Document 2JX9LMpyXaY5q0aJXpBxbeJDR

. . . ...- ... 06-20-~00l 12:54pr Frm-UNC HEALTH LIBRARY *.I -1' I- Europtm Jolnnolof Gmcm Revenha 1995.4 3 1 4 3 REVIEW 919-g66-1 537 .. .. . T-222 P.002/014 F-034 P Epidemiological evidence on hair dyes and the b .risk of cancer in humans C La Vecebia'3, A T a d ' (Receiwd 25 July 1994; accepted 23 September 1994) Introduction Hair colouriry products (hair dyes) an widely used, and it has bear cotimatcd that in Europe, North America and Japan over one third of women above the agc of18 aad over olOm ofmen above the age of40 use some type of hair dye,perm&t dyes accounnng for about three quarters of global use. This wide -* spread personal use implies considcrabIeprofcesional exposure by haitdresras andbarbns,sinocthereirn about two millian professional hairdnosas, harbcrs and beauticians in North Amaica and Europe done (Cosmetics, Toiletry and Fragr~ce Aurociation, undated). 06-20-2001 i" I i ! I 12:54pm From-WIC HEALTH LIBRARY 91g-g66-1537 T-222 P.003/014 F-034 C L a Vecchia,A Tavani Tabk 1. Campoawrs of hair coloring products that have been evaluated in the IARC Monograph Series Compound Product in which found MI evrl~atimor cudnopnici~to hurmnr URCmonographs VOlUmC M. 57 57 16 48 52 57 27 16 16 n 51 57 57 Hair dyes conrain components that ~IE mutagenic in vitro ( A r m er ai, 1973, some arc &ogenic in anhals, and several have beep evaluated by working groups of rht Inkrnarional Agency for Rcscarcb on Cancer of the World Health organizadon (IARC/ -0, ' d in IARC, 1993) for their carciuogmkityb hurnaus, A summary tabulation of Lhest components is given in Table 1. Epidsmiologicaldata on cancer risk amongsubjects occupad~dlye x p ~ s e dto, and 10w-kmp e n 0 4 UELC~Sof hair dyes have been mrudat least since rhc 1960% but only during rhe last two decades has Lhe issue attracted widespread attention in the scientific and lay pms. Af" a sum~llsvynvicw of experimental data on cancer in animals, rhc present ovcrvicw will wnsider and discuss epidemiological data in humans with reference IOvarious CSIzIccr sites, and with spccific focus on two goups of neoplasms whose assodation with hair dyes has been widely debated in the last few years, ie bladder cancer and lymphoid w o p l a L m o . Evideace of carciuogenichy in experimmtal animals In at kast three sNdics which investigated skin applications of pumanent (Burnen et ui, 1975) or sani-pumaacnt( S e d 8 and Jouc6,1977; J m b s et d, 1984) hair dye fonndauous m mice, no skin tumour wai-obscnnd, and no s-ti diairaacc emerged for l u g carurn.lymphomas or any other neoplasm .%onsidered.However, the number of afiimnls in each '-&camemgroup was small (appmximarely 20-60), a d this precludes any codusion. ' In three studiea also comprising 20-60 animals per treatment p u p , and byed 011 skin applicanons in rats (Kinkel and Hokmann, 1973: Rojanapo ez d, 1986; Bunnsrt and Guldenrhal, 1988). again no skin rumour was observed at the Site of application. Likewise, there was no sigaificant difference in body weight or sumival rim. One d y (Rojanapo er al, 1986) found an exccss of mammary tunours (fibrorucomu, adenoman and fibroadenomas) in the 32 cumpan J O of c~ . ~-aria. VOI *. 199s 06-20-2001 12:55pm FronrUNC HEALTH LIBRARY `I 818-866-1 537 T-222 P.004/014 F-034 Hair dyes mrd cancer treated group, and a n o k (Burnerr and Ooldcnhal, 1988), an elevated incidmce ofpinr;rarY adenomas in f d c s . No excess risk ac any other cancer sire was reported. A smaIl study (based on 10 male and IO female m per ueauncnt group), cwpidmd subcurancous injec- tion of mem-phenylene diamine wcry other week for 18 months. Thre was a0 ofmammary lepions (d-`a and admo&), and two sarcomas and two lymphomas ame at rhe site ofinjection (Rojanap et uf, 1986). However, the snall absolute numbers of animals again prccludt any C ~ l u s i o n . Thus, cvidmce of the carcinogcniciry of hair dye cornpaneats in.rodentp i s limittd an& from several methodological aspens, it can be criticizd for instancc, m most experiments only SJeCtcd organs wen:exarm*ncd htologically. Although a r d c amines act W y , tbe urinary bladder and liver o b king their targets (hence skin KumOuII) may not be their predomi~nttarget), the general atmeace of skin tumours at the sites of application is wonh noting. Some excess cancer incidence has b a n reported bur, as a whole, the data am largely inconclusive and offcr no clear focus for epidcmiolo- gical studies in humans. E p i a e m i o l ~data in bmnnlnr Lymphoid neop&nw At least nine cohort studies reported some dam on OccupBtioIAal exposure to bair dyes and rmbSEqumt risk Oftymphoid neoplasms(Alderson, 1980; Kono et al, 1983; Teta et al, 1984; G u h et al, 1985; McLaughiin et d,1988; Shibara at d,1989; Skov et al, 1990; pulrkala et 4 1992; Hnrkc er d,1992). Theirmain d t s are rcponedin Table 2.All studies were based on very small absolute numben of cases (1-22 ob-); still, -t RR estimates WCR close to unity, and all wen within the rrlsxivdymrrowmngc of 0.4-2.5, with no dear exceptions. This means that any subscantially elevated risk can k Confidmtly excluded. when all the data were pooled, a total of 100 lymphoid Iztoptasrm, was observed vs 84.4 ex+ (mrmrmy,RR = 1.2). The corresponding figures WCR 35 observed vs 31.3 expoctcd (RR = 1.1) for leuksermas, one obsmnd vs 0.7 urpened for Hodgkin's disease, 17 obsemd vs 11.2 ;expected for nm-Hodglrio's lymphomas (RR= 1.9, 24 obsavcd vo 20.1 expected for lymphomas unspeded (RR = l.Z), and 19 obaerved vs 16.8expected for multiple myeloma (RR = 1.1). T.bL 2. Summary results fnuD selectcd cohort szudits providine information on occupational exposure to hair dyn and lymphoid neoplasmr A h , yar, -ly kx No.of sub- Lymphoid neoplpnnr R&UW risk Type Obrrrnd Exp&ted Aldaon. 