Sunlight & Cancer

7/30/2019 Sunlight & Cancer

1/13

REVIEW PAPERS

Sunlight andcancer Cancer Causes and Control. Vol 8. 1997

DallasR English, BruceK Armstrong, AnneKricker,andClaireFleming

(Received3 June1996; accepted in revised form19August 1996)

Epidemiologicevidenceontherelationbetweensunlightandcancerisreviewed.Strongevidenceimplicatessunlight

asacauseofskincancer,although,for melanomaandbasal cell carcinoma,therelationship iscomplex.Bothtypes

ofcancer areassociatedmorestronglywithnonoccupational exposurethan withoccupational exposure,andthe

patternandamountofexposureeachappeartobeimportant.Squamouscell carcinomaappearstoberelatedmore

strongly to total (i.e., both occupational andnonoccupational) exposureto thesun.Theevidencethat sunlight

causesmelanomaoftheeyeisweak.Itshowsnolatitudegradientandtheresultsofcase-controlstudiesareconflicting.

Thereisinadequateevidencetosuggestthatsunlightdoesordoesnotcauseanyothertypeofcancer.Cancer Causes

and Control 1997, 8, 271-283

Key words: M elanoma, skin neoplasms, sunlight.

Introduction

Strictly speaking, sunlight is visible light from the sun.Weareusing theterm loosely here to refer to theultra-violet (UV) radiation(wavelengthsfromabout295to400nm)andvisiblelight(400to780nm)thatreachthesurfaceof theearth.UV radiationof lessthanabout295nmdoesnot reachtheearthssurfacebecauseit isabsorbed by theatmosphere.

UV radiation is probably responsible for all thecarcinogenic effectsof sunlight. UVB (280 to 315nm) ismuch more effective at producing cancer in animals,

erythema(sunburn) in humans, and DNA damage, thanis UVA (315 to 400 nm) wavelengths of about 340nmor moreare less than 1/1,000 times as potent in causingcancer in experimental animals as wavelengths of about295 nm.1 It is difficult to separate the effects of UVB,UVA, and visible light in epidemiologic studies; thus,epidemiologic studies generally deal with sunlight as awholerather than with any of its components.

Six categories of epidemiologic evidence are relevantto theproposition that sunlight causesskin cancer.Theyare that these cancers are: (i) more frequent in residentsof areasof highambientsolar irradiance; (ii) morefrequentinsun-sensitivepeople;(iii)occurmainly onsun-exposedbody sites; (iv) more frequent in people with high sunexposure; (v) more frequent in peoplewith benign sun-relatedskin conditions;and(vi) reducedbyprotectionoftheskin againstthesun.Evidencecanbegainedwithin eachof thesecategoriesbybothdescriptiveandanalyticalstudies.

Wereviewheretheepidemiologyof cancersof theskin melanoma, and the nonmelanocytic skin cancerssquamous cell carcinoma(SCC) and basal cell carcinoma(BCC) andmelanomaof theeye.Wealsobriefly discussother neoplasms for which relationships with sunlighthavebeen postulated. Our aim is to produceasynthesisof theevidencerelating sunlightto cancer andto identifyareas for further research.

Cancer Causesand Control, 1997, 8, pp.271-283

Dr English is with the Department of Public Health, The University of Western Australia. Dr Armstrong and Ms Fleming are with

the NSW Cancer Council, Woolloomooloo, NSW, Australia. Dr Kricker is with the National Breast Cancer Center, Woolloomooloo,

NSW, Australia. Address correspondence to Dr English, Department of Public H ealth, University of Western Australia, Perth, WA

6907, Australia. Financial support was provided, in part, by the Cancer Foundation of Western Australia, the Health Promotion

Foundation of WA, and theAustralian Rotary H ealth Research Fund.

1997Rapid SciencePublishers Cancer Causes and Control. Vol 8. 1997 271


2/13

Melanomaoftheskin

Descriptivestudies

Ethnic origin. Melanoma is predominantly a disease ofpeopleof European origin. In theUnited States, its inci-denceissome20-foldhigher inWhitesthanBlacks.2Ratesof melanoma are also very low in Asians in theUS andvery low in Asia.2

Placeof residence.Theincidenceof melanomaisgreat-est in theWhitepopulationof Australia, asun-sensitivepopulation in an area of high ambient sunlight2 andincreaseswith proximity to theequator in many popu-lations.3,4 In Europe, however, incidenceof melanomais higher in Norway and Sweden in thenorth than in

France, Italy, and Spain in the south.

2

This apparentanomaly may be due to the tendency for skin colorin Europeto increasein darknesswith increasingprox-imity to the equator.3

Migration. When people of European origin migratefrom areasof low incidenceof melanomaandgenerallylow ambient sunlight to areas of high incidence andhigher ambient sunlight, their subsequent rates ofmelanomaaregenerally higher thanthosein their homecountry and less than those in thehost country.5,6 Age

at arrival of these migrants in the host country is apowerful predictor of subsequent risk of melanoma

the older the age at arrival, the lower the incidence.5,7

However, it is not possibleto distinguish ageat arrivalfrom duration of residence in these studies.

Occupation. I ncidence and mortali ty rates ofmelanomaare generally higher among indoor workersthan outdoor workers,8-10 which is not consistent witha simple relationship between sun exposure andmelanoma. They are also higher in people of highsocioeconomic status than of low socioeconomic status.In theabsenceof any plausiblealternativeexplanation,this relationship with socioeconomic status is pre-

sumed to be due to complex effects of sun exposure.

Anatomic site. In a recent Australian survey,11 mela-noma had its highest density on the usually exposedparts of the head and neck and its lowest density onrarely exposed sites (buttocks and abdomen in bothgenders and scalp in women). I ts density was low onthe forearms, backs of hands, upper arms and lowerlimbs, and intermediate on the sometimes exposedshoulders and back in both genders and the chest inmales (Table 1).

Table 1. Surface density of occurrence of cutaneous melanoma, basal cell carcinoma (BCC), and squamous cell carcinoma

(SCC) by body site, relative to a density of 1.0 for the whole body in whole population seriesa

Melanoma BCC SCC

Level 1 Level 2+

Men

Usually exposed

Scalp, face, neck, ears 4.9 2.0 3.9 6.5

Forearms, backs of hands 0.8 0.5 1.4 2.0

Occasionally exposed

Shoulders, back, chest 1.7 2.5 1.8 0.3

Upper arms, lower limbs 0.3 0.5 0.4 0.3

Rarely exposedAbdomen, buttocks 0.2 0.2 0.0 0.0

Women

Usually exposed

Face, neck 7.6 2.5 6.7 8.9

Forearms, backs of hands 0.8 0.6 1.3 3.3

Occasionally exposed

Shoulders, back 0.9 1.8 1.8 0.2

Upper arms, lower limbs 0.8 1.1 0.5 0.4

Rarely exposed

Scalp, ears, chest, abdomen, buttocks 0.3 0.3 0.2 0.4

aData on melanoma are from Green et al;

11data on BCC and SCC are unpublished observations based on incident cases

from Geraldton, Western Australia.

D.R. English et al

272 Cancer Causes and Control. Vol 8. 1997


3/13

Analytic studies

Sensitivityof skin tosunlight. Peoplewho sunburn easilyandtanpoorly areat increasedrisk of melanoma.12Intwo

recent case-control studies from thewestern US,13,14

riskrelative to that in those who developed a deep tanfollowingchronic sun exposureincreased steadily to 9.0(95 percent confidenceinterval [CI] = 3.8-21.1) in thosewho developed no tan or only freckles and to 2.6(CI =1.6-4.1) in those who burnt without tanning followingacuteexposure.

Ambient sunlight at places of residence. Associationsbetween melanomaand measures of ambient sunlightat places of residence and measures of personal expo-sure to sunlight in case-control or cohort studiespublished since about 1980 are summarized in Table2. Thistableis based on recent reviews12,15 and recentlypublished studies.14,16-23

Eightout of 10studiesshowed significant associationsbetween risk of melanomaand measuresof ambientsun-light at places of residence (e.g., daily hours of brightsunlight at places of residence and history of residencecloser toor farther fromtheequator thanthestudyarea).

