Article Text

other Versions

PDF
Occupational exposure to electromagnetic fields and sex-differential risk of uveal melanoma
  1. Thomas Behrens1,
  2. Elsebeth Lynge2,
  3. Ian Cree3,
  4. Svend Sabroe4,
  5. Jean-Michel Lutz5,
  6. Noemia Afonso6,
  7. Mikael Eriksson7,
  8. Pascal Guénel8,9,
  9. Franco Merletti10,
  10. Maria Morales-Suarez-Varela11,12,13,
  11. Aivars Stengrevics14,
  12. Joëlle Févotte15,
  13. Agustin Llopis-González11,13,
  14. Giuseppe Gorini16,
  15. Galina Sharkova17,
  16. Lennart Hardell7,
  17. Wolfgang Ahrens1
  1. 1Bremen Institute for Prevention Research and Social Medicine, Bremen, Germany
  2. 2Institute of Public Health, University of Copenhagen, Copenhagen, Denmark
  3. 3Institute of Ophthalmology, University College London, London, UK
  4. 4Department of Epidemiology, University of Aarhus, Aarhus, Denmark
  5. 5Department of Chronic Disease Epidemiology, National Institute for Cancer Epidemiology and Registration (NICER), ISPM Zürich, University of Zurich, Zurich, Switzerland
  6. 6Serviço de Oncologia Médica, Instituto Portugues de Oncologia - Porto, Porto, Portugal
  7. 7Department of Oncology, Lund University Hospital, Lund, Sweden
  8. 8INSERM Unité 754, Villejuif Cedex, France
  9. 9University Paris-Sud 11, Orsay, France
  10. 10Unit of Cancer Epidemiology, University of Turin, CERMS and CPO, Piemonte, Italy
  11. 11Unit of Public Health and Environmental Care, Department of Preventive Medicine, University of Valencia, Valencia, Spain
  12. 12Research Foundation, University Hospital Dr. Peset, Valencia, Spain
  13. 13Research Group CIBER CB06/02/0045, CIBER Actions - Epidemiology and Public Health, Spain
  14. 14Latvia Cancer Registry, Riga, Latvia
  15. 15Institut de Veille Sanitaire (InVS), Département Santé-Travail (DST), Saint-Maurice, France
  16. 16Epidemiology and Public Health, Environmental and Occupational Epidemiology Unit - ISPO Cancer Prevention and Research Institute, Florence, Italy
  17. 17Department of Oncology, P. Stradin's Clinical University Hospital, Riga, Latvia
  1. Correspondence to Dr Thomas Behrens, Linzer Str. 10, Bremen 28359, Germany; behrens{at}bips.uni-bremen.de

Abstract

Objectives The association between occupational exposure to electromagnetic fields (EMF) and the risk of uveal melanoma was investigated in a case–control study in nine European countries.

Methods Incident cases of uveal melanoma and population as well as hospital controls were included and frequency matched by country, 5-year birth cohort and sex. Subjects were asked whether they had worked close to high-voltage electrical transmission installations, computer screens and various electrical machines, or in complex electrical environments. Measurements of two Scandinavian job–exposure matrices were applied to estimate lifelong cumulative EMF exposure. Unconditional logistic regression analyses, stratified by sex and eye colour were calculated, adjusting for several potential confounders.

Results 293 patients with uveal melanoma and 3198 control subjects were interviewed. Women exposed to electrical transmission installations showed elevated risks (OR 5.81, 95% CI 1.72 to 19.66). Positive associations with exposure to control rooms were seen among men and women, but most risk increases were restricted to subjects with dark iris colour. Application of published EMF measurements revealed stronger risk increases among women compared to men. Again, elevated risks were restricted to subjects with dark eye colour.

Conclusion Although based on a low prevalence of exposure to potential occupational sources of EMF, our data indicate that exposed dark-eyed women may be at particular risk for uveal melanoma.

Statistics from Altmetric.com

What this paper adds

  • Potential occupational and environmental causes of uveal melanoma are incompletely understood.

  • Some early analyses of job tasks point towards an association between uveal melanoma and occupations with high exposure to extremely low frequency electromagnetic fields (EMF).

  • The findings of this study suggest a particular risk for female subjects and subjects with dark eye colour exposed to occupational EMF.

