Australian Radiation Protection and Nuclear Safety Agency (ARPANSA), Yallambie, VIC, Australia.
Competence Center for Electromagnetic Fields, Federal Office for Radiation Protection (BfS), Cottbus, Germany; Institute of Medical Biostatistics, Epidemiology and Informatics (IMBEI), University of Mainz, Germany(1).
Environ Int. 2024 Sep;191:108983. doi: 10.1016/j.envint.2024.108983. Epub 2024 Aug 30.
The objective of this review was to assess the quality and strength of the evidence provided by human observational studies for a causal association between exposure to radiofrequency electromagnetic fields (RF-EMF) and risk of the most investigated neoplastic diseases.
Eligibility criteria: We included cohort and case-control studies of neoplasia risks in relation to three types of exposure to RF-EMF: near-field, head-localized, exposure from wireless phone use (SR-A); far-field, whole body, environmental exposure from fixed-site transmitters (SR-B); near/far-field occupational exposures from use of hand-held transceivers or RF-emitting equipment in the workplace (SR-C). While no restrictions on tumour type were applied, in the current paper we focus on incidence-based studies of selected "critical" neoplasms of the central nervous system (brain, meninges, pituitary gland, acoustic nerve) and salivary gland tumours (SR-A); brain tumours and leukaemias (SR-B, SR-C). We focussed on investigations of specific neoplasms in relation to specific exposure sources (i.e. E-O pairs), noting that a single article may address multiple E-O pairs.
Eligible studies were identified by literature searches through Medline, Embase, and EMF-Portal. Risk-of-bias (RoB) assessment: We used a tailored version of the Office of Health Assessment and Translation (OHAT) RoB tool to evaluate each study's internal validity. At the summary RoB step, studies were classified into three tiers according to their overall potential for bias (low, moderate and high).
We synthesized the study results using random effects restricted maximum likelihood (REML) models (overall and subgroup meta-analyses of dichotomous and categorical exposure variables), and weighted mixed effects models (dose-response meta-analyses of lifetime exposure intensity). Evidence assessment: Confidence in evidence was assessed using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach.
We included 63 aetiological articles, published between 1994 and 2022, with participants from 22 countries, reporting on 119 different E-O pairs. RF-EMF exposure from mobile phones (ever or regular use vs no or non-regular use) was not associated with an increased risk of glioma [meta-estimate of the relative risk (mRR) = 1.01, 95 % CI = 0.89-1.13), meningioma (mRR = 0.92, 95 % CI = 0.82-1.02), acoustic neuroma (mRR = 1.03, 95 % CI = 0.85-1.24), pituitary tumours (mRR = 0.81, 95 % CI = 0.61-1.06), salivary gland tumours (mRR = 0.91, 95 % CI = 0.78-1.06), or paediatric (children, adolescents and young adults) brain tumours (mRR = 1.06, 95 % CI = 0.74-1.51), with variable degree of across-study heterogeneity (I = 0 %-62 %). There was no observable increase in mRRs for the most investigated neoplasms (glioma, meningioma, and acoustic neuroma) with increasing time since start (TSS) use of mobile phones, cumulative call time (CCT), or cumulative number of calls (CNC). Cordless phone use was not significantly associated with risks of glioma [mRR = 1.04, 95 % CI = 0.74-1.46; I = 74 %) meningioma, (mRR = 0.91, 95 % CI = 0.70-1.18; I = 59 %), or acoustic neuroma (mRR = 1.16; 95 % CI = 0.83-1.61; I = 63 %). Exposure from fixed-site transmitters (broadcasting antennas or base stations) was not associated with childhood leukaemia or paediatric brain tumour risks, independently of the level of the modelled RF exposure. Glioma risk was not significantly increased following occupational RF exposure (ever vs never), and no differences were detected between increasing categories of modelled cumulative exposure levels.
