors report that this method covered nearly 90% of local residents. From the 9472 registered
residents in the town, 320 ‘inner’ and 647 ‘outer’ residents were selected by selecting ‘similar residential streets’
in the inner and outer area ‘at random’. For the period 1994–2004, the teams found 18 cases of cancer among
320 eligible ‘inner’ residents and 34 among 967 ‘outer’ residents (OR=2.35); this excess was most apparent for
the period 1999–2004. Only crude odds ratios were reported, although average age and proportions of females
were similar between the two areas. Observed numbers of cases were higher than expected based on Saarland
cancer registry data in the ‘inner’ area but lower than expected in the ‘outer’ area, suggesting relative risks might
have been lower than reported. [This study has an ill-defined population base and unclear completeness of case
ascertainment and selection of ‘similar residential streets’ in the inner and outer area ‘at random’ leaves open the
possibility for bias. Also, it was stated that the study was restricted to patients who had been living during the
entire observation time of 10 years at the same address, but it is unclear whether this criterion was applied to all
study subjects, and numbers of subjects who were excluded because of this criterion were not provided. As with
the study by Wolf and Wolf, rate ratios were based on heterogeneous types of cancer and were not age- and sexstandardised.]
2428
2429
2430
2431
The Northern Ireland Cancer Registry conducted an ecological study of cancer incidence and
mortality around a telecommunications mast in Cranlome, Northern Ireland, during 2001–2002, due to an
alleged cancer cluster of 11 cases (Gavin & Catney, 2006). The mast was erected in 1989, and was taken down
in 2002. The 11 reported cases were validated against the cancer registry records, and unreported cases in the
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
90
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
vicinity of the mast were identified from the same source with help of experts in geographical information
systems from Ordnance Survey Northern Ireland. Concentric circles of radius of 1, 2, 3, 4 and 5 km around the
mast were constructed. The COMPAS address database was used to highlight and list all addressing within these
geographic boundaries. Observed cancer rates in the population in an up to 5 km radius were compared with
national cancer incidence rates using indirect age standardisation. Six of the 11 alleged cases could be identified
from contacts with the community among which two were not cancer and one was a non-malignant tumour. In
addition to the three confirmed cancer cases a further 17 cases within 5 km of the mast were identified from
cancer records. The distribution of tumour types was in accordance with what one might expect from the general
population. All-cause malignancy was not raised (SIR=0.94, 95% CI: 0.88–0.99 for males and SIR=1.00, 95%
CI: 0.94–1.06) for females. For brain, lymphoma and leukaemia combined the SIRs were 1.01 for males and
0.99 for females, and analyses by other common subtypes did not show raised risks. [While the definition of the
population base and case ascertainment is reasonably well-defined, the study had very small numbers of cases
and is therefore considered uninformative.]
2445
2446
2447
2448
2449
2450
2451
2452
2453
Eger and Neppe (2009) investigated cancer occurrence for the period 2000–2007 around a mobile
phone base station in Westphalia, Germany. They identified 23 cases, of 10 specific cancer sites, by door-to-door
interviews among 575 residents who lived within 400 meter of a single base station. Cancer incidence in the five
years immediately after the installation of the mast was compared to that in later years, and it was concluded that
there was a statistically significant increase in incidence in the five years immediately after the installation
compared with the 2.5 year period that followed (OR=2.63, 95% CI: 1.14–6.10). [The lack of a well-defined
population base, the small numbers and heterogeneous types of observed cancers and the potential for biased
case ascertainment makes this study uninformative. It is also unclear why, if a true effect existed, cancer risks
would be only raised within five years of exposure.]
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
Stewart et al. (2011) investigated a cancer cluster alleged by residents in a street in Sandwell, West
Midlands, United Kingdom, which was attributed to the installation of a mobile phone base station in their
locality in 1997. Health professionals identified a total of 19 patients diagnosed 1974–2010 and resident in the
street by asking residents and from cancer registrations from the West Midlands Cancer Intelligence Unit. Their
diagnoses were of a range of cancers, all but one of them common, counter-indicative of a cancer cluster. Data
on cancer incidence and mortality at ward level were obtained for 1993–1995, 1999–2001, 2001–2003, 2002–
2004 [it is not clear why some of these periods overlap], and were compared against the West Midlands overall.
Mortality of malignant neoplasms excluding non-melanoma skin cancers was significantly raised for females
(SMR: 1.38, 95% CI: 1.08–1.74), but not males (SMR: 1.20, 95% CI: 0.92–1.55) during the period 2001–2003,
and was not raised for 2002–2004. There was a raised SMR for gastrointestinal cancer during 1993–1995, prior
to installation, and for 2001–2003, with only one case who had been resident in the street and their diagnosis
predated installation of the mast. The SIR for lung cancer was raised during 1999–2001 but was reduced in
females during 1993–1995. There was no association with colorectal, female breast and prostate cancers. The
authors commented that it was unlikely that information around a single base station could demonstrate
causality.
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
A study by Atzmon et al. (2012) investigated the prevalence of a past cancer diagnosis in relation to
distance from RF and MW antennae in the population of the Druze village of Isifya, Israel. The first radio
transmitters were erected in 1970s and mobile phone transmitters in the 1990s, but in the year 2000 all antennae
were destroyed by residents due to health concerns. An excess observed cancers during 1998–2001 had been
indicated according the National Cancer Registry, especially in men (RR=1.57, 95% CI: 1.12–2.02). A total of
348 study subjects were interviewed face-to-face and asked about demographical, medical and lifestyle factors.
After exclusion of subjects who moved to the area after 2000 and patients without documents, the study
population was 307 subjects. Forty-seven subjects reported past diagnosis of cancer. Houses were geographically
mapped and individual exposure intensity to the closest transmitter was estimated based on the distance from the
transmitter. Exposure intensity was not significantly related to prevalence of any of the main types of cancer
observed apart from colorectal cancer (OR=1.03, 95% CI: 1.01–1.05, units not provided). [The paper gives
insufficient detail in order to evaluate the methodological aspects of the study appropriately. It was not described
how study subjects were selected, what the participation rates were or what proportion of the village was
investigated; therefore the study base is undefined. Furthermore, this study is of prevalent cases and therefore is
a study of survivors from cancer. It is stated that family members were interviewed for deceased patients but it
not stated for how many and whether this would have captured all patients diagnosed in the area over a specified
time. Besides, the quality of recall from proxies is likely to be considerably poorer. It appears that exposure was
assessed at the time of recruitment to the study and not retrospective prior to diagnosis of cancer, which would
be aetiologically relevant. The presented odds ratios in relation to exposure intensity have very narrow
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
91
2488
2489
confidence intervals despite of being based on very small numbers of cases and it is unclear how they have been
modelled.]
2490
12.1.4.2.1 Studies of transmitters in larger geographical areas
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
An ecological study in Bavaria, Germany, investigated cancer incidence according to three categories
of mobile phone base station coverage (Meyer, 2006). There were 15 rural districts and independent towns
comprising a total of 2.98 million inhabitants that were considered to have complete coverage by the local cancer
registry in 2005, defined as reporting at least 90% of the expected cases of cancer. From these areas 48 small
municipalities with a total of 177428 inhabitants were selected for the study due to the amount of work
associated with geographical mapping of base stations. The majority (36) of municipalities included under 5000
residents, with only five having over 10 000 residents. Exposure levels were categorised into ‘none’, ‘little’ and
‘moderate’. ‘None’ entailed that none of the municipality’s residential areas were within 400 meters of a base
station, or that any base stations had been operational for less than five years, considered too short for an
induction period if there were an effect. ‘Moderate’ exposure entailed that the base station had been operational
for at least 8 years and at least 15% of the municipality’s residential areas were within 400 meters radius of the
transmitter or that it had been operational for 5–7 years and at least 30% of the area was within 400 meter radius.
Areas that could not be categorised under ‘none’ or ‘moderate’ were categorised as ‘little’ exposure. For thirtyone municipalities no base stations were considered, for 12 one, for 4 two and for 1 three. A total of 37
municipalities were categorised as ‘none’, 8 as ‘little’ and 3 as ‘moderate’. Cancer diagnoses were included for
2002, 2003 or for both years for each municipality depending on whether cancer registrations were judged as
complete for the respective year. A total of 1116 malignant neoplasms were reported in 242 508 person-years.
Standardised incidence ratios for all malignancies by municipality in the ‘none’ category varied between 0 and
approximately 2.5; these municipalities generally had the smallest population size. For the ‘little’ category, they
varied between 0.7 and 1.3 and for the ‘moderate’ between 0.7 and 1.0; these municipalities were considerably
larger than in the ‘none’ group. There were no significant differences between the three exposure groups in SIRs
for cancer of the breast, brain and central nervous system, thyroid and leukaemia [There was therefore no
evidence for higher incidence rates of cancer in municipalities with higher exposure levels, and indeed, the large
heterogeneity in SIRs among small municipalities draws into perspective the results for studies on single
transmitters. The study by Meyer et al. was, however, based on aggregated data, with the associated deficiencies
of ecological studies, and could not take into account patient-specific information on distance from transmitters.
The authors did not specify the criteria on which they selected the 48 municipalities from the larger population.
The age range of the population is not specified but is implied to include children and adults; SIRs were adjusted
for age differences, however].
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
The first nationwide study to investigate cancer risks in relation to mobile phone masts to date was
conducted by Elliot et al. ((2010) in Great Britain. It linked cancer registry and national birth register data in
order to investigate early childhood cancer risk in relation to proximity of residence to mobile phone base
stations. A total of 1397 cases of childhood cancer were identified diagnosed at ages 0–4 years during 1999-2001
in Great Britain. Four controls per case were selected from national birth registers, individually matched on sex
and date of birth. Analyses were conducted for all diagnostic groups and for the two main groups separately:
leukaemia and non-Hodgkin’s lymphoma (527 cases) and brain and central nervous system (251 cases). Mean
distance of registered birth address from the nearest macrocell base station was assessed from a national database
of base-station antennae. There was no statistically significant difference between cases and controls with respect
to mean distance to the nearest antenna (1107 vs. 1073 meter, p=0.31), total power output of all base stations
within 700 meter of the address (2.89 vs. 3.00 kW, p=0.54), or modelled power density from base stations within
1400 meters (-30.3 vs. 29.7 dBm, p=0.41), for cases overall or by main diagnostic group. There was no evidence
for trend when comparing exposures in a low, intermediate and high category. [Exposures were assessed for the
time period of the pregnancy since for controls only birth address was available. If postnatal exposures also
substantially contributed to aetiology, however, misclassification of exposure levels could have biased the results
because among cases, 528 (38%) moved residence between birth and diagnosis. While the design of this study
based on record linkage minimizes selection and reporting biases, there was no information on confounders or on
other relevant exposure sources such as mobile phone and DECT use during pregnancy, which in turn also
affects the ability of the study to detect true effects.]
2539
2540
2541
2542
2543
Spinelli et al. (2010) conducted a case-control study of glioma risk in relation to occupational and
environmental risk factors, including living near a mobile phone base station. The study was conducted in France
and included 122 cases diagnosed in 2005 from brain cancer treatment centres in Marseilles and Toulon and 122
controls selected from the neurosurgery department of the same hospitals. The participation rate among cases
was 71.6% and for controls was not reported. Nineteen cases and 33 controls lived in the vicinity of a mobile
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
92
2544
2545
2546
phone base station (OR=0.49, 95% CI: 0.26–0.92). [This information was apparently self-reported and it is
unclear how well people are able to report mobile phone base stations in their residential area, in particular
patients with glioma who may suffer cognitive effects.]
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
A population-based case-control study of childhood cancer in relation to mobile phone transmitters in
Taiwan was reported by Li et al. (2012a). Cases were children under age 15 years newly diagnosed between
2003–2007 for neoplasms, identified through the Inpatient Expenditures by Admission Database, which covers
all Taiwan citizens. Thirty controls who did not have neoplasms and were insured in the same year as the case
had been admitted were randomly selected matched on year of birth from all children on the Registry for
Beneficiaries database. A total of 2606 cases were newly diagnosed, including 939 cases of leukaemia and 394
cases of brain neoplasms. Information on mobile phone base stations was obtained from the Taiwan National
Communication Council. The researchers calculated the annual summarised power (ASP) in watt-year for each
mobile phone base station annually from 1998–2007 from the product of duration of operation and real emitted
power (watt). The annual power density (APD) in each of the 367 townships in Taiwan was then calculated as
the ratio of total ASP from all base stations in a township to the area of that township. For each study subject
ADP was averaged over the five years prior to the year of diagnosis of the index case. Analyses were adjusted
for age, sex, calendar year of diagnosis, urbanisation level of township and high-voltage transmission line
density of township, and accounted for correlation in the error term because subjects could be clustered in
township. The odds ratio for all neoplasms per 10 Watt-year per km2 increase in a 5-year average of APD after
adjusting for these confounders was 1.02 (95% CI: 0.96–1.08). There was a statistical significant increase in OR
among children with average APD greater or equal to the median level compared to those with below-median
levels (OR=1.13, 95 % CI: 1.01–1.28). When average APD levels were analysed in tertiles, however, the OR for
the highest tertile was not significantly raised compared with the bottom tertile (OR=1.10, 95% CI: 0.90–1.33).
The authors reported there was a sharp excess of cases at the 50–54th percentile, and therefore, a dichotomous
classification of exposure with the median as the cut-off led to raised ORs. For leukaemia, the greatest OR was
in subjects with over median level of exposure (OR=1.23, 95% CI: 0.99–1.52), but odds ratios were not
increased when exposure level was analysed in tertiles (1.00, 0.85, 0.82, respectively). For brain tumours, the
odds ratio associated with over median level of exposure was 1.14 (95% CI: 0.83–1.55), with odds ratios in
relation to tertiles being 1.00, 1.03, 1.14; none of these reached statistical significance. [This study had the
strength that it was population-based with very high completeness of ascertainment and that it encompassed a
large area. Even though exposure assessment included a measure of power density, a refinement over studies
only using distance from the transmitter, such assessment was still very crude as exposures were averaged over
towns. As with other ecological studies, information on confounders were only available at population level, no
information on individual risk factors for childhood cancer were available, and neither information on other
sources of RF].
2578
Studies with uncertainties related to inclusion criteria
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
An ecological study in Minas Gerais state, Brazil, investigated the spatial correlation between
mortality from cancer 1996–2006 and geographical concentration of mobile phone transmitter antennae (Dode et
al., 2011). Information on deaths from the City Health Department, geographical location of base stations from
the Brazilian Telecommunications Agency, and city census from the Brazilian Institute of Geography and
Statistics were geocoded according to ‘census tracts’ geographical zones. There were 7191 registered cancer
deaths at all ages and of types previously investigated with non-ionising radiation during the study period.
Residential address for cases of eligible cancer deaths were plotted against distance from the first transmitter that
the resident was possibly exposed. Mortality rates were derived within each radius of 100 meters (0–1000 m),
using the number of deaths divided by the estimated population in that radius. Mortality rate ratios were 1.35
within 100 meters, 1.08 within 500 meters, and declined with distance although no statistical test was applied to
assess statistical significance for a trend. No results were presented for cancer type-specific mortality rates.
Measured power density varied from 0.04 µW/cm2 to 40.78 µW/cm2. [While this study has the advantage that it
was conducted in a large area with a large numbers of transmitters, its ecological design and lack of data on
contributory factors still present difficulties in interpreting its results. A particular problem was the assessment of
the size of the population in each of the 100 m bands. As base stations are positioned in densely population
areas, it seems likely that the number of residents in close proximity to the transmitters have been
underestimated and those further away have been overestimated, which could have resulted in spurious raised
risks in the bands close to the transmitters. Calculations did not take into account differences in distribution of
age and sex, and other relevant factors such as socioeconomic status. Most deaths occurred within 1 to 2 years of
installation of the primary base stations; such a short induction period seems aetiologically unlikely. The study is
not included in the table, and is given little or no weight in the overall assessment.]
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
93
2600
12.1.4.3
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
Schüz et al. (2006b) conducted a population-based case-control study in the regions of Bielefeld,
Heidelberg, Mainz and Mannheim in Germany to investigate whether the low-level RF emitted from the base
station of DECT cordless phones is associated with risk of glioma and meningioma. The case-control study was
part of the Interphone Study and has been described elsewhere (e.g. in section 12.1.3, occupational exposure).
Cases diagnosed aged 30 to 59 years (October 2000–September 2001) or aged 30 to 69 years (October 2001–
October 2003) were recruited from the four neurosurgical clinics in the study region. Controls were selected
from compulsory population registers in the three regions according to the sex, age and region distribution of the
cases. Participation rates were 80% for glioma, 88% for meningioma, and 63% for controls. Information on past
exposures was collected by personal computer-assisted interview. Questions regarding cordless phones were
asked using a paper questionnaire developed from especially for the German study of Interphone. It collected
information on DECT and analogue cordless phones in the residence, make and model, start and stop dates of
use and the location of the base station within the residence. Exposure was classified into ‘low’ or ‘high’. ‘High’
was defined as a DECT base station located in the bedroom, 3 or less meters away from the bed or directly
adjacent to the wall of the neighbouring room next to the bed, and ‘low’ as it being positioned elsewhere or
absence of a DECT phone in the house. Because participants found it difficult to recall whether a phone was a
DECT phone, information on make and model of the phone was used to assess whether a cordless phone was
indeed ‘definitively a DECT phone’ or ‘possibly a DECT phone’. Two controls per case were ‘post hoc’
matched to cases by sex, birth year (± 2 years), and region. Analyses were conducted with conditional logistic
regression adjusted for sex, age, socioeconomic status, and living in a city. Interviews were conducted with a
next-of-kin for 10.9% of glioma cases, 1.3% of meningioma cases and 0.4% of controls. Three glioma cases and
five meningioma cases were considered exposed to ‘definitively a DECT phone’. Odds ratios were 0.50 (95%
CI: 0.14–1.76) for glioma and 1.09 (95% CI: 0.37–3.23) for meningioma. For possible or definitively DECT
phones, these odds ratios were 0.82 (95% CI: 0.29–2.33) and 0.83 (95% CI: 0.29–2.36) respectively, based on 5
exposed cases of each tumour type. For those in this category who reported first use more than five years ago
odds ratios were 0.68 (95% CI: 0.14–3.40) for glioma and 1.29 (95% CI: 0.37–4.48) for meningioma. [Due to
the small numbers of exposed cases in the study, and the difficulties in obtaining accurate recall of exposures,
this study only had sufficient statistical power to detect substantial risk increases, if these existed.]
Base stations of DECT phones
2628
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
94
Table 12.1.15. Ecological studies of leukaemia incidence or mortality in the vicinity of radio, TV or mobile phone transmitters
Country
Study population
Exposure
Time period
Design
source
Sydney, Australia
Incidence in 3 local
government areas
bordering antennae
(‘inner’) and 6
adjacent (‘outer’)
3 TV/FM radio
transmitters
1972–1990
135 000 (‘inner’)
450 000 (‘outer’)
residents
Exposure,
subgroup
No. of
exposed
cases
(/controls)
Relative risk (95%
CI)
All ages
Exposure area
Inner vs. outer
337/869
1.24 (1.09-1.40)
0-14 year
Exposure area
Inner vs. outer
33/101
1.58 (1.07-2.34)
Not reported
not reported
1.47 (0.98-2.19)
0.99 (0.59-1.64)
Comments
Reference
SIR in inner area for
childhood leukaemia:
Hocking et al.
(1996)
1.8, 95% CI: 1.2-2.5)
All ages
Sydney, Australia
1972–1990
Reanalysis of Hocking
(1996)
Sutton Coldfield, Great
Britain
1974–1986
Incidence in 16 local
government areas in
Sydney surrounding
transmitters
3 TV/FM radio
transmitters
0-14y
Incidence in
population resident
<10 km from
transmitter
Exposure area
all
without Lane Cove
Inner vs. outer
Inner vs. outer
McKenzie et
al. (1998)
1.38 (0.99-1.91)
0.90 (0.56-1.44)
Analyses at local
government level,
including larger
geographical area than
Hocking 1996 and
calculated RF exposure
levels
Dole et al.
(1997b)
23
304
1.83 (1.22-2.74)
1.01 (0.90-1.13)
P Stone test for
trend=0.001
(unconditional and
conditional)
2
97
O/E=2/1.1
O/E=34/29.7
P Stone test for
trend=0.001
(unconditional,0.052
(conditional)
Dolk et al.
(1997a)
RF exposure level
(continuous, units
not specified)
all
without Lane Cove
1 TV and FM radio ≥15 year
transmitter
Distance (km)
0-2
0-10
408 000 residents
0-14 year
Distance (km)
0-2
0-10
All ages
Great Britain
1974–1986
Incidence in
population resident
within 10 km of a
transmitter
20 TV and FM
radio transmitters
≥15 year
Distance (km)
0-2
0-10
79
3305
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
95
0.97 (0.78-1.21)
1.03 (1.00-1.07)
3.39 million residents
0-14 year
Distance (km)
0-2
0-10
All ages
Sutton Coldfield, Great
Britain
1987–1994
Reanalysis of Dolk et al
(1996) with updated cancer
data
Incidence in
population resident
<10 km from
transmitter
10
317
1 TV and FM radio All ages
transmitter
Distance (km)
0-2
0-10
20
333
Mortality 1987–1998
incidence 1987–1999
Mortality and
incidence in
population <10km of
transmitters
Vatican Radio
Station (multiple
transmitters 500600 kW)
49 656 residents
Mortality: > 14 year
incidence: 0-14 year
1
26
>15 year (mortality)
Distance (km)
0-2
2-4
4-6
6-8
8-10
1993–1996
Cooper et al.
(2001)
Incidence in 11 highpower and 44
adjacent control
areas, and 31 lowpower areas
ascertained from
census data
Total population size
not known
≥10 year
42 AM radio
transmitters in
total, high power
sites ≥100 kW,
low-power=50 kW
1.13 (0.03-6.27)
1.08 (0.71-1.59)
P Stone test for
trend=0.420
(unconditional).
SIR 0-6 km: 2.2 (1.0-4.1)
2
9
12
11
6
0-14 year
(incidence)
Distance (km)
0-2
2-4
4-6
6-8
8-10
South Korea
1.32 (0.81-2.05)
1.16 (1.04-1.29)
P Stone test for
trend=0.038
(unconditional) 0.409
(conditional).
All ages
0-14 year
Distance (km)
0-2
0-10
Rome, Italy
1.12 (0.61-2.06)
0.97 (0.87-1.08)
P Stone test for
trend=0.266
(unconditional)
Michelozzi et
al. (2002)
1.8 (0.3-5.5)
1.5 (0.7-2.7)
1.0 (0.5-1.7)
1.1 (0.6-1.8)
0.7 (0.3-1.5)
SIR 0-10 km: 1.2 (0.6-2.3)
1
2
5
0
0
6.1 (0.4-27.5)
2.3 (0.7-7.2)
1.9 (0.7-4.0)
-
Region according to
power of transmitter
High vs. Low-power
Not reported
1.5 (0.7-6.6)
SIRs for high-power
only, sites with:
100 kW
250 kW
500 kW
1500 kW
9
12
10
4
1.20 (0.55-2.28)
2.45 (1.27-4.29)
0.65 (0.31-1.19)
4.26 (1.16-10.89)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
96
SIRs with expected
values based on 5-year
age-specific incidence
rates of control areas.
Ha et al.
(2003)
South Korea
1994–1995)
Mortality in 10 areas
with a ≥100kW AM
transmitter, and
control areas
1.2 million exposed
6.9 million
unexposed
≥100 kW AM
transmitter
All ages
Exposed yes vs. no
55
1.70 (0.84-3.45)
0-14 year
Exposed yes vs. no
11
2.29 (1.05-5.98)
All ages
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
97
Park et al.
(2004)
Table 12.1.16. Case-control and cross-sectional studies of leukaemia incidence or mortality in the vicinity of radio, TV or mobile phone transmitters
Country
Study population
Exposure
Time period
Design
source
Hawaii, USA
Case-control study in LF radio (23.4 kHz)
3 census tracts
ascertained from
local health centre
covering >90% of
population
1979–1990, 1979–1990
Exposure,
subgroup
No. of
exposed
cases
(/controls)
Relative risk (95%
CI)
Distance (miles)
<2.6 (median)
12
2.0 (0.06-8.3)
Distance (km)
0-2
>2-4
>4-6
>6-8
>8-10
>10-20
>20
36
73
120
218
276
428
772
2.15 (1.00-4.67)
0.66 (0.44-0.99)
1.07 (0.77-1.49)
1.26 (0.96-1.65)
1.10 (0.85-1.41)
0.80 (0.65-0.99)
1.00 baseline
Comments
Reference
SIR analysis on same
cases: 2.09 (95% CI:
1.08-3.65)
Mascarinec et
al. (1994)
P trend=0.10
Ha et al.
(2007; 2008)
12 cases, 48 controls
0-15 year
South Korea
1993–1999
Hospital-based case- 31 AM radio
control study in 14
transmitters >20 kW
hospitals
1928 cases, 2064
controls
0-14 year
Total RF
(mV/m),
quartiles
<518.41
518.41-<624.35
624.35-<916.96
≥916.96
P trend=0.44
737
362
330
494
Peak RF
(mV/m),
quartiles
<274.31
274.31-<380.79
380.79-<608.35
≥608.35
1.00 baseline
0.75 (0.58-0.97)
0.70 (0.55-0.90)
0.83 (0.63-1.10)
P trend=0.43
737
362
330
494
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
98
1.00 baseline
0.95 (0.75-1.20)
0.86 (0.68-1.09)
1.02 (0.81-1.29)
West Germany
1984–2003
Population-based
case-control study
1959 cases, 5848
controls
0-14 year
16 AM and 8 FM/TV Exposure (V/m)
transmitters
<0.164 (90%)
0.164-<0.198
(90-94%)
0.198-0.185
(≥95%)
Distance (km)
0-<2
2-<6
6-<10
10-<15
≥15
Great Britain
1999–2001
Population-based
Macrocell mobile
case-control study
phone base stations
from cancer and birth
registers
527 cases of
leukaemia and nonHodgkin’s lymphoma,
2108 controls
0-4 year
Taiwan
2003–2007)
Population-based
Mobile phone base
case-control study
stations
from national hospital
admission and
insurance database
939 leukaemia
cases, 28 170
controls
Distance (m),
tertiles
<612
612-1071.7
>1071.7
5263
292
1.00 (baseline)
1.02 (0.80-1.31)
86
0.86 (0.67-1.11)
25
172
314
551
866
1.04 (0.65-1.67)
0.81 (0.66-0.99)
0.79 (0.67-0.93)
1.00 (baseline)
1.00 (0.88-1.14)
Merzenich et
al. (2008)
P trend=0.75
182
167
178
Modelled power
density (dBm),
tertiles
<-26.5
-26.5 to -17.6
≥-17.7 vs <26.5
Elliot et al.
(2010)
1.00 (baseline)
0.99 (0.78-1.27)
1.05 (0.71-1.35)
P trend=0.51
179
179
169
Average annual
power density
(Watt-year per
km2) over 5
years prior to
diagnosis year
Per 1 SD
increase
1.00 (baseline)
1.16 (0.90-1.48)
1.03 (0.79-1.34)
Li et al.
(2012a)
0.97 (0.87-1.08)
<168.7
(median)
≥ 168.7
1.00 (baseline)
1.23 (0.99-1.52)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
99
<92.2 (tertile 1)
92.2-<932.9
(tertile 2)
≥392.9 (tertile
3)
1.00 (baseline)
0.85 (0.68-1.07)
0.82 (0.59-1.13)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
100
Table 12.1.17. Ecological studies of brain and CNS tumour incidence or mortality in the vicinity of radio, TV or mobile phone transmitters
Country
Study population
Exposure
Time period
Design
source
Sydney, Australia
Incidence in 3 local
government areas
bordering antennae
(‘inner’) and 6
adjacent (‘outer’)
3 TV and FM
transmitter
1972–1990
135 000 (‘inner’),
450 000 (‘outer’)
residents
Exposure,
subgroup
No. of
exposed
cases
(/controls)
Relative risk
(95% CI)
All ages
Inner vs.
outer
740 (total)
0.89 (0.71-1.11)
0-14 year
Inner vs.
Outer
12/52
1.10 (0.59-2.06)
Comments
Reference
Hocking et al.
(1996)
All ages
Sutton Coldfield, Great
Britain
1974–1986
Incidence in
population resident
<10 km from mast
1 TV and FM radio
transmitter
408 000 residents
All ages
Great Britain
1974–1986
Incidence in
population resident
within 10 km of a
transmitter
20 TV and FM radio
transmitters
3.39 million residents
≥15 year
Distance
(km)
0-2
0-10
17
332
0-14 year
Distance
(km)
0-2
0-10
4
244
P Stone test for
trend=0.612
(unconditional)
Dolk et al.
(1997b)
P Stone test for
trend=0.465
(unconditional).
Dolk et al.
(1997a)
1.29 (0.80-2.06)
1.04 (0.94-1.16)
0.62 (0.17-1.59)
1.06 (0.93-1.20)
Results for adults not
reported.
1.8 (0.9-11.1)
Brain cancer only, SIRs
Ha et al. (2003)
with expected values
based on 5-year agespecific incidence rates of
control areas
All ages
South Korea
1993–1996
Incidence in 11 highpower and 44
adjacent control
areas, and 31 lowpower areas
ascertained from
42 AM radio
transmitters in total,
high power sites
≥100 kW, lowpower=50 kW
Region
according to
power of
transmitter
High vs.
low-power
Not reported
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
101
census data
SIRs for
high-power
only, sites
with:
100 kW
250 kW
500 kW
1500 kW
Total population size
not known
All ages
South Korea
1994–1995(
Mortality in 10 areas
with a ≥100kW AM
transmitter, and
control areas
≥100 kW AM
transmitter
16
10
16
3
Exposed,
Yes vs. no
All ages
0-14 year
2.27 (1.30-3.67)
0.86 (0.41-1.59)
1.47 (0.84-2.38)
2.19 (0.45-6.39)
Park et el.
(2004)
30
2
1.52 (0.61-3.75)
2.12 (0.44-12.06)
Not reported
No significant
differences in SIR
between three
exposure
categories,
p=0.17
1.2M exposed, 6.9M
unexposed
All ages
Bavaria, Germany
2002-2003
Incidence in 48
municipalities with
completeness of
cancer registrations
>90%
177428 residents
All ages
Mobile phone base
stations
Exposure
level of
municipality
37
municipaliti
es with no,
8 with little
and 3 with
moderate
exposure
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
102
Meyer (2006)
Table 12.1.18. Case-control studies of brain and CNS tumour incidence or mortality in the vicinity of radio, TV or mobile phone transmitters
Country
Study population
Exposure
Time period
Design
source
South Korea
Hospital-based case- 31 AM radio
control study in 14
transmitters >20 kW
hospitals
1993-1999
956 cases, 1020
controls
0-14 year
Exposure,
subgroup
No. of
exposed
cases
(/controls)
Distance
(km)
0-2
>2-4
>4-6
>6-8
>8-10
>10-20
>20
10
32
59
90
114
244
400
Total RF
(mV/m),
quartiles<5
32.55
532.55<622.91
622.91<881.07
≥881.07
Great Britain
1999–2001
Population-based
Macrocell mobile
case-control study
phone base stations
from cancer and birth
registers
251 cases of brain
and CNS
tumours,1004
controls
0-4 year
Relative risk
(95% CI)
Comments
Reference
P trend=0.76
Ha et al. (2007)
1.42 (0.38-5.28)
1.40 (0.77-2.56)
1.02 (0.66-1.57)
1.26 (0.96-1.65)
1.08 (0.73-1.59)
0.94 (0.67-1.33)
1.00 baseline
P trend=0.73
329
185
1.00 baseline
0.66 (0.47-0.92)
181
0.72 (0.51-1.01)
254
0.77 (0.54-1.10)
Distance
(m), tertiles
P trend=0.75
85
85
81
<612
612-1071.7
>1071.7
Modelled
power
density
(dBm),
tertiles
<-26.5
-26.5 to 17.6
≥-17.7 vs <26.5
1.00 (baseline)
0.95 (0.67-1.34)
0.95 (0.65-1.38)
P trend=0.33
93
80
1.00 (baseline)
0.97 (0.69-1.37)
78
0.76 (0.51-1.12)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
103
Elliot et al.
(2010)
France
2005
Hospital-based case- Mobile phone base
control study from
stations
two main centres
122 cases of glioma,
122 controls
Taiwan
2003–2007
Population-based
Mobile phone base
case-control study
stations
from national hospital
admission and
insurance database
394 brain neoplasm
cases, 11 820
controls
Residence
<500 from
mobile
phone base
station
Yes vs. no
19
0.49 (0.26-0.92)
Average
annual
power
density
(Watt-year
per km2)
over 5
years prior
to diagnosis
year
Spinelli et al
(2010)
Li et al. (2012a)
Per 1 SD
increase
1.09 (0.95-1.25)
<168.1
(median)
≥ 168.1
1.00 (baseline)
<94.0
(tertile 1)
≥392.9
(tertile 2)
≥392.9
(tertile 3)
1.00 (baseline)
1.14 (0.83-1.55)
1.03 (0.73-1.45)
1.14 (0.70-1.85)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
104
2629
2630
12.1.5
Incidence studies
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
Studies of changes in incidence trends over time can be informative provided that the exposure
prevalence is sufficiently high in the general population and has changed substantially over a short time period,
and that population based cancer registries with good coverage are available, which are not in a built-up phase.
With regard to mobile phone use, the exposure prevalence in many countries worldwide has changed from nonexistent to close to 100% of the population in less than two decades. If RF exposure from mobile phone use is
associated with an increased risk of cancer, e.g. brain tumours, this should be reflected in increasing incidence
trends unless the risk is restricted to a small subgroup of the population, e.g. extremely heavy users, or appears
only after a very long induction time period. This type of ecological studies cannot, however, replace analytic
study designs such as cohort or case-control studies, which collect exposure information on an individual level,
but can under certain circumstances be used as a consistency check for results obtained with analytic study
designs. Therefore, only incidence studies that cover a sufficiently long time period (at least a few years into the
2000s) to capture potential effects from mobile phone use of durations reported in some analytical
epidemiological studies are discussed below. Several studies have been conducted that cover earlier time
periods, and end in the late 1990s or early 2000s, which is too early to provide data of relevance for evaluation of
an effect of mobile phone use on the incidence of brain tumours, and especially of a slow growing tumour like
acoustic neuroma. These studies have in common that they find no increased incidence of brain tumours in agegroups where mobile phone use was most common (Cook et al., 2003; Röösli et al., 2007), while for acoustic
neuroma a gradual increase in the incidence was found from the 1970s/1980s that coincides with the introduction
of CT and later MRI, and the increasing availability of these diagnostic techniques (Evans et al., 2005; Hardell et
al., 2003; Nelson et al., 2006; Propp et al., 2006).
2651
2652
2653
2654
For periodic updates of incidence trends from the same cancer registry restriction is made to the most
recent publication, unless an earlier publication provide information of relevance that are not available in the
latest update. Incidence studies performed to specifically discuss potential effects of mobile phone use are
included, provided they fulfil the inclusion criteria.
2655
12.1.5.1
2656
2657
2658
2659
Brain tumour incidence studies from various countries have been published over a number of years,
and some more recent studies have been performed to specifically investigate the consistency between observed
incidence trends and results obtained in some published case-control studies of mobile phone use and glioma
risk.
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
Deltour and colleagues (Deltour et al., 2012) analysed age standardized glioma incidence trends for
men and women 20–79 years old in the Nordic countries 1979–2008, using joinpoint regression analyses.
Through simulations, estimations were also made of the probability of detecting, as a significant increase in the
glioma incidence rate, various magnitudes of increased risks of glioma related to mobile phone use, specifically
risk levels that had been reported in a few case-control studies. Data on glioma incidence were collected from the
national cancer registers, and prevalence of mobile phone use was estimated from data collected in the Nordic
Interphone studies. From the Nordic Interphone studies information was abstracted about the proportion of
persons who used a mobile phone regularly and the proportion of heavy users (at least 1640 cumulative hours)
for each year from 1980 to 2002 by sex and age group. For the period following the end of the data collection for
Interphone, i.e. 2003-2008, mobile phone use was extrapolated from 2002 as rising by 3% annually, which is
slightly slower than the average over the years 1990-2000. The results showed that over the whole period, the
glioma incidence had increased on average by 0.4% per year (95% CI 0.1–0.6%) among men and 0.3% (95% CI
0.1–0.5%) among women. For men in the age group 20–39 years, the incidence had decreased since 1987, in the
age group 40–59 it had been stable over the whole study period, and among men 60–79 years a slight increase in
the incidence was observed. Also for women the increased incidence was primarily observed in the oldest age
group. The probability to detect an increased risk in incidence trends was 100% for a relative risk of 2.0 or
higher associated with mobile phone use with an induction period of up to 15 years, and a relative risk of 1.5 or
higher with an induction period of up to 10 years, and 100% to detected a relative risk of 1.2 or higher with up to
5 years induction period. For heavy mobile phone use, corresponding to the highest exposure category in the
Interphone study (>1640 cumulative hours), the probability was 100% to detect a relative risk of 2.0 with up to 5
years induction period, and 98% to detect a relative risk of 1.5 with up to one year induction time. [This means
that if the risk increases that were reported in a few case-control studies to be associated with mobile phone use
were real, they would have resulted in a detectable increase in the glioma incidence in the Nordic countries. The
Brain tumours
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
105
2683
2684
Nordic countries were among the earliest to adopt the mobile phone technology in the general population. The
study did not provide information for longer induction periods than up to 15 years.]
2685
2686
2687
Ahlbom and Feychting presented glioma incidence trends in Sweden during the time period 1970–
2009 (Ahlbom & Feychting, 2011), stratified on age in three groups (20–39, 40–59, ≥60), based on data from the
Swedish national cancer registry. No indication of increased incidence during the later years was found.
2688
2689
2690
2691
2692
2693
2694
2695
In an earlier study, Deltour et al. (Deltour et al., 2009) reported incidence trends in the Nordic
countries for glioma and meningioma during the period 1974-2003. While the relevant data for glioma are
included in the study discussed above, meningioma incidence rates were not covered. For meningioma, the
incidence increased slightly over the whole study period among men (annual percent change 0.8), and more
pronounced among women (annual percent change 2.9 during 1974–1987, and 3.8 during 1990–2003), and did
not exhibit any change related to the introduction of mobile phones. [Improved diagnostic techniques, like CT
and MRI, are likely to affect the incidence rate for meningioma, which is a slow growing and generally benign
tumour. The study period is likely too short to detect any effect in a slow-growing tumour like meningioma.]
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
Little and colleagues (Little et al., 2012) compared the observed glioma incidence rate among adults
(18 years or older) in the US during the period 1997–2008 with projected incidence rates estimated from relative
risks reported in the study by Hardell et al. with pooled data from two case-control studies (Hardell, Carlberg &
Hansson Mild, 2011a) and from the Interphone study (Interphone Study Group, 2010). Glioma incidence data
were collected from 12 registries in the US population based Surveillance, Epidemiology, and End Results
(SEER) programme. The glioma incidence was stable over the study period with an average change of -0.02 per
year (95% CI -0.28% to 0.25%). A decreased incidence was observed for low grade glioma, -3.02% per year
(95% CI -3.49% to -2.54%), and for tumours with a poorly specified location, -2.35% (95% CI -2.81% to 1.89%), and an increased incidence for glioma in the temporal lobe, 0.73% per year (95% CI 0.23–1.23%) and
other specified sites, 0.79% per year (95% CI 0.40–1.19%). There was no accelerating increase in the incidence
of temporal lobe tumours when comparing the incidence before and after 1996. The predicted incidence rate,
from estimations based on relative risks, latency periods and cumulative hours of use from the Hardell study
showed a 44.5% higher predicted than observed incidence rate. For temporal lobe tumours the predicted
incidence rate was 30.6% higher than the observed incidence, and for astrocytoma the predicted incidence rate
was 58.3% higher than the observed. The corresponding predictions using the modest risk increases and risk
decreases in the Interphone study were compatible with the observed incidence rate. [The study provides
evidence against an association between mobile phone use and glioma risk of the magnitude and with the latency
periods reported by Hardell et al. A risk increase restricted to a small subgroup of the population, or after a
longer latency period than has so far been studied, would not be detected in the US incidence trends. Mobile
phone use in the US became widespread in the general population later than in the Nordic countries, e.g. in the
early 2000s mobile phone subscriptions per capita was around 40% in the US compared to 80% in Sweden
(Swerdlow et al., 2011).]
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
Incidence studies from the US were also reported by Inskip and colleagues (Inskip, Hoover & Devesa,
2010), covering the time period 1977–2006, and by Kohler and colleagues (Kohler et al., 2011), for the period
1975–2007. The data included in these studies are also covered in the study by Little et al., although these
studies included a wider age range, and presented some additional analyses. Inskip et al. report age-specific
incidence trends separately for the periods 1977–1991 and 1992–2006, where the early period covers the
introduction of CT and MRI, while the second period covers the introduction of mobile phones. An increased
brain cancer incidence during the early period was observed in the youngest age groups (<20 years and 20–29
years) and in the oldest (≥65 years). In the later period, brain cancer incidence showed a slightly downward trend
in most age groups, although not statistically significant. In women aged 20–29, however, an upward trend was
observed. Kohler et al. used joinpoint regression analyses, and observed an increased rate of neuroepithelial
tumours, which are mostly glioma, by 1.9% per year from 1980 to 1987, and a decreased rate by 0.4% per year
1987–2007. Incidence trends differed among histological groups. [Improvement and increased availability of
new diagnostic techniques and changes in coding and classification practices may affect incidence trends for
specific histological subgroups.]
2732
2733
2734
2735
2736
2737
Dolecek and colleagues (Dolecek et al., 2012) published brain tumour incidence trends in the US for
the time period 2005–2009 based on data from the Central Brain Tumor Registry of the United States
(CBTRUS). CBTRUS covers the largest collection of population based data on the incidence of primary brain
and CNS tumours in the US, and receives data from all healthcare data sources, for example from the SEER
program. No significant changes in the incidence of all primary malignant or benign brain and central nervous
system (CNS) tumours were observed during the study period. The age-adjusted incidence rate of malignant
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
106
2738
2739
brain and CNS tumours was 7.4/100 000 in 2005 and 7.1/100 100 in 2009, while corresponding results for
benign brain and CNS tumours was 12.7/100 000 and 13.8/100 000, respectively.
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
In a study from Israel, Barchana and colleagues (Barchana, Margaliot & Liphshitz, 2012) described
incidence trends for low and high grade glioma from 1980 to 2009 based on data from the national cancer
registry, and changes in the tumour laterality during this time period. In addition, a survey was conducted to
assess the preferred side of the head for mobile phone use in a random sample of 1000 Israeli adults. The
majority, 70%, reported that they used the mobile phone mostly or always on the right side, while 13% used it
equally often on both sides, and 17% reported use mostly or always on the left side. The penetration of mobile
phone use, measured as number of mobile phones per total population (including also children and elderly), was
2% in 1994, 29% in 1997, 67% in 2000, and reached 100% in 2003. Incidence rates were reported for 5-year
periods, three periods before 1994, and three periods during the time when mobile phone use increased
substantially in the population. Over the whole study period, the incidence of low grade glioma decreased
significantly both among men and women, with the sharpest decrease during the period 1994–2009. The
incidence of high-grade glioma increased over the whole time period 1980–2009, but the increase levelled off
during the latter part of the period, especially among men. For all glioma combined, the annual incidence among
men in the period 1990–1994 was 6.86/100 000 and in 2004–2009 7.21/100 000 population. The corresponding
numbers for women were 5.07/100 000 and 5.1/100 000, respectively. Tumour laterality changed from being
more common on the right side in the period 1980–1989 to be more common on the left side during 2000–2008
(data were not available for 2009), although differences were small, and for about one third of the patients
tumour side was not recorded. The shift towards left sided tumours was more marked among persons diagnosed
in the ages 20–49 years. [Missing data on tumour side is unlikely to differ between right and left sided tumours,
thus, the data do not indicate that right sided tumours have increased, which would have been expected if
localized exposure from mobile phone use affected glioma occurrence. The modest increase in glioma incidence,
most pronounced during the early study period, before the introduction of mobile phones, does also not support
an effect of mobile phone use, although an effect after a long latency period would not be detected. Access to
information about laterality of tumours is a strength of this study.]
2764
2765
2766
2767
2768
2769
2770
2771
2772
Brain cancer incidence time trends in England between 1998 and 2007 were reported by de Vocht and
colleagues (de Vocht, Burstyn & Cherrie, 2011), based on cancer incidence data from the UK Office of National
Statistics. Only malignant brain cancers were included, which means the majority of cancers were glioma.
Between 1990 and 2002, mobile phone use in the UK increased from 0 to 65%. No statistically significant
change in brain cancer incidence was found, either for men or women, or in different age groups. A slight
increase in the incidence of tumours in the temporal and frontal lobes was observed, and a decrease in the
incidence of tumours in overlapping lesions of the brain, parietal lobe, cerebrum, and cerebellum. [A shift in
incidence between tumour locations may reflect improvements in diagnostic methods used, e.g. increased access
to MRI scanning, which is indicated by the decrease in overlapping tumours.]
2773
2774
2775
2776
2777
2778
2779
2780
2781
Dobes and colleagues conducted a multicentre study of brain tumour incidence 2000–2008 in parts of
Australia, New South Wales and the Australian Capital Territory (Dobes et al., 2011). Data were collected
retrospectively through pathology databases serving neurosurgical centres in these areas, and only histologically
confirmed cases were included. Age-standardized incidence rates were calculated and trends analysed with
joinpoint analysis. A significant increase in the incidence of malignant brain tumours was observed, with an
annual percent change of 2.3 in both men and women, which was largely due to an increased incidence in the
oldest age group, ≥65 years. [Incidence rates are still lower than in the US, which may reflect an underascertainment of cases through less intensive use of imaging technologies, retrospective case identification, and
omission of tumours that are not histologically confirmed.]
2782
2783
2784
A report from the New South Wales population-based cancer registry (Currow & Thomson, 2014)
described no significant change in the incidence for brain cancer (ICD-O-3 C71) in New South Wales during the
period 2000–2009. For men the annual percent change was -0.5 and for women -0.4.
2785
2786
2787
2788
2789
2790
2791
2792
Ding and colleagues (Ding & Wang, 2011) studied the incidence of brain and nervous system tumours
in urban Shanghai, China, 1983–2007, using data collected from the Shanghai cancer registry, to which it is
mandatory to report all new cases of cancer. Joinpoint analysis was used to describe incidence trends and detect
changes in the trends. Brain and nervous system tumour incidence increased over the whole studied period, but
no significant change in the increase was observed. The annual percent change was 1.2 in men and 2.8 in
women. Mobile phones were introduced in 1987, and there were 100 000 subscribers in 1995, 1 million in 1998
and 10 million in 2000. The authors concluded that their results do not support an effect of mobile phone use, as
the incidence rates continued a gradual increase that started well before the introduction of mobile phones.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
107
2793
2794
2795
2796
2797
Nomura and colleagues (Nomura, Ioka & Tsukuma, 2011) studied the incidence rates of intracranial
tumours in Osaka between 1975 and 2004, using data from the Osaka Cancer Registry, and analysing trends with
joinpoint analysis. The incidence rate of malignant intracranial tumours increased significantly between 1975
and 1986 (annual percent change 3.5), and decreased significantly from 1986 to 2004 (annual percent change 1.4). A similar pattern was seen for all intracranial tumours combined (malignant and benign).
2798
2799
2800
2801
Saika and colleagues (Saika & Katanoda, 2011) described brain and central nervous system (CNS)
cancer mortality rates in 11 countries 1990–2006, based on information from the WHO database. Mortality rates
were stable or decreasing in all countries except Russia and for women in Spain. [Mortality rates are affected
both by changes in incidence and in survival.]
2802
Studies with uncertainties related to inclusion criteria
2803
2804
2805
The study by Lehrer and colleagues (Lehrer, Green & Stock, 2011) is not included in the evaluation
because the authors correlated the number of brain tumour cases with the number of mobile phone subscriptions
in 19 states in the US, but did not adequately consider population size, age and sex distribution.
Table 12.1.19. Incidence studies – brain tumours
Country/
location
Time period Outcome Cancer
Cancer trend
incidence data
Age range
Comments
Reference
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
108
Denmark
1979–2008
Finland
20–79 year
Glioma
Norway
Sweden
Incidence rates Annual % change (95%
from Nordic
CI):
national cancer Men
registries
1979–2008: 0.4 (0.1, 0.6)
20–39 years:
1979–1987: 3.8 (0.5, 7.2)
1987–2008: -0.7 (-1.4, 0.1)
40–59 years:
1979–2008: 0.1 (-0.2, 0.3)
60–79 years:
1979–2008: 0.9 (0.5, 1.2)
Women
Joinpoint analyses.
High quality cancer
registration.
(Deltour et al.,
2012)
Nordic countries
among the earliest to
adopt mobile phone
technology in the
general population.
Latency period up to
15 years.
No evidence of an
impact of mobile
phone use on glioma
incidence.
1979–2008: 0.3 (0.1, 0.5)
20–39 years:
1979–2008: 0.3 (-0.2, 0.8)
40–59 years:
1979–2008: 0.0 (-0.2, 0.2)
60–79 years:
1979–2008: 0.6 (0.3, 0.9)
Projections based on
simulations:
1 year induction period:
Ever use: 100% probability
to detect RR 1.2 or higher
and 0.8 or lower
Heavy use (>1640 h):
100% probability to detect
RR 2.0 or higher, 98%
probability to detect RR 1.5
or higher
5 years induction period:
Ever use: 100% probability
to detect RR 1.2 or higher
and 0.8 or lower
Heavy use (>1640 h):
100% probability to detect
RR 2.0 or higher
10 years induction period:
Ever use: 100% probability
to detect RR 1.5 or higher,
96% probability to detect
1.2 or higher
15 years induction period:
Ever use: 100% probability
to detect RR 2.0 or higher
Sweden
1970–2009
20–39 year
40–59 year
Glioma
Incidence rates No indication of increased
from Swedish
incidence during later
cancer registry years.
≥60 year
High quality cancer
registration.
Latency period over
15 years.
(Ahlbom &
Feychting,
2011)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
109
Denmark
1974–2003
Finland
20-79 year
Norway
Glioma
and
meningio
ma
Incidence rates
from Nordic
national cancer
registries
Sweden
Glioma results included in
Deltour et al. 2012 above
Meningioma, annual %
change (95% CI):
Men
1974–2003: 0.5 (0.2, 0.8)
Women
1974–1987: 2.9 (2.2, 3.7)
1987–1990: -2.1 (-8.1, 4.2)
1990–2003: 3.8 (3.2, 4.4)
USA
1997–2008
≥18 year
Glioma
Incidence rates Annual % change (95%
from SEER
CI):
Overall: -0.02 (−0.28, 0.25)
High grade:
0.64 (0.33, 0.95)
Low grade:
−3.02 (−3.49, −2.54)
Temporal lobe:
0.73 (0.23, 1.23)
Other specified sites:
0.79 (0.40, 1.19)
Poorly specified location:
−2.35 (−2.81, −1.89)
Joinpoint analyses.
High quality cancer
registration.
(Deltour et al.,
2009)
Improved diagnostic
techniques (CT,
MRI) are likely to
affect the incidence
rate for meningioma.
Latency period up to
10 years.
No evidence of an
impact of mobile
phone use on glioma
or meningioma
incidence.
Results from Hardell (Little et al.,
et al. 2011 not
2012)
compatible with
observed incidence.
Mobile phone use in
the US became
widespread in the
general population
later than in the
Nordic countries.
Latency period over
10 years.
No evidence of an
impact of mobile
phone use on glioma
incidence.
Projections based on
results from Hardell et al.
2011:
Predicted incidence rate for
glioma overall 44.5%
higher than observed
For temporal lobe tumours:
30.6% higher than
observed
For astrocytoma: 58.3%
higher than observed
Projections based on
results from Interphone,
2010:
Compatible with observed
incidence
USA
1977–2006
Separate
trends for
1977–1991
and 1992–
2006
All ages,
age specific
analyses
Brain
Incidence rates
cancer
from SEER
(excluding
meningio
ma and
other
benign
brain
tumours)
1977–1991: Increased
brain cancer incidence in
the youngest age groups
(<20, 20-29) and oldest
(≥65)
1992–2006: Slightly
downward trend in most
age groups. Upward trend
in women 20-29 years
(frontal lobe tumours)
Latency period up to (Inskip, Hoover
10 years.
& Devesa,
2010)
No evidence of an
impact of mobile
phone use on glioma
incidence.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
110
USA
1975–2007
≥20 year
Neuroepithelial
tumours
(mostly
glioma
Incidence rates Annual % change:
from SEER
1980–1987: 1.9 (p<0.05)
1987–2007: -0.4 (p<0.05)
Joinpoint analyses.
(Kohler et al.,
2011)
No evidence of an
impact of mobile
phone use on glioma
incidence.
Improvement in
diagnostic
techniques and
changes in coding
and classification
practices may affect
incidence trends for
specific histological
subgroups.
USA
2005–2009
All ages
Brain
tumours
Incidence rates
from CBTRUS
(includes also
SEER)
No significant changes in
the incidence of all primary
malignant or benign brain
and central nervous system
(CNS) tumours
Annual incidence rate,
malignant:
Latency period over
10 years.
(Dolecek et al.,
2012)
No evidence of an
impact of mobile
phone use on glioma
incidence.
2005: 7.4/100 000
2009: 7.1/100 000
Annual incidence rate,
benign:
2005: 12.7/100 000
2009: 13.8/100 000
Israel
1980–2009
Glioma
Incidence rates
from the Israeli
national cancer
registry
Information
about tumour
laterality 19802008
Low grade glioma
decreased significantly
both among men and
women, sharpest decrease
during 1994–2009
High-grade glioma
increased slightly over the
whole time period, but the
increase levelled off during
the latter part of the period
Incidence rate, all glioma
combined:
Men:
1990–1994: 6.86/100 000
2004–2009: 7.21/100 000
Women:
1990–1994: 5.07/100 000
2004–2009: 5.1/100 000
Tumour laterality:
Tumour laterality
missing for 1/3 of
tumours.
(Barchana,
Margaliot &
Liphshitz, 2012)
Mobile phone
penetration: 2% in
1994, 29% in 1997,
67% in 2000, 100%
in 2003.
Survey showed 70%
mobile phone use on
right side.
Latency period over
10 years.
No evidence of an
impact of mobile
phone use on glioma
incidence.
Information about
tumour laterality a
strength.
1980–1989: right side
tumours more common
2000-2008: left side
tumours more common
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
111
UK
1998–2007
All ages,
age specific
analyses
Malignant
brain
cancer
Incidence rates
from the UK
Office of
National
Statistics
No statistically significant
change in brain cancer
incidence was found, either
for men or women, or in
different age groups.
A slight increase in the
incidence of tumours in the
temporal and frontal lobes,
and a decrease in the
incidence of tumours in
overlapping lesions of the
brain, parietal lobe,
cerebrum, and cerebellum.
Latency period up to (de Vocht,
10 years.
Burstyn &
Cherrie, 2011)
A shift in incidence
between tumour
locations may reflect
improvements in
diagnostic methods
used, e.g. increased
access to MRI
(indicated by the
decrease in
overlapping
tumours).
No evidence of an
impact of mobile
phone use on glioma
incidence.
Australia,
New South
Wales and
the
Australian
Capital
Territory
2000–2008
Australia,
New South
Wales
2000–2009
China,
Shanghai
1983–2007
All ages,
age specific
analyses
All ages
Malignant
brain
tumours
Brain
cancer
Brain and
nervous
system
tumours
Retrospectively
collected data
through
pathology
databases
serving
neurosurgical
centres in
these areas.
Only
histologically
confirmed
cases.
Annual % change:
Joinpoint analyses.
2.3 among both men and
women (statistically
significant)
Increased incidence
largely due to an
increase in the
oldest age group
(≥65 years).
Incidence rates
from the New
South Wales
population
based cancer
registry
Annual % change:
(Dobes et al.,
2011)
No evidence of an
impact of mobile
phone use on glioma
incidence.
Men: -0.5 (not significant)
Women: -0.4 (not
significant)
Incidence rates Annual % change (95%
from the
CI):
Shanghai
Men: 1.2 (0.4, 1.9)
cancer registry
Women: 2.8 (2.1, 3.4)
No evidence of an
(Currow &
impact of mobile
Thomson,
phone use on glioma 2014)
incidence.
Joinpoint analyses.
Incidence increased
gradually over the
whole study period,
no significant
change after
introduction of
mobile phones.
(Ding & Wang,
2011)
No evidence of an
impact of mobile
phone use on glioma
incidence.
Japan,
Osaka
1975–2004
All ages
Intracrani
al
tumours
Incidence rates Annual % change (95%
from the Osaka CI):
cancer registry Malignant tumours
1975-1986: 3.5 (1.8, 5.3)
1986-2004: -1.4 (-2.2, -0.6)
11 countries
1990–2006
Central
nervous
system
cancer
Mortality rates
from the WHO
database
Mortality rates were stable
or decreasing in all
countries except in Russia
and for women in Spain.
Joinpoint analyses.
(Nomura, Ioka
& Tsukuma,
No evidence of an
2011)
impact of mobile
phone use on glioma
incidence.
Mortality rates are
affected by changes
both in incidence
and survival.
(Saika &
Katanoda,
2011)
2806
2807
Children and adolescents
2808
2809
2810
2811
2812
Ward and colleagues (Ward et al., 2014) reported childhood cancer incidence rate trend analyses in
the US based on data from the SEER 9 registries during the period 1975–2010. No significantly increasing trend
was found in the incidence rate of brain and central nervous system tumours in children and adolescents aged 0–
19 years, confirming earlier observations made by Inskip et al., with follow-up through 2006 (Inskip, Hoover &
Devesa, 2010), and Kohler et al. with follow-up through 2007 (Kohler et al., 2011).
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
112
2813
2814
2815
2816
2817
2818
McKean-Cowdin and co-workers (McKean-Cowdin et al., 2013) studied the incidence rates of
malignant childhood (0–14 years) brain tumours in the US 1973 through 2009, using data from nine SEER
registries, and presented more detailed data than Ward et al. described above. Trends were analysed using
joinpoint regression. The incidence rate was quite stable from 1973–1982, increased significantly between 1983
and 1986, and was again stable during 1987 through 2009 (annual percent change 0.10; 95 % CI -0.39 to 0.61 for
the latter period).
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
Aydin and colleagues (Aydin et al., 2011) presented age and sex standardized brain tumour incidence
rates in Sweden for children and adolescents aged 5–19 years during the period 1990 through 2008, and
estimated projected incidence rates under the assumption of increased risks associated with mobile phone use
(OR 1.36 and 2.15, as was observed in some analyses in the case-control study described in section 12.1.2.1).
While the observed incidence rate was stable and even declined slightly during the study period, the projected
incidence rates showed a slight increase under the assumption of an increased risk by 1.36, and considerably
increased projected incidence rates during the latter part of the study period assuming an OR of 2.15. In a
separate publication, Aydin and colleagues presented age-adjusted brain and central nervous system tumour
incidence rates for children and adolescents aged 5–19 years in the Nordic countries from 1990–2009, based on
data from NORDCAN (Aydin et al., 2012). During the study period, incidence rates were stable for both boys
and girls. They also estimated the proportion of regular mobile phone users in these countries, which reached
40% in the beginning of the 2000s.
2831
2832
2833
2834
2835
2836
2837
Rosychuk and co-workers (Rosychuk et al., 2012) describe childhood central nervous system
incidence trends in Alberta, Canada, during the period 1982 to the beginning of 2004, including children and
adolescents <20 years. Data was collected from the population based Alberta Cancer Registry, to which
registration of new cases is mandatory. The incidence rates increased over time, from 2.1/100 000 in 1983/1984
(95% CI 1.3-3.4) to 4.3/100 000 in 2003/2004 (95% CI 3.1–5.9). The increase occurred as a gradually increasing
trend over the whole study period. [The time period of follow-up is likely too short for any potential effect of
mobile phone use to appear.]
2838
2839
2840
The study of brain cancer incidence time trends in England between 1998 and 2007, discussed above,
reported age-specific results, including incidence trends for children 0-9 years and adolescents 10–20 years (de
Vocht, Burstyn & Cherrie, 2011). No increase in the brain cancer incidence was observed.
2841
2842
2843
The study of brain tumour incidence 2000–2008 in parts of Australia, New South Wales and the
Australian Capital Territory, discussed above, also reported results for the age group 0-19 years (Dobes et al.,
2011), with no evidence of an increasing incidence trend in age group.
2844
2845
2846
The study of incidence rates of intracranial tumours in Osaka between 1975 and 2004, discussed
above (Nomura, Ioka & Tsukuma, 2011), found no increase of the incidence of primary intracranial tumours in
the age group 0-19 years.
Table 12.1.20. Incidence studies – brain tumours in children and adolescents
Country/
location
Time
period
Outcome
Cancer
incidence
data
Cancer trend
Brain and
central
nervous
system
tumours
Incidence
rates from
SEER
No significantly
Joinpoint analyses.
increasing trend in the
incidence rate of brain
and central nervous
system tumours.
(Ward et al.,
2014)
Primary
malignant
brain
tumours
Incidence
rates from
SEER
Stable incidence rate Joinpoint analyses.
1973–1982, significant
increase 1983–1986,
stable 1987–2009
McKeanCowdin et al.,
2013)
Age range
USA
1975–2010
0–19 year
USA
1973–2009
0–14 year
Comments
Reference
Annual % change
(95% CI) 1987–2009:
0.10 (-0.39, 0.61)
USA
1992–2007
0–19 year
Brain
cancer
Incidence
rates from
SEER
Annual % change
(95% CI):
-0.26 (-0.91, 0.40)
APCs were calculated
using weighted least
squares method.
(Kohler et al.,
2011)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
113
USA
1977–2006
0–19 year
Brain
cancer
Brain
tumours
Incidence
rates from
SEER
Denmark
1990–2008
Finland
(Sweden)
Norway
1990-2009
Sweden
(all four
countries)
Incidence
rates from
Nordic
national
cancer
registries
5–19 year
NORDCAN
Annual % change
(95% CI):
1977–1991: 1.93
(0.57, 3.31)
The three US studies
overlap, but present
different analyses and
time periods.
1992–2006: -0.42 (1.84, 1.02)
No evidence of an impact
of mobile phone use on
brain tumour incidence
from any of the three US
incidence studies.
Sweden:
Age-standardized
incidence rates.
Based on assumption
RR=1.36 projected
rates showed slight
increase, and
RR=2.15 considerable
increase, while
observed incidence
was stable or
declining.
Nordic countries among
the earliest to adopt
mobile phone technology
in the general population.
(Inskip, Hoover
& Devesa,
2010)
(Aydin et al.,
2011; Aydin et
al., 2012)
Latency period up to 15
years.
No evidence of an impact
of mobile phone use on
Four Nordic countries: brain tumour incidence.
Stable incidence rates
for both boys and girls.
Canada,
Alberta
UK
1982–
Central
beginning of nervous
2004
system
tumours
0–19 year
1998–2007
<10 year
10–20 year
Australia,
New South
Wales and
the
Australian
Capital
Territory
2000–2008
Japan,
Osaka
1975–2004
0-19 year
0-19 year
Incidence
rates from
Alberta
cancer
registry
Malignant Incidence rates
brain
from the UK
cancer
Office of
National
Statistics
Incidence rates
increased over time,
from 2.1/100 000 in
1983/1984 (95% CI
1.3-3.4) to 4.3/100 000
in 2003/2004 (95% CI
3.1–5.9).
The increase occurred as
a gradually increasing
trend over the whole
study period.
No statistically
significant change in
brain cancer
incidence was found.
Latency period up to 10
years.
Time period of follow-up is
likely too short for any
potential effect of mobile
phone use to appear.
No evidence of an impact
of mobile phone use on
glioma incidence.
Malignant Retrospectively No increase in the
brain
collected data
incidence in this age
tumours
through
group.
pathology
databases
serving
neurosurgical
centres in these
areas. Only
histologically
confirmed
cases.
Joinpoint analyses.
Intracrani
al
tumours
Joinpoint analyses.
Incidence rates
from the Osaka
cancer registry
Annual % change
(95% CI):
(Rosychuk et
al., 2012)
No evidence of an impact
of mobile phone use on
glioma incidence.
No evidence of an impact
1975-2004: -0.2 (-0.9, of mobile phone use.
0.6)
(de Vocht,
Burstyn &
Cherrie, 2011)
(Dobes et al.,
2011)
(Nomura, Ioka
& Tsukuma,
2011)
2847
2848
12.1.5.2
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
Few incidence trend studies have been conducted on acoustic neuroma, which is a benign and usually
slow growing tumour. In many countries, registration of benign tumours like acoustic neuroma is not mandatory,
which may affect completeness of registration, and many cancer registers do not record acoustic neuroma at all.
In addition, the first treatment option is often “wait and observe”, as many acoustic neuromas do not grow at all
after discovery, and more invasive treatments may have unwanted side effects. Therefore, a large proportion of
acoustic neuroma tumours are not histologically confirmed, but can be unequivocally diagnosed through CT or
MRI scanning. Without histological confirmation, however, identification of cases through pathology reports are
not possible, which may lead to underreporting of cases to cancer registers. Increased availability of advanced
diagnostic techniques like CT and MRI is likely to lead to a higher detection rate, as small asymptomatic
tumours may previously to a greater extent have remained undetected. Furthermore, coding and classification of
Acoustic neuroma
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
114
2859
2860
2861
2862
acoustic neuroma in the WHO International classification of diseases (ICD) has changed, which may also affect
reporting of this tumour. Thus, changes in acoustic neuroma incidence rates may not only reflect true changes of
the incidence but may also occur as a result of changes in detection, registration and reporting practices, and are
less reliable than observed changes in incidence rates for malignant brain tumours.
2863
2864
2865
2866
2867
2868
2869
2870
2871
Larjavaara and co-workers (Larjavaara et al., 2011a) studied the incidence of acoustic neuroma in the
Nordic countries between 1987 and 2007, using data from the national cancer registers. In the Nordic countries,
it is mandatory by law to report new cases of both malignant and benign nervous system tumours to the cancer
registers. Over the whole time period, an annual increase in the incidence rate of 3% was observed, but varied
over time, and between countries. The incidence rate increased slightly between 1987 and the late 1990s, and
stabilized or even decreased from the beginning of the 2000s. [The observed trend in the incidence rate for
acoustic neuroma is not compatible with the increased risk estimates associated with mobile phone use that have
been reported in a few case-control studies. If risk would appear only after a very long induction period, or in a
small subgroup, this would not have been detectable in the incidence trends.]
2872
2873
2874
2875
2876
2877
2878
2879
Stangerup and colleagues (Stangerup et al., 2010) studied the incidence of acoustic neuroma in
Denmark 1976–2008. This study covered all the Danish data on annual incidence rates included in the study by
Larjavaara et al. described above, and extended the calendar time with one year, but included also information
about tumour size and degree of hearing loss at diagnosis. The incidence pattern described is the same as in
Larjavaara et al., i.e. it shows a gradual increase from the start of the study period until the early 2000s, with the
peak incidence rate in 2004, and thereafter a decreasing trend. Over the study period, the average size of the
tumours decreased considerably, and hearing became less impaired at the time of diagnosis, indicating an earlier
detection of tumours over time. They also observed that age at diagnosis increased over the study period.
2880
2881
2882
Benson and colleagues (Benson et al., 2013a) reported annual incidence rates for acoustic neuroma for
men and women 20–79 years old, based on national incidence data in England between 1998 and 2008. No
increased incidence of acoustic neuroma was observed for either men or women.
Table 12.1.21. Incidence studies – acoustic neuroma
Country/
Time period Outcome
location
Age range
Denmark
1987–2007
Finland
All ages
Acoustic
neuroma
Norway
Sweden
Cancer
incidence
data
Cancer trend
Comments
Reference
Incidence
rates from
Nordic
national
cancer
registries
The incidence rate
increased slightly
between 1987 and the
late 1990s, and stabilized
or even decreased from
the beginning of the
2000s
Age-standardized
incidence rates.
(Larjavaara
et al., 2011a)
Over whole study period,
annual increase 3%
Variation between
countries
Denmark
1976–2008
All ages
Acoustic
neuroma
Tumour
size and
degree of
hearing
loss at
diagnosis
UK, England 1998–2008
20–79 year
Acoustic
neuroma
Incidence
rates from
Danish cancer
registry
Gradual increase from
1976 until the early
2000s, with the peak
incidence rate in 2004,
and thereafter a
decreasing trend.
Average size of the
tumours decreased
considerably over the
study period, and hearing
became less impaired.
National
No increase in the
incidence data incidence, either among
in England
men or women.
Nordic countries among
the earliest to adopt
mobile phone technology
in the general population.
Latency period over 10
years.
No evidence of an impact
of mobile phone use on
acoustic neuroma
incidence.
High quality cancer
registration.
(Stangerup
et al., 2010)
Latency period up to 15
years.
No evidence of an impact
of mobile phone use on
acoustic neuroma
incidence.
Indication of earlier
detection (smaller
tumours, less hearing
impairment).
No evidence of an impact
of mobile phone use on
glioma or meningioma
incidence.
(Benson et
al., 2013a)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
115
2883
2884
12.1.5.3
2885
2886
2887
2888
2889
2890
2891
Shu and colleagues (Shu, Ahlbom & Feychting, 2012) studied trends in the age standardized incidence
rate of salivary gland tumours among adults (≥20 years) in the Nordic countries and of malignant parotid gland
tumours in Sweden during the period 1970–2009. No significant changes were observed in the salivary gland
tumour incidence rate overall (annual percent change -0.10% (95% CI -0.38 to 0.18) in men and -0.19% (95% CI
-0.53 to 0.13) in women), or in the incidence rate of parotid gland tumours specifically (incidence rate in men
0.9/100 000 in 1970 and 0.8/100 000 in 2009, in women the incidence rate was 0.7/100 000 in both 1970 and
2009).
2892
2893
2894
2895
2896
2897
2898
De Vocht and colleagues (de Vocht, 2011) studied the annual incidence rate of parotid gland tumours
in England from 1986–2008, based on data from the UK Office of National Statistics. The annual age
standardized incidence rate increased from 0.5 to 0.8/100 000 in men (test for trend p<0.01) and from 0.4 to 0.6
per 100 000 in women (p<0.01) over the study period. The increase in the incidence rate started before the
introduction of mobile phones, and did not exhibit a marked shift at any time after mobile phone use had become
prevalent in the population. [The UK incidence rate in 2008 had reached about the same level as the incidence
rate in Sweden had been since the 1970s.]
2899
Studies with uncertainties related to inclusion criteria
2900
2901
2902
2903
2904
2905
2906
2907
The study by Czerninski et al. is not included in the evaluation because the trend during 1970–2006
was presented only as the number of cases of salivary gland cancer per year, without taking into consideration
population size, or age and sex distribution (Czerninski, Zini & Sgan-Cohen, 2011). This prevents any
comparisons over time, as the Israeli population size has increased considerably, and most industrialized
populations are aging. The annual incidence of salivary gland cancer was 0.8/100 000, of which approximately
60% were parotid gland cancer. A 4-fold increase in the number of parotid gland cancer cases per year between
1970 and 2006 was reported. It is unclear, however, how the age and sex adjusted rate per 100 000 persons has
changed over time. [The reported incidence rate is considerably lower than in the UK and the Nordic countries.]
2908
2909
2910
2911
2912
2913
Two incidence studies of ocular melanoma have been conducted, but they cover only a short time
period after the introduction of handheld mobile phones. A Danish study covered 1943–1996 (Johansen et al.,
2002), and a US study 1974–1998 (Inskip, Devesa & Fraumeni, 2003). None of the studies found evidence of
changes in the incidence of ocular melanoma after the introduction of mobile phones. [The studied time periods
were too early to provide data of relevance for evaluation of an effect of mobile phone use on the incidence of
ocular melanoma.]
Parotid gland tumours and eye tumours
Table 12.1.22. Incidence studies – parotid gland tumours
Country/
Time period Outcome
location
Age range
Denmark
1970–2009
Finland
≥20 year
Norway
Sweden
Salivary
gland
tumours
Parotid
gland
tumours
(only
Sweden)
Cancer
incidence
data
Cancer trend
Comments
Reference
Incidence
rates from
Nordic
national
cancer
registries
through
NORDCAN
Salivary gland
Age-standardized
incidence rates.
(Shu,
Ahlbom &
Feychting,
2012)
Annual % change (95% CI):
Men: -0.10 (-0.38, 0.18)
Women: -0.19 (-0.53, 0.13)
Parotid gland
Incidence rate, men:
Nordic countries among
the earliest to adopt
mobile phone technology
in the general population.
1970: 0.9/100 000
Latency period up to 15
years.
1970: 0.7/100 000
No evidence of an
impact of mobile phone
use on brain tumour
incidence.
Swedish
cancer registry 2009: 0.8/100 000
Incidence rate women:
2009: 0.7/100 000
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
116
England
1986–2008
Parotid
gland
Incidence
rates from the
UK Office of
National
Statistics
Significantly increased
incidence rates, from 0.5 to
0.8/100 000 in men and 0.4
to 0.6 per 100 000 in
women
Age-standardized
incidence rates.
(de Vocht,
2011)
The increase started
before the introduction of
mobile phones.
Incidence rate in 2008
similar to Swedish
incidence rate from 1970
onwards.
2914
2915
12.2
Animal studies
2916
2917
2918
2919
2920
Animal studies investigating the carcinogenic potential of RF radiation were reviewed by WHO
(1993b). At that time there was only a limited number of studies available. WHO concluded that there was no
definite evidence that RF exposure has an effect on carcinogenesis, but that there clearly was a need for further
studies. Many of such studies haven been performed since then. In order to get a complete picture, also the
papers published before 1992 that were described in WHO (1993b) are discussed in this review.
2921
2922
2923
2924
The present search resulted in 300 papers, of which 237 were not fulfilling the specifications. Of the
resulting 63 papers, 4 were in a language that could not be understood, 7 were comments to other papers or
otherwise not relevant, and 3 papers could not be obtained. One paper was retrieved by another search. That left
50 papers to be discussed.