1980 (UK) K m n rl. 1983 ( J a w Tep el ai. 1984 (USA) Gubbrncr ul, 1985 (sllircerlmd) Mchghlia cf ul, 1988 @weden) !#ihar 41989 (ftpm) Sov er d, 1990 Okmark) RJJrrl. cr d.,1992 (PinLd) M F F MSF M McF .M+F F M M+F T o d (all hidogiu) MtF 1.83 1 7,736 11,845 1,380 3,067 016 14.371 3.637 740 Y923 52,923 LEU LEU LEU Mh4 LYM All RAM LEU, LYM. N?lL LEU MM 'W H L HD MM LEU LEU HD NM LYU MM All 3 6 14 3 22 4 I1 3 2 13 4 1 4 1 4 5 35 1 17 24 19 I00 2.7 1.1 4.4 1.4 11.6 1.2 4.3 0.7 17.1 1.3 4.3 0.9 8.5 1.3 . 3.8 0.8 3.0 0.7 8.1 1.6 4.2 1.o 24 0.4 3.1 1.3 0.7 1.4 1.6 2.S 4.6 1.1 31.3 1.1 0.7 I.4 11.2 I.5 20.1 1.2 16.8 1.1 84.4 1.2 %Eu. lairkarmiu: MM.mukipla mvdonu: LYM. IymDhhOmu (unplrcificd);HD. Hodgldn'm diwure. NHL.non Hodgkin's IpPhOmu. ~ur0p.n~ o u r n so~rc a m-tion. VOI 4.1995 33 06-20-2001 ,f 2 :55pm From-UNC HEALTH L I BRARY CLa Vccchia, A Tavani F M+F M+F M+F M+F F M M F 918-Q66-1537 T-222 P.005/014 F-034 32 116 78 59 131 128 54 106 256 607 622 360 -321 2 5 1 0 1 0 1 I 2 I 13 I 12 -- - 0 3.3 4 27 2.2 0.1 0.7 2.4 1.5 -1-3 Thus, froma staljsrical and biological viewpoint, the data on occupational exposure from cohort studies am compatible both with a moderatelyincreassd risk (parucularly of non-Hodgkin's lymphomas) and with the absence of association. However, since v a y littk is known about the aetiology of these neuplasma, no plausible sugwstian can be made 011 the priorities for identification or allowance for potential confoundingfactors. Data on o c c u p t i d exposare to hair dyes and rhe risk oflymphoid acoplasms from casbconuol sNdics arc given in Table 3. At lcast eight SNdieS wen micwed (Gilaer at,1984)Fl& er al, 1987; Boffetm et 01,1989; PeMon et d, 1989; Eriksson and Kalspon, 1992;Pottern et d,1992; Blair et al, 1993; Heningtnn cr al. 1994). Of these, five (Gilcs et ol, 1984, Flodm cr d, 1987; PeMOn er d, 1989; Blair et d, 1993; Hmiogtoa et d, 1994) showed some incrrascd ris4 although most RR esdnufes were d y modcratcdy above unity, and none of them was signihmt. Tho rtmaioiLLg three (Eoffctta er of, 1989; Eriksson and Karlsscm, 1992;Pottun et d,1992)found no evidence of usociauon. Only two studies (persson er al, 1989; Blair er al, 1993) includeddata and risk estimrteS on non-Hodgkds lymphamss, for a total of 14 exposed axd a RR of around 2.4. Case-conaol studies of occupational exposure to hair dyes and lymphoid ncoplssnu have at lam two major limitations: the extremely small number of exposed cascs (only a),and the fact &at none of them was spedfically d e s i p d to address this issue. Thus, no iirm conclusions can be drawn, although, taken together with cohort d e s , the results arc ~mpatiblewith some moderate assochuon between past employment as a hairdrrsser or b a r k and lympboid neoplasms a6 a group, and p a r t k ~ h l ynmHodgb's lymphomas. These epidrrmological hiding gain plausibiliry in -thc light of the Ialwldge that urposurr to s c l e c r e d chemicals may increase tbe risk of 1eUkaemraJ W & 1985) and perhaps other lymphoid ncoplasms (za Vccchia er 01, 1989). Our knowledge of the aetiology of these neoplasms in humans is, however, roo scanty to allow any mon precisc inference to be drawn on potential pathogenic links. In any case, it musf be strclsed that any infmncc is b a d essmtislly on exposure to hair dyes (or otha potcnrial risk factors) by- rs and bprbtrs in thc distant past, considaing the long duration (latency) of the process of cardnogenesis. sin# even tht presence of any iwociation is open to discussion, it is not possible to formulaze any more prrcitc inference on latmcy,or any other dmt time faccor in tho process of -lymphoid carcinogenesis. Additional information on the occupational expo- surcs KO hair dyes of haitdrsssers, beauticians or barbuy and the risk of lymphoid neoplasms can be derived from several descriptive suuiits on poputa- tians. A report for the office of Population Censuses and Sumys (1986). covering occupabal exposure in England and Wales in 1979-80and 1982-83, found uon-Simrifirantlyincreased ramfor Hodgkk's disease and leukacmias in males, based, bowever, on five cascs or fewer. A larger CXCCSP was observed for lung cancer (21 deaths observed vs 7.9 expected), again suggesting elevated smokiug prevalma in rhost 06-20-2001 12:56pm Frwn-UNC HEALTH LIBRARY f 91 W 6 6 - 1 5 3 7 T-222 P.O06/014 F-034 Haw dyes and cmtxr T.bk A Summary mults from sclccrcd cohort studinr providing infomarion on pasonal arposun to hair dycs and lymphoid ncoplanar Hcnnrlccns er ul. 1979 (USA) Thun ez ai, 19w (US4) imsn ny6e LYM NHL HD Mu LLEU umu LEU All 10 0.6 (0.3-1.1,) 37 1.O (0.7-1 2) 6 0.6 (0.2-1.4) SI 16 1 1.1 (0.8-1.S) 0.8 (0.61.4) 49 0.9 (0.6-1.3) 18 1.0 (0.61.8) 227 a9 (0,s-1.1) occupational categoies in rhe past. For females, only breast cancer was in CXCCSII (seven obscrvd, 1.6 expected). This, however, might be refated tu ditrixeaces in reproductive factors or otha conrlafes of breast cancer risk (Boyle, 1988) between hair- drrssm and the general populaTion. In a Danish study (Qcmmcsm, 1977, 1981), no ex- risk ofall cancer sites was obsavtd for males (477 naalignant cancrrs observed vs 512.4 expectad), w h data for females wcrc diflidt to interpm due to difcrences in the dcfinirion of the azupational categories in thc census and cancer registranan data. Some cxoc56 of multiple myeloma or lymphomas was also observed in descriprivc occupational studies from Los Angel- County (Cuidotti et 01, 1982), Washington State (Milham, 1983) and British Columbia (Gallagher et ul, 19891, but the ligures were geneally b a d on small absolute numbers and the results wm o h inconsistcnr betaretn rhe two sexes. otha snrdiesshowed no or link evidence of mwiatim( ~ s s t Hk 4,ign;~ e u b e r egz ~d,~1991). lnfoimntion on peusoml use of hair dya and tha risk of lymphoid ncoplaanu can k obrained from at leas two cohort