In the sensitivity analyses of glioma, meningioma, and acoustic neuroma risks in relation to mobile phone use (ever use, TSS, CCT, and CNC) the presented results were robust and not affected by changes in study aggregation. In a leave-one-out meta-analyses of glioma risk in relation to mobile phone use we identified one influential study. In subsequent meta-analyses performed after excluding this study, we observed a substantial reduction in the mRR and the heterogeneity between studies, for both the contrast Ever vs Never (regular) use (mRR = 0.96, 95 % CI = 0.87-1.07, I = 47 %), and in the analysis by increasing categories of TSS ("<5 years": mRR = 0.97, 95 % CI = 0.83-1.14, I = 41 %; "5-9 years ": mRR = 0.96, 95 % CI = 0.83-1.11, I = 34 %; "10+ years": mRR = 0.97, 95 % CI = 0.87-1.08, I = 10 %). There was limited variation across studies in RoB for the priority domains (selection/attrition, exposure and outcome information), with the number of studies evenly classified as at low and moderate risk of bias (49 % tier-1 and 51 % tier-2), and no studies classified as at high risk of bias (tier-3). The impact of the biases on the study results (amount and direction) proved difficult to predict, and the RoB tool was inherently unable to account for the effect of competing biases. However, the sensitivity meta-analyses stratified on bias-tier, showed that the heterogeneity observed in our main meta-analyses across studies of glioma and acoustic neuroma in the upper TSS stratum (I = 77 % and 76 %), was explained by the summary RoB-tier. In the tier-1 study subgroup, the mRRs (95 % CI; I) in long-term (10+ years) users were 0.95 (0.85-1.05; 5.5 %) for glioma, and 1.00 (0.78-1.29; 35 %) for acoustic neuroma. The time-trend simulation studies, evaluated as complementary evidence in line with a triangulation approach for external validity, were consistent in showing that the increased risks observed in some case-control studies were incompatible with the actual incidence rates of glioma/brain cancer observed in several countries and over long periods. Three of these simulation studies consistently reported that RR estimates > 1.5 with a 10+ years induction period were definitely implausible, and could be used to set a "credibility benchmark". In the sensitivity meta-analyses of glioma risk in the upper category of TSS excluding five studies reporting implausible effect sizes, we observed strong reductions in both the mRR [mRR of 0.95 (95 % CI = 0.86-1.05)], and the degree of heterogeneity across studies (I = 3.6 %).
Consistently with the published protocol, our final conclusions were formulated separately for each exposure-outcome combination, and primarily based on the line of evidence with the highest confidence, taking into account the ranking of RF sources by exposure level as inferred from dosimetric studies, and the external coherence with findings from time-trend simulation studies (limited to glioma in relation to mobile phone use). For near field RF-EMF exposure to the head from mobile phone use, there was moderate certainty evidence that it likely does not increase the risk of glioma, meningioma, acoustic neuroma, pituitary tumours, and salivary gland tumours in adults, or of paediatric brain tumours. For near field RF-EMF exposure to the head from cordless phone use, there was low certainty evidence that it may not increase the risk of glioma, meningioma or acoustic neuroma. For whole-body far-field RF-EMF exposure from fixed-site transmitters (broadcasting antennas or base stations), there was moderate certainty evidence that it likely does not increase childhood leukaemia risk and low certainty evidence that it may not increase the risk of paediatric brain tumours. There were no studies eligible for