A Report on Non-Ionizing Radiation

Summary and Conclusions of EPA's EMF Cancer Report


Reprinted below are the "Summary and Conclusions" of the Environmental Protection Agency's (EPA) draft report, An Evaluation of the Potential Carcinogenicity of Electromagnetic Fields (EMFs). This summary, dated June 1990, will accompany the draft of the report which will be distributed to the external review panel, assembled by the EPA.


This information originally appeared in the 1990 May/June issue of Microwave News.


7.1. Introduction


In this chapter each of the major chapters in the document are summarized; a final section presents a discussion of the relationships among the individual chapters and the overall conclusions.


7.2. Mechanisms of Interaction Between Tissue and EMFs


The basic processes by which energy from EMFs of RF and ELF frequencies is coupled to the body are described in this section. The frequency dependence of the RF power absorbed by an organism is dominated by the body size, so that mice, rats and humans have different RF absorption characteristics. For ELF fields and the lower RF frequencies near the source, the relationship between the electric and magnetic fields is not fixed, as it is for RF fields, and they are evaluated separately in this document. From the point of view of EMFs, the body is composed of a solution of ions; it is an electrical conductor and the penetration of electric fields into the body is very poor at ELF frequencies. Since the body is composed of nonmagnetic materials, an external time-varying magnetic field permeates the body, inducing ionic currents.


The human evidence, as described in the next section, suggests that magnetic fields rather than electric fields are associated with cancer incidence, and mechanisms have been sought to explain how weak currents induced by ELF magnetic fields could interact with cells and body tissue in such a way as to induce a carcinogenic response. Three classes of models for this interaction are reviewed: (1) The surface compartment model deals with the movement of ions towards and away from the inner and outer surfaces of the plasma membrane of the cell, and deals with ion-selective membrane channels, ionic pumps and membrane ion fluxes. The model describes the movement of ions in response to perturbations of electric fields and magnetically induced currents around the cell. (2) The ion cyclotron resonance hypothesis was developed in part to explain the frequency sensitivity of calcium ion efflux studies of brain tissues. If the relationship among the frequency of time-varying magnetic field, the strength of a parallel static magnetic field and the ionic charge to mass ratio of an ionic species is correct, then the ion will resonate, or synchronously follow circular paths in a plane perpendicular to the field. In one experiment demonstrating this effect, conditions were set up for calcium cyclotron resonance, and the movement of bethnic diatoms was measured. The experiment showed that calcium ions entered into the cell under these specific conditions and stimulated the motion of the cells, whereas the cell is normally impermeable to calcium. This type of mechanism could be the basis of an induced selective ion permeability of the plasma membrane and might ultimately be capable of explaining both frequency selectivity of these effects and the sensitivity to small induced currents. (3) Another class of models deals with cooperative motions of an ordered array of lipid bilayer molecules and describes how a weak field affecting the motion of the whole array could be transferred to just one site in the array. These theories have not yet been tested in the context of ELF biological processes. At the present time, these basic models of tissue interaction with EMFs cannot be linked to the biochemical or cellular processes involved in the development of malignant growth.


7.3. Human Evidence


The effects of human exposure to EMFs from several sources have been reported. This document discusses ELF fields separately from higher frequency exposure where possible. Children with residential exposure are more appropriate subjects than adults for evaluating the effects of ELF fields, since children have relatively little exposure to higher frequency fields and occupational chemicals as a consequence of their normal activity patterns. Consequently, studies of childhood cancer associated with residential exposure to 60 Hz power frequency fields are discussed separately from occupational exposure to adults, which involves a mixture of both ELF and RF fields.


