Document Ne0R3rgB15NN55BzyRLv1vaYb

PLAlNTIFF'S > EXHIBIT T I I E NEW I:NGJ,ANI>JOUKNAI. 01: MEDICINE April 23, 1911 VIII. BENZENE AND LEUKEMIA An Epidemiologic Risk Assessment ROBERT A. RINSKYM, .S., ALEXANDER B. SMITH, M.D., M.S., RICHARD HORNUNGD,R.P.H., THOMAS G . FILI-OONB,.S., RONALD J. YOUNG, M.S., ANDREA H. OKUN, M.S., AND PHILIP J. LANDRICAN, M.D., M.Sc. Abstract To assess quantitatively the association be- tween benzene exposure and leukemia, we examined the mortality rate of a cohort with occupational exposure to benzene. Cumulative exposure for each cohort member was estimated from historical air-sampling data and, when no sampling data existed, from interpolation on the basis of existing data. The overall standardized mortality ratio (a measureof relative risk multipliedby 100) for leukemia was 337 (95 percent confidence interval, 154 to 641), and that for multiple myeloma was 409 (95 percent confidence interval, 110 to 1047). With stratification according to levels of cumulative exposure, the standardized mortality ratios for leukemia increased from 109 to 322, 1186, and 6637 with increases in cumulative ben- zene exposure from less than 40 parts per million-years (ppm-years), to 40 to 199, 200 to 399, and 400 or more respectively. A cumulative benzene exposure of 400 ppm years is equivalent to a mean annual exposure of 10 pprr over a 40-year working lifetime: 10 ppm is the currentlj enforceable standard in the United States for occupa tional exposure to benzene. To examine the shape of the exposureresponse relation, we performed a conditiona, logistic-regression analysis, in which 10 controls were matched to each cohort member with leukemia. From this model, it can be calculated that protection from benzene induced leukemia would increase exponentially with any reduction in the permissible exposure limit. (N Engl J Med 1987; 316:1044-50.) Nketiologic association between benzene and leuemia was suggested by a series of case repbrts beginning more than 50 years ago."6 Those clinical observations were corroborated subsequently by epi- demiologic and, more recently, by car- cinogenesis b i o a s ~ a y s . ' ~B` 'e~nzene is now generally considered by national and international scientific bodies to be a human carcinogen.I6-I8 To reduce the risk of leukemia in industrial workers exposed to airborne benzene, the U.S.Occupational - Safety and Health Administration (OSHA) promul- '=at,--=I gated in 1978 an occupational-exposure standard that reduced the permissible workplace concentrations of F benzene I O - f ~ l d , 'f~rom the previously acceptable s ca new eight-hour time-weighted average of 1 ppm. This . -+c action was based on the qualitative demonstration of U0J m the carcinogenicity of .;r\idemiologic studies." benzene in case reports and epi- ! @ 3 & ly with the kisk of dcath from leukemia. All thiee anal- I-& : .% t i ` ' From the National Institute for Occupational Safety and Health. Centers for Disease Control. Division of Surveillance. Hazard Evaluations, and Field Stud1a ieeASCFinnali, OH 45226, where reprint requests should be addressed to Mr. yses, however, were based on estimates of group expo. sure rather than on estimates of the exposure ol individual workers. The resultant risk estimates werc therefore subject to wide variances. To reduce the uncertainties of those assessments, we have reexamined the mortality experience of a co. hort of rubber workers with previously documented exposure to benzene. This cohort is the largest to br studied in a risk assessment, and i t olTers the most extensive historical record of exposures to airbornr benzene. Since our previous evaluation of the mortal, ity rate of this cohort," the life-table analysis system of the National Institute for Occupational Safety and Health ( N I O S H ) has been modified to allow incor. poration of data on individual exposures24; previous. ly, only the duration of employment could be used asa surrogate for exposure. Also, an additional 6.5 years d observation had elapsed since the previous evaluation. allowing us to update this analysis to 1982. We repon, here on the relation in this cohort between occupation. a1 exposure to benzene and death from leukemia. BACKGROUND This study was based on the experience of worken at three plants at two locations in Ohio that manufar. tured a natural rubber film (rubber hydrochloride). T h e details of the process have been described prr viously.'o In brief, -natural rubber was dissolved benzene and spread on a conveyor. The benzene then evaporated and recovered, and the resultant film was stripped from the conveyor, rolled, milled according .I to sp. ecifications. Rubber ride was manuhctured at April 1976. Production at Location 2 was in two separate plants. At the first, it search and development project; tion then began in 1936 or 1937 1949, when the second plant operation