A review of the scientific evidence · 2020. 12. 7. · 3.2 Effect of EMF on DNA 19 3.2.1 Genotoxic...

Do electric and magnetic fields cause childhood leukaemia?

A review of the scientific evidence

Prepared for CHILDREN with LEUKAEMIA by Dr Adrienne Morgan and Katie Martin BSc

Registered Charity No. 298405. Inaugurated in 1988 by Diana, Princess of Wales in memory of Jean and Paul O’GormanCHILDREN with LEUKAEMIAFighting Britain’s biggest child killer disease

2

Contents

Contents page

Executive summary 4

1 Introduction 51.1 Electric and Magnetic fields 51.1.1 Magnetic fields 61.1.2 Electric fields 61.2 Childhood leukaemia 61.2.1 Types of childhood leukaemia 61.2.2 The genetic basis of leukaemia 71.2.3 Incidence of childhood leukaemia 71.2.4 The causes of childhood leukaemia 7

2 Epidemiology 102.1 Limitations of the epidemiological approach 102.2 Exposure assessment 112.3 Statistical methodology 112.4 Epidemiological evidence 122.4.1 Greenland et al 2000 122.4.2 Ahlbom et al 2000 142.4.3 Wartenberg 2001 152.4.4 Draper et al 2005 162.5 Epidemiology: conclusions 16

3 Biology – Experimental Evidence 183.1 Effects of the electromagnetic spectrum on biological material 183.2 Effect of EMF on DNA 193.2.1 Genotoxic and promoter pathways 193.2.2 Vijayalaxmi and Obe, 2005 203.2.3 The REFLEX project 213.2.4 Effects of EMF on DNA: Conclusions 223.3 Effect of EMF on cell function 233.3.1 Intracellular calcium ion levels 233.3.2 B cell signalling pathways 233.3.3 Gross transcription 243.3.4 Specific gene expression 243.3.5 Effects of EMF on Cell Function: Conclusions 243.4 Effect of EMF on tumour development in animal models 253.4.1 Spontaneous tumours and progression 253.4.2 Rat mammary carcinomas 253.4.3 Effects of EMF in animal models: Conclusions 26

4. Biology – Theories 274.1 Free radicals 274.2 The melatonin hypothesis 274.3 Effect of EMF on airborne particles 28

5 Main conclusions and recommendations for further research 305.1 Is the association real? 305.2 Recommendations for future work 315.2.1 Epidemiology 315.2.2 Biology 315.2.3 Theories 325.2.3 And Finally 32

6 Appendix: Description of all childhood epidemiological leukaemia studies 33

7 References 35

Registered Charity No. 298405. Inaugurated in 1988 by Diana, Princess of Wales in memory of Jean and Paul O’GormanCHILDREN with LEUKAEMIA

Tables and figures

Figures page

Figure 1 The Electromagnetic Spectrum 5Figure 2 Electrical and Magnetic Fields 6Figure 3 White blood cell lineages and the major leukaemias 6Figure 4 Trends in childhood leukaemia incidence, survival and

mortality, 0-14 years, England and Wales 1911-2000 7Figure 5 Relative risks at 0.3 µT and their 95% confidence intervals

reported in studies with children in the high magnetic fieldexposure group (from Greenland et al, 2001) 13

Figure 6 Relative risk against exposure level (Greenland et al, 2000)In their discussion, Greenland et al recognised a number ofproblems with their study 13

Figure 7 Relative risk by exposure levels – all studies. From Ahlbom et al,2000 14

Figure 8 From Draper et al, 2005. Distance of address at birth fromnearest National Grid line, relative risk and EMF exposurelevels 16

Figure 9. Mechanisms of Carcinogenesis 19

Tables page

Table 1 Possible causative factors in childhood leukaemia 8Table 2 Possible protective factors in childhood leukaemia 9Table 3 Other factors associated with childhood leukaemia 9Table 4 From Greenland et al, 2000. Numbers of cases and controls

in high exposure category (≤0.3 µT) and relative risk for eachstudy 12

Table 5. From Ahlbom et al, 2000. Number of cases and controls in highexposure category from each study and relative risks. Referencelevel:

Executive Summary

4

Electric and magnetic fields (EMF)are created by the presence ofelectricity. They surround us inmodern life and are produced invarying degrees and strengths by allelements of the electricity supplysystem – from high voltage powerlines to the electrical appliances in ourhomes. EMF have come underscrutiny as a possible source of harmand have been blamed for a widerange of adverse health effects. Agreat deal of research has beencarried out investigating thesepossible effects, with mixed results.Perhaps the largest body of evidencerelates to childhood leukaemia wherethere is now the strongest evidence ofa link.

EpidemiologyThe bulk of the evidence comes fromepidemiological studies, looking atthe pattern of childhood leukaemia inrelation to levels of exposure to EMF.Since the first study was published in1979, more than 25 epidemiologicalstudies around the world haveinvestigated the association betweenchildhood leukaemia and exposure toEMF. These studies have varied inquality, size and in the way that theyhave assessed exposure, leading toconclusions that are sometimesinconsistent. Most of the individualstudies have had very limited power todetect associations since they areconstrained by the relative rarity ofchildhood leukaemia and further bythe relatively low number of childrenexposed to high levels of EMF.

A number of researchers have pooleddata from several individual studies toincrease the ability to detect anassociation. The three most recent ofthese all report an increase in risk andthese studies are discussed in detail in

this report. The most recent study weinclude was published in 2005 in theBritish Medical Journal. It meritsinclusion by virtue of being thelargest single study of childhoodcancer and power lines, with roughlytwice the number of children livingclose to power lines than the nextlargest study. The authors reportedan increased risk of leukaemia inchildren in England and Wales whosebirth address fell within 600m of ahigh voltage power line.

No consensusThere remains a wide spectrum ofopinion within the scientificcommunity as to the meaning of theresults of the epidemiological studies.Epidemiology has limitations butmethods have become moresophisticated over time, and laterstudies address some of theshortcomings of earlierinvestigations. Interestingly, thishasn’t significantly altered therelative risks reported and theconsistency of findings in differentpopulations strengthens the evidenceof an association. However, there isstill much debate as to whether theresults are real or whether theobserved relationships may reflect acoincidental association with someother factor. In any event, it isdifficult to infer causal relationshipsbased on epidemiological studiesalone. The growing body of evidencerelating to the biology and causalmechanisms should help to furtherour understanding of therelationship between EMF exposureand childhood leukaemia.

Biological effectsThere are well-establishedmechanisms by which EMF couldproduce biological effects but resultsfrom many hundreds of experiments

since 1979 are overall contradictoryand inconclusive. Exposure to EMFhas been shown to damage DNA, toalter cell function and acceleratetumour development. However, foreach study which shows an effect,there are many good andreproducible studies which show noeffect. This presents difficulties forthe scientific community and there isstill considerable controversyamongst the scientists working in thisarea.

Mechanistic theoriesA number of theories have beenadvanced to explain both thescientific and the epidemiologicalevidence for a link betweenchildhood leukaemia and EMFexposure. The three which have themost supporting evidence are thatEMF could act: by directly increasingthe level of free radicals within thebody; by decreasing the level of thehormone melatonin; by affectingexposure to airborne pollutants.These theories aren’t necessarilymutually exclusive.

Do electric and magnetic fields causechildhood leukaemia?Following our review of the evidence,we have to say we don't know - yet.We believe that there is goodepidemiological evidence for adoubling of risk of childhoodleukaemia in children exposed toEMF above a certain level (0.4 µT).To progress from this to a proof thatEMF are a cause of childhoodleukaemia is a big jump and, at thisstage, not clearly supported by thebiological evidence although we haveperhaps moved on from ‘implausible’to ‘plausible’. More research workneeds to be done and this reportends with some recommendations forfuture studies.


1. Introduction

Leukaemia is the most commonchildhood cancer. Developments intreatment and care have led to adramatic increase in survival rates inrecent years but leukaemia still killsmore children than any other diseasein the UK.

The causes of leukaemia areuncertain. There are some factorswhich are known to cause the disease(such as exposure to high doseradiation) but these cannot accountfor all cases. Incidence of the diseaseincreased throughout the 20thcentury, suggesting that aspects of ourchanging lifestyle may be partly toblame.

The first study linking electric andmagnetic fields (EMF) and childhoodleukaemia was published in theUnited States in 1979. Since then,more than 25 epidemiological studiesinvestigating the association havebeen carried out around the world.There is also an accumulating body ofscientific evidence relating to thepossible biological mechanisms thatmight lie behind such an association.This report sets out to bring togetherthe available evidence into onecomprehensive document.

This report is not as daunting as it firstappears. Primarily, we present theepidemiological (section 2.4) andbiological (section 3) evidence forand against electric and magneticfields increasing the risk of childhoodleukaemia. But first we need tointroduce a number of the conceptsand ideas in those two sections: thenature of EMF (section 1.1);information about childhoodleukaemia (section 1.2); and anumber of issues related to theinterpretation of the epidemiology(sections 2.2 and 2.3). Afterdescribing and discussing theepidemiology and biology we go on todiscuss some of the theories proposedto explain the effects of EMF (section4). Finally, we present ourconclusions and some suggestions forfurther work.

1.1 Electric and Magnetic fieldsElectric and magnetic fields arecreated by the presence of electricity.The electric and magnetic fieldscreated by the electricity supply systemare part of the electromagneticspectrum, which also includes X-rays,ultraviolet light, visible light, infraredlight, microwaves and radio-frequencyenergy. The parts of the

electromagnetic spectrum arecharacterized by their frequency ortheir wavelength which are inverselyrelated - as the frequency rises thewavelength gets shorter. Thefrequency is the rate at which theelectromagnetic field goes throughone complete cycle and is measuredin Hertz (Hz), where one Hz is onecycle per second.