7.3.1. Studies of Children: There have been six case-control studies of cancer in children examining residential exposure from power transmission and distribution systems and two additional studies examining childhood cancer in relationship to fathers' occupations. Five of the six residential exposure studies showed positive associations with ELF field exposure; three were statistically significant and the other two had odds ratios greater than one but not statistically significant. Where different cancer sites were evaluated, leukemia, nervous system cancer and, to a lesser extent, lymphoma were found to be in excess in the five residential studies showing positive associations. (Electric fields are not believed to be a critical factor.) The measure of magnetic field exposure was variable in the six studies. Wire code configurations and proximity to distribution lines were used for five of the six studies, and measurements were taken in two of the six studies. There is a good, but not perfect, correlation between measured magnetic fields and wire code configurations. In two of the studies in which magnetic field measurements were made, cases were observed in those exposed above 2-3 mG (0.2-0.3 µT) but not in people exposed below that level. Although one study found a significant dose-response relationship with respect to surrogate measures of exposure, reliable dose-response information is not available due to the use of dichotomized exposure categories and due to small numbers of cases within the exposure groupings. Issues pertaining to personal exposure and latency have not been ad dressed. The two paternal occupational studies found statistically significant associations between neuroblastoma and brain cancer, respectively, in children and their fathers' exposure to EMFs. There was also one report of a cluster of a rare tumor type (endodermal sinus tumors) in adolescent girls exposed to EMFs from power distribution lines....


The consistently repeated pattern of leukemia, nervous system cancer and lymphoma in the childhood studies and the ruling out of several confounding exposure factors in the Savitz et al. (1988) study argue in favor of a causal link between these tumor types in children and exposure to ELF magnetic or electric fields. However, the fact that the odds ratios are small and in many cases not statistically significant indicates that the association may not be strong and there fore argues against a causal relationship.


7.3.2. Studies of Adults Residential Exposure to Power Frequency Fields:In a total of four reports of cancer mortality or incidence in adults residing near electrical power distribution equipment, three were nonpositive and the fourth was clearly positive. In a case-control study of adult leukemia in Los Angeles, there was no association with electric blanket usage in the home. In contrast to childhood cancer, the adult studies were more difficult to interpret because they lacked the statistical power to detect a relative risk of the magnitude typically observed in the childhood studies and because there was little evidence that the cases were exposed to levels of EMFs higher than fields to which the controls were exposed. Therefore, no conclusions can be made about a cancer response and adult exposure to residential fields. Occupational Exposure to ELF and Mixed Frequency Fields: Over 30 reports dealing with cancer incidence or mortality in workers in electrical and electronic occupations have been reviewed. These exposures have involved 50 or 60 Hz power frequency fields as well as mixtures of higher frequency fields which are typically poorly defined. The studies have been carried out in Europe, New Zealand and the United States. Many of them were re-examinations of previous studies or evaluations of vital records, cancer registry or occupational data bases, and thus the populations were not formed to test the specific hypothesis of whether EMF exposure is associated with increased cancer risk. Most of them used death certificates as a source of occupational information; this information furnishes only a very crude indicator of actual exposure to EMFs. Many of these are proportional mortality studies, which are less informative than studies of cohort and case-control designs because their results are affected by extraneous causes of death.


In these studies three types of cancer predominate: (1) hematopoietic system, especially leukemia and specifically acute myeloid leukemia; (2) nervous system cancer, including brain tumors; and (3) malignant melanoma of the skin. These cancer sites are found consistently across different geographic regions, age groups, industries, occupational classifications and study designs. Given this diversity of studies, in addition to the likelihood that across broad job categories the exposures to various chemicals is not uniform, it is difficult to identify any single agent or group of confounding exposures that could explain the consistent finding of these same cancer sites. RF Exposure: Reports that focused primarily on exposures to RF radiation have shown mixed results, but most of the studies were difficult to interpret. Two early reports concerning microwave [MW] exposure of the U.S. embassy personnel in Moscow and radar exposure of U.S. Navy personnel showed only a slight tendency for increased cancer risk at all sites, and somewhat higher odds ratios for hematopoietic system cancers. A study of personnel in a World War II radar research and development laboratory found no convincing evidence of increased cancer incidence, but errors of exposure misclassification are likely. A series of reports of ham radio operators found a statistically elevated incidence of acute myeloid leukemia and other neoplasms of the lymphoid system, but no clear dose-response trend was seen with longer exposure, where the degree of exposure was inferred by FCC operator license class. One report of military exposure to radar found increasing rates of hematopoietic cancer of specific sites, but a lack of detail limits the ability to interpret the results.