2925
2926
2927
2928
2929
2930
2931
2932
2933
Evaluating carcinogenicity in laboratory rodents has remained a cornerstone in identifying agents
likely to cause cancer in humans. According to IARC, agents for which there is sufficient evidence of
carcinogenicity in experimental animals are considered to pose carcinogenic hazard to humans, unless there is
scientific evidence that the agent causes cancer through a species-specific mechanism that does not operate in
humans (IARC, 2011). However, despite the similarities in many cancer characteristics between humans and
laboratory rodents, interspecies differences need to be taken into account when extrapolating data from rodents
to humans: many agents that are carcinogenic in rodents (often only at very high doses) are not carcinogenic to
humans, and some human carcinogens do not affect rodents (Ames & Gold, 1990; Anisimov, Ukraintseva &
Yashin, 2005; Trosko & Upham, 2005)
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
The effects of stress resulting from restraint and related daily handling has been seen in many animal
cancer studies as a lower body weight among the sham-exposed (restrained) animals than among the cagecontrol (unrestrained) animals (see for instance (Heikkinen et al., 2003; Oberto et al., 2007; Shirai et al., 2007;
Smith et al., 2007; Yu et al., 2006; Zook & Simmens, 2006). In many of these studies, tumour incidence has also
been lower and survival higher in the sham-exposed (restrained) group than in the cage-control (unrestrained)
group, which may be related to the observations that reduced energy intake inhibits the development of tumours
(Keenan et al., 1996; Klurfeld et al., 1991; Sinha, Gebhard & Pazik, 1988). Immobilization has not caused
experimental bias in studies assessing carcinogenicity of RF radiation, as both the RF exposed and the shamexposed animals have been restrained, but it can be argued that stress could act as an effect modifier and obscure
possible RF-induced effects. However, there is no evidence of such modifying effects: many of the studies
reviewed below have used freely moving animals, and the majority of studies have produced negative findings
independent of the handling (restrained or unrestrained) of the animals.
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
Classical carcinogenicity bioassays involve exposure of animals over most of their lifetime to the
agent being tested. Such studies are potentially capable of revealing whether the tested agent alone could act as a
complete carcinogen or serve to increase the incidence of spontaneous tumours. This type of studies are,
however, not sensitive in detecting weak carcinogenic effects (because of the low number of tumours induced)
and co-carcinogenic effects (interaction with other carcinogens). To overcome these limitations, several studies
have used tumour-prone animal strains or combined exposure to RF radiation and known carcinogens. The
animal studies are classified here as studies with exposure to RF field alone (Table 12.2.1), including studies
using tumour-prone animals strains (Table 12.2.2); studies using exposure to RF radiation combined with a
known genotoxic/carcinogenic agent (Table 12.2.3); and studies evaluating effects of RF radiation on implanted
or injected tumour cells (Table 12.2.4).
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
117
2956
12.2.1.
RF radiation alone
2957
12.2.1.1. Conventional laboratory animal strains
2958
Lymphoma
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
Utteridge et al. (2002) performed a study that was planned as a replication experiment of an earlier
study (Repacholi et al., 1997) reporting enhanced development of lymphoma in Eµ-Pim1 transgenic mice
exposed to RF radiation (see below). In addition to transgenic animals, Utteridge and colleagues used also
corresponding wildtype C57BL/6Ntac mice. In contrast to the Repacholi study where the animals were free
roaming, in this study they were restrained during the exposures, which allowed for more accurate dosimetry.
Groups of 120 animals were whole-body exposed to 898 MHz GSM signals for 1 h per day, 5 days per week,
during 104 weeks at four SAR levels ranging from 0.25 to 4 W/kg. Animals that died during this period and all
those that survived until the end of the exposure period were examined histopathologically. There were only a
few lymphoblastic leukaemias in the wild-type animals, which precluded a meaningful analysis. For nonlymphoblastic leukaemias there were no statistically significant differences between the sham and RF-exposed
animals.
2970
Heamatological tumours
2971
2972
2973
2974
2975
2976
Jin et al. (2012a) assessed in Sprague Dawley rats the effect of simultaneous exposure to two types of
RF fields, an 849 MHz code division multiple access (CDMA), and an 1900 MHz wideband code division
multiple access (WCDMA) signal, each at an SAR of 2 W/kg, so the combined SAR was 4 W/kg. The animals
were exposed for 45 min per day, 5 days per week up to 8 weeks. The main thrust of the study was on effects on
the immune system of rats (discussed in Section 10.3), but histopathological changes in the spleen were also
assessed evaluated. No tumours were observed in the spleen.
2977
Skin tumours
2978
2979
2980
2981
2982
Sanchez et al. (2006a) exposed the skin of hairless IFFA Creda rats to 900 MHz and 1800 MHz GSM
signals for 2 h per day, 5 days per week and 12 weeks. The SAR was either 2.5 or 5 W/kg, and the animals were
restrained during exposure. After the last exposure skin biopsies were taken and the histology of the skin
investigated. No differences were observed between RF-exposed and sham-exposed animals. [The number of 8
animals per group in this study was rather small.]
2983
2984
2985
2986
2987
In an experiment that was primarily aimed at studying the effect of RF exposure on chemicallyinduced skin cancer (see section 12.2.2), Paulraj and Behari (2011) exposed male Swiss mice to either 112 MHz
EMF, amplitude modulated at 16 Hz, at an SAR of 0.75 W/kg, or to 2.54 GHz at an SAR of 0.1 W/kg without
the chemical induction. Neither type of exposure resulted in any effect on skin tumours. [The number of animals
in this study is not clear: the authors use an unspecified ‘effective’ number of animals in their analysis.]
2988
Central nervous system tumours
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
Using a so-called ‘carousel’ exposure set-up for well-defined RF exposure levels in the brain, five
studies failed to find evidence of enhanced development of brain tumors in RF field-exposed rats (Adey et al.,
1999; Adey et al., 2000; Anderson et al., 2004; La Regina et al., 2003; Zook & Simmens, 2001) at average SARs
in the brain up to about 1.5 W/kg. In the carousel set-up, the rats are restrained head first in cylindrical tubes
arranged in a radial configuration with the RF antenna at the centre of the carousel, where the head is
preferentially irradiated. The SAR in other body parts is much lower, and the ratio of brain average SAR to
whole body average SAR may be up to 10:1 at mobile phone frequencies (Schönborn, Pokovic & Kuster, 2004).
The animals, in groups ranging from 56 to 100 per study, were exposed for most of their lifetime, and three of
the studies also included in utero exposures (Adey et al., 1999; Adey et al., 2000; Anderson et al., 2004). There
was a non-significant (p>0.05) decreased CNS glial tumour development in the group exposed to NADCmodulated RF fields (Adey et al., 1999). [The unexpectedly high incidence of spontaneous CNS tumours in the
sham exposed group suggests that this difference might be a consequence of chance.] No other effects on brain
tumours were observed in any of the studies. The studies that involved histopathological evaluation of other
organs also provided no evidence of enhanced development of tumours in other tissues, exposed at considerably
lower SAR values than the brain (Anderson et al., 2004; La Regina et al., 2003; Zook & Simmens, 2001).
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
118
3004
Multiple tumours
3005
3006
3007
3008
3009
3010
3011
3012
3013
The combined incidence of malignant tumours (all tumour types combined) was statistically
significantly increased (p<0.008) in male Sprague-Dawley rats exposed to radar-type pulsed 2.45 GHz RF fields
for 21.5 h per day during 25 months (100 animals per group) (Chou et al., 1992). The whole-body average SARs
varied from 0.15 to 0.4 W/kg depending on the size of the animals. The organ-specific tumour incidences were
low (except those in some endocrine organs). The incidence of any single type of primary malignant or benign
neoplasm, the combined incidence of benign neoplasms, or survival were not statistically significantly affected.
Overall, the study did not show any definite biologically significant effects. The incidence of benign
pheochromocytoma was reported to be higher in RF-exposed rats, but the difference did not reach statistical
significance.
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
Two studies evaluated carcinogenicity of both a GSM (Global System for Mobile Communication)
signal at 902 MHz and a DCS (Digital Personal Communications System) signal at 1747 MHz in B6C3F1 mice
(Tillmann et al., 2007) and in Wistar rats (Smith et al., 2007). Three exposure levels from 0.4 to 4 W/kg (and
sham exposure) were used. In the mouse study (Tillmann et al., 2007), using groups of 130 animals, no
significant increase in the incidence of any particular tumour type in the RF-exposed groups was observed.
Interestingly, in both studies and with both RF signals the incidence of liver adenomas in males decreased with
increasing exposure level, with a statistically significant difference (p<0.05) between the highest exposure and
the sham-exposed group. However, comparison to published tumour rates in untreated mice revealed that the
observed tumour rates were within the range of historical control data. In conclusion, the studies produced no
evidence that exposure at whole-body SARs of up to 4.0 W/kg increased the incidence or severity of neoplastic
or non-neoplastic lesions, or resulted in any other adverse health effects. The rat study (Smith et al., 2007) was a
combined chronic toxicity and carcinogenicity study, and some of the animals (15 males and 15 females per
group) were killed at 52 weeks from the start of the study. There were no significant differences in the incidence,
multiplicity, latency or severity of neoplasms, or any other adverse responses to RF field exposure.
3028
3029
3030
3031
3032
In a study using a mouse model of multi-organ tumour development (see below) Saran et al. (2007)
also exposed wild-type siblings to 900 MHz GSM-type radiation at an SAR of 0.4 W/kg for 30 min twice a day
for 5 days (starting on postnatal day 2). Brains, any visible tumours and preneoplastic skin lesions were
examined histopathologically. No statistically significant differences in survival and tumour incidence were
found between exposed and sham-exposed animals.
3033
Studies not included in the analysis
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
Not many studies have been performed using frequencies higher than those used in mobile
telecommunication. Ivanov et al. (2005) studied the effects of long-term exposure to 37 GHz RF EMF in inbred
albino and C57Bl/6 mice, as a simulation of SHF (super high frequency) therapy in patients. The authors stated
that the animals were repeatedly exposed for 1 month, that the total exposure time was 3.5 h, and that the
exposure level was <10 mW/cm2 (100 W/m2). Under these conditions they observed a decreased survival in
C57Bl/6 adults and offspring, and increased tumour formation in albino adults and offspring. [This study suffers
from a number of inadequacies. Many experimental details, such as the age and number of animals and the exact
exposure level are missing. Moreover, the authors state that part of the animals had been ‘specially irradiated’
and part had been exposed to SHF EMF in previous, unspecified, experiments. Therefore the results of this study
cannot be interpreted.]
Table 12.2.1. Studies on carcinogenesis in non-tumour-prone rodents using RF alone.
Animals, number
per group, age at
start
Exposure: source,
schedule, level,
freely moving or
restrained
Response
Comment
Reference
No effect.
Results for
lymphoma-prone
mice in Table 12.2.2.
Utteridge et al.
(2002)
Lymphoma
Mouse: C57BL/6Ntac GSM 898 MHz
(n=120)
1 h/d, 5 d/week, 104
4-6 weeks + 10 d
weeks
acclimatization
WBA SAR 0.25, 1.0,
2.0, 4.0 W/kg
Restrained
Haematological tumours
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
119
Rat: Sprague
Dawley(n=80)
849 MHz CDMA and
1900 MHz WCDMA
9 weeks
45 min/day, 5
days/week, 8 weeks
No effect.
(Jin et al., 2012a)
combined WBA SAR
4 W/kg
Free
Skin tumours
Rat: hairless (n=8)
5 weeks + 1 week
acclimatization
GSM 900 MHz, 1800
MHz
No effect on skin
histology.
Small number of
animals.
Sanchez et al.
(2006a)
No effect on skin
tumour.
Small number of
animals. Meaning of
‘effective’ number of
animals not clear.
Paulraj & Behari
(2011)
Non-sigificant
reduced incidence of
tumours.
See Table 12.2.3 for
results regarding
initiated tumours.
Adey et al. (1999)
No effect on tumour
incidence and
survival.
See Table 12.2.3 for
results regarding
initiated tumours.
Adey et al. (2000 )
860 MHz pulsed, CW No effect on
6 h/d, 5 d/week, 22
neurogenic or other
months
tumours.
See Table 12.2.3 for
results regarding
initiated tumours
Zook & Simmens
(2001)
2 h/d, 5 d/week, 12
weeks
Skin SAR 2.5, 5 W/kg
Restrained
Mouse: Swiss albino, 112 MHz, 16 Hz AM;
normal ( n=18)
2.54 GHz
7-8 weeks
Skin tumour
2 h/d, 3 d/week, 16
weeks
112 MHz: WBA SAR
0.75 W/kg; 2.54 GHz:
WBA SAR 0.1 W/kg
Free
CNS tumours
Rat: Fischer 344,
(n=56, 60)
Gestation d 15 + 3 d
acclimatization
NADC 835 MHz
2 h/d, 7 d/week
(prenatal), 4 d/week
(postnatal), 2 years
Brain SAR 0.33-0.53
W/kg
Free (prenatal),
restrained (postnatal)
Rat: Fischer 344,
(n=90)
Gestation d 15 + 3 d
acclimatization
FM 836 MHz
2 h/d, 7 d/week
(prenatal), 4 d/week
(postnatal), 2 years
Brain SAR 0.33-0.53
W/kg
Free (prenatal),
restrained (postnatal)
Rat: : Sprague
Dawley, (n=60)
Gestation d 15
Brain SAR 1.0±0.2
W/kg
Restrained
Rat: Fischer 344,
(n=160)
FDMA 835 MHz,
CDMA 847 MHz
4 weeks + 2 weeks
acclimatization
4 h/d, 5 d/week, 2
years
No effect of either
type of signal on
brain or other
tumours.
La Regina et al.
(2003)
CNS & other tumours Brain SAR 1.3±0.5
W/kg
Restrained
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
120
Rat: Fischer 344,
(n=180; control: 160)
4 d before birth
Brain and other
tumours
Iridium 1.6 GHz
2 h/d, 7 d/week
(perinatal); 2 h/d, 5
d/week, 2 years
(postnatal)
No effect on survival,
brain or other
tumours.
Anderson et al.
(2004)
No overall effect on
tumour development;
but more primary
malignant tumours in
exposed; no effect on
benign tumours and
overall survival.
Chou et al. (1992)
Brain SAR: 0.16
W/kg (perinatal);
0.16, 1.6 W/kg
(postnatal)
Free
Multiple tumours
Rat: Sprague
Dawley, (n=100)
2450 MHz, pulsed
3 weeks + 5 weeks
acclimatization
WBA SAR 0.15-0.4
W/kg
21.5 h/d, 25 months
Free
Mouse: B6C3F1,
(n=130)
GSM 902 MHz, DCS
1747 MHz
4-5 weeks + 4 weeks 2 h/d, 5 d/week, 24
acclimatization
months
No effect of GSM,
decreased number of
tumours with DCS
exposure.
Tumour rates were
within the range of
historical control
data.
Tillmann et al. (2007)
WBA SAR 0.4, 1.3,
4.0 W/kg
Restrained
Rat: Wistar, (n=130)
GSM 900 MHz, DCS
4-5 weeks + 2 weeks 1800 MHz
acclimatization
2 h/d, 5 d/week, 104
weeks
No effect on
carcinogenesis and
survival.
Smith et al. (2007)
No effect on
carcinogenesis and
survival.
Results for Patched1, Saran et al. (2007)
transgenic, X-ray
tumour prone mice in
Table 12.2.2.
WBA SAR 0.44, 1.33,
4.0 W/kg
Restrained
Mouse: Patched1
wild type (n=50-63)
GSM 900 MHz
2d
WBA SAR 0.4 W/kg
2x0.5 h/d, 5 d
Restrained
Abbreviations: CDMA: Code Division Multiple Access; CW: continuous wave; DCS: Digital Communication Signal;
FDMA: Frequency Division Multiple Access; FM: frequency modulation; GSM: Global System For Mobile
Communication; NADC: North American Digital Cellular; WBA SAR: whole-body averaged SAR
3044
3045
12.2.1.2. Studies using genetically predisposed animal models
3046
3047
3048
3049
3050
3051
3052
3053
3054
Animal strains developing tumours (in some organs) with particularly high frequency and/or early in
life are classified as ‘tumour-prone strains’. These strains include animals engineered to be more vulnerable via
gene manipulation (transgenic animals), as well as strains with exceptionally high tumour incidence due to their
genetic background. The division between ‘tumour-prone’ and ‘other’ strains is somewhat arbitrary, because
spontaneous tumour frequency varies greatly between different animal strains. The spontaneous incidence of
tumours in this kind of experimental models is important: if nearly all animals in the unexposed control group
develop tumours, there is not much room for an additional effect from RF field exposure. Note, however, that
accelerated development of tumours can be detected even if the final incidence is 100%, if the tumours are
externally observable during the experiment. This is the case for, e.g., skin tumours and mammary tumours.
3055
Lymphoma
3056
3057
3058
3059
3060
3061
Transgenic Eµ-Pim1 mice overexpressing the Pim1 oncogene in their lymphoid cells are prone to
malignant lymphoma. In the first study with this model with exposure to RF EMF (Repacholi et al., 1997) EµPim1 mice were exposed for 2x30 min per day during 24 months to 900 MHz GSM-type fields at whole-body
SARs ranging from 0.13–1.4 W/kg (if all possible animal orientations are included, the range was 0.008–4.2
W/kg). The RF-exposed animals (n=101) had an increased lymphoma incidence compared to sham-exposed
controls (n=100), with an odds ratio of 2.4 (95% confidence interval: 1.3–4.5). At the time the study was
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
121
3062
3063
3064
3065
3066
terminated, lymphoma incidence was increasing rapidly in both exposed and sham-exposed animals. The authors
emphasize that even if the observed effect were established, the relevance of the animal model for human cancer
risk assessment needs to be carefully considered. [Histopathology of lymphoma was only performed on animals
that died before the end of the experiment. The lymphoma incidence in surviving animals is unknown, and this is
thus also the case for the overall lymphoma incidence.]
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
The findings of the Repacholi et al. study were not confirmed in a replication study by Utteridge et al.
(2002), who used the same strain of mouse from the same supplier as the earlier study. The investigators also fed
the same food to the mice. The later study had some refinements in experimental design: four SAR levels (0.25,
1.0, 2.0 and 4.0 W/kg) were used instead of one in the original study; animals were restrained during the
exposure for better control of variations in exposure level; in the 104 weeks lasting study the animals were
exposed once per day during 5 days per week instead of in two episodes of 30 minutes per day, 7 days per week
as in the Repacholi study; and full necropsy was performed on all 120 mice per group at the end of the study. In
RF field-exposed animals enhanced development of lymphoma was not observed. The incidence of
lymphoblastic leukaemia was slightly lower in all RF-exposed groups compared to that of the sham-exposed
animals, and the difference was statistically significant (p=0.02) at the SAR of 0.25 W/kg. In contrast, the
incidence of non-lymphoblastic leukaemia was slightly higher in RF-exposed groups, but these differences were
not statistically significant either in pairwise comparisons or in a trend test. The incidence of lymphomas in the
RF-sham-exposed group was surprisingly high, and the publication stirred debate whether some critical features
of the original experiment had been changed (Goldstein et al., 2003a; b; Kundi, 2003a; b; Lerchl, 2003).
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
The study reported by Oberto et al. (2007) was also a replication and an extension of the Repacholi et
al. (1997) study. Eµ-Pim1 transgenic mice were exposed for 1 h per day, 7 days per week to pulsed 900 MHz
(pulse repetition rate 217 Hz, i.e. a GSM-type signal) RF radiation at whole-body SARs of 0.5, 1.4 or 4.0 W/kg.
50 animals per sex per group were exposed, sham-exposed or used as cage controls. There were several
methodological improvements compared to the original study by Repacholi et al. (1997), including use of several
exposure levels, well-defined dosimetry and more uniform exposure (achieved through restraint of the animals)
and necropsy and extensive histopathology of all animals. Compared to the sham-exposed controls, survival was
reduced in the animals exposed to RF radiation. The intergroup differences were statistically significant in the
male animals (p<0.05), but there was no trend with increasing exposure level (the lowest survival was found at
0.5 W/kg). In females, a decreased survival was only observed with an SAR of 0.5 W/kg (p<0.05). No increase
in lymphoma incidence was observed in any of the RF-exposed groups. Concerning other neoplastic findings,
Harderian gland adenomas were increased in male mice, with a significant dose-related trend (p<0.01).
However, this trend was not supported by the findings on female animals, in which no tumours in the highest
exposure groups were observed. [The statistical analysis used in this study can be criticized. The cage-control
and the sham-exposed control groups were combined for statistical comparisons, which is not a valid procedure
given the differences in body weight development and tumour incidence between these groups (these differences
are most likely related to restraint of the sham-exposed animals). However, based on the data reported in the
paper, a different analysis strategy (comparison to the sham-exposed group only) would not essentially change
the interpretation that there was no effect of RF radiation on tumour incidence at any site. The reduced survival
in RF field-exposed animals is not thoroughly discussed by the authors; this finding remains unexplained and
difficult to interpret without detailed information about the causes of death.]
3102
3103
3104
3105
3106
3107
3108
3109
3110
Development of lymphoma in female AKR/J mice did not change with GSM-type RF exposure at a
nominal SAR of 0.4 W/kg applied continuously for 46 weeks (Sommer et al., 2004). The AKR/J strain is prone
to develop lymphoma due to expression of an AKV retrovirus in all of their tissues. About 90% of the 160
animals in both the sham-exposed and RF-exposed groups developed lymphoma by the end of the 10-month
study. Essentially all mortality was reported to be related to the development of lymphoblastic lymphoma, and
survival was not different between RF field and sham exposure. No effects of exposure were seen in differential
leukocyte counts of blood samples collected 5–10 months after the beginning of RF exposure. [The nominal
SAR was 0.4 W/kg, but as in other studies using several freely moving animals per cage, the variation in
exposure level would undoubtedly have been large.]
3111
3112
3113
3114
3115
3116
3117
In another study by the same group (Sommer et al., 2007), unrestrained AKR/J mice, 160 animals per
group, were chronically sham-exposed or exposed to a generic UMTS test signal for 24 h per day, 7 days per
week at a SAR of 0.4 W/kg. Additionally, 30 animals were kept as cage controls. The animals were checked
visually each day and were weighed and palpated weekly to detect swollen lymph nodes. Starting at the age of 6
months, blood samples were taken from the tail every 2 weeks to perform differential leukocyte counts and to
measure the hematocrit. Visibly diseased animals or those older than 43 weeks were killed and tissue slices were
examined for metastatic infiltrations and lymphoma type. Cage-control animals had a significantly lower growth
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
122
3118
3119
3120
3121
rate than those kept in the radial waveguides. Incidence of lymphoma, survival time and the severity of the
disease indicated that there was no effect from exposure to RF radiation. Cage-control animals had significantly
lower body weights and higher occurrence of metastatic infiltrations in liver and meninges than the other groups.
This difference was most likely related to different housing conditions and stress level.
3122
3123
3124
3125
3126
3127
3128
Lee et al. (2011a) used the same AKR/J mouse strain and exposed the animals simultaneously to two
mobile phone signals used in Korea: 848.5 MHz CDMA (GSM-like) and 1950 MHz WCDMA (UMTS-like).
Exposure of the unrestrained animals (40 per group) was for 45 min per day, 5 days per week during 42 weeks at
a whole-body SAR of 4.0 W/kg. The animals were examined weekly for the presence of a large spleen or large
lymph nodes and were killed when visibly ill. The surviving animals were killed at the end of the experiment at
48 weeks of age. Necropsy was performed on all mice. No difference in survival and lymphoma incidence was
observed between exposed and sham-exposed animals.
3129
Mammary tumours
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
Although the study by Szmigielski et al. (1982) has been discussed in WHO (1993b) it will be
described here as well for completeness. The authors exposed groups of 40 female C3H/HEA mice, that have a
high spontaneous mammary cancer incidence, to 2450 MHz RF fields with a power density of 5 or 15 mW/cm 2,
corresponding to whole-body SAR levels of 2–3 and 6–8 W/kg, respectively. They observed a dose-dependent
acceleration of mammary cancer development resulting in shortening of life time. The response to the lower RF
level was reported to be similar to that of confinement stress. [Whole-body exposure to the highest SAR level
can certainly result in heating of the animals, but also an SAR of 2 W/kg is already a significant systemic
challenge to a mouse with a metabolic rate of 4–7 W/kg under thermal neutral conditions. An increased body
temperature can result in acceleration of tumour growth (Dickson, 1977; Wile, Nahabedian & Mason, 1983)
although not all tumours seem to respond similarly to moderate heating (Sen et al., 2011).]
3140
3141
3142
3143
Similarly to Szmigielski et al., Frei and co-workers used continuous 2.45 GHz RF radiation (Frei et
al., 1998a; Frei et al., 1998b). They exposed groups of 100 mammary tumour-prone C3H/HeJ mice for 20 hours
per day, 7 days per week for 18 months at SARs of 0.3 and 1.0 W/kg. No effect was observed of these exposures
on the number and rate of growth of the mammary tumours.
3144
3145
3146
3147
3148
3149
3150
3151
3152
Two other studies used signals consisting of short pulses (Jauchem et al., 2001; Toler et al., 1997).
Toler et al. (1997) exposed C3H/HeJ mice to a 435 MHz pulsed radar signal for 22 hours per day, 7 days per
week and 21 months. The unrestrained animals (200 per group) were sham exposed or exposed at a whole-body
SAR of 0.32 W/kg. Jauchem et al. (2001) investigated in the same mouse strain the effects of an ultra-wideband
(UWB) pulsed RF signal. This type of signal is studied by the military for use against vulnerable military targets.
Free roaming animals were exposed for 2 min per week during 12 weeks (group size: 100 animals). The peak
power of the pulses was very high, 40 kV/m, but because of their short duration the overall whole-body SAR
was calculated to be only 0.0098 W/kg. In both studies, no differences between exposed and sham-exposed
groups were observed in mammary tumour incidence, nor were differences found in survival.
3153
3154
3155
3156
3157
3158
3159
3160
Although the four later studies (Frei et al., 1998a; Frei et al., 1998b; Jauchem et al., 2001; Toler et al.,
1997) were designed specifically to examine mammary tumours, they included histopathological analyses of
other main tissues. Overall, RF field exposure did not affect the development of tumours or survival of the
animals. The only statistically significant differences in tumour incidence reported in these studies were a
smaller number (0 vs. 4) of alveolar-bronchiolar adenomas in RF field vs. sham-exposed animals (p=0.04) in one
study (Frei et al., 1998a), and an increased incidence of bilateral ovarian tumours (10 in exposed vs. 2 in shamexposed animals; p=0.03) in another study (Toler et al., 1997). The latter was, however, not accompanied with
increase in the number of mice developing an ovarian tumour.
3161
Multiple tumours
3162
3163
3164
3165
3166
3167
3168
3169
Saran et al. (2007) used Patched1 heterozygous knockout mice, an animal model of multi-organ
tumour development in which exposure of newborn animals to ionizing radiation greatly enhances development
of brain tumours (medulloblastoma). Newborn Patched1 heterozygous mice and their wild-type siblings
(discussed above) were exposed to 900 MHz GSM-type radiation at an whole-body SAR of 0.4 W/kg for 30 min
twice a day for 5 days, starting on postnatal day 2 (50–63 animals per group). Brains, any visible tumours and
preneoplastic skin lesions were examined histopathologically. No statistically significant differences in survival
were found between exposed and sham-exposed animals. Medulloblastomas (in 7 animals) and
rhabdomyosarcomas (in 56 animals) were found in the Patched1 mice but not in the wild-type animals. The
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
123
3170
3171
3172
3173
incidence of rhabdomyosarcoma was higher (68%, 36 animals) in the exposed group than in the sham-exposed
group (51%, 20 animals), but this difference was not statistically significant. The incidences of
medulloblastomas, other (unspecified) tumours or preneoplastic skin lesions did not differ between the exposed
and sham-exposed groups.
3174
Studies not included in the analysis.
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
Anghileri et al. (2005) exposed non-transgenic Ico:OFI(I.O.P.S. Caw) mice with a high lymphoma
incidence to a GSM 800 MHz field for 1 hour per week during 4 months. They observed a decreased survival
and more lymphomas in exposed compared to sham-exposed animals. In another publications of the same group
(Anghileri et al., 2006) the same data are provided and tumour induction by RF is compared to that by the ferricATP complex studied by this group as a promotor of carcinogenesis because of its induction of cellular calcium
ion influx. In a third study using the same experimental system, Anghileri et al. (2009) studied the effect of RF
exposure combined with calcium chloride, aluminium lactate or aluminium citrate administration. These
compounds affect the calcium homeostasis. They observed a decreased survival in animals exposed to the
combined treatments compared to RF alone.[Since in none of these publications exposure levels are provided,
these studies cannot be interpreted.]
Table 12.2.2. Studies on carcinogenesis in tumour-prone rodents using RF alone.
Animals, number
per group, age at
start
Exposure: source,
schedule, level,
freely moving or
restrained
Response
Comment
Reference
GSM 900 MHz
Higher risk in
exposed,
OR=2.4(1.3-4.5).
Large variability in
exposure, therefore
difficult to interpret.
Incomplete
histopathology.
Repacholi et al.
(1997)
Lower incidence of
lymphoblastic
leukaemia at 0.25
W/kg.
No effect on nonlymphoblastic
leukaemia.
Despite small
modifications in
design, adequate
replication of
Repacholi study.
Utteridge et al.
(2002)
Decreased survival in
males, no doseresponse; in females
only with 0.5 W/kg.
No effect on
lymphomas and other
tumours.
Despite small
modifications in
design, adequate
replication of
Repacholi study.
Lymphoma
Mouse: Eµ-Pim1,
transgenic
lymphoma-prone
(n=101; controls:
n=100)
4-6 weeks + 10 d
acclimatization
Mouse: Eµ-Pim1,
transgenic
lymphoma-prone
(n=120; cage control
n=30)
4-6 weeks + 10 d
acclimatization
Mouse: Eµ-Pim1,
transgenic
lymphoma-prone
(n=100)
6 week s+ 20 d
acclimatization
Multiple tumours
Mouse : AKR/J,
normal, but with high
lymphoma incidence
(n=160)
4-5 weeks
2x30 min/d, 24
months
WBA SAR 0.008-4.2
W/kg (indiv); 0.13-1.4
W/kg (in cage)
Free
GSM 898 MHz
1 h/d, 5 d/week, 104
weeks
WBA SAR 0.25, 1.0,
2.0, 4.0 W/kg
Restrained
GSM 900 MHz
1 h/d, 7 d/week, 18
months
WBA SAR 0.5, 1.4,
4.0 W/kg
Restrained
GSM 900 MHz
Continuous, 46
weeks
Replication of
Repacholi et al.
(1997)
See Table 12.2.1 for
results for wildtype
mice.
Oberto et al. (2007)
Replication of
Repacholi et al.
(1997)
Sham and cagecontrols combined in
analysis.
No difference in
lymphoma incidence,
survival.
Sommer et al. (2004)
No difference in
lymphoma incidence,
survival.
Sommer et al. (2007)
WBA SAR 0.4 W/kg
± 40%
Free
Mouse: AKR/J,
normal, but with high
lymphoma incidence
(n=160)
8 weeks
UMTS
Continuous, 43
weeks
WBA SAR 0.4 W/kg
± 40%
Free
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
124
Mouse: AKR/J,
normal, but with high
lymphoma incidence
(n=40)
5 weeks
CDMA, WCDMA
45 min/d, 5 d/week,
42 weeks
No difference in
lymphoma incidence,
survival.
Lee et al. (2011a)
Acceleration of
Heat stress may
development of
have accelerated
mammary cancer;
tumour growth.
effect chronic stress
same as of 2-3 W/kg.
Szmigielski et al.
(1982)
No effect on number
and growth of
tumours.
Frei et al. (1998a)
No effect on number
and growth of
tumours.
Frei et al. (1998b)
No effect on number
and growth of
tumours.
Toler et al. (1997)
No effect on tumour
incidence, survival.
Jauchem et al.
(2001)
WBA SAR 4.0 W/kg
Free
Mammary tumours
Mouse: C3H/HeA,
normal, but with high
mammary cancer
incidence (n=40)
6 weeks
2450 MHz
2 h/d, 6 d/week, 10.5
months
WBA SAR 2-3, 6-8
W/kg
Free
Mouse: C3H/HeJ,
normal, mammary
tumour-prone
(n=100)
3-4 weeks + 10 d
acclimatization
Mouse: C3H/HeJ,
normal, mammary
tumour-prone
(n=100)
3-4 weeks + 10 d
acclimatization
Mouse: C3H/HeJ,
normal, mammary
tumour-prone
(n=200)
3-4 weeks + 1 month
acclimatization
Mouse: C3H/HeJ,
normal, mammary
tumour-prone
(n=100)
3-4 weeks + 10 d
acclimatization
2450 MHz
20 h/d, 7 d/week, 18
months
WBA SAR 0.3 W/kg
Free
2450 MHz
20 h/d, 7 d/week, 78
weeks
Replication of Frei et
al. (1998a)
WBA SAR 1.0 W/kg
Free
435 MHz pulsed
22 h/d, 7 d/week, 21
months
WB SAR 0.32 W/kg
Free
UWB
2 min/week, 12
weeks
40 kV m-1 peak; WBA
SAR 0.0098 W/kg
Free
Multiple tumours
Mouse: Patched1,
transgenic, X-ray
tumour-prone (n=5063)
GSM 900 MHz
2d
Restrained
2x0.5 h/d, 5 d
No effect on survival
and carcinogenesis.
WB SAR 0.4 W/kg
See Table 12.2.1 for
results for wildtype
mice.
Saran et al. (2007)
Abbreviations: CDMA: Code Division Multiple Access; GSM: Global System For Mobile Communication; UMTS:
Universal Mobile Telecommunications Signal; UWB: Ultra Wide Band; WBA SAR: whole-body average SAR;
WCDMA: Wideband Code Division Multiple Access
3185
3186
12.2.2.
Combined RF and known genotoxic/carcinogenic agents
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
Both theoretical considerations (low photon energy) and experimental evidence indicate that direct
DNA-damaging effects of weak RF electromagnetic radiation are not likely. Therefore, there has been
considerable interest in testing RF radiation as a non-genotoxic carcinogen or a co-carcinogen that enhances the
effects of known carcinogenic agents. Methods for detecting non-genotoxic carcinogens and co-carcinogens are
less well developed than those for detecting genotoxic carcinogens. It can be argued that classical animal
carcinogenicity bioassays should identify carcinogens independently of the mechanisms. However, because of
the very low number of tumours induced, such studies (involving exposure to the agent alone, without coexposures) may suffer from low statistical power to detect co-carcinogens. Animal studies on co-carcinogenic
effects have usually been designed based on the concepts of ‘initiation’ and ‘promotion’. Such studies involve a
short-term exposure to an ‘initiator’ (a known DNA-damaging agent), followed by long-term exposure to the
putative ‘cancer promoter’. However, it has been questioned whether the initiation-promotion approach is
sufficient for describing the complex interaction of genotoxic and non-genotoxic agents (Juutilainen, Lang &
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
125
3199
3200
3201
3202
Rytomaa, 2000). Although most of the studies on co-carcinogenicity of RF radiation have tested RF radiation as
a possible promoter after a single dose or short-term treatment with a known initiator, a few studies have used
different approaches such as long-term simultaneous exposure to RF radiation and the known carcinogen, or RF
field exposure before treatment with the known carcinogen.
3203
3204
3205
3206
3207
Details of studies evaluating combined exposure to RF radiation with known genotoxic/carcinogenic
agents are shown in Table 12.2.3. Like in case of genetically predisposed models (see section 12.2.1.2), the
incidence of tumours in the control group (exposed only to the known carcinogen) should be at an appropriate
level to allow detection of a possible further increase related to RF field exposure. Therefore, information of
tumour incidence in the control (known carcinogen-only) group is included in Table 12.2.3.
3208
Lymphoma
3209
3210
3211
3212
3213
3214
3215
3216
3217
Heikkinnen et al. (2001) exposed groups of 50 female CBA/S mice irradiated with X-rays to 902 MHz
RF fields, either continuous (frequency modulated) at a whole-body SAR of 1.5 W/kg or pulsed (GSM
modulation) at an SAR of 0.35 W/kg for 1.5 hour per day, 5 days per week. They did not observe an effect on
the development of lymphomas or other tumours and on survival. The X-rays were delivered during the first
three weeks of the study in three fractions, and the exposures to RF radiation continued for 1.5 years. The only
statistically significant differences in tumour incidences (out of several dozen investigated) were a decreased
incidence of glandular polyps in the continuous wave group (p=0.011), and a decreased incidence of benign
pheochromocytomas of the adrenal glands in both RF field groups (continuous field: p=0.041; pulsed field:
p<0.039). The results were corrected for multiple comparisons.
3218
Skin tumours
3219
3220
Topical application of two different chemical initiators have been used to induce skin tumours:
benzo(a)pyrene (BAP) and 7,12-dimethylbenz(a)anthracene (DMBA).
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
Szudzinski and co-workers reported significant dose-dependent acceleration of the development of
BAP-induced skin tumours in groups of 40 Balb/c mice irradiated with 2.45 GHz RF fields at about 2–6 W/kg
(Szudzinski et al., 1982). Just like the study on mammary tumours by the same group (Szmigielski et al., 1982)
this study is included in WHO (1993b), but will be discussed here also for completeness. Exposure to both BAP
and RF fields were long-term (6 months), 2 hours daily for 6 days per week. Enhanced development of skin
tumours was reported also if the RF exposure was for 1–3 months prior to the beginning of BAP exposures.
[Some of these results appear to have been reported by Szmigielski et al. (1982). There are some inconsistencies,
however, between these two reports (e.g. the group size and the exact handling of the sham-exposed animals)
which complicate the interpretation of the results.] The authors detected no increase of rectal temperature, but
admitted that at the highest exposure level formation of significant ‘hot-spots’ was possible due to non-uniform
absorption of RF energy. [As mentioned previously with the study of Szmigielski et al. (1982), whole-body
exposure to the highest SAR level can certainly result in heating of the animals, but also an SAR of 2 W/kg is
already a significant systemic challenge to a mouse with a metabolic rate of 4–7 W/kg under thermal neutral
conditions. An increased body temperature can result in acceleration of tumour growth (Dickson, 1977; Wile,
Nahabedian & Mason, 1983) although not all tumours seem to respond similarly to moderate heating (Sen et al.,
2011).]
3237
3238
3239
3240
3241
3242
3243
Chagnaud and Veyret exposed BAP-treated female Sprague Dawley rats to RF fields at whole-body
SARs of 0.075 or 0.27 W/kg (Chagnaud & Veyret, 1999). Neither exposure was observed to have an effect on
the appearance of BAP-induced subcutaneous sarcomas or on survival. Similarly, the levels of antiphosphatidylinositol auto-antibodies, a suggested marker of malignant transformation, were not higher in RF
field-exposed animals than in sham-exposed animals. The exposure to RF radiation was for 2 hours per day for 2
weeks beginning on day 20, 40 or 75 after BAP injection. [The exposed groups were small: only 7–9 animals per
group.]
3244
3245
3246
3247
3248
3249
3250
RF field exposures have also not been observed to induce tumours in the skin of DMBA-treated CD-1
mice (Imaida et al., 2001) or ICR mice (Huang et al., 2005). The animals were subjected to topical application of
DMBA on dorsal skin a week before the beginning of RF field exposures. Imaida et al. exposed groups of 30 or
48 animals at a whole-body SAR of 2 W/kg, while Huang et al. used groups of 20 animals for exposure to a
whole-body SAR of 0.4 W/kg. No skin tumours were observed either in sham or RF-field-exposed animals
during the 19-week studies, or not even after a one-year follow-up (Huang et al., 2005), whereas a clear tumour
response was observed in the positive control animals exposed to repeated topical treatment with the classical
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
126
3251
3252
3253
3254
tumour promoter phorbol-12-myristate-13-acetate (PMA). RF field exposure was not found to affect either the
epidermal thickness (Huang et al., 2005; Imaida et al., 2001) or expression of proliferating cell nuclear antigen
(Huang et al., 2005). No difference was observable in the incidence of lymphoma (Imaida et al., 2001) (only
liver, kidney, adrenal glands and spleen were evaluated for lymphomas).
3255
3256
3257
3258
3259
3260
3261
3262
Paulraj and Behari (2011) exposed Swiss albino mice to either a 16-Hz amplitude-modulated 112
MHz signal or to a 2.54 GHz RF field for 2 hours per day, 3 days per week and 16 weeks. Exposures started
1 week after application of DMBA to the skin. The SARs were calculated to be 0.75 W/kg and 0.1 W/kg, for the
112 MHz and 2.54 GHz signals, respectively. No tumours were observed in the groups with DMBA only, with
RF only or in the DMBA+RF group, while application of DMBA and croton oil (that contains the tumour
promotor TPA) induced skin tumours in 10/14 animals. [Although the experimental groups initially each
contained 18 animals, the ‘effective’ number given in Table 1 in the publication is 2–4 less. It is not clear why
this is the case.]
3263
3264
3265
3266
3267
3268
3269
3270
Heikkinen et al. (2003) reported that exposure for 1.5 h per day, 5 days per week during 52 weeks to
pulsed 849 or 902 MHz RF radiation with two modulation characteristics (GSM or DAMPS) did not
significantly affect the development of skin tumours induced by UV radiation in female ODC-transgenic mice
(K2) or in their non-transgenic littermates (n=45–49 per group). Skin tumours were induced by exposure to
solar-simulating UV radiation three times a week during the whole study. The development of skin tumours was
faster in RF field-exposed animals than in the control group exposed to UV radiation only. This was consistently
seen with both RF signals and in both transgenic and non-transgenic animals, but it did not reach statistical
significance even in a combined analysis.
3271
Mammary tumours
3272
3273
DMBA is also used to induce mammary gland tumours in rodents. Several studies have investigated
the effects of RF field exposure in this experimental system.
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
The study of Bartsch and co-workers (2002) differs from most other RF field studies published this
far, in that the daily RF exposure time was long (nearly all-time exposure). The study involved exposures of
freely moving Sprague Dawley rats (12 per cage) at low whole-body SAR levels (0.08 W/kg or below) to a GSM
900 MHz signal. The study consisted of three experiments, each started exactly at the same time of the year on
three consecutive years. The animals were exposed continuously until “practically all animals had developed a
macroscopic mammary tumour” and the last experiment was conducted in a blinded fashion. In one of the three
experiments median latency for the development of the first malignant tumour was statistically significantly
extended in the RF field-exposed group. This finding was not supported by the two other experiments. The
overall conclusion is that in these experiments long term exposure to RF radiation had no significant effect on
the development of DMBA-induced mammary tumours in Sprague Dawley rats.
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
The study of Anane et al. (2003) consisted of two separate experiments. Female Sprague Dawley rats
were administered DMBA 10 days before the beginning of the GSM 900 MHz exposures, which were given for
2 hours per day, 5 days per week during 9 weeks. Each experiment contained 14–16 rats per group. The wholebody SAR levels were 1.4, 2.2 and 3.5 W/kg in the first experiment and 0.1, 0.7 and 1.4 W/kg in the second. In
the first experiment, an increased incidence rate of malignant mammary tumours was observed with 1.4 and 2.2
W/kg (p=0.02 and 0.04), but not with the highest exposure level. In the second experiment a decreased rate of
incidence was observed with 1.4 W/kg (p=0.04) and no effect with the two lower exposure levels. No effect was
observed on tumour latency (time of detection of the first tumour). For tumour multiplicity (the number of
tumours per animal) no effects were observed, except for the group exposed to 1.4 W/kg in the second
experiment, where the multiplicity was lower than in the sham-exposed group (p<0.01). The two experiments
were thus not consistent, but overall showed no effect of exposure.
3295
3296
3297
3298
3299
3300
3301
3302
3303
The study by Yu et al. (2006) did not provide evidence for effects of exposure to GSM 900 MHz
fields on the development of DMBA-initiated development of mammary tumours in Sprague Dawley rats.
Exposure levels up to 4 W/kg were covered and 100 animals per group were used. Exposure was for 4 h per day,
5 days per week and 26 weeks. The incidence of mammary gland adenocarcinomas was slightly lower in the
group exposed to the lowest SAR (0.44 W/kg), but the tumours were slightly larger compared to the animals
exposed to DMBA only. A slightly enhanced development of adenocarcinomas was found at the highest SAR
level. However, none of these differences were statistically significant. Significant differences were observed
between the cage-controls and the other experimental groups, with increased body weight and higher number
and more rapid development of mammary tumours in the cage-control group. These differences are most likely
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
127
3304
3305
3306
3307
3308
3309
related to restraint of the sham-exposed and RF-exposed animals. [The statistical analysis of tumour appearance
was apparently done without making a distinction between tumours observed during the study by palpation and
tumours detected in histopathological evaluation. While this could in principle mask differences between the
groups (also small non-palpable tumours are detected in histopathology), the data shown in the paper suggest
that a different statistical analysis would not essentially change the conclusion that RF radiation did not promote
mammary tumour development.]
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
Hruby et al. (2008) used a study design similar to that used by Yu et al. (2006), exposing the 100
Sprague Dawley rats per group also for 4 h per day, 5 days per week and 6 months. There were several
statistically significant differences between the GSM 902 MHz field-exposed groups and the sham-exposed
group. All RF-exposed groups had significantly more palpable mammary gland tissue masses than the shamexposed group (p<0.05), but there were no differences between the three groups exposed with whole body SARs
of 0.4, 1.3, or 4.0 W/kg respectively. The incidence of malignant mammary tissue tumours was lowest in the
sham-exposed group, and significantly increased in the high-exposure group (p<0.05). However, the incidence of
benign tumours was significantly lower in the three RF-exposed groups than in the sham-exposed group
(p<0.05). The number of animals with benign or malign neoplasms was similar in the sham-exposed group and
in the three RF-exposed groups. The cage-control group had the highest incidence and malignancy of neoplasms
among all groups. Given that the DMBA mammary tumour model is known to be prone to high variations in the
results, the authors’ interpretation was that the differences between the groups were coincidental. Comparison
with the results of the almost identical study by Yu et al. (2006) supports this conclusion: both studies reported
similar development of mammary tumours in three groups, but lower rate of development (seen in the
appearance of palpable tumours and/or reduced malignancy) in one group. Hruby et al. found the lowest rate of
development in the sham-exposed group, while Yu et al. found it in the 0.44 W/kg group. [Both studies
consistently reported the highest incidence of tumours in the cage-control group, which is most likely related to
the different handling of the cage-control animals (different stress level, differences in food intake).]
3328
Liver tumours
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
Exposure to 1.439 GHz (Imaida et al., 1998a) or 929.2 MHz (Imaida et al., 1998b) pulsed RF
radiation at whole-body SARs of 0.4–0.8 W/kg (maximum local SAR in the liver 0.9–2.0 W/kg) was used to
study effects on promotion of liver cancer in Fisher 344 rats. Groups of 48 rats were exposed to
diethylnitrosamine (DEN), partially hepatectomised a few weeks later, and then exposed to RF fields for 1.5 h
per day, 5 days per week during 6 weeks. The development of gluthathione S-transferase positive liver foci, a
preneoplastic rat liver lesion used as an endpoint marker in this assay, was slightly decreased in the RF-exposed
when compared to the sham-exposed animals in both studies, the difference being statistically significant at
1.439 GHz (p<0.05). When compared to the unrestrained cage-control animals, the level in the RF-exposed
animals was about the same. The level in the sham-exposed animals was slightly but significantly higher than
that in the cage-controls (p<0.05).
3339
Brain tumours
3340
3341
Several animal studies have evaluated the effects of low-level RF fields on the development of
tumours initiated by transplacental administration of the known genotoxic agent n-ethylnitrosourea (ENU).
3342
3343
3344
3345
3346
3347
Adey et al. (1999) exposed Fisher 344 rats before and after birth to an NADC 835 MHz signal at
SARs in the brain of 0.33–0.53 W/kg. The groups of 56 or 60 animals were exposed for 2 hours per day, 7 days
per week for 2 years. They observed a reduced incidence of ENU-induced CNS tumours in the RF-exposed
group (a similar tendency was seen also in spontaneous tumours, see section 12.2.1.1), but the difference was not
statistically significant. RF exposure did not statistically significantly affect the mortality of ENU-treated
animals, although survival was slightly increased in the RF-exposed group.
3348
3349
3350
In similar experiments but using another type of signal, FM 836 MHz at brain SARs of 1.0 or 1.2
W/kg, and groups of 90 animals, Adey et al. (2000) did not observe any effect of RF exposure on survival and
CNS tumour incidence.
3351
3352
3353
3354
3355
Shirai et al. (2005) exposed groups of 100 Fisher 344 rats born from ENU-treated mothers to a 1439
MHz TDMA signal at brain SARs of 0.67 or 2.0 W/kg for 90 min per day, 5 days per week for 104 weeks. They
observed no significant effect of RF exposure on brain tumours, although the incidence in females was slightly
lower in both exposed groups compared to the sham-exposed. The incidence of pituitary tumours showed a
tendency for increase in both sexes treated with ENU compared to the cage-control animals, the differences
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
128
3356
3357
3358
3359
3360
3361
3362
3363
being more consistent for females. Compared to the sham-exposed group, incidence of pituitary tumours was
decreased in males of both RF-exposed groups. At the higher RF-exposure level (2.0 W/kg) the decrease was
statistically significant (p<0.01), and the incidence was slightly but not significantly decreased in females. The
authors questioned the biological meaning of high pituitary tumour incidence in their study, and stated that the
incidence may still be within the wide range of background data of this strain. An earlier study did not report any
effect of RF on tumour development in pituitary glands of ENU treated Sprague Dawley rats (Zook & Simmens,
2001) (see below), but the proportion of pituitary gland tissues available for histology was only about 80% in
this study.
3364
3365
3366
3367
3368
3369
3370
3371
3372
A later study from the same group (Shirai et al., 2007) had a similar protocol, but a different mobile
phone signal was used (1.95 GHz WCDMA versus 1.439 GHz TDMA used in the first study). In contrast to the
previous study, brain tumour incidence in both females and males tended to be higher in the RF-exposed than in
the sham-exposed group, but no statistically significant differences were reported. [The statistical method used
(two-group comparisons with Fisher’s exact test) is not sensitive for detecting trends with increasing exposure
level. Using combined female and male data from the paper, chi-squared test for trend shows a p-value of 0.0395
for an increasing trend from the sham-exposed group to the highest exposure group.] No differences in pituitary
tumours were observed in this study. [Given the inconsistent findings and opposite trends observed in these two
studies (Shirai et al., 2005; Shirai et al., 2007), the differences observed are most likely incidental.]
3373
3374
3375
3376
3377
3378
3379
Zook and Simmens (2001) treated pregnant female Sprague Dawley rats with two different doses of
ENU and exposed the offspring to 860 MHz pulsed or continuous-wave RF EMF, at a brain SAR of 1 W/kg. The
exposures were for 6 h per day, 5 days per week and 22 months and the groups counted 60 animals. Neither
continuous nor pulsed field exposure significantly affect incidence, volume, multiplicity, malignancy or fatality
on ENU-induced brain tumours or development of tumours in eight other organs. There was a slight statistically
non-significant tendency toward higher incidence of fatal brain tumours in the group treated with the higher level
of ENU and exposed to the pulsed RF field.
3380
3381
3382
3383
3384
3385
3386
3387
3388
In a follow-up study, Zook and Simmens (2006) investigated further any promoting effect of the
pulsed RF signal. Latency and other characteristics of neurogenic tumours were investigated in the progeny of
pregnant Sprague Dawley rats treated with two doses of ENU. The offspring were exposed to pulsed RF, sham
exposed or kept as cage-controls. The exposure to the 860 MHz RF field was for 6 h per day, 5 days per week at
a SAR of 1.0 W/kg averaged over the brain (0.27–0.42 W/kg averaged over the whole body). An equal number
of rats from each group were killed every 30 days between the ages of 171 and 325 days; 32 rats died and 225
rats were killed when they were moribund. All rats were necropsied and the brain and spinal cord were examined
histopathologically. No evidence was found of an effect of RF exposure on the incidence, malignancy, volume,
multiplicity, latency or fatality associated with any kind of brain tumour.
3389
Colon tumours
3390
3391
3392
3393
Wu et al. investigated the effect of exposure to 2.45 GHz RF radiation on the development of
dimethylhydrazine (DMH)-induced colon tumours in groups of 26–32 Balb/c mice (Wu et al., 1994). Exposure
was at a relatively high whole-body SAR of 10–12 W/kg for 3 h per day, 6 days per week and 5 months. No
difference in tumour incidence was observed between exposed and sham-exposed groups.
3394
Multiple tumours
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
Heikkinen et al. (2006) evaluated possible effects of exposure to RF fields on tumours induced by the
mutagen and multisite carcinogen 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX) given
continuously during the experiment in drinking water. Female Wistar rats were exposed for 2 h per day, 5 days
per week, and 104 weeks to 900 MHz RF (GSM modulation) at whole-body SARs of 0.3 or 0.9 W/kg (72 rats
per group). No statistically significantly different mortality or organ-specific incidence of any tumour type was
observed in the exposed compared to sham-exposed groups. The only statistically significant difference was an
increase in the combined frequency of vascular tumours of the mesenteric lymph nodes in the high-RF group
compared to the sham-RF group (p<0.05). However, comparison to cage-control animals suggested that this
difference was due to unusually low frequency of this type of tumours in the sham-RF group, rather than high
frequency in the high-RF group.
3405
3406
3407
Tillman et al. (2010) continuously exposed groups of 54–60 female C57Bl/6N and male C3H/HeN
mice for 24 months to a UMTS-like signal at power densities of 4.8 or 48 W/m2; exposure started at the 6th day
of pregnancy. The genotoxic agent n-ethylnitrosourea (ENU) was given to pregnant females at the 14th day of
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
129
3408
3409
3410
3411
3412
3413
pregnancy. An increased number of lung tumours was found after exposure to the lower power density
(p<0.001–0.05), but not after the high one when compared to the ENU-only group. No effect of the UMTS
exposures was seen on any other type of tumour. A striking finding was the increase in liver tumours in all
groups, including controls. This could be explained by an initially undetected Helicobacter hepaticus infection
that appeared to be already present in the parent mice. [The lack of an ENU+sham exposure group makes it
difficult to draw definitive conclusions on this study.]
Table 12.2.3. Studies on carcinogenesis in rodents using RF and genotoxic/carcinogenic agents.
Animals, number
per group, age at
start
Exposure: source,
schedule, level,
freely moving or
restrained
Response
Comment
Reference
Mouse: CBA/S,
normal (n=50)
NMT 902 MHz, GSM
902 MHz
Heikkinen et al.
(2001)
3-5 weeks
1.5 h/d, 5 d/week, 78
weeks
No effect on X rayinduced lymphomas,
other neoplasms or
survival.
Acceleration of skin
Heat stress may
cancer; effect chronic have accelerated
stress same as WBA tumour growth.
SAR 2-3 W/kg.
Szmigielski et al.
(1982)
Dose-dependent
acceleration of skin
cancer; in part same
data as Szmigielski
et al. (1982);
comparison of
different treatment
regimes.
Szudzinski et al.
(1982).
Lymphoma
Lymphomas and
other tumours
Induction by X rays
WBA SAR: NMT: 1.5
W/kg; GSM: 0.35
W/kg
(Lymphoma
Restrained
incidence 24%
without RF exposure)
Skin tumours
Mouse: Balb/c,
normal (n=40)
6 weeks
2450 MHz
2 h/d, 6 d/week, 10.5
months
Initiated by BAP
WBA SAR 2-3, 6-8
W/kg
(Skin tumour
incidence 50%
Free
without RF exposure)
Mouse: Balb/c,
normal (n=100)
Age not provided
2450 MHz
2 h/d, 6 d/week, 1, 2,
3, 6 months
Initiated by BAP
WBA SAR 2-3, 6-8
W/kg during BAP; 4
(Skin tumour
W/kg before BAP
incidence 95%
without RF exposure) Free
Rat: Sprague
GSM 900 MHz
Dawley, normal (n=7- 2 h/d, 2 weeks
9; control: n=6)
WBA SAR
2 months
0.075±0.025,
Initiated by BAP
0.27±0.09 W/kg
(Malignant sarcoma
development 100%)
Heat stress may
have accelerated
tumour growth.
No effect on tumour
development or
growth, and on
survival.
Chagnaud & Veyret
(1999)
No effect on skin
tumour development.
Imaida et al. (2001)
No effect on skin
tumour.
Huang et al. (2005)
Restrained
Mouse: CD-1, normal TDMA 1.5 GHz
(n=30, 48)
1.5 h/d, 5 d/week, 19
5 weeks
weeks
Initiated by DMBA
Skin peak SAR 2.0
(No macroscopic skin W/kg
tumours without RF
Restrained
exposure)
Mouse: ICR, normal
(n=20)
849, 1763 MHz
6 weeks
2x45 min/d, 5
d/week, 19 weeks
Initiated by DMBA
WBA SAR 0.4 W/kg
(Skin tumour
Free
incidence 0% without
RF exposure)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
130
Mouse: Swiss albino, 112 MHz, 16 Hz AM;
normal (n=18)
2.54 GHz
7-8 weeks
Skin tumour
Initiated by DMBA
(Skin tumour
incidence DMBA:
0%; DMBA+TPA:
71%)
Mouse: K2 (ODC
transgenic) and wildtype (n=45-49; cage
control: n=20)
12-15 weeks
Initiated by UV
No effect on skin
tumour.
Small number of
animals.
Paulraj & Behari
(2011)
2 h/d, 3 d/week, 16
weeks
112 MHz: WBA SAR
0.75 W/kg; 2.54 GHz:
WBA SAR 0.1 W/kg
Free
GSM, DAMPS
1.5 h/d, 5 d/week, 52
weeks
No effect on skin
tumour.
Heikkinen et al.
(2003)
No effect on tumour
latency and
incidence.
Bartsch et al. (2002)
No effect on tumour
latency, volume,
multiplicity; in 1st of 2
experiments inverse
dose relation for
incidence.
Anane et al. (2003)
WBA SAR 0.5 W/kg
Restrained
(Skin tumour
incidence 22%
without RF exposure)
Mammary tumours
Rat: Sprague
Dawley, normal
(n=60)
38+13 d
acclimatization;
43+8 d
acclimatization;
34+17 d
acclimatization
Initiated by DMBA
GSM 900 MHz
Continuously until
“practically all
animals had
developed a
macroscopic
mammary tumour”
WBA SAR max 0.08
W/kg
Free
(Malignant mammary
tumour incidence 7991% without RF
exposure)
Rat: Sprague
Dawley, normal
(n=16; cage control:
8)
GSM 900 MHz
2 h/d, 5 d/week, 9
weeks
55 d
WBA SAR 0.1, 0.7,
1.4, 2.2, 3.5 W/kg
Initiated by DMBA
Restrained
(Malignant mammary
tumour incidence
60% without RF
exposure)
Rat: Sprague
Dawley, normal
(n=100)
GSM 900 MHz
36+12 d
acclimatization
WBA SAR 0.44, 1.33,
4.0 W/kg
Initiated by DMBA
Restrained
4 h/d, 5 d/week, 26
weeks
No effect on tumour
latency, size,
multiplicity,
incidence.
Incidence and
latency higher in
cage-controls than in
exposed.
Yu et al. (2006)
(Mammary tumour
incidence 45%
without RF exposure)
Rat: Sprague
Dawley, normal
(n=100)
GSM 902 MHz
5 weeks + 12 d
acclimatization
WBA SAR 0.4, 1.3,
4.0 W/kg
Initiated by DMBA
Restrained
(Mammary tumour
incidence 60%
without RF exposure)
4 h/d, 5 d/week, 6
months
More palpable
Incidence higher in
mammary glands but cage-controls than in
no dose response
exposed.
relation; more
malignant mammary
tumours in 4 W/kg
group; fewer benign
tumours in all
exposed groups; no
effect on number of
animals with benign
or malign neoplasms.
Hruby et al. (2008)
Liver tumours
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
131
Rat: Fischer 344,
normal (n=48)
5+1 week
acclimatization
Initiated by
diethylnitrosamine
TDMA 929 MHz
1.5 h/d, 5 d/week, 6
weeks
No effect on number
of foci and size.
Serum hormones
discussed in section
07.2.
Imaida et al. (1998b)
No effect on number
of foci and size.
Serum hormones
discussed in section
07.2.
Imaida et al. (1998a)
Non-significantly
reduced tumour
incidence and
increased survival.
See Table 12.2.1 for
results regarding
spontaneous
tumours.
Adey et al. (1999)
No effect on survival
and tumour
incidence.
See Table 12.2.1 for
results regarding
spontaneous
tumours.
Adey et al. (2000)
Lower incidence of
Serum hormones
pituitary tumours in
discussed in section
males after high level 07.2.
RF, no other RF
effect on neurogenic
or other tumours.
Shirai et al. (2005)
No effect on
neurogenic or other
tumours.
Shirai et al. (2007)
WBA SAR 1.7-2.0
W/kg
Restrained
(Liver foci incidence
100% without RF
exposure)
Rat: Fischer 344,
normal (n=48)
5+1 week
acclimatization
Initiated by
diethylnitrosamine
TDMA 1439 MHz
1.5 h/d, 5 d/week, 6
weeks
WBA SAR 0.9371.91 W/kg
Restrained
(Liver foci incidence
100% without RF
exposure)
CNS tumours
Rat: Fischer 344,
normal (n=56, 60)
Gestation d 15 + 3 d
acclimatization
Initiated by ENU
NADC 835 MHz
2 h/d, 7 d/week
(prenatal), 4 d/week
(postnatal), 2 years
Brain SAR 0.33-0.53
W/kg
(CNS tumour
incidence 17%
Free (prenatal),
without RF exposure) restrained (postnatal)
Rat: Fischer 344,
normal (n=90)
Gestation d 15 + 3 d
acclimatization
FM 836 MHz
2 h/d, 7 d/week
(prenatal), 4 d/week
(postnatal), 2 y
Initiated by ENU
Brain SAR 1.0, 1.2
W/kg
(CNS tumour
incidence 22%
Free (prenatal),
without RF exposure) restrained (postnatal)
Rat: Fischer 344,
normal (n=30)
Gestation d 18
Initiated by ENU
(Brain tumour
incidences 24% in
males and 30% in
females without RF
exposure)
Rat: Fischer 344,
normal (n=100)
Gestation d 18
Initiated by ENU
(Brain tumour
incidences 10% in
males and 8% in
females without RF
exposure)
TDMA 1439 GHz
90 min/d, 5 d/week,
104 weeks
Brain SAR 0.67, 2.0
W/kg
Restrained
WCDMA 1.95 GHz
90 min/d, 5 d/week,
104 weeks
Brain SAR 0.67, 2.0
W/kg
Restrained
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
132
Rat: Sprague
Dawley, normal
(n=60)
860 MHz pulsed, CW No effect on
6 h/d, 5 d/week, 22
neurogenic or other
months
tumours.
Gestation d 15
Brain SAR 1.0±0.2
W/kg
Initiated by ENU
(Brain tumour
incidence 10-16%
(low-ENU) and 58%
(high-ENU) without
RF exposure)
Restrained
Rat: Sprague
Dawley, normal
(n=60)
860 MHz pulsed
Gestation d 15
Brain SAR 1.0±0.2
W/kg
Initiated by ENU
6 h/d, 5 d/week, 325
d
See Table 12.2.1 for
results regarding
spontaneous
tumours.
Zook & Simmens
(2001)
No effect on
incidence,
malignancy, volume,
multiplicity, latency or
fatality of neurogenic
tumours.
Zook & Simmens
(2006)
No effect.
Wu et al. (1994)
Replication of Zook
& Simmens (2001)
(Brain tumour
Restrained
incidence 50%
without RF exposure)
Colon tumour
Mouse: Balb/c,
normal (n=26-32)
4-5 wk
2450 MHz
3 h/d, 6 d/week, 5
months
Initiated by DMH
WBA SAR 10-12
W/kg
(Colon tumour
incidence 46%
Restrained
without RF exposure)
Multiple tumours
Rat: Wistar, normal
(n=72)
7 weeks
GSM 900 MHz
2 h/d, 5 d/week, 104
weeks
Initiated by MX
WBA SAR 0.3, 0.9
W/kg
(Malignant tumour
incidence 51%
Free
without RF exposure)
Mouse: C57Bl/6N &
C3H/HeN , normal
(n=54-60)
6 d post conception
Initiated by ENU
UMTS continuously
24 months
0, 4.8, 48 W/m2
Free
Increase in frequency
of combined vascular
tumours of the
mesenteric lymph
nodes with 0.9 W/kg.
No effect on organ
specific tumours and
survival.
Unusually low
Heikkinen et al.
frequency of vascular (2006)
tumours of the
mesenteric lymph of
tumours in the shamexposed group.
Increased number of ENU+sham exposure Tillmann et al. (2010)
lung tumours after
group missing.
low, but not after high
exposure; no effect
on other tumours.
(Malignant tumour
incidence up to 55%
(lung carcinomas)
without RF exposure)
Abbreviations: AM: amplitude modulation; BAP: benzo(a)pyrene or 3,4 benzopyrene; CNS: central nervous system;
CW: continuous wave; DAMPS: Digital Advanced Mobile Phone System; DMBA: 7,12-dimethybenz[a]anthracene;
DMH: dimethylhydrazine; ENU: ethylnitrosourea; FM: Frequency Modulation; GSM: Global System For Mobile
Communication; NADC: North American Digital Cellular; NMT: Nordic Mobile Telephony; MX: 3-chloro-4(dichloromethyl)-5-hydroxy-2(5H)-furanone; WBA SAR: whole-body averaged SAR; TDMA: Time Division Multiple
Access; TPA: 12-O-tetradecanoylphorbol-13-acetate; UMTS: Universal Mobile Telecommunications Signal; UV:
ultraviolet (radiation); WCDMA: Wideband Code Division Multiple Access
3414
3415
12.2.3.
Effects of RF radiation on transplanted tumour
3416
3417
3418
3419
3420
3421
3422
Szmigielski et al. (1982) exposed BALB/c mice injected with sarcoma cells to 2.45 GHz microwaves
for 2 h per day, 6 days per week, during 1, 2 or 3 months at whole-body SARs of 2–3 and 6–8 W/kg.
Significantly elevated numbers of neoplastic lung foci nodules after both 1 and 3 months were reported at both
RF exposure levels (p<0.01 with the highest level). [The interpretation of this study is complicated by the fact
that methods are not described in detail. As mentioned previously, whole-body exposure to the highest SAR
level can certainly result in heating of the animals, but also an SAR of 2 W/kg is already a significant systemic
challenge to a mouse with a metabolic rate of 4-7 W/kg under thermal neutral conditions. An increased body
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
133
3423
3424
temperature can result in acceleration of tumour growth (Dickson, 1977; Wile, Nahabedian & Mason, 1983)
although not all tumours seem to respond similarly to moderate heating (Sen et al., 2011).]
3425
3426
3427
3428
Santini and co-workers (1988) exposed groups of 15 C57/6J mice lifelong for 2.5 h per day, 6 days
per week to continuous or pulsed 2.45 GHz RF fields at an whole body SAR of 1.2 W/kg. After 15 exposure
days animals were subcutaneously implanted with B16 melanoma cells. No significant effects of RF radiation on
tumour development or survival times were observed.
3429
3430
3431
3432
3433
Paulraj and Behari (2011) injected Swiss albino mice intraperitoneally with ascites cells and exposed
or sham-exposed them for 2 h per day during 14 days to either 112 MHz RF EMF amplitude modulated at 16 Hz
(SAR=0.75 W/kg), or 2.54 GHz (SAR=0.1 W/kg). No difference was observed between sham and exposed
groups in terms of number of ascites cells, body weight and survival. [With only 8 animals per group, the groups
were very small, however.]
3434
3435
3436
3437
Higashikubo et al. (1999) implanted male Fischer 344 rats with gliosarcoma cells and exposed them
for 4 h per day, 5 days per week, during 4 weeks to an 835 MHz continuous wave FM signal or a 847 MHz
CDMA signal at a brain SAR of 0.75±0.25 W/kg. They observed no effect of RF exposure on survival,
independent of the number of cells injected.
3438
Studies not included in the analysis
3439
3440
3441
3442
3443
3444
3445
3446
Salford et al. (1993; 1997) pairwise implanted male and female Fischer 344 rats with glioma cells.
One animal was sham exposed and one exposed to a 915 MHz RF field either continuous or pulse modulated at
4, 8, 16, 50 or 200 Hz. The output of the pulses was kept constant, therefore the SAR varied from 0.0077 to 1.67
W/kg. The animals were exposed for 7 h per day, 5 days per week for up to 3 weeks. In some pairs tumour
growth in the exposed animal was faster than in the sham control, but overall there was no significant effect of
exposure on tumour growth. [The animals used in these studies were not well-defined; they had a range of body
weights and presumably of ages. The use of inoculated cells with varying viability resulted in a large variation in
tumour growth. Consequently the experimental conditions show too much variation for a meaningful analysis.]
Table 12.2.4. Studies on effects of RF on growth and development of implanted tumours in rodents.
Animals, number
per group, age at
start
Exposure: source,
schedule, level,
freely moving or
restrained
Response
Comment
Reference
2450 MHz
Increased incidence
of metastatic tumour
colonies on lung
surface, more
pronounced at the
higher exposure
level.
Difficult to interpret
Szmigielski et al.
because methods are (1982)
not described in
detail, group size
unknown.
Lung tumour
Mouse: Balb/c,
normal, injected L1
sarcoma cells
(n=unknown)
6 weeks
(2.8±1.6 lung
colonies without RF
exposure)
2 h/d, 6 d/week, 1, 2,
3 months
WBA SAR=2-3, 6-8
W/kg
Free
Heat stress may
have accelerated
tumour growth.
Skin tumours
Mouse: C57/6J
normal, implanted
melanoma (n=15)
5 weeks
(Average survival
with tumour below 4
weeks)
2450 MHz, pulsed,
CW
2.5 h/d, 6 d/week,
lifelong
No effect on
melanoma tumour
growth, survival.
Santini et al. (1988)
No effect on number Small number of
of ascites cells,
animals.
survival, body weight.
Paulraj & Behari
(2011)
WBA SAR 1.2 W/kg
Free
Ascites tumour
Mouse: Swiss albino, 112 MHz, 16 Hz AM;
normal (n=8)
2.54 GHz
7–8 weeks
2 h/d, 14 d
Ascites tumour
112 MHz: WBA SAR
0.75 W/kg; 2.54 GHz:
WBA SAR 0.1 W/kg
Free
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
134
CNS tumours
Rat: Fischer 344,
normal (n=12-67
(sham), 10-49
(FMCW), 22-38
(CDMA), groups from
2 experiments
combined)
FMCW 835 MHz,
CDMA 847 MHz
Inoculated 9L
gliosarcoma cells
Restrained
4 h/d, 5 d/week, 4
weeks
No effect on survival,
independent of
number of cells
injected.
Higashikubo et al.
(1999)
Brain SAR 0.75±0.25
W/kg
28 d
(Survival 0-27%
without RF exposure,
depending on
number of inoculated
cells)
Abbreviations: CW: continuous wave; CDMA: Code Division Multiple Access; FMCW: Frequency-Modulated
Continuous Wave; WBA SAR: whole-body averaged SAR
3447
3448
12.3
In vitro studies
3449
12.3.1
Genotoxicity
3450
3451
3452
3453
3454
In the previous WHO report (WHO, 1993a) very few studies relevant for genotoxicity were identified.
In most of these studies researchers did not observe significant effects of RF EMF exposure on DNA integrity in
cells, even under mild hyperthermic conditions, and the conclusion of the report was that RF EMF exposure is
not genotoxic. There were no studies examining the genotoxic effect of combined exposure of the cells to RF
EMF and chemical or physical agents.
3455
3456
3457
3458
3459
3460
The present literature search identified a large number of papers devoted to evaluating the effect of RF
EMF exposure, either alone or combined with chemical and/or physical agents. The aim of the latter type of
studies was to examine whether RF exposure modifies the genetic damage induced by known genotoxic agents.
Investigators have used recently developed genotoxicity techniques as well as classical cytogenetic methods,
such as, DNA single/double strand breaks (SB, mostly assessed using the comet assay), chromosomal
aberrations (CA), micronuclei (MN), sister chromatid exchanges (SCE) and mutations (MUT).
3461
3462
3463
3464
3465
3466
3467
3468
The search criteria are described in Appendix Y. The present search resulted in 75 papers. Moreover,
25 papers were obtained from other sources. Altogether 100 relevant publications were identified. Of these, 25
were excluded because they did not meet the inclusion criteria for in vitro studies (see Appendix X). These
studies are not described in the text but they are listed at the end of this section. The remaining 75 papers are
described in the text, but 12 of these did not comply with the quality criteria for inclusion as described in
Appendix X, e.g. because of inadequate description of the exposure system and/or dosimetry, or an inadequate
number of experiments. These studies are presented in a separate section at the end of the section and are not
included in the overall analysis. This section is thus based upon 63 publications.
3469
3470
Unless specifically mentioned, papers did not report on blinding of the investigators to the exposure
conditions.
3471
12.3.1.1
3472
3473
The great majority of studies on human cells have been carried out using freshly collected peripheral
blood lymphocytes, but other cells of different origin have also been investigated.
3474
Human peripheral blood lymphocytes
3475
3476
3477
3478
A great majority of lymphocytes (about 99.6 %) circulating in human blood are in the resting stage, or
G0 phase, of the cell cycle. They can be stimulated to enter the cell cycle with the addition of a mitogen,
phytohemagglutinin (PHA). In most of the studies reported below, samples of whole blood collected from
healthy volunteers (containing resting lymphocytes) were exposed to RF EMF. In some studies, isolated
Human cells
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
135
3479
3480
lymphocytes (either unstimulated or stimulated with PHA for varying periods of time) were also used in
experiments of such exposure.
3481
3482
3483
3484
3485
3486
Vijayalaxmi et al. (2000) investigated the effect of 2 h exposure at 2450 MHz RF EMF, pulsed wave
(pulse width of 10 µs and pulse repetition rate of 10 kHz) at a SAR of 2.135 W/kg. An alkaline comet assay was
used to assess the extent of SB in lymphocytes from blood samples of three healthy donors immediately after
exposure and at 4 h post exposure to assess DNA damage and repair, respectively. After blind exposure, no
significant differences between RF- and sham-exposed samples were observed at any of the time points. Cultures
treated with gamma rays served as positive controls and gave positive findings.
3487
3488
3489
3490
3491
3492
3493
3494
3495
The extent of SB was also investigated by Chemeris et al. (2006) to test the potential genotoxicity of
high power microwave pulses (HPMP; 8.8 GHz, 180 ns pulse width, average SAR 1600 W/kg, pulse repetition
frequency 50 Hz) in whole-blood leukocytes and isolated lymphocytes from 36 donors. Samples of whole blood
from 18 donors were exposed for 40 min and leukocytes were examined immediately after exposure while those
from eight donors were tested at 30 min post exposure. Isolated lymphocytes from the remaining 10 donors were
examined immediately after RF exposure. After blinded analysis, no significant differences in SB were observed
between RF EMF exposed samples and sham controls. In cultures treated with ethylmethanesulphonate (EMS,
an alkylating agent) as positive control, a significant increase in SB was recorded. [Although the authors refer to
control samples, they are actually sham-exposed samples].