continued until thrr droc amo only coulc icity. withi ly dtr areas rubbc Ind pherir availa the O North These cause r made r to iden in the d iurcs (5 in unu: `etrospr ndicate ncrease ked dc evision5 our tin enzene 1 any g rocesses love thc irsions I `esented tlmatlon letailedj c 1y person1 1tiI-y the ( rochloridt wide j o b I Ird) was a `3 were tht )er hydroc :odes in t t b titles), c osurc clas ring area: h industrl .. --- - `. - --. _--_e_ I , 19117 Virl. 316 No. 17 HI':NXNE ANI) LEUKICMIA- HINSKY m AI.. 1045 -years more, 1 PPm1 PPm -rently :cupaof the itional were m this zene- h any J Med expore of were nents, a coented to be most borne ortalystem y and incordiousd as a ars ol ation, .eporl ition a. xker: iufac ride) L pre ed ir e wa, t thii an( chlo unti d ou a re 3duc unti Thi i t a1 thee plants were essentially identical. Although hydrochloric acid, soda ash, natural rubber, and small mounts of plasticizers were used, benzene was the only chemical in the rubber hydrochloride plants that could reasonably be associated with hematologic toxicity. T h e rubber hydrochloride plants were located \vi thin larger industrial facilities. Employees were likely during their working careers to have worked in areas of these facilities where materials other than rubber hydrochloride were produced. Industrial-hygiene records describing past atmospheric concentrations of benzene at the plants were available from the Industrial Commission of Ohio, the Ohio Department of Health, the University of North Carolina, N I O S H , and company records. These records have been described previously. Lo,2o Because most of the measurements in these records were made not to support an epidemiologic study but rather io identify problems with compliance, there are gaps in the data. These gaps were filled by estimating exposures (see Methods). Nonetheless, the data represent an unusually complete record of past exposures for a retrospective cohort study covering this period. They indicate that as knowledge of the toxicity of benzene increased and recommended exposure levels were revised downward, the company kept pace with those revisions. Thus, for the most part, employees' eighthour time-weighted average exposures to airborne benzene were within the limits of the standard in effect at any given time. As is characteristic of industrial processes, however, there were occasional excursions above these limits. A detailed description of those excursions in relation to individual exposures has been presented previously. METHODS Estlmatlon of Past Exposures Detailed job histories for each employee were obtained from company personnel records. Each employee's record was reviewed to identify the department symbol that indicated work in a rubber hydrochloride plant a t either location. Each unique rubber hydrochloride j o b title (described in a short narrative on the personnel rccord) was assigned a numerical code. Job codes and employment dates were then abstracted for each employee who had worked in a rubber hydrochloride department. Because of the large number of pb codes in the record system (resulting from numerous variations injob titles), codes were fitted to broader categories, referred to as `exposure classes," which could be associated with specific manufacturing areas. In general, exposure classes represented areas in ahich industrial-hygiene data had been collected. In some instances, job titles did not readily fit into a single area; in such situations, hybrid exposure classes were developed. A job-exposure matrix, which tabulated exposure-class codes by )ear, was constructed for each of the two locations. For Location 1, actual results from past industrial-hygiene measurements" were tntered in their respective cells in this matrix. Cells for which no data were available were completed by interpolation between available previous and subsequent values. When interpolation could not bc performed because no measured value existed for an exposure class in the first or last year of the study, the nearest measured value br that exposure class was projected forward or backward. Indus[rial-hygienemeasurements obtained at Location 2 were applied to he matrix in similar fashion. Processes and job assignments were tlscniially identical at both locations, so benzene exposure levels measured at Location I were assumed to be naturally occurring simulations of exposure levels in corresponding areas at Location 2, when actual exposure measurements did not exist. Population All nonsalitried white men employed in a rubber hydrochloride department for at least one day tietweenJanu;try I , 1910,and December 31, 1965, were eligible for the study. Men whose initial exposure occurred after December 3I , 1965, were excluded, because 1965 was the year in which production of rubber hydrochloride ceased at Location 2. Very few men were lirst