7.3.3. Summary of Human Evidence: The strongest evidence that there is a causal relationship between certain forms of cancer, namely leukemia, cancer of the nervous system and, to a lesser extent, lymphoma, and exposure to magnetic fields comes from the childhood cancer studies. Several studies have consistently found modestly elevated risks (some statistically significant) of these three site-specific cancers in children. In two of the studies in which magnetic field measurements were made, cases were observed in those exposed above 2-3 mG (0.2-0.3 µT) but not in people exposed below that level. This is supported by the fact that children have relatively few confounding influences that could explain the association. In fact, the few potential confounders and biases that might have had an effect on the results were examined by one of the authors in some detail and found not to be a serious problem. No other agents have been identified to explain this association. However, there are contradictory results within these same studies, and dose-response relationships could not be substantiated. Furthermore, there is little information on personal exposure and length of residency in the EMFs.


Additional, but weaker evidence that there is an elevated risk of leukemia, cancer of the nervous system and perhaps other sites comes from occupational studies of EMF exposure. Although many of these studies have found an excess risk of these forms of cancer with employment in certain jobs that have a high potential for exposure to EMFs, few or no measurements have actually been taken in those occupations. Furthermore, information about occupation has come generally from sources that could be characterized as sketchy. The likelihood that misclassification or information bias is present in these studies is high. However, exposure misclassification, if random, tends to bias relative risks toward the null. Despite these weaknesses, the occupational studies tend to support the results of the childhood studies, since the excess relative risks occur at the same sites.


The studies of residential adult exposures to EMFs provide little evidence of a risk of leukemia, mainly due to lack of statistical power and/or probably little exposure to levels of EMFs which have been found to be associated with cancer in children. These studies cannot be interpreted as evidence either for or against a causal association between cancer and EMF exposures. On the other hand, the case-control study of cancer in Colorado residents does support an association of central nervous system cancer and lymphoma if proximity to high-current electrical wiring configurations is assumed to be an adequate surrogate for exposure.


The studies of adults exposed to RF radiation produced mixed results, primarily because of limited sample size, inadequate length of follow-up, imprecise exposure data and lack of information on potential confounders. These problems prevent conclusions to be made about causal relationships with RF exposures. However, the statistically significant excess risks of leukemia in amateur radio operators require further examination.

7.4. Animal Carcinogenicity Evidence


7.4.1. ELF Fields: No lifetime animal carcinogen bioassay studies of ELF fields have been reported in the literature. Two studies currently in progress are designed to observe the induction of a carcinogenic response to chronic magnetic field exposures.


7.4.2. RF Radiation: Two chronic studies in mice have used unmodulated RF radiation at 800 MHz and 2450 MHz, respectively. Two studies in rats have used pulse modulated 2450 MHz of low power density and pulsed RF of all frequencies from 0 to about 20 MHz of high power density, respectively. One mouse study used pulsed RF radiation of 9270 MHz. Unmodulated RF Radiation: For unmodulated RF radiation, one of the mouse studies (Szmigielski et al., 1982) shows that the radiation enhances the growth rate of spontaneous mammary tumors and in a separate experiment enhances the growth rate of skin tumors initiated by a chemical carcinogen, benzo(a)pyrene. In a shorter test (3 months), the same authors showed that the radiation also enhances the growth rate of transplanted lung carcinoma cells, an effect attributed to the lowering of cell-mediated immunity. Unfortunately, histopathology was not reported in the other mouse study (Spalding et al., 1971), so conclusions about carcinogenicity from that study are difficult to make.