Therelativerisks(RR) for thehighestexposurecategoriesvaried between 1.7 and 8.0. The remaining two studiesshowed weakly positiveassociations.Thestrongestasso-ciation (RR = 8.0, CI = 2.0-34.7) was in US veterans ofWorld War I I who served in thetropicscompared with

other men of draft agewho didnot serveor didnot servein thetropics.24

Exposure of skin to sunlight.

Lifetime total exposure. By total exposure we meanoccupational andnonoccupational exposuretogether.Sixcase-control studies of melanoma have attempted toestimate lifetimetotal exposure to sunlight (Table 2). Inthree,23,25,26 statistically significant positive associations

were observed, and a statistically significant negativeassociation was observedin one. Thethree positive asso-ciations werefromstudies in Franceand Spain.

Recent total exposure. The measures of recent orusual total exposure to sunlight are a heterogeneousand often crude group mainly concentrated on sunexposure in the 10 to 20 years before diagnosis ofmelanoma. Studies using them have found no consis-tent pattern of association with melanoma (Table 2).

Occupational exposure.Twenty case-control studiesand one cohort study have examined the relationshipbetween estimated occupational exposure to sunlightand melanoma; they haveshown negative associationsas often aspositiveones (Table2). Two of thestudies

showingpositiveassociations were based on an analy-sis of occupational titles, not on specific estimation ofoccupational exposure. The remaining three studieswere all fromEuropeand included twoof thesouthernEuropean studies23,26 in which positive associationswith lifetimetotal sun-exposure were observed.

Someunderstanding of theseseemingly contradictoryresults isofferedby studieswhichsuggest that melanomaon body sitesthat usually arecovered ishighest in officeworkers, whereas the incidenceon sites that usually areexposedisashighor higher in outdoor workers.27,28Thus,contradictory resultswell may ariseas aresult of differ-encesin thedistributionsof indoor andoutdoor workersin the population, and resulting differences in distribu-tions of melanomaby body site.

Nonoccupational exposure. A majority of studies ofnonoccupational exposure to sunlight and melanomahas observed statistically significant, positive associa-tions(Table2). Theseresultshavecomefromavarietyof different measures of exposure such as time spentsunbathing, time spent in sunny vacations, estimated

Table 2. Summary of results of case-control or cohort studies of sunlight exposure and melanoma reported from 1980 to 1996a

Mea sur e of s un ex po sur e Number of studies As soc ia tio ns be twe en s unli ght an d me la noma

Number positive Number null Number negative

Ambient sunlight at places of residence 10 8 2 0

Personal exposure to sunlight

Lifetime total exposure 6 3 2 1

Recent or usual total exposure 6 1 4 1

Occupational exposure 20 5 10 5

Nonoccupational exposure 21 12 8 1

History of sunburn 24 21 3 0

Other sunlight-related skin damage 8 7 1 0

aTable includes results of studies summarized in reviews by the International Agency for Research on Cancer,

15Armstrong

and English,12

and those of other recently published studies.

Sunlight and cancer

Cancer Causes and Control. Vol 8. 1997 273


4/13

hours of nonoccupational sun exposure in summer,time spent in particular outdoor recreations (fishing,boating). Only onestudy found asignificant, negativeassociation, and all but two of the remaining studies

found RRs for the highest exposure categories thatweregreater than 1.0. SomeRRswerevery high eightor greater.

Despite thedifficulty presented by disparatemeasuresof exposure amongdifferent studies, onegroup of inves-tigators29 has attempted a pooled analysis of publishedresults with respect to nonoccupational (intermittent)and occupational (chronic) exposureto sunlight. In sevenpopulation-basedstudies,whichshowedlessheterogene-ity inRR estimatesthanhospital-basedstudies,thepooledestimatesof RR for melanomawere1.6(CI = 1.3-1.9) forthehighestcategory of nonoccupational exposureand0.7

(CI = 0.6-0.9) for the highest category of occupationalexposure.

History of sunburn. Almost all studies that haveinvestigated it have found a positive associationbetween past history of sunburn and risk of melanomawith RRs typically abovetwo and up to 12 (Table2).

Other sunlight-related skin damage. Exposure tosunlight is believed to cause nonmelanocytic skincancers, as reviewed here, and a range of benign skinconditions (e.g., solar elastosis, solar lentigines, solarkeratoses).30 Presence or history of these conditionsoften has been documented in case-control studies ofmelanoma; positive associations have almost alwaysbeen found (Table 2).

Exposure-responserelationship. What few dataexist onthe exposure-response relationship for sunlight andmelanoma suggest that risk increases sharply at lowexposures, reaches a plateau and, if anything, falls athigher exposures.31,32 The matter is complicated bypossible contributions of both amount and pattern ofexposure to causing the disease and confoundingbetween these two. A s Elwood and Gallagher32 have

pointed out: I t may therefore be inappropriate toproduce a unified dose response curve for the wholerelationship. Intermittentandconstantexposure(amountand pattern) may be intrinsically different, withconflicting effects, so that the risk for an individualdependson thebalancebetween thesetwo exposures.

Is total amount of exposure important? Therelation-ship between ambient sunlight and melanoma suggeststhat, given a particular pattern of exposure, as amountincreasesrisk increases.That melanoma ismost dense onsome more-or-less continuously exposed sites and thatmelanoma on exposed sites appears to increase with

increasing occupational exposure also suggests that risk

increases with increasing total amount of exposure. Onthe other hand, melanoma is associated strongly withnonoccupational exposureto thesun and sunburn, bothof which probably reflect an intermittent pattern of

exposure,andrisk overall generally doesnotriseandevenmayfall asoccupational exposureincreasesandincreasingamount of exposureis traded off against fallingintermit-tencyof exposure.Thus,pattern of exposurealsoappearsto beimportant.

We now postulate that, given a particular pattern ofexposure to sunlight, risk of melanoma increases withincreasing amount of exposure and, given a particularamount of exposure, risk increases as exposure becomesmoreintermittent.Theindependenceof effectsofamountand pattern of exposureand theshapesof their relation-ships with melanoma have not yet been shown

empirically.

Protection against sunlight. Most case-control andcohort studies have found positiverather than negativeassociations between sunscreen use and cutaneousmelanoma30,33 which have not been eliminated bycontrol for sun sensitivity and sun exposure. Tworecent case-control studies, however, have observedprotectiveeffects. In astudy fromthesouth of Spain,23

theRR, adjusted for sun exposure and sun sensitivity,fell from 1.0 in those who never used sunscreens to0.6 (CI =0.3-1.4) in occasional users and 0.2 (CI =0.0-0.8) in those who always used them when in thesun.Similarly,amongwomen in theSan Francisco Bayarea (California, US) in the 1980s,14 those who neverused sunscreens had an RR for melanoma of 2.3 (P =0.001) when compared with those who always usedthem, adjusted for sun exposure, sunburn, and hostfactors. Lack of information on the sun protectionfactors of sunscreens used in these various studiesmakes it difficult to explain the inconsistencies.

Basal cell carcinomaoftheskin

Basal cell carcinoma (BCC) is more common than

squamous cell carcinoma (SCC) of the skin. These twotypes of nonmelanocytic skin cancer have often beengrouped in epidemiologic studies, although recent evi-dencesuggeststhat they may differ in their relationshipto sunlight. Their epidemiology has been reviewed indetail by Kricker et al.34

Descriptive studies

Ethnicorigin.Nonmelanocyticskin cancer israrein non-Whitepopulations. InaUSsurvey conductedin 1977-78,theincidencerateof SCC and BCC combined was 232.6per 100,000 person-years in Whites but only 3.4 among

Blacks.35

In populationsof mainly European origin, inci-

D.R. English et al



5/13

dence rates are lower in people with ethnically darkerskins.35,36

Place of residence. I n Australia and the US, incidence

is greatest in regions closest to the equator.35,37 Theincidenceof BCC in Australiain 1990 increased three-fold from latitudes south of 37Sto latitudes north of29S.37

Occupation. Early reports associated nonmelanocyticskin cancer with outdoor occupation.38-40 More recentand more rigorous studies have shown much lessevidence of such an association.27,28 N one of thesestudies has distinguished between BCC and SCC.