Introduction

Although uveal melanoma is a rare tumour in absolute terms, it is the most common cancer of the eye in adults. Age-standardised incidence rates in Europe range from 7.5 per million in Denmark to 2.6 per million in Spain among men and from 6.3 per million in Denmark to 1.7 per million in Spain among women. However, temporal trends do not show a consistent pattern across countries.1

The aetiology of uveal melanoma is largely unknown. Geographical patterns reveal lower incidences in countries at lower latitudes as well as among US African-Americans, which points towards a protective influence of darker hair and skin pigmentation and dark iris colour.2 In this context chronic exposure to UV light and sunlight has been associated with the risk of uveal melanoma.3

Furthermore, weak and inconsistent associations have been observed for certain occupational factors such as welding,4 cooking,4 5 chemical exposures,6 agriculture and pesticide exposure,6 7 and working as a sailor or fisherman.4 6 Some studies, including those mentioned above, reported positive associations between extremely low frequency electromagnetic fields (EMF) and radio frequency (RF) exposure or jobs with potential EMF exposure and the risk of uveal melanoma. However, results are inconsistent.4 6 8–11

In contrast to higher frequency radiation, the energy transfer resulting from EMF (0.1 Hz to 30 kHz) and RF (300 kHz to 300 GHz) is not sufficient to break covalent bonds.12 13 However, Stevens proposed a hormonally mediated EMF mechanism which may indirectly promote oestrogen secretion from the ovaries and therefore could affect susceptibility to hormone-dependent tumours which could lead to differential risks between men and women.14 As was pointed out in laboratory experiments, melanoma cells have oestrogen receptors.15 16

We therefore investigated the sex-specific association between occupational exposure to EMF/RF and risk of uveal melanoma, using the international case–control dataset of the Study of Occupational Risk Factors for Rare Cancers of Unknown Aetiology.17

Methods

Detailed study methods have been described elsewhere.17 Briefly, the study population comprised the national populations of Denmark and Latvia, the population of certain administrative regions in France, Germany, Italy and Sweden, hospital recruitment areas in Spain and Portugal, and an eye clinic in the UK.

Cases

Incident cases of uveal melanoma were included if they were diagnosed between 1 January 1994 and 31 December 1997. The majority of cases (93.5%) were diagnosed in 1995 and 1996, a period when the aim was to include all incident cases. Recruitment was most successful in countries with comprehensive cancer registries, such as Sweden, Portugal and Denmark.17 All other countries relied on regular contacts with clinical or pathology departments, so potential cases could have been missed. However, case recruitment in these countries was considered nearly complete in 1995/96 (with the exception of the UK which provided only a few additional cases).

Eligible cases were defined by cancer site according to the International Classification of Diseases (ICD) 9th and 10th revisions. Melanoma of the eyeball, choroid and eye not otherwise specified, as well as of overlapping locations of the eye were considered eligible (ICD-9 190.0, 190.6, 190.8 and 190.9 and ICD-10 C69.3, C69.4, C69.6, C69.8 and C69.9). Tumours of the conjunctivae were excluded. Diagnoses were verified based on pathological specimens for enucleated patients or, in all other cases, on ophthalmology reports. Cases with a definite or possible diagnosis were considered eligible. In addition, cases had to be between 35 and 69 years of age at diagnosis.

Control subjects

Control subjects were frequency matched by region, sex and 5-year age group.

Population controls were randomly selected from population or municipality registries in Denmark, Italy, France, Sweden and Germany. Hospital based colon cancer controls were selected from hospitals in Portugal and Denmark, and from one Spanish centre. The second Spanish centre recruited control subjects from among patients attending the emergency ward. Latvia used population based colon cancer controls. In the UK, one control per case was selected from the patient list of the case's general practitioner. All eligible controls were retained in the analysis.

Data collection

Face-to-face or telephone interviews were conducted with every subject who agreed to participate. Cases and cancer controls were interviewed as soon as possible after the date of diagnosis. Population controls were interviewed concurrently with cases. If a study subject had died or was too ill to be interviewed, a surrogate interview was solicited.

