UNITED STAn:S ENVIRONMENTAL PROrECTtON AGENCYWASHINGTON, D.C. 20460

~l,fSJ "'4')-.

i ft l,.\SA, '. i>o(, ,.,.p.(}""

January 29. 1992

EPA-SAB-RAC-92-013

Mr. William K. Reilly, AdministratorU.S. Environmental Protection Agency401 M street SWWashington, DC 20460

Re: Potential Carclnogenicity of Electromagnetic Fields

Dear K~. Reilly:

The enclosed is the report of the Nonionizing Electric andMagnetic Fields Subcommittee of the Science Advisory Board'sRadiation Advisory Committee. The report has been endorsed bythe Science Advisory Board on the recommendation of its RadiationAdvisory Committee.

The Subcommittee was set up in response to the October 12,1990, memorandum from the Directors of the EnvironmentalProtection Agency's Offices of Health and EnvironmentalAssessment (William Farland), Health Research (Ken Sexton), andRadiation Programs (Richard Guimond). The letter requested apeer review of the draft report Evaluation of the Potentialcarcinogenicity of Electromagnetic Fields (EPA/600/6-90/aaSB).

The Subcommittee met three times. Its first meetinq, ~nJanuary 14-16, 1991, in Washinqton, D.C., elicited an uncommonlystrong public participation: over 200 people, 19 formal andseveral informal presentations, and a lively debate betweenSubcommittee members (sitting as a panel) and members of theaudience. Among the formal presentations were those made byCongressman F~ank Pallone (Dem.-New Jersey), Mayor J. Connors(Scranton, Pennsylvania), Dr. D. N. Erwin of the United statesAir Force Armatronq Laboratory for Human Systems, and Dr. RobertAdair, sterllnq Professor of Physics at Yale University. At itssecond meet1nq on April 12-13, 1991, in San Antonio, Texas, theSubcommittee received preliminary drafts prepared by its threesubgroups (on physics, biology, and epidemiology) and heardinvited presentations by biophysicist Dr. Arthur Pilla of Mt.Sinai Hospital in New York City, biologist Dr. Russel Reiter ofthe University of Texas Health Science Center in San Antonio, andphysiologist Dr. Asher Sheppard of the Pettis Memorial VeteransHospital in Loma Linda, California. At its third meeting, July23-25, 1991, in Washington, the Subcommittee reviewed a draft ofthe present report which was subsequently approved by mailwith some amendments.

B1-104

The charge to the Subcommittee was to review the doc~~e~: ~3

to the accuracy and completeness of the information provided, ~5

well as the interpretation of the scientific data. The EPAdocument has serious deficiencies and needs to be rewritten. Asthe result of many internal inconsistencies, it is oftendifficult to tell what the EPA's position. is when reading theEvaluation of the Potential Carcinogenicity of ElectromagneticFields. Portions of the draft lead to conclusions which are ~otthe conclusions stated elsewhere in the document. Suchinconsistencies appear not only between the body and theexecutive summary, but also between different pages of thedocument itself.

Consequently, the document requires a logical reorganizationand complete rewriting with particular attention to careful andprecise use of language. A simple editing of the present textwould not be sufficient. The Subcommittee consciously refrainedfrom providing a list of particular inconsistencies because itdoes not want to mislead the EPA into believing that editingalone will address the Subcommittee's concerns.

Six specific questions were posed i~ the charge to theSubcommittee. Before responding to these questions explicitly,the Subcommittee feels it important to express its viewpoint onthree underlying scientific issues, concerning (a) epidemiology,(b) biOlogical effects, and (c) carcinogenicity. These issuesare critical for the understanding of scientific issues inresearch regarding electric and magnetic fields and must beaddressed in any future EPA discussion of the potentialcarcinogenicity of electric and magnetic fields.

a. The Subcommittee concluded that some epidemiologicalevidence is suggestive of an association between surrogatemeasurements of magnetic-field exposure and certain canceroutcomes. In such studies, the existence of confounders isalways a possibility, but since no common confounder has yet beenidentified, the existing evidence cannot be dismissed. In theabsence of mµch better exposure information and an understandingof which exposures are significant, no precise exposure-responserelationship has yet been adduced. This lack, together withlimited understanding of possible biOlogical mechanisms, preventsthe inference of cancer causality from these associations at thistime.

2

b. The Subcommittee accepts that effects on somebiological systems have been shown to occur at moderate f:e~;intensities. (An example of such effects is the well-docu~e~:g;work on phosphenes.) However, the evidence for effects at ':e=~low field strengths is not so widely accepted. Even if ef:e~:~on living systems at lower fields do occur~ the assumptionsleading to estimations of physical constraints thought topreclude effects on isolated small spherical cells withoutferromagnetic structures may not be applicable to larger ce::s -­cell systems such as neurons or neur9nal networks. Manyintervening steps must be clarified before the biologicalphenomena so far shown can be taken as direct ev~dence of ~eal~~impairment or carcinogenesis in the human.

c. The EPA document does not present a holistic model ~:carcinogenesis within which the strength of existing evidenceconcerning the carcinogenic properties of electric and magneticfields can be assessed. The revised document should do so. Low­frequency electric and magnetic fields do not carry enough energyto cause mutations directly. The Subcommittee recognizes thatthe incidence of cancer might well be affected by an agent that ~does not produce mutations. The known influence of factors suchas hormonal imbalance and nutrition on cancer promotion is anexample of epigenetic effects.

. With respect to the Subcommittee's charge and the sixquestions that were posed, the Subcommittee responds as follows.

1. Is the interpretation of the human and animal evidenceof carcinogenicity supported by the available information?

currently available information is insufficient to concl~dethat the electric and magnetic fields are carcinogenic. Somehuman epidemiologic data report an association between surrogatesfor electric and magnetic field exposure and an increasedincidence of some types of cancer, but the conclusion ofcausality is currently inappropriate because of limited evidenceof an exposu~e-response relationship and the lack of a clearunderstanding of biologic plausibility.

2. Does the animal or biological effects informationprovide a basis for postulating that there is a humanhazard from exposure to extremely low-frequency fieldsor either modulated or un~odulated radiofrequencyradiation?

Some of the in vitro and in vivo data on unmodulated RF havesuggested the existence of mechanisms by which human health

]

effects, but not carcinogenicity, may be inferred. Bothunmodulated and extremely low frequency (ELF)-modulatedradio frequency (RF) fields of sufficient intensity can give :~3eto thermal effects. Nonthermal biological effects also have :ee~reported in some animals exposed to RF fields. It is wellestablished that some in vitro and in vivo experiments on E~Fhave shown nonthermal biologic effects at-tields of moderateintensity, and there are some suggestive effects at lower levels.Furthermore, ELF-modulated RF fields assumed to be nonthe~al canproduce many of the biological effects of ELF fields alone. ::should be noted that the distinction between biological effec~sand health effects is an important one. This is especially=elevant in the context of the question of possible healtheffects from electric and magnetic field exposure. Hypotheticalconstructs relating observed biological effects to possiblehealth effects (specifically, increased cancer riSk) have beendelineated. However, there are at present insufficient data onmany of the critical steps in the linkage to infer causality onthe basis of animal or cellular data.

3. Has the Agency properly evaluated the way in whichthe findings on biological effects and field-tissueinteraction mechanisms affect the interpretation of thehuman studies?

No, the EPA has not evaluated how the findings on biologicaleffects and nonionizing field and tissue interaction mechanismsrelate to the interpretation of human studies. The strenqth ~fepidemiologic data depends on identification of supportinqevidence from in vitro and in vivo data. This relationship hasnot been developed in the report. The critiques of studies ofbiological effects are contained in the discussions of theseveral chapters.

4. Is.the choice of topics in Chapter 5 appropriateand is the interpretation of the biological effectsliterature as it relates to carcinogenesis supported bythe available information?.The heading of chapter 5, "Supporting Evidence of

carcinogenicity,1t is inappropriate. The interpretation of thebiOlogical effects in the in vitro systems as presented by thereport doe. not make a case for carcinogenicity. TheSubcommittee found a lack of balance in the analysis andpresentation of evidence in this chapter. Specific individualexperiments need critical review .

4

5. Is the Agency's carcinogen classification systemapplicable to electromagnetic fields?

Nonionizing electric and magnetic fields should not beclassified under EPA's chemical carcinogenesis system because ofpresent major uncertainties. These involve an incompleteunderstanding of which aspects of field-tissue interactions giverise to biological effects. Properties of the various electricand magnetic fields such as phase angle, polarization,transients, and frequency range may contribute to differentbiological effects. For these reasons, the use of the EPA'sclassification scheme at this time would be inappropriate andconfusing. .

6. Does the information cited in the document supportthe conclusion that there is not enough information todesignate specific values of magnetic field strength asbeing hazardous to human health?

Yes, there is insufficient information to designate specifiC.values of magnetic-field strength that may be hazardous to human:health, for two reasons.

a. There is insufficient evidence from the humanepidemiology data and from animal/cell experiments toestablish cause-and-effect relationships between low­frequency electric and magnetic field exposure andhuman health effects and cancer.

b. The precise nature of the environmental lowfrequency electric and magnetic field potentiallyrelated to human disease remains to be elucidated. Inaddition to field strength, parameters such as thetime-varying nature of the magnetic fields and therelevant time/exposure parameters need to badetermined.