The special nature of the response indicates that unmodulated RF radiation might be a promoter or cocarcinogen, since the growth rate of spontaneous breast tumors, BaP-induced skin tumors and transplanted lung sarcoma cells is enhanced by the radiation. There is a remote possibility that body heating could have contributed to this response, since the absorbed RF power is estimated to be at least one-half of the basal metabolic rate of the animals. Modulated RF Radiation: For modulated RF radiation of relatively low power density [i.e., excluding the high power electromagnetic pulse (EMP) experiment of Baum et al. (1976)], the mouse experiment (Prausnitz and Susskind, 1962) showed a reversible pattern of lymphoma and leukemia which, in serial sacrifices, occurred toward the end of the 14-month exposure period but was not present in animals after a 5-month recovery period. However, the short 4.5-minute daily exposure was intense enough to raise the body core temperature by 3ƒC, raising the possibility that thermal effects were a contributing factor in the response. The rat study (Guy et al., 1985) showed the induction of benign adrenal medulla pheochromocytomas and a statistically significant increase in carcinomas of all organ and tissue sites. There was also a higher incidence of glandular organ carcinomas which was unaccompanied by an in crease in the incidence of benign tumors of these sites. Although an increase of tumors of all types in the aggregate is regarded as only minimal evidence of carcinogenic action by chemical agents, the fact that the RF radiation affects all tissues in the body independently of their individual biochemical characteristics constitutes a reason to consider the aggregate count a relevant finding.


7.5. Supporting Evidence of Carcinogenicity


Section 5.11. presents a summary of the effects of EMFs on a variety of basic biological phenomena relevant in some way to mechanisms of carcinogenesis; that information is not repeated here. ELF fields of relatively high intensity (producing induced body currents on the order of 10 µA/cm2 have enhanced DNA synthesis, altered the transcription of DNA into mRNA, altered the molecular weight distribution during protein synthesis, delayed the mitotic cell cycle, induced chromosome aberrations, blocked the action of parathyroid hormone at the site of its plasma membrane receptor, induced enzymes normally active during cell proliferation, inhibited differentiation and stimulated the growth of carcinoma cell lines, inhibited the cytotoxicity of T-lymphocytes (which indicates an impairment of the immune system) in vitrobut not in vivo, inhibited the synthesis of melatonin (a hormone that suppresses the growth of several types of tumors), disrupted the morphology of neurons and glial cells in the central nervous system and caused alterations in the binding of calcium to brain tissues. The large variety of exposure conditions and the lack of detail on the geometry of the biological samples in these studies preclude a systematic evaluation of the actual induced currents and field strengths at the tissue and cellular level that are causing these effects. In addition, the lack of reproducible results between laboratories limits the interpretation of much of this literature.


RF fields modulated at the same ELF frequencies that cause some of the effects noted above also result in the same responses, indicating that the ELF component may be responsible for these effects. Unmodulated RF radiation has not caused any of the effects noted above except for chromosome aberrations. None of the EMFs have caused gene mutations, sister chromatid exchanges or DNA damage (as measured by DNA breaks, DNA repair or differential killing of repair defective organisms) in a large number of studies.


Only three ELF effects have been induced at field strengths comparable to the low environmental exposures at which human cancer has putatively been caused: (1) the calcium efflux from brain tissue preparations using 16 Hz crossed electric and magnetic fields; (2) calcium efflux from chick brain tissue after exposure of the developing embryo to electric fields; and (3) the inhibition of melatonin synthesis by the pineal gland when a static magnetic field of approximately the strength of the earth's magnetic field is changed through a small angle of rotation. The results of the first experiment are one of several phenomena that show a complex dependence on frequency, intensity and orientation with respect to the earth's magnetic fields.


In view of these laboratory studies, there is reason to believe that the findings of carcinogenicity in humans are biologically plausible. However, the explanation of which of the biological processes is involved and the way in which these processes causally relate to each other and to the induction of malignant tumors is not understood. Most of the effects have been observed at field strengths that are many times higher than the ambient fields which are the putative cause of the childhood cancers in residential situations; as a consequence, many of the candidate mechanisms may not be really involved in the response to low environmental fields. The same issue of low-dose extrapolation arises in the evaluation of chemical agents.