Migration. In 1990, the incidence of BCC in migrants

to Australia was half that in those born in Australia.37

Most migration to Australia is from areas of lowerambient sunlight.

Anatomic site. Table 1 shows the relative density onthe skin of incident cases of BCC in a recent whole-population study in Western Australia. The highestdensity of BCC was on the usually exposed sites andthe lowest on the rarely exposed sites. Intermediatedensitiesoccurred on sometimesexposed sites, includ-ing the trunk.

Analytic studies

Earlier cross-sectional and case-control studies of BCCgenerally were not population-based and had a numberofmethodologicdifficulties,includingfewsubjects,crudemeasurement of exposure to sunlight, and inadequatecontrolofconfounding.34Thefollowingreview,therefore,is based on themorerecent, population-based studies.

Sensitivity of skin to sunlight. RRs of 2.0or morehavebeen found for BCC with a skin that burns ratherthan tans.36,41-43 L ight skin color was associated signifi-cantly with BCC in one study only.44

Consistent with the descriptiveobservations referred

to above, case-control studieshaveshown that peopleofsouthern-European ethnic origin born in Australia orCanadahavehalf or less therisk of BCC of people ofother (generally lighter skinned) ethnic origins born inthesamecountry.36,44

Ambient sunlight at places of residence. In WesternAustralia, risk of BCC increased with increasingintensity of ambient solar radiation at all places ofresidence.45 In a cohort study of US nurses,43 thosewho had lived in California or Florida (US) hadincreased rates of BCC compared with thosewho had

lived in northeastern states.

Exposure of skin to the sun.

Total exposure. The evidence linking reported total(occupational and nonoccupational) sun exposureto BCCis weak: of five studies,42-47 none showed a statisticallysignificant, positiveassociation. Earlier reportsof strongassociationscamefrom studies in whichconfoundingoftheassociation by ageand gender waslikely.34

Only theWestern Australianstudy48hasexaminedsunexposureto thesiteof theskincancer.Whileriskof BCCdid not correlatewell with total site-specific exposuretothesun, thepatternsof RRsdiffered by body-site. Rela-tiverisk of BCC of thehead and neck and thelimbsfellwith increasing lifetimetotal exposure, but, onthetrunk,it increasedwith increasingtotal exposure;theRR for thehighest category of total exposure to the trunk was 2.4(CI = 1.2-4.8).Amountsof exposuredifferedacrossbody

sitesand theentirerangeof exposureof thetrunk fittedinto thefirst quarter of exposureto thehead and neck.

Occupational exposure.Exposureto thesunat workis not associated strongly with BCC. Two41,49 of fivestudies showed statistically significant positive asso-ciations with occupational exposure, but with onlysmall increases in risk (RRs of 1.3 and 1.4) for whatwere quite crude summary variables. Two44,45 of threestudies not supportive of occupational exposure as arisk factor for BCC were based on well-definedquantitative measurements.

Nonoccupational exposure. All studies of nonoccu-pational exposure, measured as summer holiday orweekend exposure, which have used quantitativemeasurementsof sun exposure haveshown statisticallysignificant positiveassociationswith BCC. Theseresultswere obtained in Western Australia,48 Canada,44 andsouthern Europe(i.e., theHeliosstudy46). In addition,a specific measure of intermittent exposure, con-structed by K ricker et al 48 from estimates of exposureon working days and nonworking days in each week,showed strongly increasingrisksof BCC with increas-ing intermittency of sun exposure at 15to 19years of

age, with a high RR (3.9, CI =1.9-7.8) for the mostintermittent pattern of exposure.

Sunburn. Risk of BCC increased significantly withlifetime measures of sunburn in two43,46 of six popula-tion-based studies. RRs ranged from 2.9 for six ormore sunburns in one study43 to 1.7 in two studies46,48

and around 1.0 in three.44,49,50

Other sun-related skin damage. Strong positive re-lationshipshavebeenobservedbetweenBCC andsolarkeratoses36,50 and more moderateassociations between

BCC and telangiectasia and solar elastosis.36

Sunlight and cancer



6/13

Exposure-response relationship. The quantitativerelationship between sunlight exposure and BCC hasbeen modeled in threestudies: asurvey of skin cancerin Chesapeake Bay (Maryland, US) watermen,51 the

Western Australiancase-control study,45andtheHeliosstudy.46The three studies are reasonably consistent insuggesting that risk of BCC first rises with increasingexposure but then reaches a plateau.

In the Western Australian study,45 risk of BCCincreased with increasing sun exposure in those whotanned well, but the risk was initially flat and then fellwith increasing exposure in those who tanned poorly.

This pattern isconsistent with thereported exposure-response relationships because, for any given incidentexposure, actual exposure of the basal layer of the skinwould beless in those who tan well than in those who

tan poorly.

Protection against sunlight. I n Western Australia, useof sunscreensand thewearing of ahat wereassociatedwith an increased risk of BCC rather than adecreasedrisk.48This association was strongest in the 10 yearsbefore diagnosis of the cancer, thus suggesting that itmay have been due to recent adoption of protectionagainst the sun in those already at high risk of skincancer. An apparently paradoxical association wasalsofound in US nurses;43 risk of BCC was 40 percenthigher in thosewhoregularly spent timeoutdoorsandused sunscreen than in thosewho regularly spent timeoutdoors and did not use sunscreen.

Squamouscell carcinomaof theskin

Descriptive studies

Ethnic origin. Nonmelanocytic skin cancer, includingSCC, is rareamong populations not of European origin.Among populations of mainly European origin, theincidence rate of SCC is lower in ethnic groups withdarker skin pigmentation.35,36

Place of residence. Within Australia, the US, andNorway, the incidence rate of SCC increases withincreasing proximity to the equator.35,37,52The patternis less clear among countries, but is likely to beconfounded by ethnic origin and by varying degreesof completeness of ascertainment.34 SCC is relatedmore strongly to latitude37 or measured UVB radia-tion35 than is BCC.

Migration. Migrantsto AustraliafromtheUnited King-dom, an area of lower sun exposure, have incidencerates of SCC about half those of the Australian-born

population.37

Occupation. No separate descriptive studies of SCChavebeen reported. Analytic studiesof occupation arediscussed below.

Anatomic site. SCC has its highest density on usuallyexposed sites. U nlike melanoma and BC C it has lowdensity on sometimesexposed sites(Table1). Urbach53

examined SCC occurring on the head and neck andfound that, in contrast to BCC, they rarely occurredon the more sheltered parts.

Analytic studies

Thefollowingreview isbasedmainly onthemorerecent,population-based studies.

Sensitivity of skin to sunlight. Estimates of RR havebeen between 1.5and 4.5 for comparisonsof themostsensitive skin with the least sensitive skin.34

Ambient sunlight at places of residence. Grodstein etal 54 observed an association between risk of SCC andplaceof residencein acohort of USnurses. Comparedwith women livinginnortheastern states,womenlivingin California or Florida had increased rates. U Sveterans of World War I I who served in the Pacificwere more likely than their counterparts who servedin Europe to havean SCC.55

Exposure of skin to the sun.

Total lifetimeexposure. Total lifetimeexposure to thesun, measured in hours, showed astrongdose-responserelationship with SCC in an early hospital-based case-control study.56However,becauseagewasnotcontrolled,substantial confounding is likely. Much weaker positiveassociations have been observed in more adequatestudies,34 although a strong association was seen in theHeliosstudy.46 No increased risk was seen in thecohortstudy of US nurses (although the data were based on asinglequestion, Do you regularly spend timeoutdoors

in the summer?),54

nor in a case-control study fromCanada57 that had extensivedataon exposure.