studies ( H eer al, 1979;Thm cc ui, 1994)and fivc case-conuol sNdies (Smvraky er d,1981;cantorer 4 1988;zahm et tal, 1992;Moms Brown er 01, 1992;Herrinum et u l , 1994). Data fim cohort studin arc summand in Table 4.Among 120,557 fmralr nurses acdw in 1972 and recruited in the Americln Nurses H d h Srudy (Hmnefrens et d, 1979), 10 lymphomas were re&tered up to 1976 among hair dye USCIS vs 15.7 expected, correspondinpto a RR of 0.6. The American cancer society cancer prrvention Study 11 (CSS II) included 573,369 people who wezc followed-up between 1982 and 1989 (Thun er ai, 1994). A total of 941 hawratopoa'tic cancaa wore observed overall, and 227 smong ever USQS of hair dycs. The overall RR for all lymphoid neoplasms was 0.9 (95% CI 0.8-1.1) and thar of non-Hodgkin's lymp&om was 1.0 (95% CI 0.7-1.2), and 110 excess was observed for any other type of neoplasm. Likewise, thcn was 00 iadicatian for the RR to increase w i b increasing duration ofexposum, and the only strata of elevated risk wcrc for long r n m use of black dyes in relation u) mm-Hodgkia's lymphoma (RR= 4.4, based on three cases), and multiple myeloma (RR= 4.4, based on two cases). Taken as a whole, thcrdore, the~tdata provide reassuring evidence on the ab- of risk of lymphoid canerain uscrsof hair dyes, even for 20 or mare years (Thun et ul, 1994;Colditt, 1994). Thc summary renrlb of the fivecascumtml S N d i e S an given in Table 5. An earlier study from Canada (Srawakyer d, 1981),comidaing 70casesofall types. of lymphoid neoplnsmr, showed no evidence of risk. Some modcratesspociarion, however, was appuentin thctblW~~tCtMMODQOlstudisr,~ in varioqi areas of thc United States (Cantor et uf, 1988;Zahm er d,1992; M o d s Brown et ul, 1992). maltchaow.u*ugghfuolnnlyumfobrernsoonf-Hc oxdgplind'scalsyemspahvoamilaabsleweforer analysis. A population-basedstudy c d u c ~ c idn 1980-83 00 malts from Iowa and MinncsoCs (Cantoret ul, 1988) found an a v d RR of 2.0 (9P?confiAcncr inttrval (a)1.3-3.0) for non-Hodgkin's lymphomss, and showed some indication of duration-risk relatianship. The risk esrimamwere adjusudfor smu of tcsidonce and age only in two groups. Another popuhcion-based -0nrrOl d y was cod& between 1983and 1986in Nebraska (7nhm 36eumpclln J O ~ of~cI~ o Pnrevention. VOI 4.199s ----.. -- ._ 06-20-2001 I. _ r 1 2 5 6 ~ ~From-UNC HEALTH LIBRARY 91Q-Q66-1531 1-222 P.007/014 F-034 C L n Vecchiu,A Tavani Table 5. Summay resultsfram selectedcasscoarrol smdics of lymphoid neoplasms providing infoxmationon personal use of hair dyer Aurhan. year. country sa Typed No.of No.of a n d uree crrpord- odds d o slavraky r. d, 1981 (Canada) F All m NR 0.9 C.nror et ul, 1988 (USA) M LEU 578 43 1.8 NEIL 622 53 2.0 zphm CI at I992 (USA) M NHL 201 I1 0.8 F "L 184 106 1.5 (1.7 pumancllr. p) M BD 35 3 I .7 F HD 35 I6 1.7 0.0 -5 F) M MM 32 4 I.8 P MM M CLL 40 37 24 1.8 @.a Pam--tS F) 3 I.o F cu 19 9 1.0 (0.8 pamoamt, F) Monir B m et a/, 1992 (USA) M MM I73 14 1.9 Heninton el 4 19% (LISA) M NM 360 17 13 F MM 319 114 1.1 `HD. HadJan'r d i m , MHL, ma Hod&Ss lymphornu; MM,multiple mydomr: LEU, Ieukramir. NR,aot rrpoMd `et al, 1992), d included 201 mea and 184 women ., on the Surveillance, Epidemiology, and End Results with histologicallyconfinned non-Hodglin's lympho- program found no material asodariw in women mas: 11 m~aand106womtnhadevcrustdhairdyts, (RR=1.1) and only a modest nondlgnificant associa- the RR estimates being 0.8 and 1.5 (significant), tion inmar (RR=1.3) nspccdvely. The RR in women was 1.7 for use of In a leuex dimming the results on non-Hodgkin's permanent hair dyes. Although this study c o M d lymphomas, no assadation was reported between idomatiou OR several risk factors. these risk ocnrparional exposure to hair dyes and lymphoma estimates werc adjusted for age only. Whcn total risk in people with acquired immunodcficitncy hair dye use w89 considered, no relationship was syndrome (AIDS)(Cort er al, 1993). evident with duration (RR1.2 for rhelo-t durahon A popularion-based caso-can~olstudy of multiple caegory, 1 21 years), age at first use, or frequency of myeloma conducted benuem 1977 and 1981 in four use (RR 1.4 for the most frequent category, 2 52 anas participating in the Surveillance, Epidemiology times per year). Some association with duration was and E n d - R d m Programme (Seattle, Urab. Detroit apparcnt in the subgroup of black, browrJbnurerrc or and Arlanra) included infomuon on occupational nd dyes only, but this can be cons&red an a and personal exposure to hair dyes (Heninton et ai, parreriori subgroup andysis, if not due to selection 1994). Among womm employed as haitdrrssap far 6 and confounding bias, since greyiug hair might be a months or more, the OR waa 1.3, in rhs aboeace, correhzc of both hair dye use and lymphomas risk. however, of any duraumrisk rclattonship(ORfor > Howmr,eveah tbis subgroup Chaewasnoevidence s,ycan (xpomre-0.7). only one man rcportcd of a fqucuc-risk relationship. The same study occupational exposure. The ORs for personal use of reported data 011 32 male and 40 female cases of hair dyes were 1.1 for womco and 1.3 for men when muhiple myeloma The RR &marcs were 1.8 (not ~ all data were cowidered,and 1.0 and 1.3 when all sclf- significant) for both sacs and, again, rhae was no -. rtspondents were coasidsnd. None ofthcw Csrimates evidnrce of any relationship with duration, latency or was signidarnt and tho auhors concluded that the hqaency of use. The data were scanty on Hodgkin's study provided littie evidence rhat exp0sure.mhair disease, and there was no evidence of BsSociatioawith dyca incream myeloms risk. - 1 The results of epidemiological studies on hair dyes Another srudy of 173 ma& cases of multiple should be discussed With reference to our knowledp myeloma f m Iowa (Moms Brow et d, 1992) of the aetiology of lymphomas and myeloma. The suggested a moderate assodadon (RR = 1.9,950/aCI beat recognized risk factor for nou-Hodgkids lym- 1.O-3.l), but proVided no details Ot the type of colour phomas is severe immunodqmssion, sincs the risk is u9c4 and made no allowance for any covaxiate, substantially increased afkr (kidney) rransplanmdon except age and vital status. A canconml study based and in subjtcul wirh human imxnun- virus 36 EUMP.