7.6. Integrated Discussion of Separate Chapters


The occurrence of cancer in humans exposed to low frequency EMFs has been observed under several different conditions in different populations. Residential exposure of children, but not adults, has been associated with leukemia, lymphoma and brain cancer, and the same sites occur in multiple studies of children. The fields involved in these associations are magnetic and not electric fields, made up of primarily 60 Hz components but with inevitable high frequency components introduced by electric motors and the switching of currents on and off. In a few studies, these effects have been observed in children exposed to average magnetic fields above 2-3 mG (0.2-0.3 µT) but have not been seen in children exposed to smaller fields. The types of EMF exposures in the occupational studies are variable according to job category, with some jobs involving pulsed and modulated RF fields as well as 60 Hz power frequency components.


There is some, but not well-established, evidence that higher frequency components have different effects than 60 Hz components. Electrical switchyard workers exposed to spark discharges just before blood samples are taken have chromosome aberrations, whereas similar workers with no such exposure do not. Chromosome aberrations have been induced by unmodulated RF fields as well as by ELF fields. A recent preliminary report of an epidemiologic study of telephone workers shows a different effect (rare breast tumors in males) in people working in the "central office," where switching equipment is typically concentrated, than in cable splicers (leukemia) who presumably are exposed to predominantly 60 Hz power frequencies. Both electric and magnetic fields are more effective in inducing currents in the body if their frequency is higher, so that if induced currents are responsible for these effects, then the higher frequency components are expected to be more effective. If it is true that, as two studies indicate, the fathers' occupations in electrical jobs is a factor in the development of leukemia in their children, then the question is raised whether the effect could be transmitted via heritable genetic damage in sperm. This speculative hypothesis needs to be investigated.


Although there are several candidate EMF-induced biological phenomena...that could explain how a cancer response is caused in the whole organism by these fields, none of these or any combination of them has been verified experimentally, either in laboratory animals or in humans. Without understanding which combination of these is relevant to the carcinogenic process, it is not possible to hypothesize what aspect of EMF exposure is responsible for biological effects: i.e., frequency, average peak field strength, duration, time of day, whole-body average versus local critical site, electric versus magnetic fields, orientation with respect to the earth's static magnetic field. The choice of which aspect of the fields is the most relevant could be based on either knowledge of the correct mechanism of action or on empirical epidemiology correlations, but, given the current lack of information, neither method can serve as a basis for a dose-response analysis.


There are several indications that EMFs might contribute to the induction of cancer via indirect mechanisms, in contrast to a direct mutagenic action of DNA as is the case with nitrosamines, polycyclic aromatic hydrocarbons or other DNA-alkylating agents.


First, EMFs have not caused gene mutations in any of the large number of experiments carried out with both ELF and RF fields.


Second, there is no indication from the animal studies that RF fields cause a de-novo induction of tumors. On the contrary, the mice experiments by Szmigielski et al. (1982) indicate that unmodulated RF radiation acts as a growth stimulator for preexisting tumors. The same growth-stimulating or promotion characteristics of RF fields could explain the induction of glandular tumors in the Guy et al. (1985) lifetime rat study of modulated RF radiation, since many of the glandular tumors in that study had a naturally high spontaneous incidence.


A third factor indicating that there may be multiple causes of carcinogenic action is that 120 Hz-modulated 2450 MHz radiation can act as an initiator of phorbol ester-promoted cell transformation in mouse embryo cell cultures.


Finally, there are possible cancer induction mechanisms mediated by the central nervous system causing neuroendocrine influence on cellular proliferation. These mechanisms involve possible extremely sensitive detection of magnetic fields by the retina with resulting neural control of pineal melatonin activity, which in turn modulates estrogen and prolactin levels in the blood supply to the breast, prostate and other hormonally-sensitive tissues. Other speculative chains of events could be fabricated from the existing information in this document, and this one is mentioned here only as an example that there are many possible explanations but no verified ones.


In view of this, it is likely that if EMFs do contribute to the induction of cancer, the causal relationship will probably turn out to be dependent on many chemical factors and physiological conditions that are currently poorly understood.


There are two issues in the hazard evaluation of chemical carcinogens that are analogous to issues for EMFs. It may be helpful to explore whether the assumptions and conventions developed for chemicals are applicable to the EMF problem.