Topics covered by the study questionnaire included demographic and personal characteristics such as skin, hair and eye colour, tobacco and alcohol consumption, and the subject's detailed occupational history including each job that lasted for at least 6 months. Specific sets of questions addressed occupational exposure to EMF.

Exposure quantification

In each national centre, jobs were coded according to the International Classification of Occupations (ISCO 1968),18 and industries were coded according to the National Industrial Classification of all Economic Activities (NACE) in the European Community.19

Exposure to power lines and generators, electrical machines, and complex electrical environments such as control or computer rooms, were assessed via questionnaire.

Study subjects were asked whether they had worked close to high-voltage overhead lines, underground power lines, transformers, substations, electrical machine tools, electric ovens or furnaces, trains, lifting trucks or other electrical machines. In addition, the questionnaire solicited information on work exposure in complex control rooms, in computer rooms, near telephone switchboards, and in rooms with complex electrical medical device(s), or whether the subject had operated a computer screen for several hours each day. Working ‘close’ to a machine was further defined by self-reported distance to the specific source in metres. If a study subject reported a maximum voltage of zero for high-voltage lines or transformers, the electrical transmission installation was considered ‘switched off’, and the subject was recoded as unexposed to that particular source. Data were also recoded when information given in free text fields suggested exposure to a different electrical machine or source.

Duration of exposure was calculated as the difference between the start and end year of exposure and categorised into 10-year intervals (1–10 years, >10 years) with no exposure or exposure less than 1 year as referent. We evaluated EMF sources separately and as a pooled category of the main exposure groups.

Exposure to electric ovens and furnaces was evaluated using questions from the EMF section of the questionnaire as well as a separate question soliciting information about working close to a furnace. Exposure to welding equipment was derived either from the job history (employment as a welder, ISCO-68 codes 8-72.10 to 8-72.90), from one separate question regarding arc welding, or from self-reported use of welding equipment reported in the free text field for ‘other’ electrical work machines.

We lagged exposure in order to account for induction periods of 3 and 5 years, respectively, that is exposure was only considered if it took place more than 3 or 5 years before the reference date (the date of diagnosis for cases and cancer controls and the interview date for all other controls).

Classification of exposure to radar unit was based on assumptions in a previous article from a related study using the same questionnaire in which the effect of occupational exposure to RF/EMF on the risk of testicular cancer was analysed.20 This article obtained expert opinion and measurements from unpublished expert reports to quantify potential occupational radiation exposure based on job titles, and classified occupational exposure to radiation as ‘most likely’, ‘probable’ and ‘unlikely’. For this analysis, jobs with ‘most likely’ and ‘probable’ exposure were considered relevant.

In addition, we used job–exposure matrices from two Scandinavian reports21 22 to assign typical field strengths (geometric mean (GM) in μ-Teslas) to occupations reported by subjects. We applied the most specific code for both sexes. If different field strengths were reported for males and females, most likely due to sex-specific job tasks within the same occupation, sex-specific field strengths were applied. In this case, measurements reported by Floderus et al21 were assigned to males and values reported by Forssén and co-workers22 were assigned to females. Two additional Scandinavian studies reporting EMF measurements were used to impute exposure levels for workers on ships and workers in the garment industry.23 24

To estimate lifelong time-integrated occupational exposure, we calculated lifelong cumulated μT-years by applying the following equation (assuming 220 workdays of 8 h duration per year):

μTyears=GM field strength[μT]*years of exposure*220/365*8/24

μT-years were divided into 5% percentiles and cut-off points for percentiles were based on the distribution among controls.

In addition, we defined average EMF exposure as the number of cumulative μT-years divided by the total years given in the job biography.

Statistical analysis

Unconditional logistic regression analyses were carried out, adjusting for country and 5-year age groups. For adjustment by country, we merged the small UK dataset with the German dataset, as done in a previous publication from this study.4 Sex, eye colour (blue or grey, and green or hazel vs brown or black as reference), skin tone (white vs non-white), frequency of lifetime ocular damage due to intense UV exposure (>5 times, 1 to ≤5 times burned during lifetime vs no reported eye burns) were also included. Since inclusion of alcohol consumption (beer, wine, spirits, sherry/port/aperitifs) together with education (with less than 9 years at school as reference) changed results by more than 10% in some analyses (particularly among women), alcohol consumption and education were retained in the model for analyses estimating the association between uveal melanoma and occupational EMF sources, but not when calculating quantitative associations based on μT-years. Since smoking was not associated with uveal melanoma, as previously reported from this study,4 nor confounded on the associations with EMF exposure, it was not included in the final model.