~Joumrl of evnCer Rmnrion. Val 4 iWS 06-20-2001 12:57pn1 From-UNC HEALTH LIBRARY 91g-g66-1537 T-222 P.006/014 F-034 Hair &es a d mar I (HIV)infedon (Kinlmer ai, 1979; La Vecchia et ai, mas and orhu lymphoid ncoptaansmay be of intenst r 1992; Hangc and Dcvcm, 1992; Rarce and for fum eprdaniologkd stpdics, but available Berhwairc, 1992). Myeloma is related u) ionizing evidence is inadequate for any public health or )I radiation (Cuzick, 1981), but link is knownabout the regulatory implication. role of infections, autoimmune diseases, immunode? prwion or other potential risk fanors ( G r a m 4 cz B M G Tcancer al, 1991). At least seven cohort studies based on 10 d B " t D C&d incidence and mortality for both types of popuIati003 provided informadon on occupatiaaal ncoplapms have substantially increased over the last exposure to hair dyes as bairdrcssns1 b a h t 3 P few decades, airhougb it is not clear how much of the beauticians and cosmet4logisu, and sub8cquent h a # is real and how much is due to improved bladder cancer risk (Alderson, 1980; Koa0 cr ui, b instrumemandcriteria for diagnosis and certification 1983; T- e?d,1984; Gub- et ai,1985; Skov et ai, (Doll and Peco, 1981; Harts and D c v a 1992). A 1990; Hrubec et d,1992; Pukkala et al, 1992). The P recent and@ and quantification of the role of main finding arc summana zsd in Table 6. Fourstudies various risk f a a m on time trends for non-Hodgkin's wen from Europe (one was a c4opcfBfive study from lymphoma incidcncc concluded that about 80% of four Nordic counnics), two from the USA and ODC these trends rrmain u1Iu[p18iI1cd (IHartgc and Devesa, h m Japan, for a total of over 81,OOO subjccrs(63,828 P 1992). females and 17,247 males). In seven o m of 10 studiw, z In panicular, there is no clear evidence of any some moderate assodap;on was reponed. A total of association bctwcen exposure 10 chemicals and the 183 cases of bladder cancer were o k e d us 129 risk of disease, even at rhe relatively high doses found expected, conespondingto a summary RR of 1.4. 'Ihe in the occupational environment. Thus, it is diKcult association waa apparcndy, though not omrifiunrly, 10 draw any inference on the roles of the chemical SUO&= in mlcs (RRa1.6) than in female b constimu of hair dyes, particularly in the a h a of (RR = 1.1). Thus the results of cohort studies arc reliable -ts of exposure a d of allowance for compatible with some moderate association between indimon of socioeconomic status, which is o h a paa professional exposure to hair dyes (or other correlate in hair dye use. There is also no information subsrancts in the working envinmmcnt) by hair- on the doscs, biadionibution and bioavailabiliry of dressers and barbers and subsequent bladder cancer potential carcinogens in hair dyes, with particular risk. D reference to thcir conccnrrations in lymphoid tissues. There is,however, some indication that, at least in In cmclusion, therefore, two studies (Cantorer ai, the past and in selected populations, rhe prevalence of 1988; Zahm er ul, 1992) reported a positive, though cigarette smoking, a recognized risk factorfor bladder moderate, association bctwccn personal use of hair cancer (US Depanmmt of Health and Human dyes and non-Hodgkids lymphomas, and two d e r Services, 1982). w h@er in rhcsc occupaKiod datasets mggested a linlr with multiple myeloma groups thau in &e gcneral popularion, and cohort (Zahm et al, 1992; Moms Brown ez al, 1992). The studies were g c n d y unable to allow for this major d m howma, h an be criticized a epidemiologi- confounding factor in the anaIysM. It is thus possible caI grauads, prtidarly because no allowance was that at least part of the apparent ZLuOciafion is due to made far potenrial confounding facton, and the confounding by tobacco spoldD& as olso rugocsted d t s lackad a plau6ibk biological inmpmation. bYbwwF- * 'oninmab,whowatmorr In addition, epidemiological evidence rnay be influ- frequently 9mbJecR,than females. enced by sclmilm publication- of positive resula Some mdication on the smoking issue can be ('publication bias'), since, panicularly in the absence duiucd from casbcoatroi studice, which p e r a l l y of plausible biological hypotheses, there was little inclbdui information on smoking, and could therefore rationale for publishing negative studies, ie those allow for this variable in the d y s i s . The main hding no &tion. findings from casbconuol sfll(iicsam mmmarkedin since thc -on was appaxmtly stronger or Table 7. Out of 11 sfudies (Wyndm et a1, 1963; r e d 4 to subjects using black hrmnm/brunetta or Dunham et ai, 1968; Anthouy and Thomas, 1970; red dyes, it is also at least theoretically conceivable Cole er d, 1972; Viadana et ai, 1976; H o w er d, that gnying hair m a y be a corrclatc both of hair dye 1980; Vieis and Magnani, 1985; Morrison et d, usc and of the subsequent risk of lpphoid 1985; Risch er 01, 1988; Silvumau et ai, 1989, 1990), ncoplarms. seven showed some assodation between occuparid Thus, suggestions of a positive association between exposure to hair dyes and bladder cancer risk. Most hair dye use and the risk of non-Hodgkin's Iympho- estimates, however, were around or only slightly 37f2-n Journal of aWefPRVmWOn. vOl4 1995 * .0.6-2,0-2001 12 :5 7 ~ a F rom-UNC HEALTH L IBRARY CLa Vecchh, A Tovani Auzhom yar. -w .al- 1980 (UK) KOM 81 4 1983 (Japan) Tstr CI a& I984(USA) 09#rrm 8I d. )-s( slrav er a/,1990 (Norway) srrcdg FirJIld DemMlk PlrldFllr 81 Id, 1992 (FinlraQ) Hrpbcert d. 1992 (USA) Td srm M F F P M P M F M P M F M Q M F M FtM 916-966-1 537 1-222 P.000/014 F-034 Na of -9 1.831 7,736 11,845 m m3 4,356 16942 -2 9,138 &a 9,497 4374 3,637 740 63,828 17,247 81,075 ~~ Blrdderancsr ~- Okmcd p-'ilwcud 7 5.6 0 1.o 14 10.3 2 1.5 11 5.3 11 7.2 23 15.1 6 13.5 54 36.6 3 1.8 0 0.3 7 4.0 41 20.0, I 2.5 3 4.3 44 41.8 139 87.2 183 129.0 kluiw *. 