Document em6KzvZo4b0ZpEG8BdLw6dqqg

ik- Risk Analysis, Vol. No. 6, 1996 ."",., Determination of Leukemogenic Benzene Exposure Concentrations: Refined Analyses of the Pliofilm Cohort A. Robert Schnatter,!,2 Mark J. Nicolich l and Michael G. Birdl t ~'0 AA Received March 18, 1996; revised August , 1996 Biologic data on benzene metabolite doses, cytotoxicity, and genotoxicity often show that these effects do not vary directly with cumulative benzene exposure (i. , concentration times time, or cleXukefm). iaTmo oerxtaamlitiynientPhleioeffiflemctwoofrkaenrsa.ltTerhneawteoerxkphoissutorerymoeftreiacc, hwPelaionfaillmyzewdorckeellr-twypaes supseedciftioc define each worker s maxim~lly exposed job/department combination over time and the associated long-term average concentration associated with the maximally exposed job (LTA- MEJ). Using this measure, in conjunction with four job exposure estimates, we calculated SMRs for groups of workers with increasing LTA-MEJs. The analyses suggest that a critical concentration of benzene exposure must be reached in order for the risk of leukemia or, more specifically, AMML to be expressed. The minimum concentration is between 20 and 60 ppm depending on the exposure estimate and endpoint (all leukemias or AMMLs only). We. believe these analyses are a useful adjunct to previous analyses of the Pliofilm data. They suggests that (a) AMML risk is shown only above a critical concentration of benzene exposure, measured as a long-term average and experienced for years, (b) the critical concentration is between 50 and 60 ppm when using a median exposure estimate derived fi:om three previous exposure assessments, and is between 20 and 25 ppm using the lowest exposure estimates, and (c) risks for total leukemia are driven by risks for AMML, suggesting that AMML is the cell type related to benzene exposure. KEY WORDS: Benzene; leukemia; exposure concentration; threshold; epidemiological data. 1. INTRODUCTION blood following an equivalent cumulative exposure of lO-week, 6-hour/day, 5-day/week exposure at 10 ppm. The leukemogenic potential of benzene has been well-documented.o-15) Biologic evidence suggests that the timing, duration, and concentration of exposure are important independent factors in predicting benzene hematotoxic, genotoxic, and possibly leukemogenic ef- fects. Specifically, Green et al. (16) reported that severe toxicity to the bone marrow, spleen, and peripheral blood was observed in CD- l mice following a 5-day, 6- hour/day, 100-ppm exposure (125 ppm-days) while no toxicity was observed to the bone marrow or peripheral Toft et al. (17) exposed male NMRI mice to various ex- posure regimens and reported several examples of dose rate effects for cellularity, granulopoietic stem cells (CFU-GM), and micronuclei in polychromatic erythrocytes (MN-PCE' s). As an example, a MN-PCE were significantly higher for a 95-ppm, 2-dayregimen (190 ppm-days) compared to a 21-ppm, 10-day (21O-ppmdays) regimen, Exposure intermittency may also have a role in benzene-induced myelotoxicity. Luke et al. (18) showed that for 300-ppm exposure to benzene, suppres- sion of polychromatic erythrocyte (PCE) counts in mice I Exxon Biomedical Sciences , Inc., Mettlers Road, CN 2350 , East Millstone, New Jersey 08875-2350. 2 To whom al1 correspondence should be addressed. was more persistent for 3-day consecutive exposures rather than 5-day consecutive exposures. Some pharmacokinetic models have examined po- 833 0272-4332/9611200-0833$09.5011 \!:) 1996 Society for Risk Aoalysis GARABRANT/BERRYMAN 00332 , .'.. !.. 834 Schnatter, Nicolich, and Bird ii: tential dose rate effects of benzene exposure.(19-24) Bois 2. METHODS and PaxmanpO) and Watanabe et al. (22) predict greater internal metabolite levels for short-term, high-level ad- ministered doses compared to longer term, lower level exposures. COX(23,24) showed that the same cumulative 1. Subjects/Statistical Analysis exposure regimen can have different hematotoxic effects depending upon how benzene is administered over time. This model predicts that a high level discontinuous exposure has a greater effect on bone marrow proliferation. Premature bone marrow stem cell proliferation is a critical step in recently proposed models of leukemogene- sis. (25,26) The above research suggests that benzene dose rate effects may be important, which we define as an effect which relies on exposure concentration in a manner other than as a simple, linear product term of cumulative exposure (concentration X time). The above data apply to internal benzene metabolite doses, cytotoxicity, and some measures of genotoxicity. Comparable data on leu- kemogenesis is lacking. However, since the induction