5

The Subcommittee also wishss to express two specific ;:::.~:".'recommendations that in its view follow inescapably from ~~e .scientific recommendations.

POLICY RECOMMENDATION #1:, The Subcommittee is unan~~::~sin its belief that the question of electric and magnetic fieljeffects on biological systems is important and exceptionallychallenging, and that the Subcommittee's advice to the EPA sr.ouljbe that the report should be rewritten by EPA, and then re­reviewed by the Science Advisory Board.

POLICY RECOMMENDATION #2: EPA should complete' itsefforts with regard to RF electromagnetic fields (includingmicrowaves) and issue exposure guidelines independent of preser.tissues pertaining.to lower frequencies. The current EPA repor~inadvertently leads even the careful reader to conclude that thepotential carcinogenicity of electric and magnetic fields of £L:(i.e., powerline) frequencies is the only--or at least theprincipal--subject of concern with regard to nonionizing fields.Such a conclusion would reinforce the skewed and somewhatsensationalized picture presented to the public in recent yearsby the news media and government agencies responding to thispublicity. The report should therefore declare explicitly thatthe attention given to nonionizing electric and magnetic fieldsderives in the first place from long-standing concern over thehazards of RF (including microwave) radiation. EPA has expendedsubstantial resources on the study of such radiation over aperiod dating back to the EPA's inception, and EPA shouldcomplQte its efforts directed toward the issuance of RF exposureguidelines. RF fields present long-known and well-understoodhazards such as temperature elevation in tissue and heat stressresulting from acute exposures against which users and thegeneral public must be warned and protected. Any publishedexposure guideline should specifically identify the hazards fromRFexposure.

6

The Science Advisory Soard appreciates the oppor~~~i:: ::review issues ot this importance and looks forNard to a H~:::2~response from the Aqency concerning its schedule for revis~~;:~aEvaluation 0: the Potential Carcinogenicity of ;lec~ric a~dMagnetic Fields and it's availability for subsequen~ Scie~=eAdvisory Board review.

Sincerely,

~o.f~rmanExecutive co~teeScience Ad~Ory. Board

~.<~C"Oddvar F. Nyqaard, ChairmanRadiation Advisory CommitteeScience Advisory Board

~

....... , --"0. 1_' ",

~f"enevieve M. ~tan6ski, ChairmanNonionizinq Electric and~aqne~ic Fields SUbcommittee

Radiation Advisory Committee

ei

Dr. ~~V~ Sates, Vice Chai~anNoni9hiz~nq Electric andMagnetic Fields Subcommittee

Radiation Advisory Committee

7

UnIta4 St.telIEnwonmen~Protection Arney

sa-nce AdWlM'fSoud lAo 101 I

EP"..s~8~;: 3.2 .J'.!~a""¥"1 . 19~

&EPA AN SAB REPORT:POTENTIALCARCINOGENICITY OFELECTRIC AND MAGNETICFIELDS

REVIEW OF THE ORO'SCABCINOGENJCLIY OElliCI.ROMAGNETJCS EIELDSBY THE RADIATIONADVISORY COMMITTEE'SNONIONIZING ELECTRICAND MAGNETIC FIELDSSUBCOMMITTEE

NOTICE

This report has been written as a part 9f the activities ~:the Science Advisory Board, a public advisory group providingextramural scientific information and advice to the Administr~t=:and other officials of the Environmental Protection Agency. :~eBoard is structured to provide a balanced, expert assessment ofscientific matters related to problems facing the Agency. Thisreport has not been reviewed for approval by the Agency; hence,the comments of this report do not necessarily represent theviews and policies of the Environmental Protection Agency or ofother Federal agencies. Any mention of trade names or commercialproducts does not constitute endorsement or recommendation foruse.

i

ABSTRACT

This review constitutes comments by an EnvironmentalProtection Agency (EPA) appointed subcommittee of seventeenexperts in twelve disciplines to review a draft version of ~?A'5report Evaluation of the Potential Carcinogenicity ofElectromagnetic Fields (EPA/600/6-90/005B)-. The reviewerssuggest numerous changes in emphasis, coverage, and wording;comment on some policy considerations: and conclude that thedraft report ir.affect will have to be rewritten if all of thesesuggestions and comments are to be taken into account. TheSubcommittee also presents its conclusions on the substantivescientific questions discussed in the EPA report.

Keywords: electri~, magnetic, eleqtromagn~tic, cancer

ii

August 2, 1991

U. S. ENVIRONMENTAL PROTECTION AGENCYSCIENCE ADVISORY BOARD

RADIATION ADVISORY COMMITTEENONIONIZING ELECTRIC AND MAGNETIC F~ELDS SUBCOMMITTEE

ROSTER

Chairman

Dr. Genevieve M. Matanoski, The Johns Hopkins UniversitySchool of HY1iene and Public Health,Saltimore, Maryland

vice Chairman

Dr. David V. Bates, Vancouver, CANADA

MEMBERS

Dr. A. Karim Ahmed, Princeton, New Jersey

Dr. Patricia A.Buttler, University of Texas HealthCenter in Houston, Houston, Texas

Dr. Craig V. Byus, University of california, Riverside,California

Dr. Kelly H. Clifton, University of Wisconsin ClinicalCancer Center, Madison, Wisconsin

Or. John DiGiovanni, University of Texas, M. D. AndersonCancer Center, Smithville, Texas

Mr. William E. Feero, Electric Research and Management,State College, PA 16804

,Dr. Robert L. Harris, School of PUblic Health,

University of North Carolina, Chapel Hill, North Carolina

Dr. Clark W. Heath, American Cancer Society, Atlanta, Georgia

Dr. Nan M. Laird, Harvard School of Public HealthBoston, Massachusetts

Dr. M. Granger Morgan, Carnegie-Mellon University,Pittsburgh, Pennsylvania

iii

Dr. Mary Ellen O'Connor, University of Tulsa,Tulsa, Oklahc~a

Dr. Donald A. Pierce, oregon State University, Corvallis, C~eg:~

Dr. charles Susskind, College of Engineering,University of California, Berkeley, California

Dr. Bary W. Wilson, Battelle Pacific Northwest LaboratoryRichland, Washington

Dr. Richard Wilson, Department of PhysicsHarvard University, Cambridge, Massachusetts

DESIGNATED FEDERAL OFFICIAL

Mrs. Kathleen W. ConwayScience Advisory BoardU.S. Environmental Protection Agency401 M street, S.W., A-101FWashington, D.C. 20460

STAfF SECRETARY

Mrs. Dorothy M. ClarkSecretary, Science Advisory BoardU.S. Environmental Protection Agency401 M street, S.W., A-la1FWashington, D.C. 20460

;"'1

TABLE O~ CONTENTS

Executive Summary . .

Background • • • • • 6

1.0

2.0 Introduction

2.1

. .. . .

. . .

.......· . . . .

. 1

. 6

2.2 Charge . .......... 8

3.0 Review of the Document . . · · .. · · ·3.1 Response to the Charge . · · · · · ·3.2 Comments on the Executive Summary

3.3 Comments on Chapter 2 · · · · · · ·3.4 Comments on Chapter 3 · · · · · · ·3.5 Comments on Chapters 4 and 5

4.0 Additional Comments on the Science · ·5.0 POlicy Recommendations . . · · · · ·

• . .10

· • 10

.12

· .12

· .16

.19

.24

.26

Detailed Technical Comments

APPENDICES

Appendix A:

Appendix B:

Charge •

DISTRIBUTION LIST

'I

III,III!

I\IIIII

III\\III

\I\I

\

1. 0 EXECUTIVE SUMMARY

The charge to the Subcommittee was to review the docu~e~: ~3

to the accuracy and completeness of the information provided, ~5well as the interpretation of the scientific data. The EPAdocument has serious deficiencies and needs to be rewritten. Asthe result of many internal inconsistencies, it is oftendifficult to tell what the EPA's position is when reading t~eEvaluation of the Potential carcinogenicity of Electric andMagnetic Fields. Portions of the draft lead to conclusions N~:=~are not the conclusions stated elsewhere in the document. Suc~inconsistencies appear not only between the body and theexecutive summary, but also between different pages of thedocument itself.

Consequently, "the document requires a logical reorganizationand complete rewriting with particular attention to careful andprecise use of language. A simple editing of the present textwould not be sufficient. The Subcommittee consciously refrainedfrom providing a list of particular inconsistencies because itdoes not want to mislead the EPA into believing that editingalone will address the Subcommittee's concerns.