Occupational exposure. Crude measures of occupa-tional exposure were considered in four studies;34 allRRs were greater than 1.0. M ore recently, Gallagheret al 57 estimated the number of hours of exposure atwork. A lthough little effect was seen for total lifetimeexposure at work, odds ratios (OR) increased withincreasing exposure at work in the last 10 years theOR in the highest category was 4.0 (CI =1.2-13.1).Rosso et al 46 found increasing risks with increasing

hours of exposure at work.

D.R. English et al



7/13

Nonoccupational exposure. Gallagher et al 57 esti-mated hours outdoors during recreation over thelifetime, in the first 20 years of life and in thelast 10years. O nly for exposure in the first 20 years were

ORs greater than unity and even then, thehighest ORwas only 1.6. Similarly, Rosso et al 46 found little orno association with a number of measures of nonoc-cupational exposure.

Sunburn. Kricker et al 34 reviewed three studies ofsunburn. Two found strong associations (ORs =3 orgreater)50,56 and the other found a weak association(OR of 1.5).58 Three studies have been completedsubsequently: rate ratios increased with lifetimenumber of burns in the cohort study of US nurses54

andmost,butnotall,measuresof sunburnwererelated

to risk of SCC in the Canadian case-control study.57

However,noassociation wasseen in theH eliosstudy.46

Other sun-related skin damage. Strong associationshave been seen with biological markers of long-termsun exposure.34Thesemarkers includesolar lentigines,facial telangiectasia, elastosis of the neck and dorsumof the hands, and presence of solar keratoses. Theassociation with solar keratoses is particularly strong.

Exposure-response relationship. Two reports46,51

provide quantitative data on the exposure-response

relationship. Thefirst, of ChesapeakeBay watermen,51

included only 35 men with SCC. The prevalence ofSCC was modeled as a power function of average,annual estimated UVB-exposure and age (i.e., Preva-lence = Exposurea Ageb). When four cases with thelowest exposure levels were excluded, a strong rela-tionship was seen with annual exposure(theexponentwas 1.7). Although thesedata suggest that the preva-lence of SCC, and presumably also the incidence,increases with increasing exposure, the effect wouldhave been much weaker had the arbitrary exclusionof cases not been made. Data from theHelios study46

are more convincing theRR increased exponentiallywith increasing total hours of exposure.

Protection against sunlight. The effect of sunscreenson risk of SCC hasbeen considered in only onestudy,the N urses H ealth Study.54The OR for regular timeoutdoors with use of sunscreens, compared with noregular timeoutdoors, was 0.9(CI =0.6-1.2). TheORfor time outdoors without use of sunscreens was 0.7(CI =0.4-1.1). Persuasive evidence that sunscreensprotect against solar keratoses (probableprecursors toSCC) is provided by the results of two randomized

trials.59,60

Melanomaoftheeye

Most U V radiation incident on the eye is absorbed bythecornea and the lens. Approximately four percent of

it reachestheretinain early childhoodandthisproportionfallswith increasing age to lessthan onepercent of radia-tion below 340nm andtwo percent of radiation between340 and 360 nm.61Thus, whileexposure of theretina islow, induction of choroidal melanoma by solar U Vradiation is still possible.

Descriptive studies

Ocular melanoma is classified and often reported underthesamerubric asall other cancers of theeye. However,since ocular melanoma comprises some 80 percent ofcancers of theeye, thedescriptivepatterns of all cancers

of the eye probably give a reasonable reflection of itspatterns, particularly in adults.

Ethnic origin. Ocular melanoma is primarily a diseaseof populations of European origin. A mong mixedpopulations residing at thesamelatitude, its incidenceis higher in Whites than Blacks and Asians.62-64

Place of residence. No evidence has been found of alatitude gradient for ocular melanoma or cancers ofthe eye as a whole.65-67 Incidence rates for cancers ofthe eye as a whole in nine countries showed a ruralexcess in males.68

Migration. Whereas rates of cutaneous melanoma arehigher in Jews born in Israel than in Jews born inEuropeand A merica,6 there is little difference in ratesof ocular melanoma between these two groups.69

Occupation. Ocular melanoma has been associatedinconsistently with farming in descriptivestudies. Twostudies found a positive association,70,71 but three didnot.72-74

Anatomic site. Ocular melanomas are located most

frequently in the central posterior choroid and theinferior and temporal iris.75 The central posteriorchoroid corresponds to the region of maximal lightfocusing by the refractive components of the eye.Studies using model eyes haveshown that theinferiorand temporal regions of the iris are relatively unpro-tected from incident light by surrounding anatomicstructures.76,77

Analytic studies

Sensitivityofskin tosunlight. Peoplewhoburneasily andtan poorly had moderately increased risk of ocular

melanoma (RR = 2 or less) in two78,79

of four relevant

Sunlight and cancer



8/13

case-control studies. No association was seen with hairand eyecolor when ethnicity was taken into account,78-82

but in oneof thesestudies,82 an increased risk with lightskin color persisted after adjustment for ethnicity.

Ambient sunlight at placesof residence. Two case-con-trol studies from the US81,82 showed an associationbetween ocular melanoma and birth or residence inthe southern US; but in two other studies,79,80 noassociations were observed with place of residence.

Exposureof eyeto thesun. There is littleevidence foran association between personal sun exposure andocular melanoma. Total cumulative sunlight exposurewas not associated with ocular melanomain two case-control studies.80,82 Four case-control studies80,82-84

foundnosignificantassociation with farming,butone84

found an increased risk in sailors, ships officers, andfishermen. Three studies78,80,82 found no associationwith personal leisure timeor vacation exposure to thesun. One other study,81 however, found weak associa-tionswith gardeningand frequent sunny vacationsbutnot with high leisure timeoutdoors.

Protection against sunlight. In one case-control study,81

those who occasionally, rarely, or never used sun-glasses, hats, or sun visorswhen in the sun had increasedrisks of ocular melanoma (up to RR =1.9 with rare

use) relative to those who almost always used them.However, another study82 found no evidenceof a pro-tective effect with useof sunglasses or a sun visor.

Other cancers

Conjunctival cancer

Squamous cell carcinoma of the conjunctiva usuallypresents in theexposed area of theeyebetween thelids.85

It has been shown in a study of incidence rates in 47populations86 to decreasein incidenceby 49percent per10-degreeincreasein latitudeandby 29 percent for aone

minimal erythemal-dose-fall in estimated ambient UVirradiance. Additionally, a recent study87 of conjunctivaland limbal epithelial dysplasia (carcinoma in situ andsquamouscell carcinoma)reportedincreasedrisksfor fairskin, being outdoorsat < 30 latitudefor greater than 50percent of thetimebeforeage12years, estimated cumu-lativeUV exposureand ahistory of skin cancer.

Non-Hodgkins lymphoma

Studies in Europe88-93 have shown an increased risk ofnon-Hodgkins lymphoma subsequent to non-melano-cytic skin cancer and malignant melanoma. Theparallel

dramatic increasein incidenceratesin EuropeandtheUS

over recent decadesand thegenerally similar geographicvariation for non-H odgkins lymphomaand skin canceralso suggest that these malignanciesmay have a commonrisk factor,possibly sunlight exposure.94 However, in the

US, themortality rates from non-Hodgkins lymphomaarehighest in thenorth, not thesouth.95

Breast and colon cancer

It has been hypothesized that sunlight might preventbreast cancer and colon cancer96,97 on the grounds thatvitamin D might protectagainst thesecancers.Mortalityratesfor both breast andcolon cancer andincidenceratesfor colorectal cancer were found to behighestin parts oftheUSwiththelowestambientsunlight.96-98Theincidenceof breast cancer in theformer Union of Soviet SocialistRepublics (USSR) also was inversely related to ambient

sunlightlevels,99butthecorrelation betweenbreastcancerandsocioeconomic statuswasstronger thanthatbetweensunlightandbreast cancer.No studiesin individualshavebeen reported.