hired at Lcicatio~I~ aftrr 1965. Vital status was ascertained Ihr f h e CCJhOrt through IJecembcr 31, 1981. Follow-up for vital status was accomplished primarily through th'e Social Security Administration, but also through the Ohio Bureau of Motor Vehicles and a commercial agency specializing in locating people. Cohort members who were not traced were considered to be alive as of the study's ending date. Death certificates for all the known deaths were obtained and coded by a qualified nosologist according to the rules of the /n!crndionn[ ChrQ5rnlion of Disrflses, Adaplcd for U J i~n l e Unikd Slates, that were in effect at the time of death. Each code was then converted to one of 8Y "death categories" for use i n the NIOSH life-table analysis system. Analytical Methods T h e NIOSH life-table analysis system2' was used to generate expected numbers of cause-specific deaths, within five-year age and five-year calendar periods. These calculations were based on death rates for U.S. white males specific for the same five-year age and calendar periods, applied to the number of person-years at risk of dying. Person-years were further stratified according to cumulative benzene exposure and five-year latency periods (intervals since the initial exposure). To determine cumulative benzene exposure, a person's daily benzene exposure was obtained from the appropriate cell in the exposure class-year matrix. These daily values were then summed for a man's entire working career. T h e accumulation of observed deaths and of person-years at risk of dying began on January I , 1950, or on the first day on which his cumulative personal exposure to benzene reached 1.0 ppm-day (one day of employment in an exposed department), whichever occurred later. Observation ceased on December 31, 1981, or on the date of death - whichever occurred earlier. The person-years in the cohort were divided into four categories of exposure. These exposure strata were I ppm-day to 40 ppmyears, 40 to 199.99 ppm-years, 200 to 399.99 ppni-years, and more than 400 ppm-years. These boundaries correspond to the cumulative exposures that would result from average annual exposures to less than I , I to 4.99, 5 to 9.99, and I O or more ppm of benzene, respectively, accumulated over a 40-year working lifetime. To obtain cause-specific standardized mortality ratios, the observed numbers of deaths from each cause were divided by the expected numbers and multiplied by 100. Ninety-five percent confidence intervals were calculated for each cause of death examined." In addition, a matched case-control analysis was performed with use of conditional logistic regression. This analysis was intended to (1) evaluate the exposure terms that govern the relation between the risk of death from leukemia and exposure to benzene, (2) evaluate the erect of potential confounders and erect modifiers on this relation, and (3) identify the functional form of the exposure-response relation. The exposure terms evaluated were cumulative exposure, duration of exposure, and rate of exposure (cumulative exposure divided by duration of exposure). T e n controls were matched to each cohort member who died of leukemia for the year of birth and the year first employed. As suggested by Thomas,ZGthese controls were selected from among the cohort members still alive at the time of death of the corresponding case. Because i t is generally believed that some latency period subsequent to an initial exposure is required for leukemia to develop, a separate analysis that "lagged" exposures was also performed. jVithin each matched set, all bcnzene exposures within tltr Iiw-year Table 1. Observed and Expected Deaths from All Causes, All 5 to over 30 years; however, seven of the ni Malignant Neoplasms, and Lymphatic and Hematopoietic Cancers in Rubber Workers Exposed to Benzene. persons with leukemia had less than 20 years latency. CAUSOEF DEATH NO. OF DEATHS ORSERVEO EXPECTED 5 rANDARD1ZED hloRTAt.lTv RATIO (9.7% C.I.l* Because this observation was based on only n deaths from leukemia, there was the possibility tl this strongly positive trend in leukemia-associal mortality might be an artifact produced by our cho All causes All malignant neoplasms Lymphatic and hematopoietic cancers Leukemia Multiple myeloma 330 331.6 99 (89-1 I I ) 69 66.R 103 (80-130) 15 6.6 227 (127-376) 9 2.7 337 (154-641) 4 I .o 409 (I 10-1047) of boundaries for the exposure categories. T o I amine this possibility, we arbitrarily changed t sizes of the categories, first by halving the ori nal ranges to 0 through 19, 20 through 99, I through 199, and 200 or more ppm-years, and th `C.I. dcnotcs conlidcncc isrcrvel. by doubling the original ranges to 0 through 80 through 399, and 400 or more ppm-years. T resultant standardized mortality ratios continued period I d o r e tlic drnth of the case were ignored in calculating each