Six specific questions were posed in the charge to theSubcommittee. Before responding to these questions explicitly,the Subcommittee feels it important to express its viewpoint onthree underlying scientific issues, concerning (a) epidemiology,(b) biological effects, and (c) carcinogenicity. These issuesare critical for the understanding of scientific issues inresearch regarding electric and magnetic fields and must beaddressed in any future EPA discussion of the potentialcarcinogenicity of electric and magnetic fields.

a) The SUbcommittee has concluded that some of theepidemiological evidence is suggestive of anassociation between surrogate measurements of magnetic­field exposure and certain cancer outcomes. In suchstudies, the existence of confounders is always aPQ~sibility, but since no common confounder has yetbeen identified, the existing evidence cannot bedismissed. In the absence of much better exposure'information and an understanding of which exposures aresignificant, no precise exposure-response relationshiphas yet been adduced. This lack, together with limitedunderstanding of possible biological mechanisms,prevents the inference of cancer causality from theseassociations at this ti~e.

b. The Subcommittee accepts that effects on somebiological systems have been shown to occur at moderatefield intensities. (An example of such effects is thewell-documented work on phosphenes.) However, theevidence for effects at very low field strengths is notso widely accepted. Even if effects on living systems

to thermal effects. Nonthermal biological effects also h:'/e =~e~reported in some animals exposed to RF fields. It is wellestablished that some in vitro and in vivo experiments on ELFhave shown nonthermal biologic effects at fields of moderateintensity, and there are some suggestive effects at lower levels.Furthermore, ELF-modulated RF fields assumed to be nonthermal =a~produce many of the biological effects of 'ELF fields alone.Hypothetical constructs relating observed biological effects :~possible health effects (specifically, increased cancer risk)have been delineated. However, there are at present insufficien~data on many of the critical steps in the linkage to infercausality on the basis of animal or cellular data.

3. Has the Agency properly evaluated the way in which t~efindings on biological effects and field-tissueinteraction,mechanisms affect the interpretation of thehuman studies?

No, the EPA has not evaluated how the findings on biologicaleffects and nonionizing field and tissue interaction mechanismsrelate to the interpretation of human studies. The strength ofepidemiologic data depends on identification of supporting ,evidence from in vitro and in vivo data. This relationship has'not been developed in the report. The critiques of studies ofbiological effects are contained in the discussions of theseveral chapters.,

4. Is the choice of topics in Chapter 5 appropriate and isthe interpretation of the biological effects literatureas it relates to carcinogenesis supported by theavailable information?

The heading of chapter 5, "supporting Evidence ofCarcinogenicity," is inappropriate. The interpretation of thebiological effects in the in vitro systems as presented by thereport does not make a case for carcinogenicity. TheSubcommittee found a lack of balance in the analysis andpresentation of evidence in this chapter. Specific individualexperiments.need critical review.

5. Is the Agency's carcinogen classification systemapplicable to electromagnetic fields?

Nonionizinq electromagnetic fields should not be classifiedunder EPA's chemical carcinogenesis system because of presentmajor uncertainties. These involve an incomplete understandingof which aspects of field-tissue interactions give rise tobiological effects. Properties of the various electric andmagnetic fields such as phase angle, polarization, transients,and frequency range may contribute to different biologicaleffects. For these reasons, the use of EPA's classificationscheme at this time would be inappropriate and confusing.

6. Does the information cited in the document suppc=~ :~~conclusion that there 'is not enough information :~designate specific values of magnetic field stre~;:~ 35beinq hazardous to human health?

Yes, there is insufficient information to designate speci::cvalues of magnetic-field strength that may be hazardous to hu~anhealth, for two reasons.

a. There is insufficient evidence from the humanepidemiology data and from animal/cell experiments toestablish cause-and-effect relationships between lo~frequency electric and magnetic field exposure andhuman h~alth effects and cancer.

b. The precise nature of the environmental low frequencyelectric and magnetic field potentially related tohuman disease remains to be elucidated. In addition tofield strength, parameters such as the time-varyingnature of the magnetic fields and the relevanttime/exposure parameters need to be determined.

*The Subcommittee also wishes to express two specific policy

recommendations that in its view follow inescapably from thescientific recommendations.

POLICY RECOMMENDATION #1: The Subcommittee is unanimous inits belief that the question of electric and magnetic fieldeffects on biological systems is important and.exceptionallychallenging, and that the Subcommittee's advice to the EPA shouldbe that the report should be rewritten by EPA, and then re­reviewed by the Science Advisory Board.

POLICY RECOMMENDATION #2: EPA should complete itsefforts witq regard to RF electromagnetic fields (includingmicrowaves) and issue exposure guidelines independent of presentissues pertaining, to lower frequencies. The current EPA reportinadvertently leads even the careful reader to conclude that thepotential carcinogenicity of electric and magnetic fields of ELF(i.e., powerline) frequencies is the only--or at least theprincipal--subject of concern ~ith regard to nonioniz~ng fields.Such a conclusion would reinforce the skewed and somewhatsensationalized picture presented to the public in recent yearsby the news media and government agencies responding to thispublicity. The report should therefore declare explicitly thatthe attention given to nonionizing electric and magnetic fieldsderives in the first place from long-standing concern over thehazards of RF (including microwave) radiation. EPA has expended

substantial resources on the study of such radiation over l

period datinq back to the EPA's inception, and EPA shouldcomplete its efforts directed toward the issuance of RF exp~5_~~guidelines. RF fields present long-known and well-underst~~jhazards such as temperature elevation in tissue and heat s~=e55resulting from acute exposures against which users and thegeneral public must be warned and protected. Any publishedexposure guideline should specifically identify the hazards :=:~RF exposure.

:;

2.0 INTRODUCTION

2.1 Backqround

At the request of the Environmental Protection Agency'sOffice ot Radiation Programs, the Office of Health andEnvironmental Assessment prepared Evaluation of the PotentialCarcinogenicity of Electromagnetic Fields. In January 1990, E?Astaff requested orally that the Science Advisory Board (SAB)review this document. At its next meetinq, in May 1990, theSAB's Radiation Advisory Committee (RAC) responded byestablishing a Nonionizing Electric and Magnetic FieldsSubcommittee under the chairmanship of RAe member Dr. GenevieveMatanoski. An earlier draft of the Evaluation was reviewed enJune 23, 1990, by a panel chaired by Dr. Richard Griesemer of theNational Institute-of Environmental Health Sciences: this reviewwas not a Science Advisory Board review.

After wide consultation and consideration of more than 250scientists, the Director of the Science Advisory Board selectedseventeen members for the Nonionizing Electric and MagneticFields Subcommittee:

Dr. A. Karim Ahmed, Princeton, New JerseyDr. David Bates,* University of British ColumbiaDr. patricia A. BUffler, University of Texas Health Center

in Houston, School of Public HealthDr. Craig V. Byus, Biomedical Sciences and Biochemistry,

University of Calitornia-RiversideDr. Kelly H. Clifton, Department of Human oncology andRadiology, University of Wisconsin Clinical Cancer CenterDr. John DiGiovanni, Department of Carcinogenesis,

M.D. Anderson Cancer CenterMr. William E. Feero, Electric Research and Management,

State College, PADr. Robert Harris, Department of Environmental Science and

Engineerinq, School of Public Health,University of North Carolina

Dr. Clark He~th, American Cancer SocietyOr. Nan Laird, Department of Biostatistics, Harvard School of

Public HealthDr. Genevieve Matanoski,** School of Hygiene and Public Health

The Johns Hopkins UniversityDr. M. Granger Morgan, Department of Engineering and PublicPolicy, carnegie-Mellon UniversityDr. Mary Ellen O'Connor, Psychology Department, University

of TulsaDr. Donald Pierce, Department of Statistics, Oregon State

UniversityDr. Charles Susskind, College of Engineering,

University of California-BerkeleyDr. Bary Wilson, Battelle Pacific Northwest Laboratory

6

Dr. Richard Wilson, Department of Physics, Harvard Univers._1

*Vice Chairman of the Nonionizing Electric and Magnet:cFields Subcommittee.

**Chairman of the Nonionizing Electric and Magnetic FieljsSubcommittee.

On October 12, 1990, the Directors of"the Office of Healthand Environmental Assessment, Health Research, and Radiationprograms formally requested SAB review of EPA's Evaluation of t~ePotential Carcinogenicity of Electromagnetic fields and a relatedresearch agenda. (This memorandum is Appendix A.) Thememorandum contained the charge to the Subcommittee. (The chargealso appears in Section 2.2)

At its October 23-24, 1990 meeting, the Executive Committeeaccepted a workplan for the SAB which included four Fiscal Year1991 meetinqs of the Subcommittee to undertake and complete itsreviews of the carcinogenicity and research agenda documents.

The Federal Register published a notice December 18, 1990,announcinq both the availability of the Evaluation of thePotential Carcinogenicity of Electromagnetic Fields and the firs~meeting of the Subcommittee. At the first meeting, January 14- .16, 1991, in Washington, D.C., EPA staff and contractors briefedthe Subcommittee on the document, the public provided more than aday of oral comment, and time was allotted for Subcommitteedis~ussion. Before adjourning, the Subcommittee formed intothree groups which would prepare papers for consideration by theSubcommittee at its next meeting. These groups were:Physics: Mr. Feero, Dr. Susskind,* Dr. Richard WilsonBiology: Dr. Ahmed, Dr. Byus, Dr. Clifton, Dr. DiGiovanni,

Dr. O'Connor, Dr. Bary Wilson*Epidemiology: Dr. Bates, Dr. Buffler, Dr. Harris, Dr. Heath,*

Dr. Laird, Dr. Pierce* Authored paper after discussion with group.