3496
3497
3498
3499
Sannino et al. (2006) also failed to detect significant effects on SB in isolated leukocytes from six
donors exposed in blind for 24 h to a 1950 MHz UMTS RF EMF signal, at a SAR of 0.5 or 2 W/kg. Hydrogen
peroxide was used as positive control and showed positive findings. In this study cell viability was investigated
and resulted unaffected.
3500
3501
3502
3503
3504
Baohong et al. (2007) carried out blind exposures of isolated leukocytes from three healthy donors to
an 1800 MHz GSM signal at a SAR of 3 W/kg. Two exposure durations were tested, 1.5 and 4 h. The comet
assay did not reveal any significant effect of RF exposure on SB in any of the exposure conditions. Treatments
with chemical mutagens (mitomicin-C, bleomicin, methyl-methane sulphonate) employed as positive controls,
induced SB, as expected. In this study, the effect of co-exposures was also investigated.
3505
3506
3507
3508
3509
Zhijian et al. (2009) exposed isolated leukocytes from four healthy donors to an 1800 MHz GSM
signal in blind conditions. Intermittent exposures (5 min on/10 min off cycles) were carried out for 24 h at a SAR
of 2 W/kg. The exposure did not induce SB, as assessed by the comet assay. In cultures exposed to x rays as
positive controls a significant increase in SB was recorded. In this study the effect of combined exposures was
also investigated.
3510
3511
3512
3513
3514
3515
3516
In a series of studies, Belyaev and his colleagues investigated the effect of RF EMF on chromatin
conformation and DNA repair proteins. The anomalous viscosity time dependence (AVTD) assay used is based
on changes in chromatin conformation due to the formation of large DNA-protein complexes which prevent the
access of DNA repair proteins to repair sites in the DNA. Moreover, the rapid phosphorylation of two checkpoint proteins, such as, 53BP1 and γ-H2AX, and their congregation in the vicinity of double strand breaks in the
DNA to provide scaffolding to repair sites have been demonstrated using specific antibodies. Such repair sites
can be visualized in cells as discrete foci which can be counted using a fluorescent microscope.
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
Sarimov et al. (2004) exposed isolated lymphocytes from seven donors and transformed cells
(immortalization of lymphocytes with Epstein–Barr virus from 1 donor) to RF EMF at different frequencies in
the range of 895–915 MHz (GSM modulation). The exposures were 30 to 60 min at an SAR of 5.4 mW/kg. The
results from the AVTD assay showed that 30-min exposure at 900 and 905 MHz resulted in a statistically
significant increase in condensation of chromatin in lymphocytes from one of three donors, while the analysis of
pooled data from all donors resulted in statistically significant effect (p<0.01). A stronger effect was observed in
four of five donors exposed for 1 h to 905 MHz and, the pooled data from all donors showed a statistically
significant effect (p<0.01). Transformed cells exposed to 895 and 915 MHz for 30 min also showed significant
chromatin condensation (p<0.05). Treatments with camptothecin as positive controls induced significant
chromatin condensation. [As stated by the authors, the measured effects are strongly dependent on frequencies
and vary among donors.]
3528
3529
3530
Belyaev et al. (2005) collected blood samples from seven healthy donors and seven electromagnetic
hypersensitive (EHS) individuals and exposed isolated lymphocytes to 915 MHz RF (GSM, SAR=37 mW/kg)
for 2 hours, in blind conditions. A significant increase in chromatin condensation (AVTD assay) was detected
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
136
3531
3532
3533
3534
3535
immediately after exposure (p<0.005) and at 2 h post-exposure (p<0.05). Analysis with the immunostaining
method indicated a distinct reduction in 53BP1 foci in cells from all individuals (p<0.05). These observations
were similar to those in cells subjected to heat-shock treatment (positive controls). No difference in response was
observed between lymphocytes from EHS and healthy subjects. Thus, the overall data suggested decreased
accessibility of 53BP1 protein to double strand breaks repair sites due to increased chromatin condensation.
3536
3537
3538
3539
3540
3541
Markova et al. (2005) used 905 or 915 MHz RF EMF (GSM signal, SAR=37 mW/kg) to expose cells
from five EHS and five healthy individuals for 1 h in blind conditions. A significant increase in chromatin
condensation and reductions in 53BP1 and γ-H2AX foci were found in cells exposed to 915 MHz (p<0.05) and
in cltures treated with gamma rays, as positive control. No effect was detected in cells exposed to 905 MHz.
There was no significant difference between healthy and EHS donors. [However, there was a large interindividual variation in cells exposed to 905 MHz RF.]
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
In a similar study, Belyaev et al. (2009) exposed the cells from five EHS and five healthy individuals
to 905 or 915 MHz RF EMF (GSM signal, mean SAR=37 mW/kg) as well as 1947 MHz (UMTS signal, mean
SAR=40 mW/kg) for 24 and 72 hours. RF exposure was performed in blind conditions. For all frequencies
tested, the SAR ranged from 15 to 145 mW/kg and about 50% of cells were exposed to a SAR between 20 and
40 mW/kg. The increased chromatin condensation and significantly reduced numbers of 53BP1 and γ-H2AX
foci were more pronounced in cells exposed to 1947 MHz compared to those exposed to 915 MHz, and the
decreased foci persisted up to 72 hours following RF exposure (p<0.05), indicating that the effect was not only
on double SB but also on their repair process. No significant differences were observed between cells exposed to
905 MHz RF and sham-exposed ones. As for the previous study, positive controls were set up by treating
cultures with gamma rays and worked properly. In this study also, no difference in response was observed
between lymphocytes from EHS and healthy subjects.
3553
3554
MN.
3555
3556
3557
3558
3559
3560
3561
3562
3563
Zeni et al. (2007b), used a Free Electron Laser equipment to deliver 120 and 130 GHz RF EMF (a
“train” of micropulses, each 50 ps long and 330 ps pulse width) for 20 min to whole blood samples from 17
healthy subjects. The 120 GHz exposure was tested at calculated SAR of 0.4 W/kg, while the 130 GHz exposure
was tested at SARs of 0.24, 1.4 and 2 W/kg. The extent of genotoxicity was evaluated in cultures established
after blind exposure by means of the cytokinesis block MN technique. Moreover, blood samples exposed to 130
GHz at SARs of 1.4 and 2 W/kg were also tested for SB using the comet assay immediately after exposure. The
results of both endpoints indicated that RF EMF exposure did not induce significant damage. [No positive
controls were included in the study design. In this investigation, cell proliferation has also been measured, as
reported in Section 12.3.6].
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
The same research group investigated the effect of a 900 MHz GSM signal at SARs of 0.3 and 1 W/kg
by exposing unstimulated human lymphocytes from nine donors for 2 h (Zeni et al., 2005). No differences
between RF EMF exposed and sham controls were detected in SB (five donors) and CA and SCE (four donors).
In a separate study, the researchers did not observe changes in MN in unstimulated (whole blood) or PHAstimulated lymphocytes from six donors intermittently exposed (6 min on/2h off cycles) to 1950 MHz (SAR=2.2
W/kg). The exposure duration was from 24 to 68 h to cover different stages of the cell cycle. Moreover, no
effects on SB were detected in unstimulated cells after 24 h exposure (Zeni et al., 2008). In both investigations
blind exposures were carried out and positive controls, assessed by treating cultures with mitomycin-C (MMC)
or methyl-methane sulphonate (MMS), gave the expected reults. [These studies have also been described in
Section 12.3.6 cell proliferation].
3574
3575
3576
3577
3578
3579
3580
No significant genotoxic effects were reported by Schwarz et al. (2008), in human peripheral blood
lymphocytes from three donors exposed in blind to a 1950 MHz UMTS signal (SAR=0.1 W/kg) for 16 h (5 min
on/10 min off cycles). Exposures were carried out on unstimulated cells to evaluate SB and on PHA-stimulated
cells to evaluate both SB and MN. Treatments of cell cultures with UV light and vincristine served as positive
controls for SB and MN, respectively and gave the expected results. In this paper the effect of RF exposure on
primary human fibroblasts has also been investigated and the results are reported in the next section (Other
human cells).
3581
3582
3583
McNamee and co-workers used whole blood samples from five healthy donors to investigate the
effect of 2 h exposure to RF EMF at a frequency of 1900 MHz (SAR=0.1–10 W/kg) on SB and MN induction
and did not find any significant difference between RF EMF and sham-exposed cells with CW (McNamee et al.,
In some studies the extent of SB was evaluated together with chromosomal damage, such as CA or
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
137
3584
3585
3586
3587
3588
3589
2002b) or pulsed wave exposure (1/3 duty cycle, 50 Hz, 3.54 W peak forward power, 1.18 W average forward
power) (McNamee et al., 2002a). In a follow up study, the authors confirmed the lack of effects when RF EMF
exposure duration was increased to 24 h (McNamee et al., 2003), for both CW and pulsed wave fields. In all
cases, the experiments were performed in blind and treatments with gamma rays served as positive controls.
[The results of the latter study do not confirm the findings reported by Tice and co-workers (Tice et al., 2002)
which have been described in the section Studies not included in the analysis].
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
Scarfi et al. (2006) conducted an investigation with the participation of researchers from another
independent laboratory. Lymphocytes from blood samples of ten donors were exposed for 24 h to a 900 MHz
GSM signal, at SARs of 1, 5 and 10 W/kg and then cultured for 72 h. None of the two laboratories found
significant changes in MN frequency at all SARs examined when compared with sham-controls. The same
research group also failed to find significant increase in MN frequency in human peripheral blood lymphocytes
from 14 donors exposed to a 900 MHz GSM signal (SAR=1.25 W/kg) for 20 h in several stages of the cell cycle
(Sannino et al., 2009b; 2011). In a follow-up study the authors also failed to find effects when 20 h-exposures to
a 1950 MHz, UMTS signal were given in the S phase of the cell cycle at SAR values of 0.15, 0.3, 0.6 and 1.25
W/kg (Zeni et al., 2012a). In these investigations RF exposure was carried out in blind conditions and cultures
treated with MMC served as positive controls and worked properly. Moreover, cell proliferation was also
investigated, as reported in Section 12.3.6. In the latter three studies the effect of combined exposures was also
evaluated.
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
In another study involving the participation of researchers from two independent laboratories, Stronati
et al. (2006) investigated the effect of a 935 MHz GSM signal at SAR values of 1 and 2 W/kg. Lymphocytes in
whole blood (G0 phase of the cell cycle) from 14 healthy donors were blindly exposed to RF EMF for 24 h and
DNA damage was evaluated by several assays, such as, the alkaline comet technique to detect SB (10 donors),
metaphase analyses to detect chromosomal aberrations (14 donors) and sister chromatid exchanges (4 donors),
and micronuclei in cytokinesis-blocked binucleate lymphocytes (14 donors). Comparison with appropriate shamexposed and control samples indicated no effect of RF EMF for any of the endpoints investigated. On the
contrary, treatments with x rays as positive controls gave the expected increase for all the endpoints. [In this
study the effect of combined exposures was also investigated. Results on cell proliferation are reported in
Section 12.3.6].
3612
3613
3614
3615
3616
3617
3618
Manti et al. (2008) exposed isolated leukocytes from four donors to a 1950 MHz UMTS signal
(SAR=0.5 and 2 W/kg) for 24 hours in blind conditions. Cells were then stimulated to divide and the
fluorescence in situ hybridization (FISH) technique was used with molecular probes specific for whole
chromosomes 1 and 2, which account for about 16% of the total human genome. There was no significant effect
at 0.5 W/kg while a small but statistically significant increase in exchange aberrations/cell was observed at 2
W/kg (p<0.05). Treatments with x rays as positive controls worked properly. In this study the effect of combined
exposures was also investigated.
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
Vijayalaxmi et al. did not find significant effects in CA and MN when human peripheral blood
samples were exposed to different RF EMF conditions in several separate investigations. (i) Cells from two
donors were exposed to 2450 MHz, CW, at an SAR of 12.5 W/kg, for 90 min delivered continuously or
intermittently (30 min on/30min off cycles) (Vijayalaxmi et al., 1997). (ii) Cells from 4 donors were exposed for
24 h to test the effect of two different modulations, FDMA at 835.6 MHz (Vijayalaxmi et al., 2001b) and CDMA
at 847.74 MHz (Vijayalaxmi et al., 2001a). The SAR values in these studies were 4.4 or 5 W/kg (835 MHz) and
4.9 or 5.5 W/kg (847 MHz), (iii) PHA stimulated (for 24 h) and unstimulated cells from 3 donors were exposed
for 2 hours to 2450 MHz (pulse width 10 µs, pulse repetition rate 10 kHz, duty factor 0.1) and 8.2 GHz (pulse
width 8 ns, pulse repetition rate 50 kHz, and duty factor 0.0004) at SARs of 2.13 and 20.71 W/kg, respectively
(Vijayalaxmi, 2006). In all the above studies, blind procedure was followed and cultures treated with gamma
rays served as positive control and gave the expected findings. Moreover, cell proliferation was also
investigated, as reported in Section 12.3.6].
3631
3632
3633
3634
3635
3636
3637
3638
In two separate studies, Hansteen and co-workers (Hansteen et al., 2009a; 2009b) investigated the
incidence of CA in stimulated human peripheral blood lymphocytes from six donors exposed for 53 h to RF
EMF. In the first study the RF exposure was at 2.3 GHz, 10 W/m2 power density, CW or pulsed wave (200 Hz
pulse frequency, 50% duty cycle) (Hansteen et al., 2009a). In the second study, higher frequencies were tested,
such as, 16.5 GHz, 10 W/m2 power density, pulsed wave (1 kHz pulse frequency, 50% duty cycle) and 18 GHz,
1 W/m2 power density, CW (Hansteen et al., 2009b). In both studies, CA frequencies were also recorded in cells
where DNA synthesis and repair were inhibited with hydroxyurea and caffeine, respectively (positive controls).
Moreover, blind procedure was followed. In all experiments RF exposure did not induce statistically significant
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
138
3639
3640
3641
3642
3643
increase in CA. In this study combined exposures were also carried out. [Although the authors do not mention it
clearly, the controls were sham-exposed. The authors did not include dosimetric details but provided the
necessary information so that other groups can use it for numerical analysis to repeat the study: the justification
was that from rough calculations, the assessed exposure level was close to the ICNIRP recommended safety
limits.]
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
Korenstein-Ilan et al. (2008) stimulated lymphocytes isolated from whole blood samples of nine
donors with PHA for 1–6 hours, followed by exposure to 100 GHz RF EMF, CW (0.03 mW/cm2) for 1, 2 and 24
hours in an incubator in which CO2 levels were not controlled. The cells were then cultured in a regular
incubator with CO2. The cells collected after a total culture period of 69–72 hours were examined for aneuploidy
of chromosomes 1, 10, 11 and 17. The results: (a) chromosomes 11 and 17 were most vulnerable (about 30%
increase in aneuploidy after 2 and 24 hours exposure; p<0.05) while chromosomes 1 and 10 were not affected,
and (b) asynchronous mode of replication of centromeres 11, 17 and 1 was increased by 40% after 2 hour
exposure and of all four centromeres by 50% after 24 hour of exposure (p<0.05). The authors mentioned that
fiber-optic sensors were used to measure the temperature difference between RF-exposed and control cells,
which never exceeded 0.3°C. In another study carried out by the same research group (Mazor et al., 2008), PHAstimulated cells from ten donors were exposed to continuous wave 800 MHz RF EMF (SAR=2.9 or 4.1 W/kg),
for 72 hours in an incubator maintained at 36–37°C. The observations included increased aneuploid cells
carrying chromosomes 11 and 17 at lower SAR and chromosomes 1 and 10 at higher SAR (p<0.05). Multisomy
(chromosomal gain) appears to be the primary contributor to the increase in aneuploidy. Control cells from a
separate group of four donors did not show increased aneuploid cells when cultured over the temperature range
of 33.5–40°C. [Thus, the data in these 2 investigations showed differential effect of RF exposure on different
chromosomes. In both studies, positive controls have not been included].
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
In a previous study, the same research group exposed PHA-stimulated lymphocyte cultures from five
donors in blind to 830 MHz RF EMF, CW (mean SAR 1.6–8.8 W/kg) for 72 hours in an incubator maintaining a
temperature of 33.5–37.5 °C (Mashevich et al., 2003). Cells from a separate group of five donors were cultured,
without RF, in incubators whose temperatures ranged from 34.5–41.0 °C: these cells served as temperature
controls. The incidence of aneuploidy of chromosome 17 was determined using an alpha-satellite DNA repeat
sequences present in its centromeric region. The data indicated a linear and SAR-dependent increase in
aneuploidy in RF exposed cells (6–9% at SARs of 2.0–8.2 W/kg) as compared with control cells (4–5%). In
temperature controls, there was an increase in aneuploidy in cells kept at 40 °C (7.8%) while no such increase
was observed in cells maintained from 34.5–38.5 °C (4.4%). [Sham samples were kept in the same incubator as
the exposed samples, but they seem not to have been in contact with the metal plates of an identical parallel-plate
resonator (i.e. a thermal mass) to that used with the exposed sample. The precision of sham-exposure is difficult
to assess since it is dependent on the internal design of the incubator. In addition, Chou and Swicord commented
that the experimental set up and the conditions used in this study would create a non-uniform heating within the
samples owing to inefficient heat exchange and that localized elevation in temperature within the sample may
well have exceeded 40 °C (Chou & Swicord, 2003). As a result, the observed increase in anueploidy of
chromosome 17 might be due to a thermal effect. However, the response from the authors was that the observed
effect was non-thermal since cells cultured at 40-41 °C showed a robust increase in aneuploidy (about 80%)
respect to controls grown at 37°C.]
3679
3680
3681
3682
3683
3684
3685
Schrader et al. (2011) conducted a study using the well-established human-hamster hybrid (FC2) cells
containing a single copy of human chromosome 11. They exposed cell cultures for 30 min to RF EMF at
frequency of 900 MHz (SAR=10.7-17.2 mW/kg, E-field strengths 45 or 90 V/m). The exposure equipment had
separate electric- (E) and magnetic-field (H) components. Two experiments were carried out at room
temperature (20-22 °C) and the third at 37°C. A blind analysis was performed. The results of pooled data were
that the E component was able to induce significantly increased spindle disturbances (p<0.05) while the H
component did not. [No positive controls were included in the study design].
3686
Studies not included in the analysis
3687
3688
3689
3690
3691
Baohong et al. (2005) exposed isolated leukocytes from a healthy donor to RF EMF (1800 MHz,
GSM signal) at an SAR of 3 W/kg. Two hours of exposure did not induce SB, as assessed by the comet assay
carried out immediately after exposure and also 21 h later. In this study combined exposures were also carried
out. [The interpretation of the results reported in this study is difficult due to the inclusion of data from one
donor only.]
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
139
3692
3693
3694
3695
Tiwari et al. (2008) exposed whole blood from six healthy donors for 1 h to an 835 MHz CDMA
signal at an SAR of 1.17 W/kg. No effects on SB were detected with the comet assay. [In this study neither
dosimetric information nor details on the exposure system are provided. Moreover, it is not clear if the sham
controls are actually sham.]
3696
3697
3698
3699
3700
3701
3702
Figueiredo et al. (2004) exposed whole blood samples from four healthy donors to 2450 MHz RF
EMF (power output 3 W) or to 10.5 GHz (power output 15 mW) for 40 sec and 5 min, respectively. After RF
exposure the cells were cultured for 48 h and the incidence of CA was recorded. No effect of exposure was
observed at any of the two frequencies investigated. In this study combined exposures were also carried out. [In
this study sham exposure is reported, although it is defined “negative control”. However, exposures to 2450
MHz were performed by means of a microwave oven, while no information is reported about the exposure
system at 10.5 GHz.]
3703
3704
3705
3706
3707
Scarfì et al. (2003) exposed whole-blood samples from nine healthy donors for 20 min to 120 and 130
GHz at 1 and 0.6 mW average power, respectively. After exposure, lymphocytes were cultured for 72 h to
determine the incidence of MN. No statistically significant differences were detected between exposed and
sham-exposed cultures. In this study cell proliferation was also investigated. [In this study no sufficient
dosimetric information is reported.]
3708
3709
3710
3711
3712
3713
Esmekaya et al. (2011) reported a statistically significant, time-dependent increase of SCE in cultured
lymphocytes isolated from blood of healthy donors and exposed to an 1800 MHz GSM signal for 6, 8, 24 and 48
h at an SAR of 0.21 W/kg. Cell viability and the effect of combined exposures were also evaluated. [The
relevance of this study is questionable since no proper dosimetric evaluation was performed. The SAR was
estimated by using electric field measured along the horn antenna. Moreover, the number of donors included in
the study is not clear.]
3714
3715
3716
3717
3718
Hintzsche et al. (2011) observed statistically significant increased spindle disturbances in anaphase
and telophase of the cell division in human-hamster hybrid cells exposed for 30 min to CW 106 GHz RF EMF at
power density of 0.43 and 4.3 mW/cm2 (p<0.05), but not at 0.043 mW/cm2. However, there was no powerintensity-dependent effect. [In this study the results have been obtained from just two replicate cultures and
dosimetry is not reported.]
3719
3720
3721
3722
3723
3724
3725
Tice et al. (2002) exposed whole-blood samples from two healthy donors to RF EMF at 837 MHz or
1907.8 MHz at SARs ranging from 1 to 10 W/kg, for 3 or 24 hours. The results indicated no significant effect on
SB. However, a significant increase in MN indices was observed in cells exposed for 24 h at 5 W/kg (p<0.05)
and 10 W/kg (p<0.01). The authors mentioned localized ‘hot-spots’ which indicated temperature variations
during RF exposure. [In this study sham exposure was mentioned as “negative control”. The results reported are
uninterpretable since blood samples from one donor for each frequency was tested, therefore individual
variability is not taken into account.]
Table 12.3.1. In vitro studies assessing genotoxic effects of RF EMF exposure on human peripheral blood cells
Cell type
Biological endpoint
Exposure conditions
Results
SB
2450 MHz, pulsed (10
µs pulses at 10 kHz)
No effect directly
or at 4 h after
exposure.
Comment
Authors
Number of
independent
experiments
Blood
lymphocytes
n=3
SAR 2.135 W/kg
(Vijayalaxmi et
al., 2000)
2h
Blood
lymphocytes
SB
n=36
8.8 GHz, pulsed (180 ns No effect directly
pulses at 50 Hz)
or 30 min after
exposure.
Average SAR 1600
(Chemeris et al.,
2006)
W/kg
40 min
Blood
lymphocytes
n=6
SB
1950 MHz, UMTS
No effect.
SAR 0.5 and 2 W/kg
Cell viability also
investigated.
(Sannino et al.,
2006)
24 h
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
140
Blood
lymphocytes
SB
n=7
SB
1800 MHz, GSM
n=14a
(Baohong et al.,
2007)
No effect.
Combined exposures
also investigated.
(Zhijian et al.,
2009)
Increase in
chromatin
condensation
after 30 min and
1 h exposure
from analysis of
pooled data.
Effect depending on
donors and
frequency.
(Sarimov et al.,
2004)
Average SAR 2 W/kg
24 h (5 min on/10 min
off cycles)
chromatin
condensation (AVTD
assay)
895-915 MHz, GSM
Average SAR 5.4
mW/kg
30-60 min
Blood
lymphocytes
Combined exposures
also investigated.
1.5 or 4 h
n=4
Blood
lymphocytes
No effect.
Average SAR 3 W/kg
n=3
Blood
lymphocytes
1800 MHz, GSM
chromatin
condensation (AVTD
assay) and 53BP1 foci
915 MHz, GSM
Increase in
Average SAR 37 mW/kg chromatin
condensation
2h
immediately and
2 h after
exposure.
Similar results in
transformed
lymphocytes.
No information on
blinding of staff
No difference
between healthy and
EHS donors.
(Belyaev et al.,
2005)
Increase in
No difference
between healthy and
Average SAR 37 mW/kg chromatin
condensation and EHS donors.
1h
decrease of
53BP1 and γH2AX foci at 915
MHz.
(Markova et al.,
2005)
Decrease in
53BP1 foci.
Blood
lymphocytes
n=10b
chromatin
condensation (AVTD
assay)
53BP1 and γ-H2AX
foci
905 or 915 MHz, GSM
No effect at 905
MHz.
Blood
lymphocytes
n=10b
chromatin
condensation (AVTD
assay)
53BP1 and γ-H2AX
foci
905, 915 MHz, GSM
Increase in
No difference
between healthy and
Average SAR 37 mW/kg chromatin
condensation and EHS donors.
1947 MHz, UMTS
decrease of
Average SAR 40 mW/kg 53BP1 and γH2AX foci at 915
24, 72 h
and 1947 MHz.
(Belyaev et al.,
2009)
No effect at 905
MHz.
Blood
lymphocytes
MN, SB
120 GHz
No effect.
For cell proliferation
see Section 12.3.6.
(Zeni et al.,
2007b)
No effect.
For cell proliferation
see Section 12.3.6..
(Zeni et al.,
2005)
No effect.
For cell proliferation
see Section 12.3.6..
(Zeni et al.,
2008)
No effect.
Human fibroblasts
also investigated.
(Schwarz et al.,
2008)
SAR 0.4 W/kg
n=17
130 GHz
SAR 0.24, 1.4, 2 W/kg
20 min
Blood
lymphocytes
SB, CA, SCE
900 MHz, GSM
Average SAR 0.3 and 1
W/kg
n=9
2h
Blood
lymphocytes
MN, SB
1950 MHz, UMTS
SAR 2.2 W/kg
n=6
SB: 24 h
MN: 24-68 h
(6 min on/2 h off cycles)
Blood
lymphocytes
n=3
MN, SB
1950 MHz, UMTS
SAR 0.1 W/kg
16 h (5 min on/10 min
off cycles)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
141
Blood
lymphocytes
MN, SB
No effect.
(McNamee et
al., 2002b)
No effect.
(McNamee et
al., 2002a)
No effect.
The results do not
(McNamee et
confirm the ones
al., 2003)
reported by Tice et al.
(Tice et al., 2002).
No effect.
For cell proliferation
see Section 12.3.6.
(Scarfi et al.,
2006)
No effect.
Combined exposures
also investigated.
(Sannino et al.,
2009b)
SAR 0.1-10 W/kg
n=5
Blood
lymphocytes
1900 MHz, CW
2h
MN, SB
n=5
1900 MHz, pulsed (1/3
duty cycle, 50 Hz, 3.54
W peak forward power,
1.18 W average forward
power)
SAR 0.1-10 W/kg
2h
Blood
lymphocytes
MN, SB
n=5
1900 MHz, CW, pulsed
(1/3 duty cycle, 50 Hz,
3.54 W peak forward
power, 1.18 W average
forward power)
SAR 0.1-10 W/kg
24 h
Blood
lymphocytes
MN
SAR 1, 5, 10 W/kg
n=10
Blood
lymphocytes
24 h
MN
For cell proliferation
see Section 12.3.6.
20 h
MN
900 MHz, GSM
No effect.
Average SAR 1.25 W/kg
n=9
Blood
lymphocytes
900 MHz, GSM
Average SAR 1.25 W/kg
n=5
Blood
lymphocytes
900 MHz, GSM
1950 MHz, UMTS
No effect.
SAR 0.15, 0.3, 0.6, 1.25
W/kg
n=9
SB, CA, SCE, MN
CA
SAR 0.5, 2 W/kg
CA, MN
CA, MN
n=3
(Stronati et al.,
2006)
Combined exposures
also investigated.
(Manti et al.,
2008)
24 h
Increase in
exchange
aberrations at 2
W/kg.
2450 MHz, CW
No effect.
For cell proliferation
see Section 12.3.6..
(Vijayalaxmi et
al., 1997)
No effect.
For cell proliferation
see Section 12.3.6.
(Vijayalaxmi et
al., 2001b)
No effect.
For cell proliferation
see Section 12.3.6.
(Vijayalaxmi et
al., 2001a)
No effect.
For cell proliferation
see Section 12.3.6.
(Vijayalaxmi,
2006)
835.62 MHz, FDMA
SAR 4.4, 5 W/kg
24 h
CA, MN
847.74 MHz, CDMA
SAR 4.4, 5 W/kg
n=4
Blood
lymphocytes
Combined exposures
also investigated.
90 min continuous or
intermittent (30 min
on/30 min off cycles)
n=4
Blood
lymphocytes
No effect at 0.5
W/kg.
SAR 12.5 W/kg
n=2
Blood
lymphocytes
1950 MHz, UMTS
(Zeni et al.,
2012a)
For cell proliferation
see Section 12.3.6.
24 h
n=4
Blood
lymphocytes
No effect.
Average SAR 1,2 W/kg
n=14
Blood
lymphocytes
935 MHz, GSM
Combined exposures
also investigated.
For cell proliferation
see Section 12.3.6.
20 h
Blood
lymphocytes
(Sannino et al.,
2011)
For cell proliferation
see Section 12.3.6.
20 h
MN
Combined exposures
also investigated.
24 h
CA, MN
2450 MHz, pulsed
SAR 2.13 W/kg
8.2 GHz, pulsed
SAR 20.8 W/kg
2h
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
142
Blood
lymphocytes
CA
n=6
2.3 GHz, CW or pulsed
(200 Hz pulse
frequency, 50% duty
cycle)
No effect in
proliferating and
DNA-synthesis
inhibited cells
No information
provided to perform
numerical dosimetry
(Hansteen et al.,
2009a)
16.5 GHz, pulsed (1 kHz No effect in
pulse frequency, 50%
proliferating and
duty cycle)
DNA-synthesis
power density 10 W/m2 inhibited cells
No information
provided to perform
numerical dosimetry
(Hansteen et al.,
2009b)
Different effect on
different
chromosomes.
(Korenstein-Ilan
et al., 2008)
Increased aneuploidy
attributed to
multisomy.
(Mazor et al.,
2008)
RF exposed samples
at 37.5 °C.
(Mashevich et
al., 2003)
power density 10 W/m2
53 h
Blood
lymphocytes
CA
n=6
53 h
Blood
lymphocytes
aneuploidy
100 GHz, CW
power density 0.031
mW/cm2
n=9
1, 2 and 24 h
Increase for
chromosome 11
and 17
No effects for
chromosomes 1
and 10.
Asynchronous
replication of all
chromosomes
investigate.
Blood
lymphocytes
aneuploidy
800 MHz, CW
SAR 2.9-4.1 W/kg
n=10
72 h
Increase for
chromosome 11
and 17 at 2.9
W/kg
Increase for
chromosome 1
and 10 at 4.1
W/kg
Increased
multisomy
Blood
lymphocytes
Aneuploidy of
chromosome 17
n=5
830 MHz, CW
average SAR 1.6-8.8
W/kg
72 h
Humanhamster
hybrid cells
Spindle disturbances
n=3
a
7 healthy and 7 EHS individuals
b
5 healthy and 5 EHS individuals
900 MHz, CW, GSM
SAR 0.01-0.017 W/kg
E: 45 or 90 V/m
30 min
Linear and SARdependent
increase in
aneuploidy
Precision of shamexposure difficult to
assess
Increased spindle Separate E and H
disturbances due components.
to the E
component
(Schrader et al.,
2011)
No effect of H
component.
“No effect” means no statistically significant effect.
3726
3727
Other primary and cultured human cells
3728
3729
In several studies other primary and cultured human cells have been examined for genotoxicity
induced by RF EMF exposure.
3730
3731
Two studies have been carried out on amniotic cells collected during amniocentesis and two more
studies on trophoblast cells.
3732
3733
3734
3735
3736
3737
3738
Bourthoumieu et al. (2010) cultured amniotic cells from four separate donors for 15 days and then
cells were exposed to a 900 MHz GSM signal (SAR=0.25 W/kg) for 24 hours. Blind examination of R-banded
chromosomes, immediately after and 24 h after RF exposure, did not reveal significant changes in the incidence
of chromosomal breaks, rearrangements, structural and numerical aberrations. Cultures treated with bleomycin
(BLM) as positive controls gave positive findings. In a subsequent study, the authors applied the same exposure
protocol on amniotic cells from three donors to test four SAR values, 0.25, 1, 2, 4 W/kg, for 24 hours
(Bourthoumieu et al., 2011). The temperature during RF exposure ranged from 36.3 ± 0.4 to 39.7 ± 0.8°C. The
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
143
3739
3740
3741
cells in interphase were hybridized with centromere-specific probes for chromosomes 11 and 17 (FISH
technique). No significances changes were observed in monosomic and trisomic cells and in the total number of
aneuploid cells. [No positive controls were included in the study].
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
Valbonesi et al. (2008) observed no changes in SB in a human trophoblast cell line (HTR-8/neo)
following a 1 h exposure to a 1817 MHz GSM signal (SAR=2 W/kg) in blind conditions, as assessed by alkaline
comet assay performed in six independent experiments [In this study, gene and protein expression has also been
investigated, as reported in Section 12.3.3]. In a follow up study, the same research group investigated the effect
of longer exposure duration and different signals in the same cell model (Franzellitti et al., 2010). In particular,
in six independent experiments HTR-8/neo cells were intermittently exposed (5 min on/10 min off cycles) for 4,
16 and 24 h at 1800 MHz RF EMF, both CW and GSM modulated (basic and talk mode) in blind conditions. No
variation in SB was detected after CW exposure. GSM modulated signals induced an increase in SB following
16 and 24 h RF exposure (p<0.05). The effect was transient, since it disappeared 0.5–2 h after RF exposure.
However, the reported increase in SB was dependent on the comet parameters analyzed. [In these papers,
hydrogen peroxide was used as positive control and induced increased SB, as expected].
3753
3754
Glioblastoma and neuroblastoma cells were investigated in six studies and no significant effect of RF
exposure was detected on primary DNA damage (SB) in any of the studies.
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
In two separate studies, Malyapa and co-workers exposed exponentially growing U87MG
glioblastoma cells to RF EMF for 2, 4 and 24 h to evaluate the induction of SB by applying the comet assay. In a
first study (Malyapa et al., 1997a) cell cultures were exposed to 2450 MHz, CW, at an SAR of 0.7 W/kg (three
independent experiments). In the second investigation (Malyapa et al., 1997b) two types of frequency
modulations were studied: frequency-modulated continuous-wave (FMCW), with a carrier frequency of 835.62
MHz, and code-division multiple-access (CDMA) centered on 847.74 MHz, both at mean SARs of 0.6 W/kg
(three independent experiments). No significant effect was detected when RF-exposed cells were compared with
their respective sham-controls. The studies were carried out in blind conditions and exposure to gamma rays
were used as positive controls and induced SB. Both exposure conditions were also applied to evaluate the effect
of RF EMF exposure on murine cell lines (see next section for details).
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
Miyakoshi et al. (2002) did not find an effect on SB in MO54 glial cells derived from a patient with
brain tumor when exposed to 2450 MHz RF EMF, CW, at SARs of 50 and 100 W/kg for 2 hours (three
independent experiments). Cultures exposed to x rays served as positive controls. The same research group
evaluated the effects of 2 and 24 h exposure to RF from mobile radio base stations employing the International
Mobile Telecommunication 2000 (IMT-2000) cellular system at 2.1425 GHz in two cell lines, A172
glioblastoma and normal IMR-90 fibroblasts from fetal lungs (Sakuma et al., 2006). A172 cells were exposed to
a Wideband Code Division Multiple Access (W-CDMA) modulated signal at SARs of 80, 250, and 800 mW/kg
and CW radiation at 80 mW/kg, while IMR-90 cells were exposed to both W-CDMA and CW at a SAR of 80
mW/kg (three independent experiments for each condition tested). No significant differences in SB were
observed between the exposed and the sham exposed controls, as evaluated in blind immediately after the
exposure by alkaline comet assay. In positive controls (cultures treated with MMS) a significant increase in SB
was recorded. [In this paper numerical dosimetry is not reported, while experimental dosimetry has been carried
out.]
3778
3779
3780
3781
3782
3783
3784
3785
Luukkonen et al. (2009) exposed SH-SY5Y neuroblastoma cells for 1 h to 872 MHz RF EMF, CW
and GSM signal, (SAR=5 W/kg). The results from three independent experiments performed in blind indicated
no significant difference in SB compared to sham-exposed cultures. Treatments with menadione served as
positive controls. In a further study (Luukkonen, Juutilainen & Naarala, 2010) the authors applied the same
protocol but the exposure duration was 3 h and positive controls were obtained by treating cell cultures with
MMS. Also in this case no significant changes in SB were detected (4 experiments in blind). [In both studies
oxidative stress was also investigated, as reported in Section 12.3.5. Moreover, combined exposures were carried
out.]
3786
3787
3788
3789
3790
3791
Lens epithelial cell cultures were exposed to intermittent (5 min on/10 min off cycles) GSM 1800
MHz RF EMF (SAR=1, 2, 3, 4 W/kg) for 2 hours by Yao et al. (2008a). In three independent experiments, a
significant increase in single SB was observed at SARs of 3 and 4 W/kg (p<0.001), while no differences were
detected on double SB, as assessed by γ-H2AX foci formation. Induction of DSB was detected in positive
controls, treated with 4-nitroquinoline-1-oxide (4NQO). In this study (also reported in Section 6.4.2 Ocular
functions, and 12.3.5 Oxidative stress) and the effect of combined exposures were also investigated.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
144
3792
3793
3794
3795
3796
3797
3798
Lymphoblastoid cells were employed in four studies. Phillips et al. (1998) exposed Molt-4
lymphoblastoid cells to RF EMF at a frequency of 813.56 MHz (iDEN signal, SAR=2.4 and 24 W/kg) or 836.55
MHz (TDMA signal, SAR=2.6 and 26 W/kg ) for 2, 3 or 21 hours. At lower SARs, there was a significant
decrease in SB at 2 and 21 hours (p<0.01) but not at 3 hours. At higher SARs, depending on the type of RF
signal, power intensity and duration of exposure, significant increases as well as significant decreases in SB were
observed (p<0.001). The number of experiments for each condition varied from one to eight. [No positive
controls were included in the study].
3799
3800
3801
3802
3803
3804
3805
3806
In a replication study, Hook et al. (2004b) used the same Molt-4 cells, and the same as well as other
RF signals under the same and higher SAR. In particular, they exposed cells for 2, 3 or 21 h to RF EMF as
follows: a) 847.74 MHz, CDMA, SAR=3.2 W/kg; b) 835.62 MHz, FDMA, SAR=3.2 W/kg; c) 813.56 MHz,
iDEN, SAR=2.4 and 24 W/kg; d) 836.55 MHz, TDMA, SAR=2.6 and 26 W/kg. The data, from three
independent experiments for each condition, showed no significant differences on SB between sham-treated cells
and cells exposed to RF radiation for any frequency, modulation or exposure time. Cultures exposed to gamma
rays as positive control gave positive findings. Therefore, the results from the original study were not confirmed.
[In this study the induction of apoptosis was also investigated, as reported in Section 12.3.4].
3807
3808
3809
3810
3811
3812
Two more recent studies failed to find effects of RF EMF on primary DNA damage. Huang et al.
(2008a) did not find an effect on SB in T-lymphoma-derived (Jurkat) cells exposed for 1, 4 and 24 h to 1763
MHz CDMA-modulated RF EMF (SAR=10 W/kg), as assessed in three independent experiments performed in
blind. Positive controls (cultures exposed to gamma rays) gave positive findings. [This study has also been
reported in Sections 10.3 (immune system and haematology), 12.3.2 (signal transduction), 12.3.3 (gene and
protein expression) and 12.3.6 (cell proliferation).]
3813
3814
3815
3816
3817
Zhijian et al. (2010) also failed to find any effect on SB when human HMy2-CR lymphoblastoid Bcells were intermittently (5 min on/10 min off cycles) exposed in blind to RF EMF at 1800 MHz, GSM
modulated (SAR=2 W/kg) for 6, 12, 18 and 24 hours (four independent experiments). Positive controls were
treated with BLM and gave positive findings. In this study the effect of combined exposure was also
investigated.
3818
3819
Skin fibroblasts cultures, established from a 6 year old boy (ES-1 cells), were employed by several
authors to evaluate the effect of RF exposure in terms of SB and/or MN.
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
Schwartz et al. (2008) exposed ES-1 cells to a continuous and intermittent 1950 MHz UMTS signal,
for 24 h at SAR values of 0.05, 0.1, 0.5, 1 and 2 W/kg. Moreover, cells were also exposed at 0.1 W/kg for 4, 8,
12, 16, 20, 24 and 48 hours. The tail factor (TF) analysis and the evaluation of MN frequency indicated
significant increases in SB and centromere-negative MN (p=0.02) for all the SARs investigated when
intermittent exposures were performed. Moreover, the effect resulted dependent from the exposure duration
(after 8 and 12 h exposure, for SB and MN, respectively; p=0.02. No effects were detected for short exposure
duration). In a further set of experiments the authors confirmed these findings on fibroblast cultures from three
different donors, but not on short-term lymphocyte cultures from three donors (see previous section on human
lymphocytes). Positive controls were treated with UV and vincristine for SB and MN, respectively, and gave
positive findings. [The TF, a surrogate marker for SB, was made-up by arbitrary transformation of the comets
(collected manually and subjectively) into A, B, C, D and E categories giving their weights as Ax2.5, Bx12.5,
Cx30.0, Dx67.5 and Ex97.5. In a letter to the editor, other researchers (Lerchl, 2009; Lerchl & Wilhelm, 2010;
Tuffs, 2008; Vijayalaxmi, McNamee & Scarfi, 2006) have expressed concern about the data reported in this and
the other publication reported by Diem et al. (2005) (see section “Studies not included in the analysis”) regarding
the use of tail factor method, low standard deviations, low inter-individual differences and inappropriate
statistical analysis. The response from the authors was that they believe in the results obtained since the
experiments were carried out in blind conditions and microscope slides were scored by the same operator
(Rüdiger, Kratochvil & Pilger, 2006; Rüdiger, 2009a; b). Cell proliferation was also investigated, as reported in
Section 12.3.6].
3839
3840
3841
3842
3843
3844
3845
Speit et al. (2007) performed an independent replication of the Diem et al. (2005) study, also using
ES-1 skin fibroblasts, continuous and intermittent (5 min on/10 min off cycles) 1800 MHz RF EMF (CW;
SAR=2 W/kg) exposure in the same exposure equipment supplied by the same company and the same laboratory
protocols. Some of the authors of the original investigation participated in the study. After 1, 4 and 24 h
exposure the TF method as well as computerized image analysis was utilized to examine the SB. An additional
MN end-point was also included in the experiments. The data from three independent experiments, performed in
blind, indicated no significant effect on both SB and MN. Cultures exposed to gamma rays were included as
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
145
3846
3847
positive controls and gave positive findings. Thus the original results were not confirmed. In this investigation
the effect of RF-EMF exposure was also investigated in animal cells.
3848
3849
3850
3851
3852
3853
3854
Skin fibroblasts and stem cells were investigated in three more papers. Sannino et al. (2009a) used
skin cells established from healthy individuals as well as from subjects affected by Turner’s syndrome (a genetic
disorder associated with a chromosomal imbalance). Cell cultures were exposed in blind to a 900 MHz GSM
signal (SAR=1 W/kg) for 24 hours. There was no significant effect on SB and MN in both cell types with respect
to sham-exposed controls (three independent experiments for each cell type). On the contrary, in cultures treated
with MMC as positive control an increase in both SB and MN was detected. In this study cell proliferation was
also investigated, as reported in Section 12.3.6. Moreover, combined exposures were carried out.
3855
3856
3857
3858
3859
3860
3861
3862
3863
Hintzsche et al. (2012a) exposed primary dermal fibroblasts (HDF) and a keratinocyte cell line
(HaCaT) to 106 GHz EMF with power densities between 0 and 0.88 mW/cm2 for 2, 8 and 24 h to evaluate SB
(comet assay) and chromosomal damage (MN assay). The RF EMF exposures and sham-exposures were carried
out in a modified incubator at defined environmental conditions. In a separate set of experiments, cells were also
exposed to higher power intensities of 2 mW/cm2. The results of three independent experiments, performed in
blind, showed neither SB nor MN increase in RF-exposed cultures compared with sham-exposed ones. Positive
controls, treated with MMS, gave positive findings. In this study cell proliferation was also investigated, as
reported in Section 12.3.6. [From the sham-exposure description, it can be argued that sham-exposure was
subsequent to RF exposure.]
3864
3865
3866
3867
3868
3869
3870
3871
3872
Markova et al. (2010) confirmed and extended their previous results on human lymphocytes (Belyaev
et al., 2009) to VH-10 fibroblasts and mesenchymal stem cells (MSCc). They exposed cells to GSM 905 and 915
MHz signals (SAR=37 mW/kg) or a 1947.4 MHz UMTS signal (SAR=39 mW/kg). The exposures were either
acute for 1, 2 and 3 hours or chronic for 1 hour per day, 5 days per week for 10 days. There was a significant
decrease in 53BP1 foci in 915 and 1947 MHz exposed cells (p<0.01 and p<0.0001 for VH-10 and MSC cells,
respectively) after 1 h exposure, but not in cells exposed to 905 MHz (acute exposures). Chronic exposures also
decreased foci formation in MSCc (p<0.05) but not in VH-10 fibroblasts. For each cell line three to five
independent experiments were performed in blind. Positive controls have been included in the study (cultures
exposed to gamma rays) and worked properly.
3873
Studies not included in the analysis
3874
3875
3876
3877
3878
Hintzsche et al. (2012b) did not observe significant effects on MN in keratinocytes (HaCaT cells)
exposed to RF EMF at 900 MHz, CW, at 5, 10, 30 or 90 V/m for 30 min and 22 h (three independent
experiments). The same experimental conditions failed to induce effects also in human-hamster hybrid cells. In
this study cell proliferation was also investigated. [No dosimetry has been carried out. The results are reported
as a function of the E field strength.]
3879
3880
3881
3882
3883
3884
3885
3886
3887
Lixia et al. (2006) exposed immortalized lens epithelial cells (hLEC) to 1800 MHz GSM-modulated
RF EMF SARs of 1–3 W/kg. DNA SB and their repair were measured immediately after 2 h exposure and at
incubation times of 30, 60, 120 and 240 min post-exposure, respectively. By comparing exposed and shamexposed cells the comet assay revealed no differences in SB at 1 and 2 W/kg, while a significant increase was
observed at 3 W/kg SAR immediately after RF exposure and after 30 min incubation (p<0.05). There was no RF
exposure effect in DNA repair rate assessed at 30, 60, 120 and 240 minutes after RF exposure. The increase in
SB observed at an SAR of 3 W/kg was reduced at 30 minutes and returned to control levels in 2 hours. In this
study the expression of HSPs and cell proliferation have been also investigated. [The number of independent
experiments carried out is not reported, although statistical analysis was performed].
3888
3889
3890
3891
3892
3893
3894
3895
Diem et al. (2005) exposed ES-1 cells to continuous and intermittent (5 min on/10 min off cycles) RF
EMF at 1800 MHz. The field was either CW (SAR=2 W/kg) or with two different types of modulation, GSM
basic (2 W/kg) and GSM talk (1.2 W/kg), for 4, 16 and 24 h. The results from the comet assay were analyzed in
terms of “tail factor” (TF). The conclusions were: (a) a significant increase in TF after 16 hour exposure
(p<0.01) with no further increase after 24 hours exposure, (b) intermittent exposure produced a stronger effect
than continuous exposure, and (c) the induced DNA damage cannot be based on thermal effects. The same
protocol was also applied to expose rat granulosa cells (as reported in the section on animal cells). [In this study
the number of independent experiments carried out is not reported].
Table 12.3.2. In vitro studies assessing genotoxic effects of RF EMF exposure on other primary and cultured human
cells
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
146
Cell type
Biological endpoint
Exposure conditions
Results
Comment
Authors
CA
900 MHz, GSM
No effect on
breaks,
rearrangements,
structural/numerical
aberrations, soon
after and 24 h post
exposure.
(Bourthoumieu
et al., 2010)
No effect for any of No information on
the tested SAR.
blinding of staff
(Bourthoumieu
et al., 2011)
No effect.
For gene expression
and cell proliferation
see Section 12.3.3
and 12.3.6
(Valbonesi et
al., 2008)
SB increase on the
base of the comet
parameter analysed.
(Franzellitti et
al., 2010)
Number of
independent
experiments
Amniotic cells
n=4
Average SAR 0.25
W/kg
24 h
Amniotic cells
n=3
Aneuploidy
900 MHz, GSM
(chromosome 11 and Average SAR 0.25, 1,
17)
2, 4 W/kg
24 h
Trophoblast cell SB
line (HTR8/neo)
n=6
1817 MHz, GSM
Average SAR 2 W/kg
1h
Trophoblast cell SB
line (HTR8/neo)
1800 MHz, CW and
GSM
No effect at 4 h RF
exposure.
SAR 2 W/kg
n=6
4, 16, 24 h (5 min
on/10 min off cycles)
Transient increase
at 16 and 24 h that
reverted after 0.52h.
2450 MHz, CW
No effect.
Murine cells also
investigated.
(Malyapa et al.,
1997a)
No effect.
Murine cells also
investigated.
(Malyapa et al.,
1997b)
No effect.
No information on
blinding of staff
(Miyakoshi et
al., 2002)
Glioblastoma
U87MG cells
SB
SAR 0.7 W/kg
n=3
Glioblastoma
U87MG cells
2, 4, 24 h
SB
835.62 MHz, FMCW
847.74 MHz, CDMA
n=3
Average SAR 0.6 W/kg
2, 4, 24 h
Brain tumor
SB
MO54 glial cells
2450 MHz, CW
n=3
2h
Glioblastoma
A172 cells
SB
SAR 50, 100 W/kg
2.142 GHz, W-CDMA,
No effect.
(Sakuma et al.,
2006)
SAR 80, 250, 800
mW/kg
n=3
CW, SAR 80 mW/kg
(A172)
normal IMR-90
fibroblasts
2, 24 h
n=3
2.142 GHz, W-CDMA,
or CW
SAR 80 mW/kg (IMR90)
2, 24 h
Neuroblastoma
SH-SY5Y cells
SB
n=3
872 MHz, CW and
GSM
No effect.
SAR 5 W/kg
n=4
SB
872 MHz, CW and
GSM
(Luukkonen et
al., 2009)
Combined exposures
also investigated.
1h
Neuroblastoma
SH-SY5Y cells
For oxidative stress
see Section 12.3.5
No effect.
SAR 5 W/kg
For oxidative stress
see Section 12.3.5
Combined exposures
also investigated.
3h
(Luukkonen,
Juutilainen &
Naarala, 2010)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
147
Lens epithelial
cells
SB, ɣ-H2AX foci
n=3
1800 MHz, GSM
Increased single
Average SAR 1, 2, 3, 4 SB at 3 and 4
W/kg.
W/kg
No effects on
2 h (5 min on/10 min
double SB (γ-H2AX
off cycles)
foci).
For ocular functions
(Yao et al.,
and oxidative stress
2008a)
see section 6.4.2. and
12.3.5
Combined exposures
also investigated
No information on
blinding of staff
Lymphoblastoid
Molt-4 cells
SB
813.56 MHz, iDEN
SAR 2.4, 24 W/kg
n=1-8
836.55 MHz, TDMA
Significant increase No information on
or decrease in SB blinding of staff
(p<0.01)
(Phillips et al.,
1998)
No effect.
(Hook et al.,
2004b)
SAR 2.6, 26 W/kg
2, 3, 21 h
Lymphoblastoid
Molt-4 cells
SB
813.56 MHz, iDEN
SAR 2.4, 24 W/kg
n=3
836.55 MHz, TDMA
Repetition study of
Phillips et al. (Phillips
et al., 1998).
For apoptosis see
Section 12.3.4
SAR 2.6, 26 W/kg
847.74 MHz, CDMA
No information on
blinding of staff
835.62 MHz, FDMA
SAR 3.2 W/kg
2, 3, 21 h
T-lymphoma
derived (Jurkat)
cells
SB
For gene expression
and cell proliferation
see section 12.3.3
and 12.3.6.
(Huang et al.,
2008a)
No effect.
Combined exposures
also investigated.
(Zhijian et al.,
2010)
Increased SB at
0.05 and 1 W/kg,
no further increase
at 1 and 2 W/kg.
Manual scoring (TF).
(Schwarz et al.,
2008)
1, 4, 24 h
SB
1800 MHz, GSM
Average SAR 2 W/kg
6, 12, 18, 24 h (5 min
on/10 min off cycles)
n=4
Skin fibroblasts
from 3 donors
n=3
No effect.
SAR 10 W/kg
n=3
HMy2-CR
lymphoblastoid
B-cells
1763 MHz, CDMA
SB, MN
1950 MHz, UMTS
SAR 0.05-2 W/kg
24 h
SAR 0.1 W/kg
Time-dependent
4-48 h (continuous and effect for SB (up to
5 min on/10 min off
8 h) and MN (up to
cycles)
12 h).
Skin fibroblasts
(ES-1)
SB, MN
No effect.
SAR 2 W/kg
n=3
Skin fibroblasts
from healthy
and Turner’s
syndrome
donors
1800 MHz, CW
900 MHz, GSM
Controversial data.
For cell proliferation
see Section 12.3.6
Replication study of
(Diem et al., 2005).
Animal cells also
investigated.
1, 4, 24 h (continuous
and 5 min on/10 min
off cycles)
SB, MN
Human lymphocytes
also investigated.
No effect.
Averager SAR 1 W/kg
For cell proliferation
see Section 12.3.6
(Speit, Schütz
& Hoffmann,
2007)
(Sannino et al.,
2009a)
Combined exposures
also investigated.
24 h
n=3
Primary dermal
fibroblast
SB, MN
106 GHz
No effect.
0-2 mW/cm2
For cell proliferation
see Section 12.3.6
(Hintzsche et
al., 2012a)
Keratinocyte
cell line
(HaCaT)
n=3
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
148
Diploid
53BP1 foci
fibroblasts (VH10)
Mesenchimal
stem cells
(HMSc)
905, 915 MHz, GSM
SAR 37 mW/kg
1947.4 MHz, UMTS
SAR 39 mW/kg
1-3 h
1 h/day for 2 weeks
n=3-5
Inhibition of foci
Follow-up study of
(Markova,
formation (DSB) at (Belyaev et al., 2009). Malmgren &
915 and 1947.4
Belyaev, 2010)
MHz in VH-10 and
MSC cells after 1 h
exposure. No effect
at 905 MHz.
Inhibition of foci
formation after
chronic exposures
in MSC but not in
VH-10 cells.
“No effect” means no statistically significant effect.
3896
3897
12.3.1.2
3898
3899
3900
The studies reported and discussed below have been carried out using the cells of rodent origin. In
several such studies mouse fibroblasts (C3H 10T1/2 cells) have been investigated (in some studies the effect of
RF EMF in human cells has been also evaluated, and the details have been reported in the previous section).
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
In two separate studies, Malyapa and co-workers exposed C3H 10T1/2 cells to RF EMF to evaluate
the induction of SB using the comet assay. The experiments were performed in blind. In a first study (Malyapa et
al., 1997a) exponentially growing cell cultures were exposed to 2450 MHz, CW, (SAR=0.7 W/kg and 1.9 W/kg;
three independent experiments) for 2, 4 and 24 h. The extent of SB was assessed immediately after exposure;
moreover, after 2-h exposure SB were also measured at 4 h post-exposure to assess DNA repair. In the second
investigation (Malyapa et al., 1997b) two types of frequency modulations were studied in both exponential-phase
and plateau-phase cell cultures. The latter condition was used to simulate in vivo conditions for nonproliferating (plateau) cells. Frequency-modulated continuous-wave (FMCW), with a carrier frequency of
835.62 MHz, and code-division multiple-access (CDMA) centered on 847.74 MHz (SAR=0.6 W/kg; three
independent experiments) RF EMF were used for 2, 4 and 24 h exposures; moreover, cells exposed for 2 h were
also tested for DNA repair at 4 h post-exposure. No significant effect was detected when RF EMF-exposed cells
were compared with their respective sham-controls in any of the experimental conditions tested. In both studies,
cells exposed to gamma rays served as positive controls and gave positive findings. [These results do not
confirm the induction of SB reported by Lai and Singh in in vivo studies](Lai & Singh, 1995; 1996). The above
described exposure conditions were also applied to evaluate the effect of RF EMF exposure on human
glioblastoma cell lines (see previous section for details)].
3917
3918
3919
3920
3921
3922
3923
The discrepancy between the results on SB reported by Lai and Sing (1995; 1996) and those of
Malyapa et al. (Malyapa et al., 1997a; b) could be due to the presence of RF-induced DNA-DNA or DNAproteins crosslinks. To clarify this, Lagroye et al. (2004) exposed C3H 10T1/2 cells to RF EMF at 2450 MHz,
CW, for 2 h at an SAR of 1.9 W/kg (the highest used in the Malyapa investigations) in blind. The experimental
protocol included exposure to gamma rays to induce SB (positive controls) and treatment with proteinase K to
digest crosslinked proteins. No effects of RF EMF exposure were detected for any of the experimental conditions
(three to four independent experiments). Positive controls worked properly.
3924
3925
3926
3927
3928
3929
3930
3931
Li et al. (2001) also performed experiments with exponentially growing and plateau phase C3H
10T1/2 cells. The exposures were carried out to 835.62 or 847.74 MHz RF (FDMA or CDMA signals
respectively, at average SAR of 3.8 W/kg) for 2 and 4 h (blind experiments). The extent of SB was evaluated
immediately after exposure and, in the case of 2 h RF exposure, at 4 h post-exposure. The effect of 24 h
exposure was investigated at SAR values of 4.9 and 5.1 W/kg for CDMA and FDMA signals, respectively. No
significant effects on SB and DNA repair were detected for any of the experimental conditions tested, as
assessed in three independent experiments. [No positive controls have been included. This study does not
confirm the results reported by Phillips et al. (1998) on lymphoblastoid Molt-4 cells (see previous section)].
3932
3933
3934
3935
Absence of effects in terms of MN was reported by Bisht et al. (2002). They employed the same
exposure conditions tested by Li et al. (2001) in terms of frequency, signals and SAR to expose exponentially
growing and in plateau phase C3H 10T1/2 cells for 3, 8, 16, and 24 hours (three independent experiments,
performed in blind). Cells exposed to gamma rays served as positive control and gave positive findings.
Animal cells
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
149
3936
3937
3938
3939
3940
Huang et al. (2008b) exposed a mouse auditory hair cell line (HEI-OC1) to 1763 MHz CDMAmodulated RF EMF (SAR=20 W/kg) for 6, 24 and 48 h. No changes in SB were observed in three independent
experiments following RF exposure. [Positive controls were not included in the study design. This paper has also
been reported in Section 6.4.1 auditory function, 12.3.2 signal transduction, 12.3.3 gene and protein expression,
12.3.6 cell proliferation].
3941
3942
3943
3944
3945
Kumar et al. (2011) excised femur and tibia bones from 11 rats and exposed them in blind for 30 min
to 900 MHz RF EMF, CW (SAR=2 W/kg). No significant changes were observed in SB in bone marrow cells
extracted from such bones compared to sham-exposed controls. Positive controls, treated with concanavalin-A
gave the expected increase in SB. [In this study, erythrocyte maturation and cell proliferation and were also
investigated, as reported in Section 10.3 and 12.3.6].
3946
3947
3948
Zeni et al. (2012b) failed to find changes in SB in rat neuronal cells (PC12) exposed in blind for 24 h
to a 1950 MHz UMTS signal (SAR=10 W/kg), in three independent experiments. Positive controls treated with
MMS worked properly. [In this study, apoptosis was also investigated, as reported in Section 12.3.4].
3949
3950
3951
3952
3953
3954
3955
The investigation carried out by Speit et al. (2007) on ES-1 human skin fibroblasts (see section on
human cells) aimed to replicate the experiments of Diem et al. (2005) also included V79 rodent cells. The cells
were intermittently (5 min on/10 min off cycles) exposed in blind to 1800 MHz RF EMF (CW; SAR=2 W/kg)
for 1, 4 and 24 h. After an additional18 h cultivation, MN and SB were evaluated. Moreover, MN frequency was
also measured immediately after 18 h RF-exposure. As for human fibroblasts, the data on V79 cells from three
independent experiments indicated no significant effect on both SB and MN. In positive controls, exposed to
gamma rays, SB and MN resulted increased, as expected.
3956
3957
3958
3959
3960
Kim et al. (2008) in two independent experiments reported no significant effect on SB and CA in
mouse L5178Y TK+/- lymphoma cells exposed to CW 835 MHz RF EMF (SAR=4 W/kg). The exposure was for
24 h (CA) or 48 h (CA and SB). Treatments with cyclophosphamide (CPA) or 4NQO served as positive controls
and resulted in a significant increase in both CA and SB. In this study, the effect of combined exposures was also
investigated.
3961
3962
3963
3964
3965
3966
3967
3968
Nikolova et al. (2005) examined SB, CA and SCE in mouse neural progenitor stem cells exposed to
intermittent (5 min on/30 min off cycles) 1710 MHz GSM fields (average SAR=1.5 W/kg), for 6 and 48 hours.
In six independent experiments a small but statistically significant increase in SB (p<0.05) was observed only at
6 h, that disappeared after a further incubation period of 18 h. No effect was detected immediately after 48 h
exposure and 24 and 48 h later. No effects were detected in terms of CA and SCE. [As also stated by the authors,
the increase in SB was transient. Positive controls have not been included in the study design. This study has also
been described in Section 12.3.2 signal transduction, 12.3.3 gene and protein expression, 12.3.4 apoptosis and
12.3.6 cell proliferation].
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
Koyama et al. reported the data from three separate investigations carried out on Chinese hamster K1
(CHO-K1) cells exposed to CW 2450 MHz RF EMF in different conditions. For each study, three independent
experiments were carried out in blind and treatments with BLM as positive controls gave positive results. In the
first study (Koyama et al., 2003) the incidence of MN was evaluated using kinetochore method in cells exposed
for 18 hours at SARs of 13, 39, 50, 78 and 100 W/kg. A statistically significant increase in MN frequency
(p<0.01) was detected in cells exposed to SAR values higher than 50 W/kg. However, cells exposed to 39°C
temperature alone for 18 hours also showed increased MN. In the second study (Koyama et al., 2004) CHO-K1
cells were exposed for 2 hours at SARs of 5, 10, 20, 50, 100 and 200 W/kg and a significant increase in MN was
observed in cells exposed at 100 and 200 W/kg SAR (p<0.01). In the third investigation (Koyama et al., 2007)
the latter experimental conditions were tested to evaluate mutation (MUT) frequency at hypoxanthine-guanine
phosphoribosyl transferase (HPRT) locus. A statistically significant increase in MUT frequency was detected
after exposure at an SAR of 200 W/kg (p<0.05). [As also stated by the authors, a thermal effect may be possible
in samples exposed at high SARs].
3982
3983
3984
3985
3986
3987
3988
Xu et al. (2010) dissected cortical neurons from newborn rats and after 8 days of culturing they were
exposed to an 1800 MHz GSM signal (SAR=2 W/kg) for 24 hours (5 min on/10 min off cycles). Four hours
before RF exposure the cells were treated with melatonin. The levels of 8-OHdG adducts, a common biomarker
for oxidative damage, were examined in mitochondrial DNA (mtDNA). RF exposure alone induced a significant
increase in adducts with a concomitant decrease on mtDNA copy numbers and mitochondrial RNA transcripts
(p<0.01), as reported in three independent experiments performed in blind. Each of these disturbances in mtDNA
was reversed in cells pre-treated with melatonin. Positive controls treated with hydrogen peroxide gave positive
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
150
3989
3990
findings. [Although the authors do not call it “sham” the control is actually a sham. This study has also been
presented in section 12.3.5 (oxidative stress).]
3991
Studies not included in the analysis
3992
3993
3994
3995
3996
3997
Campisi et al. (2010) exposed astroglial cells from newborn rat brains for 5, 10 or 20 min to 900 MHz,
CW or amplitude modulated at 50 Hz, at the same power density of 0.26 W/m2 (no SAR reported). A significant
increase in SB (p<0.001) was observed after modulated exposure for 20 min. No effects were detected when
shorter exposure duration or CW were used (five independent experiments). In this investigation the effect of RF
exposure on cell viability and reactive oxygen species were also evaluated. [The absence of dosimetry makes the
results of this study uninterpretable.]
3998
3999
4000
4001
4002
4003
4004
Komatsubara et al. (2005) examined the incidence of CA in murine m5S skin fibroblasts exposed to
CW 2450 MHz fields (SAR=5, 10, 20, 50, 100 W/kg) or pulsed waves for 2 hours. In the latter case, cells were
exposed to a peak SAR of 900 W/kg for 1 s, followed by resting intervals of 17 or 8 s resulting in average SARs
of 50 or 100 W/kg, respectively. There was no significant effect in exposed cells compared to sham exposed
ones for all the conditions tested, although after 2 h exposure temperature increased to 39 °C and 41 °C, in
samples exposed to 50 W/kg and 100 W/kg, respectively. [The results reported in this paper seem to refer to just
one experiment, although the authors claimed that statistical analysis has been carried out.]
4005
4006
4007
4008
4009
4010
4011
4012
4013
Diem et al. (2005) exposed SV40-transformed rat granulosa cells to continuous and intermittent (5
min on/10 min off cycles) RF EMF at 1800 MHz. The field was either without modulation (CW, SAR=2 W/kg)
or with two different types of modulation, GSM basic (SAR=2 W/kg) and GSM Talk (SAR=1.2 W/kg), for 4, 16
and 24 h exposure. The results from the comet assay were expressed as “tail factor”. A significant increase in TF
was detected after 16 hour exposure (p<0.01) with no further increase after 24 hours exposure. Moreover,
intermittent exposure produced a stronger effect than continuous exposure. The same protocol was also applied
to expose human skin fibroblasts (as reported in the section on human cells). [In this study the number of
independent experiments carried out was not reported. The results of this study have been criticized by several
authors, as reported in the section on human cells.]
Table 12.3.3. In vitro studies assessing genotoxic effects of RF EMF exposure on cell cultures of animal origin
Cell type
Biological endpoint
Exposure conditions
Results
Comment
Authors
SB, DNA repair
2450 MHz, CW
No effect on SB.
SAR 0.7, 1.9 W/kg
No effect on DNA
repair in cells
exposed for 2 h
and incubated for
additional 4 h.
Human cells also
investigated.
(Malyapa et al.,
1997a)
Human cells also
investigated.
(Malyapa et al.,
1997b)
Number of
independent
experiments
mouse
fibroblasts
(C3H 10T1/2
cells)
2, 4, 24 h
n=3
mouse
fibroblasts
(C3H 10T1/2
cells)
SB, DNA repair
No effect on SB.
847.74, CDMA
No effect on DNA
repair in cells
exposed for 2 h
and incubated for
additional 4 h.
Average SAR 0.6 W/kg
2, 4, 24 h
n=3
mouse
fibroblasts
(C3H 10T1/2
cells)
835.62, FMCW
SB, crosslinks
2450 MHz, CW
No effect.
Repetition study
(Lagroye et al.,
of (Malyapa et al., 2004)
1997a; b).
835.62 MHz, FDMA
No effect on SB.
847.74 MHz, CDMA
No effect on DNA
repair in cells
exposed for 2 h
and incubated for
additional 4 h.
Repetition study
of (Phillips et al.,
1998).
SAR 1.9 W/kg
2h
n=3-4
mouse
fibroblasts
(C3H 10T1/2
cells)
n=3
SB, DNA repair
Average SAR 3,8 W/kg
2 , 4h
Average SAR 4.9-5.1
W/kg
(Li et al., 2001)
24 h
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
151
mouse
fibroblasts
(C3H 10T1/2
cells)
MN
No effect.
(Bisht et al.,
2002)
847.74 MHz, CDMA
Average SAR 3.2 - 5.1
W/kg
n=3
HEI-OC1
Mouse
auditory hair
cells
835.62 MHz, FDMA
3, 8, 16, 24 h
SB
1763 MHz, CDMA
No effect.
SAR 20 W/kg
6, 24, 48 h
n=3
For auditory
function, signal
transduction,
gene and protein
expression and
proliferation see
Section 6.4.1,
12.3.2, 12.3.3
and 12.3.6
(Huang et al.,
2008b)
No information on
blinding of staff
Rat bone
marrow
lymphocytes
SB
SB
1950 MHz, UMTS
No effect.
For apoptosis see (Zeni et al.,
Section 12.3.4
2012b)
No effect.
Human cells also
investigated.
(Speit, Schütz &
Hoffmann, 2007)
No effect.
Combined
exposures also
investigated.
(Kim et al., 2008)
SAR 10 W/kg
SB, MN
1800 MHz, CW
SAR 2 W/kg
1, 4, 18, 24 h (5 min
on/10 min off cycles)
SB, CA
835 MHz, CW
SAR 4 W/kg
24, 48 h
No information on
blinding of staff
n=2
Mouse neural
progenitor
stem cells
1710 MHz, GSM
n=6
CHO-K1 cells
MN
n=3
Average SAR 1.5 W/kg
SB increased at
6 h but not at 48 h.
Transient SB
increase.
6, 48 h (5 min on/30 min
off cycles)
No effect on CA
and SCE.
For signal
transduction,
gene and protein
expression,
apoptosis and
proliferation see
Section 12.3.2,
12.3.3, 12.3.4
and 12.3.6
2450 MHz, CW
Increased MN
frequency at 78
and 100 W/kg.
MN increase at
39°C for 18 h.
Increased MN
frequency at 100
and 200 W/kg.
Possible thermal
effect at high
SAR.
(Koyama et al.,
2004)
Increased MUT
frequency at 200
W/kg.
Possible thermal
effect at high
SAR.
(Koyama et al.,
2007)
SAR 13, 39, 50, 78, 100
W/kg
18 h
CHO-K1 cells
(Kumar et al.,
2011)
24 h
n=3
Mouse
L5178Y TK+/lymphoma
cells
For erythrocyte
maturation and
cell proliferation
see Section 10.3
and 12.3.6
30 min
n=3
V79 Chinese
Hamster Cells
No effect.
SAR 2 W/kg
n=11
rat neuronal
cells (PC12)
900 MHz, CW
MN
n=3
2450 MHz, CW
SAR 5, 10, 20, 50, 100,
200 W/kg
(Nikolova et al.,
2005)
(Koyama et al.,
2003)
Possible thermal
effect at high
SAR.
2h
CHO-K1 cells
n=3
MUT at HPRT locus
2450 MHz, CW
SAR 5, 10, 20, 50, 100,
200 W/kg
2h
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
152
Rat cortical
neurons
8-OHdG adducts
(mtDNA)
n=3
1800 MHz, GSM
Increase in 8OHdG adducts.
Average SAR 2 W/kg
24 h (5 min on/10 min off
cycles)
Effect reverted in (Xu et al., 2010)
cells pre-treated
with melatonin for
4 h (see Section
12.3.1.2
combined
exposures).
“No effect” means no statistically significant effect.
4014
4015
12.3.1.2
4016
4017
4018
4019
4020
Several studies have been performed to evaluate the effects of combined exposure to RF EMF and
genotoxic agents. The RF exposure was applied before, during and/or after treatment with genotoxic agents. All
studies included the effect of RF exposure alone and the details are presented in Section 12.3.1.1. In most of the
studies on combined exposures, RF exposure alone had no significant effects on any of the endpoint
investigated, unless otherwise mentioned.
4021
4022
4023
In the following, in vitro studies on RF combined with exposures to chemical agents, ionizing
radiation and ultra-violet rays are presented. Treatments with the chemical or physical agent alone served as
positive control.
4024
Chemical agents
4025
4026
In most cases, chemotherapeutic drugs have been employed to perform combined exposures, although
other chemical agents have also been used.
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
Three papers were published by the same research group, showing that 20 h exposure of human
peripheral blood lymphocytes to RF EMF was able to reduce the genotoxic effects induced by a subsequent
treatment with mitomycin-C (MMC), as assessed by MN frequency. The reduction was detected with a 900 MHz
GSM signal (SAR=1.25 W/kg) (Sannino et al., 2009b) in cell cultures from five healthy donors (p<0.01) and
with a 1950 MHz UMTS signal (SAR=0.3 W/kg; six donors; p<0.05). In the latter study, SARs of 0.15, 0.6 and
1.25 W/kg were also tested and gave no effect (Zeni et al., 2012a). The authors provided further evidence that
the cells were required to be exposed to RF EMF in the S-phase of the cell cycle to exhibit the reduced DNA
damage (cell cultures from nine donors) (Sannino et al., 2011) (Sannino et al., 2011). In all the studies, blind
protocols were followed. [Taken together, these studies suggest that non-ionizing radiation is capable of
inducing ‘adaptive response’ similar to that observed in several ionizing radiation exposure studies. Cell
proliferation has also been investigated, as reported in Section 12.3.6.]
4038
4039
4040
4041
4042
4043
Zhijian et al. (2010) used doxorubicin (DOX) to treat cultured human lymphoblastoid cells before,
during and after blind intermittent exposure (5 min on/10 min off cycles) to an 1800 MHz GSM signal (SAR=2
W/kg) for 2 hours. In experiments where RF was given before DOX or concurrently, the exposure duration was
2 h, while for exposures after DOX treatments it was 6, 12, 18 or 24 h. No significant effect of the RF exposures
on DOX-induced SB was found in any of the experimental schedules, although in certain exposure conditions
combined treatments can influence DNA repair (p<0.05), as assessed in four independent experiments.
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
Koyama and co-workers reported the data from three separate investigations carried out in Chinese
hamster K1 (CHO-K1) cells treated for 1 h with Bleomycin (BLM) and then exposed in blind to CW 2450 MHz
fields in different conditions. For each study, three independent experiments were carried out. In the first study
(Koyama et al., 2003), the incidence of MN was evaluated using the kinetochore method (a fluorescence
technique that employs specific antibodies to stain centromeres) in cells exposed for 18 hours at SAR values up
to 100 W/kg. A significant increase in MN frequency was found in cultures exposed to RF alone at SAR of 100
W/kg (p<0.01). Moreover, a synergistic increase in MN frequency (p<0.01) was detected in cells co-exposed to
SAR values higher than 50 W/kg, compared to cultures treated with BLM alone. In the second study (Koyama et
al., 2004) CHO-K1 cells were treated with BLM and then exposed for 2 hours in the same range of SARs. A
significant increase in MN was observed only in cells exposed at the highest SAR (200 W/kg; p<0.01). No
synergistic effects with BLM were detected. In the third investigation, (Koyama et al., 2007), the latter
experimental conditions were tested to evaluate mutation (MUT) frequency at the hypoxanthine-guanine
phosphoribosyl transferase (HPRT) locus. A statistically significant increase in MUT frequency was detected
after exposure to RF alone at 200 W/kg (p<0.05) and after co-exposure at 100 and 200 W/kg SAR (p<0.01). [As
also stated by the authors, a thermal effect cannot be excluded in samples exposures at high SARs].
