Document OzeqR8n1d6EvgbdgQ0bO6E9pp

Jkzffp. Environmental Research Volume 42, Number 2, April 1987 CONTENTS Review L. Kiremidjian-Schumacher and G. Stotzky. Selenium and Im mune Responses.............................................................................. 277 Regular Articles Saroj Gupta. Physiological Stress Induced by Sublethal Concen trations of Phenolic Compounds in Notopterus notopterus: Mea surement of Hydrolytic Enzymes................................................... 304 Sandor Takacs and Albert Tatar. Trace Elements in the Envi ronment and in Human Organs. I. Methods and Results. . . . 312 Hitoshi Michibata, Yasuyuki Nojima, and Manabu K. Kojima. Stage Sensitivity of Eggs of the Teleost Oryzias latipes to Cadmium Exposure.......................................................... ..... . . 321 L. Begin, S. Masse, M. Rola-Pleszczynski, M. Geoffroy, M. Martel, Y. Desmarais, and P. SebjKSTIEN. The Lung Bi ological Activity of American^Attapulgite............................ .... . 328 E. YOKOYAMA, Z. NAMBU, I. UCHIYAMA, AND H. KYONp. An Emphysema Model in Rats Treated Intratracheally with Elastasc 340 Ialcolm Maclure. Asbestos and Renal Adenocarcinoma: A Case-Control Study...................................................................... 353 K. C. Hemavathy and N. B. Krishnamurthy. Evaluation of Lannate 20, a Carbamate Pesticide in the Germ Cells of Male Mice.................................................................................................. `362 Attilio Arillo and Francesca Tosetti. Denitrosation of Af-Nitrosodimethylamine and JV-Nitrosomethylurea by Liver Microsomes from Trout (Salma gairdneri Rich.)........................... .... . 366 Gary H. Heinz. Mercury Accumulation in Mallards Fed Methylmercury with or without Added DDE. ........................................ 372 Duane L. Peavy and Edward J. Fairchild II. Induction of Metallothionein Synthesis in Human Peripheral Blood Leukocytes 377 Mark Germine. Sepiolite Asbestos from Franklin, New Jersey: A Case Study in Medical Geology...................................................... 386 Katsumi Yoshida, Naoko Sugihira, Michiko Suzuki, Toshiro Sakurada, Shintaro Saito, Kaoru Yoshinaga, and Hiro shi Saito. Effect of Cadmium on T4 Outer Ring Monodeiodina- tion by Rat Liver. . . ........................................................... . 400 Jay Butterman, Phillip C. Chan, and Leo Kesner. Generation of Hemolytic Activity in Ozone-Treated Phosphatidylcholine. 406 Continued on inside back cover HWBUI0001973 and ""KorcHT isariwr / VV .itaxiaPP ' ivjuu^n,' * *' Yasuyuki Shibuya. and Kazuhide Dot. Uril He Acid Expo- ENVIRONMENTAL RESEARCH 42, 333-361 (.1987) jjftd of Mercury ice. . .... ji Enhancement' : i by Prolonged Asbestos and Renal Adenocarcinoma: A Case-Control Study Malcolm Maclure Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, Massachusetts 02115 Received February i2. J9S6 A case-control study of renal adenocarcinoma has corroborated the hypothesis that as bestos is a cause of the disease. The odds of having been moderately or heavily exposed to asbestos 30 years before diagnosis were 45:473 among cases, Whereas the comparable odds among controls were 26:492. A matched-pair analysis yielded an exposure odds ratio of 1.8 with 95% confidence limits of 1.1 and 3.1. After controlling for potential confounders and selection factors by means of logistic regression, the incidence rate ratio was estimated to be 1.6, with a one-sided 95% confidence limit of 1.0. e 087 Academic Press. Inc. INTRODUCTION The hypothesis that asbestos may be a cause of renal adenocarcinoma was first corroborated when Selikoff and colleagues (1979) reported that their continuing follow-up of 17,800 insulation workers had revealed 18 kidney cancer deaths when only 8.1 were to be expected. As always, the result could have been due to chance. Alternatively an unrecognized source of incomparabOity between the insulatiofi workers and the standard US white male population may have intro duced confounding. Coupled with evidence (Langer, 1974) thatl&sbestos fibers can spread throughout the body, however, the rate ratio of 2.2 suggested cau sality. We were in a position to test this hypothesis further in a case-control study of renal adenocarcinoma conducted in eastern Massachusetts where a sub stantial number of people were at one time employed in shipyards. METHODS During the period 1981-84, 1190 cases of renal adenocarcinoma were identified from the records of 37 participating hospitals in an area around Boston com prising some 100 townships. Patients who were diagnosed before 1976 or before age 30, or who did not reside in the specified geographic area, were formally, ineligible. Of those formally