We calculated ORs stratified by sex and eye colour for the main EMF sources and for the quantitative analyses, comparing subjects exposed to a particular source of EMF with unexposed subjects. Quantitative associations for cumulative EMF exposure in μT-years were calculated comparing subjects above the respective percentile with all others for unlagged μT-years and μT-years lagged by 5 years.

Effect modification by sex or eye colour was assessed at the 95% level by a product term in the regression model. Age (≤56 vs >56 years) was also considered as a potential effect modifier, but no interaction was observed.

Sensitivity analyses excluding cancer control subjects and surrogate interviews were also carried out. We also restricted our analyses to cases with a definite diagnosis of eye melanoma, and cases diagnosed in 1995 and 1996 as well as control subjects interviewed during the same 2 years. All analyses were performed using SAS statistical package, version 8.0.

The study was approved by the local ethics committees. Study subjects or, in the case of deceased subjects, their relatives gave written informed consent to participation in the study.

Results

Of 359 identified eye melanoma cases, 323 were considered eligible and 293 were interviewed. Of 3372 population controls, 2093 were interviewed, and of 1284 cancer control patients, 1105 were interviewed. Only 126 surrogate interviews were carried out: 13 among cases, 27 among population controls and 86 among cancer controls.

Eye tumours involved the choroid (68.3%), the iris and ciliary zone (20.1%) as well as unknown or overlapping areas of the eye (11.6%). Other demographic and descriptive information is given in table 1.

Table 1

Characteristics of the study population by case–control status

Occupational exposure to high-voltage power transmission installations (power lines and transformers/substations) revealed a more than fivefold increased risk among women (OR 5.81; 95% CI 1.72 to 19.66) but not among men (OR 1.27; 95% CI 0.69 to 2.33) in whom only exposure to transformers was associated with an increase in melanoma risk (OR 2.14; 95% CI 1.06 to 4.32). However, exposure prevalence was low among women with only six exposed cases and seven exposed controls (table 2).

Table 2

Risk for exposure to different sources of EMF (OR, 95% CI), adjusted for (OR*) and stratified by sex (OR1, OR2)

Elevated risks were seen with exposure to complex electrical environments: men showed a threefold increased risk after exposure to control rooms, and a twofold increased risk for the pooled category of control and computer rooms as well as for exposure to any complex electrical environment (control/computer room, rooms with complex medical equipment, or telephone switchboards). For women, exposure prevalence was again too small to produce significant results. Mutual adjustment for other sources of EMF did not markedly change the overall results (not shown).

Applying the exposure definition of Baumgardt-Elms and co-workers,20 we did not identify any women who were potentially exposed to radar unit. Among men, no evidence for increased risk was observed for this exposure category (table 2).

Although none of the differences in OR reached statistical significance at the 95% level, analyses of exposure by eye colour suggested higher risks for subjects with light eye colour who were exposed to electrical transmission installations, while risk after exposure to a room with complex electronic devices tended to be higher for subjects with dark eye colour (table 3).

Table 3

Risk from exposure to different sources of EMF (OR, 95% CI), stratified by eye colour

Analysing the effect of duration of exposure, we did not observe any clear duration–effect relationship. When 3- and 5-year lags with respect to the reference date were applied, overall results remained unchanged. Analysis of duration weighted by distance to the EMF source (in tertiles of distance-years) did not produce any clear dose–response relationships (all results not shown).

We observed effect modification of risk with respect to eye colour in our data. Although the interaction terms eye colour and respective EMF source were not statistically significant at the 95% level, elevated risks were mainly observed among subjects with dark eyes. In contrast, the strongly elevated risk for women after exposure to high-voltage power installations was only observed among women with light-coloured eyes since only one dark-eyed control woman was exposed to this source of EMF (table 4).