Molecular effectsofultraviolet radiation

UV radiation produces a number of photoproducts inDNA.Themostcommonarecyclobutane-typepyrimid-ine di mers and pyrimidine-pyrimidone (6-4)photoproducts formed between adjacent pyrimidines(cytosine [C] and thymine [T]).100The action spectrumfor formation of photoproducts in DNA in human skinclosely approximatesthat for theinduction of squamouscell carcinoma in mice.1,101 U V photoproducts aremutagenic if they are not repaired before cell divisionoccurs. The most common UV-induced mutations areCT transitions, which occur at dypyrimidine sites.102

TandemtransitionsCCTT alsooccur thesemutationsarealmost specific to UV.103

Patientswith xerodermapigmentosum(XP), an inher-ited disorder characterized by increased sensitivity toacute exposure to the sun and defective repair of DNAphotoproducts, havehistories of high numbers of BCC,

SCC, cutaneous melanoma at young ages on body sitescommonly exposed to sunlight, and of cancers of theanterior eye.104 These observations strongly implicateunrepairedphotoproductsin DNA in thegenesisof thesecancers, although two other syndromes of inheriteddeficiency in excision repair of DNA are not associatedwith increased incidencerates of skin cancer.105,106

In 1991, Brash and co-workers107 reported mutationsin thep53 tumor suppressor gene in 14 (58 percent) of24SCCs.ThreemutationswereCCTT transitions, fivewere CT transitions and all mutations occurred atdipyrimidinesites.Other investigatorshavealsoreported

p53 mutations at dipyrimidinesites in SCC, BCC, and

D.R. English et al



9/13

solar keratoses.108-114 Mutations of thep53 geneoccur inupto90 percentof SCC andabout50 percentof BCC.115

They alsohavebeen observedinnormal skin. Transitionsfrom CC to TT were observed in normal skin on sun

exposed sites in 17 of 23 (74 percent) samples fromAustralian skin cancer patientscompared with oneof 20(five percent) samples of skin from sites not exposed tothesun,116suggestingthatp53mutationsoccur frequentlyin responseto sunlight. However,p53mutations are rareinmelanomas.117 Mutationspossibly duetoUV radiationhavebeen observed118 in another putativetumor suppres-sor gene, theCDKN2 gene, in melanomacell lines, andinfrequently in tumors.

Conclusions

Thestrength of evidence that sunlight causes melanomaof the skin, BCC, SCC, and melanoma of the eye issummarized in Table 3 for each epidemiologic featurediscussed in the text. For all three skin cancers, thedistributionsby anatomic site,ethnic origin,andplace ofresidence, and theeffectsof migration implicatesunlightasacause.Ofall thedescriptivecharacteristicsconsidered,only occupation provides little evidence that sunlightcauses skin cancer. The increased risk of melanoma inindoor workers once was considered to be persuasiveevidenceagainst sunlight causing thedisease.

Melanomaof theskin and BCC appear to sharemanyepidemiologic features,whileSCC standsapart. Evidencethat melanoma of the skin and BCC are related to totaland occupational exposure to the sun is largely absent

(Table 3). Strong evidence, particularly for melanoma,existsfor a relationship with nonoccupational exposure.In contrast, there is evidencethat SCC is related to totaland occupational exposure, but almost no evidencethatit is related to nonoccupational exposure. A ll threeskincancers are related to sunburn (particularly melanoma)and all are related also to other indicators of sunlight-induced skin damage. Finally, studies of mutations inBCCsandSCCsprovideevidencethat they were causedby sunlight. Thereis much lessevidenceof this kind formelanoma.

Therisk of melanomaof the skinandBCC appearsto

increasewithincreasingexposuretosunlightat low levelsof exposure,beforelevelingoff andperhapsfallingathighlevels. This pattern may bedueto confounded effectsofamount and pattern of sunlight exposure in causingmelanomaand BCC. It might beargued reasonably that,as amount of exposure increases, pattern of exposurebecomesmorecontinuousandthusamelioratestheeffectof the increased amount. There are at present, however,no adequate empirical data which would enable thesecomplex interacting and confounded effects to bedisentangled.

Table 3. Summary of strength of evidence that sunlight causes cancer based on epidemiologic features discussed in texta

Epidemiologic feature Skin Eye melanoma

Melanoma BCC SCC

Descriptive studies

Ethnic origin + + + +

Place of residence + + + 0

Migration + + + 0

Occupation 0 0 0 0

Anatomic site + + + +

Analytic studies

Sensitivity of skin to sunlight + + + 0

Ambient sunlight at places of residence + + + 0

Exposure of the skin to sunlight

Total exposure 0 0 + 0

Occupational exposure 0 0 + 0

Non-occupational exposure + + 0 0

Sunburn + + + 0

Other sun-related skin damage + + + 0

Protection against sunlight 0 0 0 0

Molecular effects

Effects of DNA repair deficiency + + + 0

Mutations in tumors characteristic of UV radiation + + + 0

a++ indicates strong evidence; + indicates weak evidence; 0 indicates conflicting evidence, lack of evidence or lack of effect.

Sunlight and cancer



10/13

Data on the quantitative nature of the exposure-response relationship for SCC indicate a monotonicincreasein risk with increasingexposure. Theseobserva-tionsare supportedby theanatomic sitedistribution and

thestrong latitudegradient.No consistent evidence that sunlight causesmelanoma

of the eye emerges from descriptive studies or analyticstudies in which individual exposure to the sun wasmeasured.Only differencesbyethnicorigin,theanatomicsite of melanoma within the eye, and some conflictingresultsshowingdecreasedriskassociatedwith protectionagainst sunlight are supportive.

In 1992, the International Agency for Research onCancer (IARC)concludedthatthereissufficientevidencethat solar radiation causes melanoma of the skin andnonmelanocytic skin cancer.15 What, then, are the out-

standing issues? I dentifying the exposure-responserelationships,with particular emphasis on distinguishingbetween pattern and amount of exposure is the mostimportant outstanding issue. This is not merely ofacademic interest, because if pattern of exposure isimportant, areduction in amountof exposuremight notresult in a reduction of risk of melanoma (and possiblyalso of BCC) if pattern of exposurechanges at thesametime. Another unresolved,butnot unrelated, issueis therelationship between exposure to a particular anatomicsiteand risk of skin cancer on that site. A number of theinconsistencies seen in current literature may be duetofailure to measure exposure to theanatomic site wherethe cancer occurred. The lack of protective effects ofsunscreen and hat useis also notable, and there is limitedother direct evidencethat protection from sun exposurereduces risk of any skin cancer in thegeneral population.

Thepossibilitythatsunexposurecausesocular melanomaremainstantalizingand should not bediscarded withoutsomefurther study although, given that sun exposureinearly life couldbeof dominant importance, it will notbeeasy.

Howmightweaddresstheoutstandingissues?Becauseof the low salienceof sun exposure, measuring lifetimeexposure to particular anatomic sites, including pattern

and amount, is an error-prone task. The problem maynot be so great in young adults who may be able toremember relevant exposuresin childhood, adolescence,and early adulthood,andwhoseparentsmay beavailableto corroboratetheir memories. International, collabora-tive case-control studies, involving centers with widelyvarying ambient exposures may prove useful. Cohortstudiesareof limitedvalueunlesslargenumbersof youngsubjects can be recruited and resurveyed frequently toupdate their sun exposure. This would bea formidabletask.

What istheroleof molecular epidemiology?Weneed

arangeof short-termand long-term biomarkersthat can

beused to distinguish theeffects of pattern and amountofexposure.Weknowlittleabouttherelationshipofmostcurrent biomarkers (e.g., UV-induced mutations in thep53gene, formation of DNA photoproducts) to pattern

and amount of exposure. It would not be difficult todesign short-term experimental studies in humans toelucidatesomeof theseissues. It is doubtful that contin-ued analysisof UV-induced mutationsin thep53 geneorothergenesintumorswill beofvalueif usedintheabsenceof appropriatemeasures of exposure. Lack of associationbetween some genetic changes in BCC and sun expo-sure,119 for example, may be due to failure to measurepattern of exposure.