eligible, 30% were deceased, and 24% were not in terviewed at their own, their physicians', or their families' requests, or because they had moved. Of the 549 interviewed cases, 31 were excluded from the fol lowing analyses because of mismatches with controls in age, because of poor quality interviews, or because they or their controls were nonwhite. The repre sentativeness of the interviewed cases was judged by comparing abstracts of their medical charts with comparable information from the charts of noninterviewed cases. Each of the 518 interviewed cases was matched with a control of the same age and sex systematically selected from the case's precinct of residence using the 353 0013-935L87 S3.00 Copyright S 1987 by Academic Press, Inc. All rights of reproduction in any form reserved. haled Ozone on 1 ;Muffat-Jgly. rpunodepression ijtion and Stress iTRiOU. Some nylactic Use of Pis Exposed to j :icity Limit and I.' and BHC in the j 1UNNAR SPANG . ' Cadmium Con >Unally Exposei Jscence Anal v si: ;] Tanaka. Th< .Cipase and Eattl h :f. M. Burbach 0 :i{ Methylmercur if and Carol F 5red Release ( H :it^D Christia Intratracheal I: >:i Reduction Pla <r `' J^ 1 ' J '* ' * * * { HWBUI0001974 354 MALCOLM MACLURE annually updated residence lists of Massachusetts towns. For any case under 50 years of age, a second control was selected and interviewed, so as to compensate for the small number of young cases. Whenever a control declined to participate (45% of the time), an cilternative match was selected by the same procedure. Information was obtained from 60% of the noninterviewed controls by means of a short mail questionnaire. Detailed analyses of potential sources of selection bias based on data from this questionnaire and from the cases' medical records are to be presented elsewhere (in preparation). The hour-long interview encompassed a wide range of topics in addition to occupational history, including past intake of foods and beverages, use of drugs and vitamins, smoking and exercise, disease history, and ethnicity. Detailed ques tioning on occupations was therefore ruled out. Furthermore, asbestos was only one among a number of hypothesized occupational risk factors investigated. Nev ertheless, occupational history was the firsts major section of the interview, and the word "ASBESTOS" in I cm capitalized letters topped the list of substances on the card handed to the interviewees when they were asked about specific exposures. Subjects were therefore at their most alert when asked about jobs and could not have missed the opportunity to mention exposure to asbestos. With concern for the possibility of biased reporting by cases, due to increasingly wide spread knowledge of the carcinogenicity of asbestos and other substances, openform questions about occupational history were asked first, and lists of industries and substances were only afterwards given to the subjects to supplement their accounts. In the course of coding, data entry, and verification of the interview data, evi dence was found of underreporting and overreport.ing of asbestos exposure. The open-form questions on occupational history revealed that many people who did not report asbestos exposure had had jobs in which thJf almost certainly were exposed: 43 shipyard workers, several boilermakers, and a steamfitter. On the other hand, among those who reported continual exposure to asbestos, there were five mechanics, five firefighters, three teachers, and a variety of others whose intensity of exposure was arguably much lower. Apparently self reports of exposure to asbestos could not be wholly relied upon. At best, they could be taken only as supplementary information on specific occupations. . A list of all subjects who might have been exposed to asbestos was therefore prepared. Anyone who mentioned exposure to the substance, who reported em ployment in the shipbuilding industry, or who was a pipefitter, steamfitter, or boilermaker was included in the list. Using all available information from the interview, each subject on the list was categorized as having "high," "interme diate," "low," or "negligible" exposure by two colleagues well versed in the idiosyncracies of asbestos distribution in the workplace. Their judgements were made independently without knowing whether a gubjeqt -was-a case or a control. The rate ratio was estimated by means of the exposure odds ratio (EOR) which, in the matched-pair analysis, was computed as the ratio of the number of discor dant pairs in which the case was exposed to the number in which the control was. exposed (Flsiss, 1981). In the stratified analyses, the Mantel-Haenszel estimator of the EOR was used (Mantel and Haenszel, 1959). In the multivariate analysis. the