See Figure 1. The ElectromagneticSpectrum.

The electricity supply system in theUK and most of the world has afrequency of 50 Hz and the USA 60Hz. The electric and magnetic fieldsgiven out by the electricity supplysystem are therefore referred to asextremely low frequency (ELF)electromagnetic fields as they are atthe lowest end of the electromagneticspectrum. In this report we refer tothese extremely low frequency electricand magnetic fields as EMF for thesake of simplicity.

Electricity is transmitted from powerstations to the point of use (homes,offices, schools etc) via a network ofpower cables and substations (knownas the national grid). The substationsstep down the voltage progressively asthe electricity nears its point of use(voltage is the driving force behindthe flow of electricity, somewhat likethe pressure in a water pipe). Themajor power lines which are carriedon the large pylons across much of thecountry operate at 400,000 volts (V),stepping down to 275,000 V and then132,000 V as the supply nears thepoint of use. When the electricitysupply enters the home it has beenstepped down to 230 V.

Figure 1. The Electromagnetic Spectrum.

6

1.1.1 Magnetic fieldsA magnetic field is a force fieldgenerated by moving electricalcharges. An electric current runningthrough a cable generates a magneticfield. The strength of the fielddepends on the current (the rate offlow of electricity). Magnetic fieldsare measured in tesla (T) usuallymicrotesla (µT). The three mostcommon sources of magnetic fields inthe home are electric appliances, theearthing system of the home andnearby power lines. Ambient levels ofmagnetic fields in most homes aretypically in the range of 0.01 - 0.3 µT.Close to high voltage power lines,magnetic fields are typically above 1µT, although values can reach tens ofµT. Some domestic appliancesproduce magnetic fields of 100 µT ormore in their vicinity. The field dropsoff rapidly with increasing distancefrom the source.

1.1.2 Electric fieldsAn electric field surrounds anyelectrically charged object. If a powerline is connected to the national gridor if an appliance is plugged in to themains supply (regardless of whetherthe appliance is switched on), it is‘charged’ and therefore will besurrounded by an electric field. This isin contrast to a magnetic field whichwill only be produced if the applianceis in use and there is actuallyelectricity flowing through theappliance or cable. Electric fields aremeasured in volts/metre (V/m) andtheir strength depends upon thevoltage on the wire and the distance

from the line. The Electrical PowerResearch Institute found that theaverage personal exposure to electricfields in the home typically rangesfrom 5 to 10 V/m. Directlyunderneath a 400,000 V power line,the electric field might be as high as10,000 V/m.

1.2 Childhood leukaemiaThe term ‘leukaemia’ describes agroup of cancers involving an excessof white blood cells. Leukaemias arethe most common childhood cancers,accounting for around one third of allcases – about 500 children per year inthe UK (aged 0 to 14 years). Everyyear in the UK around 100 childrenare killed by leukaemia, morechildren than are killed by any otherdisease.

Leukaemic cells develop from cells inthe bone marrow ("haemopoieticstem cells") that go on to becomeblood cells. Leukaemia arises fromthe abnormal transformation of asingle cell. This single cell generatesan expanding clone of abnormal cellsvia repeated cell divisions. The resultis a proliferation of abnormal whiteblood cells and a disruption of theproduction of normal blood cells.

Figure 3. White blood cell lineages and the major leukaemias

Haemopoietic cells divide frequently –undergoing some 100,000,000,000divisions per day in the adult and evenmore in utero when the embryo isgrowing rapidly. The cells that go onto become white blood cells(lymphocytes) undergo DNArearrangement to create the largenumber of different types of cellsneeded by the immune system. Thisprocess is intrinsically prone to DNAerrors – which may occur eitherspontaneously or as a result ofexposure to external carcinogens.This vulnerability to cancerdevelopment might be seen as anevolutionary trade-off for theotherwise highly advantageousproperties of the immune system.

1.2.1 Types of childhood leukaemiaLeukaemia can be classified as eitherlymphoid or myeloid, denoting the typeof white blood cell affected, and aseither acute or chronic, reflecting thespeed of progression.

Almost all childhood leukaemias areof the acute form. Acutelymphoblastic (lymphoid) leukaemia(ALL) accounts for more than 80 percent of all cases of childhoodleukaemia. It is the only form ofleukaemia – and one of the few formsof cancer – that is more common in

Figure 2. Electric and Magnetic Fields.


1.2.2 The genetic basis of leukaemiaLeukaemic cells can carry a variety ofabnormalities including chromosomegain or loss or structural changes likechromosome deletions, duplications,inversions and translocations. In childhood leukaemia thesechromosome rearrangements arisemainly before birth. However, theyarise at a frequency far greater thanthe incidence rate of childhoodleukaemia1. For example, the ‘TEL-AML1’ fusion is present in aroundone per cent of new born babies butless than one per cent of babies with

the fusion will go on to developleukaemia.

It therefore appears that, whilst thedevelopment of childhood leukaemiais frequently initiated in the womb,additional post-natal events arerequired for the child to develop full-blown leukaemia. This is known as the‘two hit hypothesis’ and the very firstevidence for this was derived fromstudies of identical twins. The rate ofconcordance between leukaemia inidentical twins varies substantiallyaccording to subtype and age of onsetof disease2 3. For infant ALL (ie. onsetbefore one year of age), theconcordance rate is greater than 50%.For typical childhood ALL, theconcordance rate is much lower ataround 10%. Adult leukaemia, bycontrast, has a concordance rate ofless than one per cent in identicaltwins, suggesting an increasinginfluence of environmental factorsover genetic factors with increasingage.

Figure 4. Trends in childhood leukaemia incidence, survival and mortality, 0-14years, England and Wales 1911-2000.

Incidence and mortality rates per million for England and Wales4 5 five-yearrelative survival rates (%) for children diagnosed in South-East England 1960-886

and in Great Britain 1993-97 (Coleman, 20047).

1.2.3 Incidence of childhoodleukaemiaAnother observation which suggeststhe importance of environmentalfactors in the development ofchildhood leukaemia is that theincidence of the disease appears tohave increased steadily throughoutthe 20th century, as shown in thechart above.

Incidence data were not collectedprior to the 1970s but until theintroduction of the first combinationtherapies in the 1960s childhoodleukaemia was almost inevitably fataland so the mortality figures provide afairly accurate representation ofincidence figures until the 1960s whenthe two curves dramatically diverge assurvival rates begin to grow.

Since our genetic make-up does notchange significantly over such a shorttime scale, the increasing incidence ofchildhood leukaemia must be areflection of some aspect of ourchanging lifestyle or environment.

1.2.4 The causes of childhoodleukaemiaAs outlined in sections 1.2.1 and 1.2.2above there are a variety of differenttypes of leukaemia and a variety ofdifferent genetic mutationsunderlying these different types of thedisease. Considerable advances havebeen made in understanding thebiological mechanisms underpinningthese different disease types andscientists have mapped out many ofthe chromosome translocationsinvolved.

What this biological aetiology doesn’treveal, however, is the underlyingcausal factors. What factors (otherthan spontaneous error) cause theinitial chromosome translocations tooccur and what factors are the ‘secondhits’ which predispose children withthese translocations go on to actuallydevelop leukaemia?

children than in adults. Acutemyeloid leukaemia (AML) accountsfor most of the remaining cases.

Chronic leukaemias are very rare inchildhood. Chronic myeloidleukaemia (CML) accounts for lessthan 3 per cent of childhoodleukaemias (less than 15 children peryear in the UK) and chroniclymphocytic leukaemia (CLL) isunheard of in children.

8

Table 1. Possible causative factors in childhood leukaemia

Factor Potential importance 1. Ionising radiation It has been shown that exposure to ionising radiation

(even at quite low doses) can cause leukaemia.Epidemiological evidence for this comes from studiesof in utero irradiation of the foetus through obstetricxrays and studies of Japanese atomic bomb survivors. Itis estimated that background ionising radiation isimplicated in around 25% of cases of childhoodleukaemia, with exposure both in utero and early inlife being important. See Wakeford (2004)8 for areview. There is some evidence that paternal pre-conceptual exposure (through irradiation in the workplace) may be important (Gardner et al, 19909).

2. Non-ionising radiation

i) Light at night There is a suggestion that our increasing exposure tolight at night may be increasing the risk of childhoodleukaemia through its disruption of our circadianrhythm and suppression of the hormone melatonin(Henshaw & Reiter, 2005125).

ii) ELF electric and magnetic fields See section 2 onwards.

3. Chemical exposure

i) Air pollution The risk from air pollution is difficult to detect inepidemiological studies as a result of the ubiquitousexposure in developed countries. Knox (200510,11)found an increased risk of childhood cancer for birthaddresses within 1 km of hot spots for various airpollutants.

ii) Parental smoking Epidemiological evidence to support a link withchildhood leukaemia is inconsistent, however at leastone study has found increased chromosomalabnormalities in amniocyte cells of foetuses of smokingmothers (De la Chica et al, 200512).

iii) Pesticides Associations have been reported between childhoodleukaemia and either parental or child exposure topesticides (e.g. Infante-Rivard, 199913).

iv) Prescription drugs (pre-natal exposure) One study has reported an association betweenchildhood ALL and maternal use of antihistaminesand allergy remedies (Wen et al, 200214). Robison et al(1989)15 reported a link between maternal use ofantiemetic medication and AML.

v) Parental alcohol consumption There is weak evidence for a possible link betweenmaternal alcohol consumption during pregnancy andAML but no evidence for a link with ALL16.

v) Recreational drugs and alcohol Maternal use of marijuana has been reported to(pre-natal and pre-conceptual exposure) increase the risk of both childhood ALL (Wen et al,

200214) and AML (Robison et al, 198915). There isevidence that the risk may be higher when bothparents use the drug.

vi) DNA topoisomerase inhibitors This includes some drugs used in chemotherapy,benzene metabolites (from air pollution and cigarettesmoke), certain fruits, tea, coffee, wine, soy and cocoaand many other substances17.18 Topoisomeraseinhibitors Inhibit DNA repair and are stronglyassociated with one of the chromosomerearrangements common in infant leukaemia19.

vi) Diet N-nitroso compounds (found in cured meats and hotdogs) have been linked with childhood leukaemia in atleast one study (Peters, 199420).