1.3 0 1.4 1.3 tl 1.5 IS 0.4 I .S 1.7 0 1.B 2.1 0-4 0.7 1.1 1.6 1A A4 M M M P M M F M M M M M+F M F 300 265 812 356 105 NR 480 152 512 430 399 116 836 2.100 652 4- 4 2.8 4 4.1 4 0.6 1 1.1 -5 3 1-.s 2 9 0.9 7 1.0 2 I.3 1 1.0 2o 0.8 * 28 1.3 17 1.4 abow unity, and not statidcally signilkant. Still, tbc gmaal evidcnce from -mol S is in agreement with that fmm cohort investigations, and is cornpatable With a moderately incrrascd bladder c r n c o r r i s k i l m o n g a~nd barbas. Some excess risk of bladder amd other urinary Sites cancers was also obscrvod in descriptive occupational d i t s From New ztslond (peua and Howard. 38 buapmnJ o d of- R u e a h . Vol4.199J 1986) and hfassachusetrs pubrow and Wcgman, 1984). but thc figures were b a d 011 srmller absolute numbm, snd are incoasr*sttnrwitharherscudia (Mcnck cr aZ, 1977; Ncuhcrgcr et uZ, 1991). It must be secssed that most inFdnnarion Fmm descriptive, cohort and casecontrol studies c o o c ~ ~ l l l exposure in the nlativcIy dhtant past, when there wlls lesa strict control of carcinogenic substanas in hair 06-20-2001 12:58pm From-UNC HEALTH LIBRARY 91 g-g66-1537 1-222 P.010/014 F-034 su H w e er ai. 1980 (Canada) F M 152 NR 480 8 0-.7 Hart# ez ul, 1982 (USA) f 733 443 0.9 M 2,249 172 1.1 Ohno er ai, 1985 (Japan) F 65 42 1.6 Claude cl ul, 1986 (Gumany) F 91 NR No associalion M 340 NR Nmm a ul, 1989 (USA) F 66 41 1.5 M 195 15 1.3 m.IWCrrporrcd;--,M~parsdcOnEO~ dyes and of the working conditio^^ and activities of tionconductedby the US NationalCancer Institutein hairdrtsocrs. There is therefore a reasonable biologicaI 10 US states (Harts et ul, 1982). The study was backgmuadfor the obsmved amciation, which could originally focused on anificial SwcttePQs, but also be compatiblewhh latency and other limc-nlated included adequate infomation on hair dyes, tobacco facrors in the process of bladder ca.rcino@ smoking and other major covlviatcs of inuzcst. There @ecarli et al, 1985). Thus, a m n t IARC working was no evidence of an asrcocizrdon between hair dyes group (IARC/WHO, 1993) stared that `occupationas and bladder cancer risk,the RR estimates bdng 0.9 in hairdnsser or barber entails exposures that arc f d t s (based on 443 exposed cases) and 1.1 in males probably carcinogenic'. (based on 172 exposed cases), nor wen there any However, rhe cstimsfedrelative risks, as well as the mds in risk with fhquency or durationineither sex. o v d p l c d estimam, arc only moderately above The other four 9Ndios were much smaller, and thus unity, and hence compatible with m n and biases in were ks infonnauve, iududing only a few dozen observational epidaniol0,gical studies. A major open exposed subjects, but the RR esham were again quntion, moitovct. is whaher current cxposurc to clocle to wily. modem hair dyes is still related tp some excess risk, or Thus, the O V &~ hf- cohort and c=- w h e b the selective elimination of carciaogenic conuol eeudics consistbzrrly exciudes any appreciable compounds (Table 1) over FeccnK y t m has reduced and measurable rislr of bladder cancer from penonal any such risk m nonmcasursblc lcvels. Only further usc of hair dyes. srweiuance and futurr studies will provide answers. personal hair dye use in relation to bladder a~cs!r Other neoplannr risk was considemi by at least one cohort and five other ius, rhc bieast has bem most taswxmml studies. The cohm study ( H e n n a el exreooively smdicd, bccause of rhe high incidum of al, 1979) was &e Nunu' H d t h Study, based on the disaasc in women and thc report $a dstan-td Y 120,557 fcmafe Amencan n q active in 1972 and elevated'RR of breast cancer from a small study followed-up for incidence of d i s e k until 1976. Five published in the 1970s (Wkrand Shafer, 1976). Of cases of bladder cancer wen rcgisurd vs 7.4 100 paticnu f m a dirrical praaiOe in New York, 87 mpecred, comsponding to a (nonsislificatit)relative had-bcen l0ng-m USCIS of hair d m risk of 0.6. wirh 27% of a COmpLuiso~group, with a RR estimate ?.he main fbdings from case-conml mdies are elevated over IO-fold. 31' d in Table 8. Five d e s were considered At lean scvm caswxmrrol ONdibs (Kiden ut ai, (Howe et al, 1980; Hmgc et Ol, 1982; Ohno et 41, 1977; Shore e: al, 1979; Stavxaky et a& 1979; Nasca et 1985; Claude et ai, 1986; Nomura et al, 1989). two d,1990, 1992; Wyndcr and Goodman, 1983; Kou~ig from the USA and m e each from Canada, Germany et 01, 1991) and one cohon study (Hennehs cr a4 and Japan. By far the l a q m srudy, and hence &e 1979) wen subsequently published, including a total mOSf hfOrIIlaUVC O m , includin6 O W thrCC fOUrLhS O f of over 3,200 breast cancer cases. Some of these wcrc the cases and of thc ex& subjms, was a specificauy designed to address thc hair dye issue. multiccntsc population-bd, casccontrol investiga- N o ~ efound evidence of an assodation. and it is ~ ~ a morrcl.ncrr~eventioa.Val 1.19~s39 06-20-2001 12:58pm Frorn-UNC HEALTH LIBRARY *" I 919-866-1531 1-222 P.011/014 F-034 C& Vecchk A Tavani therefore now established that hair dye use is not a risk factor for human breast cancer. Same accuparional cohort studies (eg Teta et a!, 1984;Osorio cr al, 1986; Skov et al, 1990; Hnrbec et ul, 1992) nponed an excess lung cancer rare in cohorts of subjects occupatidy exposed to hair dyes. The m)ults,however, were not consistent acroI studies and oftea not even within studies (eg Were Observed bOKweCn d C S and females). Most sNdies, mol#)vct. were unable to allow for Qnoking, by far rbe most impOnanr risk factor for hang canccr (Doll and Peter, 1981; US Dcpacnnent of Health and Human Services, 1982), and chmfore the most reasonable intezpreCarion of &e associauon obsmed in a few studies is a somewhat elevated prcvalance of smoking among hairdressers and barben in rhc past. Likewise, no s i p i h m association with lung cancer was observed for personal use in a small sNdy (60lung cancer cases, Stavraky GI ai, 1981). An cxcm of laryngeal cancer rates in barbers (Viadana er ai, 1976) can be interpreted along tbe same lines, although in that study the RR Cstimare (2.8 based on 10cases) was not materiallymodified by allowance for smoking. The results for other neoplasms am scarered. An elevatedrisk ofcancers ofdu:cervix and of the ovary was reponed among hair dye users in a multi-aite ca8e=controlstudy conducted in Canada (Stavraky et al, 1981). Then may have been some confounding by sexual and rrproducuve factors, w h i art known comlatcs of the risk of h a c neoplasms (Brinum and Frammi, 1986; Pamahi et al, 1991) and, wnceivably, of hair dye use. However, a most careftdly designed and conducted c a s b c o n ~s~nlrdy of ovarian cancer from Atheos, Greece (Tzonou et ai, 1993) found a sigdcant association, with an RR of 2.2 for frequent hair dye users. Two studica from Awrrrli;r (Hohnoa rad Annsmug, 1983: Annstrorrg and Hoknan, 1985) and Denmark (Osrerlind et ai, 1988) found no association benveen hair dye use and ski0 melamma. ' Data are inadsrent for thc relationship between hair dye usc and brain cancer (Ahlbom er d, 1986; Burch af al, 1987). as well as for some rarer neoplasms, such as salivary ghad cancers (Spitz er d,1990), and childhood neoplasns in tbc oflspling (Ktamcr et al, 1987; Bunin e?ai, 1987; Kuiljun el d, 1990), particularly since some scattered positive findings may be influenced by publication bias. Thus, thcre is at present no convincing evidence that occupational or personal exposure to hair dyes is related to the risk of any ~ l ~ p l i ~ samt ssiw orher tban the bladder and the lymphoid tissue. C O ~ ~ ~anOd iIdcEations for tbrther teselveh A substantial amount of epidemiological evidence is available on the possible assodation between hair dyes and smraf cancer sites, permitting a number of wdlcstablished conclusia Occupationnl exposun KO hair colouraxxrs or pasonal bair dye use is not related to breast cancer, nor ia it c a d y relaud to lung cancer. Likewise, r,ture is no convincingevidence lialdng hair dye use to cancers d t h e skin, brain, fdegtnitaltrad,stomachandorhcrdigcsrivesites. Further, personal use of hair dyes docs not enrail any measurable excess bladder cancer risk. Thc two remaining opca queshns art the possible relationship between bladder cancer and (past) OCOUpQtionat exposure by hnitdr#sen, barbw and beauticians, and between bair dye sxgos~re and lymphoid neoplas;ms, with rrfmncc to bath bccup8- t i 4 and persoaal me. g e association between past occupuonal expo- sure to hair colourants and bladder cancer risk is reasonably consistent on epidedological data, and plausible on biological grounds oriand Elliott, 1981). The question arises whether the risk has been eliminated, following selective nmoval of carcinogens from hair dyes and improvements in working conditions over the last few decade, or whether some exccso iisk is still idenrifiablc. Thus, furehKmonitoring on more rrctllt cohorts will be an issue of scientific inttrcst and public health r t l a ~ ~ n a . An associauon between p o d hair dye use and non-Hodgkin's lymphoma and multiple myeloma wag nccntly reporred from rhrcc csse-conuol studies (Cantor et d,1988;Zahm et d,1992;Mom6 B m et ai, 1992). The dative risk estimsreswere moderate, ~ d belowy a facror of two, and theFc wa9 no clear evidence ofduration or a doJbrisk relatiomhip. on spirlam;rlogicol grauart, tbedkKC9 IlmJBbdiagS require fwllcr mdy, paniculalrly since rhcy may be i z l f l d by StlCctiMpublication ofposidve fbdhlgs. This publicauon bias is plausible, considering the ;hence of any rdiabk idenrificarion and measure of potential carcinogcos in hair dyes which could have reached& bonc marrow or thelymphoid tissue,and hencc of any ooavindng biologhl link for these epidemiological observations. These cautious notwithstanding, repow from unpublished investigations and from fbrthcr casecouuol studies of lymphoid neop&sms and hair dye use would be of interest, wbenas cohort studies are unlikely SO provide m~aningfdi n f o d o n ~1 such rare diseases. These ca-onuol smdics, eveu if &aped to invcsdpre; s c d potcnujl rids f-ra 40 Europan Joumrl of Cancer Pmranlim. V o l 4 199s ri , L b 06-20-2001 12:59pm From-UNC HEALTH LIBRARY 919466-1 537 1-222 P.O12/014 F-034 Hair d,vcp and cancer for lymphoid ncoplagns, considerkg thc lack of definite knowledge on their aetiology, should none- theless include a suuctured and, ideally, an (externally) validated d o n on &e history ofhair dye use. They should also include further informadon 031 poscible confounding faaors, and hmct adcquau allowanceshould be made in the analysis for the main covariatcs of potential interest. Ref-= Ahlbom A, Navier IL,N o d S,Olin R, SpirOnarrB (1986). Nmoccupati~u~rlisL indicators for asuocyumm in &In Am 3&&miOi 124: 334-7. Aldcrmn M (1980). Canccrmortality in male hairdrrssers. J Epldriol CornmvnHealth 34 182-5. Amsr BN,lkunmen H, Y m k i E (1975). Hair dyes arc mutagenic ideatidcotion of a variay of mutagenic ingredients. Pror Not1 A c d Sci USA 72: 2423-5. Anthony HM, Thomas OM (1970). Tumors of the Urinary bladder: an analyais of the OCCuPPfionsof 1,030 paticnu in EngIand. INCI 45:879-95. Annmong BK,H o b CDJ (1985). Hurdrins4n's mela- notic frrckle melanoma