Combined exposures
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
153
4059
4060
4061
4062
4063
4064
4065
4066
Kim et al. (2008) assessed SB and CA in mouse L5178Y TK+/- lymphoma cells exposed to 835 MHz
RF EMF (CW, SAR=4 W/kg) together with cyclophosphamide (CP), 4-Nitroquinoline-1-oxide (4NQO) or ethyl
methane sulfonate (EMS). For SB evaluation, cells were exposed to RF for 48 h and treated during the last 4 h
with CPA or 4NQO. To evaluate CA, cells were RF-exposed for 6 h in the presence of CPA or EMS. Culture
medium was then removed, cells were washed and reincubated in fresh medium for an additional 18 or 44 h with
or without RF. The results indicated a significant increase in SB in cultures co-exposed to CPA or 4NQO
(p<0.01). The incidence of CA was not affected in cells exposed to RF and CPA or EMS. [This study is based on
just two independent experiments, although several experimental conditions have been tested].
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
Luukkonen et al. (2009) exposed SH-SY5Y neuroblastoma cells for 1 h to 872 MHz fields, both CW
and GSM modulated, at an SAR of 5 W/kg, together with menadione. A blind protocol was applied. The results
from three independent experiments indicated a significant increase in SB in CW co-exposed cultures compared
with those treated with menadione alone (p<0.01), but not in those co-exposed to the GSM signal. In a similar
study, the same research group applied the same exposure protocol for 3 h and SH-SY5Y cells were concurrently
treated with ferrous chloride (FeCl2) and diethyl-maleate (DEM) (the latter enhances the free radicals induced by
the former, resulting in decreased antioxidant levels) (Luukkonen, Juutilainen & Naarala, 2010). No significant
effect was observed on SB in both CW and GSM exposure conditions, as assessed in four independent
experiments. [In both studies, the effect of combined exposure on oxidative stress has also been investigated, as
reported in Section 12.3.5.]
4077
4078
4079
4080
4081
4082
4083
Sannino et al. (2009a) exposed human skin fibroblasts from a healthy individual as well as a subject
with Turner’s syndrome (TS) to a 900 MHz GSM signal (SAR=1 W/kg) for 24 hours and then treated the cells
with MX (3-chloro-4-(dichloromethyl)-5-Hydroxy-2(5H)-furanone, a contaminant produced during chlorination
of water) for 1 hour. In addition, RF exposure of TS cells was also tested for 1 h. No significant effects of these
treatments in both cell types were observed on SB and MN, as assessed in three independent experiments,
performed in blind, for each condition/cell type. [The effect of combined exposure on cell proliferation was aslo
investigated, as reported in Section 12.3.6.]
4084
4085
4086
4087
4088
4089
4090
4091
Xu et al. (2010) dissected cortical neurons from newborn rats and after 8 days of culturing the cells
were exposed in blind to GSM 1800 MHz (SAR=2 W/kg), for 24 hours (5 min on/10 min off cycles). The cells
were treated with melatonin 4 h before RF EMF exposure. The levels of 8-OHdG adducts, a biomarkers for
oxidative damage, were examined in mitochondrial DNA (mtDNA). RF EMF exposure alone induced a
significant increase in adducts with a concomitant decrease on mtDNA copy numbers and mitochondrial RNA
transcripts, as assessed in three independent experiments (p<0.01). Each of these disturbances in mtDNA was
reversed in cells pre-treated with melatonin. [The control cells used by the authors were “sham-exposed” cells.
The effect of combined exposure on oxidative stress has also been investigated, as reported in Section 12.3.5.]
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
In two separate studies Hansteen and co-workers (2009a; 2009b) investigated the induction of CA in
stimulated human peripheral blood lymphocytes from six donors exposed in blind for 53 h to RF EMF and
treated with MMC after 30 hours of RF exposure. In the first study, cells were exposed at 2.3 GHz, 10 W/m2
power density, CW or pulsed fields (200 Hz pulse frequency, 50% duty cycle) (Hansteen et al., 2009a). In the
second study, RF exposure was carried out at 16.5 GHz (10 W/m2 power density, pulsed wave) and 18 GHz (1
W/m2 power density, CW) (Hansteen et al., 2009b). In both studies no significant effect of combined exposure
was detected. [Although the authors do not call it “sham” the control is actually a sham. The authors did not
include dosimetric details but provided the necessary information so that other groups can use it for numerical
analysis to repeat the study; the justification was that from rough calculations, the assessed exposure level was
close to the ICNIRP safety limits].
4102
Ionizing radiation (IR)
4103
4104
4105
4106
4107
4108
4109
Stronati et al. (2006) conducted a study in blind with the participation of researchers from two
independent laboratories to investigate the effect 1 Gy X-rays given before or after 24 h exposure to a 935 MHz
GSM signal, at SAR values of 1 and 2 W/kg, on human blood lymphocytes. Several endpoints were investigated,
such as SB (10 donors), chromosomal aberrations (14 donors), sister chromatid exchanges (four donors),
micronuclei (14 donors). The results indicated that RF exposure was not able to alter the effect of X-rays for any
of the endpoints investigated. [The effect of combined exposure on proliferation has been reported in section
12.3.6].
4110
4111
Manti et al. (2008) exposed in blind isolated leukocytes from four donors to 4 Gy X-rays and
immediately after to 1950 MHz RF EMF, UMTS modulated, at SARs of 0.5 and 2 W/kg for 24 h. Cells were
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
154
4112
4113
4114
4115
4116
4117
then stimulated to divide and the fluorescence in situ hybridization (FISH) technique was used with molecular
probes specific for whole chromosomes 1 and 2. A small but statistically significant increase in the number of
exchange aberrations per cell was observed in cells exposed to RF alone at 2 W/kg SAR (p<0.05), and a
synergistic effect was observed in cells from some donors exposed to RF and X-rays (p<0.05). The authors
suggested that RF exposure may either influence the repair of damage or alter the cell death pathways of the
damage response.
4118
4119
4120
4121
4122
Zhijian et al. (2009) exposed in blind isolated leukocytes from blood samples collected from four
healthy donors to an 1800 MHz GSM signal, at an SAR of 2 W/kg for 24 h (5 min on/10 min off cycles) and
then the cells were subjected to 0.25 or 0.5 Gy X-rays. The extent of SB and their repair was assessed at 0, 15,
45, 90, 150 and 240 minutes after X-ray exposure. No significant effects were observed on SB and their repair
when compared with the cells exposed to X-rays alone, although there was individual variation.
4123
4124
4125
4126
4127
Lagroye et al. (2004) exposed C3H 10T1/2 cells to CW RF EMF at 2450 MHz (SAR=1.9 W/kg) for 2
h followed by 4 Gy gamma irradiation. The positive control cells were treated with cis-platinum, a crosslinking
agent. The data from three to four independent experiments indicated that cells pre-treated with cis-platinum
showed an increase in gamma-rays-induced SB, while pre-exposure to RF EMF did not change the effect of
gamma-rays.
4128
Ultraviolet (UV) and other physical agents
4129
4130
4131
4132
Baohong et al. (2007) exposed isolated leukocytes from three healthy donors to different doses of UVC (0.25–2 J/m2) followed by GSM modulated 1800 MHz fields (SAR=3 W/kg) for 1.5 or 4 hours. The UVinduced SB were decreased after 1.5 h and increased after 4 h exposure to RF EMF in the range of 0.25–1.5
J/m2, respectively (p<0.05), while no effects were detected at 2 J/m2 exposure.
4133
4134
4135
4136
4137
Yao et al. (2008a) exposed cultured human SRA01/04 lens epithelial cells to intermittent (5 min on/10
min off cycles) 1800 MHz RF EMF (SAR=1, 2, 3, 4 W/kg) superposed with 2 µT electromagnetic noise (30–90
Hz magnetic fields in Helmholtz coils) for 2 hours. There was a significant increase in SB in cells exposed to RF
alone at 3 and 4 W/kg and the superposed electromagnetic noise was able to block the RF-induced DNA
damage, as assessed in three independent experiments.
4138
Studies not included in the analysis
4139
4140
4141
4142
4143
4144
Baohong et al. (2005) treated human blood lymphocytes with four different chemical mutagens,
MMC, 4NQO, BLM and MMS, for 3 h before or after a 2-h exposure to a 1800 MHz GSM signal (SAR=3
W/kg). A synergistic increase in SB was observed in cells treated with MMC and 4NQO (p<0.05), while an
inconsistent increase in SB was recorded in cells treated with BLM and MMS. Thus, the data suggested
differences in the interaction of RF with different chemical mutagens. [The interpretation of the results reported
in this study is difficult due to the inclusion of data from one donor only.]
4145
4146
4147
4148
4149
Tiwari et al. (2008) exposed human blood lymphocytes from six healthy donors for 1 h to RF EMF
(835 MHz, CDMA; SAR=1.17 W/kg) together with aphidicolin (APC) and examined SB. There was a
significant increase in SB in cells exposed to APC only and also in cells exposed to RF+APC. The increased
damage appears to be reversible, as it was repaired quickly. [In this study neither dosimetric information nor
details on the exposure system are provided. Moreover, it is not clear if the sham controls are actually sham.]
4150
4151
4152
4153
4154
4155
Esmekaya et al. (2011) observed a significant decrease in SCE in human blood lymphocytes exposed
to a GSM 1800 MHz signal (SAR=0.21 W/kg) for 6, 8 and 24 hours together with ginkgo, an anti-oxidant used
in alternative medicine. In this study, exposure of the cells to RF alone also increased the incidence of SCE. [In
this study the results are uninterpretable since no proper dosimetric evaluation was performed (SAR estimated by
using electric field measured along the horn antenna). Moreover, the number of donors included in the study and
the co-exposure protocol (RF before, after or concurrent to ginkgo) are not clear.]
4156
4157
4158
4159
4160
4161
Figueiredo et al. (2004) exposed whole blood samples from four healthy donors to RF EMF at 2450
MHz (power output 3 W; 2 donors) or at 10.5 GHz (power output 15 mW; 2 donors) for 40 sec and 5 min,
respectively. Then the cells were subjected to 1.5 Gy gamma radiation. No significant effect was observed in
conventional CA analysis. [In this study sham exposure is reported, although it is defined “negative control”.
However, exposure to 2450 MHz were performed by means of a microwave oven, while no information is
reported about the exposure system at 10.5 GHz.]
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
155
Table 12.3.4. In vitro studies assessing genotoxic effects of combined exposures to RF EMF and chemical or physical
agents
Cell type
Biological endpoint
Combined Exposure
Results
Comment
Authors
Decreased MN
frequency
RF-induced
adaptive
response.
(Sannino et al.,
2009b)
Number of
independent
experiments
Combined exposures with chemical agents
Human blood
lymphocytes
MN
900 MHz, GSM
Average SAR 1.25
W/kg
n=5
For cell
proliferation see
Section 12.3.6
20 h
RF in S-phase
MMC after RF
Human blood
lymphocytes
MN
1950 MHz, UMTS
SAR 0.15-1.25 W/Kg
n=6
20 h
RF in S-phase
Decreased MN
frequency at 0.3
W/kg but not at
0.15, 0.6, 1.25
W/kg.
SAR-dependent
RF-induced
adaptive
response.
Decreased MN
frequency if RF
given in S-phase.
RF-induced
adaptive
response,
dependent on
the phase of cell
cycle.
MMC after RF
Human blood
lymphocytes
MN
900 MHz, GSM
Average SAR 1.25
W/kg
n=-9
No effects for RF
in in G0 or G2/M
phase.
20 h
RF in G0, S or G2/M
phase
SB
1800 MHz, GSM
No effects on SB.
Average 2 W/kg
RF+DOC can
influence DNA
repair in certain
experimental
conditions.
2-24 h
n=4
DOX
RF before, concurrently
or after DOX
Chinese
hamster K1
cells
MN (kinetochore
method)
n=3
For cell
proliferation see
Section 12.3.6
(Sannino et al.,
2011)
For cell
proliferation see
Section 12.3.6
MMC after RF
human
lymphoblastoid
cells
(Zeni et al., 2012a)
(Zhijian et al., 2010)
2450 MHz, CW
Synergistic
Significant
increase at 78 and increase at 100
100 W/kg.
W/kg in cultures
1 h BLM before 18 h RF
exposed to RF
alone.
SAR 25, 78, 100 W/kg
(Koyama et al.,
2003)
Thermal effects
cannot be
excluded at
higher SAR
levels
Chinese
hamster K1
cells
MN
2450 MHz, CW
SAR 5, 10, 20, 50, 100,
200 W/kg SAR
n=3
No effect of coexposures
1 h BLM before 2 h RF
Significant
(Koyama et al.,
increase in MN
2004)
at 200 W/kg in
cultures exposed
to RF alone.
Thermal effects
cannot be
excluded at
higher SAR
levels
Chinese
hamster K1
cells
n=3
MUT at HPRT locus
2450 MHz, CW
SAR 5, 10, 20, 50, 100,
200 W/kg SAR
Synergistic effect
at 100 and 200
W/kg.
1 h BLM before 2 h RF
Increased MUT
(Koyama et al.,
frequency in
2007)
cultures exposed
to RF alone.
Thermal effects
cannot be
excluded at
higher SAR
levels
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
156
Mouse L5178Y SB, CA
TK+/lymphoma cells
n=2
835 MHz, CW
Increased SB in
CPA and 4NQO
co-exposed
cultures.
SAR 4 W/kg
24, 48 h
CPA, EMS or 4NQO
concurrent to RF
Neuroblastoma SB
SH-SY5Y cells
n=3
No information
on blinding of
staff
(Kim et al., 2008)
No effect on CA.
872 MHz, CW and GSM Increased SB in
CW, but not GSM
SAR 5 W/kg
co-exposed
1h
cultures.
Menadione
Effect dependent (Luukkonen et al.,
on RF signal.
2009)
For oxidative
stress see
Section 12.3.5
RF concurrently
Neuroblastoma SB
SH-SY5Y cells
872 MHz, CW and GSM No effect.
n=4
3h
SAR 5 W/kg
For oxidative
stress see
Section 12.3.5
(Luukkonen,
Juutilainen &
Naarala, 2010)
For cell
proliferation see
Section 12.3.6
(Sannino et al.,
2009a)
DEM or FeCl2
RF concurrently
Skin fibroblasts SB, MN
from healthy
and TS donors
n=3
900 MHz, GSM
No effect.
Average SAR 1 W/kg
24 h
1 h (SB in TS cells)
MX
RF before MX
Rat cortical
neurons
8-OHdG adducts
(mtDNA)
1800 MHz, GSM
RF-induced
increase in 8OHdG adducts
24 h (5 min on/10 min
reverted by
off cycles)
melatonin
Melatonin 4 h before RF treatment.
For cell
proliferation see
Section 12.3.6
(Xu et al., 2010)
CA
2.3 GHz, CW or pulsed
(200 Hz pulse
frequency, 50% duty
cycle)
No effect
No information
provided to
perform
numerical
dosimetry
(Hansteen et al.,
2009a)
No effect
No information
provided to
perform
numerical
dosimetry
(Hansteen et al.,
2009b)
No effect.
For cell
proliferation see
Section 12.3.6
(Stronati et al.,
2006)
Synergistic effect
at 2 W/kg
Increase in
exchange
aberrations at 2
W/kg alone
(p<0.05).
(Manti et al., 2008)
n=3
Blood
lymphocytes
n=6
Average SAR 2 W/kg
power density 10 W/m2
30 h MMC after 53 h RF
Blood
lymphocytes
CA
n=6
16.5 GHz, pulsed (1
kHz pulse frequency,
50% duty cycle)
power density 10 W/m2
30 h MMC after 53 h RF
Combined exposures with ionizing radiation
Blood
lymphocytes
SB, CA, SCE, MN
935 MHz, GSM
Average SAR 1, 2 W/kg
n=14
24 h
1 Gy X-rays before or
after RF
Blood
lymphocytes
CA
1950 MHz, UMTS
SAR 0.5, 2 W/kg
n=4
24 h
4 Gy X-rays before RF
Blood
lymphocytes
n=4
SB
1800 MHz, GSM
Average SAR 2 W/kg
No effect on SB
and DNA-repair.
(Zhijian et al., 2009)
24 h (5 min on/10 min
off cycles)
RF before 0.25 or 0.5
Gy γ-rays
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
157
Mouse
fibroblasts
(C3H 10T1/2
cells)
SB
2450 MHz, CW
No effect.
(Lagroye et al.,
2004)
SAR 1.9 W/kg
4 Gy γ-rays after 2 h RF
n=3-4
Combined exposures with Ultraviolet and other physical agents
Blood
lymphocytes
SB
1800 MHz, GSM
Average SAR 3 W/kg
n=3
1.5 or 4 h
RF after UV-C (0.252 J/m2)
Lens epithelial
(hLEC) cells
SB, γ-H2AX foci
n=3
1800 MHz GSM
Average SAR 1, 2, 3, 4
W/kg
Decrease in UVinduced SB at 1.5
h RF and increase
at 4 h RF.No
effect for 2 J/m2.
(Baohong et al.,
2007)
MF noise negated RF alone
RF-induced SB.
increased SB at
3 and 4 W/kg.
(Yao et al., 2008a)
2 h (5 min on/10 min off
cycles)
MF noise superposition
(2 µT, 30-90 Hz)
For ocular
function and
oxidative stress
see Section
6.4.2 and 12.3.5.
No information
on blinding of
staff
“No effect" means no statistically significant effect.
4162
4163
Excluded references
4164
4165
4166
4167
4168
4169
(Agarwal et al., 2009; Antonopoulos, Eisenbrandt & Obe, 1997; Ballardin et al., 2011; d'Ambrosio et
al., 2002; d'Ambrosio G., 1995; De Iuliis et al., 2009; Falzone et al., 2010; Gajski & Garaj-Vrhovac, 2009; Maes
et al., 1993; Maes et al., 1995; 1996; Maes et al., 1997; Maes, Collier & Verschaeve, 2000; 2001; Scarfi et al.,
1996; Schmid & Schrader, 2007; Schrader et al., 2008; Shckorbatov et al., 1998; Shckorbatov et al., 2010;
Syldona, 2007; Yamazaki, Matsubara & Yamada, 1993; Zeni et al., 2003; Zhang et al., 2002; Zotti-Martelli et
al., 2000; Zotti-Martelli et al., 2005)
4170
12.3.2
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
Studies on intracellular and intercellular signalling include release of calcium ions, signal transduction
pathways and cytokine expression. In the previous WHO report (WHO, 1993b) only studies on release of
calcium ions were reported and the general conclusion was that, although calcium release may be useful to
understand the mechanisms of action, insufficient information is available to suggest a potential adverse effect
on human health.
The present literature search identified 54 relevant papers on this topic. Six of them were in a
language that could not be understood. That left 48 papers to be extracted. Among the relevant publications, 13
were excluded because they did not meet the inclusion criteria for in vitro studies, and references are listed at the
end of this section. Twelve papers did not completely comply with the quality criteria for inclusion due to
methodological issues, therefore they are only presented in the text. The remaining 23 papers have been
described in the text and summarized in tables.
Unless specifically mentioned, papers did not report on blinding of the investigators to the exposure
conditions.
4184
Intracellular Calcium
4185
4186
4187
4188
4189
4190
4191
Wolke et al. (1996) studied the effects of continuous wave (CW) and differently modulated (16 Hz–
30 kHz) 900–1800 MHz RF EMF on intracellular calcium levels, measured as fura-2 fluorescence by means of
digital image analysis, in Guinea pig isolated ventricular myocytes. They used several SAR levels (0.009–0.059
W/kg) and short (500 s) and long-term (120 min) exposure times and analysed 31–35 single cells. In general,
they noticed no differences between the sham and RF exposed groups, except a significant difference detected in
the 900 MHz group which was modulated at 50 Hz (p<0.01). Treatment of cells with a solution containing KCI
instead of NaCI, served as positive control and induced an increase in calcium concentration. [As also stated
Intracellular and intercellular signalling
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
158
4192
4193
by the authors, such difference cannot be regarded as relevant, since the measured mean value lies within the
limits of the standard deviation of the sham exposure.]
4194
4195
4196
4197
4198
4199
4200
4201
Linz and co-workers (Linz et al., 1999) exposed isolated rat and Guinea pig ventricular myocytes to
RF EMF in a frequency range between 180 and 1800 MHz. In particular, exposures to 1800 and 900 MHz, GSM
modulated, were carried out at different SAR values (0.015–0.88 W/kg), while exposure at 180 MHz, CW, was
conducted at 0.08 W/kg. They applied a patch clamp technique on 5–14 single cells to evaluate action potentials
(current-clamp mode) and L-type Ca2+ and K+ membrane currents (voltage-clamp mode). No effects on
electrophysiological membrane characteristics were detected in the two cell types investigated for any of the RF
exposure conditions tested compared to sham exposed samples, while lowering the ambient temperature as a
positive control induced the expected changes.
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
The study of Cranfield et al. (2001) investigated changes in average calcium levels or trends in human
leukaemic Jurkat T-cells exposed to 915 MHz (CW and GSM) RF EMF with a weighted average SAR of 1.5
W/kg for 10 min. Intracellular calcium concentrations (spiking) were estimated by Fluo-3 fluorescence using
laser scanning confocal microscopy. Since the mitogen phytohaemagglutinin (PHA) induces repetitive transient
rises in calcium spiking within about 100 s of addition, both PHA-activated and inactivated cells were studied.
The experiments were performed blinded. No differences were detected between exposed and sham exposed
groups. When the power spectral density distribution of fluorescence was analysed, a significant difference
(p<0.05) was detected, but only in PHA-activated cells exposed to the GSM signal. For each experimental
condition, at least 100 single cells were analysed. Positive controls were not included in the study design. [As
also stated by the authors, the observed difference could be a statistical artefact, due to the large variability of the
observed phenomenon. In this study the number of independent experiments is not reported, but the authors
stated that, since each experiment yielded between 0 and 20 analysable cells, the results from sufficient
experiments were pooled to allow a comparison of at least 100 cells for each condition.]
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
Green et al. (2005) showed no evidence of any consistent or biologically relevant effect on
intracellular calcium concentration ([Ca2+]i ) when rat cerebellar granule cells (from 6 to 9 independent
experiments) or cardiac myocytes (93–177 single cells) were exposed for 20 minutes to Terrestrial Trunked
Radio signals (TETRA) at 380.8875 MHz (pulse modulated at 17.6 Hz, 25% duty cycle) with SAR levels
between 0.005 and 0.4 W/kg. Although increases in [Ca2+]i in response to potassium-induced depolarization in
TETRA-exposed cells were different from sham controls, the majority of the differences was attributable to
initial biological variation between cell cultures. The fluorescent dyes Fura-PE3, Fluo-3 or Fluo-4 were
employed to measure [Ca2+]I using digital image analysis. The study was carried out blinded. Positive controls
were not included. [Indication about the homogeneity of the SAR distribution is not given. The study is also
reported in Section 5.4 (Brain physiology and function).]
4225
4226
4227
4228
4229
4230
4231
Whole-cell current-clamp and single-channel recording was used by Marchionni et al. (2006) to study
the effect of 900 MHz CW EMF on rat dorsal root ganglion neurons (four to nine independent experiments).
Exposure at a SAR value of 1 W/kg for 10 s did not modify the frequency of action potentials, compared to sham
exposed samples and did not affect the L-type Ca2+ current and the Ca2+-activated K+ current, which are involved
in the control of the interspike interval. Positive controls have not been included in the study design. [For
dosimetric analysis the authors refer to a previous report (Liberti et al., 2004). This study has also been described
in Section 5.4 (Brain physiology and function).]
4232
4233
4234
4235
4236
4237
4238
4239
4240
Platano and colleagues (Platano et al., 2007) studied Ba2+ currents through voltage-gated calcium
channels (VGCC) using primary cultures of rat cortical neurons, which were exposed for 1–3 periods of 90 sec
to 900 MHz, CW or GSM modulated, RF EMF (SAR = 2 W/kg). They measured Ba2+ currents (to avoid Ca2+dependent inactivation of the currents) by means of whole-cell patch-clamp technique. The results, obtained
from three to seven independent experiments, indicate that current amplitude or the current-voltage relationship
were not significantly altered by single or multiple acute exposures to 900 MHz CW or GSM EMF compared to
sham samples. Samples treated with CdCl2, a specific blocker of voltage-gated calcium channels, were used as
positive controls and gave positive findings. [These findings are in agreement with the results reported by Linz et
al., (Linz et al., 1999)). This study has also been described in Section 5.4 (Brain physiology and function).]
4241
4242
4243
4244
4245
Rao and coworkers (Rao et al., 2008) employed neuronal cells differentiated from a mouse embryonic
P19 carcinoma to study calcium dynamics and cytosolic calcium concentrations and spikes after exposing the
cells to CW 700–1100 MHz RF EMF. They tested several SAR levels (0.5–50 W/kg) and the exposure duration
was 60 min. Fluo-4 AM fluorescent dye, by means of fluorescence and differential interference contrast images
and time-lapse sequences, was used to monitor Ca2+ spikes. They found that RF exposure significantly increased
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
159
4246
4247
4248
4249
4250
4251
4252
4253
4254
the number of spontaneous [Ca2+]I spikes (p<0.05) in three to four independent experiments. The change was
dependent on the frequency (with a peak effect at 800 MHz) but not on the SAR in the range 0.5 to 5 W/kg.
When 50 W/kg was tested, the change was significantly lower than with the lower SAR values and was
accompanied by a temperature increase (>5 °C), which may have introduced thermally-induced alterations in
Ca2+ dynamics. In sham-exposed cells, spontaneous Ca2+ spiking could be blocked by ω-conotoxin GV1A (a
selective blocker of the N-type voltage-gated Ca2+ channels) or U73122 (a phospholipase C inhibitor). No effect
of RF exposure at 0.5 W/kg was found. These observations indicate that N-type voltage-gated Ca2+ channels and
phospholipase C are involved in intrinsic Ca2+ spiking, and may be modulated by RF. [This study has been also
described in Sections 5.4 (Brain physiology) and 12.3.6.1 (Cell differentiation).]
4255
4256
4257
4258
4259
4260
4261
4262
O’Connor et al. (2010) exposed human endothelial cells, rat PC-12 neuroblastoma cells and rat
primary hippocampal neurons to 900 MHz, GSM modulation; for human endothelial cells a CW signal was also
tested. The study was performed blinded. The high-throughput imaging platform was employed to monitor
changes in cellular Ca2+ levels with Fura-2 AM or Fura-PE3 AM fluorescent dyes. The exposure was 30 min
long (SAR 0.012–2 W/kg). The experiments were designed to study the effects of RF on both resting and
induced Ca2+ signals. The results indicate no effect of RF exposure on basal Ca 2+ homeostasis or provoked Ca2+
signals compared to sham-controls. No positive controls were included. [The number of experiments carried out
is unclear. This study has also been described in Section 5.4 (Brain physiology and function).]
Table 12.3.5. In vitro studies assessing effects of RF EMF exposure on intracellular calcium
Cell type
Biological endpoint
Exposure conditions
Results
Comment
Reference
Intracellular [Ca2+].
900 MHz, CW, 16, 50,
217 Hz, and 30 kHz
modulated
Increase in [Ca2+] No information on (Wolke et al.,
at 900 MHz, 50 Hz blinding of staff.
1996)
modulated.
SAR 0.015,0.029, 0.03,
0.059 W/kg
No effect of the
other conditions
tested.
Number of
independent
experiments
Guinea pig
isolated
ventricular
cardiac
myocytes
n=31-55a
1300 MHz, 217 Hz
modulated
SAR 0.012 W/kg
1800 MHz, 217 Hz
modulated
SAR 0.009 W/kg
500 s and 120 min
Guinea pig
and rat
isolated
ventricular
myocytes
Action potentials, and
membrane currents (Ltype Ca2+ and K+
currents)
180 MHz, CW
No effect.
No information on (Linz et al., 1999)
blinding of staff.
No changes in
average calcium
levels.
High variability of
the observed
phenomenon
SAR 0.080 W/kg
900 MHz, GSM
Average SAR 0.015,
0.25 W/kg
n=5-14a
1800 MHz, GSM
Average SAR 0.080–
0.88 W/kg
3 min (maximum)
Human Jurkat
T-cells
n=98-107a
2+
Intracellular [Ca ] in
presence or in absence
of PHA.
915 MHz, CW and GSM
Weighted average SAR
1.5 W/kg
(Cranfield et al.,
2001)
Significant
difference in PHAactivated cells
exposed to GSM
signal.
10 min
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
160
Rat cerebellar
granule cells
Intracellular [Ca2+].
n=6-9
380.8875 MHz (pulse
modulated at 17.6 Hz,
25% duty cycle)
No effect.
For Brain
physiology see
Section 5.4
(Green et al.,
2005)
No effect.
For Brain
physiology see
Section 5.4
(Marchionni et al.,
2006)
SAR 0.4 W/ kg
20 min
Cardiac
myocytes
SAR 0.005–0.04 W/ kg
n= 93-177a
10 min
Rat dorsal root Single channel ionic
ganglion
currents and firing
isolated
frequency
neurons
900 MHz, CW
SAR 1 W/kg
10 sec
No information on
blinding of staff.
n=3-9a
Primary
cultures of rat
cortical
neurons
Ion currents through
900 MHz, CW and GSM
VGCC using Ba2+ as ion SAR 2 W/kg
carrier
1-3 x 90 sec
No effect.
700-1100 MHz, CW
Increase in
calcium spiking
frequencies.
SAR 0.5-50 W/kg
60 min
Multiple parameters of
intracellular calcium
signals (peak
amplitude, integrated
Ca2+ signal, recovery
rates)
For Brain
(Rao et al., 2008)
physiology and
differentiation see
Sections 5.4 and
12.3.6.2
No information on
blinding of staff.
n=3-4
Human
endothelial
cells, rat PC-12
neuroblastoma
cells and rat
primary
hippocampal
neurons
(Platano et al.,
2007)
No information on
blinding of staff.
n=3-7
Neuronal cells Calcium dynamics and
derived from
cytosolic [Ca2+].
mouse
embryonal
P19 carcinoma
cells
For Brain
physiology see
Section 5.4
900 MHz, CW and
GSM
No effect.
SAR 0.012-2 W/kg
30 min
2 W/kg used for all cell
types; lower SARs for
human endothelial cells
only.
(O'Connor et
al., 2010)
For Brain physiology
see Section 5.4
n not clear
a
single cells
“No effect” means no statistically significant effect
Abbreviations: [Ca2+]: calcium concentration; CW: continuous wave; GSM: Global System for Mobile Communication; PHA:
Phytohaemagglutinin; SAR: specific absorption rate; VGCC: voltage-gated calcium channels.
4263
4264
Signal transduction pathways
4265
4266
4267
4268
4269
4270
Transcriptional activity in response to stressors can be mediated by mitogen-activated protein kinase
(MAPK) pathways that include the extracellular-signal regulated kinases (ERKs), p38 and the 1/2c-Jun Nterminal kinase (JNK) cascades. These pathways regulate a variety of cellular processes including proliferation,
differentiation, metabolism and the stressor response. Upon phosphorylation of these kinases, a large number of
regulatory proteins and transcription factors (Egr-1, Elk-1) can become activated, thereby altering cellular
processes and allowing further gene transcription.
4271
4272
4273
4274
4275
4276
Lee et al. (2006) exposed Jurkat cells and rat primary astrocytes to 1763 MHz RF EMF at SARs of 2
or 20 W/kg for 30 min to 1 h in the presence or absence of the tumour promotor 12-O-tetradecanoylphorbol-13acetate (TPA). No evidence of altered phosphorylation status was observed for ERK1/2, JNK1/2 or p38-MAPK
after exposure in either the presence or absence of TPA, as assessed in three independent experiments.
Treatments with TPA alone induced a dose-dependent MAPK phosphorylation, and served as positive controls.
[This study has also been presented in Section 12.3.3 (gene and protein expression.)]
4277
4278
4279
4280
4281
4282
In a further study, the same research group exposed human mammary breast epithelial (MCF10A)
cells (Kim et al. (2012). The exposure conditions tested were 837 MHz, CDMA, (SAR = 4 W/kg) for 4 h or 837
MHz, CDMA, plus 1950 MHz, WCDMA, at SAR of 2 W/kg for each signal (SAR = 4 W/kg) for 4. Moreover,
the latter condition was also tested 2 h per day for three consecutive days. The expression levels and
phosphorylation states of several MAP kinases (ERK, JNK and p38) were analyzed by Western blot. The authors
found no evidence that single or repeated exposure of these cells could elicit expression or phosphorylation
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
161
4283
4284
4285
4286
changes in any kinases assessed (three independent experiments). In positive controls, assessed by treating cell
cultures with 1 Gy gamma rays, phosphorylation of both JNK and ERK occurred, while phosphorylation of p38
was not induced. [The results related to positive controls are not shown in the paper. This study has been also
reported in Section 12.3.3 (gene and protein expression)]
4287
4288
4289
4290
4291
4292
Huang et al. (2008b) exposed a mouse auditory hair cell line (HEI-OC1) to 1763 MHz CDMAmodulated RF EMF at a SAR of 20 W/kg for 15–120 min. No changes in the protein expression level or
phosphorylation status of ERK, JNK or p38 were observed in RF exposed cultures compared to sham, as
assessed in three independent experiments. Heating of Jurkat cells to 43 ± 0.2 °C for 30 min was included as a
positive control, whereby positive findings were detected. [This study has also been reported in Section 6.4
(Auditory functions), 12.3.1 (genotoxicity), 12.3.3 (gene and protein expression) and 12.3.6 (cell proliferation).]
4293
4294
4295
In some studies the effect of RF EMF exposure on the expression of other eukaryotic transcriptional
regulators, tumor suppressor genes, cell cycle proteins, signalling molecules and growth factors has been
examined.
4296
4297
4298
4299
4300
4301
4302
4303
4304
Nikolova et al. (2005) exposed pluripotent embryonic mouse stem cells to 1710 MHz GSMmodulated RF EMF at an average SAR of 1.5 W/kg for up to 48 h. The experiments were carried out blinded.
The authors observed transient changes in the expression of bax, GADD45 and Nurr1 at specific (isolated) times
after exposure, but no consistent alterations in the expression of these genes was observed at other time-points
(five independent experiments). The authors observed no corresponding changes on cellular proliferation or
apoptosis leading them to speculate that the effect of RF on these genes may be compensated for at the
translational and post-translational level. Positive controls were not included in the study. [This study has also
been described in Sections 12.3.1 (Genotoxicity), 12.3.3 (Gene and protein expression), 12.3.4 (Apoptosis) and
12.3.6 (Cell proliferation).]
4305
4306
4307
4308
4309
Li et al. (1999) exposed human fibroblasts (primary and WSneo1 cells) to 837 MHz continuous-wave
RF EMF at SARs of 0.9 or 9.0 W/kg for 2 h, and assessed the protein expression levels of p53 by Western blot
(three independent experiments). No differences were detected compared to sham exposed cultures at 2 to 48 h
after RF exposure, while cultures exposed to ultraviolet radiation as positive control showed a significant
increase in p53 expression. The experiments were carried out blinded.
4310
4311
4312
4313
4314
4315
4316
4317
4318
Lee et al. (2011b) investigated if single or combined RF EMF exposures interfere with cell cycle
regulation. They exposed human breast MCF7 cancer cells for 1 h to either 837 MHz, CDMA (average SAR = 4
W/kg), or 837 MHz CDMA plus 1950 MHz WCDMA (2 W/kg CDMA and 2 W/kg WCDMA). In three
independent experiments, carried out blinded, protein expression level of p53 and p21, known to regulate the
function of cyclin-dependent kinases, were unaffected by RF exposure. Similarly, the protein expression of
cyclin proteins A, B1 and D1 were also unaffected by RF exposure, leading the authors to conclude that neither
single nor combined RF exposure affected cycle progression in this cell line. The positive control group was
exposed to gamma radiation and showed changes in cell cycle distribution. [This study has also been described
in Section 12.3.6 (Cell proliferation).]
4319
4320
4321
4322
4323
4324
4325
Bourthoumieu et al. (2013) investigated the expression of the p53 protein and its activation (due to its
ability to initiate apoptosis) in human amniotic cells exposed for 24 h to 900 MHz GSM modulated RF EMF
(SARs = 0.25, 1, 2, and 4 W/kg). The results of three independent experiments performed using three different
donors showed no effect in p53 expression (Western blot assay) by comparing sham-exposed to RF-exposed
cultures. Treatments with bleomycin as positive control resulted in a significant increase in apoptotic cells. [This
study has also been described in Sections 12.3.3 (Gene and protein expression), 12.3.4 (Apoptosis), and 12.3.6
(Cell proliferation).]
4326
4327
4328
4329
4330
Sun et al. (2012) exposed cultured human amniotic (FL) cells to 1800 MHz GSM RF EMF for 15 min
at SARs ranging from 0.1 to 4.0 W/kg. The authors reported clustering of epidermal growth factor receptors
(EGFR) on the cell membrane of RF-exposed cells and a significant increase (p<0.01) in phosphorylated EGFR
protein expression at RF exposures from 0.5 to 4.0 W/kg compared to sham exposed cultures (three independent
experiments carried out blinded). No positive control was included in the study.
4331
4332
4333
4334
Tuschl et al. (2006) investigated the effect of 8 h intermittent (5 min on/10 min off cycles) exposure of
human peripheral blood mononuclear cells from 15 different donors to 1950 MHz GSM-modulated RF EMF at a
SAR of 1 W/kg, on the transcript expression of a variety of cytokines and immune-relevant genes using a PCRarray. The experiments were carried out blinded. No changes in exposed cells compared to sham controls were
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
162
4335
4336
4337
4338
4339
detected in intracellular production of IL2 and interferon  (INF) in lymphocytes and IL1 and tumour necrosis
factor-α (TNFα) in monocytes, activity of immune-relevant genes (IL1α and β, IL2, IL2-receptor, IL4,
macrophage colony-stimulating factor (MCSF) receptor, TNFα, TNFα-receptor), and cytotoxicity of
lymphokine-activated killer cells (LAK cells). Positive controls were not included in the study design. [This
paper has also been described in Section 10.4 (Immune system and haematology).]
4340
4341
4342
4343
4344
Cervellati et al. (2009) exposed a human trophoblast cell line (HTR-8/SVneo) to 1817 MHz GSMmodulated RF EMF at a SAR of 2.0 W/kg for 1 h. In five independent experiments carried out blinded, the
authors observed increased transcript expression of connexins 40 and 43 (p<0.001) in exposed cultures compared
to sham exposed ones, however no changes were observed in the protein expression levels of these membrane
proteins. Positive controls were not included in the study.
4345
Studies not included in the analysis
4346
4347
4348
4349
4350
4351
4352
Leszczynski et al. (2002) exposed EA.hy926 cells to 900 MHz GSM-modulated RF EMF for 1 h at 2
W/kg. In four independent experiments, an increased expression of p38MAPK and HSP27 was reported at 1 h
post-exposure, while decreased expression of these proteins was reported at 8 h after RF EMF exposure, relative
to the sham controls. The authors further reported that the phosphorylation of HSP27 could be inhibited by
SB203580 (a specific inhibitor of p38-MAPK). [The absence of a statistical analysis, to determine if any of these
results was significant, makes the conclusion of this paper questionable. This study has also been presented in
Sections 5.4. (Brain physiology and function), 12.3.3 (Gene and protein expression) and 12.3.4 (Apoptosis).]
4353
4354
4355
4356
4357
4358
4359
4360
Zimmerman et al. (2012) exposed hepatocarcinoma cells (HepG2 and Huh7) to 27.1 MHz (100 Hz –
21 kHz frequency-modulated) RF EMF (SAR = 0.4 W/kg) for 3 h per day for 1 week. The authors found no
significant differences in gene transcript expression between sham and RF EMF-exposed cultures in HepG2 cells
by RNA-Seq analysis. They further examined two genes with an absolute fold-change >1.8 by RT-PCR and
found decreased transcript levels for PLP2 and XCL2 in HepG2 and Huh7 cell lines, increased transcript
expression for PLP2 and XCL2 in a non-malignant cell line (THLE-2), and no effect on PLP2 expression in an
MCF7 cell line after exposure (p< 0.05; four independent experiments). [The significance of these findings is
difficult to interpret as the RNA-Seq and RT-PCR data showed inconsistent results.]
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
Buttiglione et al. (2007) assessed the expression levels of several transcription factors (Egr-1, Bcl-2,
Elk-1) downstream of MAPK pathways in a human neuroblastoma cell line (SH-SY5Y) exposed to 900 MHz
GSM-modulated RF EMF at a SAR of 1 W/kg for 5 min to 24 h. In three independent experiments, carried out
blinded, the authors observed a transient increase in Egr-1 transcript levels at 5–30 min after exposure, but the
effect was no longer evident after 6–24 h exposure. Phosphorylation of ERK1/2, JNK1/2 and Elk-1 was also
transiently increased after various RF exposure times (5 min to 6 h), while a significant decrease in the transcript
levels of Bcl-2 and Survivin were observed after 24 h of exposure. P53 protein expression and Bak and Bax
transcript expression were not affected by RF exposure. However, a significant decrease in cell viability, the
appearance of sub-G1 nuclei and a G2M-block were observed after 24 h of exposure, indicating that apoptosis
was induced in the cell culture. Positive controls were not included in the study. [The authors reported the
efficiency of their exposure system to be 0.35 W/kg/(input Watt) and aimed for a target SAR of 1 W/kg.
However, it was stated that an input power of 31.6 W was fed into the exposure system, which would lead to a
SAR of approximately 10 W/kg. Since culture temperatures were not measured during RF exposure and since
the exposure details are not clearly reported, the results of this study must be cautiously interpreted as thermal
confounding may have occurred. This study has also been described in Sections 12.3.3 (gene and protein
expression), 12.3.4 (Apoptosis) and 12.3.6 (Cell proliferation).]
4377
4378
4379
4380
4381
4382
4383
4384
Hirose et al. (2006) examined gene transcript levels in A172 and IMR-90 cells following RF EMF
exposure. A series of genes known to be components of p53-mediated apoptosis (including APAF1, TP53,
TP53BP and CASP9) were assessed after cells were exposed to 2142.5 MHz at SARs of 0.08–0.8 W/kg for up to
48 h. The authors observed no significant differences in the expression of these p53-related apoptosis genes
relative to sham-control groups under any condition tested. Heating of cultures at 42 °C served as positive
control and resulted in a significant increase in gene expression. [These observations are based on only two
independent experiments for each condition, thereby limiting the significance of these results. This study has
also been reported in Sections 12.3.3 (Gene and protein expression) and 12.3.4 (Apoptosis).]
4385
4386
4387
Jin and colleagues (Jin et al., 2012b) showed that RF exposure of human promyelocytic leukaemia
HL-60 cells to 900 MHz, CW, at 12 µW/cm2 power density for 1 h per day for 3 days did not induce differences
in intracellular Ca2+ levels with respect to sham controls. Intracellular free Ca2+ and Ca2+-Mg2+-ATPase activities
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
163
4388
4389
4390
4391
4392
4393
were measured by employing cytofluorimetric and spectrophotometric techniques, respectively. Moreover, the
effect of co-exposure with doxorubicin (DOX, a chemotherapeutic drug) were also investigated: pre-exposure to
RF protected HL-60 cells from subsequent treatments with DOX, since the intracellular Ca2+ levels were lower
in RF-exposed and DOX-treated cells as compared to cells treated with DOX alone (p<0.01). [The results refer
to only two independent experiments, thereby limiting the significance of these results. This study has also been
described in Section 12.3.4 (Apoptosis)].]
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
Yoon et al. (2011) exposed cultured human dermal cells (hDPCs) to 1763 MHz RF EMF at SARs of 2
or 10 W/kg. After 3 h exposure at a SAR of 10 W/kg, the authors observed increased transcript expression of
IGF-1 (p<0.05). They also observed increased phosphorylation of MAPK1 and protein expression of BCL-2 and
cyclin D1, but decreased expression of Bax [this data appears to results from one independent experiment]. In a
follow-up experiment, VEGF and IGF-1 transcript expression were increased after only 1 h of RF exposure at 10
W/kg (p<0.05). Similar alterations in IGF-1 transcript expression were not observed in four other cell lines
exposed to the similar RF exposure conditions. Positive controls were not included in the study design. [It is
unclear whether mRNA was pooled from multiple donors for transcript analysis, which would effectively result
in one independent sample, or whether true independent experiments were conducted. Since the experimental
approach employed is unclear, the significance of these findings is questionable. This study has also been
described in Sections 12.3.4 (Apoptosis), 12.3.5 (Oxidative stress) and 12.3.6 (Cell proliferation).]
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
Hao et al. (2010) exposed murine microglial (N9) cells to 2450 MHz pulsed (2 µs, 500 pps) RF EMF
for 20 min at a SAR of 6.2 W/kg (three independent experiments). At 1–12 h after RF exposure, cell extracts
were assessed by Western blot for the expression levels of phosphorylated STAT3, JAK1 and JAK2. Increased
expression of p-STAT3 and p-JAK2 were reported at 1–12 h after RF exposure, while p-JAK1 was only
increased at 1 h post-exposure. When cells were co-exposed to pyridine 6 (an inhibitor of JAK), p-STAT3 by RF
EMF was not increased. In a related study by the same group, Yang et al. (Yang et al., 2010) investigated
additional time-points and inhibitor conditions (three independent experiments). N9 microglia cells were
exposed to RF EMF for 20 min at a SAR of 6 W/kg, then assessed at 1–24 h thereafter. The authors reported
that N9 cells showed increased p-STAT3 immunoreactivity following RF exposure, when imaged by confocal
microscopy (p<0.01). Western blot analysis demonstrated increased levels of both p-STAT3 and p-JAK2 at
times ranging from 1–24 h after RF exposure (p<0.01), while p-JAK1 was only increased in expression at 1 h
post-exposure. [In both studies, the cultures were exposed to a relatively high RF field intensity of ~6 W/kg.
Since culture temperatures were not monitored during or after RF exposure and SAR heterogeneity was not
assessed within the culture flasks, thermal confounding in these studies cannot be excluded. Moreover, the
results cannot be interpreted since the description of the exposure system and dosimetry is not adequate. These
studies have also been described in Section 8.3 (Neurodegenerative disorders).]
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
Liu et al. (2011) exposed primary rat cortical neuron cultures to RF EMF from a Nokia mobile phone
(GSM/GPRS 900/1800 MHz) in “on” mode. Sham exposures were carried out in the “stand-by” mode. The
mobile phone was placed atop a 6-well culture plate in a tissue culture incubator and the cells were exposed for 2
h, six times a day for 2 days (three independent experiments). The authors reported a significant increase in total
Bax after mobile phone irradiation by immunoprecipitation assay (p<0.001), but no change in total Bax was
observed by Western blot assay (p>0.05). The levels of active-Bax was reported to be significantly higher than in
the sham control in both assays (p<0.01). [There is an inadequate description of the RF exposure system and
dosimetry. Use of a mobile phone in “on” mode as the exposure source does not provide appropriate control of
the exposure level. Moreover, it is questionable whether placing the mobile phone in stand-by mode is an
appropriate sham control, and therefore whether the study fulfilled the inclusion criteria. This study has also
been reported in Sections 5.4 (Brain physiology and function) and 12.3.4 (apoptosis).]
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
Yao et al. (2004) exposed cultured rabbit epithelial cells (RLEC) to 2450 MHz continuous-wave RF
EMF for 4–8 h at power densities ranging from 0.1–2.0 mW/cm2. The authors examined the expression of two
genes related to the cell cycle, namely p27Kip1 and p21WAF1. Western blot analysis (five independent
experiments) demonstrated increased protein expression of p27Kip1 (p<0.05) after 4–8 h exposure at 2.0
mW/cm2, but no changes were observed in p21WAF1 expression. RT-PCR analysis found no significant
difference in the transcript expression of these genes (four independent experiments; see section 12.3.3).
[Temperature measurements indicated a 0.6 °C increase in temperature in the exposed samples. The results of
this study cannot be interpreted since no dosimetric evaluation was performed. This study has also been
described in Sections 6.4 (Auditory function), 12.3.3 (Gene and protein expression) and 12.3.6 (cell
proliferation).]
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
164
4442
4443
4444
4445
4446
4447
4448
4449
4450
Friedman et al. (2007) reported that exposure of Rat1 cells and serum-starved HeLa cells to 875–950
MHz RF EMF at power densities ranging from 0.07–0.31 mW/cm2 for 5–30 min, caused increased
phosphorylation of ERK1/2, but not p38-MAPK or JNK1/2. The authors suggest that this activation is mediated
by ROS generation following activation of NADPH oxidase by RF EMF. [The Western blot data were derived
from only one to three independent experiments for each endpoint. In many cased densitometric analysis of
Western blots was not conducted. Statistical analysis of densitometry data was not performed. There was no
assessment of dosimetry within the samples and the SAR within the mobile phone-exposed cells is unknown.
Moreover, it is not reported how the sham-controls have been performed. This study has been also described in
Section 12.3.3 (Gene and protein expression).]
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
Natarajan et al. (2002) investigated the capability of RF EMF to promote DNA-binding activity of
nuclear factor kappa B (NF-кB), a protein complex involved in cellular responses to several stimuli, including
inflammatory cytokine, chemokines and interferon. They exposed human monocyte Mono-Mac-6 cells to
pulsed-wave radiation used in radar (8.2 GHz, 2.2 μs pulse width and pulse repetition rate of 1000 pulses/s, SAR
= 10.8 ± 7.1 W/kg at the bottom of the culture flask), for 90 min at 37 °C. During exposure, cell cultures were
maintained at 37.4 ± 0.4°C while sham-exposed cultures were kept at 37.2 ± 0.4°C. Cells were then re-incubated
at 37 °C, and harvested 4 h post-exposure. Results showed a 3.6-fold increase in DNA-binding activity of NF-кB
in exposed monocytes compared to the sham exposed ones (two independent experiments carried out in
triplicate). Cell cultures treated with IL-1 were included in the study as positive control and gave the expected
results. The authors also performed experiments aimed to investigate the effect of heating and observed a
decrease in NF-кB DNA-binding activity at 43 °C. [Nevertheless, as also stated by the authors, a broad
distribution of SAR levels in the samples cannot be excluded. The results of this investigation cannot be
interpreted due to the scanty number of experiments; moreover, data are reported as fold-changes and statistical
analysis has not been performed. This study has also been presented in Section 10.3 (Immune system and
haematology).]
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
In a follow-up study, Natarajan et al. (2006) exposed the same cell type (Mono-Mac-6) for 90 min
intermittently (30 min on, 30 min off, 30 min on) to 0.79 ns long pulses with average peak electric field of 1
kV/cm (100 kV/m), pulse repetition rate of 250 Hz and carrier frequency ranging from 0 Hz to 2 GHz. Cells
were incubated and harvested at 10 min, 3 h, 8 h and 24 h post exposure. No difference in the levels of NF-kB
DNA-binding activity was detected in cells harvested at 10 min, 3h and 8 h, while 24 h incubation after
exposure resulted in a 3.5-fold increase in NF-кB-binding activity in exposed cultures compared to sham
controls. Such an increase disappeared at 48 h incubation post exposure. However, the exposure did not
significantly affect the expression of the кB-dependent gene expression profiles, measured at 8 and 24 h post
exposure. In cultures exposed to gamma rays as positive control, positive findings were found. [The validity of
the results remains unclear since the number of independent experiments carried out is not reported. Moreover,
data are reported as fold-changes, although the authors claimed that statistical analysis has been performed. This
study has also been discussed in Section 10.3 (Immune system and haematology)]
Table 12.3.6. In vitro studies assessing effects of RF EMF exposure on signal transduction pathway
Cell type
Biological endpoint
Exposure conditions Results
Comment
Reference
1763 MHz CDMA
For gene and protein
expression see
section 12.3.3
(Lee et al.,
2006)
Number of
independent
experiments
Human TPhosphorylation:
lymphocytesp38MAPK, ERK1/2
derived (Jurkat) and JNK1/2
cells and rat
primary
astrocytes
SAR 2 or 20 W/kg
30 min or 1 h
No effect in the
presence or
absence of TPA
No information on
blinding of staff.
Combined exposures
with TPA
n=3
Human breast
epithelial
(MCF10A) cells
n=3
Phosphorylation and
protein: ERK, JNK,
p38
837 MHz CDMA
No effect.
SAR 4 W/kg
Single or multiple
exposures.
(Kim et al.,
2012)
For gene and protein
expression see
section 12.3.3
4 h;
837 MHz, CDMA, plus
1950 MHz WCDMA
No information on
blinding of staff.
SAR 2+2 W/kg
4 h or 2 h/day for 3
days
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
165
Mouse auditory Phosphorylation and
hair cell-derived protein: ERK1/2, JNK,
(HEI-OC1) cells p38
n=3
1763 MHz CDMA
No effect.
SAR 20 W/kg
15–120 min
For results on auditory (Huang et al.,
system, genotoxicity, 2008b)
gene and protein
expression and
proliferation see
sections 6.4, 12.3.1,
12.3.3 and 12.3.6.
No information on
blinding of staff.
Mouse
mRNA: bax, GADD45, 1710 MHz GSM
embryonic stem Nurr1
Average SAR 1.5
(ES) cells
W/kg
n=5
up to 48 h
Transient
expression
changes.
Human
fibroblast
(WS1neo and
primary cells)
No effect.
Protein: p53
837 MHz CW
For genotoxicity, gene (Nikolova et al.,
and protein
2005)
expression, apoptosis
and proliferation see
sections 12.3.1,
12.3.3, 12.3.4 and
12.3.6.
(Li et al., 1999)
SAR 0.9 or 9 W/kg
2h
n=3
Human breast
cancer derived
(MCF7) cells
Protein: p53, p21,
cyclin (A, B1, D1)
837 MHz CDMA
No effect.
SAR 4 W/kg
(Lee et al.,
2011b)
For cell proliferation
see Section 12.3.6
1 h;
n=3
Single and multiple
exposures.
837 MHz, CDMA, plus
1950 MHz WCDMA
SAR 2+2 W/kg
1h
Human amniotic Protein: p53,
cells (primary)
phosphorylated-p53
n=3
900 MHz GSM
No effect.
Average SAR 0–4
W/kg
24 h
For gene and protein
expression, apoptosis
and cell proliferation
see Sections 12.3.3,
12.3.4 and 12.3.6.
(Bourthoumieu
et al., 2013)
No information on
blinding of staff.
Human amniotic EGF receptor
(FL) cells
clustering and
phosphorylation
n=3
1800 MHz, GSM
Average SAR 0.1–4
W/kg
Increase in EGF
clustering from 0.5
to 4 W/kg.
(Sun et al.,
2012)
15 min
Human
mRNA: variety of
1950 MHz, GSM
peripheral blood cytokine and immune- Average SAR 1 W/kg
mononuclear
relevant genes
8h
cells (PBMC)
(5 min on/10 min off
n=15
cycles)
No effect.
Human
trophoblast
(HTR-8/SVneo)
cells
Increased Cx40 and
Cx43 transcript
expression, no
change in protein
expression.
mRNA and protein:
connexin (40 and 43)
n=5
1817 MHz, GSM
Average SAR 2.0
W/kg
1h
For immune system
see section 10.4
(Tuschl, Novak
& Molla-Djafari,
2006)
(Cervellati et al.,
2009)
“No effect” means no statistically significant effect.
Abbreviations: CDMA: code division multiple access; CW: continuous wave; EGFR: epidermal growth factor receptors; ERK:
extracellular-signal regulated kinase; GSM: Global System for Mobile Communication; JNK: 1/2c-Jun N-terminal kinase; MAPK:
mitogen-activated protein kinase; RT-PCR: Reverse transcriptase-polymerase chain reaction; SAR: specific absorption rate;
TPA: 12-O-tetradecanoylphorbol-13-acetate; W-CDMA: Wideband Code Division Multiple Access
4478
4479
Cytokine expression
4480
4481
4482
Cytokines are signalling molecules (proteins, peptides or glycoproteins) involved in cell signalling.
They are key regulators of cell activation and inhibition and their expression regulates the signals to the immune
system to promote, for example, inflammation. Inappropriate expression of cytokines can induce an immune
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
166
4483
4484
4485
4486
4487
response, and can cause diseases. Studies on the cytokine expression in vitro are quite sparse and have been
appearing only in the very recent years. Among the six recognized studies, five have been carried out on immune
competent cells and have already been reported in Section 10.3. The study by Yoon et al. (2011) has been
already quoted in Section 12.3.2.2 under the studies not included in the analysis, due to methodological issues.
Therefore, all the studies on cytokine expression will only be briefly summarized here.
4488
4489
4490
4491
4492
4493
Tuschl et al. (2006) exposed human lymphocytes, monocytes, and lymphocyte activated killer (LAK)
cells from at least 15 volunteers to RF EMFs at 1950 MHz, GSM modulated, at an SAR of 1 W/kg for 8 hours (5
min on/10 min off cycles) in blind conditions. No differences were detected relative to sham-exposed controls in
the expression of different cytokines (IL-1, IL-2, TNF-α, INF-γ), activity of immune-relevant genes and
cytotoxicity of LAK cells against a tumour cell line. Positive controls were not included in the study design.
[This study is also discussed in Section 10.4 (Immune system and haematology).]
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
Absence of effects on production of several cytokines (TNF-α, IL-1B, IL-6, IL-8, IL-10, IL-12) was
also reported by Chauhan et al. (2007a) in different human-derived cell lines (TK6, HL-60, and Mono-Mac-6)
after 6 h intermittent (5 min on/10 min off cycles) exposure at 1900 MHz pulse-modulated RF EMF at SAR
values of 1 and 10 W/kg. For each cell line, five independent experiments were carried out. Moreover, no
differences compared to sham-exposed cultures were also detected in terms of apoptosis, cell cycle progression
and viability, analysed immediately after the 6-h exposure period and 18 h after exposure. The positive control
samples (heat shock for 1 h at 43 °C) displayed a significant decrease in cell viability, an increase in apoptosis,
and an alteration in cell cycle kinetics (G2/M block). [This study has been also presented in Sections 10.4
(Immune system and haematology), 10.3.3 (Gene and protein expression), 10.3.4 (Apoptosis) and 10.3.6 (cell
proliferation).]
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
In a genomic-wide analysis of gene expression, Huang et al. (2008a) exposed human lymphoblastoid
Jurkat T-cells to a 1763 MHz CDMA signal for 24 h (SAR = 10 W/kg). The experiments were performed
blinded. By comparing exposed and sham exposed cultures, ten genes were identified with a fold-change greater
than 1.3 and, among them, two cytokine receptor genes, chemokine (C-X-C motif) receptor 3 (CXCR3) and
interleukin 1 receptor, type II (IL1R2) were down-regulated, but only the CXCR3 variation was statistically
significant (p<0.05). The results were obtained in five independent experiments. However, these results were not
confirmed by reverse transcriptase-polymerase chain reaction (RT-PCR). The authors also reported that exposure
did not produce significant changes in cell number and cell cycle distributions when assayed 24 h after
exposures. Positive controls were not included in the study design. [This study has been also discussed in
sections 10.4 (Immune system and haematology), 12.3.1 (Genotoxicity) and 12.3.3 (Gene and protein
expression).]
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
Thorlin and co-workers (2006) exposed rat primary astroglial cells to 900 MHz RF EMF. They
applied GSM-modulated EMF at an SAR of 3 W/kg for 4, 8 and 24 h, or CW at 27 W/kg for 24 h. The release
into the extracellular medium of the two pro-inflammatory cytokines IL6 and tumour necrosis factor-alpha
(TNFα) was analysed. Further, the levels of the astroglial cell-specific reactive marker glial fibrillary acidic
protein (Gfap), whose expression dynamics is different from that of cytokines, were measured in astroglial
cultures and in astroglial cell-conditioned culture medium after exposure to CW fields at SARs of 27 and 54
W/kg for 4 or 24 h. Moreover, microglial cell cultures were exposed to 900 MHz, GSM modulated, at an SAR of
3 W/kg for 8 h, and IL6, TNFα, total protein and the microglial reactivity marker ED1 (a macrophage activation
antigen) were measured. No significant differences between EMF and sham-exposed samples were detected for
any of the parameters studied at any time and for any of the exposure conditions tested, as assessed in three to
eight independent experiments performed in blind. Cell cultures incubated at 38° or 42°C were used as positive
controls and gave positive findings. [This study has also been described in Section 8.3 (neurodegenerative
disorders) and 10.4 (Immune system and haematology).]
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
Absence of effects was also reported by Hirose et al. (2010), who exposed primary rat microglial cells
to W-CDMA 1950 MHz EMF at SARs of 0.2, 0.8 and 2.0 W/kg and assessed functional changes in immune
reaction-related molecule expression and cytokine production. The duration of the RF exposure was 2 h and
assay samples were processed 24 and 72 h later in a blind manner. Results showed that the percentage of cells
positive for major histocompatibility complex (MHC) class II, which is the most common marker for activated
microglial cells, did not differ between any of the EMF-exposed groups and the sham-exposed controls.
Furthermore, no remarkable differences in the production of tumour necrosis factor-alpha (TNFα), interleukin-1b
(IL1b), and interleukin-6 (IL6) were observed (three independent experiments). Treatments with
lipopolysaccharide or interferon-Ɣ as positive controls gave positive findings. [The SAR distribution in the
exposed sample was not very homogeneous (standard deviation 57%) and a temperature increase of 0.7 °C was
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
167
4538
4539
recorded during exposure to an SAR of 2.0 W/kg. This study has also been reported in Section 8.3
(Neurodegenerative disorders) and 10.4 (Immune system and haematology).]
4540
Studies not included in the analysis
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
Yoon et al. (2011) exposed cultured human dermal cells (hDPCs) to 1763 MHz RF EMF at SARs of 2
or 10 W/kg. After 3 h exposure at a SAR of 10 W/kg, the authors observed increased transcript expression of
IGF-1 (p<0.05). They also observed increased phosphorylation of MAPK1 and protein expression of BCL-2 and
cyclin D1, but decreased expression of Bax [this data appears to result from one independent experiment]. In a
follow-up experiment, VEGF and IGF-1 transcript expression were increased after only 1 h of RF exposure at 10
W/kg (p<0.05). Similar alterations in IGF-1 transcript expression were not observed in four other cell lines
exposed to the similar RF exposure conditions. [It is unclear whether mRNA was pooled from multiple donors
for transcript analysis, which would effectively result in one independent sample, or whether true independent
experiments were conducted. Since the experimental approach employed is unclear, the significance of these
findings is questionable. This study has also been described in Sections 12.3.4 (Apoptosis), 12.3.5 (Oxidative
stress) and 12.3.6 (cell proliferation).]
Table 12.3.7. In vitro studies assessing effects of RF EMF exposure on cytokines
Cell type
Biological endpoint
Exposure conditions
Results
Comment
Authors
mRNA: variety of
cytokine and immunerelevant genes
1950 MHz, GMS
No effect.
For Immune
system see
Section 10.4.
(Tuschl, Novak &
Molla-Djafari,
2006)
No effects
immediately
after exposure
and 18 h later.
For immune
(Chauhan et al.,
system, gene and 2007a)
protein expression,
apoptosis and cell
cycle see Sections
10.4, 12.3.3,
12.3.4 and 12.3.6.
Downregulation of
CXCR3 and
IL1R2.
For immune
system, DNA
damage and gene
and protein
expression see
Sections 10.4,
12.3.1 and 12.3.3.
Number of
independent
experiments
Human
peripheral
blood
mononuclear
cells
Average SAR 1 W/kg
8h
(5 min on/10 min off
cycles)
n=15–22
Human derived Inflammatory cytokine
immune cell
release (TNF-α, IL-1B,
lines (TK6, HL- IL-6, IL-8, IL-10, IL-12)
60, Mono Mac6)
1900 MHz, Pulse
Modulated
n=5
(5 min on/10 min off
cycles)
Human
Genome-wide analysis
lymphoblastoid of gene expression
Jurkat T cells
1763 MHz, CDMA
n=5
Average SAR 1, 10 W/kg
6h
Average SAR 10 W/kg
24 h
Results not
confirmed by
RT-PCR.
Rat primary
Release of TNF- α and
astroglial and
IL-6; cellular content of
microglial cells Gfap and ED-1
n=3–8
900 MHz, GSM
No effect.
Average SAR 3 W/kg
4, 8 and 24 h
900 MHz CW
n=3–4
No effect.
SAR 27, 54 W/kg
(Huang et al.,
2008a)
For
(Thorlin et al.,
neurodegenerative 2006)
disorders and
immune system
see Sections 8.3
and 10.4.
24 h
Primary
microglial cell
cultures
Changes in immune
reaction-related
molecule expression
1950 MHz, W-CDMA
n=3
Cytokine production
(TNF-alpha, IL-1beta,
IL-6)
2h
Average SAR 0.2, 0.8, 2
W/kg
No effects 24
and 48 h after
exposure.
SAR not
homogeneous.
(Hirose et al.,
2010)
For
neurodegenerative
disorders and
immune system
see Sections 8.3
and 10.4.
No effect” means no statistically significant effect.
Abbreviations: CDMA: code division multiple access; GSM: Global System for Mobile Communication; IL: interleukin; RT-PCR:
Reverse transcriptase-polymerase chain reaction; SAR: specific absorption rate; TNFα: tumor necrosis factor α.
4552
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
168
4553
12.3.3
Gene and protein expression
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
In the previous WHO report (WHO, 1993b) no studies on gene and protein expression were reported.
The present literature search identified 72 papers on this topic. Nine studies did not meet the inclusion criteria for
in vitro studies and were not considered, but references are reported at the end of the chapter. Among the
remaining 63 papers, 19 do not completely comply with the quality criteria for inclusion due to methodological
issues (inadequate description of exposure system and dosimetry, small number of experiments, lack or
inadequate statistical analysis) and are only presented in the text. The 44 included studies deal with heat shock
proteins (Section 12.3.3.1), proto-oncogenes (Section 12.3.3.2), Ornithine Decarboxylase (ODC) activity
(Section 12.3.3.3) and the use of high-throughput genomics and proteomics technologies (Section 12.3.3.4). In
some cases combined exposures with chemical agents have also been carried out. All studies reported below
included sham exposed cultures
4564
4565
In this draft is not yet reported if studies have been carried out blinded. Moreover no information are
given on positive controls included in the study design.
4566
12.3.3.1
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
One group of proteins that have been reported to be affected by RF EMFs are the “stress proteins”,
also known as the heat shock proteins (HSP). HSPs are a family of chaperone proteins that are found in all celltypes, are highly conserved and abundantly expressed with diverse functions. They are expressed in response to
cold, heat and other environmental stresses, although some are constitutively expressed. HSPs increase heat
tolerance and perform functions essential to cell survival under these conditions. Some HSPs serve to stabilize
proteins in abnormal configurations, while others play a role in the folding and unfolding of proteins, acting as
molecular chaperones. Stress-induced transcription of HSPs requires activation of heat shock factors that bind to
the heat shock promoter element, thereby activating its transcription activity. As HSPs and their associated
factors are induced by a variety of stressors, HSP gene and protein expression have been extensively investigated
in response to RF exposure in a variety of cell models. Such studies have explored both the protein and transcript
expression of a variety of HSP, but also the phosphorylation status of these proteins.
4578
4579
4580
4581
4582
4583
In a follow-up study of Leszczynski et al. (2002) (described under Studies not included in the
analysis), Nylund et al. (2009) exposed a human endothelial cell line (EA.hy926) to 1800 MHz GSM-modulated
RF EMF for 1 h at a SAR of 2.0 W/kg. The authors reported 8 differentially expressed proteins (p<0.05) by 2Dgel electrophoresis. No statistically significant changes were observed on HSP27 (or other HSP) or vimentin
protein expression, however, phosphorylation status was not assessed. These results are based on 5 independent
experiments.
4584
4585
4586
4587
4588
4589
4590
Czyz et al. (2004) reported that exposure of mouse embryonic stem (ES) cells to 1710 MHz GSMmodulated RF EMF at SAR values of 1.5 W/kg for 48 h and 2.0 W/kg for 72 h could induce a significant
(p<0.05) and stable up-regulation in transcript levels of HSP70 in p53-deficient cells, but not in wild-type cells.
No effects were observed when the 1710 MHz signal was modulated using a “GSM-talk” paradigm at the same
slot-averaged SAR values (6 independent experiments). The authors speculated that certain signal characteristics
(e.g. 217 Hz modulation of the carrier signal) and biological/genetic conditions (e.g. p53 function) may be
important for the detection of RF-related cellular responses.
4591
4592
4593
4594
4595
4596
4597
4598
4599
Valbonesi et al. (2008) observed no changes in HSP70 protein or transcript expression in a human
trophoblast cell line (HTR-8/neo) following a 1 h exposure to 1800 MHz RF EMF at a SAR of 2 W/kg (6
independent experiments). In a follow-up study, Franzellitti et al. (2008) examined HSP70 gene and protein
expression in HTR-8/SVneo cells after prolonged exposure (4–24 h) to 1800 MHz, CW or GSM-modulated
(GSM-talk and GSM-217 Hz). Protein expression of HSP70 and HSC70 were unchanged after RF exposure.
Similarly, the transcript expression of HSP70A, HSP70B and HSC70 was unchanged. However, the transcript
expression of HSP70C varied across the test conditions. HSP70C transcript expression was unchanged after CW
exposure, but was decreased after 4–16 h exposure to GSM-talk and increased at 24 h after GSM-217 Hz
exposure (6 independent experiments; p<0.05).
4600
4601
4602
4603
Vanderwaal et al. (2006) exposed a series of cell lines (HeLa, S3 and EA.hy926) to either 837 MHz
Time Division Multiple Access (TDMA)-modulated RF EMF for 1, 2 or 24 h at a SAR of 5 W/kg or 900 MHz
GSM-modulated RF for 1, 2 or 5 h at a SAR of 3.7 W/kg and used Western blots to assess the expression of
phosphorylated-HSP27. In 3 independent experiments for each cell line, the authors found no evidence that RF
Heat Shock Proteins (HSP)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
169
4604
4605
exposure altered the expression of phosphorylated-HSP27 in any of these cell lines under any conditions tested,
while heat-shock at 45 °C was found to markedly increase the expression of phosphorylated-HSP27.
4606
4607
4608
4609
4610
4611
Lee et al. (2006) exposed human T-lymphocyte-derived Jurkat cells and rat primary astrocytes to 1763
MHz Code Division Multiple Access (CDMA)-modulated RF EMF at SARs of 2 or 20 W/kg for up to 1 h. The
authors reported no significant differences in the protein expression of HSP27, HSP70 or HSP90 in either cell
type (3 independent experiments). Similarly, no alterations in the expression of phosphorylated-HSP27, HSP70
or HSP90 were observed in rat primary astrocytes after RF exposure. [This study has already been described in
section 12.3.2 (Intracellular and intercellular signalling).]
4612
4613
4614
4615
4616
Kim et al. (2012) exposed human mammary (MCF10A) cells to 837 MHz and/or 1950 MHz RF EMF
for 2 to 4 h at a SAR of 2 to 4 W/kg. The expression levels and phosphorylation states of several HSPs (HSP27,
HSP70, HSP90) were analyzed by Western blot. In three independent experiments the authors found no evidence
that single or repeated exposure of these cells to RF could elicit expression or phosphorylation changes of any
HSPs. [This study has been already described in Section 12.3.2 (Intracellular and intercellular signalling).]
4617
4618
4619
4620
Miyakoshi et al. (2005) exposed human malignant glioma cells (MO54) to 1950 MHz CW RF-EMF
for up to 2 h at SARs of 1, 2, and 10 W/kg. Exposed cells (from 3 to 6 independent experiments) did not show
increased HSP27 or HSP70 protein expression. However, cells exposed at a SAR of 10 W/kg for 1–2 h exhibited
a significant decrease in the level of phosphorylated-HSP27 relative to the sham-exposed controls (p<0.05).
4621
4622
4623
4624
Wang et al. (2006) exposed A172 cells to relatively high intensity 2450 MHz at SARs ranging from 5
to 200 W/kg for 1–3 h. In 3 independent experiments the authors detected significantly increased expression of
HSP70 and increased expression of phosphorylated-HSP27 at SARs greater than 50 W/kg (p<0.01). [These
results are most likely attributable to thermal effects.]
4625
4626
4627
4628
Huang et al. (2008b) exposed a mouse auditory hair cell line (HEI-OC1) to 1763 MHz CDMAmodulated RF EMF at a SAR of 20 W/kg for 6–24 h. No changes in the protein expression levels of HSP27,
HSP70 or HSP90 were observed following 6, 12 or 24 h RF exposure (3 independent experiments). [This study
has been already quoted in section 12.3.2 (Intracellular and intercellular signalling).]
4629
4630
4631
4632
Lantow et al. (2006a) exposed human umbilical cord blood-derived monocytes and lymphocytes to
1800 MHz CW or GSM-modulated RF EMF at 2 W/kg for 1 h. No effects on the protein expression of HSP70
were observed in the exposed monocytes at 0, 1 or 2 h post-exposure, relative to the sham controls, as assessed
in 3 independent experiments, under any RF exposure condition tested.
4633
4634
4635
4636
4637
4638
Similar results were observed in a series of studies by another research group. Chauhan et al. (2006a;
2006b; 2007b) and Qutob et al. (2006) exposed a variety of human-derived cell lines (Mono-Mac-6, U87MG,
HL60, TK6) to 1900 MHz pulse-modulated RF EMF for 4 to 24 h at SARs ranging from 0.1 to 10 W/kg. In each
of these studies, conducted by performing 5 independent experiments, concurrent sham and heat-shock positive
controls were included. The authors found no evidence of altered transcript expression for HSP27, HSP40,
HSP70, HSP90 or HSP105 in any of these cell lines following any RF condition tested.
4639
4640
4641
4642
4643
Several studies have evaluated alterations in HSP expression or activating in primary human cells.
Capri et al. (2004a) exposed human peripheral blood mononuclear cells (PBMC) from 6 young and old donors to
three different GSM-modulations of 1800 MHz RF EMF for 44 h (10 min on, 20 min off) at SARs of 1.4 and 2.0
W/kg. Using flow cytometry, the extent of fluorescence-labelling of HSP70 on exposed PBMC was not found to
be significantly different than that of sham-exposed PBMC for any of the conditions tested.