4. Infectious exposure There is evidence of an excess of cases of childhoodleukaemia in locations with an unusual type ofpopulation mixing. It has been proposed that this isdue to an, as yet unidentified, infectious agent andthat leukaemia is a rare consequence of exposure tothis agent. In isolated communities a higherproportion of the population would have beenpreviously unexposed to such an infection and a rapidinflux of newcomers into such a community wouldlead to an increased level of contact between theinfected newcomers and the susceptible originalresidents (Kinlen, 1995)21.

There is no single external factorwhich is either ‘necessary’ or‘sufficient’ to cause leukaemia inchildren. That is, there is no singlefactor to which a child must beexposed if they are to developleukaemia and there is no singlefactor, exposure to which isguaranteed to result in developmentof leukaemia.

Rather it is thought that leukaemia ismulti-causal and multi-factorial, thatis, there are many different factorswhich may cause leukaemia andexposure to more than one of these isprobably necessary – probably atdifferent stages of a child’s life –perhaps, as already described, one‘hit’ in utero and a second ‘hit’ in earlychildhood.

Most of the environmental andlifestyle factors which may beimplicated in the causes of childhoodleukaemia are extremely difficult toinvestigate in epidemiological studies.The difficulties which are explored inSection 2 in relation to EMF alsoapply to many of the otherenvironmental and lifestyle factorswhich have been linked withchildhood leukaemia.

The factors which have been linked tochildhood leukaemia can be dividedinto three categories: exposure tocausative factors increases the risk of achild developing the disease; exposureto protective factors reduces the risk ofa child developing the disease; linkedfactors are correlated with theincidence of leukaemia but are notdirectly causative or protective, morelikely they reflect the likelihood ofexposure to another causative orprotective factor.

Some of the main factors for whichassociations have been identified areset out in the three tables whichfollow. Space does not permit adetailed discussion of these factors,and we do not claim to have providedan exhaustive list of the evidence forand against each of the suggestedfactors, more we have aimed to sign-post the reader towards some of therelevant research.


Table 2. Possible protective factors in childhood leukaemia

Factor Potential importance1. Diet Evidence from one study suggests that there is a strong

protective effect of consumption of oranges andbananas in early life (Kwan et al, 200422).

2. Folate and folate metabolism Folate metabolism is thought to be important in thedevelopment of leukaemia. There is some evidence tosuggest that maternal folate supplementation duringpregnancy may protect against childhood leukaemia(Thompson et al, 200123). There are also differences inthe way that individuals metabolise folate and this maybe important (Wiemels et al, 200124).

3. Infectious exposure Lack of exposure to infections in early life results in animmature immune system and this may increase risk ofchildhood leukaemia (Greaves, 1997)25. For example, arecent study found that children attending day carefrom young age are less likely to develop leukaemia.(Gilman et al, 200526).

4. Breast feeding There is a fairly substantial body of evidence pointingtowards a protective effect of breast-feeding. A recentmeta-analysis reported a relative risk of 0.76 (0.68 –0.84) (Kwan et al, 200427).

Table 3. Other factors associated with childhood leukaemia

Factor Potential importanceMaternal age >35years Most studies have observed an increased risk for

childhood leukaemia with advanced maternal age (e.g.Reynolds, 200228) although one study has reported arisk association in ALL with young maternal age (Shu,200229).

Miscarriage history Maternal history of previous miscarriages is afrequently reported risk factor for development of ALL– and in some cases AML - in a subsequent child (e.g.Perrillat, 200230).

Birth order Being the first born child has been associated withincreased risk in some studies (e.g. Dockerty, 200131)although the opposite has also been reported (e.g.Shu, 200229). One study identified being the only childas a risk factor (van Steensel-Moll, 198632).

Socioeconomic status Incidence of ALL is higher in areas of high social class(e.g. Alexander 199133, Stiller & Parkin, 199634).Evidence from developing countries suggests thatincidence of ALL in children aged 1-4 years is risingwith improved socio-economic conditions (Hrusak,200235).

Birth weight Higher birth weight is associated with increased risk ofchildhood leukaemia (Hjalgrim, 200336).

Congenital disorders Children with Down’s Syndrome and certain othergenetic syndromes are much more susceptible toleukaemia (Reynolds, 200228).

Gender The gender effect in incidence of ALL is well-established, with boys being approximately 20% morelikely to develop ALL than girls (e.g. Pearce & Parker,200137. Males also have a worse prognosis (Eden,200038).

Ethnicity Incidence of ALL is significantly lower among blackchildren in the US (Reynolds, 200228). This may be asocial class effect.

Many of these factors may well be correlated with each other.

2. Epidemiology2. Epidemiology

10

Epidemiology is the study of thedistribution of disease in populationsand of the factors that influence thisdistribution. It is concerned with thepatterns of disease among groups ratherthan with treating the disease at theindividual level. Epidemiology is veryvaluable in assessing risks to humanhealth from certain factors but it doeshave limitations and these arediscussed under 2.1 below.

To investigate an association in a raredisease such as childhood leukaemiaand EMF, epidemiologists generallyuse a case-control study. This is a way ofcomparing people who already have aparticular disease (cases) with peoplewho do not (controls). In the case ofEMF, epidemiologists will recruit agroup of people who have leukaemiaand a group of people who do not andcompare the exposure of the twogroups to EMF to see if there are anydifferences. The controls are chosento be as similar to the cases as possiblein all characteristics except theirexposure.

Since publication of the first study byWertheimer and Leeper in 197939,more than 25 epidemiological studieshave investigated the associationbetween childhood leukaemia andexposure to magnetic fields. Therehave also been a number of publishedreviews and pooled analyses of theseindividual studies and we focus onthese overviews because they providestronger evidence than individualstudies. Overall, a significant statisticalassociation is found with the incidenceof childhood leukaemia and exposureto power frequency magnetic fields of0.3 - 0.4 µT, levels commonly foundunder or close to high voltageoverhead power lines. More recently,associations have been found at greaterdistances from power lines suggesting apossible effect of electric as well asmagnetic fields.

There remain deep divisions ofopinion as to the meaning of theepidemiological evidence. In any eventit is difficult to infer causalrelationships based on epidemiologicalstudies alone and the growing body of

evidence relating to the biology andcausal mechanisms (see sections 3 and4) should help to further ourunderstanding of the relationshipbetween EMF exposure and childhoodleukaemia.

Key considerations

2.1 Limitations of the epidemiologicalapproachEpidemiology is observational ratherthan experimental, meaning that theepidemiologist does not control whoreceives exposure. This may introducebias (a systematic tendency to error as aconsequence of design or conduct ofthe study) or confounding (spuriousfindings due to the effect of a variablethat is correlated with both theexposure and the disease under study).These problems can be overcome byappropriate study design, includingcareful selection of controls, andanalysis. But, it is not always possible toeliminate them completely.

Control selection biasIn a case-control study, cases areselected on the basis that they haveleukaemia. Investigators then set out torecruit controls that ‘match’ the caseson key characteristics (such as age,gender, ethnic background). If thecontrols are not properlyrepresentative of the population as awhole, any association found in thestudy might not be real. As anexample, in much Americanepidemiology the control group isrecruited by telephone. This couldresult in socio-economic differencesbetween the case and control groups aspoorer families are less likely to havetelephones and are therefore likely tobe under-represented in the controls.This would introduce a bias. Thisproblem of potential bias has to beconsidered in all case-control studiesand it is not possible to know howmuch it affects the results.

Non-response biasNot all controls will agree to take partin the study (nor all cases, although theresponse rate is higher amongst cases).Differential participation of controls asa function of their socio-economic

status has been proposed as a possibleexplanation of the association betweenEMF and childhood cancer i.e. thosewith lower socio-economic status areless likely to take part and this may biasthe results. This does not apply instudies where contact is not necessary –for example, the Nordic studies whichare based on national populationregistries.

ConfoundingA confounding factor is a factor thatmay also affect the relative risk but itseffect cannot be separated out fromthe factor under investigation. Forexample, an association has beenfound between magnetic fields andcancer. On the face of it this maysuggest that the fields increase the riskof cancer. But it could be that the riskof cancer is increased by somethingelse that is also associated both with thefields and with the cancer.

For example, homes with highmagnetic fields tend to be older,smaller properties, in areas of higherhousing density, on busier roads,occupied by families with lowereducation, lower socio-economic statusand younger children. If any of thesefactors – such as socio-economic status -are also associated with thedevelopment of cancer then they willconfound the results of the study.

An association exists between thedivorce rate in Britain and the level ofsales of Golden Delicious apples. Thisassociation is due to the correlation ofboth factors with levels of prosperity. Ifwe don’t take prosperity into accountas a confounding factor then we wouldend up with the conclusion that eatingGolden Delicious apples increases thelikelihood of your marriage ending indivorce!

Well designed studies attempt to adjustfor the effects of possible confoundingfactors.