The Subcommittee met for the second time April 12-13, 1991,in San Antonio, Texas. The Subcommittee considered three grouppapers, hea~d three invited speakers (Dr. Russel Reiter of theUniversity of Texas Health Science center at San Antonio, Dr.Asher Sheppard of the Pettis Memorial Veterans Hospital in LomaLinda, California, and Dr. Arthur Pilla of Mt. Sinai School ofMedicine in New York City), listened to oral comment from threemembers of the public, and appointed a writing group to prepare adraft Subcommittee report for consideration at the Subcommittee'sJuly 23-25, 1991, meeting in Washington, DC. (This July meetingalso began the Subcommittee's review of the research agenda.)

During the course of its review, the Subcommittee receivedalmost a thousand pages of written public comment from aboutthree dozen individuals and organizations. The SUbcommitteelistened to oral comment from about 40 individuals. There was

..,I

some overlap between those providing oral and written pub:~:comments.

The Subcommittee edited this report at the July 23-25, ~J):,meeting, subsequently approved it by mail, and forwarded ic ~~the Radiation Advisory Committee. The Radiation AdvisoryCommittee addressed the Subcommittee report at its Septembe~ :3­20, 1991, public meeting and forwarded it to the SAB's Exec~ti'/eCommittee to be considered at its October 29-30, 1991, publicmeeting. After approval by the Executive Committee and minereditorial corrections as suggested by the Executive Committee,the report was transmitted to the Administrator of theEnvironmental Protection Agency.

2.2 Charge to the Subcommittee

The Agency seeks the advice of the Board on theaccuracy and completeness of the entire document and onthe question of whether the interpretation of theavailable information reflects current scientificopinion. In addition, we would like the Board toaddress the following specific issues:

1. Is the interpretation of the human andanimal evidence of carcinogenicity supportedby the available information?

2. Does the anima~ or biological effectsinformation provide a basis for postulatingthat there is a human hazard from exposure toextremely low frequency fields or eithermodulated or unmodulated radiofrequencyradiation?

3. Has the Agency properly evaluated theway in which the findings on bioloqicaleffects and field-tissue interactionmechanisms affect the interpretation of theh~an studies?

4. Is the choice of topics in Chapter 5appropriate and is the interpretation of thebiological effects literature as it relatesto carcinogenesis supported by the availableinformation?

5. Is the Agency's carcinogenclassification system applicable toelectromagnetic fields?

6. Does the information cited in thedocument support the conclusion that there is

8

not enough information to designate specificvalues of magnetic field strength as beinghazardous to human health?

3 • 0 REVIEW OF THE OOCUMENT

3.1 Response to the Charge

The charge to the Subcommittee was to review the docu~e~: :5to the accuracy and completeness of the information provided, =5well as the interpretation of the scientific data. The EPAdocument has serious deficiencies and needs to be rewritten: -­is often difficult to tell what the EPA's position is whenreading the Evaluation due to many internal inconsistencies.Portions of the draft lead to conclusions which differ from :~econclusions stated elsewhere in the document. Suchinconsistencies appear not only between the body and theexecutive summary, but also between different pages of thedocument itself. .

Consequently, the document requires a logical reorganizationand complete rewriting with particular attention to careful andprecise use of language. A simple editing of the present textwould not be sufficient. The SUbcommittee consciously refrainedfrom providing a list of particular inconsistencies because itdoes not want to mislead the EPA into believing that editi~~alone will address the Subcommittee's concerns.

Six specific questions were posed in the charge to theSubcommittee. With respect to the Subcommittee's charge and thesix questions that were posed, the Subcommittee responds asfollows.

1. Is the interpretation of the human and animal evidenceof carcinogenicity supported by the availableinformation?

Currently available infornation is insufficient to concludethat the electric and magnetic fields are carcinogenic. Somehuman epidemiologic data report an association between surrogatesfor electric and magnetic field exposure and an increasedincidence of some types of cancer, but the conclusion ofcausality i~ currently inappropriate because of limited evidenceof an exposure-response relationship and the lack of a clearunderstanding of biologic plausibility.

2. Dce. the animal or biological effects informationprovide a basis for postulating that there is a humanhazard from exposure to extremely low-frequency fieldsor either modulated or unmodulated radiofrequencyradiation?

Some of the in vitro and in vivo data on unmodulated RF havesuggested the existence of mechanisms by which human healtheffects, but not carcinogenicity, ~ay be inferred. Bothunmodulated and extremely low frequency (ELF)-modulated

:0

radio frequency (RF) fields of sufficient intensity can gi','-2 _to thermal effects. Nonthermal biological effects also r.a·:e ::-::;:.-.reported in some animals exposed to RF fields. It is wel:established that some in vitro and in vivo experiments on ~:?have shown nonthermal biologic effects at fields of modera~eintensity, and there are some suggestive effects at lower :eve:s.Furthermore, ELF-modulated RF fields assumed to be nonthe~a: :~~produce many of the biological effects of ELF fields alone.Hypothetical constructs relating observed biological effects ~~possible health effects (specifically, increased cancer risk)have been delineated. However, there are at present insufficier.~data on many of the critical steps in the linkage to infercausality on the basis of animal or cellular data.

3. Has the,Agency properly evaluated the way in which ~~efindings on biological effects and field-tissueinteraction mechanisms affect the interpretation of t~ehuman studies?

No, the EPA has not evaluated how the findings on biologicaleffects and nonionizing field and tissue interaction mechanismsrelate to the interpretation of human studies. The strength of ~epidemiologic data depends on identi~ication of supportingevidence from in vitro and in vivQ data. This relationship hasnot been developed in the report. The critiques of studies ofbiological effects are contained in the discussions of theseveral chapters.

4. Is the choice of topics in Chapter 5 appropriate and :5the interpretation of the bioloqical effects literat~~eas it relates to carcinogenesis supported by theavailable information?

The heading of chapter 5, "Supportinq Evidence ofcarcinogenicity," is inappropriate. The interpretation of thebiological effects in the in vitrQ systems as presented by thereport does not make a case for carcinogenicity. TheSubcommittee found a lack of balance in the analysis andpresentation of evidence in this chapter. SpecifiC individualexperiments need critical review.

5. Is the Agency's carcinogen classification systemapplicable to electromagnetic fields?

Nonionizing electromagnetic fields should not be classifiedunder EPA's chemical carcinogenesis system because of presentmajor uncertainties. These involve an incomplete understandingof which aspects of field-tissue interactions give rise tobiological effects. Properties of the various electric andmagnetic fields such as phase angle, polarization, transients,and frequency range may contribute to different biological

11

II

!IIII11

I!,I,I,!III,;IIi

Ii

Ii\iI\

effects. For these reasons, th~use of EPA's classificati~~scheme at this time would be inappropriate and confusing.

6. Does the information cited in the document suppor~ ~~econclusion that there is not enough information todesignate specific values of magnetic field strength ~sbeing hazardous to human health?:

Yes, there is insufficient information to designate specif:cvalues of magnetic-field strength that may be hazardous to hu~anhealth, for two reasons.

a. There is insufficient evidence from the humanepidemiology data and from animal/cell experimentsto establish unequivocal cause-and-effectrelationships between low frequency electric andmagnetic field exposure and human health effectsand cancer.

.b. The precise nature of the environmental lowfrequency electric and magnetic field potentiallyrelated to human disease remains to be elucidated.In addition to field strength, parameters such asthe time-varying nature of the magnetic fields andthe relevant time/exposure parameters need to bedetermined.

3.2 Comments on the Executive Summary

The executive summary of Potential carcinogenicity ofElectromagnetic Fields is not adequate. Special care should betaken to ensure that it reflects the contents of the revisedreport exactly, without drawing any conclusions not substantiatedby the body of the report. Not a few readers (and most newsmedia) are sure to give only cursory attention to the full 366­page text and to depend for the gist of it on the executivesummary, so special car~ .m..':l.~.~--.~~-.~~J~~J1..t.q...ma_ke_it _readable.and

--..-...---.....--accurate·.. _.--------.-------

3.3 Comments on Chapter 2

3.3.1 Blectric and Maqnetic Fields and Mechanisms forBiological Interac~ions

Chapter 2 attempts to do four things: describe the physicalcharacteristics of electromagnetic fields; explain how fieldscouple with the body; quantify ambient exposures; and discussmechanisms of biOlogical interactions. A comprehensive treatmentof these four areas would constitute a book-length text. Theauthors have compounded their difficulties by attempting to makethe presentation relevant to the entire range of frequencies,

12

from ELF to microwaves. Trying to present all the above :~ ~~=3

than forty pages was destined to result in a superficial a~jspotty treatment. The authors should reduce their scope :0 ~more manageable level by emphasizing the lower frequencies, ::~which quasistatic analysis is appropriate. The report shouljattempt to teach field theory. Instead, it should presentmaterial toward the physical understanding needed in laterchapters. Simplified but logical and correct relationships ~us:be drawn from the large body of knowledge that encompasses ~~etheory of electricity and magnetism, together with the second:~~of thermodynamics. In particular, relationships of charges,forces, motion, and time rates of change should be presented :~rsimple situations. It is an assumption, not yet disproved, ~hatthey also apply to more complex (biological) situations. Thedifficulties lie in calculating such complex situations andknowing the boundary conditions. .