Table 4

ORs of uveal melanoma in different exposure groups, stratified by sex and eye colour

Estimation of cumulative EMF exposure ranged from 0.008 to 11.82 μT-years in women and from 0.01 to 13.16 μT-years in men. Only 110 subjects (69 female control subjects, six female cases and 32 male control subjects, three male cases) were classified as unexposed. This finding is not surprising given that the job–exposure matrices on which this quantification concept was based assigned an EMF value to most professions, including ‘unlikely’ ones such as teachers. Estimation of μT-years was not possible due to missing values in 21 female (two cases, 19 control subjects) and eight male subjects (one case, seven controls).

The strongest risk increase was observed for women with dark eyes whose risk increased with increasing lifelong accumulated EMF dose. Men with dark eyes also showed an increase of risk in the upper exposure percentiles, while the risk among male and female subjects with light-coloured eyes remained close to unity. In addition, elevated risks in the lower exposure categories were demonstrated for men and women with dark eyes (table 5). Upon categorisation into quintiles, women with dark eyes showed an OR of 5.97 (95% CI 1.25 to 28.44) when comparing the highest with the lowest quintile of cumulative EMF exposure, while for dark-eyed men an OR of 1.50 (95% CI 0.49 to 4.65) was observed. Analysis of log-EMF years and lagging of EMF exposure by 5 years yielded similar results (results not shown).

Table 5

ORs for cumulative EMF exposure (μT-years) according to job–exposure matrix EMF measurements: OR by percentiles and stratified by sex and eye colour

Analysis of average EMF exposure (ie, the number of cumulative μT-years divided by total job years) resulted in elevated risks, which were again most pronounced among women with dark eye colour, of 2.76 (95% CI 0.98 to 7.76) for exposure intensities above the median and 3.61 (95% CI 0.93 to 13.96) for exposure intensities above the 95th percentile. Among dark-eyed men, risk estimates of 2.43 (95% CI 0.82 to 7.23) for exposure intensities above the median and 1.27 (95% CI 0.65 to 2.50) for exposure intensities above the 95% percentile were observed (not shown).

Sensitivity analyses excluding cancer control subjects and surrogate interviews produced similar results compared to the analyses using the full dataset. For example, for females the risk for exposure to any electrical transmission installation increased to 14.2 (95% CI 2.35 to 85.53), while for males no association with this exposure was observed. For rooms with a complex electronic device, ORs were 1.37 for females and 1.78 for males. No substantial changes in results were observed when the analysis was restricted to cases diagnosed in 1995/1996 and control subjects interviewed during the same period (eg, women exposed to electrical transmission installations had an OR of 6.50 (95% CI 1.91 to 22.10) and men exposed to electrical transmission installations had an OR of 1.34 (95% CI 0.73 to 2.46). When we restricted the analyses to cases with a definite diagnosis of eye melanoma, similar results were observed compared to the analyses also including ‘possible’ cases. Females showed ORs of 4.71 (95% CI 1.28 to 17.32) for electrical transmission installations and 1.04 (95% CI 0.47 to 2.30) for rooms with a complex electronic device, while among men ORs of 1.50 (95% CI 0.81 to 2.76) and 1.95 (95% CI 1.17 to 3.23), respectively, were observed. In addition, the overall picture of a greater risk for female subjects with dark eye colour remained (results not shown).

Individual countries did not appear to have influenced results. When restricting analyses to face-to-face interviews, risk estimates weakened and the power markedly decreased. However, the overall impression of risk increases for exposure to electrical transmission installations among females and for exposures to rooms with a complex electronic device among male subjects remained.

Discussion

This article examined the impact of occupational EMF exposure on the risk of uveal melanoma. Positive associations between uveal melanoma and some sources of electromagnetic fields were observed. Particularly among women, we found a fivefold increased risk with exposure to high-voltage transmission installations (power lines, transformers and substations). Eye melanoma risk for exposure to more complex electrical environments, particularly control rooms, tended to be elevated in men, while electrical work machines and computer screens did not show any overall association with disease risk. However, occupational exposure to computer screens emerged as a significantly elevated risk factor among male and female subjects with dark eye colour.