Statistical powerAn important aspect of anepidemiological study is its statisticalpower – that is, the probability that ifthere really is an increase in risk of a


certain size the study is large enough todetect this. Small studies only have thepower to detect large risks. This can bea major problem in epidemiologicalstudies looking at the associationbetween childhood leukaemia andEMF since both the disease and theexposure are relatively rare so thenumber of both cases and controls arelimited.

This means that many studies have lowstatistical power and therefore alimited ability to detect small increasesin risk. One way to overcome this is topool results of individual, comparablestudies into a meta-analysis. Thisincreases the statistical power of thecombined data to detect a smaller riskthrough increasing the number ofsubjects.

2.2 Exposure assessmentThere are many difficulties inherent intrying to estimate the exposure of thechild to EMF, including that i)exposure has multiple sources and canvary greatly over time and shortdistances; ii) the exposure period ofrelevance is before the date at whichmeasurements can realistically beobtained and of unknown durationand induction period; and iii) theappropriate exposure metric is notknown and there are no biological datafrom which to impute it (ICNIRP,2001)40.

Period of inquiryThere is little consensus as to whichtime period may be the most crucial interms of the damaging effects ofexposure to EMF and the ‘period ofinquiry’ varies across the differentstudies. Some focus on the birthaddress, others look at the periodleading up to diagnosis and some focuson the address at diagnosis. Onlyaround 50 per cent of addresses arethe same at diagnosis as at birth andthis may dilute any observed effect.There is evidence that the mother’sexposure during pregnancy may beimportant but this has so far receivedlittle attention in epidemiologicalstudies. In whatever period isconsidered, it is difficult to reconstructhistorical exposure accurately.

Exposure metricsThere has been a wide range ofexposure metrics used in the studiescarried out to date. The first study byWertheimer and Leeper in 197939 (andmany of the subsequent studies) usedwire codes – an indirect measure ofEMF exposure based upon proximityto particular types of transmission linesand distribution equipment. Studies inthe Nordic countries especially haveused available data to calculatehistorical fields. And more recentstudies in other countries have utilisedadvances in methods for measuringexposure. The different approaches allhave their strengths and weaknesses.

Wire codes. This indirect measure wasused in the first epidemiologic study ofexposure to magnetic fields and cancerrisk (Wertheimer and Leeper, 197939)and in many of the early residentialstudies of EMF in the US. Wire codecalculations are based on the proximityto power lines and the voltage of thelines. Wire coding has become moresophisticated over time and it remainsthe most commonly used approach.Wire codes are not strong predictors ofmagnetic field strengths in homes butthey do tend to distinguish relativelywell between the very high and very lowfield homes.

Distance. Some of the initial Europeaninvestigations examined risk of cancerin relation to distance of subjects’residences from electricity generatingor transmission equipment. Becausethe magnetic field decreases withdistance from such equipment,distance can be used as a crudepredictor of the field. The electricitycompany will usually have recordsshowing whether a particular line wasoperating during a particular period.Although this method is fairly crude, itmay be the most appropriate predictorwhen investigating the effects of powerlines on human health since there isstill so much uncertainty about whichaspect of fields may be damaging andshould therefore be measured.

Calculated historical fields. Magneticfields from power lines can be

calculated accurately if the relevantvariables are known. The mostimportant variables for calculations arewire spacing and height, the current ineach wire and distance from the home.Electricity companies rarely haverecords of the current in each wire forpast years but they often have a recordof power line load from which thecurrent can be calculated – morereliably for high-voltage than low-voltage lines. This has been a popularapproach in the Nordic studies wherethe relevant data is available.

Measured fields. As researchers inmost countries do not have access tothe information necessary to calculatehistorical fields, the most commonapproach is to estimate historical fieldlevels based on measurements takenfrom the residence during the study.Two types of measurements are usedmost often: spot measurements (theaverage magnetic field over a period ofseconds or minutes) and time-weightedaverages or TWA (a long termmeasurement extending 24-48 hours).These two different types ofmeasurement do not appear tocorrelate well.

In some studies children are givenpersonal EMF meters to wear in orderto ascertain their EMF exposure fromall sources, including residential,school and other exposures. Althoughsuch measurements are the onlypractical way to determine the fieldfrom multiple sources, their majorweakness is that they might notaccurately reflect the conditions whichprevailed years before – during theperiod in which the disease developed -since a child’s lifestyle may changedramatically pre- and post-leukaemia.

2.3 Statistical methodologyEpidemiological studies always involvetaking a sample of people with diseaseto obtain an estimate of risk associatedwith a particular exposure. Thisobserved risk is an estimate of the truerisk that we would see if we couldmeasure the whole population,

12

everybody who has the disease. Becausewe have taken a sample there isuncertainty associated with ourestimate and statistical techniquesinvolve trying to quantify thisuncertainty.

Relative riskRisk itself is the probability that anevent will happen. The risk of a childdeveloping leukaemia (before the ageof 16 years) is approximately 1 in1,500. If a study investigating anassociation between magnetic fieldsand childhood leukaemia reported arelative risk (RR) of 2, this means thatan exposed child is twice as likely as anunexposed child to develop leukaemia.The risk of an exposed childdeveloping leukaemia would thereforebe approximately 1 in 750. RelativeRisks similar to an Odds Ratio (OR) incase control studies.

Confidence intervals Most studies looking at the relationshipbetween childhood leukaemia andmagnetic fields have been based onrelatively small samples (due to thelimited availability of cases andcontrols) and are thus statisticallyimprecise. Statisticians can calculate arange (interval) in which we can befairly confident that the "true value"lies. The true value is the value that wewould get if we had data for the wholepopulation. The observed relative riskis calculated from the particularsample of children taking part in thestudy and is an estimate of the truerisk.

The precision of the study can begauged by the width of the confidenceinterval (CI). Larger studies usuallyhave a narrower CI because the largernumbers give a greater degree ofcertainty. The CI is usually shown inbrackets after the relative risk e.g. 2.0(1.65 - 2.7). Unless otherwise stated,the CIs shown in this paper are the95% CIs, i.e. there is a 95% chance thatthe true value lies within the rangeshown.

Statistical significanceStatistical significance is a measure ofhow likely it is that the effect is real andnot just a chance occurrence in thisparticular sample. If there were noeffect (of exposure) the p-value tells ushow likely we would be to see an effectas large as the one we have foundpurely by chance. A p-value of 0.05means that the probability of an effectthis large having happened by chanceis 1 in 20. This is frequently the figurequoted as being statistically significanti.e. unlikely to have happened bychance. The lower the p-value, the lesslikely it is that the effect happened bychance and so the higher thesignificance of the finding.

2.4 Epidemiological evidenceSince the first study by Wertheimer andLeeper in 1979 which reportedevidence of an association betweenchildhood cancer and electrical wiring,more than 25 further studies have beenpublished.

A number of pooled (meta) analyseshave been carried out using data fromvarious of these studies, the mostrecent of which are Greenland et al200041, which included 15 studies,Ahlbom et al 200042 which includednine of the individual studies andWartenberg 200143, which reviewedseven previous meta-analyses andextended them by adding data fromfour further, more recent studies.

We include details of just oneindividual study – by Draper et al200544. This recent study, conducted inEngland and Wales, is the largest everstudy of childhood cancer and powerlines. Details of most of the earlierindividual studies can be found in theAppendix.

2.4.1 Greenland et al 2000.Greenland et al41 carried out a meta-analysis designed to answer thequestion: Are magnetic fields or wirecodes consistently associated withchildhood leukaemia?

The authors identified 24 relevantstudies. To be eligible for inclusion,each study had to have obtainedquantitative magnetic field measuresfor individual subjects or enoughinformation to approximateWertheimer-Leeper wire codes.Nineteen of the studies met thesecriteria. One study group refused toparticipate, two were unable to supplydata in time and two from the samecountry were combined into one studyfor the purposes of the analysis. Theinvestigators therefore had data from15 studies.

Table 4 shows the numbers of casesand controls in the high exposurecategory (defined as ≤ 0.3 µT) and therelative risk (reference group


In their discussion, Greenland et alrecognised a number of problemswith their study:

• Confounding factors. Although theauthors raise the possibility thatconfounders may partly account forthe observed effect, they point outthat a confounding explanationrequires the confounder to have aneffect considerably larger than theobserved association, as well as astrong association with exposure.These attributes have not beendemonstrated for the hypothesisedconfounders that display positiveassociations. They recommended thatfuture studies attempt to collect morecomplete data on possibleconfounders – especially on trafficdensity and ambient pollution levels –since their data was incomplete.

• Bias. They acknowledge that biasesare present in all studies and pointout that one of the major problems inassessing bias is that there is noagreed-upon definition of targetexposure. They suggest that all of themeasures used are likely to sufferconsiderable error since they are allproxies for the yet-to-be-identifiedbiologically relevant exposuremeasure.

They were unable to assess the impactof selection bias.

They concluded that it was likely thatmeasurement and validity differenceswere responsible for some of thevariation in study-specific results andacknowledge that this is a flaw in theirstudy – in that they pooled differentmagnetic field measures withoutdemonstrating that all of the measureswere comparable or combinable.Although they observed that thedegree of variation in the study-specific results was limited, they pointout that other choices could lead tovery different degrees of variation.This adds uncertainty to their results.

Pooling these studies of power linefields and childhood leukaemia theauthors reported a summary oddsratio of 1.7 (1.2 – 2.3) for children inthe highest magnetic field exposuregroup of 0.3 µT or more comparedwith those in the lowest exposurecategory (0.1 µT or less).

Figure 5. Relative risks at 0.3 µT and their 95% confidence intervals reported instudies with children in the high magnetic field exposure group (fromGreenland et al, 2000).