4644
4645
4646
4647
Lim et al. (2005) exposed diluted (1:1) human peripheral whole blood to 900 MHz CW or GSMmodulated RF EMF at SARs of 0.4 to 3.6 W/kg for times ranging from 20 min to 4 h. No statistically significant
differences were detected in the percentage of lymphocytes or monocytes expressing elevated HSP27 or HSP70
after any RF exposure conditions. The data are based upon three independent experiments.
4648
4649
4650
4651
4652
4653
Sanchez et al. (2006b) exposed isolated human skin cells (4 to 5 independent experiments) and
reconstructed human epidermis (4 to 7 independent experiments) to 900 MHz GSM-modulated RF-EMF at a
SAR of 2 W/kg for 48 h. Microscopy analysis of fluorescently-labelled HSC70, HSP27 and HSP70
demonstrated no detectable changes in the protein expression of HSC70, HSP27 or HSP70 in keratinocytes
following RF exposure. However, HSC70 protein expression was reported to be significantly decreased in
fibroblasts after exposure (p<0.05). Similar changes in HSC70 expression were not observed in normal human
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
170
4654
4655
4656
4657
4658
4659
4660
4661
4662
fibroblasts purchased from a commercial supplier following similar RFR conditions. The authors reported a
slight, but significant, increase in HSP70 expression in reconstructed human epidermis cells at 3–5 weeks after
RF exposure, relative to the sham controls (p<0.05). However, this observation was based upon declining HSP70
expression in the sham group over time, rather than an increase in HSP70 expression in the exposed group over
time. In a follow-up study, the same investigators found that primary human skin cells (keratinocytes and
fibroblasts) did not display any alterations in HSP27, HSP70 or HSC70 protein expression following a 48 h
exposure to 1800 MHz GSM-modulated RF (Sanchez et al., 2007). This study had 8 independent experiments
for fibroblasts and 3–6 independent experiments for fibroblast cells. [The authors did not speculate on the
differential responses observed between these two studies.]
4663
4664
4665
4666
4667
4668
4669
Terro et al. (2012) exposed cultured cerebral cortical cells to 900 MHz GSM-modulated RF EMF for
24 h at a SAR of 0.25 W/kg. Protein expression of HSC70 was increased by ~26% (p<0.01) and HSP90 was
decreased by ~10% (3 independent experiments; p<0.05), however similar changes were observed to occur after
relatively modest temperature alterations of 0.3-0.5 °C in these experiments. [Since the temperature within the
RF exposed cultures increased by ~0.5 °C during the 24 h exposure period relative to the sham controls, it is
likely that the observed changes on HSP expression are due to temperature elevations in the RF-exposed
samples.]
4670
4671
4672
4673
4674
4675
Laszlo et al. (2005) exposed hamster (HA-1), mouse (C3H 10T1/2) and human (HeLa, S3) cell
cultures to 835.62 MHz or 847.74 MHz frequency domain multiple access (FDMA) or CDMA-modulated RF
EMF at SARs of ~0.6 W/kg and ~5 W/kg for 5 min to 24 h. The heat shock factor (HSF) DNA-binding activity
was examined using a 32P-labelled HSF gel shift assay. The authors reported no evidence of altered in HSF
DNA-binding ability in response to RF exposure in any of the cell lines for any of the exposure conditions (3
independent experiments).
4676
4677
Two studies have assessed the effect of millimeter wave RF EMF exposures on HSP expression in
cultured cell lines.
4678
4679
4680
4681
Zhadobov et al. (2007) exposed a human glial cell line (U-251MG) to 60 GHz RF-EMF for 33 h at
power densities of 5.4 μW/cm2 or 0.54 mW/cm2.In 3 to 7 independent experiments, the authors found no
evidence that low level millimeter-wave exposure altered the transcript or protein expression of clusterin or
HSP70.
4682
4683
4684
4685
Nicolaz et al. (2009) assessed the effect of millimeter wave (59–61 GHz) on endoplasmic reticulum
stress-responsive chaperone proteins. After 24 h exposure at SARs of 2.64–3.30 W/kg, the transcript levels of
BiP/GRP78, ORP/GRP170 and HSP70 were assessed by RT-PCR and found to be unaffected by RF exposure (3
to 4 independent experiments).
4686
4687
4688
4689
4690
4691
4692
4693
4694
Only two studies have assessed the effect of RF EMF exposure combined with another agent on HSP
protein expression. Simkó et al. (2006) exposed a human monocyte-derived cell line (Mono-Mac-6) to 1800
MHz CW or GSM-modulated RF EMF for 1 h at a SAR of 2 W/kg, either alone or in conjunction with ultra-fine
particles. The authors reported that RF exposure alone or co-exposure with ultra-fine particles had no effect on
altering HSP70 protein expression (3 to 6 independent experiments). In a follow-up study, Lantow et al. (2006c)
investigated whether 1800 MHz CW or GSM-modulated RF exposure at SARs of 0.5 to 2.0 W/kg for 45 min,
either alone or in the presence of phorbol 12-myristate 13-acetate (PMA), could cause altered expression of
HSP70 in Mono-Mac-6 and K562 cells. No significant effects were detected in HSP70 protein expression in
either cell-line from RF exposure, under any of the conditions tested (4 to 6 independent experiments).
4695
Studies not included in the analysis
4696
4697
4698
4699
Cleary et al. (1997) exposed HeLa cells and CHO cells to 27 MHz or 2450 CW MHz RFR at a SAR
of 25 or 100 W/kg under isothermal exposure conditions (37 ± 0.2 °C) for 2 h. The authors reported no evidence
of altered stress protein induction was observed in either cell line under any of the RFR exposure conditions
tested (2 independent experiments). [No statistical analysis was performed on these results.]
4700
4701
4702
4703
4704
Leszczynski et al. (2002) exposed a human endothelial cell line (EA.hy926) to 900 MHz global
system for mobile communication (GSM)-modulated RFR for 1 h at an average SAR of 2 W/kg. The authors
reported altered phosphorylation status for several proteins (4 independent experiments). Specifically, HSP27
was reported to undergo a transient increase in expression and phosphorylation immediately after exposure. At 8
h after exposure, decreased expression of HSP27 was observed. [However, none of these observations were
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
171
4705
4706
subjected to statistical analysis. This paper has been also described in Sections 5.4.1 (Brain physiology and
function) and in 12.3.2 (Intracellular and intercellular signalling).]
4707
4708
4709
4710
4711
4712
4713
Tian et al. (2002) exposed human glioma (MO54) cells to 2450 MHz RF EMF at a SAR of 5–100
W/kg for exposure times of 2–16 h. Significant increases in HSP70 protein levels were observed at SARs
ranging from 25 to 100 W/kg (p<0.05), but no effect on HSP70 protein expression was observed at SARs below
20 W/kg. [Thermal confounding cannot be ruled out in this study at the higher SARs tested due to the highly
non-uniform SAR distributions within the exposure system and the extreme loss of cell viability (~70%)
observed among some samples during RF exposure. Moreover, the number of independent experiments is not
reported and statistical analysis was not performed on much of the data.]
4714
4715
4716
4717
4718
Lixia et al. (2006) exposed human lens epithelial cells (hLEC) to 1800 MHz GSM-modulated RF
EMF for 2 h at SARs of 1–3 W/kg. The authors noted increased HSP70 protein expression at higher doses
(p<0.05), but no corresponding change was observed in mRNA expression. [The number of independent
experiments is not reported although statistical analysis was performed. This study has been described in detail
in Section 12.3.1 (genotoxicity).]
4719
4720
4721
4722
4723
4724
Hirose et al. (2007) examined the effect of RF exposure on HSP27 phosphorylation, transcript and
protein expression in two cell lines. Human glioblastoma-derived A172 cells and human fetal lung-derived IMR90 fibroblasts were exposed for 2–48 h to 2142.5 MHz CW or wideband-CDMA modulated RF EMF at SARs of
0.08 to 0.8 W/kg. No evidence of altered HSP27-phosphorylation or transcript expression of a variety of HSPs
was observed in either cell line under any conditions tested. [Only two independent experiments from each cell
line were conducted in this study.]
4725
4726
4727
4728
4729
Gurisik et al. (2006) exposed a human glioblastoma cell line (U937) to 900 MHz GSM-modulated RF
EMF for 2 h at a SAR of 0.2 W/kg. The authors found no evidence of altered HSP60, HSP70 or HSP90 protein
levels at 2 h post exposure (4 independent experiments). [However, the authors also reported that these cell
cultures demonstrated low viability (~65%) in both the sham and RF-exposed samples at 24 h post-exposure.
Furthermore, no statistical analysis was performed on Western blot data.]
4730
4731
4732
4733
4734
4735
4736
Ding et al. (2009) exposed three human glioma cell lines (MO54, A172, and T98) to 1950 MHz CW
RF EMF for 1 h at a SAR of 1 or 10 W/kg. The authors reported that RF field exposure did not alter the
distribution or expression of HSP27, or the protein expression of HSP27 or HSP70 under any of the RF exposure
conditions or cell lines tested (2 independent experiments). However, a significant decrease in p-HSP27 was
observed in MO54 cells after a 1 h exposure to RF EMF at 10 W/kg (p < 0.05). [There is no description of the
RF exposure system or dosimetry. The results are based upon an inadequate number of independent
experiments.]
Table 12.3.8. In vitro studies assessing the effect of RF fields on Heat Shock Proteins (HSP) in cell lines
Cells
Biological endpoint
Exposure conditions
Results
Comment
Authors
protein: HSP27
1800 MHz, GSM
No effect.
Follow up study to Nylund et al.
Leszczynski et al. (2009)
(2002)
Number of
independent
experiments
Human umbilical
vein (EA.hy926)
cells
Average SAR 2 W/kg
1h
n=5
Mouse embryonic
stem (ES) cells
mRNA: HSP70
n=6
Phosphorylation
not assessed.
1710 MHz, GSM
Average SAR 1.5 W/kg
48 h
Average SAR 2.0 W/kg
Increased
For ProtoHSP70 transcript oncogenes see
expression in
12.3.3.2
p53-deficient
cells.
Czyz et al. (2004)
No effect.
Valbonesi et al.
(2008)
72 h
Human
mRNA and protein:
trophoblast-derived HSP70 and HSC70
cell line (HTR-8/SV
neo)
1817 MHz, GSM
Average SAR 2 W/kg
1h
n=6
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
172
Human
protein and mRNA:
trophoblast-derived HSP70 (A, B, C) and
cell line (HTR-8/SV HSC70
neo)
n=6
1800 MHz, CW or GSM No effect on
protein or
SAR 2 W/kg
transcript
4–24 h
expression for
HSP70(A and
B) and HSC70.
Follow up study to Franziletti et al.
Valbonesi et al.
(2008)
(2008)
HSP70C had
varied transcript
expression in
response to
1800 MHz GSMtalk and GSM217 Hz.
Human cervical
cancer-derived
(HeLa S3) cells,
human umbilical
vein (EA.hy926)
cell lines
protein: p-HSP27
No effect.
Vanderwaal et al.
(2006)
900 MHz, GSM
SAR 5 W/kg
1, 2 or 24 h
Average SAR 3.7 W/kg
n=3
Human Tlymphocytederived (Jurkat)
cells, rat primary
astrocytes
837 MHz, TDMA;
1, 2 or 5 h
protein: HSP27,
HSP70 and HSP90
1763 MHz, CDMA
No effect.
For signal
transduction see
Section 12.3.3.
Lee et al. (2006)
No effect.
Single and
repeated
exposures.
Kim et al. (2012)
SAR 2 or 20 W/kg
30 min or 1 h
n=3
Human mammary
epithelial
(MCF10A) cells
n=3
Human gliomaderived (MO54)
cells
phosphorylation and
837 MHz, CDMA;
protein: HSPs (27, 70, 1950 MHz W-CDMA
90)
SAR 2–4 W/kg
protein: HSP27,
HSP70 and p-HSP27
1950 MHz, CW
SAR 1, 2, 10 W/kg
10–30 min, 1 or 2 h
n=3 or 6
Human
glioblastomaderived (A172)
cells
protein: HSP27,
HSP70 an p-HSP27
2450 MHz
SAR 5–200 W/kg
1–3 h
protein: HSP27,
HSP70, HSP90
For cell
proliferation see
Section 12.3.6.
Miyakoshi et al.
(2005)
No effect at
SARs less than
50 W/kg.
Thermal
confounding is
possible at SAR
>50 W/kg.
Wang et al.
(2006)
1763 MHz, CDMA
No effect.
SAR 20 W/kg
6–24 h
n=3
Human umbilical
cord blood
(primary)
monocytes and
lymphocytes.
No effect on
HSP27 or
HSP70 protein
expression.
Decrease in pHSP27 at 10
W/kg.
Increased
expression of
HSP70 and pHSP27 at SARs
> 50 W/kg.
n=3
Mouse auditory
hair cell-derived
(HEI-OC1) cells
For signal
transduction see
Section 12.3.2.
2–4 h
protein: HSP70
1800 MHz, GSM
Cell viability also
evaluated.
For signal
transduction see
Section 12.3.2.
No effect.
Huang et al.
(2008a)
Lantow et al.
(2006a)
Average SAR 2 W/kg
1h
n=3
Human
lymphoblastoidderived (TK6) cells
n=5
mRNA: HSP27,
HSP70
1900 MHz, PW
No effect.
SAR 1, 10 W/kg
6 h (5 min on/10 min off
cycles)
For Protooncogenes see
Section 12.3.3.2.
Chauhan et al.
(2006a)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
173
Human
mRNA: HSP27,
promyelocytic
HSP70
leukemia-derived
(HL60) cells,
human monocytederived (Mono-Mac
6) cells
1900 MHz PW
No effect.
For Protooncogenes see
Section 12.3.3.2.
Chauhan et al.
(2006b)
No effect.
For Protooncogenes see
Section 12.3.3.2.
Chauhan et al.
(2007b)
SAR 1, 10 W/kg
6 h (5 min on/10 min off
cycles)
n=5
Human
mRNA: HSP27,
glioblastomaHSP40, HSP70,
derived (U87MG)
HSP86 and HSP105
cells, human
monocyte-derived
(Mono-Mac 6) cells
1900 MHz, PW
SAR 0.1, 1, 10 W/kg
6–24 h (5 min on/10
min off cycles)
n=5
Human
glioblastomaderived (U87MG)
cells
mRNA: HSP27,
HSP40, HSP70,
HSP86 and HSP105
1900 MHz PW
No effect.
Qutob et al.
(2006)
No effect.
Capri et al.
(2004a)
No effect.
Lim et al. (2005)
No effect on
keratinocytes.
Sanchez et al.
(2006b)
SAR 0.1, 1, 10 W/kg
4h
n=5
Human peripheral protein: HSP70
blood mononuclear
cells (PBMC)
1800 MHz GSM
n=6
44 h (10 min on/20 min
off)
Human peripheral
whole blood
protein: HSP27 and
HSP70
n=3
Average SAR 1.4 , 2
W/kg
900 MHz, CW or GSM
Average SAR 0.4, 2,
3.6 W/kg
20 min, 1, 4 h
Human primary
keratinocytes,
fibroblasts,
reconstructed
human epidermis
protein: HSP27,
HSC70 and HSP70
900 MHz, GSM
Average SAR 2 W/kg
48 h
Significant
decrease in
HSC70 in
fibroblasts.
1800 MHz, GSM
No effect.
n=4–5
(keratinocytes)
n=4–7
(reconstructed
epidermis)
Human primary
keratinocytes and
fibroblasts
protein: HSP27,
HSP70 and HSC70
Average SAR 2 W/kg
48 h
n=8 (keratinocytes)
protein: HSC70,
HSP90
900 MHz, GSM
HSC70
expression
increased.
Average SAR 0.25
W/kg
HSF protein-DNA
binding activity
0.5 °C
Terro et al.
temperature
(2012)
difference
between sham/RF
cultures.
24 h
HSP90
expression
decreased.
835.62 MHz, FDMA;
847.74 MHz, CDMA
No effect.
Lazlo et al.
(2005)
No effect.
Zhabodov et al.
(2007)
n=3
Human cervical
cancer-derived
(HeLa S3) cells,
hamster ovary
(HA-1) cells,
mouse embryo
fibroblast derived
(C3H 10T1/2) cells
Sanchez et al.
(2007)
Results differ from
2006 study.
n=3–6 (fibroblasts)
Cultured primary
cerebral cortical
cells from
embryonic Wistar
rats.
Follow up to
Sanchez et al.
(2006b)
SAR 0.6 - 5 W/kg
5–60 min; 24 h
n=3
Human
glioblastomaderived (U-251
MG) cells
mRNA and protein:
HSP70 and clusterin
60 GHz, CW
5.4, 0.54 mW/cm2
1–33 h
n= 3–6
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
174
Human
astrocytomaderived (U-251)
cells
mRNA: BiP/GRP78,
ORP/GRP170 and
HSP70
59–61 GHz
No effect.
Nicolaz et al.
(2009)
SAR 2.64–3.3 W/kg
24 h
n= 3–4
Studies including co-exposures
Human monocyte- protein: HSP70
derived (Mono-Mac
6) cells
n=3–6
1800 MHz, CW or GSM No effect.
Simkó et al.
(2006)
SAR 2 W/kg
1h
Co-exposures with
ultra-fine particles.
Human monocyte- protein: HSP70
derived (Mono-Mac
6) cells, human
myelogenous
leukemia-derived
(K562) cells
1800 MHz, CW or GSM No effect.
SAR 0.5, 1, 1.5, 2 W/kg
45 min
Follow-up to
Simkó et al.
(2006)
Lantow et al.
(2006c)
Co-exposures with
PMA.
n=4–6
“No effect” means no statistically significant effect.
4737
4738
12.3.3.2
4739
4740
4741
4742
Several studies have investigated the ability of RF-EMF to mediate the expression of protooncogenes. Proto-oncogenes are genes whose altered expression has the capability to induce cellular
proliferation and/or transformation. While these genes are constitutively expressed at low levels, they are rapidly
and transiently induced in response to external stressful stimuli.
4743
4744
4745
4746
4747
Ivaschuk et al. (1997) exposed nerve growth factor (NGF)-stimulated rat PC12 pheochromocytoma
cells to 836.55 MHz TDMA-modulated intermittent (20 min on, 20 min off) RF for 20 to 60 min at SARs of
0.5–4.6 mW/kg. Northern blot analysis, performed on 5 independent experiments, indicated no change in c-fos
expression after exposure, however c-jun expression was significantly decreased after 20 min exposure at 4.6
mW/kg (p<0.05), but not at any of the other exposure intensities or times after exposure.
4748
4749
4750
4751
4752
4753
4754
4755
Goswami et al. (1999) reported that exposure of C3H 10T1/2 cells to 835–850 MHz modulated RFEMF at an average SAR of 0.6 W/kg significantly increased c-fos transcript expression (p<0.001 for exponential
growth phase and transition phase cells, p<0.04 for plateau phase cells). No changes were observed in the
transcript expression levels of c-jun and c-myc or in the DNA binding activity of the transcription factors AP1,
AP2 or NF-κB (3 independent experiments). In a follow-up study, Whitehead et al. (2005) reassessed the
potential effect of exposure of these cells to 835-850 MHz modulated RF at SARs of 5 or 10 W/kg on c-fos
expression using RT-PCR (three independent experiments). The authors did not confirm their earlier
observations, although higher SARs were used in the follow-up study.
4756
4757
4758
4759
4760
Czyz et al. (2004)
modulated RF-EMF caused a
increase in c-myc expression
changes in the expression of
cells.
4761
4762
4763
4764
Chauhan et al. (2006a; 2006b) exposed several human-derived cell lines to intermittent (5 min on/10
min off) 1950 MHz RF-EMF at SARs of 1 or 10 W/kg for 6–24 h and proto-oncogene transcript expression was
assessed by RT-PCR. In 5 independent experiments no significant differences were observed on the relative
expression level of the proto-oncogenes c-jun, c-fos and c-myc in any of the cell lines examined.
4765
4766
4767
4768
4769
Merola et al. (2006) exposed human neuroblastoma cell (LAN-5) to 900 MHz GSM modulated RFEMF at an average SAR of 1.0 W/kg for 24 to 72 h. Western blot analysis of B-myb and N-myc, sensitive
markers of proliferation and differentiation, indicated no effect of RF exposure on the expression of these
oncogenes, either in the presence or absence of retinoic acid or camptothecin stimulation, as assessed in 3
independent experiments.
Proto-oncogene expression
reported that p53-deficient embryonic stem cells exposed to 1710 MHz GSMtransient increase in c-jun expression immediately after a 48 h exposure and an
at 5 days after exposure (p<0.05), in 6 independent experiments. However, no
these proto-oncogenes were detected in the wild-type exposed embryonic stem
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
175
Table 12.3.9. In vitro studies assessing the effect of RF fields on the expression of proto-oncogenes
Cells
Biological endpoint Exposure conditions Results
Comment
Authors
Number of
independent
experiments
Rat adrenal medulla mRNA: c-fos, c-jun
pheochromocytomaderived (PC12)
cells
- NGF-stimulated
836.55 MHz, TDMA
SAR 0.5–4.6 mW/kg
20–60 min (20 min
on/20 min off)
– serum deprived
protein, mRNA: cfos, c-jun, c-myc
835.62 MHz, FMCW
847.74 MHz, CDMA
DNA binding: AP1,
AP2, NF-κB
Average SAR 0.6
W/kg
n=3
mRNA: c-fos
835.62 MHz, FDMA
836.55 MHz, TDMA
847.74 MHz, CDMA
– serum deprived
SAR 5 or 10 W/kg
n=3
0.5–2 h
Mouse embryonic
stem (ES) cells
mRNA: c-jun, c-myc
1710 MHz, GSM
SAR 2.0 W/kg for 72
h
Human
lymphoblastoidderived (TK6) cells
mRNA: c-fos, c-myc
and c-jun
Goswami et al.
(1999)
No effect.
Follow-up to
Goswami et
al. (1999)
Whitehead et al.
(2005)
Higher SARs
than original
study.
SAR 1.5 W/kg for 48
h
n=6
Increased c-fos
transcript expression
after RFR exposure.
No effect on c-myc or
c-jun.
24 h
Mouse embryo
fibroblast-derived
(C3H10T1/2) cells
Ivaschuk et al.
(1997)
No effect on c-fos or
c-jun at other
exposure conditions.
n=5
Mouse embryo
fibroblast-derived
(C3H10T1/2) cells
Decrease in c-jun
expression after 20
min exposure (4.6
mW/kg).
1900 MHz PW
Increased c-jun, cmyc and p21
expression in p53deficient cells at
various timepoints.
Similar results
not observed
in p53competent
cells.
Czyz et al. (2004)
For HSP see
12.3.3.1.
No effect.
For HSP see
12.3.3.1.
Chauhan et al.
(2006a)
No effect.
For HSP see
12.3.3.1.
Chauhan et al.
(2006b)
No effect.
For Apoptosis
see Section
12.3.4.
Merola et al.
(2006)
SAR 1, 10 W/kg
6 h (5 minon/10 min
off)
n=5
Human
mRNA: c-fos, c-myc
promyelocytic
and c-jun
leukemia-derived
(HL60) cells, human
monocyte-derived
(Mono-Mac 6) cells
1900 MHz, PW
SAR 1, 10 W/kg
6 h (5 min on/10 min
off)
n=5
Studies including co-exposures
Human
neuroblastomaderived (LAN-5)
cells
n=3
protein: B-myb, Nmyc
900 MHz GSM
Average SAR 1 W/kg
24–72 hCo-exposures
with retinoic acid and
camptothecin.
“No effect” means no statistically significant effect
4770
4771
12.3.3.3
4772
4773
4774
ODC is the rate-limiting enzyme in polyamine synthesis and thereby acts to regulate cell proliferation.
Alterations in ODC activity have been linked with both uncontrolled growth of malignant cells and reduced
apoptosis.
4775
4776
4777
4778
4779
Desta et al. (2003) investigated the ability of 835 MHz TDMA-modulated RF EMF at SARs of <1 to
15 W/kg for 8 h to modulate ODC activity in L929 cells. This study found no evidence of altered ODC activity
at RF exposures up to 5 W/kg that did not induce sample heating (3 to 6 independent experiments). However
decreased ODC activity was observed in RF and non-RF exposed samples where sample temperatures increased
by 1.5-2.0 °C (p<0.05).
Ornithine Decarboxylase (ODC) Activity
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
176
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
Höytö et al. (2006) exposed L929 murine fibroblasts to 900 MHz CW or GSM-modulated RF-EMF at
SARs of 0.2–0.4 W/kg for 2 to 24 h and assessed the impact of RF exposure on ODC activity. The authors in 3
to 5 independent experiments found no evidence of RF-induced alterations in ODC activity but noted that small
increases in temperature (<1 °C) caused decreased ODC activity. The authors commented that ODC activity is
remarkably temperature-sensitive and that precise temperature control is essential in studies on cellular ODC
activity. In a follow-up study, Höytö et al. (2007a) exposed L929 fibroblasts from 2 different cell repositories to
835 and 872 MHz CW and TDMA-modulated (50 Hz burst modulated, 33% duty cycle) RF at SARs of 2.5 or
6.0 W/kg for 2–24 h. Exposures were conducted using a Crawford exposure apparatus and a waveguide exposure
system (3 to 7 independent experiments). Using the Crawford exposure system, ODC activity was unaffected by
RF exposure except for a slight decrease after 2 h exposure at a SAR of 6 W/kg (p<0.05), which the authors
attributed to a thermal effect. When L929 cells were exposed using the waveguide system, ODC activity was
increased after 8 h of exposure at 6 W/kg relative to the sham-exposed samples (p<0.05). No evidence of
amplitude modulation-specific effects were observed. The authors speculated that the use of adjusted water-bath
temperatures for different SARs (to accommodate for RF-induced heating of the sample at 6 W/kg) using the
waveguide system may have decreased the basal ODC activity which was then increased as RF-exposure
‘heated’ the sample to 37 °C. In addition to L929 fibroblasts, Höytö et al. (2007b) assessed the ability of 872
MHz RF-EMF to modulate ODC activity in rat C6 glioblastoma cells, human-derived SH-SY5Y neuroblastoma
cells and rat primary astrocytes. The cell lines were exposed to 872 MHz CW and GSM-modulated RF at SARs
of 1.5, 2.5 and 6.0 W/kg, in a water-jacketed waveguide for 2 to 24 h. ODC activity was reported to be
significantly decreased in rat primary astrocytes at SARs of 1.5 and 6.0 W/kg with both CW and GSMmodulation (p<0.005). The authors reported no consistent effects on ODC activity in the other cell lines (5 to 12
independent experiments). In attempt to understand the variability in ODC responses among and within studies,
Höytö et al. (2008b) investigated the influence of a variety of physiological conditions on ODC activity. Serumstarved and serum-enriched cultures of L929 cells were exposed to 872 MHz CW and GSM-modulated RF at a
SAR of 5 W/kg for 1 or 24 h. No consistent evidence of altered ODC activity was observed following RFR
exposure. Similarly, the authors reported no evidence that physiological conditions of the exposed cells may alter
their responsiveness to RF EMF (3 to 6 independent experiments).
4807
4808
4809
4810
4811
4812
4813
4814
Billaudel et al. (2009b) exposed L929 murine fibroblasts to 835 MHz TDMA-modulated RF-EMF at
SARs of 0.5–2.5 W/kg for 8 h, 900 MHz GSM-modulated at SARs of 0.5–2.0 W/kg for 2 h or 1800 MHz GSMmodulated RF at 2.5 W/kg for 2–24 h and ODC activity was assessed using a 14C-radiolabel assay. The authors,
in 3 to 5 independent experiments, found no evidence of altered ODC activity in any exposed group under any
exposure condition tested. In a related study, Billaudel et al. (2009a) exposed human neuroblastoma-derived SHSY5Y cells to 835 MHz TDMA-modulated (50 Hz, 33% duty cycle) RF-EMF or 1800 MHz GSM-modulated
RF at SARs of 1 or 2.5 W/kg for 8 or 24 h. Similarly, the authors found no evidence of altered ODC activity
following any RF EMF exposure condition (4 to 5 independent experiments).
Table 12.3.10. In vitro studies assessing the effect of RF fields on ornithine decarboxylase activity (ODC)
Cells
Biological endpoint
Exposure conditions
Results
Comment
Authors
ODC activity
835 MHz, TDMA
No effect.
Observed
decrease in ODC
activity in
samples with
temperature
increase.
Desta, Owen &
Cress (2003)
No effect.
Noted that small
increase in
temperature
(<1 °C) resulted
in decreased
ODC activity.
Höytö et al.
(2006)
Number of
independent
experiments
Mouse
fibrosarcomaderived (L929)
cells
SAR <1–15 W/kg
8h
n=3–6
Mouse
fibrosarcomaderived (L929)
cells
ODC activity
900 MHz, CW or GSM
SAR 0.2–0.4 W/kg
2–24 h
n=3–5
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
177
Mouse
fibrosarcomaderived (L929)
cells (from 2
sources)
ODC activity
835 and 872 MHz, CW
and TDMA
SAR 2.5, 6 W/kg
2, 8, 24 h
Increased
activity after 8 h
exposure at 6
W/kg (waveguide
exposure
system).
n=3–7
Rat
ODC activity
glioblastomaderived (C6)
cells, rat
primary
astrocytes,
human
neuroblastomaderived (SHSY5Y) cells,
mouse
fibrosarcomaderived (L929)
cells
No effect in
Crawford
exposure
system.
872 MHz, CW and GSM
SAR 1.5, 2.5, 6 W/kg
2, 8, 24 h
Decreased
activity for all RF
conditions in rat
primary
astrocytes.
No consistent
effect in other
cell lines.
Replication study
of Höytö et al.
(2006).
Höytö, Juutilainen
& Naarala
(2007a)
Used 2 types of
exposure
systems.
Possible thermal
confounding in
waveguide
exposure data.
Temperature of
RF exposed cells
were 0.8 °C
higher than sham
for some
samples.
Höytö, Juutilainen
& Naarala
(2007b)
n=5–12
Mouse
fibrosarcoma
L929 cells
(starved and
enriched)
ODC activity
872 MHz, CW and GSM
No effect.
SAR 5 W/kg
Höytö et al.
(2008b)
1, 24 h
n=3– 6
Mouse
fibrosarcoma
L929 cells
(starved and
enriched)
ODC activity
n=3–5
Human
ODC activity
neuroblastomaderived (SHSY5Y) cells
n=4–5
835 and 872 MHz,TDMA
900 MHz GSM
1800 MHz GSM
No effect.
4 different
Billaudel et al.
exposure systems (2009a)
used, no effects
observed in any
system.
No effect.
Billaudel et al.
(2009a)
SAR 0.5–2.5 W/kg
8 h; 2 h; 2–24 h
835 MHz, TDMA
1800 MHz, GSM
SAR 1, 2.5 W/kg
8, 24 h
“No effect” means no statistically significant effect.
4815
4816
12.3.3.4
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
The use of high-throughput genomics and proteomics technologies is becoming increasingly popular
for analysis of differential gene/protein expression in response to a variety of chemical, pharmaceutical and
environmental exposures/conditions. These technologies, while extremely useful for screening large numbers of
genes/proteins for potential interaction with a test agent, are particularly prone to artefactual results in studies
with poor experimental design, improper data acquisition/normalization procedures and statistical analyses
(correction for multiple comparisons). Such studies require validation of the data using other techniques (e.g.
semi-quantitative RT-PCR and/or Western blotting) to provide additional scientific rigor. In recent years, several
high-throughput genomics and proteomics techniques have been applied to bio-electromagnetics research to
analyse differential gene and/or protein expression in human and mammalian cells in response to RFR exposure,
however only those studies which have met the minimum criteria for data acceptance are included in this section.
4827
Proteomics studies
4828
4829
4830
4831
Nylund and Leszcynski (2004) exposed a human endothelial cell line (EA.hy926) to 900 MHz GSMmodulated RF EMF for 1 h at an average SAR of 2.4 W/kg. The authors reported altered expression of 38
protein spots (p<0.05) and identified 4 proteins by MALDI-MS (10 independent experiments, p<0.05). Two of
the identified spots were isoforms of the cytoskeleton protein, vimentin. Increased expression of vimentin was
High-throughput studies on gene/protein expression
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
178
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
also observed by SDS-PAGE/Western blotting, although statistical analysis was not applied to these results.
[Correction for multiple comparison testing was not performed in this study.] These authors also examined the
effect of RF exposure in two closely related cell lines (EA.hy926 and EA.hy926v1) exposed to 900 MHz GSMmodulated RF for 1 h at a SAR of 2.8 W/kg (Nylund & Leszczynski, 2006). In this study, the authors observed
that 1 gene was significantly down-regulated in the EA.hy926 cell line and 13 genes were significantly
upregulated in a related EA.hy926v1 cell line following RF exposure (p<0.05, 10 independent experiments).
Proteome analysis indicated 38 differentially expressed proteins in the EA.hy926 cell line and 45 altered proteins
in the EA.hy926v1 cell line (p<0.05). The identity of the differentially expressed proteins was not determined.
[The transcriptomic and proteomic responses were not validated with other techniques. Correction for multiple
comparison testing was not performed in this study.]
4842
4843
4844
4845
4846
4847
4848
4849
In recent studies by this group, EA.hy926 cells, human brain microvascular endothelial cells
(HBMEC) and human umbilical vein endothelial cells (HUVEC) were exposed to 1800 MHz GSM-modulated
RF-EMF for 1 h at a SAR of 2.0 W/kg (Nylund et al., 2009; Nylund, Kuster & Leszczynski, 2010). The authors
observed numerous differentially expressed spots between sham and RF EMF exposed samples by 2D gel
electrophoresis (11 to 13 independent experiments). However these protein spots were either not identified and
validated by Western blotting or were not significantly different after correction for multiple comparison testing.
[Unlike earlier studies by this group, neither vimentin nor HSP27 were significantly affected by exposure to
1800 MHz GSM-modulated RF EMF exposure.]
4850
4851
4852
4853
4854
4855
4856
Gerner et al. (2010) assessed relative protein expression in Jurkat cells, human fibroblast (ES1) and
primary mononuclear cells after exposure to intermittent (5 min on, 10 min off) 1800 MHz GSM-modulated RFEMF for 8 hours at 2 W/kg. No significant differences were observed in the expression of any proteins between
sham and RF exposed cells using 2D gel electrophoresis (3 independent experiments). However, cells exposed
for 8 h to RF EMF showed a number of proteins with increased de novo protein synthesis in the Jurkat and ES1
cell cultures (p<0.05). While several stress-responsive proteins were found to be affected, including HSP’s,
Annexin proteins and T-complex protein 1, none of these observations were validated using other techniques.
4857
Studies not included in the analysis
4858
4859
4860
4861
4862
4863
4864
The first high-throughput study of protein expression in response to RF EMFs exposure was
completed by Leszczynski et al. (2002). In this study, a human endothelial cell line (EA.hy926) was exposed to
900 MHz GSM-modulated RF for 1 h at a SAR of 2.0 W/kg and protein expression was assessed by 2D-gel
electrophoresis (4 independent experiments). A three-fold increase in the number of phosphoprotein spots were
observed in exposed samples, relative to the sham controls. Several phosphoproteins, including HSP27 and
p38MAPK, were identified as demonstrating increased expression after RF exposure. [Statistical analysis was
not performed to determine if any of these results were statistically significant.]
4865
4866
4867
4868
4869
Kim et al. (2010) employed 2D gel electrophoresis to examine the proteome of a human-derived
breast cancer cell line (MCF7) after exposure to 849 MHz CDMA-modulated RF EMF for 1 h/day for 3 days at
SARs of 2 or 10 W/kg. At 24 h after the final RF exposure, no differences in protein expression were reported
between RF-exposed and sham-treated cells (3 independent experiments). [Statistical analysis was not performed
to determine if any of these results were statistically significant.]
4870
4871
4872
4873
4874
4875
4876
Zeng et al. (2006) exposed MCF7 cells to continuous or intermittent (5 min on/10 min off cycles)
1800 MHz GSM-modulated RF at a SAR of 3.5 W/kg for 1 to 24 h. Differential expression was defined as a fold
change greater than 2 in reference to the sham control. In 3 independent experiments the authors reported that a
small number of spots demonstrated differential expression between the RF and sham-exposed groups, but none
of these were identified and none of the proteins were differentially expressed at more than one time-point,
leading the authors to conclude that the differential spots occurred due to chance. [Statistical analysis was not
performed on the proteomic data to determine if any of these results were statistically significant.]
4877
4878
4879
4880
4881
4882
4883
4884
Li et al. (2007) exposed human lens epithelial cells (HLEC) to 1800 MHz GSM-modulated RF EMF
for 2 h at SARs of 1–3.5 W/kg. Immediately after exposure, the proteome was extracted and analyzed by 2D gel
electrophoresis (3 independent experiments). The authors observed 4 protein spots were upregulated by more
than 3-fold after exposure to 3.5 W/kg and 2-fold after exposure to 2.0 W/kg (no p values reported). No proteins
demonstrated altered expression after a 1 W/kg RF exposure. The authors used mass spectroscopy to identify
these spots as stress-related proteins, namely HSP70 and ribonucleoporin K. However, differential expression of
these proteins was not confirmed by Western Blot. [It is unclear whether differential expression was determined
by fold-changes or statistical analysis.]
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
179
4885
Transcriptomics studies
4886
4887
4888
4889
4890
4891
Nylund and Leszczynski (2006) exposed two endothelial cell lines (EA.hy926 and EA.hy926v1) to
900 MHz GSM-modulated RF EMF for 1 h at a SAR of 2.8 W/kg. Gene expression analysis (3 independent
experiments per cell line) indicated that 1 gene was significantly down-regulated in the EA.hy926 cell line and
13 genes were significantly up-regulated in a related EA.hy926v1 cell line following RF exposure (p<0.05).
Differential expression of these genes was not confirmed using RT-PCR. [Correction for multiple comparison
testing was not performed in this study.]
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
Huang et al. (2008b) exposed a mouse auditory hair cell line (HEI-OC1) to 1763 MHz CDMAmodulated RF EMF at a SAR of 20 W/kg for 24 h and RNA was extracted 5 h later. Gene expression analysis
identified 29 genes that were statistically significant (p<0.05) after correction for multiple comparison testing
and displayed fold changes >1.5 (3 independent experiments). Of these, 14 were downregulated and 15 were
upregulated. None of the differentially expressed genes were confirmed by RT-PCR. In a related study, Huang et
al. (2008a) exposed human T-cell derived (Jurkat) cells to 1763 MHz CDMA-modulated RF at a SAR of 10
W/kg for 24 h. Microarray analysis (5 independent experiments) did not identify any genes that were
differentially expressed with a fold-change greater than 2.0, but 10 genes were identified with a fold-change
greater than 1.3 (p < 0.1). [Differential expression of these genes was not confirmed using RT-PCR. The latter
study has been also described in Section 12.3.2 (Intracellular and intercellular signalling).]
4902
4903
4904
4905
4906
4907
Sakurai et al. (2011) assessed differential gene expression in a normal human glial cell line (SVGp12)
exposed to 2.45 GHz RF EMF, CW, at SARs of 1, 5 and 10 W/kg for 1–24 h. Microarray analysis identified 23
differentially expressed genes in response to RF exposure after correction of the analysis for multiple
comparison testing (3 independent experiments; p < 0.05). RT-PCR validation was conducted on 22 of these
genes, however none were validated as being differentially expressed in the RF-exposed samples compared to
the sham control group.
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
Qutob et al. (2006) exposed human glioblastoma-derived (U87MG) cells to 1900 MHz pulsemodulated RF EMF for 4 hours at SARs of 0.1, 1.0 and 10.0 W/kg. In this study, 5 independent experiments
were conducted, matched heat-shock positive controls were performed for each experiment and microarray
analysis was corrected for multiple comparison testing. No evidence of differential gene expression was
observed in any of the RF-exposed samples relative to the sham control. RT-PCR validation was conducted on a
panel of HSP and a lack of responsiveness to RF exposure observed in the microarray analysis was confirmed by
RT-PCR. In a follow-up study, Chauhan et al. (2007b) exposed U87MG cells to intermittent (5 min on/10 min
off cycles) 1900 MHz pulse-modulated RF EMF for 24 hours at SARs of 0.1, 1.0 and 10.0 W/kg and RNA was
harvested 6 hours after exposure. A human-derived monocyte cell line (Mono-Mac-6) was also exposed to
similar RF conditions for 6 hours and RNA was harvested either immediately or at 18 hours post-exposure. For
both cell lines 5 independent biological experiments were conducted and microarray analysis found no
differentially expressed genes in response to any RF-exposure condition. RT-PCR validation confirmed a lack of
response for a variety of HSP in response to RF exposure. In summary, these studies found no evidence of
differential gene expression in either cell line at any SAR or time-point tested following RF exposure.
4922
4923
4924
4925
4926
4927
4928
4929
Whitehead et al. (2006) exposed mouse embryonic (C3H 10T1/2) cells to 835.62 MHz, FDMA, or
847.74 MHz, CDMA, RF EMF at a SAR of 5 W/kg for 24 hours. Three independent experiments were
conducted for each of the exposure conditions tested and matched samples exposed to X-radiation (0.68 Gy)
were used as positive controls. Sham-sham comparisons were conducted to estimate the false-discovery rate
(FDR) in the experimental model. While the authors indicated that approximately 200 genes were differentially
expressed between RF- and sham-exposure conditions (for both CDMA and FDMA exposures; p < 0.05), the
authors indicated that this number was lower than that between sham-sham comparisons (~400) and therefore
that expected by chance. [These results were not confirmed with RT-PCR.]
4930
4931
4932
4933
4934
4935
4936
Roux et al. (2011) exposed normal human epidermal keratinocytes (NHEK) to 900 MHz RF EMF,
CW, at a SAR of 2.6 mW/kg for 10 min or 73 mW/kg for 30 min. Four independent experiments were conducted
for the microarray analysis whereby 16–17 genes for each exposure condition were identified as differentially
expressed (p<0.05) after correction for multiple comparison testing. None of these genes was affected under both
exposure conditions. RT-PCR validation was conducted for 15 known genes but the authors failed to validate
differential expression for any of the genes examined. The authors concluded that the cultured keratinocytes
were not significantly affected by RF EMF exposure.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
180
4937
Studies not included in the analysis
4938
4939
4940
4941
4942
4943
4944
Harvey and French (2000) exposed a human mast cell line (HMC-1) to 864.3 MHz RF EMF at a SAR
of 7 W/kg for 20 min, 3 times a day for 7 days. The authors reported 3 genes were significantly different in the
RF-exposed samples compared to the sham controls (2 independent experiments). These genes were the protooncogene c-kit, the transcription factor NME-B and the apoptosis-associated gene DAD-1. [The authors reported
a 0.7 °C temperature difference between the RF- and sham-exposed samples. The results were not confirmed by
RT-PCR and were based on differences in fold- change. Statistical analysis of the gene expression data was not
performed].
4945
4946
4947
4948
4949
4950
4951
Pacini et al. (2002) examined gene expression in human skin fibroblasts exposed to 902.4 MHz GSMmodulated RF EMF for 1 h from a commercial mobile phone on the benchtop at an estimated SAR of 0.6 W/kg.
The authors reported 14 genes were differentially expressed in the RF-exposed samples, including several
mitogenic signal transduction genes (1 independent experiment). [There is an inadequate description of the RF
exposure system dosimetry and an inadequate number of independent experiments. The results were based on
differences in fold-change. Statistical analysis of the gene expression data was not performed and results were
not confirmed by RT-PCR.]
4952
4953
4954
4955
4956
4957
4958
Lee et al. (2005) exposed cultured human HL-60 cells to 2450 MHz pulse-modulated RF EMF at 10
W/kg for 2–6 h. The authors used serial analysis of gene expression (SAGE) to quantify gene transcript levels
after RF exposure and reported 221 genes demonstrated altered expression after a 2 h exposure and 759 after a
six hour exposure (1 independent experiment). [There is an inadequate number of independent experiments. A 2
h sham-exposed sample was used as a reference to compare against the 6 h RF-exposed sample. The results were
based on differences in fold-change. Statistical analysis of the gene expression data was not performed and
results were not confirmed by RT-PCR.]
4959
4960
4961
4962
4963
4964
4965
4966
Remondini et al. (2006) exposed a variety of human derived cell lines to 900 or 1800 MHz GSM or
DTX-modulated RF EMF for 1–44 h, at SARs ranging from 1.0–2.5 W/kg. While multiple independent
experiments were conducted for each exposure condition, a single RNA pool was generated for each RF- or
sham-exposed group (1 independent experiment per comparison). The authors reported 32 differentially
expressed genes in EA.hy926 cells exposed to 900 MHz RF EMF, 34 differentially expressed genes in U937
cells exposed to 900 MHz RF EMF and 12 differentially expressed genes in HL-60 cells exposed to 1800 MHz
RF EMF. [There is an inadequate number of independent experiments for each gene expression comparison and
the results were not confirmed by RT-PCR.]
4967
4968
4969
4970
4971
Zeng et al. (2006) exposed MCF7 cells to intermittent (5 min on/10 min off cycles) 1800 MHz RF
EMF for 24 hours at 2.0 and 3.5 W/kg. No differences were observed at a SAR of 2.0 W/kg, but 5 differentially
expressed genes were identified in cells exposed at a SAR of 3.5 W/kg. RT-PCR validation failed to confirm
these differentially-expressed genes. [These results are based upon only 2 independent experiments and
differential expression was based only on fold-changes. No statistical analysis was performed in this study.]
4972
4973
4974
4975
4976
4977
Zhao et al. (2007) exposed primary rat neurons to 1800 MHz GSM-modulated RF EMF intermittently
(5 min on/10 min off cycles) for 24 h at a SAR of 2 W/kg. Using fold-change analysis, the authors reported that
24 genes were upregulated and 10 genes were downregulated after RF exposure (number of independent
experiments not reported). RT-PCR analysis was conducted on 25 genes and 19 were reported to be
differentially expressed (p<0.05). [These results are based upon an unknown number of independent
experiments for microarray analysis and differential expression was based only on fold-changes.]
4978
4979
4980
4981
4982
4983
4984
Hirose et al. (2006; 2007) assessed gene expression in human glioblastoma-derived A172 cells or fetal
lung-derived IMR-90 cells after exposure to 2142.5 MHz CW- or CDMA-modulated RF EMF for 2–48 hours at
SARs ranging from 0.08 to 0.8 W/kg. Despite assessing a variety of exposure conditions including exposure
duration, signal modulation and SAR levels, the authors reported no consistent evidence of differential gene
expression in either cell line tested. [While RT-PCR was used to validate the results in each of the above studies,
microarray analysis was based upon only 2 independent experiments thereby limiting the significance of these
results.]
4985
4986
4987
4988
In a similar study, Sekijima et al. (2010) exposed three cell lines (A172; IMR-90; and neurogliomaderived H4 cells) to 2142.5 MHz, CW or W-CDMA-modulated RF EMF at SARs of 0.08–0.8 W/kg for up to 96
hours. The authors observed differential expression of a small number of genes in each cell line after correction
for multiple comparison testing (p<0.05). Ribosomal protein S2, growth arrest specific transcript 5 and integrin
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
181
4989
4990
β5 were differentially expressed in H4 cells at the two highest SARs tested. These results were not validated by
RT-PCR. [Microarray analysis was based upon only 2 independent experiments.]
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
Zhao et al. (2007) exposed cultured mouse primary neurons and astrocytes to 1900 MHz GSMmodulated RF EMF from a mobile phone for 2 h, by placing the mobile phone atop petri dishes containing the
cells with the lid removed. Sham exposure was defined as placing the mobile phone in ‘stand-by’ mode. The
SAR for the exposure was not reported. The authors reported microarray analysis identified 8 genes were
upregulated and 1 was downregulated in neurons in comparison to unexposed (incubator) control cells, based
upon differences in fold-change (2 independent experiments). Three of these genes were assessed by RT-PCR
analysis, which confirmed the microarray results. However, RT-PCR analysis showed no differences between
the RF- and sham-exposed groups. [There is an inadequate description of the RF exposure system dosimetry. It
is unclear whether placing the mobile phone in ‘stand-by’ mode is an appropriate sham control. Differential
expression was based upon fold-change and not statistical analysis. No comparison was made between the RFand sham-exposed samples for either the microarray or RT-PCR analysis.]
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
Le Quément et al. (2011) exposed primary human skin cells to 60 GHz millimeter-wave RF EMF at a
SAR of 42.4 W/kg for 1, 6, or 24 h. Temperature recording indicated that the RF-exposed samples displayed a
slightly elevated temperature (~0.8 °C) relative to the non-exposed controls. This study employed four
independent experiments and microarray analysis was corrected for multiple comparisons. No significant
differences in gene expression were identified when the Benjamini-Hochberg correction was applied for multiple
comparisons testing. When the authors re-evaluated their data using a less stringent statistical (t-test) approach
(no correction for multiple comparison testing), 130 transcripts were identified as potentially differentially
expressed (p<0.05). RT-PCR analysis identified 5 of 24 selected genes as differentially expressed. [In this study,
the sham controls were not time-matched with the RF-exposed samples and therefore are not true sham controls.
Temperature confounding in this study is also possible.]
Table 12.3.11. In vitro high-throughput studies assessing the effect of RF fields on gene/protein expression
Cells
Biological endpoint
Exposure conditions
Results
Comment
Authors
protein
900 MHz GSM
Altered
expression of
38 protein
spots.
Similar results
observed with
SDS-PAGE/
Western blot.
Nylund &
Leszczynski
(2004)
Identified 4
proteins, of
which 2 were
isoforms of
vimentin.
No correction for
multiple
comparison
testing.
Altered
expression of
38 (EA.hy926)
and 45
(EA.hy926v1)
protein spots.
Follow-up to
Nylund and
Leszcynski
(2004).
Number of
independent
experiments
Human
umbilical vein
(EA.hy926)
cells
Average SAR 2.4 W/kg
1h
n=10
Human
umbilical vein
(EA.hy926 and
EA.hy926v1)
cells
protein
900 MHz GSM
Average SAR 2.8 W/kg
1h
n=10
Did not identify
proteins.
Human
umbilical vein
(EA.hy926)
cells
protein
1800 MHz GSM
Average SAR 2 W/kg
1h
Nylund &
Leszczynski
(2006)
Results not
confirmed by
Western blot.
No correction for
multiple
comparison
testing.
Altered
Results not
expression of 8 confirmed by
protein spots.
Western blot.
Nylund et al.
(2009)
No correction for
multiple
comparison
testing.
n=10
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
182
Human
umbilical vein
endothelial
(HUVEC) cells,
human brain
endothelial
(HBMEC) cells
protein
1800 MHz GSM
Average SAR 2 W/kg
1h
No effect after
results
corrected for
multiple
comparison
testing.
Some spots
identified as
artifacts (dust).
Nylund, Kuster &
Leszczynski
(2010)
n=11
Jurkat (T-cell
leukemia),
human
fibroblast (ES1)
and
mononuclear
cells
protein
1800 MHz GSM
Average SAR 2 W/kg
8 h (5 min on/10 min off
cycles)
n=3
No difference in Results not
proteome
confirmed by
expression, but Western blot.
RF- exposed
cells had
greater de novo
synthesis of
some stress
proteins.
Gerner et al.
(2010)
1 gene
downregulated
in EA.hy926.
Results not
confirmed by RTPCR.
Nylund &
Leszczynski
(2006)
13 genes
upregulated in
EA.hy926v1.
No correction for
multiple
comparison
testing.
Transcriptomic studies
Human
umbilical vein
(EA.hy926 and
EA.hy926v1)
cells
mRNA
900 MHz GSM
Average SAR 2.8 W/kg
1h
n=3
Mouse auditory mRNA
hair cell-derived
(HEI-OC1) cells
n=3
Human T-cell
mRNA
derived (Jurkat)
cells
n=5
1763 MHz CDMA
SAR 20 W/kg
24 h
1763 MHz CDMA
SAR 10 W/kg
24 h
29 differentially Results not
expressed
confirmed by RTgenes identified PCR.
with foldchange >1.5.
Huang et al.
(2008a)
No effect on
gene
expression for
fold change
>2.0.
Results not
confirmed by RTPCR.
Huang et al.
(2008b)
Altered
expression of
23 genes.
RT-PCR did not
confirm these
results.
Sakurai et al.
(2011)
No effect on
gene
expression.
Matched heatshock positive
controls.
Qutob et al.
(2006)
10 genes had
fold change >
1.3.
Human fetal
astroglial
(SVGp12) cells
mRNA
SAR 1, 5, 10 W/kg
1, 4 or 24 h
n=3
Human
glioblastomaderived
(U87MG) cells
2450 MHz CW
mRNA
1900 MHz PW
SAR 0.1, 1, 10 W/kg
4h
RT-PCR
validation on
HSPs. Lack of
response
confirmed.
n=5
For HSP see
12.3.3.1.
Human
glioblastomaderived
(U87MG) cells,
human
monocytederived (MonoMac 6) cells
mRNA
1900 MHz pulsed-wave
SAR 0.1, 1, 10 W/kg
6 h (MM6), 24 h (U87MG)
(5 min on/10 min off)
n=5
No effect on
gene
expression.
RNA harvested at Chauhan et al.
various times
(2007b)
after RFR
exposure.
RT-PCR
validation on
HSPs. Lack of
response
confirmed.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
183
Mouse
mRNA
embryonic (C3H
10T1/2) cells
835.62 MHz FDMA
847.74 MHZ CDMA
n=3
24 h
Human
epidermal
keratinocytes
(NHEK)
mRNA
SAR 5 W/kg
900 MHz CW
SAR 2.6, 73 mW/kg
10,30 min
n=4
~200 genes
differentially
expressed in
RF samples,
but this was
lower than
sham-sham
comparisons
(~400 genes.
Matched X-ray
positive controls.
Small number
of genes
differentially
expressed for
each condition
RT-PCR did not
validate any
genes.
Whitehead et al.
(2006)
Differential
expression in
RFR samples less
than expected by
chance.
No results
validated with RTPCR.
Roux et al. (2011)
5012
5013
Excluded papers
5014
5015
(Calabrò et al., 2012; Cao et al., 2009; Litovitz et al., 1993; Litovitz et al., 1997; Penafiel et al., 1997; Perez et
al., 2008; Port et al., 2003; Trivino Pardo et al., 2012; Verma & Dutta, 1993)
5016
12.3.4
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
The number of cells in a multicellular organism is tightly regulated by controlling the rate of cell
division and cell death. Apoptosis, or programmed cell death, is an evolutionarily conserved mechanism for the
selective removal of aging, damaged or otherwise unwanted cells. It is an essential component of many normal
physiological processes such as embryogenesis, normal tissue development and the immune response.
Unintentional cellular insults may also trigger cell death such as those caused by ultraviolet light or chemical or
physical agents. Apoptosis is distinguishable from cell death by necrosis, which is a considered as a random
event causing a potentially damaging inflammatory response. Apoptosis can be triggered by intracellular or
extracellular signals whereby two main pathways exist: the intrinsic or mitochondrial pathway which transmits
intracellular received death signals, and the extrinsic or death receptor pathway relaying apoptotic messages via
receptors. Although these pathways act independently to initiate the death machinery, there is a delicate
coordination and cross talk between the extrinsic and intrinsic pathways, which leads to the activation of the
executioner caspase (cysteinyl, aspartate-specific proteases) cascade. The morphological features of apoptosis
consist of chromatin condensation, cell shrinkage and membrane blebbing, which can be clearly observed by
light microscopy. The biochemical features include DNA fragmentation, protein cleavage at specific locations,
increased mitochondrial membrane permeability, and the externalization of phosphatidylserine (PE) on the cell
membrane surface.
5033
5034
5035
5036
5037
5038
5039
Laboratory assays that are commonly employed to investigate the induction of apoptosis are based on
these specific features and include tests to measure e.g. the DNA fragmentation (comet assay, double staining
with TdT-mediated dUTP nick-end labeling (TUNEL) and propidium iodide (PI)), the formation of apoptotic
bodies (haematoxylin and eosin (H&E) staining), the cell membrane permeability (YOPRO-1/PI staining), the
appearance of PE on the cell membrane surface (Annexin-V-FITC/PI), mitochondrial membrane potential
modifications (—ΔΨm—, JC-1), protein cleavage, and/or executioner caspase activity but also other apoptosisrelated protein expressions, such as Bcl2.
5040
5041
5042
5043
5044
5045
5046
5047
5048
The previous WHO report on the effects of RF exposure (WHO, 1993b) did not report any studies on
apoptosis. The present literature search identified 46 relevant papers in this area, addressing the effect of RF
EMF exposure, either alone or in combination with chemical agents. Nine papers were obtained from other
sources and five papers were in languages that could not be understood. That left 50 papers to be extracted.
Among the relevant publications, fourteen were excluded because they did not meet the inclusion criteria for in
vitro studies, and references are listed in Appendix X, and thirteen papers did not completely comply with the
quality criteria for inclusion due to methodological issues, therefore they are only presented in the text. The
remaining 23 papers have been described in the text. Unless specifically mentioned, papers did not report on
blinding of the investigators to the exposure condition.
5049
5050
Terro et al. (2012) examined the effects of 24 h exposure to a 900 MHz GSM signal (SAR = 0.25
W/kg) on apoptosis and chaperone-mediated autophagy in primary cerebral cortical cells (neurons and
Apoptosis
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
184
5051
5052
5053
5054
5055
5056
5057
astrocytes) of rat embryos. In at least three independent experiments, no induction of apoptosis was observed, as
evaluated by morphological (chromatin condensation and fragmentation) and biochemical (analysis of caspase-3
cleavage and activation) techniques. Moreover, no effect was observed on autophagy (self-eating), another
eukaryotic system for degradation in lysosomes. Serum deprivation for 48 h served as positive control for
autophagy and induced alteration in active lysosome distribution. [In this study, already described in sections 8.3
and 12.3.3, the effect of RF EMF exposure on neurodegenerative disorders and protein expression has also been
investigated.]
5058
5059
5060
5061
5062
5063
5064
Simon et al. (2013) analyzed the influence of exposure to a 900 MHz GSM signal (SAR = 2 W/kg) on
pigmented and non-pigmented skin cells and the influence of melanocytes on this response. Cell cultures were
exposed for 6 h and analyzed 2, 6, 18 and 24 h after exposure. In a set of two to twenty experiments, no effects
were detected on apoptosis, evaluated by morphological (H&E staining) and biochemical (cleaved caspase-3
expression) techniques. No positive controls were included in the study design. [This study has also been
reported in sections 12.3.5 and 12.3.6, where the effects of RF exposure on oxidative stress, cell proliferation and
differentiation are discussed.]
5065
5066
5067
5068
5069
5070
Zeni et al. (2012b) failed to find changes in apoptosis, evaluated by flow-cytometric techniques
assessing the appearance of PE on the cell membrane surface (Annexin-V-FITC-PI method) in rat neuronal cells
(PC12) exposed for 24 h to an 1950 MHz UMTS signal (SAR = 10 W/kg). The study was performed blinded.
The absence of effect was detected either immediately after exposure or 24 h later (three independent
experiments). Treatments with etoposide as positive control resulted in a significant increase in apoptosis. [In
this study, the effect on DNA damage has also been investigated, as reported in section 12.3.1.]
5071
5072
5073
5074
5075
5076
Höytö et al. (2008b) applied different experimental conditions to test inconsistences in effects of RF
by altered physiological states (stimulation or stress) of L929 murine fibroblasts. The cells were exposed for 1 h
to 872 MHz RF EMF, CW and GSM-modulated, at an SAR of 5 W/kg. In all the experimental conditions
examined, in three independent experiments, no evidence of effects on caspase-3 activity was found. Serum
deprivation induced a significant increase in apoptosis (positive control). [In this study the effect of RF exposure
on ODC activity and cell proliferation was also investigated (see sections 12.3.3 and 12.3.6).]
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
Joubert et al. evaluated in a series of investigations the induction of apoptosis in neuronal cells
exposed to 900 MHz fields. In a first study (Joubert et al., 2006) human neuroblastoma SH-SY5Y cells were
employed to test the effect of 24 h exposure to CW (SAR = 2 W/kg) and GSM modulation (average SAR = 0.25
W/kg). No change in apoptosis rate was detected either immediately or 24 h after exposure, as assessed by
applying three different tests: 4',6-diamino-2-phenylindole (DAPI) staining, flow cytometry using double
staining with TdT-mediated dUTP nick-end labeling (TUNEL) and propidium iodide (PI), and measurement of
caspase-3 activity by fluorimetry (three independent experiments). In a second study (Joubert et al., 2007) the
authors exposed primary rat cortical neuron cells of embryonic Wistar rats for 24 h to a 900 MHz GSM signal at
0.25 W/kg. No statistically significant difference was detected in exposed cells compared to sham-exposed
controls, assayed immediately after exposure or 24 h later by employing the three tests described above (three
independent experiments). The same experimental conditions were applied in a further investigation to test the
effect of CW (average SAR = 2 W/kg) (Joubert et al., 2008). In this case a statistically significant increase in
apoptosis rate, evaluated by means of DAPI and TUNEL assays, was found in the RF-exposed neurons (five
independent experiments; p<0.001). The effect was not caspase-3-dependent, whereas the percentage of
apoptosis-induced-factor (AIF)-positive nuclei in exposed neurons was increased (five independent experiments;
p<0.001). In all the investigations, treatments with staurosporine, a protein kinase inhibitor, were used as
positive controls and gave a significant increase of apoptosis. [In Joubert et al. (2006; 2008), samples exposed to
CW exhibited 2 °C temperature increase with respect to sham-exposed controls, but the authors demonstrated, by
ad hoc experiments, that such an increase was not responsible for the observed effect. In Joubert et al. (2008) the
authors followed a blind procedure, while in the previous papers they did not mention such a procedure.]
5097
5098
5099
5100
5101
5102
Moquet and co-workers (2008) performed a study to examine whether RF EMF induced apoptosis in
murine neuroblastoma (N2a) cells in both proliferating and differentiated states. Cell cultures were exposed for
24 h to 935 MHz, CW, GSM-basic or GSM-talk (SAR = 2 W/kg). In three independent experiments, carried out
in blind conditions, no changes in apoptosis rate were found, as assessed at several time-points between 0 and 48
h post-exposure by applying three different assays: caspase-3 activity, TUNEL and Annexin-V-FITC-PI. In cell
cultures exposed to X rays as positive control the level of apoptosis significantly increased.
5103
5104
Palumbo et al. (2008) investigated the induction of apoptosis in human peripheral blood lymphocytes,
either in G0 stage and proliferating and in human lymphoblastoid Jurkat T-cells after a 1-h exposure to 900 MHz
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
185
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
RF EMF, GSM modulated, SAR = 1.35 W/kg. The experiments were carried out blinded. A slight but
statistically significant increase in caspase-3 activity was observed 6 h after exposure in Jurkat cells (five
experiments, p<0.05) and in proliferating lymphocytes (four donors, p<0.05), but not in resting lymphocytes
(G0-phase). To assess if the increased activity was related to a RF-induced apoptosis, two more endpoints were
investigated: the cleavage of PARP, a protein crucial for several processes, including apoptosis, and PE. No
differences were observed between RF- and sham-exposed cells, for both parameters. A significant increase in
apoptosis was detected in Jurkat cells treated with Fas-L and in human lymphocytes treated with cytosine
arabinose as positive controls. The authors concluded that RF-induced caspase-3 activation occurs only in
proliferating cells and is independent from apoptosis. [In this study, effects on the cell cycle were also
investigated (Section 12.3.6).]
5115
5116
5117
5118
5119
5120
5121
5122
Belyaev et al. (2005) collected blood samples from seven healthy donors and seven alleged
electromagnetic hypersensitive (EHS) individuals and exposed isolated lymphocytes to 915 MHz RF (GSM
modulation, SAR = 37 mW/kg) for 2 hours, in blind conditions. No differences between RF- and sham-exposed
cells were detected. No difference in response was observed between lymphocytes from EHS and healthy
subjects. Apoptosis was evaluated by measuring DNA fragmentation and by morphological analysis, carried out
24 and 48 h post-exposure. Treatments with gamma rays served as positive controls and resulted in a significant
increase in apoptosis. [In this study the effect of RF exposure on DNA damage has also been investigated
(Section 12.3.1).]
5123
5124
5125
5126
5127
5128
5129
5130
Hook et al. (2004b) exposed human lymphoblastoid Molt-4 cells for 2, 3 or 21 h to RF EMF as
follows: a) 847.74 MHz, CDMA, SAR = 3.2 W/kg; b) 835.62 MHz, FDMA, SAR = 3.2 W/kg; c) 813.56 MHz,
iDEN, SAR = 24 W/kg; d) 836.55 MHz, TDMA, SAR = 26 W/kg. The data, from three independent experiments
for each condition, showed no significant differences between sham-treated cells and cells exposed to RF
radiation for any frequency, modulation or exposure time, as assessed by Annexin-V/FITC/PI method.
Treatments to 44 °C for 20 min were used as a positive control and resulted in a significant increase in the
percentage of apoptotic cells. [In this study the effect of RF exposure on DNA damage was also investigated
(Section 12.3.1).]
5131
5132
5133
5134
5135
5136
5137
Chauhan et al. (2007a) used exponentially growing human leukaemia (HL-60), monocyte (MonoMac-6) and lymphoblastoma (TK6) cells to assess apoptosis (neutral comet assay) after exposure to 1900 MHz,
pulse-modulated fields for 6 h (5 min on/10 min off cycles) at SARs of 1 and 10 W/kg. In five independent
experiments, carried out in blind conditions, no change of apoptosis was detected immediately after exposure or
6 and 24 h later in any of the cell lines tested. Positive controls (treatments at 43 °C for 1 h) displayed a
significant increase in apoptosis. [In this study, also described in sections 10.3, 12.3.2 and 12.3.6, the effect of
RF EMF exposure on cytokine expression and cell cycle was also investigated.]
5138
5139
5140
5141
5142
5143
5144
5145
5146
Sanchez et al. (2006b) exposed isolated human skin cells and reconstructed human epidermis (five
independent experiments for both cell lines) to 900 MHz GSM-modulated RF EMF at an SAR of 2 W/kg for 48
h. No apoptosis was detected by evaluating PE. In a follow-up study (Sanchez et al., 2007), the same
investigators exposed primary human skin cells (keratinocytes and fibroblasts) for 48 h to 1800 MHz GSMmodulated RF (SAR = 2 W/kg). Also in this case no effect was detected by PE in both cell types, as assessed in
six independent experiments. In these studies, carried out blinded, positive controls (treatments with UV light)
worked properly. [In these studies, the effect of RF EMF exposure on cell proliferation (Sanchez et al., 2006b)
and protein expression (Sanchez et al., 2006b; 2007) was also investigated; they are also described in sections
12.3.3 and 12.3.6.]
5147
5148
5149
5150
5151
5152
5153
Hirose et al. (2006) exposed human glioblastoma-derived A172 cells and fetal lung-derived IMR-90
cells to 2142.5 MHz RF EMF, CW or CDMA-modulated, for 24 and 48 hours at SARs ranging from 0.08 to 0.8
W/kg. The experiments were performed in blind conditions. No apoptosis was detected, either by evaluating PE
and measuring the expression of p53, a tumor suppressor gene and regulator of apoptotic cell death (three
independent experiments). Treatments with doxorubicin as positive control resulted in a significant increase in
apoptotic cells. [This study has also been described in sections 12.3.2 (Signal transduction) and 12.3.3 (Gene and
protein expression).]
5154
5155
5156
5157
5158
Bourthoumieu et al. (2013) investigated the expression of the p53 protein and its activation (due to its
ability to initiate apoptosis) in human amniotic cells exposed for 24 h to 900 MHz GSM modulated RF EMF
(SARs = 0.25, 1, 2, and 4 W/kg). The results of three independent experiments performed using three different
donors showed no effect in p53 expression (Western blot assay) by comparing sham-exposed to RF-exposed
cultures. Treatments with bleomycin as positive control resulted in a significant increase in apoptotic cells. [This
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
186
5159
5160
study has also been described in sections 12.3.2 (Signal transduction), 12.3.3 (Gene and protein expression) and
12.3.6 (Cell proliferation).]
5161
5162
5163
5164
5165
5166
5167
5168
Nikolova et al. (2005) exposed mouse nestin-positive progenitor cells for 48 h (5 min on/30 min off
cycles) to a 1710 MHz GSM signal (average SAR = 1.5 W/kg) to analyse apoptosis (DNA fragmentation and
mitochondrial function) and different transcript levels of apoptosis-related genes (bax and GADD45). The RF
exposure resulted in upregulation of bax and GADD45 mRNA levels (p<0.05). No effect was detected in the
other parameters investigated (5 independent experiments), suggesting that RF exposure may not affect cell
physiological functions. Positive controls were not included in the study design. [In this paper, also quoted in
sections 12.3.1, 12.3.2, 12.3.3 and 12.3.6, the effect of RF exposure on DNA damage, signal transduction, gene
expression and proliferation was also investigated.]
5169
5170
5171
5172
5173
5174
5175
5176
5177
Yoon and co-workers (2011) studied the effect of exposure to a 1763 MHz CDMA RF EMF (SAR =
10 W/kg, 1 h per day for seven days) on cultured human dermal papilla cells by evaluating changes in the
expression of protein marker-related to hair growth and apoptosis. In three independent experiments, the
expression of insulin-like growth factor-1 (IGF-1) mRNA in human dermal papilla cells was significantly
induced upon RF exposure, resulting in an increased expression of Bcl-2, cyclin and increased phosphorylation
of MAPK-1 protein (p<0.05). Exposure significantly suppressed apoptosis in hair matrix keratinocytes and
enhanced hair shaft elongation in ex vivo hair organ cultures (p<0.05). Positive controls were not included in the
study design. [In this study the effect of RF exposure on oxidative stress and cell cycle has also been investigated
(see sections 12.3.5 and 12.3.6).]
5178
5179
In five studies the effects of RF EMF exposure alone as well as in combination with chemical agents
on apoptosis have been assessed.
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
Höytö et al. (2008a) exposed human neuroblastoma (SH-SY5Y) cells and mouse fibroblasts (L929) to
872 MHz CW and GSM-modulated RF EMF (SAR = 5 W/kg) for 24 h to investigate caspase-3 activity and
DNA fragmentation. Moreover, the effect of combined exposure with menadione (Vitamine K3, a cell death and
oxidative stress inducer) was also evaluated in both cell lines. In three independent experiments, no effect of RF
exposure alone was detected, in both cell types. Treatments with menadione alone also served as positive
controls and gave significant increase in caspase-3 activity in both cell lines. The results of co-exposures showed
a statistically significant increase in caspase-3 activity in L929 cells co-exposed to GSM modulation (p<0.01)
but not to CW. No significant effects were detected in SH-SY5Y cells. [Although in the figures the authors do
not call it “sham” the control is actually a sham. In this study the effect of RF on cell proliferation and oxidative
stress was also investigated (sections 12.3.5 and 12.3.6).]
5190
5191
5192
5193
5194
5195
5196
Lantow et al. (2006c) investigated whether exposure to an 1800 MHz GSM-DTX signal at an SAR of
2.0 W/kg for 12 h, either alone or in the presence of phorbol 12-myristate 13-acetate (PMA, a necrosis-inducer),
or glyotoxin (an apoptosis-inducer) could altered apoptosis in Mono-Mac-6 cells. No significant effects of RF
exposure alone or in combination with the chemicals were detected on apoptosis and necrosis, as evaluated by
Annexin-V-FITC/PI in three independent experiments performed blinded. Treatments with chemicals alone
resulted in a significant increase in apoptosis (positive controls). [In this study the effects of RF EMF on cell
cycle and protein expression were also investigated (sections 12.3.3 and 12.3.6).]
5197
5198
5199
5200
5201
5202
5203
5204
Merola et al. (2006) exposed human neuroblastoma cells (LAN-5) to 900 MHz GSM modulated RFEMF at an average SAR of 1.0 W/kg for 24 to 72 h. RF was given either alone or in combination with the
differentiative agent retinoic acid or the apoptosis-inducer camptothecin. No significant alteration in apoptosis
induction was detected in all cases, as assessed in three independent experiments performed blinded, by caspase
activation analysis and by molecular detection of Poly (ADPribose) polimerase (PARP) cleavage. Treatments
with camptothecin alone induced a significant increase in apoptosis (positive control). [In this study the effect of
RF exposure and co-exposure on the expression of proto-oncogenes and on cell proliferation and differentiation
was also investigated (sections 12.3.3 and 12.3.6).]
5205
5206
5207
5208
5209
5210
5211
Capri et al. (2004a) evaluated the induction of apoptosis in human peripheral blood mononuclear cells
(PBMC) from 28 young and eight old donors. Cell cultures were exposed blinded to RF fields using three
different GSM-modulations of 1800 MHz (GSM basic, GSM talk and DTX) for 44 h (10 min on/20 min off
cycles) at SARs of 1.4 and 2.0 W/kg. Moreover, the effect of combined exposures with the apoptosis-inducing
agent 2-deoxy-D-ribose (dRib) was also investigated. No effect was detected in all cases, as assessed by the
Annexin-V-FITC/PI method and mitochondrial membrane potential modifications, using the specific lipophilic
cationic probe JC-1. Treatments with dRib alone gave positive findings (positive controls). [In this study the
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
187
5212
5213
5214
5215
5216
5217
5218
effect of RF exposure and co-exposure on HSP was also investigated, as reported in section 12.3.3.] In another
study the same research group (Capri et al., 2004b) exposed PBMC from 39 healthy donors to 900 MHz RF
EMF, CW or GSM modulated, SAR= 70 and 76 mW/kg, respectively, for 1 h/day for three days. Also in this
case treatments with dRib were carried out to assess the effect of combined exposure. No changes in apoptosis
(Annexin-V-FITC/PI) and in mitochondrial membrane potential (JC-1) were detected in exposed and co-exposed
samples compared to their respective sham-controls. [In this study the effect of RF EMF on cell cycle and
proliferation was investigated (see Section 12.3.6).]
5219
Studies not included in the analysis
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
The first high-throughput study of protein expression in response to RF EMFs exposure was
completed by Leszczynski et al. (2002). In this study, a human endothelial cell line (EA.hy926) was exposed to
900 MHz GSM-modulated RF for 1 h at a SAR of 2.0 W/kg and protein expression was assessed by 2D-gel
electrophoresis (4 independent experiments). A three-fold increase in the number of phosphoprotein spots were
observed in exposed samples, relative to the sham controls. Positive controls were not included in the study
design. Several phosphoproteins, including HSP27 and p38MAPK, were identified as demonstrating increased
expression after RF exposure, whereby the authors suggested that the apoptotic pathway regulated by
hsp27/p38MAPK might be the target of RF-EMF radiation. [Statistical analysis was not performed to determine
if any of these results were statistically significant. This study has been also reported in Section 12.3.3 (Gene and
protein expression).]