EOR was c regression (Bi matched with recent because presentation ( figures and tat sion. The full elsewhere (in ; There were posure, 24 rat who reported negligible in e sure rating sc hypothesis am sure are used, 1 and 2, allow alternative rat The figures manner which period of any years of expo cancer or, for diagnosed. Fq high by either subjects who ' Table 1 is a rated as interi stratified by t posed, count! sponding crud nosis appearei % 1), a 30-yet The operatl agreement by sure to asbest on this definit 09, and 34 in the control's e trols' exposur analyses were The discord dence limits o: from an analy: 8*1 Stef MmMm HWBUI0001975 e under 50 ompensate participate procedure. Imeans of a petion bias addition to se of drugs ailed ques ts was only :ated. Nevrview, and substances ut specific ut jobs and :stos. With ngly widetces. openindustries ment their data, eviosure. The e who did tainly were ter. On the stos, there of others reports of could be therefore ported emnfiiter, or from the "intermeed in the ents were } control. R) which, of discormtrol was estimator analysis. Continuedfrom outside back cover y kotCHl Kono, Yasuhisa Yoshjda Hrencut 'V'MAQATA, Ml DDE Dot. oric Acid ; ASBESTOS AND RENAL ADENOCARCJXNOMA 355 the EOR was computed as the antilog of the coefficient from a conditional logistic regression (Breslow and Day, 1980). Since only 6 of the 87 second controls matched with cases under age 50 had reported exposure to asbestos, all of it recent because they were young, these subjects added little but complexity to the presentation of the data. For clarity, therefore, they were excluded from all figures and tables shown below and all analyses except the overall logistic regres sion. The full analysis and interpretation of the logistic model is to be presented elsewhere (in preparation). RESULTS There were 79 subjects rated by both reviewers as intermediate or high in ex- posure, 24 rated by only one reviewer as intermediate or high, and 38 subjects who reported exposure to asbestos but were rated by both reviewers as low or I negligible in exposhre. In recognition of the inevitable subjectivity of any expo sure rating scheme, as well as the rarity of opportunities to test this etiologic hypothesis and the diverse ways in which epidemiologic data on asbestos expo.* sure are used, we have chosen to present the raw data on all 141 subjects in Figs. | 1 and 2, allowing readers to scrutinize them critically and to reanalyze them with j .7 alternative rating schemes. } ' The figures show the distribution of exposure years for cases and controls in a ) manner which permits visual assessment of the EOR while allowing an induction j period of any length to be assumed. Each horizontal line traces a singly subject's j years of exposure across the decades preceding his or%er diagnosis of renal | - cancer or, for a control, preceding the date when his or her matched case was diagnosed. Figure 1 shows the exposure years of subjects rated intermediate or ? high by either of the reviewers. Figure 2 shows the reported exposure years of : subjects who were rated by both reviewers as having had low exposure. \ Table 1 is a condensation of Fig. 1 showing the numbers of cases and controls > . rated as intermediate or high in exposure by both reviewers. The numbers are stratified by the decade when the subject was first moderately or heavily ex; posed, counting backwards from the date of the case's diagnosis. The correj sponding crude EORs are shown. Since exposure within three decades of diag; nosis appeared no more common among the cases than among the controls (EOR = 1), a 30-year induction period was assumed. ' ' 3'he operational definition of exposure adopted for all subsequent analyses was' agreement by both reviewers that the subject had bad intermediate or high expo sure to asbestos at least 30 years prior to the date of.the case's diagnosis. Based on this definition, there were three concordant pairs: matched-pair numbers 03, 09, and 34 in Fig. 1. (Matched-pair number 35 was counted as discordant since the control's exposure was recent.) There was little correlation of cases' and con. trols' exposures due to matching (correlation coefficient = 0.02), but matched analyses were done since the data in Fig. 1 lend themselves to such an approach. The discordant pair ratio was 42:23 meaning the EOR was 1.8 with 95% confi dence limits of 1.1 and 3.1 (Fleiss, 1981). Mantel-Haenszel estimates of the EOR from an analysis of the unmatched data, stratified by age, sex, and either years of )od of Me lice. . . s Enhanct ;s by Prot. raled Ozo Muffat- lunoaepn tion and . rraiou. ylactic l s Expos city Lim .nd BHC NNAR S -admiutr .ally Ex ence Ar ANAKA. ase and 4. Bur lethylrro d Car d Rele Chri atraeht due tier J? HWBUI0001976 356 MALCOLM MACLURE -60 Yrs unnuuu 22 CASES -40 yrs -Uni*01 ?lurber<40);velder!navy yd)<- *02 ?i-jnber;shipyd{?){n) 2222222222222222222222222222 22 2 22 222222222222 222222 22 22 222 222 22222 lumunum 22 *03 3o i I eranking/c leanine{ navy ydM-){h} *04 stea^fixterl-) fx) *05 K*i?