pcrsoii's cuniul;iiivc lol;il. both instances to show a strongly positive trend increasing risk with increasing exposure. (Standai Cohort Analysis RJLSULTS ized mortality ratios for exposure ranges X Vz 134, 277, 0 deaths, and 2338; ratios for expos1 ranges X 2 = 141, 609, and 6833.) These findin A total of 1165 white men with at least I ppm-day are evidence of the robustness of the observed i of cumulative exposure to benzene through December socia tion. 3 1, 1965, were included in the cohort. They contribut- Standardized mortality ratios for multiple myelor ed 3 1,612 person-years at risk. O n December 3 1, I98 1, over the four original exposure strata did not increa a total of 819 (70.3 percent) were alive, 330 (28.3 with increasing exposure (Table 3). Three of the fa percent) were dead, and 16 (1.4 percent) were lost to men who died of myeloma had less than 40 ppm-ye: follow-up. Those lost to follow-up were considered to of exposure (on the basis of our assumptions of dosi be alive as of the study's ending date. and all four deaths occurred after 20 years of latenc Neither the mortality from all causes of death com- All four were from Location 1. Case descriptions o f t bined (330 observed vs. 331.6 expected) nor the mor- deaths due to leukemia and multiple myeloma a tality from all malignant neoplasms combined (69 ob- given in Table 4. served vs. 66.8 expected) was above the expected rate (Table I ) . There was, however, a statistically sig- Case-Control Analysts nificant increase in deaths from all lymphatic and Examination of data from the case-control analy! hematopoietic neoplasms (15 observed vs. 6.6 ex- indicated that the mean cumulative exposure w pected; standardized mortality ratio, 227; 95 percent higher for cases than for controls (254 vs. 50 ppi confidence interval, 127 to 376). This increase was years). Also, the average duration of exposure w due mainly to excess numbers of deaths from leu- longer for the cases (8.7 vs. 2.6 years). Finally, the kemia (9 observed vs. 2.7 expected; standardized mor- was a difference in rates of exposure between cas tality ratio, 337; 95 percent con- fidence interval, 154 to 641) and from multiple myeloma (4 observed vs. 1 expected; standardized mortality ratio, 409; 95 percent confi- Table 2. Observed and Expected Deaths from Leukemia in 1165 White Men with Least One Day of Exposure to Benzene from January 1,1940, through December 3 1965, According to Cumulative Exposure and Years of Latency. dence interval, 1 IO to 1047). LATENCI YYR ) Standardized mortality ratios for TOTAL. leukemia, over the four exposure strata ( 1 ppm-day to 39.99 ppmyears, 40 to 199.99 ppm-years, 200 to 399.99 ppm-years, and more than 400 ppm-years) demonstrated a marked, progressive increase c5 >IO 10-15 15-20 20-25 uo.10 W.16 010.22 010.27 010.32 . 010.02 010.05 1/0.07 1lO.09 Ml.10 - 010.01 110.02 110.03 010.03 - - 010.00 2lo.01 110.01 uo.I2 010.22 m.31 310.39 110.46 with increasing cumulative expo- 25-30 010.37 010.12 010.04 0/0.01 010.54 sure to benzene (standardized mor- >30 W.40 0/0.16 110.04 010.0I 1iQ.62 tality ratios were 109, 322, 1186, Total* U1.83 U0.62 U0.17 310.04 9i2.66 and 6637, respectively) (Table 2). No apparent pattern was evident for these deaths with regard to la- Standardized mortality ratio Confidence interval 109 12-394 322 36-1 165 I I86 133-4285 6637 1334-19.393 331 154-641 tency, which ranged from under `The numbers of expeckd deaths have been rounded. I. Table 3. Observed and Expected Deaths from Multiple Myeloma in 1165 White Men wilh at Least One Day of Exposure to Benzene from January 1.1940, through Decem- ber 31, 1965,According to Cumulative Exposure and Years of Latency. founders, or efTect modifiers, and Bi coeficients be estimated, Otlds ratios calculated by this technique L.ATl:NCY ( Y R ) EXKISURc: IPPM-YR) are expressed relative to that ol' an 0.IXll-40 JO-?IJu 200-1l10 241x1 rorAi unexposed workcr, in wliicli X i is obrcrvrdle.cpccml rlrathhs considered to be 0. <5 W0.02 O/O.Ol 0/0.02 We examined several moclels to 5-10 wo.04 0/0.01 0/0.00 0/0.05 identify one that woultl adequately 10-15 W0.07 w0.02 O/O.OI o/o.oo om IO explain the risk of death from leu- 15-20 wO.09 010.03 O/O.Ol o/o.oo 010.14 kemia with the minimal numher 20-25 I/O.I2 WO.04 O/O.Ol 1/0.00 ao.I8 of terms. In our first examination, 25-30 uO.15 W0.05 0/0.02 WO.00 a0.22 we considered three cxposure vari- 230 Total Slandardized mortality ratio Cwlidence interval 010.17 3/0.65 4511 92-1339 W0.07 0/0.24 I OIO.02 0/0.07 * wo.01 110.02 5347 70-29.753 010.27 410.98 3911 I IO-IO47 ables separately - cumulative ex- posure, duration of exposure, and average exposure rate -and we lit- ted three separate models, one for *Nav Jcaihr were obvrved wilh thcv exposure levels. each of these variables. I n these three models, cumulative exposure (ppm-years) was found to be the ;iiid controls, with cases averaging approximately 24 strongest single predictor of death from leukemia 111)ni of benzene per day, as compared with approxi- ( p = 0.0126; 95 percent confidence interval, 0.0028 to iii;itely 16.5 ppm per day for controls. 