The authors also found a suggestionof a dose response to increasing levelsof magnetic field i.e. an increase inthe relative risk of childhoodleukaemia with increasing exposurelevel, as illustrated in figure 5 below.

Figure 6. Relative risk against exposure level - results from individual studieswhich contributed to the pooled analysis (Greenland et al, 2000).

0.10.20.4µT

14

They conclude that "In light of theabove problems, the inconclusivenessof our results seems inescapable;resolution will have to awaitconsiderably more data on highelectric and magnetic-field exposures,childhood leukaemia and possiblebias sources. It also appears to us that,if an effect exists below 0.2 µT, it isprobably too small to reach consensusabout it via epidemiologicinvestigation alone. In contrast, bothour categorical and trend analysesindicate that there is some associationcomparing fields above 0.3 µT tolower exposures, although there are asyet insufficient data to provide morethan a vague sense of its form andpossible sources. We believe thatindividual-level studies that focus onhighly exposed populations would beneeded to clarify this association."

2.4.2 Ahlbom et al 2000Ahlbom et al 200042 conducted apooled analysis based on individualrecords from nine studies. Theiranalysis was designed to answer twomain questions:

1. Do the combined results of thesestudies indicate that there is anassociation between EMF exposureand childhood leukaemia risk,which is larger than one wouldexpect from random variability?

2. Does adjustment for confoundingfrom socioeconomic class, mobility,level of urbanization, detached/notdetached dwelling and level oftraffic exhaust change the results?

The nine studies, which included atotal of 3,247 children with leukaemiaand 10,400 controls, were chosenbecause they met specified qualitycriteria as opposed to the Greenlandstudy which included any study whoseauthors would provide data. Studieswith 24/48-hour magnetic fieldmeasurements or calculated magneticfields were included. These studiesare described in the following table:

Table 5. From Ahlbom et al, 2000. Number of cases and controls in highexposure category from each study and relative risks. Reference level: 0.10.2


2.4.3 Wartenberg 2001Wartenberg 200143 carried out athorough review of seven previousmeta-analyses, including his own 1998paper, before going on to extend this1998 paper with the inclusion of foursubsequently published studies. Hiscriteria for inclusion were that eachstudy reported an exposure measurefor the possible association ofchildhood leukaemia with residentialexposure to magnetic fields andprovided data on the exposuremeasure used. This resulted in theinclusion of 19 studies. These are setout in tables 6 and 7. Table 6 includesthose studies which presented data oncalculated and measured magneticfields (14 studies) and table 7 includes

Table 6. From Wartenberg (2001). Meta-analysis and individual study results forstudies of calculated and measured magnetic fields and childhood leukaemia. Study Exposed Exposed Unexp Unexp Individual OR

cases controls cases controls (95% CI)

Tomenius, 1986 4 10 239 202 0.34 (0.10-1.09)Myers et al, 1990 6 6 174 271 1.56 (0.49-4.91)Savitz et al, 1988 5 16 31 191 1.93 (0.66-5.63)London et al, 1991 20 11 144 133 1.68 (0.78-3.64)Feychting & Ahlbom, 1993 7 46 31 508 2.49 (1.04-5.98)Olsen et al, 1993 3 4 830 1662 1.50 (0.34-6.73)Verkasalo et al, 1993 3 1.93 - - 1.55 (0.29-3.81)Linet et al, 1997 83 70 541 545 1.19 (0.85-1.68)Tynes & Haldorsen, 1997 1 14 147 565 0.27 (0.04-2.10)Michaelis et al, 1997a 9 8 167 406 2.74 (1.04-7.21)McBride et al, 1999 49 42 248 287 1.35 (0.86-2.11)Dockerty et al, 1998 5 2 35 38 2.71 (0.49-14.9)Green et al, 1999 25 44 58 142 1.30 (0.78-2.48)UKCCS, 1999 21 23 1052 1050 0.96 (0.50-1.66)

Combined 241 298 3697 6000 1.31 (1.09-1.59)

Table 7. From Wartenberg (2001). Meta-analysis and individual study results forstudies of proximity to electrical facilities and childhood leukaemia. Study Exposed Exposed Unexp Unexp Individual OR

cases controls cases controls (95% CI)

Wertheimer & Leeper, 1979 52 29 84 107 2.28 (1.34-3.91)Savitz et al, 1988 27 52 70 207 1.54 (0.90-2.63)London et al, 1991 122 92 89 113 1.68 (1.14-2.48)Linet et al, 1997 111 113 291 289 0.98 (0.72-1.33)McBride et al, 1999 122 128 229 234 0.97 (0.72-1.32)Green et al, 1999 12 25 67 100 0.72 (0.34-1.52)Fulton et al, 1980 42 56 131 169 0.95 (0.60-1.50)Feychting & Ahlbom, 1993 6 34 32 520 2.87 (1.12-7.33)Tynes & Haldorsen, 1997 9 55 139 524 0.62 (0.30-1.28)Fajardo-Gutierrez et al, 1993 3 2 43 47 1.64 (0.26-10.29)Coleman et al, 1989 3 3 81 138 1.70 (0.34-8.64)Petridou et al, 1993 11 14 106 188 1.39 (0.61-3.18)Myers et al, 1990 7 7 173 270 1.56 (0.54-4.53)

Combined 527 610 1535 2906 1.18 (1.02-1.37)

those which presented data onproximity to electrical facilities (12studies).

For the dichotomous exposureclassifications (i.e. exposed vsunexposed), the combined relativerisk for the calculated and measuredmagnetic field data was 1.3 (1.1 – 1.6)which is statistically significant. Thecombined risk for the proximity toelectrical facilities data (table 7) wasfound to be elevated but notstatistically significant at 1.2 (1.0 - 1.6).However Wartenberg reported thatthe difference in design between theseproximity studies was such that thecombined risk was not likely to be anadequate representation of thestudies.

Wartenberg found that in most casesnone of the individual studies weredisproportionately influential. He says"This is important because manypeople believe there are no data tosupport an association betweenresidential magnetic field exposureand childhood leukaemia. To thecontrary, the data strongly andrelatively consistently support such anassociation, although the estimatedmagnitude of the risk is moderate.Limitations due to design,confounding, or other biases maysuggest alternative interpretations."

In conclusion he comments "Overall, Isee largely positive results with smallto moderate effect sizes…. Thesesummaries are unlikely to be changedby additional studies unless thosestudies are extremely large andproduce markedly different results. Ifone chooses to use these summaryestimates for interpretation, given thewidespread exposure to magneticfields they suggest perhaps as much asa 15-25 % increase in the childhoodleukaemia rate, which is a large andimportant public health impact."

16

2.4.4 Draper et al 2005This study merits inclusion since it isthe largest single study of childhoodcancer and power lines, with roughlytwice the number of children livingclose to power lines than the nextlargest study (Feychting and Ahlbom199342). Draper et al 200544 set out todetermine whether an associationexists between distance of homeaddress at birth from high voltagepower lines and incidence ofleukaemia and other cancers inchildren in England and Wales. Theyused the UK Cancer Registry recordsof 29,081 children with cancer,including 9,700 with leukaemia, andNational Grid records to determinethe distance from birth address to thenearest high voltage overhead powerline. For each case a healthy controlwas selected from birth registersmatching sex, date of birth (to within6 months) and birth registrationdistrict.

Table 8. From Draper et al, 2005.Distance of address at birth fromnearest National Grid line for casesand controls and estimated relativerisk

Distance to Leukaemia Controls Relative line (metres) cases risk0-49 5 3 1.6750-99 19 11 1.79100-199 40 25 1.64200-299 44 39 1.16300-399 61 54 1.15400-499 78 65 1.23500-599 75 56 1.36=600(ref. group) 9,378 9,447 1.0

Total 9,700 9,700

Compared with those who lived morethan 600 m from a line at birth,children who lived within 200 m had arelative risk of leukaemia of 1.7 (1.1-2.5); those living between 200 and 600m had a relative risk of 1.2 (1.0-1.5).There was a significant (P < 0.01) trendin risk in relation to the reciprocal ofdistance from the line – i.e. the riskincreased as the distance decreased.

Adjustment for socioeconomic statushad no effect on the relative risks fordistance.

Figure 8. Distance of address at birth from the nearest National Grid line andrelative risk of childhood leukaemia. The approximate EMF exposure levels atdistances up to 200m are shown.

The authors conclude: "There is anassociation between childhoodleukaemia and proximity of homeaddress at birth to high voltage powerlines, and the apparent risk extends toa greater distance than would havebeen expected from previousstudies…. There is no acceptedbiological mechanism to explain theepidemiological results; indeed, therelation may be due to chance orconfounding."

2.5 Epidemiology: conclusions As evident from the above, there hasbeen a wealth of research in this areasince Wertheimer and Leeper firstreported a link between childhoodcancer and electrical wiring. Howeverthe results of individual studies arevariable and there remains a lack ofconsensus as to whether a link actuallyexists.

The limiting factor in most of theindividual studies is the number ofcases available for inclusion.Childhood leukaemia is a relativelyrare disease (affecting some 500children per year in the UK) and it isalso relatively rare for children to beexposed to high levels of EMF

(around 0.4% children in the UKaccording to the NRPB, 2004). Thismeans that most studies have too fewsubjects in the highest exposurecategory to have enough statisticalresolving power to detect even adoubling of risk, let alone a smallereffect.

A number of meta-analyses havepooled data from a number ofindividual studies to increase thestatistical power. This paper outlinesthree of these meta-analyses(Greenland et al, 200041; Ahlbom et al,200042; Wartenberg, 200143). All ofthese studies report a small increase inrisk.