So far, only a few studies have reported positive associations between EMF and the risk of uveal melanoma and results are inconsistent.4 6 8–11 The majority of these studies were based on job titles of ‘electrical trades’, or included only crude estimates of potential EMF exposure,6 11 which implies potential for misclassification bias.9 In contrast, in this study information on working in close proximity to sources of EMF was directly solicited during a personal interview and sources that were not directly operated by the participants were also included. According to the EMF questionnaire, 13 women exposed to high-voltage power transmission sources did not work in electrical trades but reported the presence of such a source close to their work place.

When quantifying exposure in cumulative μT-years, risks in our study were most pronounced among brown- or black-eyed female subjects. Among dark-eyed men, risk increases were indicated for the highest percentiles of exposure. More pronounced risk increases among dark-eyed subjects, particularly females, were also seen with average EMF exposure, indicating a potential hazard for continuously exposed subjects. It is still unknown which attribute of EMF (eg, time-weighted average, transient peaks of field strength, or field frequency, etc), may affect cancer risk.25 In our data we observed an initial risk increase for low cumulative exposure-years among subjects with dark iris pigmentation. Therefore, subjects exposed to high peaks of occupational EMF for a short period only may also be at greater risk of uveal melanoma as compared to continuously exposed subjects. However, we were unable to test whether subjects with dark eye colour in the 10 most highly exposed occupations, who had average exposure below the median, were at increased melanoma risk, as no case subjects were observed in these groups.

In a previous analysis of a subset of the European Rare Cancer case–control study (German data combined with an additional hospital based case–control study), an elevated risk for occupational exposure to RF transmitting devices, such as mobile phone use and radio set exposure, was observed.9 However, this finding was not confirmed in a recent update of this German case–control study.10 As we have discussed in a previous publication, classification of exposure to RF emitting sources may be extremely difficult if based on personal reports of appliance use.26 Furthermore, use of mobile phones or RF emitting devices was not solicited in the international protocol of the Rare Cancer study. Therefore, we had to restrict our analyses to jobs with possible radiatar exposure using the classification of Baumgardt-Elms and co-workers.20 Consistent with the German analysis,10 we did not detect a positive association between occupational exposure to sources of radiation and risk of uveal melanoma.

Welding was identified as a risk factor in the French section of the Rare Cancer study.4 27 We also observed risk increases for welders, but the estimates did not reach statistical significance. In the previous article,27 exposure to welding was interpreted within the context of exposure to artificial light. The posterior temporal segment of the eye is the most common site of melanoma origin,28 and it is doubtful whether melanocytes in the posterior uvea are reached by significant amounts of UV radiation.3 However, electric welding is also associated with high exposure to EMF.29

Underascertainment of incident cancers is a potentially serious problem in case–control studies. The majority of cases (93.5%) were recruited in 1995/96 when we aimed to include all eligible incident cases.17 In addition, cases were recruited in nine different European countries from a variety of sources which adds more uncertainty to the representativeness of our sample. However, the distribution of the tumour locations was comparable to the findings reported for incident eye melanoma in Europe.30 Also, the median age of cases in our study was close to the median age-at-diagnosis of 55 years reported in the literature, which suggests that our sample consisted of a typical group of eye melanoma cases.31

An important limitation of our analyses is that results were based on very low exposure frequencies, particularly among women. Misclassification of one or two case women as being exposed to high-voltage transmission installations could have easily biased the risk estimates towards a positive association. However, our data suggested that the risk of uveal melanoma increased with exposure to more complex electrical environments: control room exposure was associated with an increased risk, particularly among men, and exposure to electrical transmission installations led to an increased risk in women.

Another limitation of this study is that our EMF assessment concept was based on broad measurement categories from job–exposure matrices across all subjects dedicated to a particular job task. As has been previously discussed,32 there may be great variability in tasks and exposures within a single job category which likely results in different exposure levels between subjects involved in the same job task. Our quantification method may have therefore missed subjects highly exposed to EMF. Furthermore, high temporal variation in fields may exist during work shifts,25 but we did not take duration of shifts into account, as this information was not solicited in the questionnaire. This could also have caused misclassification of exposure which could have biased our results.

In addition, we did not consider residential EMF exposure in our analyses. Other studies that we have conducted in the past demonstrated that exposure to certain domestic electrical appliances can cause EMF exposure that may well exceed the average exposure levels of workers in electrical trades.33 However, in the USA no strong correlation between occupational and residential exposure to EMF has been noted.25 Since it is unlikely that occupationally exposed cases simultaneously experience higher exposure levels from electrical appliances or power lines close to their homes than controls, residential exposure does not confound the observed associations with occupational EMF.