*r<RAvy yd) ,Sh*sbdg{?)fmJ #06 KsUiny {shipyd)! ->;carpenter! 3/mo) (h) rG7 5fc4pfitier{40)|nj *#0359 CTiorp^pteesrspmeirtfhshaioppyr-eiHnti-c}(e.fn:}avy yd)<?){h] *10 Propeller shaft assemblyshipyd)(-)U] *11 Supervis picefi11ars,ece(shipyd)(-}}xj *12 Ke;dlng{shpyd){-);fcoller.op.l40)lmJ *13 Bojlarfatkiag (raiiroad)f-){nJ *14 Asst shl?fitter<-){hl 222222222222222ilumtUllllHU 2222 *15 NscA-.-j.jt- sh-pv i;{ - 'tr} *16 Shipwo>r5tar{-){ai *17 Soiiermakingfnavy)(0)JnJ *13 wel<5er(shlpya){40)(nl 2222 *19 -S Savy sbipbuildi.*jg!-){xJ 22 *20 Navy yard {-HnJ 22 V *21 Snipfittert-lfx) 222 *22 Snip!itter helper!r.a/y yd)(-)lml 22 22 Shipfittar{-);c&rpencer{l){m) maimmnuuiiiimmiiiniui .11 *24 Construction worker!40)[hi lunuuniuuumiinummui 12 -25 2ar?entar(40>fh) 22 *26 Carpantry!boatyd)(-)Ixl 222222222 >ipydj(15)Jx) 222222 ii.?yd)(-}(x) *29 Shipbuilding! *30 Asst.snip.sapt{r.avry yd)(-){xl 222 *32 Shipbuilding!engine room){-)[x] 22222222222222222222222222222222222222 *32 W**-. r(4Q)[m) 222 *33 Engineer!shi?yd)|-)Ih] 222 *34 Sftiomi!iding(-Hx) 2 222 *35 Stage bldr{engine rccm,shipyd)!49)(n) 2222 36 Welder(shipyd)(40)(n) 22222 #27 Sfiipyd! insulation.etc. )(30){x) 22322 222 *33 Ssvy yd:coppersmith,w%lder{-)(xj *39 Electrician,welding!sr.pbg)! -) (m) 22 22 *40 Insulator! 40) ;eiectr(shi.pY3) (40) (h) 222222221111111211113111111 *4101 erk, supply {shlpyd 1U1 * 42 Electr assetrblyl?)[n) 2222 *43 Navy ship repair!-)(n) UmiLlllllllllLlllUlUUUliili *44 Construction<40){nl 22 *45 3oilerr?aklng (r.avy v;d)(-)lh| 4) immitmmuuuuiuuuui *46 % service, heatir.9^sine5S(40)tn) 222222222222222222222222222 #47 SW%f itter;supers. (shipyd)( 3)(m) 22 *43 Sorsner laborer.cleanup!shipyd)(40){x} ^ 2222222222222 #49 -wilerrnaker!fire ermine)!-)[h] -- 111111111111111 '*=0 Construction work*r(40)(n} mnmmiuniniinunili: *51 Stationary engineer!4C){hl UninUinUUlllimr.ll!'-u *52 Tr:i< r.echar>ic(bra<*s)(?Hxl 222 *53 5oi leraaker apprer.tite{ ?) IxJ 22222222222222 2 -54 Plastering! 2B){)0 *53 s<33tyari(20) {.-!) lliUlllllllillU *36 Carpenter!:5da,'roj f n I imnuinmumu *57 SreltingUaundryj/SOM*) lllil *55 Construction suot.(4)|rij 22 *59 Cr.-pping/cauikin.g( shipyd){ 20){m) nmilUUliU -SO Navy yard hei=*r!43>fnf 22222222222222 *61 e.der(snobg)(2=1] n *62 Gen .work. ba lir.g( shipya) {-) (h f * Fig. t. Chronologic distribution of cases' and controls' years of moderate or heavy exposure to asbestos, dated in years prior to cases' dates of diagnosis. Key: 22222, rated heavy to moderate by both reviewers; 11111, rated heavy to moderate by only one reviewer; 00000, rated as light or negli gible by both reviewers; (25), subject claimed exposure 25 hr/week; (40), exposed 40 hr/week or "all the time"; (?), reported exposure but did not quantify; (-), subject did not report asbestos exposure; [n], never smoked; [x], quit smoking 10 years before diagnosis; [m], smoked up to 1 pack/day within 10 years of diagnosis; [h], smoked more than 1 pack/day within 10 years of diagnosis. schooling, or socioeconomic rank of the most recent occupation according to the Standard Occupation Classification Manual (US Dept. Commerce, 1980), were no different. From a conditional logistic regression, controlling for known renai cancer risk factors (history of smoking, obesity, and kidney stones) as well as ail -70 yrs -6Q yrs 222222222222222 222 control: other hypot phenacetin, the FOR wa The hypol adenocarcim tory as show the same as subject is int | to "never sr smoker," ar posed cases never smoki recent mode both cases f smoking catratio. Indeec to the low oi to an unusu; around the < orous testing HWBUI0001977 {navy ya>{-){n) ) LngCnavy yd)(-)(h| ipbdg{?){fsl carpentertS/ffloJJh] r-)[h|;ice(jtavy yc}(?){h| 3errfclv{sbipycn-j[nJ rSietc(thipYCH->lxl >ailr.Qp.<40)[iBj road)<-)(nl *1 aY Yd)<-){c:3 p.t.-er(i){ir) ?40){h] )fxj ys>(-)ix) ;UrOooJ.lrTr.J){'[xJ h) I ){30)fx} s(sr.hnbip9y){d-)<> t<-t"f;{hj i-H* 1 1 3 3cjHSi-n)e[hs)sUCHn] Mbyc)l?)(rr! :p(Shipydj(){xi LiOr.elh)(l-Hbi l;Oi'H7h}J{xj ref)f jcJ i^i in] Pi"r)(20)[r; r.] d><-){h] avv exposure to to moderate by as light or negli- ) hr/week or "all ,'eslos exposure: pack'day within 5. ording lo the 1980). were known renal s) as well as teiBrisp T4^ Continuedfrom outside back cover yk o' . ichi Kono, Yasuhisa Yoshida, Hiroshi Yamagat, / V.------------------- r,................ ,, __ . :de ric j asbestos and renal adenocarcinoma 357 -70 yrs -60 yrs -50 yrs yrs -20 yrs yrs Diagnosis Date 2222222222222222222222222222222222222222222222222222222 #63 Plimiber(8) [nj 222 i:nii22222uiinuniriiinn222222222222222222222222222222222222222222222222222 #64 Shipbuilding{-)[x] #65 Ironworker(?)(xj #66 Plumberfnavy yd 40-45){?)(m] 222222222222222222222222222222222222222222222 #67 Shipfitter,welder{shipyd)(?){x} 22222222222222222222222222222222222222 #68 KelderlshipydX-Hnj#68 Supervis pipefitter(shipyd){40;60}Jx] 22222222 #09 Serviceman:heating.air eo<3(22){h| 22212Z2122222222222222222222Z22222222222 22222222222222222222222221illIllil111 #70 Plumber(some shpbg)(10)(m) #7l Machinist^ navy yd,maint dept)(-)[x| #72 Kavy yerd(?){nj 22222 #73 Welder(shipyd)(2)(x] #74 Machinist(shipyd)(lO){nl #03 Coppersndth(shipyd)(-)(xJ CONTROLS 222 #75 Plumber helper(shipyd){-)fxj 2222221: mum in in mini: um mil #76 Clerk,planner(shipyd)<-)(x3 22 #77 Painter(shipyd){40)[m] 22222222222222222222222222 #78 Pipefitter(Navy shipyd)(15)txj iiiiiuimimmiiiiiiimimuiii #79 Construction worker{40)fxj 222 #60 Boilennaker(shipyd)(-)(n>I niinniinmniiiiiiinmnuim #46 Carpenter140)(n] 222 2222222 #34 Pipefitter{shipYd)(4)[tn] 222222222222222 #81 Boiler maintenance(10)(xj 2 #82 Pairitshop disp(sbipyd)(-)lml 2 #83 Secretary(shipyd) 222222222222222222222222222222222 #84 Plunth,carp,ntairttenance(4Q) [xj 22222222222222222222225222222222 #85 plumbing(40)[m) 2 #86 US Havy(shipyd)t-)th) 2221111111 #87 Shipbuilding engineer(-)[hj 222i:ilUlHil!ill #88 Telephone instal{shipyd,cellars)(-)|x) mm nnnu illinium #33 Construe superv(20)(hj minimum innnmi #89 Linoleum instal(sanded vinyl esb)(2)[h) #90 Patching(20/wk in suirmer){h) 222 91 Bailer repair(shipyd)(-)(mj luiiimiinmi #92 Hachinist{4Q)Jhl #93 Sales(total 1 yr in shipyd){-)tx] 94 Ship machinery installation-) #95 Heavy equip mechanic*;20l(h) #35 Piped tier(5hipyd){,4Q)[n} #96 jmnfiring(for gririS?whee).)( l)fxl #97 destruction IDlxJ Fig. I--Continued. & other hypothesized risk factors and potential selection biases (hypertension, phenacetin, dietary protein, years of schooling, income, and status of recent job), the EOR was estimated to be 1.6 with a one-sided 95% confidence limit of 1.0. The hypothesis that smoking modifies the association of asbestos with renal adenocarcinoma was tested by stratifying the study population by smoking his tory as shown in Table 2. The definitions of smoking subcategories in Table 2 are the same as the definitions used in Figs. 1 and 2. (The smoking status of each subject is indicated in the figures by the letter n, x, m, or h in brackets, referring to "never smoked," ``exsmoker 10 years ^before diagnosis," "moderate or light smoker," and "heavy smoker," respectively.) Table 2 shows that asbestos-ex posed cases outnumbered asbestos-exposed controls mainly in two categories-- never smoked and recent heavy smokers. By contrast, among exsmokers and recent moderate smokers, the odds of asbestos exposure were very similar for both cases and controls. The lack of a smooth trend in the odds ratios across smoking categories is readily explicable in terms of the instability of each odds ratio. Indeed the large odds ratio among heavy smokers appears to be mainly due to the low odds of asbestos exposure among heavy smoking controls, rather than to an unusually high exposure odds in the cases. The wide confidence intervals around the odds ratios, even after grouping all smokers together, preclude rig orous testing of the hypothesis that smoking modifies an asbestos effect. ad c ice. Enl by aied <1uf: unoc ion i TRIO riact ; Ex :itv 1 ndB. MNA1 adrn ally ence vnai< ase a I. Bi ethyl d C, d Re Chi atrac Juctu f i HWBUI0001978 358 CASES MALCOLM MACLURE -20 yrs OOOCQOOQoOOOOOOOOOOOOOOQOOOOaCC-OOOOOOOOOOOOCOOCCOC OOOCOOOOOOOOOOOOOOQOOOOOCCOOOCOOOOOOOOCOOOCOO OOOOOOOQOOtiOCOQQOOOOOOOOOQOOCOOOOOCOOO aOQOOOOOOOQOOOOOOCOOaCQOOOOCOOCCOOOCOCOCOU OOOOOOOOOOOOOQOOOOOOOOOOOOOOOOO oooo DOOOOOOQOOCDOOQOOOQOQCOOOOOOCCOCO oooooooooo oooo ooccoooo GOGOOOOO 00 cooc: coco Measured asbes.rolls(l)[nj Autonechanict40)[h} Autowechan ic (4) [ r. 1 Autcimeehar.icfo) f x) Stockroom {uaiv)(?){bl Flyer(asb. ir. coc.<?ic)(2Q)txl Navy( 40)[x] Firef igfiter( 40) Ih J Govsrn*iSS( room) (30) (x} Savy(?){h} Teach<fr(30)tn} Construe.lsrororer(7) (x] Bookkeeper(exp.once)[n} Hason(8 )[h| Ayent(vlsiced srspyis)(?) [m] Editorf -eis.painting)(10)[m] Ad->in. (bldg, rehab)(4)f m] -70 yrs -50 yrs -40 yrs -30 yrs -20 yrs -10 yrs oooooooooooooooGOoooooooooocooooocoooioocooco CONTROLS OOOOOOOOOOOOOOOOOQOOOOOOOOOOOOOOOOaOCJCOO ooooooo ooQooooooooaooooocoocaooco ooooooo ooooooaoaoatioaaaaooao OQOOOOOOOOOOOQOOOOQCOOOOOOQCCODCCOOOCC-' OOOOOOO COO' OOQOODOOOOOCGOOOOOOOOOGijaoOCOGCCOCC; OOOOOCOOOOOOOOOOOOQOQCOCOCCOOO OOQOO oooooo oooooooaoooccocoocccoc.:^ ooo ooocoooccoocaooocoocccccc OOOQOCQCOCOCCCOCOCOG GOCOCCCOOOOGOQO oocoocooooooccccoo OOOOODOOOOCQCCCOOOC COQGCOOOQ 115 Automechanic/1)Jx| 118 Fireighter(4Ci[nl 11? C4rp**nter.pl,.