and thc use of non-permanent hair dyes (letter to tba Editar). Br J Concer 52: 135. Blair A, Linos A, Stewan PA et aI (1993). Evaluarion of risks for non-Hodgkids lymphoms by occupation and industry urposrws from a casbcond study. Am fnd Med 23: 301-12. Boffha P,StellmM SD,Garfinkel L (1989). A caswxntrol study of mdtipk myelomanmaiin rht American Cancer sodcry prospecrive study. h t J Cancer A3: 554-9. Boyle P (1988). Epidemiology of breast cancer. Ballliere`s Clin One01 2: 1-57. Bnnm LA, Fraummi JF Jr (1986). Epidemiology of uterine arvical cancer. J Chron DU 39: 1051-106. Bunin GR, Kramer S. Mamm 0, Meadows AT (1987). Gestational & facton for W W mor: mults of a -rrOl m y . CMmRM 47: 2972-7. B\mh JD,CRib KTP, Choi BCK, Miller AB, R i d HA. H m GR (1987). An exploratory c ~ u osruldy of brain tumors in adults. JNCI 71)t 601-9. Bmwa Goldendud EX (1988). Multipnmti~n reproddon and c a n i n o ~ k k ysadier in Spragne Dawiey ra5 expored topically to oxidarivc haircolour- ing fmulauons containing p-~haykncdiamine and othrr amslutic amines. Food chmr Toxic01 26c 467-74. Bumar CM,Latunan BM, GioMecbkri R, Wolmtt R, !bhR, Keplhrp M (1975). L0w-m toXiCiw studies on axidPCion hair dyes Food CamKr Toxirol13: 353-7. CantorKP,Blair A, EverettG et d (1 988). Hair dyeuse and risk 0f-a and lymphoma. Am 3 Pnbk H d K h 78: s70-1. Claude J. KUIIZCE, Fmml-Beymc R, PaCrLowdd K, schneider J. Schubcrt H (1986). Life-aylc and -pa- ti@ risk fpctorsin cauter ofthc lowcr urinasy vocs. Am J Epdcmiol 1U: 578-89. Clemmesen J (1977). Statistical studiu in tk aetiology of rmlignurt neoplasms. V. Tmds and risks, Denmark 1943-72. ACTOPathol hfkrobiol Sand Suppl261: 1-286. C1- J (1981). Epidcmhlogiul s d i w inco cho possible OucinogeniCiry of hair d p . Murat Ru %7: 65-79. coldiu GA (1994). Hair dye and cancer. nauuring Ovidencc of no d a t i o n . JNCI 86c 164-5. Cole P, Hoavn R, FriaiCll GH (1972). Occupation and urnceroftheIowerPrtnuyrractCPuvr29: 125M1. Coanetic, Toilttry and Fragmncc Assodation (undated). lairdmumWashinwn DC. Cod TR, Dosemecl.M,Rothman N,Banks RB,Biggar RJ (1993). N ~ - W & ~ 'lYSmph0m;r yrd exposure to hdr dyes among people with AIDS. h I Public Heulth 83: 590-9. Cuzick J (1981). Radiation induced myelomatoriS.N fig1J Mcd 361: 204-10. Deauli A, Rto J, Piolrno G, La Vecehia C (1985). Bladder cancer mortality ofworken exposed to aromaic ami= ~ a l y s i sof models of carcinogen&. Br J Cheer SI: 707-12. Doll R;b o R (1981). Thc .of ~ ~ ~ lqlucartrtritacLve CJdmotesof avoidable rislrs of CIUICC~in the United States today. JNCI66: 1191-308. Dubrow R,Wegnan DH (1984). Caacer and OOCUPQUiOn~ Masrachruda:adcathccrtihtcst\rdy.AmJIndMed6: 207-30. Dunham LJ,Rabson AS. Smart NL,Frank AS, YOU6 JL (1968). Ratcs, ineavinw and pathology ptudy of CILllCcT of thdurinary bladder in New Orleans. Louisiana JNCI 41: 683-709. Eriluson M, Karlsson M (1992). O c c u p a t i d and othcr environmental fanon and multiple myeloma: a popula- tion based casc-canpol study. Br J I d Med 49.95-103. Flodin U, Redrikpon M, PeMon B (1987). Multiple myeloma and engine #hausts. hrsh wood, and CJCO- so=: a csstircfermt siudy. Am J Ind Med 12: 519-29. Gallagher RP. ThreIfaIl WJ, Band PR,Spiaclli JJ (1989). OccuptwnaA monaliry in Bratuh C o h b i a 1950-N84, Worker`s Cornpeasation Board of British Columbia, Vancouver, BC. Giles GC, Lickh JN.W e MJ,LomnthalRM, -ton J (1984). Mycloproriferodvc and lymphoprolifuative dis- orden in T d , 1972-80: occupational and fnmilial aspecrs. JNCI 72: 1 2 3 3 4 . Gmmenzi A, Batrino I, D'Avruw B, Negri E, Franceschi S. La Veechir C (1991). Medical k o r y and the risk of mdtiple mydamk Br J Concer 6 3 7-72 G u b E, Raymoad L,S~lneaumPM (1989). lnrrrased risk for& bLdda cancar amonga cohonof nulo ad fmlk~hai&at,mftom Geneva. `Inr J Qidemiol 14: 549-54. csyidotti S, Wnght WE, percn JM (1982). Mulaple myeloma in cosmnolofim. Am J 1M Med 3:169-7 1. H a r ~P~,eDewsa SS (1992). m a o n o f the impact of known risk -f OD time umds in non-Hodgkin's lymphoma incidence. Cancer Rrs S2 SS66s-9~. Hange P. Hoover R, Alcman R et d (1982). Use of hair dycs and &k of bladder cancer. Gmcw Res 4% 4784-7. Hrzmekm~CH, R ~ m e rB, Ba- C, S- F& Bain CJ.Peu,R (1979).Ust ofpanloeat hair dyuand cancer among nglswredn m . Lunar i:1390-3. Hcrrinm U,Weiss NS, Kocpsell TD et d (1994). Exposure to baircoIoriag pmducus an4 the risk of multiple myeloma. Am J Pub1 Hralrh 84: 11424. Holman CDJ, ~rmsaongBK (1983). Hutchinson's mcla- notic frccklc melanoma avraciattd with non-pmnanmt hair dpJ. Ur J Cmccr 48:59-1. EumpanJoumd or can- m u o n VOI 4 1995 41 06-20-2001 12 :59pm From-UNC HEALTH L IBRARY I.. P18-966-1537 T-222 P.013/014 F-034 C L a Vecchia, A Tavani Howe GR,Burch JD, Miller AB et a1 (1980). Tobacco use, Manchester, UK and Nagoya, Japan. I &ldaniol occupah, coffee,various auticnts and btaddcr cancer. Commvniry Health 39:294-300. JNCl64t 701-13. Nvca PC.Baptiste MS.fieldNA,Meager BB,DcMiurino Hrubec Z Blair AE, Rogot E, Vaught J (1992). Mortdiiy R (1990). An epidemiologic Qubconpd study of h a s t lpIjkr by Omcpotion Among US Vereraru of Knows cancer. Am J EpicSmrial I#: 7-1. shraking h t w 1934-19&) Voi 1 (NMPub1 NO 92- Nascs PC,ElrpttDc MS,Field NA,Mcty~crBB, DcMartino 3407); N a h a l Innnimta of Health, Bechda, MD. R (1992). An epidaiologic ese-contrd study of breast h ~ t i Adgency for Raiaucb on Cancer (1993). cancer and uposurc to hair dyes. Ann &i&nwI E 577- oceupauonal exposum of bairdresrrn and brrrbm and 86. pmonal we of hair colourann; 6omc hair dyes, com& Neubcrgcr JS, Browaaoa RC, Moronrs RA, Chm TDY colouranu, industrial dyestutTs and aromatic umber. (1991). Auociati~nof brain ca~uz1witb dental x-ray^ IARC Monogr Eva1 Cbchog Risks Humans 57. and accupruiOnin Miseouri. CutaceDetect Arv 1%314. Jacob MM, B v n m CM, Penicnak AJ er al (1984). Nomura