0.0224; chi-square = 6.4; I' = 0.01 1 ) . Then, for a '1'0 evaluate the exposure terms that govern the ex- more complex examination of the same three intercor- Iwsure-response relation between benzene and leuke- related exposure variables, we constructed another iiiia and to assess potential confounders and eKect model in which all three were entered simultaneously. rriodiliers, we analyzed the case-control data using In this model, only cumulative exposure was found to coiiditional logistic regression." This analysis pro- contribute materially to the risk of death from leuke- tluccs odds ratios of the general form mia, although i t is not possible with only nine cases to OR = exp(BIX, + . . . + B,,X,,), establish cumulative exposure as the unqualified best expression. Interactions among cumulative cxposurr, \here X i represents exposure variables, potential con- duration, and rate of exposure wcre also examined. Table 4. Descriptions of Deaths from Leukemia and Multiple Myeloma in 13 Workers Exposed to Benzene. CASE NO. I 2 3 4 5 6 1 8 9 IO II 12 13 AGE AT DEAri+ 36 29 60 65 62 YEAROF DEATH I958 I950 IOS8 I960 1%1 57 1961 51 I957 28 I954 67 I979 69 1980 52 1963 62 I968 68 1981 LATENCY. (YR) CAUSE OF DEATH* 17 2 I3 VI ISE 22 20 15 3M 37 25E 22E 24M 26% Monocytic leukemia I204) Chronic myelogenous leukenii;~( 2 0 4 ) Acutc n~yclc~cytic leukemia (2W) Acute myelogenous leukemia (20.1) Di Gugliclmo's acute myelocytic leukemia (204) Acute gr:inultrylic leukemia (204) Acute monocytic leukemia (204) Myelogent~usleukemia (2114) Acute myeloblastic leukemia ( 2 0 4 ) Multiple myeloma (2031 Multiple myeloma (203) Plasma-cell sarcoma (201) Mulliplc myeloma (203) CORROBORIAN (I. M~DICAI Ktmars None ;~vail;ihlc Hospital. sutopsy. l i w i c slide\ II(irpilii1. IICII~;Itologist Hematologist. hospital. tissuc slides Hospital, physician llorpit;~l.tissuc slides. autopry Tissue slides None availahlc None available None available Hospital Hospital None availablc PI ANT LOCATION: ~ U R A l I O N(11 EhIH I1YhIl:N I Liration I; 1 % yr I.w;ition I; I Ill,> i.ic;itioii 2 : I II/: yr Liratiiin 2; 14 yr Loc-ation 2; 13 yr Lir;itim 2; 20 yr Location 2; S yr Liration I ; IV? yr Location 2; I1yr Locution I ; I E yr lrxxion I; 4 days Location I ; 22 yr Ixxation I: Y 1111) cuMu1.mI V E H P N Z ~ N I E. XIWS~IRP III~M-I'Y) 49.99 0.I O 1SU.511 49x.23 17x.45 hS9.XJ `In .55 10.Ih 252.66 19.50 0.I I h52.66 7.15 None of those intcraction terms were fioltnd to contributc significantly to the model in predicting the risk or drat11 from leukemia. 'l'ltc s h p c or tlic rxposurc-response function was tlirn cvaluatccl with use of several models. l:irst, since thc distribution or cumulative exposures was highly skewcd, wc examined a logarithmic transformation of cumulative exposures. T h e results or this analysis indicated that the fit ror this model (chisquare = 4.86; P = 0.027) was less adequate than that determined above for the untransformcd measure (chi-square = 6.4; P = 0.01 I ) . Then, to investigate the possibility of a morc general Form or curvature, we adtlcd a quadratic tcrm for cumrilativc cxposurc. 'Iltis maneuver ditl not significantly improve the fit of tlic model, however ( P = 0.91). From these findings, we determined that the untransrormed model provided the best representation or thc exposure-response relation. With this model, the equation best describing the odds ratio Tor lcukcmin in relation to cumulative exposure to 1,cnzrnr was drtcrmincrl t o I)r intervals were calculated Tor cumulativc lifetime cxp surcs to benzene ranging from 0 to 400 ppm-ycar From this cquation, ~ h arvcr;ige cumulative. c x p ~ s r t i of thc cascs niicl controls (69 plm-yc;irs) was fi)itttd I produce an otltls ratio rclativc to tlir uiicxposccl worl ers of 2.4 (95 pcrcent confidence interval, 1.2 to 4.7 To ensure that thc odds ratios Tor the matchcd scts ( cases and controls werc homogeneous (a prercquisii for the above analyses), in terartions I)ctween (-11 ti1til; tive exposure and the matching vari;il,lcs (ycar I birth and year of first exposure) wcre introtlucrd i i t t the model. Neither o f these intcractions was round I be significant. '1'0 takc into account a n incluction period Tor Icr kcmia, bcnzcnc cxposurcs occurring within tlic fivc year period berore the death o r a caw wcrc climinatr from the calculated cumulative cxposure or earh ma in a matched sct. Wc thcn reexamined thc clTcct c cumulative exposure. T h e odds ratios incrrasrd sliglii ly ( p = 0.0169), as ditl thr statistical significnncr I 111col)srrviition (chi-sqrtiirr = fj.7; I' = 0.010). OR = cxp(0.0126 x ppm-years). DISCUSSION T