The draft report separates electric and magnetic fieldsaccording to frequency, but the way in which these fields arelikely to differ is poorly described. The following should behighlighted.

a) Ionizing frequencies (e.g., gamma and x radiation>: theproduct hf (Planck's constant times frequency) is abovethe ionization threshold, so that ionization anddestruction of a simple cell are possible (even at lowintensity, although with correspondingly lowerprobability).

b) High nonionizinq frequencies (e.g., microwave and otherradio frequency radiation): hf is below threshold and noionization takes place, but both electric and magneticfields penetrate insulatinq and partially conductingbodies and can produce heating,. as do microwave ovens.

c) Extremely low frequencies (e.g., powerlinefrequencies): electric fields are stronqly attenuatedin partially conducting bodies. The electric field Ein.such bodies derives primarily from changes in themagnetic field B, according to the equationcurl E - - ~B/at, or in words, the spatial rate ofchanqe of the vector E equals the negative of the ti~erate of change of the vector B, so that it is not alonethe magnitude of the magnetic field that is importanc,but also the rate at which it changes with time.

The draft report lacks a discussion of the actual parametersof the total electromagnetic fields that are likely to beimportant for subsequent chapters. There is no recognition thatthe effects on a system of charges (inclUding the human body) arein force~, as summarized by the Lorenz force law

13

IiI

\i

IIIiII1iII\I1IIi

\I

I1!

F ~ q(! + v x B), where v is th~ velocity vector and q is :~~charge; or in words, that the force consists of two compor.e~:3:(1) the product of charge and electric field, and (2) the ~r:d~=:of charge and velocity times the magnetic field and sine of :~eangle between them. This point is not made in the draft repc=:.There are many implications.

Under some circumstances charges in some bodies actcoherently; this provides one potentially promising route =~rinvestigating mechanisms of biological interactions with electr:=and magnetic fields. Another route for inv~stigation is theexistence in biological tissues of permanent magnets, such asthose found in some soil microorganism and in honey bees. Insome circumstances these magnetic structures appear important ~~livinq organisms; .however, there has not been systematicinvestigation of the prevalence of such structures or theirimportance.

3.3.2 Exposure

In the discussion of epidemiology, all mention of exposureis couched in the vaque terms of "electromagnetic fields" or even"magnetic fields." The EPA's Human Health Assessment Group .normally relates cancer incidence to a long-term average of someexposure parameter. Although that may be appropriate for mostcancer-inducing agents, the averaging time may well be shorterfor an agent that increases cancer in a new way. However, ifthere are risks, it is not clear which parameters of electric ormagnetic fields will be the important ones. It is appropriatefer the report to focus on the time-integrated exposure (or dose)metric, but it is important as well that other parameters beadequately treated.

Which critical parameter of the magnetic-field intensity Hshould be discussed? It could be dH/dt (the time rate of changeof H), or dH/dt above some threshold applied randomly or in some~articular sequence over some period of time. It could be)H(t) dt (the integral or sum of H over a time t), as implied by'~he draft report: or the amount above a threshold Hc'j~H(t) - Hel dt: _~ the amou~ with a saturation Hs' ]H(t) dt(for H < Kg> an~~{t) dt + -1Hs dt (for H > Hs)' It may not befeasible to be all-inclusive, but some discussion of theplausible possibilities is reco~~ended.

The report also fails to discuss background fields otherthan fluctuations. If we accept there is no new force, theexposure parameter is E + v x B: and inside the body,curl E = -~B/at. One.of the principal contributions tobackground exposure comes as we move with a velocity v throughthe earth's magnetic field B. for many of us, this component isvastly increased when v is the velocity of an aircraft. Yet thedraft report is written as if the only background of importance

14

were the ambient 60-Hz field. That is clearly wrong ar.d 3~:._:be corrected. And of ;ourse this varying background is ~ 5:_;:~of confusion in any epidemiological study in ways tha~ the ~~1~~

report fails to discuss.

3.3.3 Models

The rather extensive discussion of models in the repc~~should be reduced to a tabulation o~ the prevalent hypothes:2edmodels, with reference to when each was introduced and to Nr.~~extent attempts to explain experimental findings by each ha~ebeen successful. The strengths and weaknesses identified in ~~eliterature should be tabulated. For example, the cyclotron­resonance model has been criticized on two counts in a simplepaper by John Sandweiss (Bioelectromagnetics 11: 203-205, 1990):.(1) the cyclotron radius at the assumed field exceeds the size ~:the object, yet several revolutions inside are needed to get aresonance 1 and (2) the mean free path of an electron in themedium is so short that nQ resonance seems possible--a problem ~owhich reference is made, but whos~ fundamental nature is notbrought out.

3.3.4 Fluctuations

A graph on p. 2-15 of the draft report shows the currentdensities at which biological effects are expected, as well asthe noise fluctuations. There are other papers on fluctuations.A recent publication by Robert Adair in the Physical Review (A4J:1039-1048, 1991), which also served as the basis of hispresentation at the Subcommittee's 15 January 1991 meeting, is agood summary of the apparent points of conflict between thebiological effects attributed to weak electric and magneticfields and the constraints of known physical principles.

If a findinq is repeatable but not explainable by existingphysical theory, it must be clearly labeled as such.Hypothesized mechanisms in such cases are desirable and should bepresented. However, they must be discussed in terms of strengt~sand weakness~s in predicting the observed phenomena, and thepoints at which they do not fit present understanding. Forexample, the calculation of the~al noise in a single cell wouljprove to be inapplicable if it could be demonstrated that thebodies under consideration are not isolated cells butagglomerations of cells of substantially larger dimensions thanindividual cells, so that calculations based on single cellswould have to be revised1 or if ferromagnetic materials areinvolved. An additional point is that biological systems areinherently nonlinear in hierarchies that extend from molecules :~cells, to tissues, and finally to organs and orqan systems, so·that appropriate physical models must account ror nonlinearbehavior, as well as nonegyilibrium physical characteristics,commonly seen at the atomic level rather than the molecular.

, --:)

3.4 Comments on Chapter 3

3.4.1 Comments

The Subcommittee feels that the report's discussion ofepidemiologic findings is seriously deficient, given the ce~:~l~importance of such data in evaluating possible human healt~ ~:3~5

in this perplexing field. Extensive revisions are thereforgnecessary both in Chapter 3 and in the report's Summary ar.dConclusions. Although the EPA report achieves nearly comple:ecoverage of pertinent epidemiologic work published through :j39and properly approaches workplace exposures separately fromresidential exposures, the manner in which these data aredescribed and evaluated is inadequate in major respects.

a) In general, the report's review and analysis ofepidemiologic findings is unfocused and diffuse. :tswriting is often repetitive and imprecise, anddescriptions of data are frequently mingled withinterpretative comments. As a result, the report lackscohesiveness, is difficult to read, and loseseffectiveness in communicating its findinqs andconclusions.

b) The manner in which studies are reviewed is uneven.Some studies are clearly less substantial than others,yet they often receive equal or greater attention.Discussion of findings includes too much unwarrantedspeculation about causal interpretation. Often suchspeculation appears unbalanced, giving emphasis topositive findings while de-emphasizinq negative ones.

c) The critical importance of exposure asseSSMent does notreceive sufficient attention, particularly in relationto surrogate measures and to potentialmisclassification.

d) In evaluating ir~ividual studies, and then reachingcQnclusions for ~he field as a whole, the report failsto'focus coherently on the major epidemioloqic issuesthat ultimately shape any assessment of risk. Inaddition to exposure assessment, these issuesparticularly involve ability to discern exposure­response relationships, to make precise measurements c:risk levels, and to evaluate the potential influence ofconfounding variables. The report greatly needs toaddress these particular issues in an organized andcritical manner.

16

3.4.2 Recommendations

a) Chapter 3 should be extensively re-worked.

Major consideration should be given to fullpublished studies or studies in .which manuscripts havereceived peer review and are in press. Data presentedin abstract form, as letters to journals, or as caseseries anecdotes should be clearly labeled as such.Where other data deserve review (such as the recentwork by Peters et al.), their pre-pUblication natureshould be recognized.

The Subcommittee suggests focusing detaileddiscussion of findings on the several studies ofgreatest importance and, in those instances, perhapsnot exceeding three or four pages each. Other studies(abstracts, letters, etc.) could be summarized morebriefly. An appendix might be used it it seemsnecessary to include extensive detail.

b) Coverage of the literature should be extended through ~1990, and beyond as available, especially with respect:to the study by Peters et ale The 1989 Coleman et alestudy (Brit J. Cancer) deserves fuller consideration,as does information suggesting a role for trafficdensity as a confounding factor.