References

1. de Gruijl FR. Photobiology of photocarcinogenesis.Photochem Photobiol 1996; 63: 372-5.2. ParkinDM, MuirCS,WhelanSL,GaoYT,Ferlay J, Powell

J, eds. Cancer I ncidencein Five Continents. Lyon, France:International Agency for ResearchonCancer,1992; IARCSci. Pub. No. 120.

3. A rmstrongBK. Melanomaof theskin.Br Med Bull 1984;40: 346-50.

4. BulliardJL,CoxB, ElwoodJM. Latitudegradientsin mela-nomaincidenceandmortality in thenon-Maori populationof New Zealand. Cancer CausesControl 1994; 5: 234-40.

5. Khlat M, Vail A, Parkin DM, Green A. Mortality frommelanoma in migrants to Australia: variation by age atarrival and duration of stay. Am J Epidemiol 1992; 135:1103-13.

6. Iscovich J, Andreev H, Steinitz R. Incidenceof cutaneousmalignant melanoma in Israel, 1960-1989. Public H ealthRev 1995; 23: 1-23.

7. Mack TM , FloderusB. Malignantmelanomariskbynativ-ity, placeof residence at diagnosis, and age at migration.Cancer CausesControl 1991; 2: 401-11.

8. HolmanCD,Mulroney CD, ArmstrongBK.Epidemiologyof pre-invasive and invasive malignant melanoma inWestern Australia. I nt J Cancer 1980; 25: 317-23.

9. L ee JA, Strickland D. Malignant melanoma: social statusand outdoor work. Br J Cancer 1980; 41: 757-63.

10. Cooke KR, Skegg DC, Fraser J. Socio-economic status,indoor and outdoor work, andmalignant melanoma. Int JCancer 1984; 34: 57-62.

11. GreenA, MaclennanR, Youl P,MartinN. Sitedistribution

of cutaneousmelanomain Queensland. I nt J Cancer 1993;53: 232-6.12. Armstrong BK , English DR. Cutaneous malignant mela-

noma. In: Schottenfeld D, Fraumeni JF Jr, eds. CancerEpidemiology andPrevention. SecondEdition. New York,NY (USA): Oxford University Press, 1996.

13. White E, K irkpatrick CS, Lee JA. Case-control study ofmalignantmelanomain WashingtonState. I.Constitutionalfactorsandsunexposure.AmJ Epidemiol1994;139:857-68.

14. H olly EA, Aston DA, CressRD, AhnDK , Kristiansen JJ .Cutaneous melanoma in women. I. Exposure to sunlight,ability to tan, and other risk factors related to ultravioletlight. Am J Epidemiol 1995; 141: 923-33.

15. International Agency for Research on Cancer. Solar andUltraviolet Radiation. Lyon, France: IARC , 1992: IARC

Monogr Eval CarcinogRisksHumans: 1-316.

D.R. English et al



11/13

16. PionIA, Rigel DS,Garfinkel L , Silverman MK , Kopf AW.Occupation and therisk of malignant melanoma. Cancer1995; 75: 637-44.

17. A utier P, Dore JF, LejeuneF, et al. Recreational exposureto sunlight and lack of information as risk factors forcutaneous malignant melanoma. Results of an EuropeanOrganization for Research and Treatment of Cancer(EO RTC) case-control study in Belgium, France andGermany. TheEORTC MalignantMelanomaCooperativeGroup. MelanomaRes1994; 4: 79-85.

18. Autier P,Dore JF, SchifflersE, et al. Melanomaanduseofsunscreens: an EORTC case-control study in Germany,Belgiumand France. TheEORTC Melanoma CooperativeGroup. I nt J Cancer 1995; 61: 749-55.

19. Westerdahl J, Olsson H, Ingvar C, Brandt L, Jonsson PE,Moller T. Southern travellinghabitswith special referenceto tumour site in Swedish melanoma patients. AnticancerRes1992; 12: 1539-42.

20. Westerdahl J,OlssonH, Ingvar C.At what agedosunburnepisodes play acrucial role for thedevelopment of malig-nant melanoma? Eur J Cancer 1994; 30A: 1647-54.

21. Marghoob AA, SladeJ, Salopek TG, K opf AW, Bart RS,Rigel DS. Basal cell and squamous cell carcinomas areimportant risk factors for cutaneousmalignant melanoma.Screening implications. Cancer 1995; 75: 707-14.

22. Goodman KJ, Bible ML, London S, Mack TM. Propor-tional melanoma incidenceand occupation among Whitemales in Los Angeles County (California, United States).Cancer CausesControl 1995; 6: 451-9.

23. RodenasJM,Delgado-Rodriguez M,HerranzMT,TercedorJ, Serrano S. Sun exposure, pigmentary traits, and risk ofcutaneousmalignant melanoma acase-control study in aMediterraneanpopulation.Cancer CausesControl 1996;7:275-83.

24. Brown J , Kopf AW, Rigel DS, Friedman RJ. Malignantmelanomain World War II veterans. I nt J Dermatol 1984;23: 661-3.

25. L MG, Cabanes PA, Desvignes V, Chanteau MF, MlikaN, Avril MF. Oral contraceptiveuseandrisk of cutaneousmalignant melanoma in a case-control study of Frenchwomen. Cancer CausesControl 1992; 3: 199-205.

26. Grob JJ, Gouvernet J, Aymar D , et al. Count of benignmelanocyticneviasamajor indicatorof risk for nonfamilialnodular andsuperficial spreading melanoma. Cancer 1990;66: 387-95.

27. Vger D, Ringbck G, KivirantaH . Melanomaand othertumoursof theskinamongoffice,other indoorandoutdoorworkersinSweden1961-1979.Br J Cancer1986;53:507-12.

28. Beral V, RobinsonN. Therelationship of malignant mela-

noma, basal and squamous skin cancers to indoor andoutdoor work. Br J Cancer 1981; 44: 886-91.29. Nelemans PJ, Rampen FH, Ruiter DJ , Verbeek AL. A n

additiontothecontroversyonsunlightexposureandmela-noma risk: a meta-analytical approach. J Clin Epidemiol1995; 48: 1331-42.

30. World Health Organization. Ultraviolet Radiation (Envi-ronmental H ealth Criteria 160). Geneva, Switzerland:WHO, 1994.

31. ElwoodJM, GallagherRP,H ill GB, PearsonJC.Cutaneousmelanoma in relation to intermittent and constant sunexposurethe Western Canada Melanoma Study. Int JCancer 1985; 35: 427-33.

32. ElwoodJM,GallagherRP.Sunexposureandtheepidemiol-ogy of melanoma. In: Gallagher RP, Elwood ME, eds.

Epidemiological Aspects of Cutaneous Malignant Mela-

noma. Boston, MA (USA): Kluwer Academic Publishers,1994: 15-66.

33. Westerdahl J, O lssonH, Masback A, Ingvar C, JonssonN.Is theuse of sunscreens a risk factor for malignant mela-noma? Melanoma Res1995; 5: 59-65.

34. Kricker A, ArmstrongBK, English DR. Sunexposure andnon-melanocyticskincancer.Cancer CausesControl 1994;5: 367-92.

35. Scotto J, Fears TR, Fraumeni JF Jr. I ncidence of Non-melanomaSkin Cancer in theUnitedStates.Bethesda,MD(USA): US Department of Health and H uman Services,1983; N IH Pub. No. 82-2433.

36. Kricker A , Armstrong BK, English DR, Heenan PJ. Pig-mentary and cutaneous risk factors for non-melanocyticskin cancera case-control study. I nt J Cancer 1991; 48:650-62.

37. MarksR, StaplesM, GilesGG. Trendsin non-melanocyticskincancer treated in Australia:thesecondnational survey.I nt J Cancer 1993; 53: 585-90.

38. MolesworthEH .Rodentulcer.MedJ Aust 1927;1:878-99.39. Blum H. Sunlight as a causal factor in cancer of the skin

in man.JNC I 1948; 9: 247-58.40. EmmettEA .Ultravioletradiationasacauseof skin tumors.