^riher{3/yr)j8i) weaver (c)|x) :;3 Cierk(auto ir.aus)(3)(nj 119 Security gjarc(SteelCo;(40)[hJ 120 Auton;echanic(20)[hJ 121 Mari.ne{ 24/da} ;schi.nluuib{6)ltn} .22 AccountantCHlec.Co){40)(h) 123 Firefighteri40)lnl 124 Fhysicist{F!ooringCo){10Hxi 125 Carpenter(?)[x) 126 Truckdriverft)(} 127 Highscheol Teacher^ ?){xj 123 sbsetaeraivorkertIwk/yr ){riJ '.I? lhd.Arts.Tfcach(ciothg)(2Q)(h] 120 Welder(i)[ni 131 Firefiebteri40)[nj 132 SiUscreer. printer{40)[xj 10"* Assembler* Dim] 161 7ruckariver(4Ci{nJ Fig. 2. Chronologic distribution of reported years of asbestos exposure for subjects judged as X* having had light or negligible exposure, dated in years prior to cases' dates of diajgpsis. See Fig. 1 for key. '% DISCUSSION There are four alternatives to an etiologic explanation for the observed associa tion between asbestos and renal adenocarcinoma: random sampling variation, se lection bias, information bias, and confounding. By the conventional criterion of the 95% confidence interval excluding the null, sampling variation can be virtually ruled out as an explanation. In doing so, no account is taken of probable but unquantifiable nondifferential random error in classifying subjects with regard to exposure. This would only cause underestima- TABLE I Distribution of Exposed Subjects by Decade When They' First Became Moderately or Heavily Exposed, according to Both Reviewers Time of first exposure (in decades before cme's diagnosis) 6-8 5 4 1-3 Total Cases 2 ii 32 4. 49 Controls 3 9 14 4 30 Crude exposure odds ratio 0.7 1.2 2.4 1.0 Odds of As Never smoked E ver smoked !. ..smokers (qu Recent modem Recent heavy'' " One pack y '' More than lion of both misclassifict planation. f overly restri number of s more inclus exposed by of 2.0 (95% estimate of The possi abstracted f data from t viewed co: might b-;- stain, by the .. (her cento baum, Kupi the EOR es selection bi; If the sub tional defini for positive causes of c; however, th obscured bj ential miscl; reported as! or negligibli often as ca: only as a c: data was to JSp.l i! HWBUI0001979 F0.)rlchljisdHn} })(jc?] )(*} octeit)(20)x} I'nJ (?Hx) or.ce) {n ] ;ting)(lC)[m] aiH4){nt} xj '<3/yr>[m] <2)|nJ r~]"-eelCo) {40) (h} r.i.plunb(6)[jp] nl r.cco)(10){x] *"< ?Hx) >:-tn*g/yHr)2[csnt]nJ ibjects judged as sis. See Fig. I for rved associa'variation, se eling the null, doing so, no idem error in jnderestima- df.rately or Total 49 30 T M. .'j Continued from outside back cover /Roichi Kono, Yasuhisa Yoshida, Hiroshi Yamagata. / . m.-s . v. . - x/-,------------------------^------------------------ DE DO ic Act 'fSKfs; ASBESTOS AND RENAL ADENOCARCINOMA 359 im TABLE 2 Odds of Asbestos Exposure 30 Years before Case's Diagnosis, after Stratifying by Smoking History Asbestos exposure odds (exposed monexposed) Cases Controls Crude odds ratio 95% Confidence interval Never smoked Ever smoked Exsmokers (quit 10 years+ ) Recent moderate" smokers Recent heavy* smokers 11:174 34:298 14:114 8:94 12:90 4:187 22:305 12:114 8:103 2:88 2.9 1.6 1.2 1.1 5.9 (1.0, 8.9) (0.9, 2.6) (0.5, 2.6) (0.4,2.9) (1.4, 24.) " One pack per day or less, within 10 years of case's diagnosis. h More than one pack per day. within 10 years of case's diagnosis. tion of both the point and interval estimates. Any adjustment for such random misclassification would only make sampling variation seem less likely as an ex planation. Moreover, one might consider that the definition of exposure was overly restrictive and increased the problem of sampling variation by reducing the number of subjects classified as exposed. Repeating the above analyses usiijg the more inclusive criterion---that a subject must be rated as moderately or heavily exposed by at least one reviewer, rather than by bolh-^3ne obtains an odds ratio of 2.0 (95% confidence limits: 1.2 and 3.4) from the matched-pair analysis and an estimate of 1.8 (95% confidence limits: 1.1 and 3.1) from the logistic regression. The possibility of selection bias was studied by multivariate analyses of data abstracted from medical records of interviewed and noninlerviewed cases, and of data from controls' interviews and mail questionnaires returned by noninterviewed controls. These analyses showed that the main selection factors which might have biased a crude analysis were education, current socioeconomic status, cardiovascular disease, and age. These factors were partially controlled by the matched-pair design and analysis, and any residual selection bias was fur ther controlled by stratified analyses and conditional logistic regression (Kleinbaum, Kupper