h e exposure-eiTcct curve defined by this equation T h e principal findings or this analysis arc t h ; ~ was plotted (Fig. I ) . Nincty-five percent confidence (1 ) there is a strongly positivc cxposurc-rcspons relation Iwtwccn bcnzenc ant1 Ici /100 kcniia; (2) on tlic basis of our mot1 cl, this rrlation can bc pro.jcctc'1 downward to mcan annual cxlx 80 surc Icvels of less than I ppm vi 0n mulated over a 40-year workin Iirctinic; and (3) in thc poprtlatio 60 stutlicct, there was ;dso a statistical R A T 40 I 0 20 SEE OETAIL EXP[O.OlZB X PPM-YEARS] ly signilicant excess of dcaths rror multiple myeloma. T h e environmental data used i this risk assessment are admittcdl incompletc. Mcasurctl cnviron mcntal lcvcls or bcttzenc did no 0 cxist for all j o b in all yrars. Gap 100 150 200 250 300 350 . 400 hac1 to be lillrd h y interpolatioi PARTS PER MILLION YEARS rrom existing data. This interpci lation was prrlbrnird according 11 4.515.04 OETllL OF ABOVE 4.0 8 3.5I3.0- R 2.5- ., prccst;iblislicrl rulcs and witliotr knowlcrlgc of a person's tliscas outcome. I n somc c;iscs this pro ccclurc required t h a t w e allow ;I sill glc nirasurctl cxposurc to scrvr ;I ati intlcx of cxposurc liir ii nutn1)r ol' years. Episotlcs or high rxpositrl d u e 10 such tcmporary circuit1 2.0- st;inccs as spills and procrss iipsrt 0I 1.5- ._.............._.._................*........-...---.--..w--c-r-c---probably overlookctl hy 1111 1.0- 0.54 Figure 1. The Risk of Leukemia in Relaiio to Cumulative Exposure to Benzene. 0.0 0 20 40 60 BO 100 + 120 The lower panel shows an enlarged deta of Ihe lower left-hand section of PARTS PER MILLION YEARS the upper graph. III<N%INII ANI) l.l:utiI~hl ititlustrial-liygierie surveys and are therefore not re- 1957. His estimated cumulative benzene exposure llcctetl iii our calculations of exposure. Percutaneous was 90.56 ppm-years. Thirty-four years had elapsed h o l . l ) t i ( J l i Of beliZeIie, a route Of exposure that has between liis first exposure and his death. Because die rrcciitly lieen shown to be of potential importance, vital status was not determined for members of tlie iiied.'" Nevertheless, tlie existing envi- coliort beyond 198I, inclusion 01' this case in the series roiinieiitiil data are unusually comlxeliensive in com- is not entirely consistent with the selection criteria, se typically available liir retrospective wliicli called lor controls to be selected from among cpitlemiologic studies. 'l'liey permit a reasonable esti- cohort members still alive at the time 01' the death initit: oi' cumulative beiizene exposure during rubber 01' the case. 'I'lierefore, these aclclitional data are not hytlrocliloridc productioii Iiw cacli nicniber ol' this iiicludecl in our estiniatcs of' risk but serve as corrobo- 11. I~xmii~ratio01i'i these cxposurc data ratiiig evidence. utlc tlia t cniployces' historical eiglit- Multiple myeloma, tlie ciiuse of tleatli iii four mem- Iioiir time-wciglitccl average exposures to airborne bers of this coliort, Iias tieen observed previously in Ixiizeiie had generally i i ( ~et xceeded the limits in ellect persons exposed to benzene, although also in small dl ;illy given time. numbers.29 In addition, several recent toxicologic II' tlie environmental data are i n error, we believe studies have demonstrated lymphoid cancers in both tlicy probably err i n overestimating actual average ex- rats and mice exposed to benzene."-15 I t is of interest posures, inasmuch ;is tlie majority ol' measurements that three ol' the four deaths liom multiple myeloma were taken by industrial hygienists looking for trouble that were observed in this cohort occurred in the rlx)~swithin the process rather tlian trying to docu- group with the lowest cumulative exposure to benzene nicrit typical occupational exposures. There were (<40 ppm-years), and that all four persons who died slrotig economic incentives in tlie rubber hydrochlo- had exceptionally long latency periods for hematologic ride inanulicturing process to recover the costly ben- cancers (>20 years). These two observations raise the wile elliciently; indeed, much of tlie process was dedi- possibility that relatively low cumulative exposures to cated to that end. Continuous contamination of tlie benzene may produce a relatively well-dili'erentiated work area by benzene a t tlie average levels of exposure cancer such as multiple myeloma, whereas higher ex- lave been economi- posures may lead to leukemia. I n conclusion, tlie results of this risk assessment 'l'lie 95 percent confidence interval becomes ex- indicate that a n exponential decrease in the risk of lrrriiely wick in the higher dose ranges considered death from leukemia could be achieved by lowering (Fig. I ) . 