CRC Critical Reviewsin Toxicology 1973; 2: 211-55.41. Marks R, J olley D, Dorevitch AP, Selwood TS. The inci-

dence of non-melanocytic skin cancers in an Australianpopulation: results of afive-year prospectivestudy. Med JAust 1989; 150: 475-8.

42. Vitasa BC, Taylor HR, Strickland PT, et al. Association ofnonmelanoma skin cancer and actinic keratosis withcumulativesolar ultraviolet exposure in Maryland water-men. Cancer 1990; 65: 2811-7.

43. Hunter DJ, Colditz GA, Stampfer MJ, Rosner B, WillettWC, Speizer FE. Risk factorsfor basal cell carcinomain a

prospectivecohort of women.AnnEpidemiol 1990;1:13-23.44. Gallagher RP,H ill GB, BajdikCD ,etal.Sunlightexposure,pigmentary factors,andrisk of nonmelanocyticskincancer.I. Basal cell carcinoma. Arch Dermatol 1995; 131: 157-63.

45. K ricker A, A rmstrong BK, English DR, H eenan PJ . Adose-responsecurvefor sun exposureand basal cell carci-noma. I nt J Cancer 1995; 60: 482-8.

46. Rosso S, Zanetti R, Martinez C, et al. The multicentreSouth European Study, Helios. 2. Different sunexposurepatterns in the aetiology of basal cell and squamous cellcarcinomas of theskin. Br J Cancer 1996; 73: 1447-54.

47. Zanetti R, Rosso S, Martinez C, et al. The multicentreSouth EuropeanStudy, H elios.1. Skincharacteristicsandsunburns in basal cell and squamouscell carcinomas of theskin. Br J Cancer 1996; 73: 1440-6.

48. Kricker A, ArmstrongBK, English DR, Heenan PJ. Doesintermittent sun exposure cause basal cell carcinoma? Acase-control study in Western Australia. I nt J Cancer 1995;60: 489-94.

49. H ogan DJ , To T, Gran L, WongD, LanePR. Risk factorsfor basal cell carcinoma. I nt J Dermatol 1989; 28: 591-4.

50. GreenA,BattistuttaD. Incidenceanddeterminantsof skincancer in ahigh-risk Australian population. I nt J Cancer1990; 46: 356-61.

51. Strickland PT, Vitasa BC, West SK, et al. Quantitativecarcinogenesis in man:solar ultraviolet B dosedependenceof skin cancer inMarylandwatermen.JNC I 1989;81:1910-3.

52. MagnusK. TheN ordicprofileof skin cancer incidence. Acomparativeepidemiological study of the three main types

of skin cancer. I nt J Cancer 1991; 47: 12-9.

Sunlight and cancer



12/13

53. UrbachF.Geographicpathologyof skincancer. In:UrbachF,ed. BiologicEffectsof UVR with Emphasison theSkin.Oxford, UK: Pergamon Press, 1969: 635-50.

54. Grodstein F, Speizer FE, Hunter DJ. A prospectivestudyof incident squamous cell carcinoma of the skin in theNurses Health Study.JNCI 1995; 87: 1061-6.

55. Ramani ML, Bennett RG. High prevalenceof skin cancerin World War II servicemenstationed in thePacifictheater.J Am Acad Dermatol 1993; 28: 733-7.

56. Urbach F, Rose DB, Bonnem M. Genetic and environ-mental interactions in skin carcinogenesis. EnvironmentalCancer. Baltimore, MD (USA):Williams& Wilkins,1972:355-71.

57. Gallagher RP,H ill GB, BajdikCD,etal.Sunlightexposure,pigmentation factors,andriskofnonmelanocyticskincancer.II.Squamouscell carcinoma.ArchDermatol1995;131:164-9.

58. Hogan DJ, Lane PR, Gran L, Wong D. Risk factors forsquamous cell carcinoma of the skin in Saskatchewan,Canada.J Dermatol Sci 1990; 1: 97-101.

59. Thompson SC, Jolley D, Marks R. Reduction of solarkeratoses by regular sunscreen use. N Engl J Med 1993;329: 1147-51.

60. Naylor MF, Boyd A, Smith DW, Cameron GS, HubbardD, Neldner KH . Highsunprotection factor sunscreensinthesuppression of actinic neoplasia. Arch Dermatol 1995;131: 170-5.

61. LermanS.L ightinducedchangesinoculartissues.In:MillerD, ed. Clinical Light DamagetotheEye. New York, NY(USA): Springer-Verlag, 1987: 183-215.

62. Neugut AI , Kizelnik-FreilichS, Ackerman C. Black-whitedifferences in risk for cutaneous, ocular,andvisceral mela-nomas. Am J PublicHealth 1994; 84: 1828-9.

63. Polednak AP, Flannery JT. Brain, other central nervoussystem, and eye cancer. Cancer 1995; 75: 330-7.

64. Raivio I. Uveal melanomain Finland. An epidemiological,clinical, histological and prognostic study. Acta Ophthal-mol 1977; Suppl: 1-64.

65. Scotto J, Fraumeni JF Jr, Lee JA. Melanomas of the eyeandothernoncutaneoussites:epidemiologicaspects.JNCI1976; 56: 489-91.

66. GislasonI,MagnussonB,TuliniusH.Malignantmelanomaof the uvea in I celand 1955-1979. Acta O phthalmol(Copenh) 1985; 63: 389-94.

67. Schwartz SM, Weiss NS. Place of birth and incidence ofocular melanoma in theUnitedStates. I nt J Cancer 1988;41: 174-7.

68. Doll R. Urban andrural factors in theaetiology of cancer.I nt J Cancer 1991; 47: 803-10.

69. Iscovich J, Ackerman C, Andreev H , Peer J, Steinitz R.

An epidemiological study of posterior uveal melanomainIsrael, 1961-1989. I nt J Cancer 1995; 61: 291-5.70. Saftlas AF, Blair A, Cantor KP, Hanrahan L , Anderson

HA. Cancer and other causes of death among Wisconsinfarmers. AmJ I nd Med 1987; 11: 119-29.

71. Gallagher RP, Elwood JM, Rootman J. Epidemiologicaspects of intraocular malignant melanoma. Cancer TreatRes1988; 43: 73-84.

72. Milham SJ. Occupational M ortality in Washington State1950-1979. Cincinnati, OH (USA): National Institute forOccupational Safety and H ealth. 1983; DHSS (N IOSH)Pub. No. 83-116.

73. Officeof thePopulation Censuses and Surveys. Occupa-tionalMortality:theRegistrarGeneralsDecennial Supplementfor Great Britain 1979-80, 1982-83. London, UK: Her

MajestysStationery Office, 1986; SeriesDS No. 6.

74. Vger D, Swerdlow AJ, Beral V. Occupation and malig-nant melanoma: a study based on cancer registration datain England andWales and in Sweden. Br J Ind Med 1990;47: 317-24.

75. Horn EP, H artgeP, Shields JA, Tucker MA. Sunlight andrisk of uveal melanoma.JNCI 1994; 86: 1476-8.

76. Coroneo MT, M uller-StolzenburgNW, H o A. Peripherallight focusing by the anterior eye and theophthalmohe-lioses. OphthalmicSurg 1991; 22: 705-11.

77. Narayanan P, MerriamJC, Vazquez ME, Dillon J. Experi-mental model of light focusing of the peripheral cornea.I nvest Ophthalmol Vis Sci 1996; 37: 37-41.

78. Holly EA, Aston DA, Char DH , Kristiansen JJ, AhnDK.Uveal melanoma in relation to ultraviolet light exposureand host factors. Cancer Res1990; 50: 5773-7.

79. VanHeesCL , DeBoer A, Jager MJ , etal.Areatypical nevia risk factor for uveal melanoma? A case-control study. JI nvest Dermatol 1994; 103: 202-5.