A possible confounder we may have missed was total personal lifetime sun exposure. It has been queried whether sun exposure is an important factor in uveal melanoma carcinogenesis, as UV radiation is unlikely to reach the melanocytes of the posterior uvea.3 In addition, as noted in a previous analysis of the international Rare Cancer dataset, occupational sunlight exposure was not associated with an increased risk of uveal melanoma.4

Two biological mechanisms have been suggested that might explain an association between EMF and eye melanoma.

According to the first mechanism, EMF may decrease melatonin release from the pineal gland. Melatonin in turn may slow the turnover of normal cells at risk of malignant transformation or may even be directly oncostatic. In experimental studies melatonin was reported to inhibit the proliferation of uveal melanoma cells in vitro.15 Reduced levels of melatonin could therefore lead to a cancer-promoting effect in individuals exposed to EMF.

According to a second mechanism, EMF may act along the hypothalamic–pituitary axis and enhance prolactin production in the pituitary gland as well as oestrogen production in the ovaries, thus increasing susceptibility to sex hormone related cancers particularly among women. Indeed, increased prolactin and oestrogen levels have been observed in female experimental animals after exposure to extremely low-frequency EMF,14 and melanoma cells have melatonin as well as oestrogen receptors.15 16

Nevertheless, human data and epidemiological evidence for such an effect are sparse,14 and the current prevailing opinion does not support an effect of EMF on, for example, breast cancer with respect to the melatonin hypothesis.34

However, one epidemiological study is supportive of a possible link between hormonal exposures and uveal melanoma risk.35 This study demonstrated that women with proxies of increased hormone levels (ie, women who took hormone replacement therapy, women with intact ovaries, and women with liver cirrhosis) showed stronger risks for uveal melanoma compared to women without these exposures.

Our observation that subjects with brown or black eyes are more susceptible to the influence of EMF compared to subjects with blue, grey, green or hazel eyes is surprising since dark eye pigmentation is commonly considered to be protective against uveal melanoma.2 However, it has been pointed out that the process of melanin synthesis may become dysregulated during carcinogenesis, which could result in variant melanin polymers with pro- instead of antioxidant properties.36 Animal models suggest that melatonin may interfere with the de novo synthesis of melanin by inhibiting the induction of its primary catalyst tyrosinase.37 Suppression of melatonin release by EMF would thus indirectly promote melanin synthesis. EMF could therefore increase the synthetic cycle of melanin, which would particularly affect subjects with highly pigmented eyes. In these subjects this mechanism could therefore involve a greater likelihood of dysregulated synthesis pathways during early carcinogenesis and a cancer-promoting effect which would explain the higher risks of uveal melanoma seen in dark-eyed subjects. This effect would therefore be independent of a general protective effect of dark eye colour which was also observed in these data (ORs adjusted for sex, 5-year age group and country were 2.49 (95% CI 1.77 to 3.50) for blue/gray eye colour and 1.98 (95% CI 1.35 to 2.91) for green/hazel eyes), and would only be tumour promoting if a person was simultaneously exposed to EMF.

Conclusion

We found that occupational EMF exposure may possibly increase the risk of uveal melanoma, particularly among dark-eyed women. Although biological mechanisms have been proposed that may explain the differential effect between men and women as well as between dark- and light-coloured subjects, it cannot be ruled out that these results are due to chance. Despite the pooling of international case–control data, results were based on very low exposure frequencies. Therefore, our findings should be interpreted with great care. However, we were able to exclude several important sources of bias. Restricting analyses to definite cases, non-cancer controls or interviews with the index subject yielded similar results. A clearer picture might be obtained when pooling existent study data in a meta-analysis of all case–control studies conducted so far.

Acknowledgments

We gratefully acknowledge collaboration from patients, control subjects, participating hospitals and data providers.