Some of the methodologicalshortcomings of earlier investigationshave been addressed and informationhas been collected on a broad rangeof possible confounding factors in anattempt to rule out any bias.Interestingly, this hasn’t significantlyaltered the relative risks reported butthere is still debate as to whether theresults are real or whether theobserved relationships may reflectconfounding factors or other biases.


As far as exposure metrics areconcerned, all studies have beencriticised to some extent for using aless than perfect exposuremeasurement. However, the nature ofthese problems is such that they wouldnot result in spurious associationsbetween EMF and disease risk – ifanything they would mask realassociations or lead tounderestimations of their magnitude.This also applies to discussions aboutthe ‘period of inquiry’. If an effect isstill found, despite the fact that wemay not be looking at the relevantwindow of exposure, then presumablythe effect found is diluted comparedto that which would be found if weknew what was the relevant window ofexposure.

Hatch et al (2000)45 published a paperin which they set out to identify theeffects of any confounding factors andselection bias which may have beenpresent in their earlier study (Linet etal, 199746) into the associationbetween childhood ALL and wirecodes. They found that when theyincluded rather than excluded partialparticipants (i.e. those who failed tocomplete all phases of the study) andcontrolled for all identifiedconfounding variables the relative riskwas reduced but still elevated (from1.9 to 1.5).

In his 1965 paper "The Environmentand Disease: Association orCausation?" Hill set out a systematicapproach for using scientificjudgment to infer causation fromstatistical associations, listing thecriteria (set out below) to beconsidered when judging whether anobserved association reflects a causalrelationship. Hill’s criteria are widelyused in the evaluation of modernepidemiological research. It isimportant to note, however, thatwhilst each criterion that is metincreases our confidence in claiming acausal relationship, failure to meet allcriteria does not disprove such arelationship.

• Temporal relationship ie. exposurealways precedes the outcome. This isclearly the case here, if exposurehistory precedes the disease.

• Strength. The stronger the association,the more likely it is causal. At higherexposures a doubling of risk hasbeen demonstrated (Greenlandreported a RR of 1.7, Ahlbom 2.0,Wartenberg 1.6 and Draper 1.7 inthe highest versus lowest exposurecategories).

• Dose-response relationship. Anincreasing amount of exposure increasesthe risk. It is important to note thatnot all causal relationships exhibit adose-response. Some factors mayhave a ‘switching’ effect whereexposure over a certain levelswitches on the damaging effect.Others may have damaging‘windows’ of exposure. There issome evidence, however, of a doseresponse relationship between EMFand childhood leukaemia.Interestingly, the strongestevidence of a linear dose responserelationship comes from Draper etal (2005) who reported a significanttrend in risk in relation to thereciprocal of distance from the line(if this can be considered a ‘dose’).Other individual studies and eventhe meta-analyses are limited bysmall numbers of cases in the highexposure groups althoughGreenland found a suggestion of adose response in his analysis.

• Consistency. The association isconsistent when results are replicated indifferent studies using different methods.The results of the individual studiesvary widely but the lack of statisticalpower, combined with the use ofdifferent exposure metrics, in thesestudies is bound to lead to widevariance in the results. Combiningthe data in his 2001 meta-analysishowever, Wartenberg concludedthat ‘the data strongly and relativelyconsistently support such anassociation’.

• Plausibility. The association agreeswith currently accepted understandingof pathological processes. This is

perhaps where the EMF/childhoodleukaemia association has come infor the most criticism since the lackof any established biologicalmechanism by which EMF couldcause childhood leukaemia hasbeen cited by many as a reason forrejecting the association. Howeverit is equally valid to suggest that ademonstrated epidemiologicalassociation should force a re-evaluation of accepted belief.

• Consideration of alternateexplanations. Have researchers ruledout other possible explanations for theresults? Every effort is made tocontrol for confounding factorsand studies have been carried outto assess their possible influence.When controlling for selection biasas well as confounders, Hatch et al(2000)45 reported a decrease inobserved risk from 1.9 to 1.5.

• Experiment. The condition can bealtered/prevented by an appropriateexperimental regimen. It is notpossible to test this in relation tothe childhood leukaemia/EMFassociation.

• Specificity. This is established when asingle putative cause produces a specificeffect. Like all environmentalexposures thought to increase therisk of cancer, including smoking,EMF do not produce leukaemia inall those who are exposed. This iscompatible with existing knowledgewhich suggests that someindividuals are born with apredisposition to developleukaemia, subject to exposure to‘triggering factors’ later in life. EMFmay be such a triggering factor.

• Coherence. The association iscompatible with existing theory andknowledge. In this case, theepidemiological association isperhaps ahead of existing theoryand knowledge when it comes tothe mechanisms of the disease.

3. Biology - Experimental Evidence

18

How could electric and/or magneticfields increase the risk of cancer? Whatcould the mechanism be?

Prompted by Wertheimer and Leeper’soriginal observation in 1979, of anincreased incidence of leukaemia inchildren living near certain types ofpower lines, many hundreds of papershave been published on the biologicaleffects of electric and magnetic fields(EMF). As reviewed earlier in thisreport, a number of epidemiologicalstudies have indicated an associationbetween exposure to EMF and anincreased risk of childhood leukaemia.However, these observations haveraised considerable scepticism in thescientific community due to theabsence of reproducible and relevantlaboratory experimental evidence thatEMF exposure is carcinogenic, and theabsence of any clear demonstratedbiophysical mechanisms through whichEMF could affect biological systems.

This part of this report concentrates onthe experimental evidence of theeffects of EMF in model systems: oncells grown in the laboratory (in vitro)and in animal models (in vivo). Wereview the evidence relating to thethree main areas of EMF research: theeffects of EMF on the genetic materialof the cell – the DNA (3.2); the effect ofEMF on the functions of the cell, inparticular intracellular calciumregulation and protein production(3.3); the effects of EMF on wholeanimals (3.4). Many different celltypes, animal model systems andexperimental methods have been usedand the results are overall contradictoryand inconclusive. Some laboratorieshave found clear evidence for EMFeffects and other laboratories havefailed to reproduce these results andtheir scientists are critical of techniquesused. But, the fact that some effectshave been observed is of concern andneeds further investigation.

3.1 Effects of the electromagneticspectrum on biological materialThe effect of electromagnetic radiationon biological material depends on thefrequency of the source. Theelectromagnetic spectrum produceswaves of energy but can also behave likeparticles or photons - particularly athigh frequencies. The particle natureof electromagnetic energy is importantbecause it is the energy per photon thatdetermines what biological effectselectromagnetic energy will have.

At the very high frequencies, photonshave sufficient energy to breakchemical bonds and are thereforeknown as ionizing radiation. The well-known biological effects of X-rays andgamma radiation are associated withbreaking chemical bonds in molecules.At lower frequencies the energy of aphoton is below that needed to breakchemical bonds. This part of theelectromagnetic spectrum is termednon-ionizing. Because non-ionizingelectromagnetic energy cannot breakchemical bonds there is no analogybetween the biological effects ofionizing and non-ionizingelectromagnetic energy.

However non-ionizing electromagneticfields can produce biological effects.The biological effects of high frequencyelectromagnetic radiation, includingvisible and infrared radiation, dependon the photon energy, and involveelectronic excitation (adding energy toa molecule) rather than ionization.Radio- and microwave-frequencysources can cause effects by inducingelectric currents in tissues, which causeheating. Continuing down thespectrum, EMF have frequencies thatinteract poorly with the bodies ofhumans and animals, and areinefficient at inducing electric currentsand causing heating.

Electric fields associated with powersources have very little ability topenetrate buildings or even skin. Incontrast, magnetic fields associated withpower sources are difficult to shield,and easily penetrate buildings andpeople. Because electric fields do notpenetrate the body, it is generallyassumed that any direct biologicaleffect from residential exposure to EMFmust be due to the magneticcomponent of the field, or to theelectric fields and currents that thesemagnetic fields induce in the body.

Although the EMF photon energy istoo weak to break chemical bonds,there are well-established mechanismsby which EMF could produce biologicaleffects without breaking chemicalbonds48 49 50 51. EMF can exert forces oncharged and uncharged molecules orcellular structures within a tissue. Theseforces can cause movement of chargedparticles, orient or deform cellularstructures, orient dipolar molecules, orinduce voltages across cell membranes.EMF can exert forces on cellularstructures; but since biologicalmaterials are largely non-magneticthese forces are usually very weak.

3.2 Effect of EMF on DNAGenerally, agents that damage thegenetic information within the cell –the DNA - can cause cancer. Thissection of the report reviews the effectsof EMF on DNA. Does EMF cause DNAdamage? Can this damage lead tocancer? This is the largest part of theexperimental biology section as it iswhere most investigators haveconcentrated their efforts. In general,there are many good and reproduciblestudies that demonstrate no DNAdamaging effects of EMF, but there arealso those that do show effects.


3.2.1 Genetic and Promoter pathwaysAs discussed earlier in this report, theinitiating event for childhoodleukaemia appears to be particularDNA translocations (bits of onechromosome attached incorrectly toanother) or hyperdiploidy (multiplecopies of genes) in blood cellprecursors (haematopoetic stem cells)in the foetus. These cells divide rapidlyand frequently and are particularlysusceptible to DNA damaging agentsand to natural errors. One hundredtimes more infants are born with theseDNA changes than go on to developchildhood leukaemia – so there mustother additional biological events tocause clinical leukaemia – the "two hithypothesis". It is possible that exposureto EMF may play a part in either orboth of these events in some children.