and Morgenstern, 1982). Th lack of substantial differences among the EOR estimates from crude, stratified and multivariate analyses indicates that selection bias, at least by these factors, is an unlikely explanation. If the subjects' own reports of asbestos cxpo^rc had-been taken as the opera tional definition of exposure in the analyses, there would have been opportunity for positive information bias due to some of the cases having read more about the causes of cancer and thought more about their past exposures. The data show, however, that if this happened in a few instances, it would have been more than obscured by the negative information bias due to the above-mentioned nondiffer ential misclassification. Figure 2 lists all the subjects who, it might be said, over reported asbestos exposure insofar as both reviewers rated their exposure as low or negligible in intensity. It can be seen that controls overreported exposure as often as cases. To reduce such misclassification, self-reported exposure served only as a criterion for inclusion on the shortlist of subjects whose occupational data was to be screened by the reviewers. As it turned out, the major contribution d of I ce. . Enhar by Pr iled O lUFFA inodej on am RIOU. lactic Exp< ity Li id BH INAR admiu illy E ;nce / iNAICr ise an I. Bu: ethyir 3 Ca i Rei< Chk a tract: iuctio I HWBUI0001980 360 MALCOLM MACLURE to the operational definition of exposure was employment in shipyards and people so employed more often than not denied ever being exposed to asbestos. There fore information bias is a very unlikely explanation for the association. Such a conclusion is consistent with the results of a psychological study of mesothelioma patients by Lebovits and colleagues (1983) who found most patients believed their asbestos exposures had not contributed to their getting the disease. The hypothesis that the association was wholly or partially due to confounding by other renal cancer risk factors was tested by multivariate analyses using a wide variety of combinations and transformations of variables encompassing smoking history, nutrient intake, ethnicity, drug use, and disease history. The asbestos exposure odds ratio estimates from these models remained in the range 1.5 to 2. This does not eliminate the possibility that unknown risk factors, or suspected risk factors measured imperfectly and therefore only partially con trolled, might account for the association! For example, exposure to lead and other metals also occurs in many of the jobs listed in Fig. 1. Separate analyses of the data on lead and metal exposure, however, tended to refute the hypotheses that they are risk factors. On balance, confounding appears an insufficient expla nation for the association with asbestos. Insofar as the evidence disfavors the four alternative explanations, it corrobo rates the hypothesis that asbestos is a cause of renal adenocarcinoma. The kidney was the first site outside the digestive and respiratory tracts in which a significant excess of fatal cancers occurred among Selikoff's insulation workers, so there was reason to be skeptical of a causal interpretation. Such an interpretation would have been especially untenable if asbestos fibers could not be shown ca pable of reaching the kidney. Laboratory experiments with mice and rat&* how ever, demonstrate that asbestos fibers can be widely disseminat# by means of the lymphatic pathways and the circulation (KanazavH et al., 1970; Sebastien et al., 1980). The'same process is inferred to occur in humans since autopsies of asbestos workers have revealed asbestos bodies in various organs remote from* the.respiratory and digestive systems (Langer, 1974). Among the remote organs, the kidney was the one in which Langer observed the highest concentrations of asbestos bodies. Moreover, asbestos fibers have been found in human urine (Cook and Olson, 1979). It seems likely, therefore, that many of the people in the present study, who reported working in shipyards 30 years ago, harbor asbestos fibers in their kidneys today. Whether it is only because these people were ex posed to much higher dust levels back then, or whether asbestos fibers need 20 years or more to manifest their carcinogenicity in the renal parenchyma, the timing of their exposures appears to hav#-been important. The lack of an associa tion with recent asbestos exposure, which was the basis for the working assump tion of a 30-year induction period in the analysis, adds support to a causal inter pretation by further disfavoring selection bias and information bias as explana tions; biasing factors would tend to be associated with recent exposures more than with exposures in the distant past. A further test of the causal