~) A succinct summary table displaying only key findings(Note: There is a "summary table" on pp. 3-127 to J­131 of the EPA's October 1990 document.) andlimitations in major studies may help to focus theassessment and might become a nucleus for formal meta­analysis of data. It may also be useful to constructgraphs that display odds ratios and confidenceintervals for various population/exposure comparisonsin major studies.

d) It, may be useful to compare data across studies inrelation to cancer site, e.g., risks observed forchildhood cancers as a whole as opposed to leukemia orcentral nervous system tumors in particular.

e) Throuqhout the report, special care should be taken toavoid gratuitous speculation or surmise which may favoreither negative or positive interpretations of databeyond the actual limits of those data. Theconservative i-~ent of the report can only benefit byevenhanded caU~lon and careful precision in itswording.

17

J. 4.3 Exposure Assessment -

A separate, expanded section should be developed address~~jissues of exposure assessment. Such a section shouldparticularly discuss the value and limitat~ons of exposuresurrogates, gradients, and models in relation to epidemiologicinvestigations.

a} Exposure Surrogates. The use of surrogates forexposures, particularly for past unmeasured exposures,is not uncommon in occupational epidemiologic studies.The use of wire codes in community studies, as done :nthe Wertheimer, Savitz, and Peters studies, isappropriate and defensible. The use of surrogates ~ayvery we~l result in less misclassification than occurswhen inherently variable spot present-day measurementsare used to represent unmeasured average exposures thatoccurred in past years.

b) Exposure Gradients. A gradient in exposures of studysubjects is necessary in environmental or occupationalepidemiologic studies. The choice of cut-points indichotomous or multistage exposure classifications mayinfluence the results of statistical analyses. Thismatter may deserve some summary comment relative toreported studies.

c} Exposure Models. Use of exposure models may besuperior to the direct use of exposure measurements f~rcurrent exposures. Exposure models based on availablehistoric information that" can be validated by use ofcurrent measurement data are particularly useful forstudies dealing with past unmeasured exposures. Wire.code surrogate data have been indexed to magnetic fieldlevels in a relatively unsophisticated approach to~odelinq. Progress is being made in this area (M. R.Flynn et al., "Validation of expert judqment inassessing occupational exposure to maqnetic fields intlleutility industry," Appl. occup. Environ. Hyg. 6:141-145, 1990). Exposure modelinq has been addressedin the course of the Peters study and one hopes thiswill be discussed in the forthcoming publication of thework.

3.4.4 criteria for Assessing Significance of Data

The revised report should summarize the epidemiologic andstatistical principles that govern the report's assessment ofepidemiologic data. Such a section should include bothmethodologic aspects and interpretive considerations and shouldapply those principles to the overall evaluation of epidemiologicevidence in this field.

Important methodologic topics useful for assessingindividual studies include how a study measures exposure ~ex~e~:of exposure misclassification), how subjects are selected t~insure comparability, how potential confounding variables areaddressed, choice of statistical measures, and consideratior.s :fstudy size and power. Criteria for interpretation of results inthe field as a whole include consistency ~mong different studies,strength of association, degree of evidence for exposure-rescor.serelationships, and potential clinical specificity. .

Also important for assessment of epidemiologic data isconsideration of the biological plausibility of epidemiologicfindings in relation to findings from other research discipli~es.Although the Subcommittee did not directly address this largerissue in detail, the Subcommittee does recognize the criticalimportance of such issues in reaching a valid assessment ofpossible cause-effect relationships between nonionizing-radiationexposures and human cancer. Such considerations will ultimatelyneed to balance the strengths and weaknesses of epidemiologicfindings in relation to the strengths and weaknesses of evidencein allied disciplines (experiments in animal toxicology, incellular and molecular systems, and in the physics of nonionizingradiation). :

3.5 Comments on Chapters 4 and 5

3.5.1 OVerview

Both chapters mix review and evaluation in an inappropriatemanner. Studies should first be summarized objectively. Theyshould then be evaluated in a discussion that covers both strengpoints and weak points.

Biological and potential health effects from ELF exposurenow constitute an active area of research. The report should beupdated to include data that have' become available since therelease of the present draft.

ELF and.RF data should be presented and discussed incompletely separate sections of the report, with comment andconclusions provided separately for each frequency range.However, where both ELF and RF are present (e.g., ELF-modulatedRF: ELF transmission lines carrying or picking up RF), such datashould be presented and discussed ~nder both headings.

The summary tables included in the chapters were helpful.Such tables should be expanded and updated to reflect recentresearch findings.

From the information presented in Chapters 4 and 5, it isclear that hypothetical constructs relating observed biologicaleffects to possible health effects (specifically, increased

19

cancer risk} can be delineated. However, there are ins~::~:. ;-:data on many of the critical steps in the linkage to infa~causality on the basis of animal or cellular data.

3.5.2 Comments on Chapter 4

This c~apter deals with biological effects observed :~animals exposed to electric and magnetic fields in the ELF :3~~eand to RF fields assumed to be nonthermal. ~

Taken as a whole, animal data reviewed and discussed i~ :~~schapter, together with data from more recent animal experi::e:;:s,.strongly suggest (and in some instances can be said todemonstrate) that ELF magnetic and electric fields are capab~eof eliciting bioleqical effects.

The distinction between biological effects and healtheffects is an important one. It is especially relevant in thecontext of the question of possible health effects from electricand magnetic field exposure. The report must be precise indistinguishing between these two concepts and carefully guardagainst the tacit assumption that the observation of one(biological effects) automatically implies that the other (healtheffects) exists.

In response to the specific charge to the Subcommittee tocomment on the value of animal data in interpretinq humanepidemiologic studies, the Subcommittee believes that the ani~aldata provided are not easily applied to the interpretation ofresults from human epidemiologic studies for the followingreasons:

a) There are few data from different laboratories thatdemonstrate consistent biological effects using thesame experimental protocols. However, newly publishedinformation is emerging that will address thisshortcoming.

b) There have been no lifetime animal studies in whichcarcinogenesis was specifically investigated as aconsequence of magnetic-field or electriC-fieldexposure.

c) With few exceptions, ani~al studies were not directedat testinq of a specific ~odel or hypothesis relevantto possible mechanisms of electric and magnetic fieldeffects in biological systems.

d) There is insufficient evidence of linear ormonotonically increasing exposure response from the .animal studies reviewed, and where response is linked

20

to exposure (or dose), the effects appear to be "~~~none" above some threshold.

e) Field intens ities used in the experiments revie'....ed -:~::-"/widely and are often above those commonly encoun~e=ejby humans.

3.5.3 Comments on Chapter 5:

The title of Chapter 5 should be changed. The studiesreviewed and cited should not be presented as "supportingevidence for carcinogenicity." The report, in fact, reviews ja~3that are relevant to and consistent with the hypothesis thatelectric and magnetic fields may increase cancer risk. Apositive lifetime carcinogenicity study would constitutesupportinq evidence. Such a study has not yet been done. Infact, this chapter deals with potential mechanisms of interacti~nof electromagnetic fields with biological systems. The revisedreport should make it clear that these cellular effects can onlyestablish relationships and mechanisms of interaction.

In view of the testimony to the Subcommittee regarding thetheoretical (mainly thermodynamic) constraints on biOlogicaleffects of low-frequency low-intensity fields, Chapter 5 shouldinclude a discussion of the underlying assumptions regardingsize, structure, and conductive and dielectric properties of thebiological systems (cells, organs, or whole animals) underconsideration. The lumped transmission line model for signaldetection in collections of cells as presented to theSubcommittee in testimony by Or. Arthur Pilla, for example, mayserve as such a set of assumptions. These should be contrastedwith ,the assumptions used by Dr. Robert Adair in his recentPhysical Review paper (A43: 1039-1048, 1991).

EPA should consider deleting discussion of data on electric.and magnetic field exposure of plant cells, but adding thevoluminous results available in connection with electric andmagnetic field stimulated bone repair.

,Not mentioned in this document is the lack of effects

reproducible among laboratories, and lack of agreement orunderstandinq of what constitutes exposure or dose. Until thereis progress in these two areas it will remain impossible tosummarize the results of this field in a manner that willapproach "a consensus viewpoint. II

A carefully worded statement regarding the difficulty ofsome of the assay techniques should be included in this chapter.Training in one aspect of biological assays may not qualify oneto conduct an assay on another system. Many laboratories havereported failure to replicate a result when in fact the .laboratory lacked the necessary expertise to conduct the assay

21

appropriately. Conversely, a laboratory may report pos it':' ';eresults because it lacks the necessary expertise to carry ~u:reliable experiments.

A number of studies are presented dealing with DNA/fieldinteraction. No effects were found on breakage of strands orrepair of damaged strands. There is no evidence that geneticmutations are induced by ELF electromagnetic fields. However,effects of ELF magnetic fields on gene transcription andtranslation have been reported.

EPA should carefully review epigenetic factors in cancerrisk. Such factors include possible effects ontranscription/translation, hormonal effects, and effects on theimmune system.

Fairly small fields have been reported to alter the flux ofcalcium ions across cell membranes. Although some of the invitro calcium efflux data are open to concern about experimentaldesign and physiological interpretations, these effects appearsto be characterized by "windows" in frequency and intensity.Although lower in amplitude, this alteration in calcium ion flux~appeared in the same direction as changes caused by known cancerpromoters. An important consequence of altered calcium fluxwould be a change in production of parathyroid hormone andornithine-decarboxylase.