There appear to be at least twopathways to the development of cancerand these are not mutually exclusive:

i) A genetic pathway where a single or aseries of very specific damages to theDNA occurs in a cell. This could becaused by exposure to anenvironmental agent or as a result ofnatural error-prone processes.Gamma radiation is a knowngenotoxic agent. The huge varietyand power of the immune system isgenerated by rearranging genes inimmune cells in early infancy and isnecessarily an error prone processwhich often goes wrong. The cellswith damaged DNA either die or theDNA is repaired. If the DNA isrepaired correctly there are nofurther problems but any unrepairedor misrepaired damage will lead tochanges in chromosomes (deletionsor hyperdiploidy), micronuclei(fragmentation of the nucleus),sister chromatid exchanges(translocations), and mutations,some of which may lead to thedevelopment of cancer.

ii) A Promoter pathway whereenvironmental agents, which maynot be genotoxic or carcinogenic bythemselves, can contribute to cancerby increasing the genotoxic potentialof other agents, interfering with theDNA repair processes, allowing a cellwith DNA damage to survive andstimulating the cell division resultingin alteration of the normal functionsof the cell52. The evidence for thepromoting action of an agent,especially for its relevance to humancarcinogenicity under real lifeexposure conditions, must beevaluated carefully since very fewenvironmental agents are known topromote carcinogens53.

There are more than 150 peer reviewedscientific publications on the effects ofEMF on DNA in animal model systemsand on cells grown in the laboratory. Adetailed review of all of these papers isbeyond the scope of this report, whichwill summarise five, representative,recently published reviews:

McCann et al 199354 reviewed 55 studiesand concluded that "the preponderanceof evidence suggests that neither EMFnor static electric or magnetic fieldshave a clearly demonstrated potential tocause genotoxic effects. However, theremay be genotoxic activity from exposureunder conditions where phenomena

auxiliary to an electric field, such asspark discharges, electrical shocks, orcorona can occur."

Murphy et al 199355 stated that"although most of the available evidencedoes not suggest that electric and/ormagnetic fields cause DNA damage, theexistence of some positive findings andlimitations in the set of studies carriedout suggest a need for additional work."

Moulder 199856 summarised that "thereare approximately 100 publishedreports that have looked for evidencethat power frequency fields havegenotoxic or epigenetic activity. Thesestudies have found no replicatedevidence that power frequency fieldshave the potential to either cause orcontribute to cancer."

McCann et al 199857 added a further 23studies to their original 55 andconcluded that "…the preponderanceof evidence suggests that EMF do nothave genotoxic potential. Nevertheless,a pool of positive results remains, whichhave not yet been tested byindependent replication."

Vijayalaxmi & Obe 200558 say "Amongthe total of 63 reports published during1990–2003, the conclusions from 29investigations (46%) did not identifyincreased cytogenetic (chromosomal)

CarcinogenesisGenetic or Initiators

Agents that cause direct genetic damageusually to DNA that may predispose cells tobecoming cancerous

Promoters

Agents that cause cancer by some othermechanism other than direct DNA damageand by promoting the effects of carcinogens

Figure 9. Mechanisms of Carcinogenesis.

EMF could act as a Genetic carcinogen or be a Promoter.

20

Table 9. Effects of EMF on DNA single and double strand breaks and theirrepair in mammalian cells (from Vijayalaxmi & Obe, 2005).Study Cells used Damage

Whole body exposure: animalsLai et al 1997a Rat, whole brain IncreasedLai et al 1997b Rat, whole brain IncreasedSingh et al 1998 Rat, whole brain IncreasedSvedenstal et al 1999a Mice (outdoor) brain cortex IncreasedSvedenstal et al 1999b Mice (laboratory) brain cortex IncreasedMcNamee et al 2002 Mice, brain cerebellum Not increasedIn vitro: human cellsFiorani et al 1992 Human tumour cells Not increasedAhuja et al 1997 Human blood lymphocytes IncreasedAhuja et al 1999 Human blood lymphocytes IncreasedPacini et al 1999 Human neuronal cells Not increasedKindzelskii et al 2000 Human blood lymphocytes IncreasedMaes et al 2000 Human blood lymphocytes Not increasedIvancsits et al 2002 Human skin fibroblasts Increased

Table 10, summarises the findings ofstudies looking at chromosomalaberrations, micronuclei and sisterchromatid exchanges (translocations)in cells exposed to EMF.Chromosomal aberrations,micronuclei and sister chromatidexchanges are all indications ofserious damage to DNA – usuallycaused by the normal DNA repairprocess going wrong. This type ofdamage is very dangerous and verysimilar to many of the translocationsseen in cells from children withleukaemia. In general, theseaberrations are rarely seen aftertreatment with EMF. One group,Scarfi et al, sometimes find changesand at other times do not, indicatingthe difficulties inherent in replicatingthese experiments.

damage following EMF exposure per sewhile those from 14 studies (22%)indicated a genotoxic potential of EMFexposure. The observations in 20 otherreports (32%) were inconclusive.Among the 23 combination exposureinvestigations (i.e. where cells have beenexposed to EMF in combination with aknown carcinogen), the data from 10studies did not identify epigeneticeffects of EMF, while the data from onestudy indicated such an effect. Theresults from 12 other reports wereinconclusive." This data is reproducedin table form in this report (tables 9, 10and 11 overleaf)

Vijayalaxmi & Obe58 continue"Considering the ‘weight of scientificevidence’ approach for genotoxicityinvestigations, as adopted by IARC(2002), the preponderance of data thusfar available in the literature shows thatEMF exposure per se is not genotoxic(and little evidence for epigeneticinfluences) in mammalian cells.However, research must continue toresolve the controversial data publishedin the literature."

All of these reviews pointed tounconfirmed and/or inconclusivepositive reports and suggestedadditional research.

3.2.2 Vijayalaxmi and Obe, 2005A detailed review of the literature onEMF and DNA damage is beyond thescope of this report which presents datataken from the 2005 review byVijayalaxmi & Obe58 as an example orsnapshot of the available data.

Vijayalaxmi & Obe present data fromexperiments in different laboratories inwhich mammalian cells were exposed invivo and in vitro to EMF and the effectson DNA damage measured.

The three tables which followsummarise the evidence for differenttypes of cell damage. The term"increased", "no increased" or"inconclusive (increase and decrease)"for each type of damage indicates acomparison of the data in samplesexposed to EMF with those observed insham exposed and/or unexposedcontrols. The full references for eachstudy can be found in Vijayalaxmi &Obe58.

Table 9, summarises the findings ofstudies looking at single and doublestrand breaks in the DNA of cellsexposed to EMF. The level of DNAstrand breaks is important as it canrepresent the level of damage done tothe genetic material of the cell. (It canalso represent the level of cell division).Any damage is usually repaired by thecell but mistakes can happen. Genescan be changed and very occasionallythe change is sufficient to produce acancer cell. Some laboratories find aneffect of EMF on DNA strand breaksand others do not. Where increaseshave been found these results have notbeen repeated in other laboratories.


Table 10. Effects of EMF on chromosomal aberrations, micronuclei and sisterchromatid exchanges in mammalian cells (from Vijayalaxmi & Obe, 2005).Study Cells examined Damage

Whole body exposure: animalsZwingberg et al 1993 Rat, blood lymphocytes Not increasedSingh et al 1997 Mice, blood erythrocytes Not increasedSvedenstal et al 1998 Mice, blood erythrocytes Not increasedWhole body exposure: humansCiccone et al 1993 Human blood lymphocytes Not increasedSkyberg et al 1993 Human blood lymphocytes Not increasedKhalil et al 1993 Human blood lymphocytes InconclusiveValjus et al 1993 Human blood lymphocytes InconclusiveSkyberg et al 2001 Human blood lymphocytes Not increasedIn vitro: animal and human cellsGarcia-Sagredo et al 1990 Human blood lymphocytes Not increasedGarcia-Sagredo et al 1991 Human blood lymphocytes InconclusiveKhalil et al 1991 Human blood lymphocytes InconclusiveLivingston et al 1991 Human blood lymphocytes Not increasedLivingston et al 1991 Chinese hamster ovary cells Not increasedNordenson et al 1994 Human amniotic cells InconclusiveScarfi et al 1994 Human blood lymphocytes Not increasedAntonopoulos et al 1995 Human blood lymphocytes Not increasedD'Ambrosio et al 1995 Human blood lymphocytes InconclusiveGalt et al 1995 Human amniotic cells Not increasedPaile et al 1995 Human blood lymphocytes Not increasedJacobson-Kram et al 1997 Chinese hamster ovary cells Not increasedScarfi et al 1997a Human blood lymphocytes IncreasedScarfi et al 1997b Human blood lymphocytes IncreasedScarfi et al 1997c Human blood lymphocytes Not increasedSimko et al 1998a Human tumour cells InconclusiveSimko et al 1998a Human amniotic cells InconclusiveScarfi et al 1999 Human blood lymphocytes Not increasedZeni et al 2002 Human blood lymphocytes Increased

Table 11. Genotoxic potential of EMF on mammalian cells exposed in vitro togenotoxic mutations (from Vijayalaxmi & Obe, 2005). Study Cells used Damage

DNA strand breaks and repairAnimal and human cellsFrazier et al 1990 Human blood lymphocytes Not increasedFairbairn et al 1994 Human tumour and blood cells Not increasedCantoni et al 1995 Chinese hamster cells Not increasedCantoni et al 1996 Chinese hamster cells Not increasedMiyakoshi et al 2000 Human tumour cells InconclusiveZmyslony et al 2000 Rat blood lymphocytes Inconclusive

Chromosomal aberrations, micronuclei and sister chromatid exchangesAnimal cellsOkongi et al 1996 Chinese hamster lung cells Not increasedLagroye et al 1997 Rat tracheal epithelial cells IncreasedYaguchi et al 1999 Mouse embryonic skin cells InconclusiveYaguchi et al 2000 Mouse embryonic skin cells InconclusiveSimko et al 2001 Syrian hamster embryo cells InconclusiveNakahara et al 2002 Chinese hamster cells InconclusiveHuman cellsScarfi et al 1991 Human blood lymphocytes Not increasedScarfi et al 1993 Human blood lymphocytes Not increasedHintenlang et al 1993 Human blood lymphocytes InconclusiveTofani et al 1995 Human blood lymphocytes InconclusiveSimko et al 1998b Human amniotic cells InconclusiveSimko et al 1999 Human amniotic cells InconclusiveMaes et al 2000 Human blood lymphocytes Not increasedHeredia-Rojas et al 2001 Human blood lymphocytes Not increasedHone et al 2003 Human blood lymphocytes Not increasedCho et al 2003 Human blood lymphocytes InconclusiveVerheyen et al 2003 Human blood lymphocytes Inconclusive

Table 11, brings together all thestudies on the effects of EMF onanimals treated with a known cancercausing agent – looking for anepigenetic or cancer-promoting effect.Most of the studies were inconclusiveand no clear effect of EMF could bedemonstrated.