hypothesis was to examine the data for evidence of modification by smoking. However, when the study population was subdivided into four strata by smoking history, the odds ratios hypothesis. The analyses defined in this ; adenocarcinom; carcinoma whic diagnoses, to a !940s and early elusion lies rigt used in toxic tc was "'as likely I am very gratel Godfredsen, Mich; Parted by a grant fellowship from th Urcslow. N. E., ai Publ. 32. Cook, P. M-, and 204, 195--198. Heiss, I. L. (1981 Harter, P. J. (198* Issues." Mor Washington, Kanazawa, K.., B from subcuta Kleinbaum, D. G and Qurinlita Langer, A. M. (I ease. Enviroi Lebovits, A. H., . responses of Mantel, N,, and studies of di' Sebastien, P.,.M: i, intestinal lyr "Selikoff, 1. M,, t in the Unita U.S. Dept. Com Publ. No. SI ? Wmm HWBUI0001981 s and people stos. Thercion. Such ... esothcliomu dieved their .onfoumJinj; ses using a compassing listory. The in the ranee factors, or irtially cunto lead and analyses of hypotheses cient expla it corrohoThe kidney i significant `s, so there erp relation shove n ca rats. hov-y means of ebastien i ' utopsies o! mote from ote organs, itrations of man urine ople in the Hr asbestos c were exirs need 20 jbynta, the in associug assumpusal inter5 explanaures more lypothesis However, ig history. / Continued from outside back cover / //Koichi Kono, Yasuhisa Yoshida, Hiroshi Yamagata, Mist. U/aTAKUOC VAPtrvm/T Crrrnmi *"" v ... ......''E Dot. I c Acid Ex t i ASBESTOS AND RENAL ADENOCARCINOMA 361 1 of Metr :e. . . . >he 0dds ratios were too imprecise to provide further evidence for or against the. hypothesis. The analyses presented here suggest that asbestos exposure, as operationally defined in this study, causes a 50 to 100% increase in the incidence rate of renal adenocarcinoma. The data suggest that as many as half the cases of renal adeno carcinoma which occur among people who were exposed, 30 years before their diagnoses, to asbestos levels comparable to those in US shipyards during the 1940s and early 1950s may be attributable to their asbestos exposures. This con tusion lies right on the borderline of the "preponderance of evidence criterion" used in toxic tort law, namely, that evidence should indicate a claimant's disease was "as likely as not" due to the exposure in question (Harter, 1984). Enhancem by Prolon; led Ozone UFFAT-Jo nodepress m and Sir riou. Sc ACKNOWLEDGMENTS - I am very grateful for the advice of Charles Poole and wish to thank him, Ronald Bruno, Morag (iodfredsen, Michael Ellenbecker, and-Rebecca Kolodny for their contributions. This work was sup- ported by a grant from the National Cancer Institute (5 R01 CA29723) and partially supported by a fellowship from the DuPont Companv. Wilmington, Delaware. REFERENCES Bresknv. N. E., and Day, N. E. (1980). "Statistical Methods in Cancer Research." Vol I., IARC Sci. - Pub!. 32. ' * Cook, P. M., and Olson, G. F. (1979). Ingested mineral fibers: Elimination in human unite. Science 204, 195-198. * . Fleiss, I. L. (1981). "Statistical Methods for Rates and Proportions." 2nd ed. Wiley, New York. Harter, P. J. (1984). The dilemma of causation in toxic torts. In "Selected Studies in Health Policy Issues." Monograph 101. Institute for Health Policy Analysis. Georgetown Univ. Med. Center, Washington, D.C. Kanazawa, K., Birbeck, M. S., Carter, R. L., and Roe, F. I. C. (1970). Migration of asbestos fibers from subcutaneous injection sites in mice. Brit. J. Cancer 24, 96-106. Kieinbaum, D. G., Kupper, L. L., ana Margenstem, H. (1982). "Epidemiologic Research: Principles and Quantitative Methods." Wadsworth, Belmont, Calif. Langer, A. M. (1974). Inorganic particles in human tissues and their association with neoplastic dis- . ease. Environ. Health Perspect. 9, 229-233. Lebovits, A. H., Chahinian, A. P., and Holland, J. C. (1983). Exposure to asbestos: Psychological responses of mesothelioma patients. Amer. J. Ind. Med. 4, 459-466. Mantel, Nr.. and Haenszel, W. (1959). Statistical aspects <jf the analysis of data from retrospective studies of disease. J. Natl. Cancer Inst. 22, 719-748. Sebastien, P., Masse. R.. and Rignon. J. 11980). Recovery of ingested asbestos fibers from the gastro intestinal lymph in rats. Environ. Res. 22, 201-216. Selikoff, 1. M., Hammond, E. C\, and Seidman. II. (19/9). Mortality experience of insulation workers in the United States and Canada. 1943-1976. Ann. .V. Y. Acad. Sci. 330, 91-116. U.S. Dept. Commerce (1980). "Standard Occupation Classification Manual." U.S. Govt. Print. Off. Pnbi. No. 80-332-946, Washington. D.C. lactic Use Exposed . . ., . Limit . d BHC in NAR SpA; idm.ium C .lly Expo nee Anal; NAKA. tse and Ft . Burba Mhylmerc 5 Carol 1 Releasi Christ: atracheal iuction P i . ..' HWBUI0001982