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
Buttiglione et al. (2007) assessed the expression of genes involved in apoptosis (egr-1, p53, apoptosis
inhibitor: bcl-2, survivin) and enzyme activities in human neuroblastoma (SH-SY5Y) cells after exposure to
GSM modulated 900 MHz RF EMF (SAR = 1 W/kg), at different time points (5 min to 24 h). In three
independent experiments, the egr-1- gene expression increased in 5 minutes, reaching maximum after 15 minutes
(p<0.01), and declined to baseline levels after 6 h. The enzyme activities of MAPK subtypes showed a similar
pattern (p<0.01). Cells exposed for 24 hours exhibited cell cycle progress typical for apoptosis (G2/M arrest)
accompanied by significant decreased gene expression of the investigated apoptosis inhibitor bcl-2 and survivin
(p<0.01). Positive controls were not included in the study design. [It was stated that an input power of 31.6 W
was fed into the exposure system, which would lead to an SAR of approximately 10 W/kg. Since the temperature
of the cell cultures was not measured during RF exposure and since the exposure details are not clearly reported,
the results of this study must be cautiously interpreted, as thermal confounding may have occurred. This paper
has also been described in section 12.3.2 (signal transduction).]
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
Jin et al. (2012b), in two independent experiments, exposed human promyelocytic leukemia HL-60
cells to CW 900 MHz RF EMF at a power density of 12 µW/cm2 for 1 hour per day for 3 days (calculated
average SAR = 2.5 10-5 W/kg) to evaluate apoptosis and the mitochondrial membrane potential (MMP). The
results indicated no effect when RF-exposed cultures were compared to sham-exposed controls. In this study the
effect of combined exposure was also investigated: on the day after the last RF exposure, cells were treated with
doxorubicin (DOX), a chemotherapeutic drug. Cells pre-exposed to RF and treated with DOX exhibited a
significantly decreased apoptosis (p<0.01) and an increase in MMP (p<0.01) compared to cultures treated with
DOX alone, indicating a protective effect of pre-exposure to RF. [The results of this study are based on just two
experiments. In this study viability, intracellular free Ca2+ and Ca2+-Mg2+-ATPase activity were also investigated
(see Section 12.3.2).]
5252
5253
5254
5255
5256
5257
Gurisik et al. (2006) exposed neuronal SK-N-SH cells to a 900 MHz GSM signal (SAR = 0.2 W/kg)
for 2 h. After 24 h from the RF exposure no effects were observed on the induction of apoptosis, as assessed in 6
independent experiments (YOPRO-1/PI staining). [The authors also reported that the cell cultures demonstrated
low viability (~65%) in both the sham and RF-exposed samples at 24 h post-exposure. Therefore, the validity of
these results is questionable. In this paper the effects of RF on gene expression and cell cycle distribution were
also investigated (see sections 12.3.3 and 12.3.6).]
5258
5259
5260
5261
5262
5263
5264
5265
Peinnequin et al. (2000) exposed human T-lymphoma-derived (Jurkat) cells to CW 2450 MHz RF
EMF at a power density of 5 mW/cm2. The exposure system consisted of an RF source, an amplifier and a horn
antenna irradiating cells hosted in 96-well culture plates inside a cell culture incubator. Sham samples were
hosted in the same incubator, separated by a RF-absorbing screen. After 48 h RF exposure, cell cultures were
treated with apoptotic agents, i.e. antagonist anti-Fas receptor antibody or sodium butyrate. The results obtained
on three independent experiments indicated that RF interacts with the Fas-induced apoptotic pathway (p<0.001)
but not with the ceramide butyrate-induced apoptotic pathway. [In this paper neither numerical nor experimental
dosimetry is reported. The SAR has been evaluated calorimetrically, but the authors do not describe the methods
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
188
5266
5267
5268
adopted to gain this result. The power density has been obtained by using a field meter, therefore measurements
were supposedly performed in absence of the sample. In this study the effect of RF EMF exposure on cell
proliferation was also investigated (see Section 12.3.6).]
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
Zhijian et al. (2013) exposed human B lymphoblastoid HMy2.CIR cells to an 1800 MHz GSM signal
for 24 h (5 min on/10 min off cycles) at 37 °C (SAR = 2.0 W/kg). The authors performed one experiment to
analyze the differential protein expression after RF exposure by using a protein microarray. Changes in the
expression of 27 proteins were detected related to apoptosis, DNA damage repair, oncogenesis, cell cycle and
proliferation (ratio >1.5-fold, p<0.05). In a further step, three independent experiments were carried by applying
Western blot analysis and a significant down-regulation of the Replication Protein A 32 (RPA32) (ratio>1.5fold, p < 0.05) was detected, while the expression of p73 (a transcription factor of p53 family) was significantly
upregulated (ratio >1.5-fold, p < 0.05). No difference of the Hypoxia Inducible Factor 1-a (HIF-1a) expression (a
protein involved in apoptosis induced hypoxia) was seen. [While RT-PCR was used to validate the results of
three experiments, microarray analysis was based upon only one experiment, thereby limiting the significance of
these results. This paper is also quoted in sections 12.3.1 and 12.3.6, were the effects of RF exposure on DNA
repair and cell proliferation are described.]
5281
5282
5283
5284
5285
5286
5287
5288
5289
Lee et al. (2005) exposed cultured human HL-60 cells to 2450 MHz pulse-modulated RF EMF at 10
W/kg for 2–6 h. The authors used serial analysis of gene expression (SAGE) to quantify gene transcript levels
after RF exposure. They found that 221 genes demonstrated altered expression after a 2-h exposure and 759 after
a 6-h exposure and among the up-regulated genes there were genes related to apoptosis (one independent
experiment). [There are an inadequate number of independent experiments. A 2-h sham-exposed sample was
used as a reference to compare against the 6-h RF-exposed sample. The results were based on differences in
fold-change. Statistical analysis of the gene expression data was not performed and results were not confirmed
by RT-PCR. This study has been also quoted in sections 12.3.3 and 12.3.6, were the effects of RF exposure on
protein expression and cell proliferation were described.]
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
Zhao et al. (2007) used a mobile phone either in the “on” mode (exposed) or in the “stand-by” mode
(sham) to expose primary mice neurons or astrocytes for 2 h to an 1900 MHz GSM signal. Array analysis and
real-time RT-PCR of three independent experiments showed the upregulation of caspase-2, caspase-6 and Asc
(apoptosis associated speck-like protein containing a card) gene expression in neurons and astrocytes. The
upregulation occurred in both “on” and “stand-by” modes in neurons, but only in “on” mode in astrocytes. Also
the up-regulation of the Bax gene was shown in astrocytes. No effects were detected on the expression of
caspase-9 in either neurons or astrocytes, or Bax in neurons. [There is an inadequate description of the RF
exposure system and dosimetry. Use of a mobile phone in “on” mode as the exposure source does not provide
appropriate control of the exposure level. Moreover, it is questionable whether placing the mobile phone in
stand-by mode is an appropriate sham control, and therefore whether the study fulfilled the inclusion criteria.]
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
Liu et al. (2011) investigated the possible protective effects of green tea polyphenols against RF EMF
in cultured rat cortical neurons exposed for 24 h to 1800 MHz. Also in this study a mobile phone in the “on”
mode was employed, while sham exposures were carried out in the “stand-by” mode. They found that RF
exposure induced cell death, evaluated with the MTT (3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyl-tetrazolium
bromide) and TUNEL assay. A protective effect of green tea polyphenols on the RF-exposed cortical neurons
was demonstrated by testing the content of Bcl-2 Assaciated X protein (Bax), as assessed by the
immunoprecipitation assay and Western blot assay. [There is an inadequate description of the RF exposure
system and dosimetry. Use of a mobile phone in “on” mode as the exposure source does not provide appropriate
control of the exposure level. Moreover, it is questionable whether placing the mobile phone in stand-by mode is
an appropriate sham control, and therefore whether the study fulfilled the inclusion criteria.]
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
Wu and coworkers (2012) used microwave exposure to study the interference of the fields with rat
spermatogenesis. Primary rat Sertoli cells were exposed to S-band microwave radiation (power density = 100
mW/cm2) for 4 min and the levels of certain cytokines were investigated. The authors suggest that elevated
levels of cytokines can affect the induction of apoptosis. The results showed that TNF, IL-1β and IL-6 were
increased in Sertoli cells after exposure. Germ cells co-cultured with exposed Sertoli cells showed a significantly
higher apoptosis rate than the control germ cells. In addition, germ cell apoptosis was associated with the
upregulation of Bax/Bcl-2 and caspase-3, which can interrupt spermatogenesis. The authors reported that the
microwave exposure induced elevated level of cytokines from Sertoli cells induced the lipid peroxidation
significantly in germ cells membrane. Also the MDA content in medium of germ cells co-cultured with radiated
Sertoli cells was up-regulated. The authors summarize that the results suggest that cytokines produced by
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
189
5320
5321
microwave-radiated Sertoli cells may disrupt spermatogenesis. [In this study no information is provided on the
exposure system and dosimetry.]
5322
5323
5324
5325
5326
5327
5328
5329
(Calabrò et al., 2012) exposed for 2 h and 4 h neuron-like cells, obtained by retinoic acid-induced
differentiation of human neuroblastoma SH-SY5Y cells, to 1800 MHz at an average SAR of 0.086 W/kg. For
exposure, cell cultures were placed at 3 cm from the antenna of a mobile phone. In four independent
experiments, no changes in the expression of caspase-3, measured by Western blotting, were detected in exposed
cultures compared to sham exposed controls. [A mobile phone was used to perform the exposure. No dosimetry
was carried out: the authors measured the electric and magnetic field due to the mobile phone in absence of the
sample, by means of a broadband measurement system. Moreover, it is not reported how the sham-exposures
have been carried out.]
5330
5331
5332
5333
5334
5335
5336
5337
Pacini and coworkers (Pacini et al., 2002) exposed cultured human skin fibroblasts to 902.4 MHz,
GSM signal (SAR=0.6 W/kg) to evaluate gene expression, determined by DNA microarray analysis. The
exposure consisted of placing culture plates for 1 h above a commercial cellular telephone; sham-exposed
samples were placed above a cell phone switched off. The results of four experiments indicated an increase in
genes related to apoptosis. [There is an inadequate description of the RF exposure system and dosimetry. Use of
a mobile phone in “on” mode as the exposure source does not provide appropriate control of the exposure level.
Moreover, it is questionable whether placing the mobile phone in stand-by mode is an appropriate sham control,
and therefore also whether the study fulfilled the inclusion criteria.]
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
Chinese Hamster V79 fibroblasts were employed by Ballardin et al. (2011) to evaluate the induction
of apoptosis after 15 min exposure to 2450 MHz, CW, at power densities of 50 and 100 W/m 2. Apoptotic cells
were visualized with the terminal deoxynucleotidyl transferase (TdT) FragEL DNA fragmentation detection
method, which is analogous to the TUNEL method. The results of three independent experiments showed a
statistically significant increase in the number of apoptotic cells at both power densities tested (p<0.001)
compared to sham exposed cultures. The effect was transient since no sign of apoptosis was seen when cells
were allowed to proliferate further 24 h after the cessation of RF exposure. [The authors refer to sham-exposed
cultures. However, from the description of the exposure system it looks like that only one GTEM was used as RF
applicator. This casts doubt on the fact that real sham exposures were actually carried out and therefore if the
study fulfilled the inclusion criteria. In this study the effect of RF exposure on spindle disturbances and cell
proliferation was also investigated (see sections 12.3.1 and 12.3.6).]
Table 12.3.12. In vitro studies assessing the effects of RF-EMF exposure on apoptosis
Cell type
Biological endpoint
Exposure conditions
Results
Comment
Reference
For results on
protein expression
and
neurodegenerative
disorders see 8.3
and 12.3.3.
Terro et al.
(2012)
Number of
independent
experiments
Primary
cerebral cortical
cells of rat
embryos
Morphological and
900 MHz, GSM
No effect.
biochemical (caspase- Average SAR 0.25W/kg
3) hallmarks;
24 h
autophagy
n=3
No information on
blinding of staff.
Primary human
melanocytes
and
keratinocytes
cells
Morphological (H&E
900 MHz, GSM
staining) and
Average SAR 2 W/kg
biochemical (caspase6h
3) hallmarks
No effect.
n=2-20
Rat neuronal
cells (PC12)
n=3
For oxidative stress Simon et al.
and cell
(2013)
proliferation and
differentiation see
12.3.5 and 12.3.6.
No information on
blinding of staff.
PE
1950 MHz, UMTS
SAR 10 W/kg
24 h
No effect immediately
after and 24 h post-RF
exposure.
For genotoxicity
see 12.3.1.
Zeni et al.
(2012b)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
190
Murine
fibroblasts
(L929)
Caspase-3 activity
872 MHz, CW and
GSM
No effect.
SAR 5 W/kg
n=3
1h
Human
DAPI, TUNEL-PI,
neuroblastoma Caspase-3 activity
(SH-SY5Y) cells
n=3
900 MHz, CW
SAR 2 W/kg
900 MHz, GSM
For ODC activity
and cell
proliferation see
12.3.3 and 12.3.6.
Höytö et al.
(2008b)
No information on
blinding of staff.
No effect immediately
after and 24 h post-RF
exposure.
No information on
blinding of staff.
Joubert et al.
(2006)
No effect immediately
after and 24 h post-RF
exposure.
No information on
blinding of staff.
Joubert et al.
(2007)
Increased apoptosis
(DAPI and TUNEL-PI)
immediately after and
24 h post-RF exposure.
2 °C increase in
RF-exposed
cultures.
Joubert et al.
(2008)
Average SAR 0.25
W/kg
24 h
Primary rat
DAPI, TUNEL-PI,
cortical neurons Caspase-3 activity
n=3
900 MHz, GSM
Average SAR 0.25
W/kg
24 h
Primary rat
DAPI, TUNEL-PI,
cortical neurons Caspase-3 activity,
AIF
n=3-5
900 MHz, CW
SAR 2 W/kg
24 h
No effect on caspase-3
activity.
Thermal effects
excluded by ad hoc
experiments.
Increase in AIF-positive
nuclei soon after and
24 h post-exposure.
Murine
neuroblastoma
(N2a) cells
Caspase-3 activity,
TUNEL-PI, PE
935 MHz, CW, GSMbasic, GSM-talk
SAR 2 W/kg
n=3
24 h
Human
lymphocytes
Caspase-3 activity,
900 MHz, GSM
PARP cleavage,
n=4
PE
Average SAR 1.35
W/kg
Human
lymphoblastoid
(Jurkat) cells
No effect immediately
after and at several
time-points between 0
and 48 h post-RF
exposure.
Increase in caspase-3
activity in proliferating
but not in resting cells.
1h
No effects on PARP
cleavage and PE.
915 MHz, GSM
No effect measured 24
and 48 h postexposure.
Moquet et al.
(2008)
Increase in
caspase-3 activity
not related to
apoptosis.
Palumbo et al.
(2008)
For cell cycle see
12.3.6.
n=5
Blood
lymphocytes
N=14a
DNA fragmentation
and morphological
analysis
Average SAR 37
mW/kg
813.56 MHz, iDEN
n=3
836.55 MHz, TDMA
Belyaev et al.
(2005)
For genotoxicity
see 12.3.1.
2h
Lymphoblastoid PE
Molt-4 cells
No difference
between healthy
and EHS donors.
No effect
SAR 24 W/kg
For genotoxicity
see 12.3.1.
Hook et al.
(2004b)
No information on
blinding of staff.
SAR 26 W/kg
847.74 MHz, CDMA
835.62 MHz, FDMA
SAR 3.2 W/kg
2, 3, 21 h
Human-derived
immune cell
lines (HL-60,
Mono-Mac-6,
TK6)
n=5
Neutral comet assay
1900 MHz, pulse
modulated
SAR 1 and 10 W/kg
No effects immediately
after and 6 and 24 h
post-RF exposure.
6h
For cytokine
expression and cell
cycle see 10.3,
12.3.2 and 12.3.6.
Chauhan et al.
(2007a)
(5 min on/10 min off
cycles)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
191
Human skin
cells and
reconstructed
human
epidermis
PE
900 MHz, GSM
No effect.
For protein
expression and
proliferation see
12.3.3 and 12.3.6.
Sanchez et al.
(2006b)
No effect.
Follow up study to
(Sanchez et al.,
2006b).
Sanchez et al.
(2007)
Average SAR 2 W/kg
48 h
n=5
Primary human
skin cells
PE
1800 MHz, GSM
Average SAR 2 W/kg
n=6
For protein
expression see
12.3.3.
Human
PE, p53 expression
glioblastoma
(A172) cells and
lung fibroblasts
(IMR-90)
2142.5 MHz, CW and
CDMA
No effect.
For signal
transduction and
gene expression
see 12.3.2 and
12.3.3.
No effects.
Blind procedure not Bourthoumieu et
indicated.
al. (2013)
Upregulation of some
genes.
For signal
transduction, gene
expression,
genotoxicity and
cell cycle see
12.3.1, 12.3.2,
12.3.3 and 12.3.6
SAR 0.08–0.8 W/kg
24–48 h
Hirose et al.
(2006)
n=3
Primary human
amniotic cells
Activation and
expression of p53
n=3
900 MHz, GSM
Average SAR 0.25, 1,
2, 4 W/kg
24 h
Mouse neural
progenitor stem
cells
n=6
Human dermal
papilla cells
DNA fragmentation,
mitochondrial
function, apoptosisrelated gene
expression
1710 MHz, GSM
Apoptosis-related
protein expression
1763 MHz, CDMA
n=3
Average SAR 1.5 W/kg
48 h
(5 min on/30 min off
cycles)
SAR 10 W/kg
1h/day for 7 days
No effect on other
parameters
investigated.
Increased expression of
Bcl-2 and
phosphorylation of
MAPK-1.
Nikolova et al.
(2005)
For signal
Yoon et al.
transduction,
(2011)
oxidative stress and
proliferation see
12.3.2, 12.3.5 and
12.3.6.
No information on
blinding of staff.
Studies including co-exposures
Human
Caspase-3 activity,
neuroblastoma DNA fragmentation
(SH-SY5Y) cells
872 MHz, CW and
GSM
No effect of RF
exposure alone.
SAR 5 W/kg
n=3
24 h
Mouse
fibroblasts
(L929)
Combined exposure
with menadione
(concurrent)
Increase in caspase-3
activity only in L929
cells co-exposed with
GSM signal.
For oxidative stress Höytö et al.
and cell
(2008a)
proliferation see
12.3.5 and 12.3.6.
No information on
blinding of staff.
n=3
Human
monocytes
(Mono-Mac-6)
cells
PE
n=3
No effect of RF alone.
Average SAR 2 W/kg
No effect of coexposures.
12 h
For protein
expression and cell
cycle see 12.3.3
and 12.3.6.
Lantow et al.
(2006c)
For gene
expression and cell
proliferation and
differentiation see
12.3.3 and 12.3.6.
Merola et al.
(2006)
Combined exposure
with PMA or glyoxin
(concurrent)
n=3
Human
neuroblastoma
(LAN-5) cells
1800 MHz, GSM-DTX
Caspase-3 activity,
900 MHz, GSM
No effect of RF alone.
PARP cleavage
Average SAR 1 W/kg
No effect of coexposures.
24 to 72 h
Combined exposure
with retinoic acid or
camptothecin
(concurrent)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
192
Human blood
mononuclear
cells
PE, MMP modification 1800 MHz, GSM-basic,
talk, DTX
No effect of RF alone.
No effect of coAverage SAR 1.4 and 2 exposures.
W/kg
n = 36
For protein
expression see
12.3.3.
Capri et al.
(2004a)
44 h
(10 min on/20 min off
cycles)
Combined exposure
with dRib (concurrent)
Human blood
mononuclear
cells
PE, MMP modification 900 MHz, CW and
GSM
n = 39
SAR 70 and 76 mW/kg
No effect of RF alone.
No effect of coexposures.
1 h/day for 3 days
For cell proliferation Capri et al.
see 12.3.6.
(2004b)
No information on
blinding of staff.
Combined exposure
with dRib (concurrent)
“No effect” means no statistically significant effect
Abbreviations: AIF: apoptosis inducing factor; CW: continuous wave; DAPI: 4',6-diamino-2-phenylindole; dRib: 2-deoxy-D-ribose;
DTX: Discontinuous transmission, EHS: electromagnetic hypersensitive; CDMA: code division multiple access; FDMA: frequency
division multiple access; iDEN: Integrated Digital Enhanced Network; IL: interleukin; Fas-L: Fas-ligand; GSM: Global System for
Mobile Communication; H&S: haematoxylin and eosin; MAPK: Mitogen-activated protein kinases; MMP: mitochondrial membrane
potential; MTT: (3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyl-tetrazolium bromide; PARP: Poly (ADP-ribose) polymerase; PE:
Phosphatidylserine externalization; PI: propidium iodide; PMA: phorbol 12-myristate 13-acetate; RT-PCR: Reverse transcriptasepolymerase chain reaction; SAGE: serial analysis of gene expression; SAR: specific absorption rate; TDMA: time division multiple
access; TNFα: tumor necrosis factor α; TUNEL: TdT-mediated (2'-deoxyuridine 5'-triphosphate) nick-end labeling; UMTS: universal
mobile telecommunications system.
5349
5350
Excluded papers
5351
5352
5353
(Agarwal et al., 2009; Atasoy et al., 2009; Avendano et al., 2012; Cao et al., 2009; Caraglia et al., 2005; Falzone
et al., 2010; Liu et al., 2012; Lu, Huang & Huang, 2012; Marinelli et al., 2004; Port et al., 2003; Shckorbatov et
al., 2010; Song et al., 2011; Yang et al., 2012; Zhou et al., 2008).
5354
12.3.5
5355
5356
5357
5358
Reactive oxygen species (ROS) are molecules containing oxygen that are highly chemically reactive
because they contain free radicals. The generation of ROS is a naturally occurring process in cell metabolism
and ROS levels are normally controlled by specific enzymes and antioxidants. Although ROS are required by
cells, an excess can be damaging and they have the potential to damage DNA, lipids and other biomolecules.
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
Oxidative stress is the imbalance between the production of pro-oxidants and antioxidants. ROS refer
to pro-oxidants and oxygen free radicals. Reactive nitrogen species (RNS), derived from nitric oxide (NO), and
superoxide are produced via the enzymatic activity of inducible nitric oxide synthase 2 (NOS2) and NADPH
oxidase, respectively. The formation and release of ROS is closely connected to the immune defense system and
especially to phagocytic processes. ROS formation is part of the cascade of events in the antimicrobial action of
phagocytic cells, called oxidative burst, which results from the assembly of a complex electron transport system
in the plasma-membrane. High levels of ROS can lead to a number of damaging pathological consequences in
cells and the organism, including lipid peroxidation, protein damage, deactivation of enzymatic activities, and
DNA modification as well as pro-inflammatory processes. In the normal cellular biochemistry there is a balance
between free radical formation and the action of an antioxidative system. A number of primary antioxidant
enzymes, such as different dismutases (SOD), catalases, reductases or peroxidases are known to neutralise the
amounts of ROS. Moreover, several compounds, such as reduced glutathione (GSH) and its oxidized form
(GSSG), scavenge free radicals. Irregularity or disturbance in the redox homeostasis by increased quantities of
ROS or by the inhibition of the action of antioxidants, can lead to cellular oxidative stress causing direct
oxidative damages in cells and tissues, and may also initiate inflammatory processes. Other modulations in cell
functions via signal transduction processes can furthermore be induced. Therefore, oxidative stress caused by the
formation of radical oxygen and nitrogen species plays a decisive role in cytotoxicity and inflammation
eventually leading to the onset of pathophysiological alterations and pathogenesis.
5377
5378
5379
Oxidative stress and/or ROS production can be investigated in vitro directly by studying the release of
the intermediates (flow cytometry) or by measuring the activation or expression of the involved proteins such as
antioxidants (SDS-PAGE). ROS production is usually measured by using the 2’,7’-dichlorofluorescin diacetate,
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
Oxidative stress
193
5380
5381
5382
oxidized by ROS to 2’,7’-dichlorofluorescein (DCFH-DA assay). On the other hand, these effects can also be
investigated indirectly by detecting e.g. molecular damage on the DNA or on proteins in different cell types by
various exposure conditions.
5383
5384
5385
5386
5387
5388
5389
5390
The previous WHO report on the effects of RF exposure issued in 1993 reported no study on oxidative
stress. The present literature search identified 32 relevant papers in this area, addressing the effect of RF EMF
exposure, either alone or in combination with other agents. Five of them were in a language that could not be
understood. Five papers were obtained from other sources. That left 32 papers to be extracted. Among the
relevant publications, 10 were excluded because they did not meet the inclusion criteria for in vitro studies, and
references are quoted at the end of this chapter. Two papers did not completely comply with the quality criteria
for inclusion due to methodological issues therefore they are only presented in the text. The remaining 20 papers
have been described in the text.
5391
5392
Unless specifically mentioned, papers did not report on blinding of the investigators to the exposure
conditions.
5393
5394
5395
5396
5397
5398
Hong et al. (2012) (2012) studied oxidative stress related effects in human MCF10A mammary
epithelial cells after exposure to either a single frequency (837 MHz alone or 1950 MHz alone) or multiple
frequencies (837 and 1950 MHz) at SAR value of 4 W/kg (4 W/kg for single signals and 2 W/kg plus 2 W/kg for
multiple signals) for 2 h (Hong et al., 2012). After exposure, cell cultures were kept in a cell culture incubator for
10, 24 and 48 h before evaluating SOD activity, GSH/GSSG ratio and ROS formation (DCFH-DA assay),
respectively. In five independent experiments, the authors were not identifying any effect.
5399
5400
5401
5402
5403
Yoon and co-workers (2011) studied the effect of 1763 MHz RF EMF, CDMA signal (SAR=10
W/kg) on cultured human dermal papilla cells by evaluating ROS release. ROS formation was not affected, as
measured in 3 independent experiments using DCFH-DA assay. Positive controls were not included in the study
design. This study has also been reported in Sections 12.3.2 (Signal transduction), 12.3.4 (Apoptosis) and 12.3.6
(cell proliferation).
5404
5405
5406
5407
5408
Poulletier de Gannes et al. (2011) (2011) investigated the induction of ROS in different neuronal cell
types after RF exposure to 1800 MHz, EDGE signal, (SAR =2 and 10 W/kg) for 1 and 24 h. The experiments
were carried out blinded. The production of ROS was measured (DCFH-DA assay) at the end of the 24-h
exposure or 24 h after the 1-h exposure. No increase in ROS production was detected after RF exposure.
Rotenone was used as positive control and induced significant ROS increase.
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
Xu and co-worker (2010) (2010) used primary cultured cortical neurons which were exposed to pulsed
RF fields at 1800 MHz, modulated by 217 Hz (SAR=2 W/kg) for 24 h, to study ROS induction (DCFH-DA
assay) and the indirect ROS effect, namely the mitochondrial DNA (mtDNA) damage. Blinded RF exposure
induced a significant increase in ROS production after 24 h of exposure (6 independent experiments; p<0.01)
and in the 8-hydroxyguanine (8-OHdG) level, a common biomarker for DNA oxidative damage in the
mitochondria of neurons (3 independent experiments; p<0.01). In parallel, the copy number of mtDNA and the
levels of mitochondrial RNA (mtRNA) transcripts showed an obvious reduction after RF exposure (6
independent experiments; p<0.01). These mtDNA disturbances could be reversed by pre-treatment with
melatonin, which is known to be an efficient antioxidant in the brain. This study has also been described in
section 12.3.1, where the effect of RF exposure on DNA fragmentation is reported. In a further study, the same
research group exposed mouse spermatocyte-derived (GC-2) cells to RF EMF at 1800 MHz, GSM signal
(average SAR = 1, 2 and 4 W/kg), for 24 h (5 min on/10 min off cycles) and measured ROS levels by applying
the DCF-DA assay, and oxidative DNA base damage by measuring the levels of 8-oxoG (Liu et al., 2013). RF
exposure, carried out blinded, increased the generation of ROS in a SAR-dependent manner (p<0.05), and 4
W/kg exposure also increased the levels of the DNA adduct 8-oxoG (p<0.05), as assessed in three independent
experiments. Pre-treatments with the antioxidant tocopherol blocked RF exposure-increased ROS production and
8-oxoG levels at a SAR value of 4 W/kg. This study has been also described in section 12.3.1 were the effect of
RF on genotoxicity has been reported. In both studies, treatment with hydrogen peroxide as positive control
resulted in a significant increase in ROS production.
5428
5429
5430
5431
5432
Lens epithelial cell cultures exposed to intermittent (5 min on/10 min off cycles) 1800 MHz RF EMF,
GSM signal (SAR = 1, 2, 3, 4 W/kg) for 2 h were used by Yao et al (2008a) to evaluate the induction of
oxidative stress. Moreover, RF was also superposed with 2 µT electromagnetic noise (30-90 Hz electromagnetic
fields in Helmholtz coils) for 2 hours. In three independent experiments, a significant increase in ROS formation
was detected after exposure at 2, 3 and 4 W/Kg (p<0.05), as assessed by the DCFH-DA assay. The superposed
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
194
5433
5434
5435
electromagnetic noise was able to block the RF-induced ROS formation. As positive controls, the cells were
treated with 4-nitroquinoline-1-oxide, which resulted in positive findings. This study has already been described
in detail in Section 6.4.2 (Ocular functions) and 12.3.1(genotoxicity).
5436
5437
5438
5439
5440
Simon et al. (2013) analyzed the influence of acute exposure to 900 MHz RF EMF, GSM signal (SAR
= 2 W/kg) on pigmented and non-pigmented skin cells and the influence of melanocytes on this response. Cell
cultures were exposed for 6 h and analyzed 2, 6 and 24 h post exposure. In a set of 2 to 20 experiments, no
effects were detected on ROS production at any examined time point, evaluated by looking at global protein
oxidation. This study has also been quoted in Sections 12.3.4 (apoptosis) and 12.3.6 (cell proliferation).
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
Ni et al (2013) investigated the induction of oxidative stress in human lens epithelial B3 (HLE-B3)
cells intermittently exposed (5 min on/10 min off cycles) to a 1800 MHz RF EMF, GSM signal (average SAR=2,
3 and 4 W/Kg). The ROS levels were measured (DCFH-DA assay) in cells exposed for 0.5, 1, and 1.5 h. Lipid
peroxidation was detected by a Malondialdehyde test (MDA, a member of a family of final products of lipid
peroxidation) in cells exposed for 6, 12, and 24 h. The mRNA expression of SOD1, SOD2, CAT, and
GPx1genes and the expression of SOD1, SOD2, CAT, and GPx1 proteins were measured by qRT-PCR and
Western blot assays in the cells exposed for 1 h. For all the experimental conditions tested, in the RF exposed
cultures ROS and MDA levels increased (P<0.05) and mRNA and protein expression significantly decreased
(P<0.05) in comparison to sham-exposed ones; cell viability also resulted decreased (three independent
experiments for each exposure condition/endpoint examined). Positive controls have not been included in the
study design. This study has also been described in Section 6.4.2 (Ocular functions).
5452
5453
5454
5455
5456
Xu and co-workers (2013) (2013) exposed human skin fibroblasts to RF EMF at 1800 MHz, GSM
signal (average SAR = 3 W/kg), for 24 h (5 min on/10 min off cycles). Intracellular ROS levels (DCF-DA assay)
were not affected by RF exposure, as assessed in seven independent experiments carried out blinded. As positive
controls, the cells were treated with 4-nitroquinoline-1-oxide, which resulted in positive findings. This paper is
also quoted in Sections 12.3.1 (Genotoxicity) and 12.3.6 (Cell proliferation).
5457
5458
In nine studies the effects of RF EMF exposure alone as well as in combination with other agents on
oxidative stress have been assessed.
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
Simkó et al. (2006) (2006) used human Mono Mac 6 cells which are capable to internalize e.g. ultra
fine particles (UFP). This phagocytic activity is connected to an increased release of free radicals. RF EMF at
1800 MHz, CW, 217Hz or GSM-nonDTX2 (SAR = 2 W/kg) was given for 1h, alone or in combination with
UFP. None of the employed RF exposure conditions induced any effect on free radical levels, as assessed by
measuring superoxide radical anions release (cytochrome c assay). Moreover, RF did not potentiate the
internalization of UFP (3 to 6 independent experiments). Heat treatment (42 °C) was used as positive control and
worked properly. This study, performed blinded, has also been described in section 12.3.3 (protein expression).
The same research group also examined ROS release in human umbilical cord blood-derived monocytes and
lymphocytes after blind exposure to 1800 MHz, CW or GSM modulated (GSM-DTX and GSM-Talk) at 2 W/kg
SAR for 30 or 45 min of continuous or intermittent exposure (Lantow et al., 2006a). In three independent
experiments, no RF effects were seen on ROS production with one exeption: The GSM-DTX signal (2 W/kg)
induced a significantly different ROS production if the data were compared to sham (1.3 fold increase), but not
when the comparison was carried out with respect to control cultures (incubator). However, this difference
appeared due to the lowered value of ROS release during sham exposure. In human lymphocytes, no differences
were detected at any RF exposure condition. Moreover, the authors also investigated the effect of co-exposures
to phorbol ester phorbol-12-myristate-13-acetate (PMA), a ROS inducer. No additional increase was detected
after co-exposures. Treatment with MPA alone also served as positive control. This study has also been
described in section 12.3.2 (signal transduction). In a further study the authors, using Human Mono Mac 6 and
K562 cells, reported similar findings, namely no RF induced ROS release with one exeption: the GSM-DTX
signal at 2 W/kg, on Mono Mac 6 (p<0.05) but not on K562 cells, where the sham value was lower for unknown
reason (4 independent experiments for each cell type investigated, carried out blinded). The result was also
confirmed when Mono Mac 6 cells were assayed for superoxide radical anion production, as assessed in 4-6
independent experiments (Lantow et al., 2006c). In co-exposure experiments to PMA or lipopolysaccharides
(LPS) Mono Mac 6 cells did not show additional ROS increase with respect to treatments PMA or LPS alone
(positive controls). This study has also been described in Section 12.3.3 (gene and protein expression).
5484
5485
5486
In a comprehensive study carried out by Hook et al. (2004) (2004a) oxidative stress was evaluated in
J774.16 mouse macrophages after exposure to RF at 835,62 MHz, FMCW, and at 847,74 MHz, CDMA, for 20
or 22 h (SAR = 0.8 W/kg). Moreover, J774.16 cells were stimulated with Ɣ-interferon (IFN) and LPS prior to
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
195
5487
5488
5489
5490
5491
5492
5493
exposure. Oxidative stress was evaluated by measuring oxidant levels, antioxidant levels, oxidative damage and
nitric oxide production. No changes in glutathione disulfide (GSSG) concentration, SOD activity or in catalase
and glutathione peroxidase activity were detected in RF exposed and co-exposed cultures with respect to their
respective sham samples (2 to 4 independent experiments for each condition investigated). Treatments with IFN
and LPS alone also served as positive control and worked properly. Consistent with the lack of an effect on
oxidative stress parameters, no change in viability was observed in J774.16 cells after either optimal (with or
without inhibitors of nitric oxide synthase) or suboptimal stimulation.
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
The effects of RF exposure and co-exposure were investigated by Hoyto et al. (2008) (2008a) in
human SH-SY5Y neuroblastoma cells and in L929 mouse fibroblasts. Cells were exposed to 872 MHz RF EMF,
CW or GSM signals (average SAR = 5 W/kg), for 1 or 24 h. RF was given alone or in combination with
menadione (Vitamin K3, a cell death and oxidative stress inducer) to induce ROS, or with tertbutylhydroperoxide (t-BOOH) to induce lipid peroxidation. In three independent experiments, no effects of RF
exposure alone were observed. Treatments with menadaion or t-BOOH alone also served as positive control and
gave positive findings. After co-exposure, the lipid peroxidation induced by t-BOOH was increased in SH-SY5Y
(p<0.05) but not in L929 cells. The combined effect was detected after GSM exposure. The authors concluded
that the positive findings may be due to chance, but they may also reflect effects that occur only in cells
sensitized by chemical stress. This study includes also data on apoptosis and cell proliferation and is therefore
quoted in section 12.3.4 and 12.3.6. [Although in the figures the authors do not call it “sham” the control is
actually a sham]. The same research group studied the effects of 1 h exposure to 872 MHz RF EMF, CW or
GSM, (SAR = 5 W/kg) in human SH-SY5Y neuroblastoma cells. RF exposure was carried out in the presence or
absence of menadione, a chemical inducing intracellular ROS production and DNA damage. ROS formation was
measured at time-points of 0, 5, 10, 15, 30 and 60 min after the end of RF exposure by applying the DCF-DA
assay (Luukkonen et al., 2009). The results of three independent experiments, carried out blinded, indicated no
differences in cultures exposed to RF alone compared to sham controls. Co-exposure to the CW RF EMF
showed an increased ROS level at 30 and 60 min (p < 0.01) in comparison to the cells exposed to menadione
only, but not in cultures co-exposed to the GSM signal. Treatments with menadione alone served as positive
control and induced a significant increase in ROS formation. In a further study (Luukkonen, Juutilainen &
Naarala, 2010) the authors applied the same exposure protocol but for 3 h and co-exposures were carried out
with ferrous chloride (FeCl2) and diethyl-maleate (DEM) (the latter enhances the free radicals induced by the
former, resulting in decreased antioxidant levels). No effects from either CW or modulated RF exposure or from
co-exposure with DEM or FeCl2 were observed (4 independent experiments, carried out blinded). These studies
have also been described in section 12.3.1 (genotoxicity).
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
In murine L929 fibrosarcoma cells no elevated ROS release was detected after exposure to RF EMF at
900 MHz, CW or GSM modulated (SAR = 0.3 and 1 W/kg) for 10 or 30 min, in the presence or absence of 3chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX) in the study of Zeni et al. (2007) (2007a). However,
exposure to MX alone resulted in a statistically significant increase in ROS formation since MX is an inductor of
oxidative stress in L929 cells. The formation of ROS (DCFH-DA probe ) was evaluated soon after the RF EMF
exposure and at several times up to 1 h from exposure/co-exposure (at 10, 20, 40 and 60 min in case of the 10min RF exposure and at 30, 45 and 60 min in the case of the 30-min RF exposure) (three independent
experiments for each condition tested). Treatments with MX alone served as positive control and resulted in a
significant increase in ROS production. In a further study, the same research group investigated ROS formation
in Jurkat cells exposed 1950 MHz RF EMF, UMTS signal (SAR = 0.5 and 2.0 W/kg) for 5, 60 min and 24 h and
co-exposure to ferrous ions (Brescia et al., 2009). Several co-exposure protocols were applied to test if RF
radiation is altering ROS formation induced by FeSO4 (RF given before or concurrently to FeSO4). In 3
independent experiments, carried out blinded, no effects were detected in either exposed and co-exposed
cultures, compared to their respective sham samples. Cell viability was also measured after 24 h RF exposure
and was not affected in any of the experimental protocols applied. Treatments with FeSO 4 alone served as
positive control and induced significant increase in ROS formation.
5535
5536
5537
5538
5539
5540
5541
5542
Del Vecchio et al (2009) investigated oxidative stress-related cell survival of SN56 neural cells and rat
primary cultured neurons after exposure to 900 MHz, GSM signal (average SAR=1 W/ kg), up to 72 h. The
exposure, performed blinded, was carried out in the presence or absence of 25-35 AA beta-amyloid fragments (a
major toxic event in Alzheimer disease) or hydrogen peroxide (H 2O2) by using the MTT assay. RF exposure
alone did not cause any effect. The same results were obtained when cell cultures were exposed to RF in
combination with 25-35 AA beta-amyloid fragments, as assessed in 2 to 4 independent experiments. Coexposure to H2O2 resulted in a strong reduction of the number of living cells in SN56 (p<0.001) but not in
primary neurons (3 independent experiments). Treatments with H2O2 alone served as positive control and
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
196
5543
5544
worked properly. This study has been also described in section 12.3.6 (cell proliferation) and 8.3
(neurodegenerative disorders).
5545
5546
In one paper millimiter waves were employed to evaluate the induction of oxidative stress in exposed
and co-exposed cultures.
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
Safronova et al (2002) investigated the effect of 20 min CW irradiation at 41.95 GHz on neutrophils
from the NMRI strain mice. The exposure was carried out at a calculated SAR of 0.45 W/kg for incident power
density of 19.5 mW/cm2. Exposure of resting neutrophils did not influence the basal ROS level (by the luminol
dependent chemiluminescence technique), in 22 independent experiments. However, the RF exposure with Nformylmethionyl-leucyl-phenylalanine (fMLP) primed cells resulted in a significant increase of the fMLP
response (in 14 independent experiments). Treatment with fMLP alone served as positive control and induced
significant increase in ROS production. The use of the inhibitors of serine/threonine protein kinases (H-7),
tyrosine protein kinases (W-7), and Ca2+ /calmodulin-dependent enzymes (tyrphostin 51) showed that H-7 and
tyrphostin 51 inhibited the RF induced cell response, whereas W-7 increased it. The authors summarized that the
RF induced effect on primed neutrophils but not in resting cells is due to the shift from one state to the other via
changes in the protein kinase activity leading to altered cell susceptibility to RF exposure.
5558
Studies not included in the analysis
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
Friedman and co-worker (2007) investigated the molecular mechanism for RF-induced MAPK
activation by exposing Rat1 cells and serum-starved HeLa cells to 800, 875 and 950 MHz RF EMF at power
densities of 0.005, 0.03, 0.07, 0.110 and 0.310 mW/cm2 for 5-30 min). The observed results showed that ERKs
were rapidly activated in response to RF exposure at various frequencies and intensities (p values not reported).
By using different inhibitors their result showed that this effect was mediated by EGFR and ROS, which is
produced by NADH oxidase after RF exposure. [Statistical analysis of densitometry data was not performed.
There is an inadequate description of the RF exposure system and dosimetry. Use of a mobile phone in “on”
mode as the exposure source does not provide appropriate control of the exposure level. Moreover, it is not
reported how the sham-controls have been performed. This study has also been reported in section 12.3.2 (signal
transduction).]
5569
5570
5571
5572
5573
5574
5575
Campisi et al. (2010) (2010) exposed astroglial cells from new-born rat brains for 5, 10 or 20 min to
900 MHz, CW or amplitude modulated at 50 Hz, at the same power density of 0.26 W/m2 (no SAR reported). A
significant increase in ROS levels (p<0.001) was observed after modulated exposure for 20 min. No effects were
detected when shorter exposure duration or CW were used (five independent experiments). In this investigation
the effect of RF exposure on genotoxicity and cell viability were also evaluated (section 12.3.1). Positive
controls have not been included in the study design. [The absence of dosimetry makes the results of this study
uninterpretable.]
12.3.13. In vitro studies assessing effects of RF EMF exposure on oxidative stress
Cells
Biological endpoint
Exposure conditions
Results
Comment
Authors
SOD, GSH/GSSG,
ROS (DCFH-DA
assay)
837 or 1950 MHz, or
multiple frequencies
(837 and 1950 MHz).
No effect.
No information on
blinding of staff.
Hong et al.
(2012)
No effect
For Signal
transduction,
apoptosis and
proliferation see
Sections 12.3.2,
12.3.4 and 12.3.6
Yoon et al.
(2011)
Number of
independent
experiments
Human MCF10A
mammary epithelial
cells
n=5
SAR 4 W/kg (single).
SAR 2 W/kg + 2 W/kg
(multiple)
2h
Human dermal
ROS (DCFH-DA
papilla cells (hDPC) assay)
1763 MHz, CDMA
n=3
1h
SAR 10 W/kg
No information on
blinding of staff.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
197
Human brain cell
lines (SH-SY5Y,
U87 and CHME5)
ROS (DCFH-DA
assay)
1800 MHz, GSM
(EDGE)
SAR 2, 10 W/kg SAR
primary cortical
neuron cultures
1 or 24 h
n=3
Rat primary cortical
neurons
n=3-6
ROS (DCFH-DA
1800 MHz pulse
assay), Mitochondrial
modulated by 217 Hz
DNA oxidative damage SAR 2 W/kg
(8-OHdG)
24 h
mtRNA transcripts
Mouse
ROS (DCFH-DA
spermatocyteassay)
derived (GC-2) cells oxidative DNA base (8n=3
oxoG)
1800 MHz, GSM
average SAR 1, 2 or 4
W/kg
24 h (5 min on/10 min
off cycles)
No effect
immediately
after 24 h RF
exposure or 24
h after 1 h RF
exposure
Poulletier de
Gannes et al.
(2011)
Increase in
ROS and 8OHdG levels;
reduction in
mtDNA and
mtRNA
transcripts,
reverted by
pretreatment
with melatonin
For gentoxicity see Xu et al. (2010)
section 12.3.1
SARdependent
increase in
ROS
generation;
For gentoxicity see Liu et al. (2013)
section 12.3.1
Increased
adduct 8-oxoG
at 4 W/kg
Effect blocked
by pretreatments with
tocopherol
Lens epithelial
(hLEC) cells
ROS (DCFH-DA
assay)
n=3
1800 MHz, GSM
average SAR 1, 2, 3, 4
W/kg
2 h (5 min on/10 min off
cycles)
MF noise superposition
RF alone
For gentoxicity see Yao et al.
increased ROS section 12.3.1
(2008a)
at 2, 3 and 4
W/kg
MF noise
negated RFinduced ROS.
(2 µT, 30-90 Hz)
Primary human
melanocytes and
keratinocytes cells
ROS production (global 900 MHz, GSM
protein oxidation)
average SAR 2 W/kg
6h
n=2-20
No effect, as
assessed 2, 6
and 24 h post
RF exposure
For apoptosis and
cell proliferation
and differentiation
see sections
12.3.4 and 12.3.6
Simon et al.
(2013)
No information on
blinding of staff.
Human lens
epithelial (HLE-B3)
cells
n=3
ROS (DCFH-DA
assay)
1800 MHz, GSM
Increase of
average SAR 2, 3 and 4 ROS and lipid
peroxidation
Lipid peroxidation
W/kg
(MDA test)
0.5, 1, and 1.5 h (ROS) Decrease of
gene and
Gene and protein
6, 12, and 24 h (lipid
protein
expression of SOD1,
peroxidation)
expression
SOD2, CAT, and GPx1
1 h (gene and protein
expression)
Decreased cell
viability
Ni et al. (2013)
No information on
blinding of staff.
(5 min on/10 min off
cycles)
Human skin
fibroblasts
ROS (DCFH-DA
assay)
n=7
1800 MHz, GSM
No effect
For gentoxicity and Xu et al. (2013)
cell proliferation
see Sections
12.3.1 and 12.3.6
No effect
For gene and
protein expression
see section 12.3.3
average SAR 3 W/kg
24 h (5 min on/10 min
off cycles)
Studies including co-exposures
Human monocyte
Mono Mac 6 cells
n=3-6
superoxide radical
anion (cytochrome C
assay)
1800 MHz, CW or 217
Hz or GSM-nonDTX
SAR 2 W/kg
Simkó et al.
(2006)
1h
Combined exposure
with ultrafine particles
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
198
Human umbilical
cord blood-derived
monocytes and
ROS (rhodamine
assay)
1800 MHz GSM (DTX
hearing only and Talk)
SAR 2 W/kg
lymphocytes
30, 45 min
n=3
continuous or
intermittent (5 min on/5
min off cycles)
Continuous or
intermitten
exposure to
GSM-DTX
signal caused
an increase in
ROS
production in
monocytes. No
effect in
lymphocytes
The effect seems
to be due to the
lowered ROS
value during sham
exposure.
Lantow et al.
(2006a)
For gene and
protein expression
see section 12.3.3
Co-exposures with PMA No effects of
co-exposures
Human Mono Mac
6 and K562 cells
ROS (rhodamine
assay)
n=4-6
1800 MHz GSM, non
DTX (speaking only),
DTX (hearing only), Talk
(34% speaking and 66%
hearing)
SAR 0.5, 1.0, 1.5 and
2.0 W/kg
GSM-DTX
signal at 2
W/kg produced
an increase in
ROS
production.
30 or 45 min
No effects of
Co-exposures with PMA co-exposures
or LPS
Mouse J774.16
macrophage cells
n=2-3
NO, GSSG, SOD,
catalase and
gluthatione peroxidase
activity
835.62 MHz, FMCW
847.4 MHz, CDMA
n=3
For gene and
protein expression
see section 12.3.3
No information on
blinding of staff.
Hook et al.
(2004a)
For apoptosis and
cell proliferation
see Sections
12.3.4 and 12.3.6
Höytö et al.
(2008a)
20 or 22 h
GSH, lipid peroxidation 872 MHz CW or GSM
mouse L929
fibroblast
Lantow et al.
(2006c)
SAR 0.8 W/kg
Co-exposures with IFN
or LPS
Human SH-SY5Y
neuroblastoma
cells;
No effect
The effect seems
to be due to the
lowered ROS
value during sham
exposure
No effects of
co-exposures
No effect
SAR 5 W/kg
1 or 24 h
Co-exposures with
No information on
menadione and t-BOOH Increased tblinding of staff.
BOOH-induced
lipid
peroxidation in
SH-SY5Y (but
not in L929)
cells after coexposure with
GSM. No
effects on
GSH.
Human SH-SY5Y
ROS (DCFH-DA
neuroblastoma cells assay)
872 MHz CW or GSM
n=3
1h
No effect
SAR 5 W/kg
Co-exposure with
menadione (concurrent)
For genotoxicity
see Section
12.3.1.
Luukkonen et al.
(2009)
No information on
Co-exposure to blinding of staff.
CW increased
ROS formation
30 and 60 min
after RF
exposure.
No effects with
the GSM signal
Human SH-SY5Y
ROS (DCFH-DA
neuroblastoma cells assay)
872 MHz CW or GSM
n=4
3h
SAR 5 W/kg
Co-exposure with DEM
or FeCl2 (concurrent)
No effect
For gene and
protein expression
see section 12.3.3
Luukkonen et al.
(2010)
No information on
blinding of staff.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
199
Murine L929
fibrosarcoma cells
ROS (DCFH-DA
assay)
n=3
900 MHz, CW and GSM No effect
No information on
blinding of staff.
Zeni et al.
(2007a)
No effect
Cell viability also
not affected.
Brescia et al.
(2009)
No effect
For
Del Vecchio et
Neurodegenerative al. (2009)
disorders and cell
proliferation see
Sections 8.3 and
12.3.6.
SAR 0.3 and 1 W/kg
10 and 30 min
Co-exposure with MX
(concurrent)
Human
lymphoblastoid T
cells (Jurkat)
ROS (DCFH-DA
assay)
1950 MHz, UMTS
SAR 0.5 and 2.0 W/kg
5, 60 min and 24 h
n=3
Co-exposure with
FeSO4
Mouse SN56 neural oxidative stress-related 900 MHz, GSM
cells
survival (MTT assay)
Average SAR 1 W/kg
Primary cultures of
up to 72 h
cortical neurons
Co-exposure with 25-35
n=2–4
AA or H2O2
No effect with
25-35 AA
Reduction of
living SN56
cells coexposed to
H2O2
Neutrophils from
the NMRI strain
mice
n=14-22
ROS
(chemiluminescence)
41.95 GHz, CW
No effect in
resting cells.
calculated SAR
No information on
blinding of staff.
Safronova et al.
(2002)
0.45 W/kg
20 min
Co-exposure with fMLP
Increased ROS
response to
fMLP in RF
exposed cells
primed with
low doses of
fMLP.
Increase or
decrease of RF
response on
the bases of
the protein
kinases
employed
“No effect” means no statistically significant effect
5576
5577
Excluded papers
5578
5579
5580
(Agarwal et al., 2009; Cao et al., 2009; De Iuliis et al., 2009; Falzone et al., 2010; Gajski & Garaj-Vrhovac,
2009; Lu, Huang & Huang, 2012; Naziroğlu et al., 2012; Orel et al., 2004; Yao et al., 2008b; Zmyślony et al.,
2004)
5581
12.3.6
5582
5583
5584
5585
5586
5587
Cell proliferation is the increase in cell number as a result of cell growth and division. Cell
proliferation, differentiation and apoptosis are fundamental processes in multicellular organisms and are tightly
connected to each other. Increased cell numbers may result from increased proliferation or from decreased cell
death. Cell proliferation can be stimulated by physiologic and pathologic conditions and is largely controlled by
signals from the microenvironment that either stimulate or inhibit proliferation. An excess of stimulators or a
deficiency of inhibitors leads to net growth and, in the case of cancer, uncontrolled growth.
5588
5589
5590
5591
5592
5593
Cell proliferation is a complex process that is under the control of multiple cell signal transduction
pathways. Maintaining the integrity of genetic information during cell proliferation is fundamental for living
systems. It is therefore vital for cells that DNA damage, induced by spontaneous hydrolytic events or by
radiation or chemical mutagens, is effectively recognized and repaired efficiently. Unrepaired or inaccurately
repaired DNA can lead to cell death (necrosis or apoptosis) as well as to genomic instability, mutations and
ultimately to cancer. The accurate assessment of cell number and cell proliferation is useful in many high content
Proliferation, cell cycle and differentiation
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
200
5594
5595
assays and is a key endpoint in cytotoxicity. Furthermore, an alteration in cell proliferation is also a very
sensitive indicator of cellular stress.
5596
5597
5598
Cellular differentiation is the process by which cells acquire the ability to perform their specialized
functions. Differentiation induces changes in cell shape, size and metabolic activity. The level of cellular
differentiation is used as a measure of cancer progression.
5599
5600
5601
Many RF EMF-related studies applied the cellular viability test or the analysis of the cell cycle and/or
the cell proliferation by using different technologies either to investigate the endpoint itself or to control the
experimental conditions at cellular level.
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
Laboratory assays employed to investigate alteration of cell proliferation are based on the evaluation
of specific features at cellular, cell physiological and/or molecular level. Such tests measure the number of viable
cells (proliferation index, mitotic index, trypan blue exclusion method, alamar blue assay, 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, MTT and (3-(4,5-dimethylthiazol-2-yl)-5-(3carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium), MTS assays), or investigate the different cell cycle
phases (propidium iodide staining, PI, bromodeoxyuridine incorporation, BrdU, tritiated thymidine
incorporation, 3[H]thymidine) by applying microscopic, colorimetric or flow cytometric techniques. Moreover,
to investigate the transcription or expression level of specific genes or proteins molecular assays are often used
(gene or protein arrays, RT-PCR, Western blot). Good candidates for investigations are cyclins and their
associated cyclin-dependent kinases (CDKs) and inhibitors that are the main components of the cell cycle
machinery. The tumor suppressor p53 is a common candidate to be investigated since it influences the cell cycle
at the G1/S regulation.
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
The previous WHO monograph (1993b) reported that an effect on cell proliferation, particularly in
relation to tumour promotion, by interactions other than heating, has not been established. The present literature
search identified 62 relevant papers in this area, addressing the effect of RF EMF exposure, either alone or in
combination with other agents. Six of them were in a language that could not be understood. Twenty-two papers
were obtained from other sources. That left 78 papers to be extracted. Among the relevant publications, 16 were
excluded because they did not meet the inclusion criteria for in vitro studies, and references are quoted at the end
of this chapter. Fifteen papers did not completely comply with the quality criteria for inclusion due to
methodological issues, therefore they are only presented in the text. The remaining 56 papers have been
described in the text and summarized in tables 12.3.6.1 (Proliferation and cell cycle) and 12.3.6.2
(Differentiation). Unless specifically mentioned, papers did not report on blinding of the investigators to the
exposure condition.
5625
12.3.6.1 Proliferation and cell cycle
5626
5627
5628
Most of the studies on cell proliferation and cell cycle have been conducted at frequencies in use for
wireless communications. Moreover, few investigations have tested the effect of higher frequencies (millimeter
waves and THz).
5629
5630
5631
5632
5633
5634
Bourthoumieu et al. (2013) investigated the expression of p53 (a tumor suppressor protein, regulator
of cell cycle) and its activation in human amniotic cells exposed for 24 h to 900 MHz RF EMF, GSM
modulation (SAR = 0.25, 1, 2 and 4 W/kg). The results of three independent experiments performed using three
different donors showed no effect in p53 expression (Western blot assay) by comparing sham-exposed to RFexposed cultures.. In this study, also described in sections 12.3.2 and 12.3.4, the effect of RF exposure on signal
transduction and apoptosis has been also investigated.
5635
5636
5637
5638
5639
5640
Palumbo et al. (2008) investigated the effect of 1 h exposure to 900 MHz RF EMF, GSM modulation,
(SAR = 1.35 W/kg) on cell cycle kinetics of human peripheral blood lymphocytes and human lymphoblastoid
Jurkat T-cells. Cell cycle analysis was performed 6, 24 and 48 h after the exposure using BrdU incorporation and
propidium iodide (PI) staining. The results of three independent experiments for each cell type, performed
blinded, indicated that cell cycle distribution was not affected by the RF exposure. Positive controls have not
been included in the study design. In this study, also quoted in Section 12.3.4, apoptosis was also investigated.
5641
5642
5643
Capri and co-workers (2004b) exposed human blood mononuclear cells from 25 healthy donors to
continuous or GSM modulated 900 MHz RF EMF (SAR= 70 and 76 mW/kg, respectively) 1 h/day for 3 days.
No changes in cell proliferation (3[H]thymidine incorporation; 25 donors) and in cell cycle progression (BrdU
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
201
5644
5645
incorporation and PI staining; 13-21 donors) were detected in exposed cultures compared to their respective
sham-controls. In this study the effect of RF EMF on apoptosis was also investigated (see section 12.3.4).
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
Zeni et al. investigated the effect of 900 MHz, GSM signal, at an average SAR of 0.3 or 1 W/kg by
exposing unstimulated human lymphocytes from 4 donors for 2 h (Zeni et al., 2005). Cell cycle analysis and
proliferation index were evaluated on microscope slides set up to investigate sister chromatid exchanges and
chromosomal aberrations, respectively. No differences between RF EMF exposed and sham controls were
detected. In a further study the same research group conducted an investigation with the participation of
researchers from another independent laboratory (Scarfi et al., 2006). Lymphocytes from blood samples of 10
donors were exposed double blind for 24 h to 900 MHz RF EMF, GSM signal, at an average SAR of 1, 5 or 10
W/kg, and then cultured for 72 h. The data from both laboratories did not indicate significant changes in cell
proliferation, as assessed by the cytokinesis-block proliferation index (CBPI) at all SARs examined when
compared with sham-controls. In these studies, the induction of genotoxic effects has been also investigated,
which is described in Section 12.3.1.
5657
5658
5659
5660
5661
Sanchez et al. (2006b) exposed reconstructed human epidermis cells to 900 MHz GSM-modulated RF
EMF at a SAR of 2 W/kg for 48 h. Cell proliferation was evaluated immediately after RF exposure by applying a
direct immunofluorescence technique. In 5 independent experiments no variation in RF and sham-exposed cell
cultures was detected. In this study protein expression and apoptosis has also been investigated (sections 12.3.3
and 12.3.4).
5662
5663
5664
5665
5666
5667
5668
5669
Moisescu et al. (2008) exposed metastatic murine B16F10 melanoma cells to 900 MHz RF EMF,
GSM signal (SAR = 2.2 W/kg) to measure the duration of every phase of mitosis and the total mitosis duration.
The exposure duration was 1 h, at the beginning or in the middle of image acquisition and two exposure
protocols were used: (i) 1 h GSM exposure + 2 h control after exposure and (ii) 2 h control before exposure + 1 h
GSM exposure + 2 h control after exposure. No significant perturbation of total mitosis duration or every mitosis
phase duration was found during the 1 h GSM-EMF exposure. Also, positive controls have not been included in
the study. [The number of independent experiments is not reported, although data are presented as mean ± SD
and statistical analysis has been performed.]
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
Kwee and Rasmark (1998b) exposed transformed human epithelial amnion (AMA) cells to a 960
MHz, GSM signal, at SAR values of 0.021, 0.21 and 2.1 mW/kg. The duration of exposure, which was carried
out at 37 °C, was 20, 30 or 40 minutes for each SAR value. After exposure, cells were allowed to grow for
another 24 h and cell proliferation was evaluated by applying the MTS assay. The results from 12 independent
experiments showed that a decrease in cell growth was induced at all three SAR values investigated (p<0.05) and
it was not dependent on the SAR value applied. In a follow up study, employing the same exposure system, the
authors exposed AMA cells for 30 min at the highest SAR (2.1 mW/kg) but at temperature two degrees higher
(39 °C) and lower (35 °C) than in the previous study (Velizarov, Raskmark & Kwee, 1999). The authors stated
that the exposure to RF EMF resulted in a significant changes in cell proliferation (p<0.1; 11 independent
experiments) and these changes were almost of the same order of magnitude as those reported in the previous
study under isothermal conditions (37 °C). In both papers positive controls have not been included in the study
design. [While in the first study the confidence level was fixed at 95%, in the latter study it was 90%, therefore a
direct comparison between studies cannot be made.]
5683
5684
5685
5686
5687
5688
5689
Stagg et al. (1997) exposed normal and transformed (C6) rat glial cells to 836.55 MHz TDMAmodulated RF EMF (average SAR = 0.59, 5.9 and 59 mW/kg). Cell proliferation was measured immediately
after RF exposure and at 1, 5, 7, 9, 12 and 14 days post exposure ( 3H-thymidine incorporation). No effects were
detected in both cell types investigated, as assessed in 2 (normal cells) and in 3 to 6 (transformed cells)
independent experiments. In this investigation DNA synthesis was also investigated. In a sub-set of experiments
(3 up to 8) increased DNA synthesis was observed in transformed but not in normal cells exposed for 24 h to 5.9
mW/kg (p<0.05). Positive controls have not been included in the study.
5690
5691
5692
5693
5694
5695
Lee et al. (2008) exposed NIH3T3 mouse fibroblasts to 849 MHz RF EMF at average SAR values of
2 or 10 W/kg for 1 h, or for 1 h per day for 3 days. Cell proliferation and cell cycle distribution were analysed at
24 and 48 h after exposure, by applying a colorimetric assay (incorporation of BrdU into genomic DNA) and by
flow cytometry (PI staining), respectively. No statistically significant differences between sham-exposed and
RF-exposed cells were detected for both cell proliferation (4 independent experiments) and cell cycle
distribution (6 independent experiments).
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
202
5696
5697
5698
5699
5700
Higashikubo et al. (2001) evaluated cell cycle progression of mouse fibroblasts C3H 10T1⁄2 and
human glioma U87MG cells exposed to a frequency-modulated CW at 835.62 MHz or a CDMA RF EMF
centered on 847.74 MHz at an average SAR of 0.6 W/kg. They investigated the effect of short (13 h) and long
(100 h) exposure duration. No variation in cell cycle distribution was detected for any of the experimental
conditions investigated (3 independent experiments), as assessed by BrdU incorporation and PI staining.
5701
5702
5703
5704
5705
5706
Schrader et al. (2011) conducted a study using the well-established human-hamster hybrid (FC2) cells
containing a single copy of human chromosome 11. They exposed cell cultures for 30 min to RF EMF at
frequency of 900 MHz (SAR = 10.7–17.2 mW/kg, E-field strengths 45 or 90 V/m). The exposure equipment had
separate electric- (E) and magnetic-field (H) components. Cell proliferation, evaluated by determining the
mitotic index, was not changed in any of the experimental conditions tested. In this study the induction of
genotoxic effects was also investigated (Section 12.3.1).
5707
5708
5709
5710
5711
5712
Kumar et al. (2011) excised femur and tibia bones from 11 rats and exposed them for 30 min to 900
MHz RF EMF, CW (SAR = 2 W/kg). Cell proliferation was evaluated in whole bone marrow cells and
lymphocytes extracted from exposed bones after 96 and 72 h of culture, respectively. No differences were
detected in exposed samples compared to sham-exposed ones, as assessed by trypan blue exclusion method. In
this study, the effect of RF exposure on erythrocyte maturation rate and DNA strand breaks was also investigated
(see sections 10.3 and 12.3.1).
5713
5714
5715
5716
5717
5718
5719
5720
Xu and co-workers (2013) exposed human skin fibroblasts to an 1800 MHz GSM signal (average
SAR = 3 W/kg), for 24 h (5 min on/10 min off cycles). Cell cycle progression was measured soon after and 6
and 12 h after RF exposure by flow-cytometry (PI staining). The results of 5 independent experiments indicated
that cell cycle distribution was not affected immediately after RF exposure. At 6 h after exposure a slight
increase in G0/G1 arrest occurred compared to sham-exposed cultures (p<0.05), that was not detected after 12 h
from the exposure. Moreover, no effects on cell proliferation were found by counting cells after 12, 24, 36 and
48 h after 24 h RF exposure (3 independent experiments). This paper is also quoted in sections 12.3.1 and 12.3.5,
were the effects of RF on genotoxicity and oxidative stress are reported.
5721
5722
5723
5724
5725
5726
Lantow et al. (2006b) investigated whether exposure to an 1800 MHz GSM-DTX signal, at an SAR of
2.0 W/kg for 12 h, could altered cell cycle progression and kinetics in Mono-Mac-6 cells. No significant effects
of RF exposure were detected, immediately and 12, 24, 36, 48 and 60 h after exposure, as evaluated by PI
staining (4 independent experiments) and BrdU incorporation (6 independent experiments). In this study, also
quoted in sections 12.3.3 and 12.3.4, the effect of RF EMF on protein expression and apoptosis was also
investigated.
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
Huang et al. (2008b) exposed a mouse auditory hair cell line (HEI-OC1) to 1763 MHz CDMAmodulated RF EMF at an SAR of 20 W/kg for 24 and 48 h. No alterations in cell cycle distribution were
observed in 3 independent experiments, as assessed by flow-cytometry (PI staining). In a related study, the same
research group exposed human T-cell derived (Jurkat) cells for 1 h per day for 1, 2 or 3 days to 1763 MHz
CDMA-modulated RF at an SAR of 2 or 10 W/kg. After the last exposure, cells were incubated for 24 h before
harvesting (Huang et al., 2008a). The results indicated that RF exposure did not produce significant changes in
cell proliferation (3 independent experiments) and cell cycle distribution (6 independent experiments), as
assessed by monitoring cell numbers and by PI staining, respectively. In these papers DNA damage (section
12.3.1), signal transduction (section 12.3.2), and HSPs and gene expression (section 12.3.3) were also
investigated. Moreover, the latter study has been also described in Chapter 6 (section 6.4.1).
5737
5738
5739
5740
5741
5742
5743
5744
5745
Yoon and co-workers (2011) studied the effect of a 1763 MHz CDMA signal (SAR = 10 W/kg) on
cultured human dermal papilla cells (hDPC) by evaluating cell cycle progression using PI staining. Changes in
the expression of protein marker related to hair growth and apoptosis were also investigated. In 3 independent
experiments, no differences in cell cycle distribution were detected between sham and RF-exposed cultures. The
expression of insulin-like growth factor-1 (IGF-1) mRNA in hDPC was significantly induced upon RF exposure,
resulting in an increased expression of Bcl-2, cyclin and increased phosphorylation of MAPK-1 protein (p<0.05).
Positive controls have not been included in the study design. This study has also been discussed in sections
12.3.2, 12.3.4 and 12.3.6, where the effects of RF exposure on signal transduction, apoptosis and oxidative stress
are reported.
5746
5747
Zeni et al. (2008) did not observe changes in cell proliferation of human lymphocytes from 6 donors
intermittently exposed (6 min on/2h off cycles) to a 1950 MHz UMTS signal (SAR = 2.2 W/kg). The exposure
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
203
5748
5749
duration was from 24 to 68 h to cover different stages of the cell cycle. Cell proliferation, evaluated as CBPI,
was calculated on microscope slides set up to measure the induction of micronuclei (see Section 12.3.1).
5750
5751
5752
5753
5754
5755
5756
Chauhan et al. (2007a) used exponentially growing human leukaemia (HL-60), monocyte (MonoMac-6) and lymphoblastoma (TK6) cells to assess cell cycle alterations (PI staining) after exposure to 1900
MHz, pulse-modulated RF EMF for 6 h (5 min on/10 min off cycles) at an SAR of 1 or 10 W/kg. In 5
independent experiments, no change in cell cycle kinetics was detected immediately after exposure and 6 and 24
h later in any of the cell lines tested. In this study, described in more detail in sections 12.3.2, 12.3.6 and in
Chapter 10, section 10.3, the effects of RF EMF exposure on apoptosis and cytokine expression were also
investigated.
5757
5758
5759
5760
5761
Miyakoshi et al. (2005) exposed human malignant glioma cells (MO54) to 1950 MHz continuous RF
EMF for 2 h at SARs of 2 and 10 W/kg. Proliferation was measured in 2 independent experiments by recording
cell number immediately after RF exposure and after 1, 2, 3 and 4 days from the exposure. No effects were
detected: exposed and sham-exposed cells demonstrated similar proliferation pattern. This study has already
been described in Section 12.3.3, where the results on gene and protein expression are reported.
5762
5763
5764
5765
5766
5767
5768
5769
5770
Trillo et al. (2011) investigated the response of two human cancer cell lines (Human hepatocarcinoma,
HepG2, and neuroblastoma, NB69, cells) to a 24-h exposure to 2200 MHz pulse-modulated RF EMF (5 µs pulse
duration, 100 Hz repetition rate) radar-like signal (average SAR = 0.023 W/kg). Cell proliferation and cell cycle
progression were investigated by applying the Trypan blue exclusion method and PI staining, respectively. A
reduction in cell number (p<0.001) together with an increased proportion of cells in G0/G1 and G2/M phase
(p<0.05) was detected in neuroblastoma, but not in hepatocarcinoma cells (6 independent experiments).
Therefore, the cytostatic response observed resulted cell-type specific. Positive controls were not included in the
study design. [In this study sham-exposed cultures were set up but were not handled in parallel to RF exposed
ones.]
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
Takashima et al. (2006) continuously exposed Chinese hamster ovary (CHO-K1) cells and MO54
cells, derived from a human malignant glioma, for 2 h to 2450 MHz RF EMF at SARs ranging from 0.05 to 200
W/kg. Furthermore, the effects of continuous and intermittent exposure at high SARs were compared, to
evaluate the differences between thermal and non-thermal conditions. In the intermittent exposure study, cells
were exposed to a peak SAR of 300 W/kg, 900 W/kg, and 1500 W/kg for 1 s, followed by resting intervals of 5
s, 17 s, and 29 s resulting in an average SAR of 50 W/kg, or 2 s, 8 s, and 14 s to provide an average SAR of 100
W/kg. In 3 independent experiments for each cell type, proliferation was investigated at 0, 1, 2, 3, 4, 5 days after
the RF exposure by counting the cell number, while cell cycle distribution was assessed immediately after and
after 6 and 24 h post exposure by applying the PI staining assay. The results indicated that exposure to a
continuous RF field at SARs from 0.05 to 100 W/kg for 2 h did not affect cell proliferation and cell cycle
distribution. Continuous exposure to an SAR of 200W/kg, which caused a temperature increase up to 44°C,
inhibited cell growth and proliferation, but not cell cycle progression. When CHO-K1 cells were exposed
intermittently, even at a peak SAR of 1500 W/kg (100 W/kg average SAR), no significant differences were
observed between these conditions and CW exposure at 100 W/kg, suggesting that the non-thermal effect of RF
radiation did not adversely affect cell proliferation. Vijayalaxmi et al. did not find significant effects in cell
proliferation, evaluated as mitotic index and CBPI (from microscope slides set up to investigate chromosomal
aberrations and micronuclei, respectively), when human peripheral blood samples were exposed to different RF
EMF conditions in several separate investigations. (i) Cells from two donors were exposed to 2450 MHz, CW, at
an SAR of 12.5 W/kg, for 90 min delivered continuously or intermittently (30 min on/30min off cycles)
(Vijayalaxmi et al., 1997). (ii) Cells from 4 donors were exposed for 24 h to test the effect of two different
modulations, FDMA at 835.6 MHz, (Vijayalaxmi et al., 2001b) and CDMA at 847.74 MHz (Vijayalaxmi et al.,
2001a). The SAR values in these studies were 4.4 or 5 W/kg (835 MHz) and 4.9 or 5.5 W/kg (847 MHz), (iii)
PHA stimulated (for 24 h) and un-stimulated cells from 3 donors were exposed for 2 hours to pulsed 2450 MHz
(pulse width 10 µs, pulse repetition rate 10 kHz, duty factor 0.1) and 8.2 GHz (pulse width 8 ns, pulse repetition
rate 50 kHz, and duty factor 0.0004) at SARs of 2.13 and 20.71 W/kg, respectively (Vijayalaxmi, 2006). In all
these studies positive controls were also assessed by exposing cells to gamma rays. In all cases, genotoxic effects
were also investigated (see Section 12.3.1).
5798
5799
5800
5801
5802
Cleary et al. (1996) exposed murine cytolytic T lymphocytes (CTLL-2) to 2450 MHz RF EMF, either
CW (average SAR of 5–50 W/kg) or pulse modulated (SAR = 5 W/kg) for 2 h to study the effects on interleukin2 (IL-2)-dependent proliferation. Pulse modulation parameters (pulse repetition rate of 50 Hz, 6.67 ms pulse
duration) simulated those of the modulation of the Personal Communication System (PCS). Exposures were
carried out in a waveguide exposure chamber. The temperature in exposed and sham-exposed cultures was
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
204
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
measured continuously during exposure and was 37 ± 0.1 °C. After exposure, cells were cultured at various
physiological concentrations of IL-2 up to 40 U/ml and cell proliferation was measured by tritiated thymidine
incorporation immediately or 24 h after exposure. In cells exposed to CW, a significant reduction in proliferation
was detected in the absence of IL-2 at SAR values greater than 25 W/kg (p<0.05). This reduction remained when
IL-2 was added after exposure at concentrations lower than 20 U/ml (IL-2 saturating concentration). In contrast,
for lower SAR values a slight increase in proliferation was recorded immediately after exposure, which became
statistically significant at 40 U/ml IL-2 and disappeared at 24 h after exposure. Results of pulse-modulated RF
exposure indicated an increase in cell proliferation at 2 and 6 h after exposure in absence of IL-2, and no effect
in the presence of 20 U/ml IL-2. On the whole, the results from 2 to 4 independent experiments indicated that RF
exposure is capable of affecting the IL-2-dependent proliferation of murine T-lymphocytes. Such effects were
not thermal, since with increased temperature (38–41 °C), cells showed qualitative and quantitative different
effects. This study has already been described in Section 10.3.