Melatonin is a hormone with important regulatory functions.A number of experiments have shown correlation between breastcancer and decreased melatonin production. Exposure of rats to avariety of low-strength electric and magnetic fields has beenshown to decrease melatonin levels in blood and pineal gland.

The melatonin data are an important reason for recognizingthe existence of bioloqical effects due to electric and magneticfields. It is not clear, however, that melatonin changes lead tohealth effects. Although the phenomena are well described in thereport, they should not be presented as "supporting evidence ofcarcinoqenic~ty.1I More recent data on breast cancer and prostatecancer in males should be included in the melatonin discussion,because increased risk for these specific cancers washypothesized on the basis of the melatonin findings.

References that are missing include the Gabriel et ale paperand the Meltz references that were supplied to the Subcommitteeby Dr. Martin L. Meltz. This chapter (or the mechanism chapter)should include mention of the work·of Dr. James C. Weaver andDr. R. Dean Astumian. The work of C. A. L. Bassett and others onpulsed electromagnetic field (PEMF) treatment of bone is alsopertinent (see, for example, C. T. Brighton and S. R. pollack,Eds., Electromagnetism in Medicine and Biology, 1991).

22

In summary, the Subcommittee believes the following:

a) The EPA's interpretation of the human and animalevidence of carcinogenicity is not supported by dat3available in the literature.

b) Biological effects and health effects must be clea~:ydistinguished, and although there may now behypothetical constructs for linking observed biologi=a~effects to cancer risk, there is insufficient basisfrom animal and cellular data tor postulating humancancer risk from exposure to ELF electric and magneti=fields.

c) In part.because of insufficient data, EPA has notproperly evaluated the ways in which proposedbiological mechanisms of electric and magnetic fieldsmay affect interpretation of human data. However, themelatonin findings as they relate to male breast andprostate cancer, and other work on DNA transcriptionleading to enhanced growth, may constitute findingsfrom which biological mechanisms to assist ininterpretation of human data may eventually bedeveloped.

d) The title for Chapter 5 is misleading and should bechanged; althouqh the topics discussed in Chapter 5 arerelevant to the issue, the underlying assumption ofthe existence of increased risk is disturbing.

e) The data presented in Chapter 5 do not constitutesupportive evidence for carcinogenic effects ofelectric and magnetic fields in humans and animals.

23

4.0 ADDITIONAL COMMENTS ON THE SCIENCE

The Subcommittee's mandate was to review the EPA repor~ ~~janswer specific questions in relation to it. In view of theopportunity that the Subcommittee had during its deliberations ~~hear testimony and discussions regarding the issue of any riskassociated with ELF, in the brief sections that follow, theSubcommittee gives a summary of its collective response to ~hatit has identified as substantive questions. Until these questionshave been answered, no opinion can be advanced on the possibili~yof a relationship between ELF and human disease.

The Subcommittee 'of 17 individuals represented expertise intwelve disciplinary areas: biostatistics, engineering,epidemiology, exp&rimental biology, exposure assessment,medicine, neuro-endocrinology, physics, physiological psychology,radiation oncology and carcinogenesis, risk analysis andtoxicology. The breadth of opinion and diversity of view withinthe Subcommittee would likely be replicated by any other groupwith a correspondingly wide spectrum of expertise.

The Subcommittee feels it important to express its viewpointon three underlying scientific issues, concerninq (a) .epidemiology, (b) bioloqical effects, and (c) carcinogenicity.These issues are critical for the understanding of scientificissues in research reqardinq electric and maqnetic fields andmust be addressed in any future EPA discussion of the potentialcarcinogenicity of electric and magnetic fields. .

a) The Subcommittee has concluded that the epidemiologicalevidence is suqgestive of an association betweensurrogate measurements of magnetic-field exposure andcertain cancer outcomes. In such studies, theexistence of confounders is always a possibility, butsince no common confounder has yet been identified, theexisting evidence cannot be dismissed. In the absenceof much better exposure information and anunderstanding of which exposures are significant, nop~~cise exposure-response relationship has yet beenadduced. This lack, together with limitedunderstanding of possible biological mechanisms,prevents the inference of cancer causality from theseassociations at this ti~e.

b) The Subcommittee accepts that effects on biologicalsystems have been shown to occur at moderate fieldintensities. (An example of such effects is the well-'documented work on phosphenes.) However, the evidencefor effects at very low field strenqths is not sowidely accepted. Even if effects on living systems atlower fields do occur, the assumptions leading toestimations of physical constraints for effects on

c)

isolated small spherical cells without ferromag~~:.:structures may not be applicable to larger cel:s ::cell systems such as neurons or neuronal networ~s.Many intervening steps must be clarified before ~~ebiological phenomena so far shown can be taken asdirect evidence of health impairment or carcinoge~e5~5in the human.

The EPA document does not present a holistic model ::carcinogenesis within which the strength of exist:~gevidence concerning the carcinogenic properties ofelectric and magnetic fields can be assessed. Therevised document should do so. Low-frequency elec~r:~and magnetic fields do not carry enough energy to ca~semutations directly. The Subcommittee recognizes tha~the incidence of cancer might well be affected by anagent that does not produce mutations. The knowninfluence of factors such as hormonal imbalance andnutrition on cancer promotion is an example ofepigenetic effect. .

25

5.0 policy Recommendations

The Subcommittee also wishes to express two specific ;:0 ~:. ::':recommendations that in its view follow inescapably from t~e ­scientific recommendations.

POLICY RECOMMENDATION #1: The Subcommittee is unanimo~5 :~its belief that the question of electric and magnetic fieldeffects on biological systems is important and exceptionallychallenging, and that the Subcommittee's advice to the EPA shouljbe that the report should be rewritten by EPA, and then re­reviewed by the Science Advisory Board.

POLICY RECO~NDATION #2: EPA should complete itsefforts with regard to RF electromagnetic fields (includingmicrowaves) and issue exposure guidelines independent of presen~issues pertaining to lower frequencies. The current EPA reportinadvertently leads even the careful reade~ to conclude that thepotential carcinogenicity of electric and magnetic fields of ELF(i.e., powerline) frequencies is the only--or at least theprincipal--subject of concern with regard to nonionizing fields.Such a conclusion would reinforce the skewed and somewhatsensationalized picture presented to the public in recent yearsby the news media and government agencies responding to thispublicity. The report should therefore declare explicitly thatthe attention given to nonionizing electric and magnetic fieldsderives in the first place from long-standing concern over thehazards of RF (including microwave) radiation. EPA has expendedsubstantial resources on the study of such radiation over a .period dating back to the EPA's inception, and EPA shouldcomplete its efforts directed toward the issuance of RF exposureguidelines. RF fields present long-known and well-understoodhazards such as temperature elevation in tissue and heat stressresulting from acute exposuresagainst which users and the generalpublic must be warned and protected. Any published exposureguideline should specifically identify the hazards from RFexposure ..

26

-. ~ . -.. -.-UNITED STATE:S ENVIRONMENTAL ?ROTECTrON AGE.'K·f

WASHINGTON, O.C. 20460

f"t'¥'f' ., .. - ..IoA.II I~ .~~J Ol"r'e£ :::I'"

~[S( .....qC:H ol. .... O :JE:V!:\..::P ....i; ....-:-

IM!MOp..;NOU'M

.....~

" .'~

StJBJ::CT:

FROM:

Seienee Advisory Board Review ot the EPA Oocument on~~. Careino~enioity of Ele~t:omagnltie ?ial- ,

W~lliam K. ?arland, Ph.C. .~01reetor ~~I _ _Ottica of Health and !nvironmental Assessmeneattica of Researe~han Oavel,pment (RD-j.S9)

Ken Sexton, Se.D.Director..Offic:. of Health '~I h V It 4attic. at Raseareh /~ Yr~~.dp. nt (ltD-saJ)

R~ehard. Gui:lloncl f. / ~/. > /tIlAA~1Clrec:tor I' ~~ ~-. -( Iottic:a of Radiation Proqrusoteici of Air and. Radiation (ANR-4S8)

TO: ~athlaen C~nway, oro, Raeliation Aclvisory Commie:.esc:iene. Advisory Soard. (A-lOll)

•"'J,.• oS

. w. are hlrGY rlquasJ:inq Seiene. Aelvisory !carel (SU) =aview~t t·"e d.ratt EPA d.CC\DIents. The first, -t:valuationof thepotlntial'carcinoqenic:ity ot Ilectrosaqnatie P1Ilds·, ~.pr.par~ ~y the Of tic. ot Ha.lea ~ Inv1ronaental As.assalnt at1:.";_ requ •• e of the Of~ic:e af lla4iaeioft Proc;rus (aU).. the'51c:on4 4~~ .Bua&n ZXposur. to Illctreaaqnatic: 11.14s:A~ation&l le','n:b 14aM.- was prep&red. bY' the ott ice o~ Kaaltha.saarca &~ •• t at the Administrato: as & c:o.p_nioft toea. tirs1: ~ne. Iceh d.OC'U:Ilanes'"are prepared. &net r.viavlc1 ~yscianeisu iJI. the OU and. the oeeic:. 0: R•• larch aM. Oevelopclne(ORD)• !cell 40caents will b. available to the SAl in u=lyNovlaCer.