80. Gallagher RP, Elwood JM, Rootman J, et al. Risk factors

for ocular melanoma: Western Canada MelanomaStudy.JNC I 1985; 74: 775-8.81. Tucker MA , ShieldsJA, HartgeP, et al. Sunlight exposure

as risk factor for intraocular malignant melanoma. N EnglJ Med 1985; 313: 789-92.

82. Seddon JM, Gragoudas ES, Glynn RJ, Egan KM, AlbertDM , Blitzer PH . H ost factors, UV radiation, and risk ofuveal melanoma. A case-control study. Arch Ophthalmol1990; 108: 1274-80.

83. A jani UA, SeddonJM, Hsieh C-c, Egan KM, Albert DM,Gragoudas ES. Occupation and risk of uveal melanoma.An exploratory study. Cancer 1992; 70: 2891-900.

84. HollyEA,AstonDA,AhnDK,etal.Intraocularmelanomalinked to occupations and chemical exposures.Epidemiol-ogy 1996; 7: 55-61.

85. N i C,Searl SS,KriegsteinHJ,WuBF.Epibulbarcarcinoma.I nt Ophthalmol Clin 1982; 22: 1-33.86. Newton R, Ferlay J, Reeves G, Beral V, ParkinDM. Effect

of ambient solar ultraviolet radiation on incidence ofsquamous-cell carcinoma of the eye. Lancet 1996; 347:1450-1.

87. Lee GA , Williams G, H irst LW, Green AC. Risk factorsin the development of ocular surfaceepithelial dysplasia.Ophthalmology 1994; 101: 360-4.

88. H all P, Rosendahl I , Mattsson A , Einhorn S. N on-H odgkins lymphoma and skin malignanciessharedetiology? I nt J Cancer 1995; 62: 519-22.

89. Adami J , Frisch M, Yuen J, Glimelius B, MelbyeM. Evi-denceof anassociationbetweennon-H odgkinslymphomaand skin cancer. Br Med J 1995; 310: 1491-5.

90. FrischM,MelbyeM.New primary cancersafter squamouscell skin cancer. Am J Epidemiol 1995; 141: 916-22.91. Levi F, Randimbison L , Te VC, La Vecchia C. Non-

Hodgkins lymphomas, chronic lymphocytic leukaemiasand skin cancers. Br J Cancer 1996; 74: 1847-50.

92. Frisch M, H jalgrim H, O lsen JH , Melbye M. Risk forsubsequent cancer after diagnosis of basal-cell carcinoma.A population-based, epidemiologic study. Ann Int Med1996; 125: 815-21.

93. WassbergC, Thorn M, Yuen J, RingborgU, Hakulinen T.Second primary cancers in patientswith cutaneousmalig-nant melanoma: apopulation-based study in Sweden. Br JCancer 1996; 73: 255-9.

94. McMichaelAJ,GilesGG.Haveincreasesinsolarultravioletexposure contributed to the rise in incidence of non-

Hodgkins lymphoma? Br J Cancer 1996; 73: 945-50.

D.R. English et al



13/13

95. HartgeP, DevesaSS, Grauman D, Fears TR, Fraumeni JFJr. Non-H odgkins lymphoma and sunlight. JNCI 1996;88: 298-300.

96. GarlandCF,GarlandFC.Do sunlightandvitaminD reducethe likelihood of colon cancer? I nt J Epidemiol 1980; 9:227-31.

97. Garland FC, Garland CF, Gorham ED, Young JF. Geo-graphic variation in breast cancer mortality in the UnitedStates: ahypothesis involvingexposureto solar radiation.Prev Med 1990; 19: 614-22.

98. EmersonJC, WeissNS. Colorectal cancer andsolar radia-tion. Cancer CausesControl 1992; 3: 95-9.

99. GorhamED,GarlandFC ,GarlandCF.Sunlightandbreastcancer incidence in the USSR. I nt J Epidemiol 1990; 19:820-4.

100. Brash DE. U V mutagenic photoproducts in Escherichiacoli and human cells: a molecular genetics perspective onhuman skin cancer. Photochem Photobiol 1988; 48: 59-66.

101. Freeman SE, H acham H , Gange RW, Maytum DJ ,

Sutherland JC, Sutherland BM. Wavelength dependenceofpyrimidinedimer formation in DNA of human skin irra-diatedin situwithultraviolet light.ProcNatl AcadSci USA1989; 86: 5605-9.

102. Sage E. D istribution and repair of photolesions in DNA:genetic consequences and the role of sequence context.Photochem Photobiol 1993; 57: 163-74.

103. HutchinsonF.Inductionof tandem-basechangemutations.Mutat Res1994; 309: 11-5.

104. Kraemer KH , LeeMM, AndrewsAD, Lambert WC. Therole of sunlight and DN A repair in melanoma and non-melanoma skin cancer. The xeroderma pigmentosumparadigm. Arch Dermatol 1994; 130: 1018-21.

105. Bridges B. Sunlight, DNA damageand skin cancer: anewperspective.Jpn J Cancer Res1990; 81: 105-7.

106. BarrettSF,RobbinsJH,TaroneRE,KraemerKH.Evidencefor defectiverepair of cyclobutanepyrimidinedimers withnormal repair of other DN A photoproductsin atranscrip-tionally active gene transfected into Cockayne syndromecells. Mutat Res1991; 255: 281-91.

107. BrashDE,RudolphJA ,SimonJA, etal.A rolefor sunlightin skin cancer: UV-induced p53 mutations in squamous

cell carcinoma. ProcNatl Acad Sci USA 1991;88: 10124-8.108. Pierceall WE, Mukhopadhyay T, Goldberg LH, Anan-

thaswamy H N . Mutations in the p53 tumor suppressorgene in human cutaneous squamous cell carcinomas. Mol

Carcinog 1991; 4: 445-9.109. Ziegler A , Leffell DJ , Kunala S, et al. Mutation hotspotsdue to sunlight in the p53 gene of nonmelanoma skincancers. ProcNatl Acad Sci USA 1993; 90: 4216-20.

110. Rady P, Scinicariello F, Wagner RF Jr, Tyring SK. p53mutations in basal cell carcinomas. Cancer Res 1992; 52:3804-6.

111. Mols JP, Moyret C, Guillot B, et al. p53 genemutationsin human epithelial skin cancers. Oncogene1993;8: 583-8.

112. Burns JE, BairdMC, Clark LJ , et al. Genemutations andincreased levels of p53 protein in human squamous cellcarcinomasandtheir cell lines.Br J Cancer 1993;67:1274-84.

113. van der Riet P, Karp D, Farmer E, et al. Progression ofbasal cell carcinomathrough loss of chromosome 9q and

inactivationof asinglep53allele.Cancer Res1994;54:25-7.114. Ziegler A , Jonason AS, Leffell DJ, et al. Sunburn and p53in theonset of skin cancer. Nature1994; 372: 773-6.

115. Ziegler A, Jonason AS, Simon J , Leffell DJ , Brash DE.Tumor suppressor genemutationsand photocarcinogenesis.Photochem Photobiol 1996; 63: 432-5.

116. Nakazawa H, English D, Randell PL , et al. UV and skincancer: specific p53 gene mutation in normal skin as abiologically relevant exposure measurement. Proc NatlAcad Sci USA 1994; 91: 360-4.

117. Daya-Grosjean L, Dumaz N, Sarasin A. Thespecificity ofp53mutation spectra in sunlight induced human cancers.J Photochem Photobiol B 1995; 28: 115-24.

118. Pollock PM, Pearson JV, Hayward N K. Compilation of

somatic mutationsof theCDKN2genein human cancers non-random distribution of base substitutions. GenesChromosomCancer 1996; 15: 77-88.

119. Gailani MR, L effell DJ, Ziegler A , Gross EG, Brash DE,Bale AE. Relationship between sunlight exposure and akey genetic alteration in basal cell carcinoma.JNC I 1996;88: 349-54.

Sunlight and cancer


Sunlight & Cancer

Documents

Transcript of Sunlight & Cancer