Appendix The European Study Group on Occupational Causes of Rare Cancers

Denmark: Herman Autrup, Henrik Kolstad, Linda Kaerlev, Elsebeth Lynge, Jorn Olsen, Lisbeth Norum Pedersen, Svend Sabroe; reference pathologists: Preben Johansen, Stein Poulsen, Peter Stubbe Teglbjaerg, Mogens Vyberg. France: Pascal Guénel, Joëlle Févotte and the members of the FRANCIM association: Patrick Arveux, Antoine Buemi, Paule-Marie Carli, Gilles Chaplain, Jean-Pierre Daurès, Jean Faivre, Pascale Grosclaude, Anne-Valérie Guizard, Michel Henry-Amar, Guy Launoy, Francois Ménégoz, Nicole Raverdy, Paul Schaffer. Germany: Wolfgang Ahrens, Cornelia Baumgardt-Elms, Thomas Behrens, Sibylle Gotthardt, Ingeborg Jahn, Karl-Heinz Jöckel, Hiltrud Merzenich, Andreas Stang, Christa Stegmaier, Antje Timmer, Hartwig Ziegler. Italy: Terri Ballard, Franco Bertoni, Giuseppe Gorini, Sandra Gostinicchi, Giovanna Masala, Enzo Merler, Franco Merletti, Lorenzo Richiardi, Lorenzo Simonato, Paola Zambon. Latvia: Irena Rogovska, Galina Sharkova, Aivars Stengrevics. Lithuania: Jolita Gibaviciene, Laimonas Jazukevicius, Juozas Kurtinaitis, Poma Pociute. Portugal: Noemia Alfonso, Altamiro Costa-Pereira, Sonia Doria, Carlos Lopes, José Manuel Lopes, Ana Miranda, Cristina Santos. Spain: M Adela Sanz Aguado, Juan J Aurrekoetxea, Concepción Brun, Alicia Córdoba, Miguel Angel Martínez González, Francisco Guillén Grima, Rosa Guarch, Agustin Llopis González, Blanca Marín, Amparo Marquina, María M Morales Suárez-Varela, Inés Aguinaga Ontoso, JM Martínez Peñuela, Ana Puras, Francisco Vega, Maria Aurora Villanueva Guardia. Sweden: Mikael Eriksson, Lennart Hardell, Irene Larsson, Hakan Olson, Monica Sandström, Gun Wingren. Switzerland: Jean-Michel Lutz. United Kingdom: Janine Bell, Ian Cree, Tony Fletcher, Alex JE Foss.

References

View Abstract

Footnotes

  • Funding Funding was supplied by the European Commission, DGXII, BIOMED research programme grants no BMH1 CT 93-1630 and ERB CIPD CT 940285, and the following national funding agencies. Denmark: The Strategic Environment Programme, grant no 92.01.015.7-06; the Danish Epidemiology Science Centre whose activities are financed by a grant from the Danish National Research Foundation. France: Ligue Nationale contre le Cancer, Fédération Nationale des Centres de Lutte contre le Cancer, Fondation de France, contract no 955368; Institut National de la Santé et de la Recherche Médicale (INSERM) contract Réseau en Santé Publique no 4R006A; French Ministry of the Environment, contract no 237.01.94.40182. Germany: Federal Ministry for Education, Science, Research and Technology (BMBF), grant no 01-HP-684/8. Italy: The Italian Association for Cancer Research (AIRC), Special Project Oncology, Compagnia di San Paolo/FIRMS, MURST, Piedmont Region. Spain: Fondo de Investigación de la Sanitarie, Ministerio de Sanidad y Consumo, Unidad de Investigación Clinico-Epidemiológica, Hospital Dr. Peset, Generalitat Valenciana (FISS. 95/0044-01, 96/0043-01); Departmento de Sanidad y Consumo, Gobierno Vasco; Fondo de Investigación de la Sanitaria, Ministerio de Sanidad y Consumo, Ayuda a la Investigación del Departamento de Salud del Gobierno de Navarra. Sweden: Swedish Council for Work Life Research; Research Foundation of the Department of Oncology in Umeå; Swedish Society of Medicine; Lund University Hospital Research Foundation; Gunnar, Arvid and Elisabeth Nilsson Cancer Foundation; Örebro County Council Research Committee, Örebro Medical Center Research Foundation; John and Augusta Persson Foundation for Scientific Medical Research; Berta Kamprad Foundation for Cancer Research.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the national ethics committees.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.