~. eirse d.oc:umeneis schedul~ fer release eo the ~ublic i~larly NOVIBer a. &ft axeemal review d.:"att. fhe Aqanc:y 1s~l&nninq eor a &O-d&y p~lic: co.-ant pari04 and a ec~entSci.nee Ad:"r1scry-!o&~ review-. ,- Jl:t-e.r1ier <iratt h•• alrM<1y- ~In-

reviewed by a panel of experts outside the Agency and"~~eexternal review draft includes revisions ~ade in response :0their comments~ The document reviews the available evide~caralatinq to the potential carcinogenicity of elect=omag~e~~=fields bet~een 3 Hz and 30 GHz. This evidence includesepidemioloqical studies, chronic liteti:e ani~al tests andlaboratory studies of biological phenomena related tocarcinogenesis.

The Agency seeks the advice of the Board on the acc~=~c'l andcompleteness o~ the entire document and on the question o~ .~hether the interpretation ot the available infor:ation =e~lec:scurrent scientific opinion. In addition, we ~ould like the 3ca~:to address tha.tollowinq specific issues:

1. Is the interpretation ot the human ana ani=al evidenceof eareinoqenicity supported by the availablein~or.nation?

2. Coe5 the anl=al or bioloqical e:~.cts in:or:ationprovide a basis ~or postulatinq that there is a humanhazard trom exposure to extremely low frequency fieldsor either :odulated or ur~odulated radio frequencyradiation?

J. Has the aqency properly evaluated the ~ay in ~hich :~etindinqs on biological effects and :ield-ti~sueinteraction :echanis:s a~:.ct the intarpracation o~the human studies?

4. Is the choice ot topics in chapter 5 appropriate and ~sthe interpretation.o! the biological 'f~eets lieerat~:eas it relates to careinoqanesis supported QY theavailable intor=ation?

s. ts the AqenCY's.carcinoqen classitieatian syste.applieaDle ta alectromaqnetie tields?

6. 'ec.. the intomtion cited in ttledceu:ent support ':haconcluaicn that there is not anouqh in:or:aeion to4aaiq,nate specific values ot 2aqnetic field stren~h as

-~~1nq h&:~ous to human healea?

The secon4 document ~ill Q. reviawed =y appropriata ra4era1Aqancie. &n4 other non-faderal interasta4 orqanizaeions durinqoceocer an4 will Ilso be availADle tor SAl reviaw in early~ovem.b.r. l'!1& purpo.e ottha ~ation&l a•• earch Aqenda on ~ is~o identity the sajor research na.d~ to: 1) accurately as••ss~'a human ae.lta ettect. ~ro. exposure to !H7 and 2) identity andrecommend appropriate exposure quideline. and control technoloqyto protect auun '11.aItlf.--nt.-' <!OCUJll.nt:·~••c:rib•• a n&~ton.al viev-"- ­.of ':henece.sU"'! r.search in the arias ot exposure ass.snent,oioloqical ettacts, and control :achnoloqy, ~i~out reqard :0 .ho- - -- -- -. ... - . - -- _.- ..- .. - -.....- - ----

;..-1. -

" -

or ...,hich orqanization :night do the resear~h. 'I'/e a::e:'=eq'~es:~~;the SAS to address the following issues i~ thei~'~~~iew o~ .:~~

t ".~l\ •.. "'~"."" .documen : !J-. "''''' u

1) Does the document identify t~e :najorresearch ~eeC5EMF? Specifically are any idantified needsinappropriate and are all the needs identi~ied:

2) Is the level of detail sufficient to set prior:::esamonq the research needs?

- ~..

:1)

-

Co any research needs stand out as higher pri~=:::issues for assessinq human health risks?

5 ~.!.._..

. ------.. .. 4_ _ .. , .... ----- ...:..- )

-- .. -

-

APPENDIX B: Detailed Technical Comments

Comments ot the individual group members on Chapter 4 fol~~~:

The described animal studies focus exclusively on RFexposures and thus bear little relevance to ELF fields prod~cedby transmission lines and household electricity. The mechan:s~sof tissue effects from 1000 Mhz exposure may not relate tomechanisms at 60 Hz. Therefore, as.pointed out previously, ~Fand ELF field effects should be handled separately in thissection of the report.

The specific presentation of experiments in this chapterintroduces problems. For example, reanalysis of the Guy et ai.data from 1985 to reach conclusions other than those presented bythe authors is inappropriate in this type of report. Scaling ~~normalize absorption rate is questionable. Some RF studies maybe excessively thermal. The relevance of all studies should bediscussed in regard to the spectrum of exposure frequencies.

The cited studies appear to have little relevance to ELFfields as produced by transmission lines and householdelectricity. For example, there is no evidence that theinteraction mechanisms involved at 1000 MHz are in any waysimilar to those at 50 Hz. Since the interaction is most likelyto depend critically on frequency, scalinq to normalizeabsorption rate is certainly inappropriate. Moreover, thegroupinq of glandular organs implied in the presentation ishighly unorthodox. No conclusions for ELF can be derived fromthese experiments.

The microwave chapter should give the conclusion asoriginally stated in the 1985 report by Guy et al., whichcombined all malignant tumors and reported a significantdifference between exposed and controls, but this differencedisappeared when benign tumors were included. The Guy reportconcludes that the observed differences are not of biOlogicalsignificanc~~

Much attention is given to the fact that the exposurecondition. ift the Guy study were 1I ••• calibrated to simulate humanexposure a~ the upper limit allowed by the ANSI standards" scaledfrom the weight of a small child exposed to 450 MHz (a standardradar). These calculations apparently do not take into accountthat resonance for animals in the circularly polarized waveguideused by Guy differs from estimates on resonance taken in ananechoic chamber or a multimode cavity.

3-1

The use of the post hoc analysis of the Guy data in :~~5document must be seriously questioned. Such an analysis s~:·~~:at the very least be submitted for publication before it ca~ 1=~any benefit to the document. Such re-analysis of data is ~~::ecommon in literature. If Kunz believes in this re-analysis, ~eshould submit it for publication with all.the appropriateco-authors. If the study had been intended as a cancer study :~if it were ELF one might think that it was important enough :~the overall theme of the report to waive the peer reviewrequirement. Since the original Guy study meets neither of t~csetwo criteria, there is no reason to include such a re-analys:s :~the present document.

The report states that the Sprague Dawley rat was chosen ~~m~mlc heterogenous variety of human population. But the Guystudy used all males and Sprague Oawleys are an albino strain.Creel has suqgested many unusual attributes of albinos. Thepaper by Creel on albino rats as experimental animals shouldtherefore be cited and discussed.

If the Guy post hoc analysis is included in Chapter 4, itshould be given much less attention and the statisticalquestionability of the results should be focused. Foursignificant differences can be found in the post hoc analysis ofthe Guy data. The fact that the exposed group lived longer thanthe control was not significant: nor was the analysis of thecause of death. With regard to the tumor incidence, the post hocanalysis showed significant differences using a one-tailedFisher·s exact test for benign pheochromocytoma of the adrenalmedulla (p < 0.023), malignant tumors at all sites (p < 0.0012),carcinomas at all cites (p < 0.018), and glandUlar carcinoma forcombined glands (p < 0.018). The EPA report makes no mention ofhow many statistical tests were done on these data. It wouldseem that multiple independent tests on all the possible effectswould have yielded a suspected significance that might evenexceed the four reported. .

The Prausnitz and Susskind study was !xcessively thermal.It is unclea~ how an exposure of 0.1 mW/cm over 4.5 minutescould induce a 3.3 degree C temperature increase, for example.This study has been criticized and these other criticisms shou~jbe included in this report. The most important to include areprobably Roberts and Michaelson (Health Physics 44: 430-433.1982) and Kirk (Life span and carcinogenesis, in J. Elder andD. Cahill, Eds., Biological Effects of Radiofrequency Radiation.EPA-600/8-83-026F, 1984). The criticisms focus on the fact thatthe colony of mice had an infection of pneumonitis during thestudy and that the leucosis reported may not have been leucosis.The authors note that the exposed animals lived longer than thecontrols (an observation also made by others) and suqgest thatthe mild heating enabled exposed animals to thwart the virus Hithwhich the colony was infected. There is also a question of how

B-2

t

many multiple comparisons were made to arrive at the very feNstatistical significant effects reported.

The Szmigielski et ale (1982) study must be criticized ferthe lack of pertinent information. The only effects were onpromotion ot skin cancer. Heating the skin could have been themechanism. This section should make it clearer that this is nota robust study or conclusion. There was no control of heatir.g.

B-3

i

I!II~7#

II!Ii

DISTRIBUTION LISTAdministratorAssistant AdministratorsRegional Adainistrators;Director, Ottice of Health and Environmental Assessment

....Director, ottice of Health Research. Director, ottice ot Radiation Programs

Director, Health Effects Research LaboratoryHeadquarters LibrarianRegional LibrariansNational Technical Information service

8-4