5815
5816
5817
5818
5819
5820
5821
5822
Sekijima et al. (2010) exposed three human cell lines (A172; IMR-90; and neuroglioma-derived H4
cells) to 2142.5 MHz, CW or W-CDMA-modulated RF EMF at SARs of 0.08–0.8 W/kg for up to 96 hours. The
results of two to four independent experiments, performed blinded, indicated no differences in cell proliferationrelated gene expression profile between exposed and sham-exposed cultures. Under some exposure conditions,
an increase or decrease (or both) in the number of cells was observed, but these changes were not consistently
repeated under the same RF field exposure conditions. In contrast, change in proliferation profile of the human
cell lines exposed to heat shock at 41 °C (positive control) was significantly different from that of cells
maintained at 37 °C. This study has been also described in Section 12.3.3.
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
Zhijian et al. (2013) exposed human B lymphoblastoid HMy2.CIR cells to an 1800 MHz GSM signal
for 24 h (5 min on/10 min off cycles) at 37 °C (average SAR = 2.0 W/kg). The authors performed one
experiment to analyse the differential protein expression after RF exposure by using a protein microarray.
Changes in the expression of 27 proteins were detected related to apoptosis, DNA damage repair, oncogenesis,
cell cycle and proliferation (ratio >1.5-fold, p<0.05). In a further step, three independent experiments were
carried out by applying Western blot analysis and a significant down-regulation of the Replication Protein A 32
(RPA32) (ratio>1.5-fold, p < 0.05) was detected, while the expression of p73 (a transcription factor of p53
family) was significantly upregulated (ratio >1.5-fold, p < 0.05). No difference of the Hypoxia Inducible Factor
1-a (HIF-1a) expression (a protein involved in apoptosis induced hypoxia) was seen. This study is also quoted in
sections 12.3.1, 12.3.3 and 12.3.4, were the effects of RF exposure on DNA repair, gene and protein expression
and apoptosis are reported.
5834
5835
5836
5837
5838
5839
5840
Nikolova et al. (2005) exposed mouse nestin-positive progenitor cells for 48 h (5 min on/30 min off
cycles) to a GSM-modulated 1710 MHz field (average SAR = 1.5 W/kg) to evaluate changes in proliferation
(BrdU incorporation) and differences in transcript levels of cell cycle-regulatory genes (GADD45). In five
independent experiments, the RF exposure resulted in upregulation of GADD45 mRNA levels (p<0.05). Positive
controls have not been included in the study design. In this paper, also quoted in sections 12.3.1, 12.3.3, 12.3.4
and 12.3.6.2, the effect of RF exposure on DNA damage, gene expression, apoptosis and differentiation was also
investigated.
5841
5842
5843
5844
5845
5846
5847
5848
French et al. (1997a) exposed human astrocyte U-87 cells to 835 MHz continuous wave RF EMF, at
power densities of 4 and 0.81 W/m2. Cells were irradiated for 7 days, three-times per day for 20 min, at regular
intervals of four hours. Cell proliferation was evaluated from day 1 to 7 by means of tritiated thymidine assay.
The results did not indicate differences in cells exposed at 4 W/m2 with respect to sham exposed controls.
Cultures exposed to 0.81 W/m2 showed a decrease in cell proliferation already after 2 days (p<0.05). In this
study, also described in Section 5.4.2, Brain - cell morphology, positive controls have not been included. [A
slight temperature increase of 0.6 °C was detected in cultures exposed to 0.81 W/m2 compared to sham controls.
The number of independent experiments is not reported.]
5849
5850
In several studies the effect of higher frequencies (from millimeter waves to THz) on cell proliferation
was investigated.
5851
5852
5853
5854
5855
5856
Szabo et al. (2001) exposed human HaCaT keratinocyte cells to 61.22 GHz (SAR = 770 W/kg) for 30
min. After exposure, cultures were further incubated at 37 °C for 24 h and proliferation was determined by MTT
colorimetric assay. The results of 6 independent experiments indicated that proliferation was not affected by the
RF EMF exposure. [Although in the figures the authors do not call it “sham”, the control is actually a sham.
Positive controls have not been included in the study. SAR was estimated by experimental dosimetry
(calorimetric measurement).]
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
205
5857
5858
5859
5860
5861
Hintzsche et al. (2013) exposed human primary dermal fibroblasts (HDF cells) and a keratinocyte cell
line (HaCaT) to terahertz radiation for 2 and 8 h at 0.380 and 2.520 THz, at power densities ranging from 0.3 to
9 W/m2. Cell proliferation, evaluated by applying the cytokinesis block proliferation index, was unaffected by
the exposure (3 independent experiments for each cell type). In this study DNA damage was also investigated, as
reported in Section 12.3.1.
5862
5863
5864
5865
5866
5867
Zeni et al. (2007b), used a Free Electron Laser equipment to deliver 120 and 130 GHz RF EMF (a
“train” of micropulses, each 50 ps long and 330 ps pulse width) for 20 min to whole blood samples from 17
healthy subjects. The 120 GHz exposure was tested at a calculated SAR of 0.4 W/kg, while the 130 GHz
exposure was tested at SARs of 0.24, 1.4 and 2 W/kg. No effect on cell proliferation was found, as assessed by
calculating the CBPI on slides set up to investigate MN frequency. Positive controls have not been included in
the study design. In this study, genotoxic effects have been also investigated (see Section 12.3.1).
5868
5869
In some investigations the effect of combined exposures to RF EMF and chemical or physical agents
have been evaluated.
5870
5871
5872
5873
5874
5875
5876
5877
5878
Sannino and co-workers conducted several studies to investigate the effect of RF EMF, given alone or
in combination with mitomycin-C (MMC). They exposed human peripheral blood lymphocytes from 14 donors
to a 900 MHz GSM signal (average SAR = 1.25 W/kg) for 20 h in several stages of the cell cycle (Sannino et al.,
2009b; 2011). No effect on cell proliferation was detected in cultures exposed to RF alone or in combination
with MMC, as evaluated by applying the CBPI from slides set up to investigate the induction of micronuclei (see
Section 12.3.1). In a follow-up study the authors also failed to find effects when 20 h RF EMF exposures were
given in the S phase of the cell cycle at a 1950 MHz UMTS signal, at SAR values of 0.15, 0.3, 0.6 and 1.25
W/kg (Zeni et al., 2012a). In these studies the induction of genotoxic effects was also investigated (see Section
12.3.1).
5879
5880
5881
5882
5883
5884
5885
Lee et al. (2011b) investigated if single or combined RF EMF exposures interfere with cell cycle and
proliferation. To this aim they exposed human breast MCF7 cancer cells for 1 h to either 837 MHz, CDMA
(average SAR = 4 W/kg), or 837 MHz CDMA plus 1950 MHz WCDMA (2 W/kg CDMA and 2 W/kg
WCDMA). In three independent experiments the authors found no differences compared to sham-exposed
cultures in cell cycle distribution (PI staining) and cell proliferation rate (BrdU incorporation assay). Moreover,
the expression of cell cycle regulatory proteins p53, p21, cyclin A, cyclin B1 and cyclin D1 was also unaffected.
This study has also been described in Section 12.3.2 (signal transduction).
5886
5887
5888
5889
5890
Del Vecchio et al. (2009) found that up to 72 h exposure to a 900 MHz GSM signal (SAR = 1 W/kg)
did not affect cell proliferation of SN56 neural cells and rat primary cultured neurons (MTT assay). The same
results were obtained when cell cultures were exposed to RF in combination with 25–35 β-amyloid fragments (a
major toxic event in Alzheimer’s disease), as assessed in 2 to 4 independent experiments. This study has also
been described in Chapter 8.3.
5891
5892
5893
5894
5895
5896
5897
Höytö et al. (2008a) exposed human neuroblastoma (SH-SY5Y) cells and mouse fibroblasts (L929) to
872 MHz, CW and GSM-modulated RF EMF (SAR = 5 W/kg) for 1 h to investigate cell proliferation.
Moreover, the effect of combined exposure with menadione (Vitamine K3, an inducer of cell death and oxidative
stress) was also evaluated in both cell lines. In 3 independent experiments, no effect of RF exposure, alone or in
combination with menadione, was detected in both cell types (Alamar blue assay). In this study, also quoted in
sections 12.3.4 and 12.3.5, the effect of RF on apoptosis and oxidative stress was also investigated. [Although in
the figures the authors do not call it “sham”, the control is actually a sham.]
5898
Studies not included in the analysis
5899
5900
5901
5902
5903
5904
Hintzsche et al. (2012b) did not observe differences in cell proliferation, evaluated by applying the
CBPI, in keratinocytes (HaCaT cells) exposed to RF EMF at 900 MHz, CW, at 5, 10, 30 or 90 V/m for 30 min
and 22 h (3 independent experiments performed blinded). The same experimental conditions failed to induce
effects also in human-hamster hybrid cells. In this study the induction of genotoxic effects was also investigated
(Section 12.3.1). [No dosimetry has been carried out. The results are reported as a function of the E-field
strength.]
5905
5906
5907
Esmekaya et al. (2011) observed a decrease in proliferation of human blood lymphocytes exposed to
an 1800 MHz GSM signal (average SAR = 0.21 W/kg) for 6, 8, 24 and 48 hours. When cell cultures were treated
with ginkgo, an anti-oxidant used in alternative medicine, such decrease was reduced. Cell proliferation was
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
206
5908
5909
5910
5911
evaluated in five independent experiments by applying the MTT assay. In this study the induction of genotoxic
effects was also investigated (Section 12.3.1). [The relevance of this study is questionable since no proper
dosimetric evaluation was performed (the SAR was estimated by using the electric field measured along the horn
antenna). Moreover, the number of donors included in the study is not clear.]
5912
5913
5914
5915
5916
5917
Lixia et al. (2006) exposed immortalized lens epithelial cells (hLEC) to 1800 MHz GSM-modulated
RF EMF at SARs of 1–3 W/kg. Cell proliferation rate was measured by BrdU incorporation immediately after
the 2-h exposure and 1 and 4 days later. By comparing exposed and sham-exposed cells no differences were
detected, for all the experimental conditions tested. In this study the expression of HSPs and DNA damage have
also been investigated, as reported in sections 12.3.1 and 12.3.3. [The number of independent experiments
carried out is not reported, although statistical analysis was performed.]
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
Marinelli et al. (2004) exposed human T-Lymphoblastoid Leukemia CCRF-CEM cells 900 MHz RF
EMF, CW (SAR = 3.5 mW/kg) for 2, 4, 12, 24 and 48 h to assess cell proliferation (MTT assay). The results
were compared to their respective control cultures. Moreover, control and sham-exposed samples were assessed
at 2, 24 and 48 h. In 15 independent experiments a statistically significant decrease (p<0.01) in cell proliferation
after 24 and 48 h of exposure to 900 MHz compared to the control cells was detected, while no effects were
found for shorter exposure times (2, 4, and 12 h). [It seems that sham exposures were carried out in a separate set
of experiments and compared to controls concurrently assessed. In the study, cell cycle progression, DNA
damage and the expression of pro-apoptotic and cell cycle regulatory genes were also measured, but the results
are not reported here since they have been compared to control samples. No numerical dosimetry is reported and
SAR calculation seems to be derived from electric field measurement in absence of the sample inside the TEM
cell. Therefore, the relevance of this study is questionable.]
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
Pavičić and Trošić (2006) exposed Chinese hamster lung (V79) cells for 1, 2 and 3 h to 864 MHz RF
EMF, CW, at a calculated average SAR per cell of 0.08 W/kg. Cell proliferation kinetics was measured by
counting cells on 1, 2, 3, 4 and 5 days post exposure. A statistically significant decrease in cell proliferation was
detected in cultures exposed for 2 and 3 h at day 3 (p<0.05; 6 experiments), while cell viability was not affected.
In a follow-up study the authors applied the same protocol and confirmed their previous findings. Moreover, the
same results were obtained when V79 cells were exposed to 935 MHz, at a calculated average SAR per cell of
0.12 W/kg (Pavičić & Trošić, 2008). Positive controls were not included in these studies. [Numerical dosimetry
was not performed and it seems that SAR was derived from and electric field measurement in the TEM cell,
carried out in absence of the sample. Furthermore, the authors refer to a SAR for a single cell, whose definition
and estimation method is questionable.]
5939
5940
5941
5942
5943
5944
Gurisik et al. (2006) exposed human neuronal SK-N-SH and human monocytoid U937 cells to a 900
MHz GSM signal (SAR = 0.2 W/kg) for 2 h. After 24 h from the RF exposure no effects were identified in cell
cycle distribution, as assessed in 6 independent experiments for each cell type investigated (PI staining). In this
paper the effect of RF on gene expression and apoptosis was also investigated (see sections 12.3.3 and 12.3.4).
[The authors also reported that the cell cultures demonstrated low viability (~65%) in both the sham and RFexposed samples at 24 h after exposure. Therefore, the validity of these results is questionable.]
5945
5946
5947
5948
5949
5950
5951
5952
5953
Lee et al. (2005) exposed cultured human HL-60 cells to 2450 MHz pulse-modulated RF EMF at 10
W/kg for 2–6 h. The authors used serial analysis of gene expression (SAGE) to quantify gene transcript levels
after exposure. They found that 221 genes demonstrated altered expression after a 2-h exposure and 759 after a
6-h exposure. Among the downregulated genes, 23 were related to cell cycle (1 independent experiment). This
study has been also quoted in sections 12.3.3 and 12.3.4, were the effects of RF exposure on gene and protein
expression and apoptosis were reported. [There is an inadequate number of independent experiments. A 2 h
sham-exposed sample was used as a reference to compare against the 6 h RF-exposed sample. The results were
based on differences in fold-change. Statistical analysis of the gene expression data was not performed and
results were not confirmed by RT-PCR.]
5954
5955
5956
5957
5958
5959
5960
5961
In a study by Yao et al. (2004), cultured rabbit lens epithelial cells (RLEC) were exposed to
continuous RF EMF at 2450 MHz and power densities of 0.10, 0.25, 0.50, 1.00, and 2.00 mW/cm2 for up to 8 h.
As assessed by PI staining, cell cycle progression was not affected in cultures exposed to power densities lower
than 0.50 mW/cm2, while cultures exposed to 0.5, 1, and 2 mW/cm2 resulted arrested in the G0/G1 phase of the
cell cycle when compared to sham-exposed cultures (p<0.01). Moreover, the expression of two genes involved
in the cell cycle, P21WAF1 and P27Kip1, was evaluated in the RLEC using Western blot analysis. A
significantly increased expression of P27Kip1 protein was detected in cultures exposed to 2 mW/cm2 for 4, 6,
and 8 h when compared to their respective sham controls. This latter finding was not confirmed by RT-PCR
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
207
5962
5963
analysis. This paper has been also reported in Section 12.3.3. [The relevance of this study is questionable since
no dosimetric evaluation was performed.]
5964
5965
5966
5967
5968
5969
5970
5971
Buttiglione et al. (2007) evaluated the effect of exposure to 900 MHz RF EMF, GSM modulated
(SAR = 1 W/kg), at different time points (5 min to 24 h) on cell cycle progression of human neuroblastoma (SHSY5Y) cells (PI staining). In 3 independent experiments, RF exposure induced a significant G2-M arrest
(p<0.01). This study has been described in details in sections 12.3.2 and 12.3.4, where the effect on signal
transduction pathway and apoptosis is reported. [It was stated that an input power of 31.6 W was fed into the
exposure system, which would lead to an SAR of approximately 10 W/kg. Since culture temperatures were not
measured during RF exposure and since the exposure details are not clearly reported, the results of this study
must be cautiously interpreted as thermal confounding may have occurred.]
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
Beneduci and co-workers carried out several separate investigations to evaluate the effect of low
power millimeter waves on proliferation and cell cycle progression of healthy and cancer cells. In a first study,
RPMI 7932 human melanoma cells were exposed to low power millimeter waves with three different
frequencies and irradiation modes: a wide-band irradiation in the range 53.57–78.33 GHz and two
monochromatic irradiations at 51.05 GHz and 65.00 GHz. Cell cultures were exposed for 1 or 3 h per day for
several days. No effect on proliferation was detected in cultures exposed at 51.05 GHz or at 65.00 GHz, while
low-power wide-band millimeter waves, in the 50–80 GHz frequency range, induced a decrease in cell
proliferation, as assessed by counting the cell number every 24 h. Such a decrease reached 29% after seven days
of treatment, compared to sham-exposed cultures (Beneduci et al., 2005). [In this study the number of
independent experiments in not reported. Moreover, although the authors stated that statistical analysis was
performed, no p-values are reported. Dosimetric analysis was not performed, and authors only provide the power
level at the output of the MMW generator, but not an estimation of either SAR or power density.] In a follow-up
study the authors exposed human erythromyeloid leukemia K562 cell line to a wide-band irradiation in the
frequency range between 53.57 and 78.33 GHz. Also in this case the exposure was carried out 3 h per day for
several days. They confirmed the inhibitory effect of RF exposure, that was statistically significant after 3 days
exposure (p<0.05) and decreased by increasing the exposure time until day 7 (3 independent experiments)
(Beneduci et al., 2007). Positive controls were not included in the study. [The authors did not perform any
dosimetric analysis, and the description of the irradiation condition is very imprecise and presents some technical
mistakes.] In a further study, RPMI 7932 human melanoma cells were exposed at 42.20 and 53.57 GHz, in far
field conditions, with an incident power density of 0.14 and 0.37 mW/cm2, respectively. The experimental
protocol was 1 h exposure per day up to a total of four treatments. Neither proliferation nor cell cycle
progression (PI staining) were affected by the exposure (Beneduci, 2009). Positive controls were not included in
the study. [No numerical dosimetry was performed, but a rough estimation of the incident power density is
reported. Moreover, the results are from two independent experiments, counted four times.]
5996
5997
5998
5999
6000
6001
6002
6003
6004
Peinnequin et al. (2000) exposed human T-lymphoma-derived (Jurkat) cells for 48 h to 2450 MHz RF
EMF, CW, (power density = 5 mW/cm2, corresponding to a SAR of 4 W/kg). Cell proliferation (Alamar Blue
assay) was evaluated immediately after RF exposure or 16 h after the addition of apoptotic agents, i.e. antagonist
anti-Fas receptor antibody or sodium butyrate. The results obtained on three independent experiments indicated
no effects of RF alone, while a reduction in cell proliferation was detected in samples treated with anti-Fas
(p<0.001), but not with ceramide butyrate. This study has been also quoted in Section 12.3.4 (apoptosis). [In this
paper neither numerical nor experimental dosimetry is reported. SAR has been evaluated calorimetrically, but
the authors do not describe the methods used. The power density has been obtained by using a field meter,
therefore measurements were supposedly performed in absence of the sample.]
12.3.14. In vitro studies assessing effects of RF-EMF exposure on proliferation and cell cycle
Cells
Biological endpoint
Exposure conditions
Results
Comment
Reference
Activation and
expression of p53
900 MHz, GSM
No effect.
Signal transduction (Bourthoumieu et
and apoptosis
al., 2013)
described in 12.3.2
and 12.3.4.
Number of
independent
experiments
Primary human
amniotic cells
n=3
Average SAR 0.25, 1,
2, 4 W/kg
24 h
Blind procedure
not indicated.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
208
Human
lymphocytes
n=3
Cell cycle (BrdU
incorporation and PI
staining)
Human
lymphoblastoid
(Jurkat) cells
900 MHz, GSM
No effect.
For Apoptosis see
Section 12.3.4.
(Palumbo et al.,
2008)
No effect.
For Apoptosis see
Section 12.3.4.
(Capri et al.,
2004b)
Average SAR 1.35
W/kg
1h
n=3
Cell proliferation (3Hthymidine
incorporation)
900 MHz, CW and
GSM
Cell cycle (BrdU
incorporation and PI
staining)
1 h/day for 3 days
Human blood
lymphocytes
Cell cycle (BrdU
incorporation)
900 MHz, GSM,
n=4
Cell proliferation (MI)
Human blood
mononuclear cells
n=13–25
SAR 70 and 76 mW/kg
Blind procedure
not indicated.
No effect.
Average SAR 0.3 and
1 W/kg
Cell proliferation
(CBPI)
n=10
(Zeni et al., 2005)
Blind procedure
not indicated.
2h
Human blood
lymphocytes
For Genotoxicity
see Section
12.3.1.
900 MHz, GSM
No effect.
For Genotoxicity
see Section
12.3.1.
(Scarfi et al., 2006)
No effect.
For Protein
expression and
apoptosis see
Sections 12.3.3
and 12.3.4.
(Sanchez et al.,
2006b)
No effect.
Blind procedure
not indicated.
(Moisescu et al.,
2008)
Significant
reduction, not
dependent on
SAR.
95% confidence
level.
(Kwee &
Raskmark, 1998a)
Significant
reduction, not
dependent on
temperature.
Follow-up study of
Average SAR 1, 5, 10
W/kg
24 h
Reconstructed
human epidermis
cells
Cell proliferation
900 MHz, GSM
(immunofluorescence) Average SAR 2 W/kg
48 h
n=5
Murine B16F10
melanoma cells
Phases of mitosis
Average SAR 2.2 W/kg
n not reported
Transformed
human epithelial
amnion (AMA)
cells
1h
Cell proliferation
(MTS test)
Cell proliferation
(MTS test)
n=2
SAR 0.021, 0.21 and
2.1 mW/kg
960 MHz, GSM
Average SAR 2.1
mW/kg
30 min
n=11
Normal rat glial
cells
960 MHz, GSM
20, 30 and 40 min
n=12
Transformed
human epithelial
amnion (AMA)
cells
900 MHz, GSM
T= 35, 39 °C
Cell proliferation (3Hthymidine
incorporation)
Transformed (C6)
rat glial cells
Blind procedure
not indicated.
(Kwee &
Raskmark,
1998a)90%
confidence level.
(Velizarov,
Raskmark & Kwee,
1999)
Blind procedure
not indicated.
836.55 MHz, TDMA
For genotoxicity
see Section
12.3.1.
24 h
No effect
immediately
and 1, 5, 7, 9,
12 and 14 days
after exposure.
849 MHz
No effect.
Blind procedure
not indicated.
(Lee et al., 2008)
No effect.
Blind procedure
not indicated.
(Higashikubo et al.,
2001)
Average SAR 0.59,
5.9, 59 mW/kg
(Stagg et al., 1997)
Blind procedure
not indicated.
n=3–6
Mouse fibroblasts
NIH3T3
n=4–6
Cell proliferation
(BrdU incorporation)
Average SAR 2–10
Cell cycle (PI staining) W/kg
1 h; 1 h per day for 3
days
Mouse fibroblasts
C3H 10T1⁄2
human glioma
cells U87MG
n=3
Cell cycle (BrdU
incorporation and PI
staining)
835.62 MHz, FMCW
847.74 MHz, CDMA
Average SAR 0.6 W/kg
13 or 100 h
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
209
Human- Hamster
hybrid FC2 cells
Cell proliferation (MI)
900 MHz, CW, GSM
No effect.
SAR 0.01-0.017 W/kg
n=3
Separate E and H
components.
(Schrader et al.,
2011)
For genotoxicity
see Section
12.3.1.
E-field: 45 or 90 V/m
30 min
Blind procedure
not indicated.
Rat bone marrow
lymphocytes
n=11
Human skin
fibroblasts
n=3-5
Cell proliferation
900 MHz, CW
(trypan blue exclusion SAR 2 W/kg
method)
30 min
No effect.
For immune
system and
genotoxicity see
sections 10.4 and
12.3.1.
(Kumar et al.,
2011)
Cell cycle (PI staining) 1800 MHz, GSM
No effects on
cell cycle
distribution
soon after and
after 24 h from
RF.
For genotoxicity
and oxidative
stress see
Sections 12.3.1
and 12.3.5.
(Xu et al., 2013)
No effect.
For Protein
expression and
apoptosis see
sections 12.3.3
and 12.3.4.
(Lantow et al.,
2006c)
No effect.
For genotoxicity,
(Huang et al.,
signal transduction 2008b)
and gene and
protein expression
see Sections
12.3.1, 12.3.2 and
12.3.3.
Cell proliferation
(Trypan blue
exclusion method)
Average SAR 3 W/kg
24 h
(5 min on/10 min off
cycles)
Increase in
G0/G1 arrest
after 6 h
following
exposure.
No effect on
cell
proliferation.
Human monocytes Cell cycle (PI staining
(Mono Mac 6) cells and BrdU
incorporation)
n=4-6
Mouse auditory
hair cell-derived
(HEI-OC1) cells
1800 MHz, GSM-DTX
Average SAR 2 W/kg
12 h
Cell cycle (PI staining) 1763 MHz, CDMA
SAR 20 W/kg
24 and 48 h
n=3
Blind procedure
not indicated.
Human T-cell
derived (Jurkat)
cells
n=3-6
Cell proliferation (cell
counting)
1763 MHz CDMA
No effect.
SAR 2, 10 W/kg
Cell cycle (PI staining) 1, 2, 3 h
For genotoxicity,
(Huang et al.,
signal transduction 2008a)
and gene and
protein expression
see Sections
12.3.1, 12.3.2 and
12.3.3.
Blind procedure
not indicated.
Human dermal
papilla cells
n=3
Cell cycle (PI staining) 1763 MHz CDMA
growth-related protein SAR 10 W/kg
expression
1 or 3 h
No effect on
cell cycle
distribution
after 1 h
exposure.
1 h/day for 7 days
For signal
transduction,
apoptosis and
oxidative stress
see Sections
12.3.2, 12.3.4 and
12.3.5.
(Yoon et al., 2011)
Increased
expression of
Bcl-2 and
Blind procedure
cyclin; increase not indicated.
in IGF1 mRNA
and in
phosphorylation
of MAPK-1.
Human blood
lymphocytes
n=6
Cell proliferation
(CBPI)
1950 MHz, UMTS
No effect.
SAR 2.2 W/kg
For Genotoxicity
(Zeni et al., 2008)
see section 12.3.1.
24–68 h
(6 min on/2 h off
cycles)
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
210
Human-derived
immune cell lines
(HL-60, MonoMac-6, TK6)
Cell cycle (PI staining) 1900 MHz, pulse
modulated
SAR 1 and 10 W/kg
6h
n=5
Human gliomaderived (MO54)
cells
Proliferation (cell
counting)
1950 MHz, CW
SAR 2 and 10 W/kg
2h
Proliferation (Trypan
blue staining)
2200 MHz, pulse
modulated
(Chauhan et al.,
2007a)
No effect
immediately
and 1, 2, 3 and
4 days after
exposure.
For Protein
expression see
Section 12.3.3.
(Miyakoshi et al.,
2005)
24 h
Chinese hamster
Cell proliferation (cell 2450 MHz
ovary (CHO-K1)
counting)
SAR 0.05–200 W/kg
and human glioma Cell cycle (PI staining)
Continuous for 2 h
(MO54) cells
Intermittent exposure
n=3
Average SAR 50 and
100 W/kg
Proliferation (MI,
CBPI)
n=2
2450 MHz, CW
Sham and RFexposed samples
not handled in
NB69:
reduction in cell parallel.
number and
increased
proportion of
cells in G0/G1
and G2/M
phase.
(Trillo et al., 2011)
CHO-K1: no
effect.
Blind procedure
not indicated.
(Takashima et al.,
2006)
For Genotoxicity
see Section
12.3.1.
(Vijayalaxmi et al.,
1997)
MO54: No
effect
immediately
and . 1–5 days
after exposure
at non-thermal
levels.
No effect.
SAR 12.5 W/kg
90 min continuous or
intermittent
Blind procedure
not indicated.
(30 min on/30 min off
cycles)
Human blood
lymphocytes
Proliferation (MI,
CBPI)
n=4
Human blood
lymphocytes
n=4
Human blood
lymphocytes
n=3
835.62 MHz, FDMA
No effect.
SAR 4.4 and 5 W/kg
24 h
Proliferation (MI,
CBPI)
For Genotoxicity
see Section
12.3.1.
(Vijayalaxmi et al.,
2001b)
Blind procedure
not indicated
847.74 MHz, CDMA
No effect.
SAR 4.4 and 5 W/kg
24 h
Proliferation (MI,
CBPI)
Blind procedure
not indicated.
HepG2: no
effect.
Cell cycle (PI staining) Average SAR 0.023
W/kg
n=6
Human blood
lymphocytes
For immune
system, protein
expression and
apoptosis, see
Section 10.4,
12.3.2 and 12.3.4.
(5 min on/10 min off
cycles)
n=2
Human
hepatocarcinoma
(HepG2) and
neuroblastoma
(NB69) cells
No effects
immediately
and 6 and 24 h
after exposure.
For Genotoxicity
see Section
12.3.1.
(Vijayalaxmi et al.,
2001a)
Blind procedure
not indicated
2450 MHz, pulsed
No effect.
SAR 2.13 W/kg
8.2 GHz, pulsed
For Genotoxicity
see Section
12.3.1.
(Vijayalaxmi, 2006)
Blind procedure
not indicated.
SAR 20.8 W/kg
2h
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
211
Murine cytolitic T
lymphocytes
(CTLL-2)
n=2–4
Human glioblastoma (A172)),
neuroglioma (H4)
and fibroblasts
(IMR-90)
Proliferation (3Hthymidine
incorporation) in
presence or in
absence of IL-2
2450 MHz
Proliferation-related
gene expression
profile
2142.5 MHz, CW or W- Increase or
For Gene
CDMA
decrease in cell expression see
number, not
Section 12.3.3.
SAR 0.08–0.8 W/kg
consistently
Up to 96 h
repeated.
(Sekijima et al.,
2010)
Proliferation-related
protein expression
1800MHz, GSM
(Zhijian et al.,
2013)
CW: reduced
proliferation at
SAR > 25 W/kg
at low IL-2
Pulse modulation (6.67 concentration.
ms pulsed at 50 Hz):
Pulse
average SAR 5 W/kg
modulation:
2h
increased
proliferation at
lower SARs
immediately
post exposure
and reduced 24
h post
exposure.
CW: average SAR 5–
50 W/kg
For immune
system see
Section 10.4.
(Cleary et al.,
1996)
Blind procedure
not indicated.
n=2-4
Human B
lymphoblastoid
HMy2.CIR cells
Average SAR 2 W/kg
24 h
n=3
(5 min on/10 min off
cycles)
Downregulation of
RPA32; upregulation of
p73.
For genotoxicity,
gene and protein
expression and
apoptosis see
Sections 12.3.1,
12.3.3 and 12.3.4.
Blind procedure
not indicated.
mouse neural
progenitor stem
cells
Upregulation of
Average SAR 1.5 W/kg GADD45.
No effect on
48 h
other
(5 min on/30 min off
parameters.
cycles)
For genotoxicity,
gene expression,
apoptosis and
differentiation see
Sections 12.3.1,
12.3.3, 12.3.4 and
12.3.6.2.
(Nikolova et al.,
2005)
Human
Proliferation (3Hastrocytoma (U-87 thymidine
MG) cell line
incorporation)
835 MHz, CW
No effect at
4 W/m2.
Cell morphology
also investigated.
n not reported
20 min, 3 h/day for 7
days
Decrease
already after 2
days at 0.81
W/m2.
Slight temperature
increase (0.6 °C)
in cultures
exposed to 0.81
W/m2.
(French, Donnellan
& McKenzie,
1997b)
n=5
Proliferation (BrdU
incorporation)
cell cycle (transcript
levels of genes)
1710 MHz, GSM
Power density 0.81 or
4 W/m2
Blind procedure
not indicated.
Human HaCaT
keratinocytes
Proliferation (MTT)
No effect.
30 min
Proliferation (CBPI)
0.380 and 2.520 THz
No effect.
Power density 0.3-9
W/m2
For genotoxicity
see Section
12.3.1.
(Hintzsche et al.,
2013)
2 and 8 h
Blind procedure
not indicated.
n=3
Human blood
lymphocytes
(Szabo et al.,
2001)
SAR 770 W/kg
n=6
Human primary
dermal fibroblasts
(HDF cells)
keratinocyte cell
line (HaCaT)
61.22 GHz
Proliferation (CBPI)
n=17
120 GHz
No effect.
SAR 0.4 W/kg
130 GHz
For genotoxicity
see Section
12.3.1.
(Zeni et al., 2007b)
SAR 0.24, 1.4, 2 W/kg
20 min
Studies including co-exposures
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
212
Human blood
lymphocytes
Proliferation (CBPI)
900 MHz, GSM
No effect.
For genotoxicity
see Section
12.3.1.
(Sannino et al.,
2009b)
No effect.
Follow-up study of
(Sannino et al.,
2009b).
(Sannino et al.,
2011)
Average SAR 1.25
W/kg
n=5
20 h
Combined exposures
with MMC in S phase
(after RF)
Human blood
lymphocytes
Proliferation (CBPI)
900 MHz, GSM
Average SAR 1.25
W/kg
n=9
For genotoxicity
see Section
12.3.1.
20 h
Combined exposures
with MMC in different
stages of the cell cycle
Human blood
lymphocytes
Proliferation (CBPI)
1950 MHz, UMTS
No effect.
For genotoxicity
see Section
12.3.1.
(Zeni et al., 2012a)
No effect.
For Signal
transduction see
Section 12.3.2.
(Lee et al., 2011b)
No effect.
For
(Del Vecchio et al.,
Neurodegenerative 2009)
disorders see
Section 8.3.
No effect.
For Apoptosis and
oxidative stress
see sections
12.3.4 and 2.3.5.
SAR 0.15, 0.3, 0.6,
1.25 W/kg
n=9
20 h
Combined exposures
with MMC in S phase
(after RF)
Human breast
Cell cycle (PI staining) 837 MHz, CDMA
MCF7 cancer cells Proliferation (BrdU
Average SAR 4 W/kg
n=3
assay)
837 MHz, CDMA +
cell cycle regulatory
1950 MHz, WCDMA
proteins (p53, p21,
4 W/kg SAR (2+2
cyclin A, cyclin B1,
W/kg)
cyclin D1)
1h
Mouse SN56
neural cells
Proliferation (MTT
assay)
Primary cultures of
cortical neurons
Average SAR 1 W/kg
Up to 72 h
Combined exposures
with β-amyloid
fragments
n=2–4
Human
neuroblastoma
(SH-SY5Y) cells
900 MHz, GSM
Proliferation (Alamar
blue assay)
872 MHz, CW and
GSM
SAR 5 W/kg
n=3
1h
Mouse fibroblasts
(L929)
Combined exposure
with menadione
(concurrent)
n=3
(Höytö et al.,
2008a)
Blind procedure
not indicated.
“No effect” means no statistically significant effect.
Abbreviations:
6005
6006
12.3.6.2
6007
6008
A limited number of in vitro studies have addressed the effect of RF EMF exposure on differentiation.
They are reported in the following.
6009
6010
6011
6012
6013
6014
6015
6016
Merola et al. (2006) exposed human neuroblastoma cells (LAN-5) to 900 MHz GSM modulated RFEMF at an average SAR of 1.0 W/kg for 24, 48 and 72 h to evaluate proliferation and differentiation. No
significant alteration in cell proliferation was detected in all cases, as assessed in three independent experiments
by a colorimetric technique (metabolic conversion of a tetrazolium salt into Formazan). Moreover, when cultures
were treated with the differentiative agent retinoic acid and exposed to RF EMF, no differences were detected
relative to cultures treated with retinoic acid alone, as assessed by evaluating the expression of two oncogenes,
B-myb and N-myc, very sensitive markers of proliferation and differentiation in the cell type investigated. This
study has been also described in sections 12.3.3 (gene and protein expression) and 12.3.4 (apoptosis).
Differentiation
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
213
6017
6018
6019
6020
6021
6022
6023
6024
Nikolova et al. (2005) exposed mouse nestin-positive progenitor cells for 48 h (5 min on/30 min off
cycles) to a 1710 MHz GSM signal (average SAR = 1.5 W/kg) to evaluate changes in the mRNA transcript
levels of the neural specific gene Nurr1 by using quantitative (Q)-RT-PCR. Nurr1 is a transcription factor
involved in the determination and development of dopaminergic neurons in vivo. RF EMF exposure resulted in
downregulation of transcript levels of Nurr1 (p<0.05; 5 independent experiments). The authors stated that this
effect might transiently affect neuronal differentiation. Positive controls have not been included in the study
design. In this paper, also quoted in sections 12.3.1 (Genotoxicity), 12.3.2 (signal transduction), 12.3.4
(apoptosis) and 12.3.6.1 (proliferation).
6025
6026
6027
6028
6029
Czyz et al. (2004) exposed mouse embryonic stem (ES) cells, wild-type and ES cells deficient for the
tumor suppressor p53, to 1710 MHz GSM-modulated RF EMF at SAR values of 2.0 W/kg for 22 or 40 h. No
effects were observed on cell cycle progression (ethidium bromide fluorescence assay; 3–4 independent
experiments) and on spontaneous or DMSO-induced cardiac differentiation. Positive controls were not included.
This study has been also described in Section 12.3.3 (gene and protein expression).
6030
6031
6032
6033
6034
6035
6036
6037
6038
Epidermal reconstructs containing either only keratinocytes or a combination of keratinocytes and
melanocytes grown on dead de-epidermized dermis, were exposed to a 900 MHz GSM signal (SAR = 2 W/kg)
for 6 hours by Simon et al. (2013). The expression and localization of various markers of keratinocyte and
melanocyte differentiation were analysed at 2, 6, 18 and 24 h after exposure, using histology,
immunohistochemistry and Western blot. No noticeable changes were found in the localization of basal markers
(cytokeratins 5, 14) and late markers of differentiation (loricrin, filaggrin), but the rate of epidermal proliferation,
was transiently decreased 2 h post-exposure. Overall, the main effect of the RF exposure was a subtle alteration
of differentiation markers level without alteration of localization of such markers and no detectable induction of
apoptosis. This study has been also described in sections 12.3.4 (apoptosis) and 12.3.5 (oxidative stress).
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
Bock et al. (2010) exposed mouse mesenchymal stem cells (MSC) to a broad spectrum THz radiation
(centered at ~ 10 THz) at an average power density of 1 W/m2 for 2, 4, 6 and 9 h. By looking at morphological
changes, a significant accumulation of lipid-like droplets in the cytoplasm was evident after 6 h exposure.
Moreover, after 9 h exposure MCS were analysed for global gene expression (Affymetrix mouse genome
microarray). Many of the MSC genes did not respond at all (89%), certain genes were activated (6%) while other
were repressed (5%) significantly after 9 h irradiation (p<0.05). In the activated group of genes, confirmed by
mRNA levels quantification by using RT-PCR, the overexpression of transcription factor peroxisome
proliferator-activated receptor gamma (PPARG) that is known to be required for adipocyte differentiation,
suggested that THz field, in the specific exposure conditions, enhanced the differentiation process towards a
adipocyte-like phenotype in MSC. This study has been also described in section 12.3.3. [The results reported
here have been obtained from two independent experiments]. In a follow-up study (Alexandrov et al., 2013) the
authors exposed MSC cells to either broadband THz radiation (centered at ~ 10 THz, 1 mJ, pulse width 35 fs
duration, i.e., high peak power per pulse 30 MW) for 2 or 12 hours, or to single-frequency (2.52 THz) for 2
hours only. Each scenario was applied in three independent experiments and the reported results were averaged.
In each experiment, the MSC cultures were synchronized to be at the same differentiation time point
immediately before the irradiation. It is reported that prolonged (12 hours) broad-spectrum THz irradiation of
MSCs resulted in overexpression of PPARG, adiponectin, GLUT4, and FABP4 (p<0.05) with dependence on the
level of stem cell differentiation, while 2 hour exposures did not have significant effects on gene expression.
Taken together, the results showed that the effect of THz exposure on adipocyte differentiation depended on
irradiation parameters such as the duration and type of THz source, and on the degree of stem cell
differentiation. No positive controls were included in the study design.
6060
Studies not included in the analysis
6061
6062
6063
6064
6065
6066
6067
6068
6069
Duranti et al. (2005) investigated the effect of RF EMF exposure on HaCaT cells, a spontaneously
immortalized human keratinocytes cell line. Cell cultures were exposed for 18 h to CW 900 MHz (average SAR
ranging from 0.04 to 0.08 W/kg, on the bases of the culture position inside the GTEM exposure chamber). Cell
proliferation was measured 24, 48 and 72 h after RF (trypan blue exclusion method). The results of three
independent experiments indicated a delay in cell proliferation (12%, 24% and 30% after 24, 48 and 72 h,
respectively), that was related to a the activation of a differentiative process, as demonstrated by the increased
expression of keratin 1 and involucrin, two hallmarks of differentiation. Positive controls have not been included
in the study design. [In this paper, although the authors stated that statistical analysis has been performed, no pvalues were reported.]
12.3.15. In vitro studies assessing effects of RF-EMF exposure on differentiation
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
214
Cells
Biological endpoint
Exposure conditions
Results
Comment
Reference
Cell proliferation and
differentiation
(Formazan and gene
expression)
900 MHz, GSM
No effect.
For Gene and
protein
expression see
Section 12.3.3.
(Merola et al.,
2006)
mRNA transcript
levels of Nurr
1710 MHz, GSM
For genotoxicity,
signal
transduction
Apoptosis and
proliferation see
Sections 12.3.1,
12.3.2, 12.3.4
and 12.3.6.1.
(Nikolova et al.,
2005)
No effect.
For Gene
expression see
section 12.3.3.
(Czyz et al., 2004)
No effect, as
assessed 2, 6
and 24 h after
exposure.
For Apoptosis
and oxidative
stress see
Sections 12.3.4
and 12.3.5.
(Simon et al.,
2013)
Number of
independent
experiments
Human
neuroblastoma
(LAN-5) cells
n=3
Mouse neural
progenitor stem
cells
Average SAR 1 W/kg
24 to 72 h
Downregulation
Average SAR 1.5 W/kg of Nurr.
48 h
n=5
(5 min on/30 min off
cycles)
Mouse embryonic
stem (ES) cells
Cell cycle (ethidium
bromide staining)
1710 MHz, GSM
n=6
Gene expression
22 or 40 h
Primary human
melanocytes and
keratinocytes
Markers of
differentiation
900 MHz, GSM
n=2-20
Average SAR 2.0 W/kg
Average SAR 2 W/kg
6h
Blind procedure
not indicated.
Mouse
mesenchymal
stem cells (MSC)
Gene expression
10 THz
Average power 1 W/m2
n=2
2-9 h
Mouse
mesenchymal
stem cells (MSC)
n=3
Gene expression
10 THz
1 mJ, pulse width 35 fs
duration
2–12h
2.52 THz
2h
Accumulation of Blind procedure
lipid-like droplets not indicated.
in the cytoplasm
and 6%
activated genes
after 9 h
exposure.
Overexpression
of a transcription
factor (PPARG)
related to
adipocyte
differentiation.
(Bock et al., 2010)
Overexpression
of PPARG,
adiponectin,
GLUT4, and
FABP4 after 12
h THz exposure.
(Alexandrov et al.,
2013)
Follow-up study
of (Bock et al.,
2010).
Blind procedure
not indicated.
“No effect” means no statistically significant effect.
Abbreviations:
6070
6071
Excluded papers
6072
6073
6074
6075
(Antonopoulos, Eisenbrandt & Obe, 1997; Atasoy et al., 2009; Ballardin et al., 2011; Cao et al., 2009;
Chidichimo et al., 2002; d'Ambrosio et al., 2002; Li et al., 2012b; Maes et al., 1996; Naziroğlu et al., 2012;
Pacini et al., 2002; Stankiewicz et al., 2006; Trošić & Pavičić, 2009; Yang et al., 2012; Zeni et al., 2003; ZottiMartelli et al., 2000; Zotti-Martelli et al., 2005)
6076
REFERENCES
6077
6078
Adey WR et al. (1999). Spontaneous and nitrosourea-induced primary tumors of the central nervous system in Fischer 344
rats chronically exposed to 836 MHz modulated microwaves. Radiat Res, 152(3):293-302.
6079
6080
Adey WR et al. (2000). Spontaneous and nitrosourea-induced primary tumors of the central nervous system in Fischer 344
rats exposed to frequency-modulated microwave fields. Cancer Res, 60(7):1857-1863.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
215
6081
6082
Agarwal A et al. (2009). Effects of radiofrequency electromagnetic waves (RF-EMW) from cellular phones on human
ejaculated semen: an in vitro pilot study. Fertil Steril, 92(4):1318-1325.
6083
6084
Ahlbom A, Feychting M (1999). Re: Use of cellular phones and the risk of brain tumours: a case-control study. Int J Oncol,
15(5):1045-1047. Epub 2000/01/05.
6085
Ahlbom A, Feychting M (2011). Mobile telephones and brain tumours. Br Med J, 343:d6605. Epub 2011/10/22.
6086
6087
Alexandrov BS et al. (2013). Specificity and heterogeneity of terahertz radiation effect on gene expression in mouse
mesenchymal stem cells. Sci Rep, 3:1184.
6088
Ames BN, Gold LS (1990). Too many rodent carcinogens: mitogenesis increases mutagenesis. Science, 249(4972):970-971.
6089
6090
Anane R et al. (2003). Effects of GSM-900 microwaves on DMBA-induced mammary gland tumors in female SpragueDawley rats. Radiat Res, 160(4):492-497.
6091
6092
Anderson LE et al. (2004). Two-year chronic bioassay study of rats exposed to a 1.6 GHz radiofrequency signal. Radiat Res,
162(2):201-210.
6093
6094
Anghileri LJ et al. (2005). Radiofrequency-induced carcinogenesis: cellular calcium homeostasis changes as a triggering
factor. Int J Radiat Biol, 81(3):205-209.
6095
6096
Anghileri LJ et al. (2006). Evaluation of health risks caused by radio frequency accelerated carcinogenesis: the importance of
processes driven by the calcium ion signal. Eur J Cancer Prev, 15(3):191-195.
6097
6098
Anghileri LJ, Mayayo E, Domingo JL (2009). Aluminum, calcium ion and radiofrequency synergism in acceleration of
lymphomagenesis. Immunopharmacol Immunotoxicol, 31(3):358-362.
6099
6100
Anisimov VN, Ukraintseva SV, Yashin AI (2005). Cancer in rodents: does it tell us about cancer in humans? Nat Rev
Cancer, 5(10):807-819.
6101
6102
Antonopoulos A, Eisenbrandt H, Obe G (1997). Effects of high-frequency electromagnetic fields on human lymphocytes in
vitro. Mutat Res, 395(2-3):209-214.
6103
6104
Armstrong B et al. (1994). Association between exposure to pulsed electromagnetic fields and cancer in electric utility
workers in Quebec, Canada, and France. Am J Epidemiol, 140(9):805-820. Epub 1994/11/01.
6105
6106
Atasoy A et al. (2009). The effects of electromagnetic fields on peripheral blood mononuclear cells in vitro. Bratisl Lek
Listy, 110(9):526-529. Epub 2009/10/16.
6107
6108
Atzmon I et al. (2012). Cancer risks in the Druze Isifya Village: Reasons and RF/MW antennas. Pathophysiology, 19(1):2128. Epub 2011/08/30.
6109
6110
Auvinen A et al. (2002). Brain tumors and salivary gland cancers among cellular telephone users. Epidemiology, 13(3):356359. Epub 2002/04/20.
6111
6112
Avendano C et al. (2012). Use of laptop computers connected to internet through Wi-Fi decreases human sperm motility and
increases sperm DNA fragmentation. Fertil Steril, 97(1):39-45 e32.
6113
6114
Aydin D et al. (2011). Mobile phone use and brain tumors in children and adolescents: a multicenter case-control study. J
Natl Cancer Inst, 103(16):1264-1276. Epub 2011/07/29.
6115
6116
Aydin D et al. (2012). Childhood brain tumours and use of mobile phones: comparison of a case-control study with incidence
data. Environ Health, 11:35. Epub 2012/05/23.
6117
6118
Baldi I et al. (2011). Occupational and residential exposure to electromagnetic fields and risk of brain tumors in adults: a
case-control study in Gironde, France. Int J Cancer, 129(6):1477-1484. Epub 2011/07/28.
6119
6120
Ballardin M et al. (2011). Non-thermal effects of 2.45 GHz microwaves on spindle assembly, mitotic cells and viability of
Chinese hamster V-79 cells. Mutat Res, 716(1-2):1-9. Epub 2011/08/11.
6121
6122
Baohong W et al. (2005). Studying the synergistic damage effects induced by 1.8 GHz radiofrequency field radiation (RFR)
with four chemical mutagens on human lymphocyte DNA using comet assay in vitro. Mutat Res, 578(1-2):149-157.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
216
6123
6124
Baohong W et al. (2007). Evaluating the combinative effects on human lymphocyte DNA damage induced by ultraviolet ray
C plus 1.8 GHz microwaves using comet assay in vitro. Toxicology, 232(3):311-316.
6125
6126
Barchana M, Margaliot M, Liphshitz I (2012). Changes in brain glioma incidence and laterality correlates with use of mobile
phones--a nationwide population based study in Israel. Asian Pac J Cancer Prev, 13(11):5857-5863. Epub 2013/01/16.
6127
6128
Bartsch H et al. (2002). Chronic exposure to a GSM-like signal (mobile phone) does not stimulate the development of
DMBA-induced mammary tumors in rats: results of three consecutive studies. Radiat Res, 157(2):183-190.
6129
6130
Baumgardt-Elms C et al. (2002). Testicular cancer and electromagnetic fields (EMF) in the workplace: results of a
population-based case-control study in Germany. Cancer Causes Control, 13(10):895-902. Epub 2003/02/18.
6131
6132
Behrens T et al. (2010). Occupational exposure to electromagnetic fields and sex-differential risk of uveal melanoma. Occup
Environ Med, 67(11):751-759. Epub 2010/08/28.
6133
6134
Belyaev IY et al. (2005). 915 MHz microwaves and 50 Hz magnetic field affect chromatin conformation and 53BP1 foci in
human lymphocytes from hypersensitive and healthy persons. Bioelectromagnetics, 26(3):173-184.
6135
6136
Belyaev IY et al. (2009). Microwaves from UMTS/GSM mobile phones induce long-lasting inhibition of 53BP1/gammaH2AX DNA repair foci in human lymphocytes. Bioelectromagnetics, 30(2):129-141.
6137
6138
Beneduci A et al. (2005). Frequency and irradiation time-dependant antiproliferative effect of low-power millimeter waves
on RPMI 7932 human melanoma cell line. Anticancer Res, 25(2A):1023-1028. Epub 2005/05/05.
6139
6140
Beneduci A et al. (2007). Antiproliferative effect of millimeter radiation on human erythromyeloid leukemia cell line K562
in culture: ultrastructural- and metabolic-induced changes. Bioelectrochemistry, 70(2):214-220. Epub 2006/09/09.
6141
6142
Beneduci A (2009). Evaluation of the potential in vitro antiproliferative effects of millimeter waves at some therapeutic
frequencies on RPMI 7932 human skin malignant melanoma cells. Cell Biochem Biophys, 55(1):25-32. Epub 2009/06/19.
6143
6144
Benson VS et al. (2013a). Mobile phone use and risk of brain neoplasms and other cancers: prospective study. Int J
Epidemiol, 42(3):792-802. Epub 2013/05/10.
6145
6146
Benson VS et al. (2013b). Authors' response to: The case of acoustic neuroma: comment on mobile phone use and risk of
brain neoplasms and other cancers. Int J Epidemiol. Epub 2013/10/01.
6147
6148
Berg G et al. (2006). Occupational exposure to radio frequency/microwave radiation and the risk of brain tumors: Interphone
Study Group, Germany. Am J Epidemiol, 164(6):538-548. Epub 2006/07/29.
6149
6150
Billaudel B et al. (2009a). Effects of exposure to DAMPS and GSM signals on ornithine decarboxylase (ODC) activity: II.
SH-SY5Y human neuroblastoma cells. Int J Radiat Biol, 85(6):519-522.
6151
6152
Billaudel B et al. (2009b). Effects of exposure to DAMPS and GSM signals on ornithine decarboxylase (ODC) activity: I. L929 mouse fibroblasts. Int J Radiat Biol, 85(6):510-518.
6153
6154
Bisht KS et al. (2002). The effect of 835.62 MHz FDMA or 847.74 MHz CDMA modulated radiofrequency radiation on the
induction of micronuclei in C3H 10T(1/2) cells. Radiat Res, 157(5):506-515.
6155
Bock J et al. (2010). Mammalian stem cells reprogramming in response to terahertz radiation. PLoS One, 5(12):e15806.
6156
6157
Bourthoumieu S et al. (2010). Cytogenetic studies in human cells exposed in vitro to GSM-900 MHz radiofrequency
radiation using R-banded karyotyping. Radiat Res, 174(6):712-718.
6158
6159
Bourthoumieu S et al. (2011). Aneuploidy studies in human cells exposed in vitro to GSM-900 MHz radiofrequency
radiation using FISH. Int J Radiat Biol, 87(4):400-408.
6160
6161
Bourthoumieu S et al. (2013). Study of p53 expression and post-transcriptional modifications after GSM-900 radiofrequency
exposure of human amniotic cells. Bioelectromagnetics, 34(1):52-60.
6162
6163
Brescia F et al. (2009). Reactive oxygen species formation is not enhanced by exposure to UMTS 1950 MHz radiation and
co-exposure to ferrous ions in Jurkat cells. Bioelectromagnetics, 30(7):525-535. Epub 2009/05/29.
6164
6165
Buttiglione M et al. (2007). Radiofrequency radiation (900 MHz) induces Egr-1 gene expression and affects cell-cycle
control in human neuroblastoma cells. J Cell Physiol, 213(3):759-767.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
217
6166
6167
Calabrò E et al. (2012). Modulation of heat shock protein response in SH-SY5Y by mobile phone microwaves. World J Biol
Chem, 3(2):34-40.
6168
6169
Campisi A et al. (2010). Reactive oxygen species levels and DNA fragmentation on astrocytes in primary culture after acute
exposure to low intensity microwave electromagnetic field. Neurosci Lett, 473(1):52-55.
6170
6171
Cantor KP et al. (1995). Occupational exposures and female breast cancer mortality in the United States. J Occup Environ
Med, 37(3):336-348. Epub 1995/03/01.
6172
6173
Cao Y et al. (2009). 900-MHz microwave radiation enhances gamma-ray adverse effects on SHG44 cells. J Toxicol Environ
Health A, 72(11-12):727-732.
6174
6175
6176
Capri M et al. (2004a). 1800 MHz radiofrequency (mobile phones, different Global System for Mobile communication
modulations) does not affect apoptosis and heat shock protein 70 level in peripheral blood mononuclear cells from young and
old donors. Int J Radiat Biol, 80(6):389-397.
6177
6178
Capri M et al. (2004b). In vitro exposure of human lymphocytes to 900 MHz CW and GSM modulated radiofrequency:
studies of proliferation, apoptosis and mitochondrial membrane potential. Radiat Res, 162(2):211-218. Epub 2004/09/25.
6179
6180
Caraglia M et al. (2005). Electromagnetic fields at mobile phone frequency induce apoptosis and inactivation of the multichaperone complex in human epidermoid cancer cells. J Cell Physiol, 204(2):539-548.
6181
6182
Cardis E et al. (2007). The INTERPHONE study: design, epidemiological methods, and description of the study population.
Eur J Epidemiol, 22(9):647-664. Epub 2007/07/20.
6183
6184
Cardis E et al. (2011a). Risk of brain tumours in relation to estimated RF dose from mobile phones: results from five
Interphone countries. Occup Environ Med, 68(9):631-640. Epub 2011/06/11.
6185
6186
Cardis E et al. (2011b). Estimation of RF energy absorbed in the brain from mobile phones in the Interphone Study. Occup
Environ Med, 68(9):686-693. Epub 2011/06/11.
6187
6188
Carlberg M et al. (2013). Meningioma patients diagnosed 2007--2009 and the association with use of mobile and cordless
phones: a case--control study. Environ Health, 12(1):60. Epub 2013/07/23.
6189
6190
Cervellati F et al. (2009). Effect of high-frequency electromagnetic fields on trophoblastic connexins. Reprod Toxicol,
28(1):59-65.
6191
6192
Chagnaud JL, Veyret B (1999). In vivo exposure of rats to GSM-modulated microwaves: flow cytometry analysis of
lymphocyte subpopulations and of mitogen stimulation. Int J Radiat Biol, 75(1):111-113.
6193
6194
Chauhan V et al. (2006a). Gene expression analysis of a human lymphoblastoma cell line exposed in vitro to an intermittent
1.9 GHz pulse-modulated radiofrequency field. Radiat Res, 165(4):424-429.
6195
6196
Chauhan V et al. (2006b). Analysis of proto-oncogene and heat-shock protein gene expression in human derived cell-lines
exposed in vitro to an intermittent 1.9 GHz pulse-modulated radiofrequency field. Int J Radiat Biol, 82(5):347-354.
6197
6198
Chauhan V et al. (2007a). Evaluating the biological effects of intermittent 1.9 GHz pulse-modulated radiofrequency fields in
a series of human-derived cell lines. Radiat Res, 167(1):87-93.
6199
6200
Chauhan V et al. (2007b). Analysis of gene expression in two human-derived cell lines exposed in vitro to a 1.9 GHz pulsemodulated radiofrequency field. Proteomics, 7(21):3896-3905.
6201
6202
Chemeris NK et al. (2006). Lack of direct DNA damage in human blood leukocytes and lymphocytes after in vitro exposure
to high power microwave pulses. Bioelectromagnetics, 27(3):197-203.
6203
6204
Chidichimo G et al. (2002). Selective inhibition of tumoral cells growth by low power millimeter waves. Anticancer Res,
22(3):1681-1688. Epub 2002/08/10.
6205
Chou CK et al. (1992). Long-term, low-level microwave irradiation of rats. Bioelectromagnetics, 13(6):469-496.
6206
6207
6208
Chou CK, Swicord M (2003). Comment on "Exposure of human peripheral blood lymphocytes to electromagnetic fields
associated with cellular phones leads to chromosomal instability," by Mashevich et al. Bioelectromagnetics 2003;24(2):8290. Bioelectromagnetics, 24(8):582; discussion 583-585.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
218
6209
6210
Christensen HC et al. (2004). Cellular telephone use and risk of acoustic neuroma. Am J Epidemiol, 159(3):277-283. Epub
2004/01/27.
6211
6212
Christensen HC et al. (2005). Cellular telephones and risk for brain tumors: a population-based, incident case-control study.
Neurology, 64(7):1189-1195. Epub 2005/04/13.
6213
Cleary SF et al. (1996). Effect of isothermal radiofrequency radiation on cytolytic T lymphocytes. FASEB J, 10(8):913-919.
6214
6215
Cleary SF et al. (1997). Stress proteins are not induced in mammalian cells exposed to radiofrequency or microwave
radiation. Bioelectromagnetics, 18(7):499-505.
6216
6217
Cook A et al. (2003). Cellular telephone use and time trends for brain, head and neck tumours. N Z Med J, 116(1175):U457.
Epub 2003/07/03.
6218
6219
Cooke R, Laing S, Swerdlow AJ (2010). A case-control study of risk of leukaemia in relation to mobile phone use. Br J
Cancer, 103(11):1729-1735. Epub 2010/10/14.
6220
6221
Cooper D, Hemming K, Saunders P (2001). Re: "Cancer incidence near radio and television transmitters in Great Britain. I.
Sutton Coldfield transmitter; II. All high power transmitters". Am J Epidemiol, 153(2):202-204. Epub 2001/02/13.
6222
6223
Corona AP et al. (2012). Risk factors associated with vestibular nerve schwannomas. Otol Neurotol, 33(3):459-465. Epub
2012/03/02.
6224
6225
Cranfield CG et al. (2001). Effects of mobile phone type signals on calcium levels within human leukaemic T-cells (Jurkat
cells). Int J Radiat Biol, 77(12):1207-1217.
6226
Currow D, Thomson W. Cancer in New South Wales: Incidence Report 2009. Sydney, Cancer Institute NSW, 2014.
6227
6228
Czerninski R, Zini A, Sgan-Cohen HD (2011). Risk of parotid malignant tumors in Israel (1970-2006). Epidemiology,
22(1):130-131. Epub 2010/12/15.
6229
6230
Czyz J et al. (2004). High frequency electromagnetic fields (GSM signals) affect gene expression levels in tumor suppressor
p53-deficient embryonic stem cells. Bioelectromagnetics, 25(4):296-307.
6231
6232
d'Ambrosio G et al. (2002). Cytogenetic damage in human lymphocytes following GMSK phase modulated microwave
exposure. Bioelectromagnetics, 23(1):7-13.
6233
6234
Genotoxic effects of amplitude modulated microwaves on human lymphocytes exposed in vitro under controlled conditions,
(1995).
6235
6236
De Iuliis GN et al. (2009). Mobile phone radiation induces reactive oxygen species production and DNA damage in human
spermatozoa in vitro. PLoS One, 4(7):e6446. Epub 2009/08/04.
6237
de Vocht F (2011). Cell phones and parotid cancer trends in England. Epidemiology, 22(4):608-609. Epub 2011/06/07.
6238
6239
de Vocht F, Burstyn I, Cherrie JW (2011). Time trends (1998-2007) in brain cancer incidence rates in relation to mobile
phone use in England. Bioelectromagnetics, 32(5):334-339. Epub 2011/02/01.
6240
6241
Degrave E et al. (2009). Causes of death among Belgian professional military radar operators: a 37-year retrospective cohort
study. Int J Cancer, 124(4):945-951. Epub 2008/11/28.
6242
6243
Del Vecchio G et al. (2009). Effect of radiofrequency electromagnetic field exposure on in vitro models of neurodegenerative
disease. Bioelectromagnetics, 30(7):564-572.
6244
6245
Deltour I et al. (2009). Time trends in brain tumor incidence rates in Denmark, Finland, Norway, and Sweden, 1974-2003. J
Natl Cancer Inst, 101(24):1721-1724. Epub 2009/12/05.
6246
6247
Deltour I et al. (2012). Mobile phone use and incidence of glioma in the Nordic countries 1979-2008: consistency check.
Epidemiology, 23(2):301-307. Epub 2012/01/18.
6248
6249
Desta AB, Owen RD, Cress LW (2003). Non-thermal exposure to radiofrequency energy from digital wireless phones does
not affect ornithine decarboxylase activity in L929 cells. Radiat Res, 160(4):488-491.
6250
Dickson JA (1977). The effects of hyperthermia in animal tumour systems. Recent Results Cancer Res, (59):43-111.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
219
6251
6252
Diem E et al. (2005). Non-thermal DNA breakage by mobile-phone radiation (1800 MHz) in human fibroblasts and in
transformed GFSH-R17 rat granulosa cells in vitro. Mutat Res, 583(2):178-183.
6253
6254
Ding GR et al. (2009). Comparison of Hsps expression after radio-frequency field exposure in three human glioma cell lines.
Biomed Environ Sci, 22(5):374-380.
6255
6256
Ding LX, Wang YX (2011). Increasing incidence of brain and nervous tumours in urban Shanghai, China, 1983-2007. Asian
Pac J Cancer Prev, 12(12):3319-3322. Epub 2011/01/01.
6257
6258
Dobes M et al. (2011). A multicenter study of primary brain tumor incidence in Australia (2000-2008). Neuro Oncol,
13(7):783-790. Epub 2011/07/06.
6259
6260
Dode AC et al. (2011). Mortality by neoplasia and cellular telephone base stations in the Belo Horizonte municipality, Minas
Gerais state, Brazil. Sci Total Environ, 409(19):3649-3665. Epub 2011/07/12.
6261
6262
Dolecek TA et al. (2012). CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the
United States in 2005-2009. Neuro Oncol, 14 Suppl 5:v1-49. Epub 2012/11/01.
6263
6264
Dolk H et al. (1997a). Cancer incidence near radio and television transmitters in Great Britain. II. All high power
transmitters. Am J Epidemiol, 145(1):10-17. Epub 1997/01/01.
6265
6266
Dolk H et al. (1997b). Cancer incidence near radio and television transmitters in Great Britain. I. Sutton Coldfield
transmitter. Am J Epidemiol, 145(1):1-9. Epub 1997/01/01.
6267
6268
Dreyer NA, Loughlin JE, Rothman KJ (1999). Cause-specific mortality in cellular telephone users. JAMA, 282(19):18141816.
6269
6270
Duan Y, Zhang HZ, Bu RF (2011). Correlation between cellular phone use and epithelial parotid gland malignancies. Int J
Oral Maxillofac Surg, 40(9):966-972. Epub 2011/04/09.
6271
6272
Duranti G et al. (2005). In Vitro Evaluation of Biological Effects on Human Keratinocytes Exposed to 900 MHz
Electromagnetic Field. Environmentalist, 25(2-4):113-119.
6273
6274
Eger H et al. (2004). The influence of being physically near to a cell phone transmission mast on the incidence of cancer.
Umwelt Medizin Gesellschaft, 17(4).
6275
6276
Eger H, Neppe F (2009). Krebsinzidenz von Anwohnern im Umkreis einer Mobilfunksendeanlage in Westfalen; Interviewbasierte Piloterhebung und Riskoschatzung. Umwelt Medizin Gesellschaft, 22:6.
6277
6278
Elliott P et al. (2010). Mobile phone base stations and early childhood cancers: case-control study. BMJ, 340:c3077. Epub
2010/06/24.
6279
6280
6281
Esmekaya MA et al. (2011). Mutagenic and morphologic impacts of 1.8GHz radiofrequency radiation on human peripheral
blood lymphocytes (hPBLs) and possible protective role of pre-treatment with Ginkgo biloba (EGb 761). Sci Total Environ,
410-411:59-64.
6282
6283
Evans DG et al. (2005). Incidence of vestibular schwannoma and neurofibromatosis 2 in the North West of England over a
10-year period: higher incidence than previously thought. Otol Neurotol, 26(1):93-97. Epub 2005/02/09.
6284
6285
Falzone N et al. (2010). Mobile phone radiation does not induce pro-apoptosis effects in human spermatozoa. Radiat Res,
174(2):169-176. Epub 2010/08/05.
6286
6287
Figueiredo ABS, Alves RN, Ramalho AT (2004). Cytogenetic analysis of the effects of 2.5 and 10.5 GHz microwaves on
human lymphocytes. Genetics and Molecular Biology, 27(3):460-466.
6288
Finkelstein MM (1998). Cancer incidence among Ontario police officers. Am J Ind Med, 34(2):157-162. Epub 1998/07/04.
6289
6290
Franzellitti S et al. (2008). HSP70 expression in human trophoblast cells exposed to different 1.8 Ghz mobile phone signals.
Radiat Res, 170(4):488-497.
6291
6292
Franzellitti S et al. (2010). Transient DNA damage induced by high-frequency electromagnetic fields (GSM 1.8 GHz) in the
human trophoblast HTR-8/SVneo cell line evaluated with the alkaline comet assay. Mutat Res, 683(1-2):35-42.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
220
6293
6294
Frei MR et al. (1998a). Chronic exposure of cancer-prone mice to low-level 2450 MHz radiofrequency radiation.
Bioelectromagnetics, 19(1):20-31.
6295
6296
Frei MR et al. (1998b). Chronic, low-level (1.0 W/kg) exposure of mice prone to mammary cancer to 2450 MHz
microwaves. Radiat Res, 150(5):568-576.
6297
6298
Frei P et al. (2011). Use of mobile phones and risk of brain tumours: update of Danish cohort study. BMJ, 343:d6387. Epub
2011/10/22.
6299
6300
French P, Donnellan M, McKenzie DR (1997a). Electromagnetic radiation at 835 MHz changes the morphology and inhibits
proliferation of a human astrocytoma cell line. . Bioelectrochem Bioenerg, 43(1):13 - 18.
6301
6302
French PW, Donnellan M, McKenzie DR (1997b). Electromagnetic radiation at 835 MHz changes the morphology and
inhibits proliferation of a human astrocytoma cell line. Bioelectrochemistry and Bioenergetics, 43(1):13-18.
6303
6304
Friedman J et al. (2007). Mechanism of short-term ERK activation by electromagnetic fields at mobile phone frequencies.
Biochem J, 405(3):559-568.
6305
6306
Gajski G, Garaj-Vrhovac V (2009). Radioprotective effects of honeybee venom (Apis mellifera) against 915-MHz
microwave radiation-induced DNA damage in wistar rat lymphocytes: in vitro study. Int J Toxicol, 28(2):88-98.
6307
6308
Gavin AT, Catney D (2006). Addressing a community's cancer cluster concerns. Ulster Med J, 75(3):195-199. Epub
2006/09/13.
6309
6310
Gerner C et al. (2010). Increased protein synthesis by cells exposed to a 1,800-MHz radio-frequency mobile phone
electromagnetic field, detected by proteome profiling. Int Arch Occup Environ Health, 83(6):691-702.
6311
6312
Goldstein LS et al. (2003a). Comments on "Long-term exposure of Eµ-Pim1 transgenic mice to 898.4 MHz microwaves does
not increase lymphoma incidence" by Utteridge et al., Radiat. Res. 158, 357-364 (2002). Radiat Res, 159(2):275-276.
6313
6314
6315
Goldstein LS et al. (2003b). Further Comments on "Long-Term Exposure of Eµ-Pim1 Transgenic Mice to 898.4 MHz
Microwaves does not Increase Lymphoma Incidence" by Utteridge et al. (Radiat. Res. 158, 357-364 2002). Radiat Res,
159(6):835.
6316
6317
6318
Goswami PC et al. (1999). Proto-oncogene mRNA levels and activities of multiple transcription factors in C3H 10T 1/2
murine embryonic fibroblasts exposed to 835.62 and 847.74 MHz cellular phone communication frequency radiation. Radiat
Res, 151(3):300-309.
6319
6320
Gousias K et al. (2009). Descriptive epidemiology of cerebral gliomas in Northwest Greece and study of potential
predisposing factors, 2005-2007. Neuroepidemiology, 33(2):89-95.
6321
6322
Grayson JK (1996). Radiation exposure, socioeconomic status, and brain tumor risk in the US Air Force: a nested casecontrol study. Am J Epidemiol, 143(5):480-486. Epub 1996/03/01.
6323
6324
Green AC et al. (2005). An investigation of the effects of TETRA RF fields on intracellular calcium in neurones and cardiac
myocytes. Int J Radiat Biol, 81(12):869-885.
6325
6326
Groves FD et al. (2002). Cancer in Korean war navy technicians: mortality survey after 40 years. Am J Epidemiol,
155(9):810-818. Epub 2002/04/30.
6327
6328
Gurisik E et al. (2006). An in vitro study of the effects of exposure to a GSM signal in two human cell lines: monocytic U937
and neuroblastoma SK-N-SH. Cell Biol Int, 30(10):793-799.
6329
6330
Ha M et al. (2003). Incidence of cancer in the vicinity of Korean AM radio transmitters. Arch Environ Health, 58(12):756762. Epub 2005/04/30.
6331
6332
Ha M et al. (2007). Radio-frequency radiation exposure from AM radio transmitters and childhood leukemia and brain
cancer. Am J Epidemiol, 166(3):270-279. Epub 2007/06/09.
6333
6334
Ha MI, H; Kim, B-C; Gimm Y-M; Pack, J-K (2008). Re: Radiofrequency radiation exposure from AM radio transmitters and
childhood leukaemia and brain cancer. Am J Epidemiol, 167:6.
6335
6336
Hallberg O, Johansson O (2002). Melanoma incidence and frequency modulation (FM) broadcasting. Arch Environ Health,
57(1):32-40. Epub 2002/06/20.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
221
6337
6338
Hallberg O, Johansson O (2005). FM broadcasting exposure time and malignant melanoma incidence. Electromagn Biol
Med, 24:8.
6339
6340
Han YY et al. (2012). Are frequent dental x-ray examinations associated with increased risk of vestibular schwannoma? J
Neurosurg, 117 Suppl:78-83. Epub 2012/12/12.
6341
6342
6343
Hansson Mild K, Hardell L, Carlberg M (2007). Pooled analysis of two Swedish case-control studies on the use of mobile
and cordless telephones and the risk of brain tumours diagnosed during 1997-2003. Int J Occup Saf Ergon, 13(1):63-71. Epub
2007/03/17.
6344
6345
Hansteen IL et al. (2009a). Cytogenetic effects of exposure to 2.3 GHz radiofrequency radiation on human lymphocytes in
vitro. Anticancer Res, 29(11):4323-4330.
6346
6347
Hansteen IL et al. (2009b). Cytogenetic effects of 18.0 and 16.5 GHz microwave radiation on human lymphocytes in vitro.
Anticancer Res, 29(8):2885-2892.
6348
6349
Hao Y et al. (2010). STAT3 signalling pathway is involved in the activation of microglia induced by 2.45 GHz
electromagnetic fields. Int J Radiat Biol, 86(1):27-36.
6350
6351
Hardell L et al. (1998). Case-control study on risk factors for testicular cancer. Int J Oncol, 13(6):1299-1303. Epub
1998/11/24.
6352
6353
Hardell L et al. (1999). Use of cellular telephones and the risk for brain tumours: A case-control study. Int J Oncol,
15(1):113-116.
6354
6355
Hardell L et al. (2001). Ionizing radiation, cellular telephones and the risk for brain tumours. Eur J Cancer Prev, 10(6):523529.
6356
Hardell L et al. (2002). Cellular and cordless telephones and the risk for brain tumours. Eur J Cancer Prev, 11(4):377-386.
6357
6358
Hardell L, Hansson Mild K, Carlberg M (2002). Case-control study on the use of cellular and cordless phones and the risk for
malignant brain tumours. Int J Radiat Biol, 78(10):931-936.
6359
6360
Hardell L, Hansson Mild K, Carlberg M (2003). Further aspects on cellular and cordless telephones and brain tumours. Int J
Oncol, 22(2):399-407.
6361
6362
Hardell L et al. (2003). Vestibular schwannoma, tinnitus and cellular telephones. Neuroepidemiology, 22(2):124-129. Epub
2003/03/12.
6363
6364
Hardell L et al. (2004a). No association between the use of cellular or cordless telephones and salivary gland tumours. Occup
Environ Med, 61(8):675-679. Epub 2004/07/20.
6365
6366
Hardell L et al. (2004b). Cellular and cordless telephone use and the association with brain tumors in different age groups.
Arch Environ Health, 59(3):132-137. Epub 2005/08/27.
6367
6368
Hardell L, Carlberg M, Hansson Mild K (2005a). Case-control study on cellular and cordless telephones and the risk for
acoustic neuroma or meningioma in patients diagnosed 2000-2003. Neuroepidemiology, 25(3):120-128. Epub 2005/06/16.
6369
6370
Hardell L, Carlberg M, Hansson Mild K (2005b). Use of cellular telephones and brain tumour risk in urban and rural areas.
Occup Environ Med, 62(6):390-394. Epub 2005/05/20.