'I'liis is primarily a function of there being occupational exposure to benzene. According to the udy iiiiie deaths from leukemia i n tlie series. However, model derived in this study, a worker occupationally I w d i the categorical and tlie conditional logistic re- exposed to benzene a t an average exposure level of 10 Rrcssion analyses indicated exponential increases in ppm for 40 years would have an increased risk of rrlative risk with increasing cumulative exposure to death from leukemia of 154.5 (95 percent confidence Iwiizene. ?'lierefore, although the exact estimate of interval, 3.1 to 7785). If the average exposure were rrl;itive risk at higher dose levels lies within a wide lowered to 1 ppm, that excess risk would decrease to rttieless tliat an ex- 1.7 (95 perccrit conficlence interval, 1.1 to 2.5). At 0. I L lioin any loweriiig ppm, the risk would be virtually equivalent to the background risk (odds ratio, 1.05; confidence interval, of an association such as that ob- 1.01 to 1.09). es if i t reniaiiis eviclent after inipojtl rcllcctiiig induction latency. 1 1 1 REFEKENCES miocl was imposed by disco~~iiting le Inatcllcd sets all bellzelle cxpoirred i n tlie live years belore tlie I . Ix'Iorc P. Burgonuno C. Lcuci-niie aigui' au cuurs de I'intoxicatiun benrCni- que: sur I'uriginc toxiquc de ccnaitics Icuci-inies aiguc's et lcurs relations avec l a ani-mics graves. J Med Lyuii 1Y28; 9:227-33. 2. Mallury TB. Call EA. Brickley WJ. Chronic exposure tu benzene (benzol). 111. The pathologic results. J Ind Hyg Tuxicol IY3Y; 21:355-77. 3. Browning E. Toxicity and tiietaholisin uf industrial solvents. New York: EIscvier Press. IY65:3-h5. 4. Cavigtieau L. Lcs intoxications cunlimices. Cah Med lnterprof 1962; 2 2 8 - , we learned of an- 36. 5 . Vigliani EC, Saita C. Benzene and leukemia. N Engl J Med 1964; 27 I:872- 6. Ollort, of il 68-yearer o u r vital-status f i ~ l l ~ ~ -ttliaptc 01' 1)ecenibcr 31, 1981). 'l'lie ca~ise 01' t1c;itli i ~ o r t l e don the death certificate was "inecl- keniia due to m y - h. Aksiiy M. Erdeni S. DinCol G . Leukemia in shoe-workers exposed chronically Io hunzcnc. IHwd I'J74: 44:X37-41, 7. Vigliuni EC. Leukcniia asauciated with twirene exposure. Ann NY Acid Sci IY76: 271:143-51. 8. lnlhiilz PF. Kinsky KA, Wagoner JK. Young RJ. Leukxmia in k n r e n e workers. Lancet 1977; 276-8. 9. Ott G M . Townwnd JC, Fishbeck W A . Langner RA. Mortality ainung indi- mast cell disease." jobs at Location 2 ;iI)oiit 3'A years, I)ctcvcen August 1952 and May viduals occupationally exposed to benzene. Arch Enviruii Health l97X: 333-10. IO. Riniky KA. Young KJ, Smith A B . Leukemia in bonzetis workers. Ant J Ind Mcd IYXI; 2:217-45. VI I I. Goldstein BD. Benzene toxicity: ;I review o f recent literature. Piscataway. N.J.: Department of Envirunniental and Community Medicine. Univcrsity nf Medicine and Dentistry of Ncw Jersey-Kutgen Medical School. 1983. 12. Maltoni C. Scamato C. First experimental demonstrationof the carcinogenic cfrects of benzene: long-term bioassays on Sprague-Dawley rats by oral adniinistration. Mcd Lav 1979: 70:352-7. 13. Snyder CA. Goldstein BD. Sellakunnr AR. Bromberg I.Laskin S, Albert RE. The inhalation toxicology of benzene: incidence of hematopoietic neoplasnis and hematotoxicity in AKRlJ and C57BU6J mice. Toxicol Appl Phnrmacol I 9 8 0 54:323-3 I. 14. Maltoni C. Conti B. Cotti G. Bcnzene: a tiiultipotcntial carcinogen: results of long-term b i u w a y s performed at the Uologna Institute oT Oncology. Alii J Ind Mcd 1983: 4:589-630. 15. NTP technical report on the toxicology and carcinogenesis studies o f benzcne in F344/N rats and B6C3Fl mice (gavage studies). Research Triangle Park. N.C.: National Toxicology Program. 1986. (NIH publication no. 86- 2545.) 16. National Institute for Occupational Sakty and Health. NIOSH revised rec- oinnicnd;ltion for an occupational exposure standard for benzcne. Cincinnati. Ohio: National lnstitutc for Okcupational Sdety and Health. 1976. IIXIEW puhlication no. (NIOSII) 76-137-A,) 17. hllxrt K l i . C;itiiiiogcn aswr'iiiicnt griiitp's linal rcptrt iin p)pul;ition risk to attihictit hcii~cticc x l ~ i b u m ~K.cscarch 'I'riaiiglc Park. N.C.: Uiiitcd Statcs Environtiiental Protection Aguncy , 1Y7YI (EPA-450/5-80-004.) 18. International Agency Tor Kesearch on Cancer. IARC monographs on the evaluation o f the carcinogenic risk of chemicals tu humans: some industrial chemicals and dyestuffs. Vul. 2Y. Lyon. France: International Agency Tor Kescurch on Canccr. IYXZ:Y3-14X. 19. Occupational Safety and Health Administration. Occupational exposurc I, henzcnc. Fed Regist. February IO. 1978; 43:5YIX-70. 20. Occupational Safety and Health Administration. Docket H-0.59. Benzenr Washington. D.C.: Occupational Safety and Health Adniinistration. Tcth nical Data Center. 1977. 2 I. Ashford NA. Risk assessment and the design of policy for worker ~ N I C C lion. Am J Ind Med 1982; 3:241-2. 22. U.S. Supreme Court. Industrial Union Department. AFL-CIO vs. Amcricai Petroleum Institute. July 2. 