A, filonel LN,Yorhip*vaCN (1989). Smoking, Evaluation of the d u t y and C&M@&Y of hair alcohol, occupadoa and hair dye use in cancer of the dyes in Swiss mice. Drug Ckrnr Toxic017: 573-86. lower urinary tract Am J &iduniollS 1159-63. Kinkel HJ, H o h a m S (1973). Study of long-mm offioc of Population Cmmres and Surveys (1986). Occupo- percu?aneous toxicity and Oorcinogeniciry of hais dyes rional mortal& nkennhl wp&ment, 1 9 M . ; Her (oxidizing dya) in rats. Food Comer Toxic01 11: 641-8. Majesty's Stationery O h , London, UK. .Kinlcn U,Hmh R, Garrod A, R o d n y a K (1977). Use of Ohno Y,Aoti K,Obam K, Morrison AS (19tq. casb hair dya by pUimrr with breast cancer a caoocontrol conid study of urinary bladder cancer in mcvopolitan study. Er hied J II:3 6 M . Nagoyp. NCI Monogr 69.229-34. Kin&n U,shdl AGR, Pet0J, Doll R (1979).Collaborative Osorio AM,B c m d L,Gambrant DH,Peters JM (1986). United ~ - A u ~ m sd yOf c~lul~erh~ P ~ T S Inv#tigadon of lung cancer among fmuk asamdo- tratrd witb immuaosuppdvc drug. Br Med J k giau. J &cup Med 28: 291-5. 1461-76. OsterlindA, Tucker MA,Scone 81, J ~ J QO!M (1988). "he Koenig KL,plutanack BS,Shore RE,Stmx P (1991). Hair Danish casbcontrol study of cutaarous maIiaa8nt dye use and breast cancer a cass-conaol study among belanoma. IV. No association with nutritional fanors, eCnening pmicipaars. Am J E$iclrmo/ 133: alcohol, moldng or hair dyes. hr J Conem 43:825-8. 985-95. ParaPini F,F~nceschiS, La Vecchip C cr d (1991). The Kow S, Tokudome S,Ikeda M, Yoshimusa T,Kuratsuat epidemiology of owuim cancer. Gynecol Oncol43: 2-23. M (1983). Caaca and other causos of doath among Peane N, P (1992). I d g b- dm ~ - fanale beauticha. JNCI 70: 443-6. Kramer S, Ward E, Meadows AT, Malonc KE (1987). HodNn's lymphoma: occupational and mviroamenral facton. crmcer Res s2 0:54963-5Ooe. Medical and drug tisk facrorr associawdwith neurobb Pearce NB,Howard IK (1986). Occupation, socialdassand UILIIIL: a casecontrol study. JNCI 78: 797404. mala cancer mortality in New Zealand, 1974-78. Inr 3 KUiiten RR, Buah GR,Nass CC, Madoan AT (1990). Epldrmlol Ts:456-62. Gestational and familial risk factors for childhood Passan B, Dahlander A-M, FredriLsoon M,B r a s HN, asmcytoma: d t a a casecontrol study. Cuncer Res so:2ma-12 Ohlson CG, Axclson 0 (1989). Malignant lymphomas and ocnrpltid exposuns. Br J Znd Med Y: 51620. La Vecchia C, Negri E,DAvanu, B,Ftanccschi S (1989). Po- LM,Hdnanan EF, O l m JH.M n E Blair A Occupation and lymphoid neoplasms. Br I Cancer M): (1992). Multiple myeloma among Danish women: 385-8. anploymznt him and workplace txposurrs. Conerr La V e a h h C, Negri E, Franceschi S (1992). Medical Cmrre~COW^ 3: 427-32. hhtw and drc ridt of ~ l l r - H o d & i ~l'y~mph-. PuWrda E,Nokro-KoivistoP,Roponcn P (1992).Changing mceotnMcSn(E1p9l8d5k)r.mTlohlemleuokmemairaks!sePplieava1m:i5o33lo-6g.d aspects. In cancer risk patterns among Finnish hairdrcrrcn. In8 Arch ocryp &v&on H8dth 60: 3942. b &ikmblogy d 4mtatl8&9. Yo1 6 (ed Riscb HA,Burda JD,Wlbr AB, Hilt G 4 $4- R., IHOwe AM); OxfordUniversity Press, New York, pp. GR (1988). Occupndonal f- and the incidence of 2065. cancer of rhe bladder in Canah Br J b d Medbs: 361-7. Mamnoski GM, Elliott EA (1981). Bladder cancer Rojanapo W,Kupradhum P,Tepsuwrn A, chudmapewin epidemiology. Ejidrmlol &v k 203-29, S, Tanyakaset M (1986). Camiuob&' ofan oxidorion McLaughh JK,Liw MS,Sumc BJ er ul(1988). Mulaple product of pphenylencdiamine. Carcinogenesit 7: 1997- myeloma and occupationin Sweden Arch Enviran Health 2002. 43: 7-10. W e CE,Jones EL (1977). Effkcu of npclredappliurionS Men& HR. Pike MC,Headnson BE,ling JS (1 977). Lung of two semi-permancnt hair dyes to the skin of A and cancer risk among beautician urd other frmalc workers DBAf mice. & J Cancer 36:467-78. brief'commuaicuion. JNCI 1423-5. Sbafer N, Shafer R W (1976). Potcnrial of Caniaoguxk Milhorn S (1983). Omrpntbrrol Morraliiy m Washingion effects of hair dyes. NY Stale J Med 76: 394-6. S W , 1950-1979. (NIOSH R C S C a d tepoK); US Dcpirn- Shibata A, %saki R, Hamajima N, Aoki K (1969). m a t of Health, Educlrtionand Welfare, Washington DC. Mortality of hematopoietic dironiars and hair dye uw Morris Brown L, E m GD, Burmeister LF, Blair A among b a r b . Acra HormaroI Jpn 52: 116-1 8. (1992). Hair dye use and mula'plemydoma in white men. Shore RE,Parccrnrrck BS,Thiesmr EU,Sadow M, Forbes Am I Publu Heal& 82: 1673-4. R,Albcn RE (1979). A uwc-controIstudy of hair dye use Momson AS, Ahlbom A, Verhock WG at uJ (1985). aand breast cancer. JNCI 277-83. O o e u p h i o m -4 b l n d d m r QMOU k Booron, Uah, Silrs- DT, Lrrin LI, Hoe*rr RN, X u r ~ ~P o(1383), 42 Eumpns lavluld C . n a r ikvmdm. Vol4.1995 06-20-2001 01 :OOpfi Frofi-UNC HEALTH LIBRARY 91W 6 6 - 1 5 3 7 o c c u rid~u o~fbladdercanawin the United Smm. -I. White mea. JNC181: 14724. S i DT,Levin LIDHoover RN, H8rtgC P (1990). Ocnrprtioualrisksofbladdercanaramongwb~ in ths Umwd Srucs. Am J Ep&emiol132:453-61. Skov T. Aadsrrm A, Maker PIu,ltkrla E, Weincr J, Lynp E (1990). Rirk for cancer of tha \Irinory bladder am% in the Nordic countries. Am J Ind M d 17:217-23. Spitz Ad& Fuegm JJ, Gocpiert 11, Nmell GR (1990). s.livuy gland cancer. A cam-control invcatiaptiOn of risk f a m a Arth OtolrrfyryolH i d Neck Skrg 116: 11634. -9 Iud,clulte EA, Donnw A (1979). -Wl study of hair dyc taw by patienn with breast cancer and eadomsairlm.JNCI 63 941-5. SUnaLy KM.ClarkeEA,DomnslA (1981). A aso-conuol smdy of hairdye we and c ~ u r c ro~f various si=. Br J Cancer 43c 236-9. Tcta MJ, Wairatb I, Wincr J, Flanrwxy R (1984). Cancer incidence among cosmcmlogists. JNCI n: 1051-7. Thun MJ,Altaknue SF, Nmboodiri MM, Calk EE. M y m IXJ;Heath CW Jr(1994). Hair dyeuseand risk of fatal cancan in US women. JNCI b6: 210-15. c T-222 P.014/014 F-034 Hair d y s and cancer