6371
6372
Hardell L et al. (2005). Use of cellular or cordless telephones and the risk for non-Hodgkin's lymphoma. Int Arch Occup
Environ Health, 78(8):625-632.
6373
6374
6375
Hardell L, Carlberg M, Hansson Mild K (2006a). Pooled analysis of two case-control studies on use of cellular and cordless
telephones and the risk for malignant brain tumours diagnosed in 1997-2003. Int Arch Occup Environ Health, 79(8):630-639.
Epub 2006/03/17.
6376
6377
6378
Hardell L, Carlberg M, Hansson Mild K (2006b). Pooled analysis of two case-control studies on the use of cellular and
cordless telephones and the risk of benign brain tumours diagnosed during 1997-2003. Int J Oncol, 28(2):509-518. Epub
2006/01/05.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
222
6379
6380
6381
Hardell L, Carlberg M, Hansson Mild K (2006c). Case-control study of the association between the use of cellular and
cordless telephones and malignant brain tumors diagnosed during 2000-2003. Environ Res, 100(2):232-241. Epub
2005/07/19.
6382
6383
Hardell L et al. (2006). Tumour risk associated with use of cellular telephones or cordless desktop telephones. World J Surg
Oncol, 4:74. Epub 2006/10/13.
6384
6385
Hardell L et al. (2007). Use of cellular and cordless telephones and risk of testicular cancer. Int J Androl, 30(2):115-122.
Epub 2007/01/11.
6386
6387
Hardell L, Carlberg M (2009). Mobile phones, cordless phones and the risk for brain tumours. Int J Oncol, 35(1):5-17. Epub
2009/06/11.
6388
6389
Hardell L, Carlberg M, Hansson Mild K (2010). Mobile phone use and the risk for malignant brain tumors: a case-control
study on deceased cases and controls. Neuroepidemiology, 35(2):109-114. Epub 2010/06/17.
6390
6391
Hardell L, Carlberg M, Hansson Mild K (2011a). Pooled analysis of case-control studies on malignant brain tumours and the
use of mobile and cordless phones including living and deceased subjects. Int J Oncol, 38(5):1465-1474. Epub 2011/02/19.
6392
6393
6394
Hardell L, Carlberg M, Hansson Mild K (2011b). Re-analysis of risk for glioma in relation to mobile telephone use:
comparison with the results of the Interphone international case-control study. Int J Epidemiol, 40(4):1126-1128. Epub
2010/12/21.
6395
6396
Hardell L et al. (2011). Case-control study on the use of mobile and cordless phones and the risk for malignant melanoma in
the head and neck region. Pathophysiology, 18(4):325-333. Epub 2011/07/19.
6397
6398
Hardell L, Carlberg M (2013). Use of mobile and cordless phones and survival of patients with glioma. Neuroepidemiology,
40(2):101-108.
6399
6400
Hardell L et al. (2013a). Case-control study of the association between malignant brain tumours diagnosed between 2007 and
2009 and mobile and cordless phone use. Int J Oncol, 43(6):1833-1845. Epub 2013/09/26.
6401
6402
Hardell L et al. (2013b). Pooled analysis of case-control studies on acoustic neuroma diagnosed 1997-2003 and 2007-2009
and use of mobile and cordless phones. Int J Oncol, 43(4):1036-1044. Epub 2013/07/24.
6403
6404
Harvey C, French PW (2000). Effects on protein kinase C and gene expression in a human mast cell line, HMC-1, following
microwave exposure. Cell Biol Int, 23(11):739-748.
6405
6406
Heikkinen P et al. (2001). Effects of mobile phone radiation on X-ray-induced tumorigenesis in mice. Radiat Res,
156(6):775-785.
6407
6408
Heikkinen P et al. (2003). Effects of mobile phone radiation on UV-induced skin tumourigenesis in ornithine decarboxylase
transgenic and non-transgenic mice. Int J Radiat Biol, 79(4):221-233.
6409
6410
Heikkinen P et al. (2006). No effects of radiofrequency radiation on 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanoneinduced tumorigenesis in female Wistar rats. Radiat Res, 166(2):397-408.
6411
6412
Hepworth SJ et al. (2006). Mobile phone use and risk of glioma in adults: case-control study. BMJ, 332(7546):883-887. Epub
2006/01/24.
6413
6414
Higashikubo R et al. (1999). Radiofrequency electromagnetic fields have no effect on the in vivo proliferation of the 9L brain
tumor. Radiat Res, 152:665-671.
6415
6416
Higashikubo R et al. (2001). Radiofrequency electromagnetic fields do not alter the cell cycle progression of C3H 10T and
U87MG cells. Radiat Res, 156(6):786-795. Epub 2001/12/14.
6417
6418
Hintzsche H et al. (2011). Terahertz radiation induces spindle disturbances in human-hamster hybrid cells. Radiat Res,
175(5):569-574.
6419
6420
Hintzsche H et al. (2012a). Terahertz electromagnetic fields (0.106 THz) do not induce manifest genomic damage in vitro.
PLoS One, 7(9):e46397.
6421
6422
Hintzsche H et al. (2012b). 900 MHz radiation does not induce micronucleus formation in different cell types. Mutagenesis,
27(4):477-483.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
223
6423
6424
Hintzsche H et al. (2013). Terahertz radiation at 0.380 THz and 2.520 THz does not lead to DNA damage in skin cells in
vitro. Radiat Res, 179(1):38-45.
6425
6426
Hirose H et al. (2006). Phosphorylation and gene expression of p53 are not affected in human cells exposed to 2.1425 GHz
band CW or W-CDMA modulated radiation allocated to mobile radio base stations. Bioelectromagnetics, 27(6):494-504.
6427
6428
Hirose H et al. (2007). Mobile phone base station-emitted radiation does not induce phosphorylation of Hsp27.
Bioelectromagnetics, 28(2):99-108.
6429
6430
Hirose H et al. (2010). 1950 MHz IMT-2000 field does not activate microglial cells in vitro. Bioelectromagnetics, 31(2):104112.
6431
6432
Hocking B et al. (1996). Cancer incidence and mortality and proximity to TV towers. Med J Aust, 165(11-12):601-605. Epub
1996/12/02.
6433
6434
Hocking BG, I.; (2003). Decreased Survival from Childhood Leukaemia in Proximity to Television Towers. Arch Environ
Health, 58(9):5.
6435
6436
Holly EA et al. (1996). Intraocular melanoma linked to occupations and chemical exposures. Epidemiology, 7(1):55-61.
Epub 1996/01/01.
6437
6438
Hong MN et al. (2012). Effects of 837 and 1950 MHz radiofrequency radiation exposure alone or combined on oxidative
stress in MCF10A cells. Bioelectromagnetics, 33(7):604-611. Epub 2012/05/03.
6439
6440
Hook GJ et al. (2004a). Evaluation of parameters of oxidative stress after in vitro exposure to FMCW- and CDMAmodulated radiofrequency radiation fields. Radiat Res, 162(5):497-504. Epub 2004/12/31.
6441
6442
Hook GJ et al. (2004b). Measurement of DNA damage and apoptosis in Molt-4 cells after in vitro exposure to radiofrequency
radiation. Radiat Res, 161(2):193-200.
6443
6444
6445
Hours M et al. (2007). [Cell Phones and Risk of brain and acoustic nerve tumours: the French INTERPHONE case-control
study]. Rev Epidemiol Sante Publique, 55(5):321-332. Epub 2007/09/14. Telephone mobile, risque de tumeurs cerebrales et
du nerf vestibuloacoustique: l'etude cas-temoins INTERPHONE en France.
6446
6447
Höytö A et al. (2006). Modest increase in temperature affects ODC activity in L929 cells: Low-level radiofrequency radiation
does not. Radiat Environ Biophys, 45(3):231-235.
6448
6449
Höytö A, Juutilainen J, Naarala J (2007a). Ornithine decarboxylase activity is affected in primary astrocytes but not in
secondary cell lines exposed to 872 MHz RF radiation. Int J Radiat Biol, 83(6):367-374.
6450
6451
Höytö A, Juutilainen J, Naarala J (2007b). Ornithine decarboxylase activity of L929 cells after exposure to continuous wave
or 50 Hz modulated radiofrequency radiation--a replication study. Bioelectromagnetics, 28(7):501-508.
6452
6453
Höytö A et al. (2008a). Proliferation, oxidative stress and cell death in cells exposed to 872 MHz radiofrequency radiation
and oxidants. Radiat Res, 170(2):235-243. Epub 2008/08/01.
6454
6455
Höytö A et al. (2008b). Radiofrequency radiation does not significantly affect ornithine decarboxylase activity, proliferation,
or caspase-3 activity of fibroblasts in different physiological conditions. Int J Radiat Biol, 84(9):727-733.
6456
6457
Hruby R et al. (2008). Study on potential effects of "902-MHz GSM-type Wireless Communication Signals" on DMBAinduced mammary tumours in Sprague-Dawley rats. Mutat Res, 649(1-2):34-44.
6458
6459
Huang TQ et al. (2005). Effect of radiofrequency radiation exposure on mouse skin tumorigenesis initiated by 7,12dimethybenz[alpha]anthracene. Int J Radiat Biol, 81(12):861-867.
6460
6461
Huang TQ et al. (2008a). Molecular responses of Jurkat T-cells to 1763 MHz radiofrequency radiation. Int J Radiat Biol,
84(9):734-741.
6462
6463
Huang TQ et al. (2008b). Characterization of biological effect of 1763 MHz radiofrequency exposure on auditory hair cells.
Int J Radiat Biol, 84(11):909-915.
6464
6465
IARC - International Agency for Research on Cancer (2011). IARC Monographs on the Evaluation of Carcinogenic Risks to
Humans. Preamble. (http://monographs.iarc.fr/ENG/Preamble/CurrentPreamble.pdf , accessed 7/13/2011).
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
224
6466
6467
Imaida K et al. (1998a). The 1.5 GHz electromagnetic near-field used for cellular phones does not promote rat liver
carcinogenesis in a medium-term liver bioassay. Jpn J Cancer Res, 89(10):995-1002.
6468
6469
Imaida K et al. (1998b). Lack of promoting effects of the electromagnetic near-field used for cellular phones (929.2 MHz) on
rat liver carcinogenesis in a medium-term liver bioassay. Carcinogenesis, 19(2):311-314.
6470
6471
Imaida K et al. (2001). Lack of promotion of 7,12-dimethylbenz[a]anthracene-initiated mouse skin carcinogenesis by 1.5
GHz electromagnetic near fields. Carcinogenesis, 22(11):1837-1841.
6472
Inskip PD et al. (2001). Cellular-telephone use and brain tumors. N Engl J Med, 344(2):79-86. Epub 2001/01/11.
6473
6474
Inskip PD, Devesa SS, Fraumeni JF, Jr. (2003). Trends in the incidence of ocular melanoma in the United States, 1974-1998.
Cancer Causes Control, 14(3):251-257. Epub 2003/06/20.
6475
6476
Inskip PD, Hoover RN, Devesa SS (2010). Brain cancer incidence trends in relation to cellular telephone use in the United
States. Neuro Oncol, 12(11):1147-1151. Epub 2010/07/20.
6477
6478
Interphone Study Group (2010). Brain tumour risk in relation to mobile telephone use: results of the INTERPHONE
international case-control study. Int J Epidemiol, 39(3):675-694. Epub 2010/05/21.
6479
6480
Interphone Study Group (2011). Acoustic neuroma risk in relation to mobile telephone use: results of the INTERPHONE
international case-control study. Cancer Epidemiol, 35(5):453-464. Epub 2011/08/25.
6481
6482
Inyang I et al. (2009). How well do adolescents recall use of mobile telephones? Results of a validation study. Bmc Medical
Research Methodology, 9.
6483
6484
Ivanov VB et al. (2005). Exposure to low-intensive superhigh frequency electromagnetic field as a factor of carcinogenesis in
experimental animals. Bull Exp Biol Med, 139(2):241-244.
6485
6486
Ivaschuk OI et al. (1997). Exposure of nerve growth factor-treated PC12 rat pheochromocytoma cells to a modulated
radiofrequency field at 836.55 MHz: effects on c-jun and c-fos expression. Bioelectromagnetics, 18(3):223-229.
6487
6488
Jauchem JR et al. (2001). Repeated exposure of C3H/HeJ mice to ultra-wideband electromagnetic pulses: lack of effects on
mammary tumors. Radiat Res, 155(2):369-377.
6489
6490
Jin YB et al. (2012a). Effects of simultaneous combined exposure to CDMA and WCDMA electromagnetic field on immune
functions in rats. Int J Radiat Biol, 88(11):814-821.
6491
6492
Jin Z et al. (2012b). The effect of combined exposure of 900 MHz radiofrequency fields and doxorubicin in HL-60 cells.
PLoS One, 7(9):e46102.
6493
6494
Johansen C et al. (2001). Cellular telephones and cancer--a nationwide cohort study in Denmark. J Natl Cancer Inst,
93(3):203-207. Epub 2001/02/07.
6495
Johansen C et al. (2002). Mobile phones and malignant melanoma of the eye. Br J Cancer, 86(3):348-349. Epub 2002/03/05.
6496
6497
Joubert V et al. (2006). Microwave exposure of neuronal cells in vitro: Study of apoptosis. Int J Radiat Biol, 82(4):267-275.
Epub 2006/05/13.
6498
6499
Joubert V et al. (2007). No apoptosis is induced in rat cortical neurons exposed to GSM phone fields. Bioelectromagnetics,
28(2):115-121. Epub 2006/09/28.
6500
6501
Joubert V et al. (2008). Apoptosis is induced by radiofrequency fields through the caspase-independent mitochondrial
pathway in cortical neurons. Radiat Res, 169(1):38-45. Epub 2007/12/28.
6502
6503
Juutilainen J, Lang S, Rytomaa T (2000). Possible cocarcinogenic effects of ELF electromagnetic fields may require repeated
long-term interaction with known carcinogenic factors. Bioelectromagnetics, 21(2):122-128.
6504
6505
Karipidis KK et al. (2007a). Occupational exposure to ionizing and non-ionizing radiation and risk of glioma. Occup Med
(Lond), 57(7):518-524. Epub 2007/08/31.
6506
6507
Karipidis KK et al. (2007b). Occupational exposure to ionizing and non-ionizing radiation and risk of non-Hodgkin
lymphoma. Int Arch Occup Environ Health, 80(8):663-670. Epub 2007/03/06.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
225
6508
6509
Kaufman DW, Anderson TE, Issaragrisil S (2009). Risk factors for leukemia in Thailand. Ann Hematol, 88(11):1079-1088.
Epub 2009/03/19.
6510
6511
Keenan KP et al. (1996). The effects of diet, ad libitum overfeeding, and moderate dietary restriction on the rodent bioassay:
the uncontrolled variable in safety assessment. Toxicol Pathol, 24(6):757-768.
6512
6513
Kim HN et al. (2012). Analysis of the cellular stress response in MCF10A cells exposed to combined radio frequency
radiation. J Radiat Res, 53(2):176-183.
6514
6515
Kim JY et al. (2008). In vitro assessment of clastogenicity of mobile-phone radiation (835 MHz) using the alkaline comet
assay and chromosomal aberration test. Environ Toxicol, 23(3):319-327.
6516
6517
Kim KB et al. (2010). Two-dimensional electrophoretic analysis of radio-frequency radiation-exposed MCF7 breast cancer
cells. J Radiat Res, 51(2):205-213.
6518
6519
Klaeboe L, Blaasaas KG, Tynes T (2007). Use of mobile phones in Norway and risk of intracranial tumours. Eur J Cancer
Prev, 16(2):158-164. Epub 2007/02/14.
6520
6521
Klurfeld DM et al. (1991). Reduction of enhanced mammary carcinogenesis in LA/N-cp (corpulent) rats by energy
restriction. Proc Soc Exp Biol Med, 196(4):381-384.
6522
6523
Kohler BA et al. (2011). Annual report to the nation on the status of cancer, 1975-2007, featuring tumors of the brain and
other nervous system. J Natl Cancer Inst, 103(9):714-736. Epub 2011/04/02.
6524
6525
Komatsubara Y et al. (2005). Effect of high-frequency electromagnetic fields with a wide range of SARs on chromosomal
aberrations in murine m5S cells. Mutat Res, 587(1-2):114-119.
6526
6527
Korenstein-Ilan A et al. (2008). Terahertz radiation increases genomic instability in human lymphocytes. Radiat Res,
170(2):224-234.
6528
6529
Koyama S et al. (2003). Effects of high frequency electromagnetic fields on micronucleus formation in CHO-K1 cells. Mutat
Res, 541(1-2):81-89.
6530
6531
Koyama S et al. (2004). Effects of 2.45-GHz electromagnetic fields with a wide range of SARs on micronucleus formation in
CHO-K1 cells. ScientificWorldJournal, 4 Suppl 2:29-40.
6532
6533
Koyama S et al. (2007). Effects of 2.45 GHz electromagnetic fields with a wide range of SARs on bacterial and HPRT gene
mutations. J Radiat Res, 48(1):69-75.
6534
6535
Kumar G et al. (2011). Evaluation of hematopoietic system effects after in vitro radiofrequency radiation exposure in rats. Int
J Radiat Biol, 87(2):231-240.
6536
6537
Kundi M (2003a). Comments on the responses of Utteridge et al. (Radiat. Res. 159, 277-278, 2003) to letters about their
paper (Radiat. Res. 158, 357-364, 2002). Radiat Res, 160(5):613-614.
6538
6539
Kundi M (2003b). Comments on "Long-term exposure of Eµ-Pim1 transgenic mice to 898.4 MHz microwaves does not
increase lymphoma incidence" by Utteridge et al., Radiat. Res. 158, 357-364 (2002). Radiat Res, 159(2):274-278.
6540
6541
Kwee S, Raskmark P (1998a). Changes in cell proliferation due to environmental non-ionizing radiation. Bioelectrochemistry
and Bioenergetics, 44(2):251-255.
6542
6543
Kwee S, Raskmark P (1998b). Changes in cell proliferation due to environmental non-ionizing radiation: 2. Microwave
radiation. Bioelectrochemistry and Bioenergetics, 44(2):251-255.
6544
6545
La Regina M et al. (2003). The effect of chronic exposure to 835.62 MHz FDMA or 847.74 MHz CDMA radiofrequency
radiation on the incidence of spontaneous tumors in rats. Radiat Res, 160(2):143-151.
6546
la Vecchia C et al. (1990). Occupation and the risk of bladder cancer. Int J Epidemiol, 19(2):264-268. Epub 1990/06/01.
6547
6548
Lagorio S et al. (1997). Mortality of plastic-ware workers exposed to radiofrequencies. Bioelectromagnetics, 18(6):418-421.
Epub 1997/01/01.
6549
6550
Lagroye I et al. (2004). Measurements of alkali-labile DNA damage and protein-DNA crosslinks after 2450 MHz microwave
and low-dose gamma irradiation in vitro. Radiat Res, 161(2):201-214.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
226
6551
6552
Lahkola A et al. (2007). Mobile phone use and risk of glioma in 5 North European countries. Int J Cancer, 120(8):1769-1775.
Epub 2007/01/19.
6553
6554
Lahkola A et al. (2008). Meningioma and mobile phone use--a collaborative case-control study in five North European
countries. Int J Epidemiol, 37(6):1304-1313. Epub 2008/08/05.
6555
6556
Lai H, Singh NP (1995). Acute low-intensity microwave exposure increases DNA single-strand breaks in rat brain cells.
Bioelectromagnetics, 16(3):207-210.
6557
6558
Lai H, Singh NP (1996). Single-strand and double-strand DNA breaks in rat brain cells after acute exposure to
radiofrequency electromagnetic radiation. Int J Radiat Biol, 69(4):513-521.
6559
6560
Lantow M et al. (2006a). ROS release and Hsp70 expression after exposure to 1,800 MHz radiofrequency electromagnetic
fields in primary human monocytes and lymphocytes. Radiat Environ Biophys, 45(1):55-62.
6561
6562
Lantow M et al. (2006b). Free radical release and HSP70 expression in two human immune-relevant cell lines after exposure
to 1800 MHz radiofrequency radiation. Radiat Res, 165(1):88-94. Epub 2006/01/06.
6563
6564
Lantow M et al. (2006c). Free radical release and HSP70 expression in two human immune-relevant cell lines after exposure
to 1800 MHz radiofrequency radiation. Radiat Res, 165(1):88-94. Epub 2006/01/06.
6565
6566
Larjavaara S et al. (2011a). Incidence trends of vestibular schwannomas in Denmark, Finland, Norway and Sweden in 19872007. Br J Cancer, 105(7):1069-1075. Epub 2011/09/08.
6567
6568
Larjavaara S et al. (2011b). Location of gliomas in relation to mobile telephone use: a case-case and case-specular analysis.
Am J Epidemiol, 174(1):2-11.
6569
6570
Laszlo A et al. (2005). The heat-shock factor is not activated in mammalian cells exposed to cellular phone frequency
microwaves. Radiat Res, 164(2):163-172.
6571
6572
le Quément C et al. (2011). Whole-genome expression analysis in primary human keratinocyte cell cultures exposed to 60
GHz radiation. Bioelectromagnetics.
6573
6574
Lee HJ et al. (2011a). Lymphoma development of simultaneously combined exposure to two radiofrequency signals in
AKR/J mice. Bioelectromagnetics, 32(6):485-492.
6575
6576
Lee JJ et al. (2008). Acute radio frequency irradiation does not affect cell cycle, cellular migration, and invasion.
Bioelectromagnetics, 29(8):615-625. Epub 2008/06/03.
6577
6578
Lee JS et al. (2006). Radiofrequency radiation does not induce stress response in human T-lymphocytes and rat primary
astrocytes. Bioelectromagnetics, 27(7):578-588.
6579
6580
Lee KY et al. (2011b). Effects of combined radiofrequency radiation exposure on the cell cycle and its regulatory proteins.
Bioelectromagnetics, 32(3):169-178.
6581
6582
Lee S et al. (2005). 2.45 GHz radiofrequency fields alter gene expression in cultured human cells. FEBS Lett, 579(21):48294836.
6583
6584
Lehrer S, Green S, Stock RG (2011). Association between number of cell phone contracts and brain tumor incidence in
nineteen U.S. States. J Neurooncol, 101(3):505-507. Epub 2010/07/01.
6585
6586
Lerchl A (2003). Comments on the recent publication on microwave effects on Eµ-Pim1 transgenic mice (Utteridge et al.,
Radiat. Res. 158, 357-364, 2002). Radiat Res, 159(2):276-278.
6587
6588
6589
Lerchl A (2009). Comments on "Radiofrequency electromagnetic fields (UMTS, 1,950 MHz) induce genotoxic effects in
vitro in human fibroblasts but not in lymphocytes" by Schwarz et al. (Int Arch Occup Environ Health 2008: doi:
10.1007/s00420-008-0305-5). Int Arch Occup Environ Health, 82(2):275-278.
6590
6591
Lerchl A, Wilhelm AF (2010). Critical comments on DNA breakage by mobile-phone electromagnetic fields [Diem et al.,
Mutat. Res. 583 (2005) 178-183]. Mutat Res, 697(1-2):60-65.
6592
6593
6594
Leszczynski D et al. (2002). Non-thermal activation of the hsp27/p38MAPK stress pathway by mobile phone radiation in
human endothelial cells: molecular mechanism for cancer- and blood-brain barrier-related effects. Differentiation, 70(23):120-129.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
227
6595
6596
Li CY et al. (2012a). A population-based case-control study of radiofrequency exposure in relation to childhood neoplasm.
Sci Total Environ, 435-436:472-478. Epub 2012/08/14.
6597
6598
Li HW et al. (2007). Proteomic analysis of human lens epithelial cells exposed to microwaves. Jpn J Ophthalmol, 51(6):412416.
6599
6600
Li JR et al. (1999). TP53 tumor suppressor protein in normal human fibroblasts does not respond to 837 MHz microwave
exposure. Radiat Res, 151(6):710-716.
6601
6602
Li L et al. (2001). Measurement of DNA damage in mammalian cells exposed in vitro to radiofrequency fields at SARs of 35 W/kg. Radiat Res, 156(3):328-332.
6603
6604
Li X et al. (2012b). Millimeter wave treatment promotes chondrocyte proliferation via G1/S cell cycle transition. Int J Mol
Med, 29(5):823-831.
6605
6606
Liberti M et al. (2004). A coplanar-waveguide system for cell exposure during electrophysiological recordings. IEEE Trans
Microw Theory Tech, 52(11):2521-2528.
6607
6608
Lim HB et al. (2005). Effect of 900 MHz electromagnetic fields on nonthermal induction of heat-shock proteins in human
leukocytes. Radiat Res, 163(1):45-52.
6609
Linet MS et al. (2006). Cellular telephones and non-Hodgkin lymphoma. Int J Cancer, 119(10):2382-2388. Epub 2006/08/09.
6610
6611
Linz KW et al. (1999). Membrane potential and currents of isolated heart muscle cells exposed to pulsed radio frequency
fields. Bioelectromagnetics, 20(8):497-511.
6612
6613
Litovitz TA et al. (1993). The role of coherence time in the effect of microwaves on ornithine decarboxylase activity.
Bioelectromagnetics, 14(5):395-403.
6614
6615
Litovitz TA et al. (1997). Bioeffects induced by exposure to microwaves are mitigated by superposition of ELF noise.
Bioelectromagnetics, 18(6):422-430.
6616
6617
Little MP et al. (2012). Mobile phone use and glioma risk: comparison of epidemiological study results with incidence trends
in the United States. BMJ, 344:e1147. Epub 2012/03/10.
6618
6619
Liu C et al. (2013). Exposure to 1800 MHz radiofrequency electromagnetic radiation induces oxidative DNA base damage in
a mouse spermatocyte-derived cell line. Toxicol Lett, 218(1):2-9.
6620
6621
Liu ML, Wen JQ, Fan YB (2011). Potential protection of green tea polyphenols against 1800 MHz electromagnetic radiationinduced injury on rat cortical neurons. Neurotox Res, 20(3):270-276.
6622
6623
Liu YX et al. (2012). Exposure to 1950-MHz TD-SCDMA electromagnetic fields affects the apoptosis of astrocytes via
caspase-3-dependent pathway. PLoS One, 7(8):e42332.
6624
6625
Lixia S et al. (2006). Effects of 1.8 GHz radiofrequency field on DNA damage and expression of heat shock protein 70 in
human lens epithelial cells. Mutat Res, 602(1-2):135-142.
6626
Lönn S et al. (2004a). Mobile phone use and the risk of acoustic neuroma. Epidemiology, 15(6):653-659. Epub 2004/10/12.
6627
6628
Lönn S et al. (2004b). Output power levels from mobile phones in different geographical areas; implications for exposure
assessment. Occup Environ Med, 61(9):769-772.
6629
Lönn S et al. (2005). Long-term mobile phone use and brain tumor risk. Am J Epidemiol, 161(6):526-535. Epub 2005/03/05.
6630
Lönn S et al. (2006). Mobile phone use and risk of parotid gland tumor. Am J Epidemiol, 164(7):637-643. Epub 2006/07/05.
6631
6632
Lu YS, Huang BT, Huang YX (2012). Reactive oxygen species formation and apoptosis in human peripheral blood
mononuclear cell induced by 900 MHz mobile phone radiation. Oxid Med Cell Longev, 2012:740280. Epub 2012/07/11.
6633
6634
Luukkonen J et al. (2009). Enhancement of chemically induced reactive oxygen species production and DNA damage in
human SH-SY5Y neuroblastoma cells by 872 MHz radiofrequency radiation. Mutat Res, 662(1-2):54-58.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
228
6635
6636
6637
Luukkonen J, Juutilainen J, Naarala J (2010). Combined effects of 872 MHz radiofrequency radiation and ferrous chloride on
reactive oxygen species production and DNA damage in human SH-SY5Y neuroblastoma cells. Bioelectromagnetics,
31(6):417-424.
6638
6639
Maes A et al. (1993). In vitro cytogenetic effects of 2450 MHz waves on human peripheral blood lymphocytes.
Bioelectromagnetics, 14(6):495-501.
6640
6641
Maes A et al. (1995). Cytogenetic Effects of Microwaves from Mobile Communication Frequencies (954 Mhz). Electro- and
Magnetobiology, 14(2):91-98.
6642
6643
Maes A et al. (1996). 954 MHz microwaves enhance the mutagenic properties of mitomycin C. Environ Mol Mutagen,
28(1):26-30.
6644
6645
Maes A et al. (1997). Cytogenetic effects of 935.2-MHz (GSM) microwaves alone and in combination with mitomycin C.
Mutat Res, 393(1-2):151-156.
6646
6647
Maes A, Collier M, Verschaeve L (2000). Cytogenetic investigations on microwaves emitted by a 455.7 MHz car phone.
Folia Biol (Praha), 46(5):175-180.
6648
6649
Maes A, Collier M, Verschaeve L (2001). Cytogenetic effects of 900 MHz (GSM) microwaves on human lymphocytes.
Bioelectromagnetics, 22(2):91-96.
6650
6651
Malyapa RS et al. (1997a). Measurement of DNA damage after exposure to 2450 MHz electromagnetic radiation. Radiat
Res, 148(6):608-617.
6652
6653
Malyapa RS et al. (1997b). Measurement of DNA damage after exposure to electromagnetic radiation in the cellular phone
communication frequency band (835.62 and 847.74 MHz). Radiat Res, 148(6):618-627.
6654
6655
Manti L et al. (2008). Effects of modulated microwave radiation at cellular telephone frequency (1.95 GHz) on X-rayinduced chromosome aberrations in human lymphocytes in vitro. Radiat Res, 169(5):575-583.
6656
6657
6658
Marchionni I et al. (2006). Comparison between low-level 50 Hz and 900 MHz electromagnetic stimulation on single
channel ionic currents and on firing frequency in dorsal root ganglion isolated neurons. Biochim Biophys Acta, 1758(5):597605.
6659
6660
Marinelli F et al. (2004). Exposure to 900 MHz electromagnetic field induces an unbalance between pro-apoptotic and prosurvival signals in T-lymphoblastoid leukemia CCRF-CEM cells. J Cell Physiol, 198(2):324-332.
6661
6662
Markova E et al. (2005). Microwaves from GSM mobile telephones affect 53BP1 and gamma-H2AX foci in human
lymphocytes from hypersensitive and healthy persons. Environ Health Perspect, 113(9):1172-1177.
6663
6664
6665
Markova E, Malmgren LOG, Belyaev IY (2010). Microwaves from mobile phones inhibit 53BP1 focus formation in human
stem cells more strongly than in differentiated cells: possible mechanistic link to cancer risk. Environmental Health
Perspectives, 118(3):394-399.
6666
6667
Mashevich M et al. (2003). Exposure of human peripheral blood lymphocytes to electromagnetic fields associated with
cellular phones leads to chromosomal instability. Bioelectromagnetics, 24(2):82-90.
6668
6669
Maskarinec G, Cooper J, Swygert L (1994). Investigation of increased incidence in childhood leukemia near radio towers in
Hawaii: preliminary observations. J Environ Pathol Toxicol Oncol, 13(1):33-37. Epub 1994/01/01.
6670
6671
Mazor R et al. (2008). Increased levels of numerical chromosome aberrations after in vitro exposure of human peripheral
blood lymphocytes to radiofrequency electromagnetic fields for 72 hours. Radiat Res, 169(1):28-37.
6672
6673
McKean-Cowdin R et al. (2013). Trends in childhood brain tumor incidence, 1973-2009. J Neurooncol, 115(2):153-160.
Epub 2013/08/09.
6674
6675
McKenzie DR, Yin Y, Morrell S (1998). Childhood incidence of acute lymphoblastic leukaemia and exposure to broadcast
radiation in Sydney--a second look. Aust N Z J Public Health, 22(3 Suppl):360-367. Epub 1998/06/18.
6676
6677
McNamee JP et al. (2002a). DNA damage in human leukocytes after acute in vitro exposure to a 1.9 GHz pulse-modulated
radiofrequency field. Radiat Res, 158(4):534-537.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
229
6678
6679
McNamee JP et al. (2002b). DNA damage and micronucleus induction in human leukocytes after acute in vitro exposure to a
1.9 GHz continuous-wave radiofrequency field. Radiat Res, 158(4):523-533.
6680
6681
McNamee JP et al. (2003). No evidence for genotoxic effects from 24 h exposure of human leukocytes to 1.9 GHz
radiofrequency fields. Radiat Res, 159(5):693-697.
6682
6683
Merola P et al. (2006). Proliferation and apoptosis in a neuroblastoma cell line exposed to 900 MHz modulated
radiofrequency field. Bioelectromagnetics, 27(3):164-171.
6684
6685
Merzenich H et al. (2008). Childhood leukemia in relation to radio frequency electromagnetic fields in the vicinity of TV and
radio broadcast transmitters. Am J Epidemiol, 168(10):1169-1178. Epub 2008/10/07.
6686
6687
Meyer MG-D, A.; Radespiel-Troger, M. (2006). Mobilefunkbasisstationen und Krebshaufigkeit in Bayern. Umweltmed
Forsch Prax, 11(2):9.
6688
6689
Michelozzi P et al. (2002). Adult and childhood leukemia near a high-power radio station in Rome, Italy. Am J Epidemiol,
155(12):1096-1103. Epub 2002/06/06.
6690
6691
Milham S, Jr. (1988). Increased mortality in amateur radio operators due to lymphatic and hematopoietic malignancies. Am J
Epidemiol, 127(1):50-54. Epub 1988/01/01.
6692
6693
Miyakoshi J et al. (2002). Effects oh high-frequency electromagnetic fields on DNA strand breacks using comet assay
method. Electrical Eng Jpn, 141:9-15.
6694
6695
Miyakoshi J et al. (2005). Effects of exposure to a 1950 MHz radio frequency field on expression of Hsp70 and Hsp27 in
human glioma cells. Bioelectromagnetics, 26(4):251-257.
6696
6697
Moisescu MG et al. (2008). Microscopic observation of living cells during their exposure to modulated electromagnetic
fields. Bioelectrochemistry, 74(1):9-15. Epub 2008/02/15.
6698
6699
Moquet J et al. (2008). Exposure to low level GSM 935 MHZ radiofrequency fields does not induce apoptosis in proliferating
or differentiated murine neuroblastoma cells. Radiat Prot Dosimetry, 131(3):287-296. Epub 2008/06/14.
6700
6701
Morgan RW et al. (2000). Radiofrequency exposure and mortality from cancer of the brain and lymphatic/hematopoietic
systems. Epidemiology, 11(2):118-127. Epub 2000/10/06.
6702
6703
Muscat JE et al. (2000). Handheld cellular telephone use and risk of brain cancer. JAMA, 284(23):3001-3007. Epub
2000/12/21.
6704
6705
Muscat JE et al. (2002). Handheld cellular telephones and risk of acoustic neuroma. Neurology, 58(8):1304-1306. Epub
2002/04/24.
6706
6707
Natarajan M et al. (2002). NF-kappaB DNA-binding activity after high peak power pulsed microwave (8.2 GHz) exposure of
normal human monocytes. Bioelectromagnetics, 23(4):271-277.
6708
6709
6710
Natarajan M et al. (2006). Nuclear translocation and DNA-binding activity of NFKB (NF-kappaB) after exposure of human
monocytes to pulsed ultra-wideband electromagnetic fields (1 kV/cm) fails to transactivate kappaB-dependent gene
expression. Radiat Res, 165(6):645-654.
6711
6712
Naziroğlu M et al. (2012). 2.45-Gz wireless devices induce oxidative stress and proliferation through cytosolic Ca(2)(+)
influx in human leukemia cancer cells. Int J Radiat Biol, 88(6):449-456. Epub 2012/04/12.
6713
6714
Nelson PD et al. (2006). Trends in acoustic neuroma and cellular phones: is there a link? Neurology, 66(2):284-285. Epub
2006/01/26.
6715
6716
Ni S et al. (2013). Study of oxidative stress in human lens epithelial cells exposed to 1.8 GHz radiofrequency fields. PLoS
One, 8(8):e72370.
6717
6718
Nicolaz CN et al. (2009). Study of narrow band millimeter-wave potential interactions with endoplasmic reticulum stress
sensor genes. Bioelectromagnetics, 30(5):365-373.
6719
6720
Nikolova T et al. (2005). Electromagnetic fields affect transcript levels of apoptosis-related genes in embryonic stem cellderived neural progenitor cells. FASEB J, 19(12):1686-1688.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
230
6721
6722
Nomura E, Ioka A, Tsukuma H (2011). Trends in the incidence of primary intracranial tumors in Osaka, Japan. Jpn J Clin
Oncol, 41(2):291-294. Epub 2011/01/29.
6723
6724
Nylund R, Leszczynski D (2004). Proteomics analysis of human endothelial cell line EA.hy926 after exposure to GSM 900
radiation. Proteomics, 4(5):1359-1365.
6725
6726
Nylund R, Leszczynski D (2006). Mobile phone radiation causes changes in gene and protein expression in human
endothelial cell lines and the response seems to be genome- and proteome-dependent. Proteomics, 6(17):4769-4780.
6727
6728
Nylund R et al. (2009). Proteomic analysis of the response of human endothelial cell line EA.hy926 to 1800 GSM mobile
phone radiation. J Proteomics Bioinform, 2(10):455-462.
6729
6730
Nylund R, Kuster N, Leszczynski D (2010). Analysis of proteome response to the mobile phone radiation in two types of
human primary endothelial cells. Proteome Sci, 8:52.
6731
6732
O'Connor RP et al. (2010). Exposure to GSM RF fields does not affect calcium homeostasis in human endothelial cells, rat
pheocromocytoma cells or rat hippocampal neurons. PLoS One, 5(7):e11828.
6733
6734
Oberto G et al. (2007). Carcinogenicity study of 217 Hz pulsed 900 MHz electromagnetic fields in Pim1 transgenic mice.
Radiat Res, 168(3):316-326.
6735
6736
Orel VE et al. (2004). Effects of mechanochemically activated doxorubicin and 40 MHz frequency irradiation on human A549 lung carcinoma cells. Exp Oncol, 26(4):271-277. Epub 2005/01/01.
6737
6738
Pacini S et al. (2002). Exposure to global system for mobile communication (GSM) cellular phone radiofrequency alters gene
expression, proliferation, and morphology of human skin fibroblasts. Oncol Res, 13(1):19-24.
6739
6740
Palumbo R et al. (2008). Exposure to 900 MHz radiofrequency radiation induces caspase 3 activation in proliferating human
lymphocytes. Radiat Res, 170(3):327-334. Epub 2008/09/04.
6741
6742
Park SK, Ha M, Im HJ (2004). Ecological study on residences in the vicinity of AM radio broadcasting towers and cancer
death: preliminary observations in Korea. Int Arch Occup Environ Health, 77(6):387-394. Epub 2004/09/01.
6743
6744
Paulraj R, Behari J (2011). Effects of low level microwave radiation on carcinogenesis in Swiss Albino mice. Mol Cell
Biochem, 348(1-2):191-197.
6745
6746
Pavicic I, Trosic I (2006). Influence of 864 MHz electromagnetic field on growth kinetics of established cell line. Biologia,
61(3):321-325.
6747
6748
Pavičić I, Trošić I (2008). Impact of 864 MHz or 935 MHz radiofrequency microwave radiation on the basic growth
parameters of V79 cell line. Acta Biol Hung, 59(1):67-76. Epub 2008/04/12.
6749
6750
Peinnequin A et al. (2000). Non-thermal effects of continuous 2.45 GHz microwaves on Fas-induced apoptosis in human
Jurkat T-cell line. Bioelectrochemistry, 51(2):157-161. Epub 2000/07/26.
6751
6752
Penafiel LM et al. (1997). Role of modulation on the effect of microwaves on ornithine decarboxylase activity in L929 cells.
Bioelectromagnetics, 18(2):132-141.
6753
6754
Perez FP et al. (2008). Electromagnetic field therapy delays cellular senescence and death by enhancement of the heat shock
response. Exp Gerontol, 43(4):307-316.
6755
6756
Phillips JL et al. (1998). DNA damage in Molt-4 T-lymphoblastoid cells exposed to cellular telephone radiofrequency fields
in vitro. Bioelec Chem Bioenerg, 45:103-110.
6757
6758
Platano D et al. (2007). Acute exposure to low-level CW and GSM-modulated 900 MHz radiofrequency does not affect Ba
2+ currents through voltage-gated calcium channels in rat cortical neurons. Bioelectromagnetics, 28(8):599-607.
6759
6760
Port M et al. (2003). Influence of high-frequency electromagnetic fields on different modes of cell death and gene expression.
Int J Radiat Biol, 79(9):701-708.
6761
6762
Poulletier de Gannes F et al. (2011). Effect of exposure to the edge signal on oxidative stress in brain cell models. Radiat Res,
175(2):225-230. Epub 2011/01/28.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
231
6763
6764
Poulsen AH et al. (2013). Mobile phone use and the risk of skin cancer: a nationwide cohort study in Denmark. Am J
Epidemiol, 178(2):190-197. Epub 2013/06/22.
6765
6766
Preece AW et al. (2007). Health response of two communities to military antennae in Cyprus. Occup Environ Med,
64(6):402-408. Epub 2007/01/30.
6767
Propp JM et al. (2006). Descriptive epidemiology of vestibular schwannomas. Neuro Oncol, 8(1):1-11. Epub 2006/01/31.
6768
6769
PTS - Post- och Telestyrelsen. Så efterfrågar vi elektronisk kommunikation - en individundersökning 2003. Report PTS-ER2003:43. 2003.
6770
6771
PTS - Post- och Telestyrelsen. Svenskarnas användning av telefoni & internet 2011, Individundersökning 2011. Report PTSER-2011:23. 2011.
6772
6773
PTS - The Swedish Post and Telecom Agency (2011). Statistics Portal. (http://www.statistik.pts.se/pts2011/index.html,
accessed 14 January 2013).
6774
6775
Qutob SS et al. (2006). Microarray gene expression profiling of a human glioblastoma cell line exposed in vitro to a 1.9 GHz
pulse-modulated radiofrequency field. Radiat Res, 165(6):636-644.
6776
6777
Rao VS et al. (2008). Nonthermal effects of radiofrequency-field exposure on calcium dynamics in stem cell-derived
neuronal cells: elucidation of calcium pathways. Radiat Res, 169(3):319-329.
6778
6779
Remondini D et al. (2006). Gene expression changes in human cells after exposure to mobile phone microwaves. Proteomics,
6(17):4745-4754.
6780
6781
Repacholi MH et al. (1997). Lymphomas in E-Pim1 transgenic mice exposed to pulsed 900 MHz electromagnetic fields.
Radiat Res, 147(5):631-640.
6782
6783
Robinette CD, Silverman C, Jablon S (1980). Effects upon health of occupational exposure to microwave radiation (radar).
Am J Epidemiol, 112(1):39-53. Epub 1980/07/01.
6784
6785
Röösli M et al. (2007). Cellular telephone use and time trends in brain tumour mortality in Switzerland from 1969 to 2002.
Eur J Cancer Prev, 16(1):77-82. Epub 2007/01/16.
6786
6787
Rosychuk RJ et al. (2012). Central nervous system (CNS) tumor trends in children in a western Canadian province: a
population-based 22-year retrospective study. J Neurol, 259(6):1131-1136. Epub 2011/11/24.
6788
6789
Roux D et al. (2011). Human keratinocytes in culture exhibit no response when exposed to short duration, low amplitude,
high frequency (900 MHz) electromagnetic fields in a reverberation chamber. Bioelectromagnetics, 32(4):302-311.
6790
Rüdiger H, Kratochvil E, Pilger A (2006). Reply to the letter by Vijayalaxmi et al. Mutat Res, 603:107-109.
6791
6792
6793
Rüdiger HW (2009a). Answer to comments by A. Lerchl on "Radiofrequency electromagnetic fields (UMTS, 1,950 MHz)
induce genotoxic effects in vitro in human fibroblasts but not in lymphocytes" published by C. Schwarz et al. 2008. Int Arch
Occup Environ Health, 82(2):279-283.
6794
6795
Rüdiger HW (2009b). Letter to the editor: doubts raised about the blinding process do not apply to the Diem et al. paper.
Mutat Res, 673(1):2.
6796
6797
Sadetzki S et al. (2008). Cellular phone use and risk of benign and malignant parotid gland tumors--a nationwide case-control
study. Am J Epidemiol, 167(4):457-467. Epub 2007/12/08.
6798
6799
Safronova VG, Gabdoulkhakova AG, Santalov BF (2002). Immunomodulating action of low intensity millimeter waves on
primed neutrophils. Bioelectromagnetics, 23(8):599-606.
6800
6801
Saika K, Katanoda K (2011). Comparison of time trends in brain and central nervous system cancer mortality (1990-2006)
between countries based on the WHO mortality database. Jpn J Clin Oncol, 41(2):304-305. Epub 2011/01/29.
6802
6803
Sakuma N et al. (2006). DNA strand breaks are not induced in human cells exposed to 2.1425 GHz band CW and W-CDMA
modulated radiofrequency fields allocated to mobile radio base stations. Bioelectromagnetics, 27(1):51-57.
6804
6805
Sakurai T et al. (2011). Analysis of gene expression in a human-derived glial cell line exposed to 2.45 GHz continuous
radiofrequency electromagnetic fields. J Radiat Res, 52(2):185-192.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
232
6806
6807
Salford LG et al. (1993). Experimental studies of brain tumor development during exposure with continuous and pulsed 915
MHz radiofrequency radiation. Bioelectrochem Bioenerg, 30(1-3):313-318.
6808
6809
Salford LG, Brun A, Persson BR (1997). Brain tumor development in rats exposed to electromagnetic fields used in wireless
communication. Wireless Netw, 3:463-469.
6810
6811
Sanchez S et al. (2006a). Effect of GSM-900 and -1800 signals on the skin of hairless rats. II: 12-week chronic exposures. Int
J Radiat Biol, 82(9):675-680.
6812
6813
Sanchez S et al. (2006b). Human skin cell stress response to GSM-900 mobile phone signals. In vitro study on isolated
primary cells and reconstructed epidermis. FEBS J, 273(24):5491-5507.
6814
6815
Sanchez S et al. (2007). In vitro study of the stress response of human skin cells to GSM-1800 mobile phone signals
compared to UVB radiation and heat shock. Radiat Res, 167(5):572-580.
6816
6817
Sannino A et al. (2006). Evaluation of cytotoxic and genotoxic effects in human peripheral blood leukocytes following
exposure to 1950-MHz modulated signal. Ieee Transactions on Plasma Science, 34(4):1441-1448.
6818
6819
Sannino A et al. (2009a). Human fibroblasts and 900 MHz radiofrequency radiation: evaluation of DNA damage after
exposure and co-exposure to 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5h)-furanone (MX). Radiat Res, 171(6):743-751.
6820
6821
Sannino A et al. (2009b). Induction of adaptive response in human blood lymphocytes exposed to radiofrequency radiation.
Radiat Res, 171(6):735-742.
6822
6823
Sannino A et al. (2011). Induction of adaptive response in human blood lymphocytes exposed to 900 MHz radiofrequency
fields: influence of cell cycle. Int J Radiat Biol, 87(9):993-999.
6824
6825
Santini R et al. (1988). B16 melanoma development in black mice exposed to low-level microwave radiation.
Bioelectromagnetics, 9(1):105-107.
6826
6827
Saran A et al. (2007). Effects of exposure of newborn patched1 heterozygous mice to GSM, 900 MHz. Radiat Res,
168(6):733-740.
6828
6829
Sarimov R et al. (2004). Nonthermal GSM microwaves affect chromatin conformation in human lymphocytes similar to heat
shock. Ieee Transactions on Plasma Science, 32(4):1600-1608.
6830
6831
Sato Y et al. (2011). A case-case study of mobile phone use and acoustic neuroma risk in Japan. Bioelectromagnetics,
32(2):85-93. Epub 2011/01/13.
6832
6833
Savitz DA et al. (1997). Lung cancer in relation to employment in the electrical utility industry and exposure to magnetic
fields. Occup Environ Med, 54(6):396-402. Epub 1997/06/01.
6834
6835
Scarfi MR et al. (1996). Genotoxic effects of mitomycin-C and microwave radiation on bovine lymphocytes. Electro- and
Magnetobiology, 15(2):99-107.
6836
6837
Scarfi MR et al. (2003). THz Exposure of Whole Blood for the Study of Biological Effects on Human Lymphocytes. J Biol
Phys, 29(2-3):171-176.
6838
6839
6840
Scarfi MR et al. (2006). Exposure to radiofrequency radiation (900 MHz, GSM signal) does not affect micronucleus
frequency and cell proliferation in human peripheral blood lymphocytes: an interlaboratory study. Radiat Res, 165(6):655663.
6841
6842
Schlehofer B et al. (2007). Environmental risk factors for sporadic acoustic neuroma (Interphone Study Group, Germany).
Eur J Cancer, 43(11):1741-1747. Epub 2007/06/30.
6843
6844
Schmid E, Schrader T (2007). Different biological effectiveness of ionising and non-ionising radiations in mammalian cells.
Advances in Radio Science, 5:1-4.
6845
6846
Schmiedel S et al. (2009). An evaluation of exposure metrics in an epidemiologic study on radio and television broadcast
transmitters and the risk of childhood leukemia. Bioelectromagnetics, 30(2):81-91. Epub 2008/11/26.
6847
6848
Schoemaker MJ et al. (2005). Mobile phone use and risk of acoustic neuroma: results of the Interphone case-control study in
five North European countries. Br J Cancer, 93(7):842-848. Epub 2005/09/02.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
233
6849
6850
Schoemaker MJ, Swerdlow AJ (2009). Risk of pituitary tumors in cellular phone users: a case-control study. Epidemiology,
20(3):348-354. Epub 2009/03/13.
6851
6852
Schönborn F, Pokovic K, Kuster N (2004). Dosimetric analysis of the carousel setup for the exposure of rats at 1.62 GHz.
Bioelectromagnetics, 25(1):16-26.
6853
6854
Schrader T et al. (2008). Spindle disturbances in human-hamster hybrid (AL) cells induced by mobile communication
frequency range signals. Bioelectromagnetics, 29(8):626-639.
6855
6856
Schrader T et al. (2011). Spindle disturbances in human-hamster hybrid (A(L) ) cells induced by the electrical component of
the mobile communication frequency range signal. Bioelectromagnetics, 32(4):291-301.
6857
6858
Schüz J et al. (2006a). Cellular phones, cordless phones, and the risks of glioma and meningioma (Interphone Study Group,
Germany). Am J Epidemiol, 163(6):512-520. Epub 2006/01/31.
6859
6860
Schüz J et al. (2006b). Radiofrequency electromagnetic fields emitted from base stations of DECT cordless phones and the
risk of glioma and meningioma (Interphone Study Group, Germany). Radiat Res, 166(1 Pt 1):116-119. Epub 2006/07/01.
6861
6862
Schüz J et al. (2006c). Cellular telephone use and cancer risk: update of a nationwide Danish cohort. J Natl Cancer Inst,
98(23):1707-1713. Epub 2006/12/07.
6863
6864
Schüz J et al. (2011). Long-term mobile phone use and the risk of vestibular schwannoma: a Danish nationwide cohort study.
Am J Epidemiol, 174(4):416-422. Epub 2011/06/30.
6865
6866
Schwarz C et al. (2008). Radiofrequency electromagnetic fields (UMTS, 1,950 MHz) induce genotoxic effects in vitro in
human fibroblasts but not in lymphocytes. Int Arch Occup Environ Health, 81(6):755-767.
6867
6868
Sekijima M et al. (2010). 2-GHz band CW and W-CDMA modulated radiofrequency fields have no significant effect on cell
proliferation and gene expression profile in human cells. J Radiat Res, 51(3):277-284.
6869
6870
Selvin S, Schulman J, Merrill DW (1992). Distance and risk measures for the analysis of spatial data: a study of childhood
cancers. Soc Sci Med, 34(7):769-777. Epub 1992/04/01.
6871
6872
Sen A et al. (2011). Mild elevation of body temperature reduces tumor interstitial fluid pressure and hypoxia and enhances
efficacy of radiotherapy in murine tumor models. Cancer Res, 71(11):3872-3880.
6873
6874
Shckorbatov YG et al. (1998). Microwave irradiation influences on the state of human cell nuclei. Bioelectromagnetics,
19(7):414-419.
6875
6876
Shckorbatov YG et al. (2010). Effects of differently polarized microwave radiation on the microscopic structure of the nuclei
in human fibroblasts. J Zhejiang Univ Sci B, 11(10):801-805.
6877
6878
Shirai T et al. (2005). Chronic exposure to a 1.439 GHz electromagnetic field used for cellular phones does not promote Nethylnitrosourea induced central nervous system tumors in F344 rats. Bioelectromagnetics, 26(1):59-68.
6879
6880
6881
Shirai T et al. (2007). Lack of promoting effects of chronic exposure to 1.95-GHz W-CDMA signals for IMT-2000 cellular
system on development of N-ethylnitrosourea-induced central nervous system tumors in F344 rats. Bioelectromagnetics,
28(7):562-572.
6882
6883
Shu X, Ahlbom A, Feychting M (2012). Incidence trends of malignant parotid gland tumors in Swedish and nordic adults
1970 to 2009. Epidemiology, 23(5):766-767. Epub 2012/08/09.
6884
6885
Simkó M et al. (2006). Hsp70 expression and free radical release after exposure to non-thermal radio-frequency
electromagnetic fields and ultrafine particles in human Mono Mac 6 cells. Toxicol Lett, 161(1):73-82.
6886
6887
Simon D et al. (2013). Exposure to acute electromagnetic radiation of mobile phone exposure range alters transiently skin
homeostasis of a model of pigmented reconstructed epidermis. Int J Cosmet Sci, 35(1):27-34.
6888
6889
Sinha DK, Gebhard RL, Pazik JE (1988). Inhibition of mammary carcinogenesis in rats by dietary restriction. Cancer Lett,
40(2):133-141.
6890
6891
Smith P et al. (2007). GSM and DCS wireless communication signals: combined chronic toxicity/carcinogenicity study in the
Wistar rat. Radiat Res, 168(4):480-492.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
234
6892
6893
Söderqvist F, Carlberg M, Hardell L (2012). Use of wireless phones and the risk of salivary gland tumours: a case-control
study. Eur J Cancer Prev, 21(6):576-579. Epub 2012/03/22.
6894
6895
Sommer AM et al. (2004). No effects of GSM-modulated 900 MHz electromagnetic fields on survival rate and spontaneous
development of lymphoma in female AKR/J mice. BMC Cancer, 4:77.
6896
6897
Sommer AM et al. (2007). Lymphoma development in mice chronically exposed to UMTS-modulated radiofrequency
electromagnetic fields. Radiat Res, 168(1):72-80.
6898
6899
Song XL et al. (2011). Microwave induces apoptosis in A549 human lung carcinoma cell line. Chin Med J (Engl),
124(8):1193-1198.
6900
6901
Speit G, Schütz P, Hoffmann H (2007). Genotoxic effects of exposure to radiofrequency electromagnetic fields (RF-EMF) in
cultured mammalian cells are not independently reproducible. Mutat Res, 626(1-2):42-47.
6902
6903
Spinelli V et al. (2010). Occupational and environmental risk factors for brain cancer: a pilot case-control study in France.
Presse Med, 39(2):e35-44. Epub 2009/12/08.
6904
6905
Stagg RB et al. (1997). DNA synthesis and cell proliferation in C6 glioma and primary glial cells exposed to a 836.55 MHz
modulated radiofrequency field. Bioelectromagnetics, 18(3):230-236. Epub 1997/01/01.
6906
6907
Stang A et al. (2001). The possible role of radiofrequency radiation in the development of uveal melanoma. Epidemiology,
12(1):7-12. Epub 2001/01/04.
6908
6909
Stang A et al. (2009). Mobile phone use and risk of uveal melanoma: results of the risk factors for uveal melanoma casecontrol study. J Natl Cancer Inst, 101(2):120-123. Epub 2009/01/15.
6910
6911
Stangerup SE et al. (2010). True incidence of vestibular schwannoma? Neurosurgery, 67(5):1335-1340; discussion 1340.
Epub 2010/09/28.
6912
6913
Stankiewicz W et al. (2006). Immunotropic influence of 900 MHz microwave GSM signal on human blood immune cells
activated in vitro. Electromagn Biol Med, 25(1):45-51.
6914
6915
Stewart A et al. (2011). Mobile telecommunications and health : Report of an investigation into an alleged cancer cluster in
Sandwell, West Midlands. Perspectives in Public Health.
6916
6917
Stronati L et al. (2006). 935 MHz cellular phone radiation. An in vitro study of genotoxicity in human lymphocytes. Int J
Radiat Biol, 82(5):339-346.
6918
6919
Sun W et al. (2012). A 1.8-GHz radiofrequency radiation induces EGF receptor clustering and phosphorylation in cultured
human amniotic (FL) cells. Int J Radiat Biol, 88(3):239-244.
6920
6921
Swerdlow AJ et al. (2011). Mobile phones, brain tumors, and the interphone study: where are we now? Environ Health
Perspect, 119(11):1534-1538. Epub 2011/12/16.
6922
6923
Syldona M (2007). Reducing the in-vitro electromagnetic field effect of cellular phones on human DNA and the intensity of
their emitted radiation. Acupunct Electrother Res, 32(1-2):1-14.
6924
6925
Szabo I et al. (2001). Reactions of keratinocytes to in vitro millimeter wave exposure. Bioelectromagnetics, 22(5):358-364.
Epub 2001/06/26.
6926
6927
Szmigielski S et al. (1982). Accelerated development of spontaneous and benzopyrene-induced skin cancer in mice exposed
to 2450 MHz microwave radiation. Bioelectromagnetics, 3:179-191.
6928
6929
Szmigielski S (1996). Cancer morbidity in subjects occupationally exposed to high frequency (radiofrequency and
microwave) electromagnetic radiation. Sci Total Environ, 180(1):9-17. Epub 1996/02/02.
6930
6931
Szmigielski S, Sobiczewska E, Kubacki R (2001). Carcinogenic potency of microwave radiation: overview of the problem
and results of epidemiological studies on Polish military personnel. European J Oncol, 6(2):6. Epub 2001.
6932
6933
Szmigielski S, Sobiczewska E, Kubacki R (2006). Cancer risks in workers exposed to radiofrequency and microwave
radiation: an epidemiologic study. Proceedings of the Latvian Academy of Sciences, 60(642):3.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
235
6934
6935
Szudzinski A et al. (1982). Acceleration of the development of benzopyrene-induced skin cancer in mice by microwave
radiation. Arch Dermatol Res, 274(3-4):303-312.
6936
6937
Takashima Y et al. (2006). Effects of continuous and intermittent exposure to RF fields with a wide range of SARs on cell
growth, survival, and cell cycle distribution. Bioelectromagnetics, 27(5):392-400. Epub 2006/04/15.
6938
6939
Takebayashi T et al. (2006). Mobile phone use and acoustic neuroma risk in Japan. Occup Environ Med, 63(12):802-807.
Epub 2006/08/17.
6940
6941
Takebayashi T et al. (2008). Mobile phone use, exposure to radiofrequency electromagnetic field, and brain tumour: a casecontrol study. Br J Cancer, 98(3):652-659. Epub 2008/02/08.
6942
6943
Terro F et al. (2012). GSM-900MHz at low dose temperature-dependently downregulates alpha-synuclein in cultured
cerebral cells independently of chaperone-mediated-autophagy. Toxicology, 292(2-3):136-144.
6944
6945
Thomsen J, Tos M (1990). Acoustic neuroma: clinical aspects, audiovestibular assessment, diagnostic delay, and growth rate.
Am J Otol, 11(1):12-19.
6946
6947
Thorlin T et al. (2006). Exposure of cultured astroglial and microglial brain cells to 900 MHz microwave radiation. Radiat
Res, 166(2):409-421.
6948
6949
Tian F et al. (2002). Exposure to 2.45 GHz electromagnetic fields induces hsp70 at a high SAR of more than 20 W/kg but not
at 5W/kg in human glioma MO54 cells. Int J Radiat Biol, 78(5):433-440.
6950
6951
Tice RR et al. (2002). Genotoxicity of radiofrequency signals. I. Investigation of DNA damage and micronuclei induction in
cultured human blood cells. Bioelectromagnetics, 23(2):113-126.
6952
6953
Tillmann T et al. (2007). Carcinogenicity study of GSM and DCS wireless communication signals in B6C3F1 mice.
Bioelectromagnetics, 28(3):173-187.
6954
6955
Tillmann T et al. (2010). Indication of cocarcinogenic potential of chronic UMTS-modulated radiofrequency exposure in an
ethylnitrosourea mouse model. Int J Radiat Biol, 86(7):529-541.
6956
6957
Tiwari R et al. (2008). Combinative exposure effect of radio frequency signals from CDMA mobile phones and aphidicolin
on DNA integrity. Electromagn Biol Med, 27(4):418-425.
6958
6959
Toler JC et al. (1997). Long-term, low-level exposure of mice prone to mammary tumors to 435 MHz radiofrequency
radiation. Radiat Res, 148(3):227-234.
6960
6961
Trillo MA et al. (2011). Cytostatic response of NB69 cells to weak pulse-modulated 2.2 GHz radar-like signals.
Bioelectromagnetics, 32(5):340-350. Epub 2011/02/01.
6962
6963
Trivino Pardo JC et al. (2012). Microwave electromagnetic field regulates gene expression in T-lymphoblastoid leukemia
CCRF-CEM cell line exposed to 900 MHz. Electromagn Biol Med, 31(1):1-18.
6964
6965
Trošić I, Pavičić I (2009). Disturbance of cell proliferation in response to mobile phone frequency radiation. Arh Hig Rada
Toksikol, 60(1):109-115. Epub 2009/03/31.
6966
6967
Trosko JE, Upham BL (2005). The emperor wears no clothes in the field of carcinogen risk assessment: ignored concepts in
cancer risk assessment. Mutagenesis, 20(2):81-92.
6968
6969
Tuffs A (2008). University calls for mobile phone research to be withdrawn after technician admits faking data. BMJ,
336(7656):1270.
6970
6971
Tuschl H, Novak W, Molla-Djafari H (2006). In vitro effects of GSM modulated radiofrequency fields on human immune
cells. Bioelectromagnetics, 27(3):188-196.
6972
6973
Tynes T et al. (1996). Incidence of breast cancer in Norwegian female radio and telegraph operators. Cancer Causes Control,
7(2):197-204. Epub 1996/03/01.
6974
6975
Utteridge TD et al. (2002). Long-term exposure of E-µ-Pim1 transgenic mice to 898.4 MHz microwaves does not increase
lymphoma incidence. Radiat Res, 158(3):357-364.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
236
6976
6977
Valbonesi P et al. (2008). Evaluation of HSP70 expression and DNA damage in cells of a human trophoblast cell line
exposed to 1.8 GHz amplitude-modulated radiofrequency fields. Radiat Res, 169(3):270-279.
6978
6979
Vanderwaal RP et al. (2006). HSP27 phosphorylation increases after 45 degrees C or 41 degrees C heat shocks but not after
non-thermal TDMA or GSM exposures. Int J Hyperthermia, 22(6):507-519.
6980
6981
Velizarov S, Raskmark P, Kwee S (1999). The effects of radiofrequency fields on cell proliferation are non-thermal.
Bioelectrochem Bioenerg, 48(1):177-180. Epub 1999/05/06.
6982
6983
Verma M, Dutta SK (1993). Microwave induced alteration in the neuron specific enolase gene expression. Cancer Biochem
Biophys, 13(4):239-244.
6984
6985
Vijayalaxmi et al. (1997). Proliferation and cytogenetic studies in human blood lymphocytes exposed in vitro to 2450 MHz
radiofrequency radiation. Int J Radiat Biol, 72(6):751-757.
6986
6987
Vijayalaxmi et al. (2000). Primary DNA damage in human blood lymphocytes exposed in vitro to 2450 MHz radiofrequency
radiation. Radiat Res, 153(4):479-486.
6988
6989
Vijayalaxmi et al. (2001a). Chromosome damage and micronucleus formation in human blood lymphocytes exposed in vitro
to radiofrequency radiation at a cellular telephone frequency (847.74 MHz, CDMA). Radiat Res, 156(4):430-432.
6990
6991
Vijayalaxmi et al. (2001b). Cytogenetic studies in human blood lymphocytes exposed in vitro to radiofrequency radiation at a
cellular telephone frequency (835.62 MHz, FDMA). Radiat Res, 155(1 Pt 1):113-121.
6992
6993
Vijayalaxmi (2006). Cytogenetic studies in human blood lymphocytes exposed in vitro to 2.45 GHz or 8.2 GHz
radiofrequency radiation. Radiat Res, 166(3):532-538.
6994
6995
Vijayalaxmi, McNamee JP, Scarfi MR (2006). Comments on: "DNA strand breaks" by Diem et al. [Mutat. Res. 583 (2005)
178-183] and Ivancsits et al. [Mutat. Res. 583 (2005) 184-188]. Mutat Res, 603(1):104-106.
6996
6997
Vrijheid M et al. (2006). Validation of short term recall of mobile phone use for the Interphone study. Occup Environ Med,
63(4):237-243. Epub 2006/03/25.
6998
6999
Vrijheid M et al. (2009a). Recall bias in the assessment of exposure to mobile phones. J Expo Sci Environ Epidemiol,
19(4):369-381. Epub 2008/05/22.
7000
7001
Vrijheid M et al. (2009b). Quantifying the impact of selection bias caused by nonparticipation in a case-control study of
mobile phone use. Ann Epidemiol, 19(1):33-41.
7002
7003
Wang J et al. (2006). Effects of a 2450 MHz high-frequency electromagnetic field with a wide range of SARs on the
induction of heat-shock proteins in A172 cells. Bioelectromagnetics, 27(6):479-486.
7004
Ward E et al. (2014). Childhood and adolescent cancer statistics, 2014. CA Cancer J Clin, 64(2):83-103. Epub 2014/02/04.
7005
7006
Warren HG et al. (2003). Cellular telephone use and risk of intratemporal facial nerve tumor. Laryngoscope, 113(4):663-667.
Epub 2003/04/03.
7007
7008
Whitehead TD et al. (2005). Expression of the proto-oncogene Fos after exposure to radiofrequency radiation relevant to
wireless communications. Radiat Res, 164(4 Pt 1):420-430.
7009
7010
Whitehead TD et al. (2006). Gene expression does not change significantly in C3H 10T(1/2) cells after exposure to 847.74
CDMA or 835.62 FDMA radiofrequency radiation. Radiat Res, 165(6):626-635.
7011
7012
WHO - World Health Organization. Electromagnetic fields (300 Hz to 300 GHz). Geneva, World Health Organization, 1993a
(Environmental Health Criteria 137).
7013
7014
WHO - World Health Organization. Electromagnetic fields (300 Hz to 300 GHz). Geneva, World Health Organization,
1993b.
7015
7016
Wile AG, Nahabedian MY, Mason GR (1983). Enhanced tumor growth in experimental whole body hyperthermia. J Surg
Oncol, 24(2):119-123.
7017
7018
Wolf R, Wolf D (2004). Increased incidence of cancer near a cell-phone transmitter station. International Journal of Cancer
Prevention, 1(2):6.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
237
7019
7020
Wolke S et al. (1996). Calcium homeostasis of isolated heart muscle cells exposed to pulsed high-frequency electromagnetic
fields. Bioelectromagnetics, 17(2):144-153.
7021
7022
Wu H et al. (2012). Cytokines produced by microwave-radiated Sertoli cells interfere with spermatogenesis in rat testis.
Andrologia, 44 Suppl 1:590-599.
7023
7024
Wu RY et al. (1994). Effects of 2.45-GHz microwave radiation and phorbol ester 12-O-tetradecanoylphorbol-13-acetate on
dimethylhydrazine-induced colon cancer in mice. Bioelectromagnetics, 15(6):531-538.
7025
7026
Xu S et al. (2010). Exposure to 1800 MHz radiofrequency radiation induces oxidative damage to mitochondrial DNA in
primary cultured neurons. Brain Res, 1311:189-196.
7027
7028
Xu S et al. (2013). Cell type-dependent induction of DNA damage by 1800 MHz radiofrequency electromagnetic fields does
not result in significant cellular dysfunctions. PLoS One, 8(1):e54906.
7029
7030
Yamazaki E, Matsubara S, Yamada I (1993). Effect of Gd-DTPA and/or magnetic field and radiofrequency exposure on
sister chromatid exchange in human peripheral lymphocytes. Acta Radiol, 34(6):607-611.
7031
7032
Yang L et al. (2012). Cellular neoplastic transformation induced by 916 MHz microwave radiation. Cell Mol Neurobiol,
32(6):1039-1046. Epub 2012/03/08.
7033
7034
Yang X et al. (2010). The role of the JAK2-STAT3 pathway in pro-inflammatory responses of EMF-stimulated N9
microglial cells. J Neuroinflammation, 7:54.
7035
7036
Yao K et al. (2004). Low power microwave radiation inhibits the proliferation of rabbit lens epithelial cells by upregulating
P27Kip1 expression. Mol Vis, 10:138-143.
7037
7038
Yao K et al. (2008a). Electromagnetic noise inhibits radiofrequency radiation-induced DNA damage and reactive oxygen
species increase in human lens epithelial cells. Mol Vis, 14:964-969.
7039
7040
Yao K et al. (2008b). Effect of superposed electromagnetic noise on DNA damage of lens epithelial cells induced by
microwave radiation. Invest Ophthalmol Vis Sci, 49(5):2009-2015.
7041
7042
Yoon SY et al. (2011). Induction of hair growth by insulin-like growth factor-1 in 1,763 MHz radiofrequency-irradiated hair
follicle cells. PLoS One, 6(12):e28474.
7043
7044
Yu D et al. (2006). Effects of 900 MHz GSM wireless communication signals on DMBA-induced mammary tumors in rats.
Radiat Res, 165(2):174-180.
7045
7046
Zeng Q et al. (2006). Effects of global system for mobile communications 1800 MHz radiofrequency electromagnetic fields
on gene and protein expression in MCF-7 cells. Proteomics, 6(17):4732-4738.
7047
7048
Zeni O et al. (2003). Lack of genotoxic effects (micronucleus induction) in human lymphocytes exposed in vitro to 900 MHz
electromagnetic fields. Radiat Res, 160(2):152-158.
7049
7050
Zeni O et al. (2005). Evaluation of genotoxic effects in human peripheral blood leukocytes following an acute in vitro
exposure to 900 MHz radiofrequency fields. Bioelectromagnetics, 26(4):258-265.
7051
7052
7053
Zeni O et al. (2007a). Formation of reactive oxygen species in L929 cells after exposure to 900 MHz RF radiation with and
without co-exposure to 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone. Radiat Res, 167(3):306-311. Epub
2007/02/24.
7054
7055
Zeni O et al. (2007b). Cytogenetic observations in human peripheral blood leukocytes following in vitro exposure to THz
radiation: a pilot study. Health Phys, 92(4):349-357.
7056
7057
Zeni O et al. (2008). Evaluation of genotoxic effects in human leukocytes after in vitro exposure to 1950 MHz UMTS
radiofrequency field. Bioelectromagnetics, 29(3):177-184.
7058
7059
7060
Zeni O et al. (2012a). Induction of an adaptive response in human blood lymphocytes exposed to radiofrequency fields:
influence of the universal mobile telecommunication system (UMTS) signal and the specific absorption rate. Mutat Res,
747(1):29-35.
7061
7062
Zeni O et al. (2012b). Radiofrequency radiation at 1950 MHz (UMTS) does not affect key cellular endpoints in neuron-like
PC12 cells. Bioelectromagnetics, 33(6):497-507.
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
238
7063
7064
Zhadobov M et al. (2007). Low-power millimeter wave radiations do not alter stress-sensitive gene expression of chaperone
proteins. Bioelectromagnetics, 28(3):188-196.
7065
7066
Zhang MB et al. (2002). Study of low-intensity 2450-MHz microwave exposure enhancing the genotoxic effects of
mitomycin C using micronucleus test and comet assay in vitro. Biomed Environ Sci, 15(4):283-290.
7067
7068
Zhao R et al. (2007). Studying gene expression profile of rat neuron exposed to 1800MHz radiofrequency electromagnetic
fields with cDNA microassay. Toxicology, 235(3):167-175.
7069
7070
Zhao TY, Zou SP, Knapp PE (2007). Exposure to cell phone radiation up-regulates apoptosis genes in primary cultures of
neurons and astrocytes. Neurosci Lett, 412(1):34-38.
7071
7072
Zhijian C et al. (2009). Influence of 1.8-GHz (GSM) radiofrequency radiation (RFR) on DNA damage and repair induced by
X-rays in human leukocytes in vitro. Mutat Res, 677(1-2):100-104.
7073
7074
Zhijian C et al. (2010). Impact of 1.8-GHz radiofrequency radiation (RFR) on DNA damage and repair induced by
doxorubicin in human B-cell lymphoblastoid cells. Mutat Res, 695(1-2):16-21.
7075
7076
Zhijian C et al. (2013). Studying the protein expression in human B lymphoblastoid cells exposed to 1.8-GHz (GSM)
radiofrequency radiation (RFR) with protein microarray. Biochem Biophys Res Commun, 433(1):36-39.
7077
7078
Zhou XR et al. (2008). The study of retinal ganglion cell apoptosis induced by different intensities of microwave irradiation.
Ophthalmologica, 222(1):6-10.
7079
7080
Zimmerman JW et al. (2012). Cancer cell proliferation is inhibited by specific modulation frequencies. Br J Cancer,
106(2):307-313.
7081
7082
Zmyślony M et al. (2004). Acute exposure to 930 MHz CW electromagnetic radiation in vitro affects reactive oxygen species
level in rat lymphocytes treated by iron ions. Bioelectromagnetics, 25(5):324-328. Epub 2004/06/16.
7083
7084
Zook BC, Simmens SJ (2001). The effects of 860 MHz radiofrequency radiation on the induction or promotion of brain
tumors and other neoplasms in rats. Radiat Res, 155(4):572-583.
7085
7086
Zook BC, Simmens SJ (2006). The effects of pulsed 860 MHz radiofrequency radiation on the promotion of neurogenic
tumors in rats. Radiat Res, 165(5):608-615.
7087
7088
Zotti-Martelli L et al. (2000). Induction of micronuclei in human lymphocytes exposed in vitro to microwave radiation. Mutat
Res, 472(1-2):51-58.
7089
7090
Zotti-Martelli L et al. (2005). Individual responsiveness to induction of micronuclei in human lymphocytes after exposure in
vitro to 1800-MHz microwave radiation. Mutat Res, 582(1-2):42-52.
7091
7092
THIS IS A DRAFT DOCUMENT FOR PUBLIC CONSULTATION. PLEASE DO NOT QUOTE OR CITE.
239