1980. (Supreme Court publication no. 4411 U: 607.) 23. White MC. Infante PF. Chu KC. A quantitative eslitnate or leukcniia mix tulity assvciatcd with occupational cxposurc to hcnzenc. Risk Anal IYH! 2:19.5-204. 24. Waxwciler RJ. BeaumontJJ, Henry JA. et al. A modified lirc-table analpi, system fur cohort studies. J Occup Mcd 1983; 25: 115-24. 25. Rothman KJ. Boice JD Jr. Epidemiologic analysis with a prograniniabk calculator. Washington. D.C.: National Institutes of Health. IY7Y:pZU (NIH publication no. 79-1649.) 26. Thomas DC. Addrftduum fo Methods o f cohort analysis: appraisal by applica tion to asbestos mining. J R Stat S ~ H1:977; 140:483-5. 27. Kleinhauni Dc.Kupper LL. Morgcnstcrn H. Epidemiologic research: prin ciplcs and quantirativc methods. Ucliiiont. Calif.: Lifctimc Lcarning Puhli c;ltiotis. lY112:4YZ-S03. 28. Sustcn AS, Dames EL. Burg JR. Nieriieier RW.Percutaneouspenctrntim of benzene in hairless mice: an estimate or dermal absorption during lirr building operations. Ant J Ind Med 1985: 7323-35. 29. Decouflc P. Blattner WA. Blair A. Mortality among chemical worlcv cxpused tu bcnzenc and other agents. Enviruri Res 1983;30:16-25. MARROW TRANSPLANTATION IN PATIENTS WITH ADVANCED THALASSEMIA GUIDOLUCAREI-MLI.,D., MARIELLGAALIMBERTI,M.D., 1'AOI.A POLCtII, M.D., CLAUDIOGIARDINI,M.D., PATRICIA POI.lTI, M.D., DONATELLUAARONCIANI, M.D., EMANIIEIAXNCELUCCI,M.D., I;LAVIA MANENrI, M.D., CONSTANTI ~EE I . F I N I , M . l I . , GIOVANNAIUKELIb,l.lJ., AND 1'1 E'I'RO M UKIXTO, M .I). Abstract In a study of the outcome of marrow transplantation in patients with advanced thalassemia, 40 patients with homozygous p-thalassemia who were 8 to 15 years of age (median, 10) received HLA-identicalallogeneic marrow after treatment with busulfan and cyclophosphamide. Twenty-eight of the 40 patients were alive and free of disease 260 to 939 days after transplantation, and 2 patients were alive with thalassemia 372 and 1133 days after transplantation. The actuarial probabilities of survival and of disease-free survival at two years were 75 percent and 69 percent, respectively. Ten patients (25 percent) died. Three died of cardiac failure, interstitial pneumonitis, or septicemia within 14 days of transplantation. Three died of infectious complications associated with acute graft-versus-host disease at 46 to 97 days, and two died of infectious complications of chronic graft- versus-host disease at 249 and 290 days. Two patients had transplant rejection and died with marrow aplasia 115 and 192 days after transplantation. One patient had rejection after four months and while the marrow was aplastic underwent a successful second transplantation; the patient was alive without thalassemia 624 days after the first transplantation. The actuarial probabilityof grade 2 or higher acute graft-versus-host disease in the 32 patients with initial sustained engraftment was 35 per. cent. Three patients had chronic graft-versus-host dis. ease, which was fatal in two and still active on day 710 in the third. We conclude that bone marrow transpJantation can po tentially save patients with advanced thalassemia from an otherwise inexorable progression to death from Ihe complicationsof blood transfusions.The ultimateoutcome in this group of patients must await a longer follow-up. (N Engl J Med 1987; 316:1050-5.) PI th W; Ye or bri V;I trc PC blc I; min tint. ana ldrt led ty-1 nor! bilit LllC I tatic was A Satii orall cycl1 cadi the I was Tor p patic intra 71CIC ind xivc MARROW transplantatioii is being explored as an alternative to blood transfusions and chelat- ing therapy for patients with Thomas and coworkers postulated that transplantation early i n the course of the disease would result in a high Froin the Division uf Cleniatology of Muruglia. the Ccnler fur Bonc Marrow Transplantation. the Service u f the General Laboratory. the Service of Anatumical Pathology. and the Hospital of Pesaro. Address rcprint requests to Dr. Lucarclli at Divisionc Einatologica. Ospedali Riuniti di Pesaro. 61 100 Pcsaru. Italy. Supported by a grant f19X5) frum the Rcgione Marche. Ancona: by the Bcrloni hiund;~tiwa~gainst lhalasretiiia. Pesaro; and by the Italian Association against Lcukcmia. Section of Pc\aru. prol>ability of disease-free siirvivaI, whereas traiiaplantation later in the course might result in a h i g h probability of treatment failure and death from trailsplant-related causes.'" T h e results in the first serics of patients given transplants for thalassemia in I'csarci. Italy, appeared to substantiate this hypothesis, sinrr tlicre were no long-term survivors among patictm undergoing tratisplantation for advanced discasr after rcgimens that included total-hotly irradiation.' I n a sul~sequentseries of patients undergoing 1raiisplantation early (hefbrr the age of eight) after p r r p aration with Iiusulfan and c.)rclo~~liospliamidter,ails- .- , .