One analogy is that EMFs are mixtures consisting of several frequencies, intensities and combinations of electric and magnetic fields, which (for ELF frequencies) occur in arbitrary proportions. One approach to the assessment of chemical mixtures is to identify hazardous components of the mixture and, assuming additivity of components, consider the risk of the mixture to be proportional to the risk of the hazardous components. If this concept were applied to the EMF problem, then magnetic fields from 60 Hz power usage in the home would be the only "hazardous component" identified, although there is some indication that occupational exposures of adults to mixed fields may cause the same effect. Laboratory studies under relatively controlled conditions of exposure have not been able to test the additivity assumption for EMF components or for chemical components except for a few rare cases, but one feels more comfortable with the latter. With chemical agents, the basic phenomenon is ultimately some chemical reaction, which is expected to have additive properties at low enough concentrations, or at least to be monotonic in the sense that more chemical produces a greater effect. With EMFs, however, the ultimate causative interaction between fields and biological systems is unknown, and there is certainly no additivity with RF and ELF fields, or with ELF electric and ELF magnetic fields. The consequence of not being able to add the risks for different exposures is that the effects for each combination must be investigated and assessed separately.


Another analogy is the similarity between the "biologically effective dose" for chemical agents and the critical electrical measure of tissue "dose" which causes the effect for EMFs. For chemical agents the relationship between "administered dose" and "effective dose" has been studied occasionally, but only rarely. In the absence of this information, the default position for chemical agents has been to assume a linear relationship. Then there are hosts of unresolved questions in determining whether the biological effect is proportional to the "effective dose." These questions arise when, as is usually the case, the mechanism of action is not known. Here again the linearity assumption is made in the absence of knowledge, and the overall default position is that the adverse effect is proportional to the administered dose of the chemical agent. For EMFs, the "tissue doses" could be calculated, typically with great difficulty and uncertainty, but the same type of questions needs to be answered about which of these dose metrics are relevant for EMF exposure. As with chemical agents, the choice of a candidate mechanism of action dictates which tissue dose metric is appropriate, and there could be several mechanisms for each of the administered agents. For EMFs, the default linearity assumption is not appropriate basically because there are frequency and intensity "windows" of activity for more than one EMF-induced biological effect and such "window" interactions cannot be ruled out as contributory to cancer causation.


In conclusion, the several studies showing leukemia, lymphoma and cancer of the nervous system in children exposed to magnetic fields from residential 60 Hz electrical power distribution systems, supported by similar findings in adults in several occupational studies also involving electrical power frequency exposures, show a consistent pattern of response that suggests, but does not prove, a causal link. Frequency components higher than 60 Hz cannot be ruled out as contributing factors. Evidence from a large number of biological test systems shows that these fields induce biological effects that are consistent with several possible mechanisms of carcinogenesis. However, none of these processes has been experimentally linked to the induction of tumors, either in animals or humans, by EMFs. The particular aspects of exposure to the EMFs that cause these events are not known.


In evaluating the potential for carcinogenicity of chemical agents, EPA has developed an approach that attempts to integrate all of the available information into a summary classification of the overall weight-of-evidence that the agent is carcinogenic in humans. At this time such a characterization regarding the link between cancer and exposure to EMFs is not appropriate because the basic nature of the interaction between EMFs and biological processes leading to cancer is not understood. For example, a real possibility exists that exposure to higher field strengths is actually less hazardous than exposure to low field strengths. Because of this uncertainty, it is inappropriate to make generalizations about the carcinogenicity of EMFs. As additional studies with more definitive exposure assessment become completed, a better understanding of the nature of the hazard will be gained. With our current understanding we can identify 60 Hz magnetic fields from power lines and perhaps other sources in the home as a possible, but not proven, cause of cancer in people. The absence of key information summarized above makes it difficult to make quantitative estimates of risk. Such quantitative estimates are necessary before judgments about the degree of safety or hazard of a given exposure can be made. This situation indicates the need to continue to evaluate the information from ongoing studies and to further evaluate the mechanisms of carcinogenic action and the characteristics of exposure that lead to these effects.