And finally, Table 12, overleaf, is asummary of the 63 studies reviewed byVijayalaxmi & Obe showing that 29 ofthe studies reported no increaseddamage as an effect of exposure toEMF, 14 showed clear effects and in20 studies the results wereinconclusive. Out of the 14 studiesshowing clear effects, 11 reportedevidence of increased DNA single anddouble strand breaks and repair.

3.2.3 The REFLEX projectThe REFLEX project59("RiskEvaluation of Potential EnvironmentalHazards From Low EnergyElectromagnetic Field Exposure UsingSensitive in vitro Methods") was fundedby the European Union under theprogramme Quality of Life andManagement of Living Resources.

The main goal of this project, whichran from 2000 to 2004, was toinvestigate the effects of EMF on singlecells in vitro at the molecular levelbelow the present safety levels. Theproject was designed to assess whetheror not any of the disease-causingcritical events (gene mutations,deregulated cell division andsuppressed or exaggeratedprogrammed cell death [apoptosis])could be induced in living cells by EMFexposure. The authors’ introductionsuggested that they set out to provide adefinitive answer "Failure to observethe key critical events in living cells invitro after EMF exposure would suggestthat further research efforts in this

22

field could be suspended and financialresources should be reallocated for theinvestigation of more importantissues."

Twelve laboratories from sevenEuropean countries contributed to thestudy. In order to compare the resultsof investigations carried out in thedifferent laboratories and to ensure theconclusiveness of the data obtained inthe studies, they took steps to ensurethat the conditions of exposure werestrictly controlled across theparticipating laboratories.

The data obtained in the course of theREFLEX project showed that EMF hadgenotoxic effects on primary cellcultures of human skin cells(fibroblasts) and on other cell lines.These results were obtained in twolaboratories and confirmed in twoadditional laboratories outside theREFLEX project, while no such effectscould be observed in a furtherlaboratory.

• Exposure to EMF down to 35 µTgenerated DNA strand breaks infibroblasts at a significant level.There was a strong positivecorrelation between both theintensity and duration of exposureto EMF and the increase in singleand double strand DNA breaks andmicronuclei frequencies.

Surprisingly this genotoxic effect wasonly observed when cells wereexposed to intermittent EMF, butnot to continuous exposure.

• Responsiveness of fibroblasts to EMFincreased with the age of the celldonor and in the presence ofspecific genetic repair defects.

• The effect differed among the othertypes of cells examined. Inparticular, cells of the immunesystem (lymphocytes) from adultdonors were unaffected by EMF.

• Chromosomal aberrations were alsoobserved after EMF exposure ofhuman fibroblasts.

• No clear effects of EMF were foundon DNA synthesis, cell cycle, celldifferentiation, cell proliferation andapoptosis (described later) but therewas some suggestion that EMF mayactivate several groups of genes thatplay a role in cell division, cellproliferation and celldifferentiation.

The authors reported that "Genotoxiceffects and a modified expression ofnumerous genes and proteins afterEMF exposure could be demonstratedwith great certainty, while effects on cellproliferation, differentiation andapoptosis were much less conclusive.Since all these observations were madeusing in vitro studies, the results neitherpreclude nor confirm a health risk dueto EMF exposure, but they supportsuch a possibility."

The authors concluded "Takentogether, the results of the REFLEXproject were exclusively obtained in invitro studies and are, therefore, notsuitable for the conclusion that EMFexposure below the presently validsafety limits causes a risk to the healthof people. They move, however, suchan assumption nearer into the range ofthe possible. Furthermore, there existsno justification any more to claim thatwe are not aware of anypathophysiological mechanisms whichcould be the basis for the developmentof functional disturbances and anykind of chronic diseases in animal andman".

So far, the only data from the REFLEXproject that has been published in peerreviewed journals is the work of oneinvestigator Ivancsits et al (200260,200361 62, 200563 64) who exposed humancultured fibroblasts to EMF andobserved DNA strand breaks withintermittent but not continuousexposure. Crumpton & Collins65 arecritical of the methods used to measureDNA strand breaks as they would alsoinclude cells undergoing cell division.From a cell biologist’s viewpoint,cultured fibroblasts are unusual cells.No tumour has ever developed from afibroblast and continuously culturedcells generally have little relevance tocells in vivo.

3.2.4 Effects of EMF on DNA:Conclusions There are many good andreproducible studies that demonstrateno genotoxic effect of EMF, but thereare also those that do show effects.This presents difficulties for thescientific community and there is stillconsiderable controversy amongst thescientists working in this area.Discussions range across the relevanceof the cells used for the experiments tothe methods used to estimate DNAdamage to the often very high levels of

Table 12. Summary of studies looking at DNA damage in cells exposed to EMF(from Vijayalaxmi & Obe, 2005).Test system employed No increased Increased Inconclusive Total no.

damage damage damage of studies

DNA single and double strand breaks and repairWhole body: animals 1 5 0 6In vitro: human cells 3 6 0 9In vitro: EMF + genotoxic mutations 4 0 2 6Chromosomal aberrations, micronuclei, and sister chromatid exchangesWhole body: animals and humans 6 0 2 8In vitro: animal and human cells 9 2 6 17In vitro: EMF + genotoxic mutations animal cells 1 1 4 6In vitro: EMF + genotoxic mutations human cells 5 0 6 11Total 29 14 20 63Percentage, % 46 22 32


magnetic fields used. Effects, quitereproducible in one laboratory, oftencan’t be repeated elsewhere. However,it cannot be ignored that in somelaboratories EMF do cause genotoxicdamage. Clearly more work needs tobe done but perhaps the questionsneed to be asked in a different way.

The finding reported in the REFLEXstudy that the genotoxic effect was onlyobserved when cells were exposed tointermittent rather than continuousEMF highlights the importance ofmetrics, raised in Section 2. Moreresearch is needed to establish whichare the most damaging aspects of EMFexposure. It does appear that it is notjust intensity and duration of exposurewhich we should be considering.

3.3 Effect of EMF on cell functionThis section of the report looks at theeffects of EMF at the cellular level. DoEMF interfere with normal cellularfunctions? Do they change the levelsof important cellular proteins or ionsand consequently affect cell functionand cause cancer cells to develop? Aswith the effects of EMF on DNA, theexperimental data is contradictory andinconclusive. Effects are observed butoften other laboratories are unable toreproduce these same effects.

Multi-cellular organisms (like humans)remove unnecessary and damaged cellsby intentional or programmed celldeath – a process known as apoptosis.If this normal biological cleaning-upprocess is impaired in some way thendamaged cells can survive and go on tobecome cancer cells. Apoptosis, andmost other cellular functions, areeffected by molecular messages beingpassed through the cell (known asintracellular signalling pathways).These signalling pathways can beinterfered with at numerous points. Ifthe natural process of cell death isprevented then a damaged cell will

survive and may go on to become acancer. There is evidence that EMFexposure has an effect on signallingpathways although this evidence iscontroversial and the effects have beendifficult to reproduce.

3.3.1 Intracellular calcium ion levelsCalcium ions (Ca2+) within the cell playa key role in intracellular signallingand cell death (apoptosis). Thecalcium levels within a cell ultimatelydetermine the rate of cell division. Anyagent that affects the levels of calciumions within a cell would have profoundeffects on that cell's function.

Work on the effects of EMF on calciumlevels in cells of the immune system(lymphocytes) was reviewed byWalleczek et al in 199266. Theyreviewed work from 9 laboratories onthe effects of EMF on Ca2+ metabolismand concluded "it is proposed thatmembrane mediated Ca2+ signallingprocesses are involved in the mediationof field effects on the immune system".

Liburdy et al 199267 and 199368 foundraised levels of intracellular calciumafter 22mT (= 22,000 µT) magneticfield exposure using stimulated humanlymphocytes. Lindstrom et al 199369

found intracellular calcium changeswere induced in a T-cell line (humanlymphocytes modified to growcontinuously in culture) by a 100 µTmagnetic field and later showed a doseresponse, with no effect below 40 µTand a plateau at 150 µT. But, theseresults could not be reproduced byWey et al 200070. Galvanovskis et al199971 reported that Ca2+ changes inhuman leukaemia T-cells are reducedby 50 Hz magnetic fields.

Attempts to reproduce the effects ofEMF on intracellular calcium levelshave been unsuccessful by Prasad et al199172, Coulton and Barker 199373 and

A review of the scientific evidence · 2020. 12. 7. · 3.2 Effect of EMF on DNA 19 3.2.1 Genotoxic...

Documents

Transcript of A review of the scientific evidence · 2020. 12. 7. · 3.2 Effect of EMF on DNA 19 3.2.1 Genotoxic...