inclusion investigating RF-EMF exposure from fixed-site transmitters and critical tumours in adults. For occupational RF-EMF exposure, there was low certainty evidence that it may not increase the risk of brain cancer/glioma, but there were no included studies of leukemias (the second critical outcome in SR-C). The evidence rating regarding paediatric brain tumours in relation to environmental RF exposure from fixed-site transmitters should be interpreted with caution, due to the small number of studies. Similar interpretative cautions apply to the evidence rating of the relation between glioma/brain cancer and occupational RF exposure, due to differences in exposure sources and metrics across the few included studies.
This project was commissioned and partially funded by the World Health Organization (WHO). Co-financing was provided by the New Zealand Ministry of Health; the Istituto Superiore di Sanità in its capacity as a WHO Collaborating Centre for Radiation and Health; and ARPANSA as a WHO Collaborating Centre for Radiation Protection.
PROSPERO CRD42021236798. Published protocol: [(Lagorio et al., 2021) DOI https://doi.org/10.1016/j.envint.2021.106828].
本综述的目的是评估人类观察性研究提供的关于射频电磁场（RF-EMF）暴露与最受调查的肿瘤性疾病风险之间因果关系的证据的质量和强度。
纳入标准：我们纳入了与三种射频电磁场暴露类型相关的三种类型的肿瘤风险的队列和病例对照研究：近场、头部局部暴露，来自无线电话使用的暴露（SR-A）；远场、全身、来自固定站点发射机的环境暴露（SR-B）；来自工作场所使用手持式收发器或射频发射设备的职业近场/远场暴露（SR-C）。虽然没有对肿瘤类型施加任何限制，但在目前的论文中，我们重点关注选定的“关键”中枢神经系统（脑、脑膜、垂体腺、听神经）和唾液腺肿瘤（SR-A）；脑肿瘤和白血病（SR-B、SR-C）的基于发病率的研究。我们关注的是与特定暴露源（即 E-O 对）相关的特定肿瘤的调查结果，注意到一篇文章可能涉及多个 E-O 对。
通过 Medline、Embase 和 EMF-Portal 进行文献检索，确定符合条件的研究。风险评估：我们使用定制版的卫生评估和翻译办公室（OHAT）风险评估工具来评估每项研究的内部有效性。在总结风险步骤中，将研究分为三个风险等级，根据其总体偏倚潜力（低、中、高）进行分类。
我们使用随机效应受限最大似然（REML）模型（对二分类和分类暴露变量的综合和亚组荟萃分析）和加权混合效应模型（对终生暴露强度的剂量反应荟萃分析）对研究结果进行综合。证据评估：使用 Grading of Recommendations, Assessment, Development and Evaluations（GRADE）方法评估证据的可信度。
我们纳入了 63 项病因学文章，发表于 1994 年至 2022 年，来自 22 个国家，报告了 119 个不同的 E-O 对。使用移动电话（曾经或经常使用与没有或不经常使用）与胶质瘤（mRR=1.01，95%CI=0.89-1.13）、脑膜瘤（mRR=0.92，95%CI=0.82-1.02）、听神经瘤（mRR=1.03，95%CI=0.85-1.24）、垂体瘤（mRR=0.81，95%CI=0.61-1.06）、唾液腺肿瘤（mRR=0.91，95%CI=0.78-1.06）或儿童（青少年和年轻人）脑肿瘤（mRR=1.06，95%CI=0.74-1.51）的风险没有关联，各研究间存在不同程度的异质性（I=0%-62%）。对于最受调查的肿瘤（胶质瘤、脑膜瘤和听神经瘤），随着移动电话使用的开始时间（TSS）、累积通话时间（CCT）或累积通话次数（CNC）的增加，mRR 没有观察到增加。无绳电话使用与胶质瘤（mRR=1.04，95%CI=0.74-1.46；I=74%）、脑膜瘤（mRR=0.91，95%CI=0.70-1.18；I=59%）或听神经瘤（mRR=1.16；95%CI=0.83-1.61；I=63%）的风险无显著关联。来自固定站点发射机（广播天线或基站）的暴露与儿童白血病或儿童脑肿瘤风险无关，独立于建模的射频暴露水平。职业射频暴露（曾经与从未）与胶质瘤风险增加无关，并且在建模累积暴露水平的递增类别之间未检测到差异。
在移动电话使用与胶质瘤风险的亚组分析中，（曾用、TSS、CCT、CNC）进行了敏感性分析，结果稳健，不受研究汇总的影响。在移动电话使用与胶质瘤风险的个体研究剔除分析中，我们确定了一项有影响力的研究。在剔除该研究后进行的随后的荟萃分析中，我们观察到 mRR 和研究间的异质性显著降低，无论是在 Ever 与（经常）使用（regular）的对比（mRR=0.96，95%CI=0.87-1.07，I=47%），还是在按 TSS 递增类别进行的分析中（<5 年：mRR=0.97，95%CI=0.83-1.14，I=41%；“5-9 年”：mRR=0.96，95%CI=0.83-1.11，I=34%；“10+ 年”：mRR=0.97，95%CI=0.87-1.08，I=10%）。在选择/损耗、暴露和结果信息等优先领域的风险评估中，各研究的 RoB 存在差异，其中 49%（分层 1 级和 51%（分层 2 级）的研究被归类为低和中度风险，没有研究被归类为高风险（分层 3 级）。研究结果（数量和方向）的偏差影响难以预测，RoB 工具也无法考虑竞争偏差的影响。然而，在分层 RoB 分层的敏感性荟萃分析中，我们在胶质瘤和听神经瘤的 TSS 上层（I=77%和 76%）的主要荟萃分析中观察到的异质性，解释了汇总 RoB 分层。在分层 1 研究子集中，长期（10+ 年）使用者的胶质瘤 mRR（95%CI；I）为 0.95（0.85-1.05；5.5%），听神经瘤 mRR 为 1.00（0.78-1.29；35%）。时间趋势模拟研究评估为外部有效性的三角方法的补充证据，一致表明，一些病例对照研究中观察到的增加风险与几个国家和长期观察到的胶质瘤/脑癌的实际发病率不相符。其中三项模拟研究一致报告，RR 估计值>1.5 且具有 10+ 年诱导期的肯定是不合理的，可用于设定“可信度基准”。在 TSS 上层类别中，将排除五项报告不合理效应大小的研究后，我们观察到 mRR [mRR 为 0.95（95%CI=0.86-1.05）]和研究间异质性（I=3.6%）的强烈降低。
根据发表的方案，我们根据风险评估的最高可信度，对每个暴露-结果组合单独制定最终结论，主要基于射频源暴露水平的证据等级推断（根据剂量学研究），并与移动电话使用时与胶质瘤相关的时间趋势模拟研究（仅限于胶质瘤）的外部一致性。对于头部近场射频电磁场暴露来自移动电话使用，有中度确定性证据表明，它可能不会增加成年人的胶质瘤、脑膜瘤、听神经瘤、垂体瘤和唾液腺肿瘤，以及儿童脑肿瘤的风险。对于头部近场射频电磁场暴露来自无绳电话使用，有低确定性证据表明，它可能不会增加胶质瘤或听神经瘤的风险。对于来自固定站点发射机（广播天线或基站）的全身远场射频电磁场暴露，有中度确定性证据表明，它不太可能增加儿童白血病的风险，低确定性证据表明，它可能不会增加儿童脑肿瘤的风险。没有研究可用于调查来自固定站点发射机的射频电磁场暴露和成年人的关键肿瘤。对于职业射频电磁场暴露，有低确定性证据表明，它可能不会增加脑癌/胶质瘤的风险，但针对白血病的纳入研究很少。
本项目