of leukemia (specifically, acute nonlymphocytic leukemia or ANLL) is likely a long term, multifactorial process The employees included in this evaluation are de- scribed in detail elsewhere. (5,6) Vital status of the cohort was determined through 1987 for most workers , and 1981 for one subgroup. Age and time period specific rates of total leukemias and acute myelocytic and mon- ocytic leukemia (AMML) were obtained for the U. S. and the state of Ohio. Since similar results were obtained using both sets of rates, only results for u.S. rates are reported here. Expected deaths were calculated by mul- tiplying distributions of person years (through 1981 or 1987) by these rates, specific to time period, age, and gender. The population was restricted to non-Black nonfemale employees, consistent with previous publications. Standardized Mortality Ratios (SMRs) were calculated in a conventional manner using the Occupational Mor- tality Analysis Program (OCMAP) program.f2S) involving metabolism, (27) and both cytotoxicity and gen- otoxicity, (25,26) a reasonable supposition is that these ef- fects also play role in benzene-induced leukemogenesis: Another corollary of the multifactorial leukemogenesis model borne out by empirical data(3,5) is that long term exposure at such critical concentrations is necessary, and it is unlikely that exposures of hours days, or weeks have any practical relevance to leuke- mogenesIs. Most epidemiologic studies, including those on benzene-exposed populations , estimate cumulative ex- posure in ppm-years (c X f). When leukemic risk is estimated by cumulative exposure, it is implicitly as- 2. Exposure Determination Three different sets of L T A exposure estimates have been developed for the Pliofilm workers.(1. 15) Analyses have been performed with each set of exposure estimates, plus an estimate derived by considering the median of the three sets of estimates. The median exposure estimate was calculated by simply selecting the inmiddle-ranked estimate for each cell the job/department/time exposure matrix. One advantage of this estimate is that it disregards extreme exposure estimates (either high or low), which have been the subject sumed that (in ppm) and (in years) are equal con- tributors to an exposure index (ppm-years) which is then used to predict leukemic risk. While some refinements have been used such as using a nonlinear combination of previous criticism. (S,29) Each worker s maximally ex- posed job/department combination over time and the long term average concentration associated with the maximally exposed job (LTA-MEJ) was calculated for of concentration and duration (en t),cS) and entering simultaneous terms for , t and f,5) these maneuvers have not been fully investigated, and still rely on model extrapolations to predict low-level effects. To further explore the possible impact of exposure the median as well each series(I 15) of exposure esti- mates. This enab1ed the enumeration of subgroups of workers and person-years who were always exposed less than or equal to specific concentrations of benzene. 3 varying the exposure concentration cutpoint in small in- rate effects on leukemogenesis, we used the Pliofilm data, (5-S) which is the basis of most risk assessments and regulatory decisions. (9-14) Effects of specific benzene crements and calculating SMRs for workers exposed below each cutpoint, exposure concentrations which are L T A exposure concentrations, rather than the estimated cumulative exposure of benzene were analyzed. The ex- posure concentrations investigated were those used by 3 To avoid immortal person-years, the maximally-exposed job (MEJ) was determined over an entire workers career. A worker s MEJ could change over time. If he first experienced jobs with low exposures previous authors(I, 15) and represent long-term averages for jobs held by employees in the Pliofilm cohort. and subsequently experienced jobs with a higher L T A concentration the worker would contribute person-years first to lower, then higher concentrations. GARABRANT/BERRYMAN 00333 ":,,j' ~,.: i:7=~'- I-e 17: fl't' ;h~: "i- t.; Benzene Concentrations and Leukemic Risk 835 Exposure concentration (ppm) '~L~ 100 120 140 150 175 200 250 260 Table I. Number of Wor:Kers aJia 'Person Years forSe1ected Cutpoints and Four Exposure Estimates Median Nat Person risk years 248 427 508 552 890 1109 1126 1158 1321 1367 1388 1458 1501 1530 1596 1596 1600 1653 1710 1710 1710 1710 1710 1710 4555. 8314. 9503. 10570. 17516. 24528.3 25004. 26310. 34090. 36616. 38414. 40380. 42932. 44449. 46696. 46696. 46744. 49989. 52453. 52453. 52453. 52453. 52453. 52453. Rinsky Nat risk II."'~ ~f1.\ 1047 1135 1206 1206 1267 2.3. I lJ18J- 1439 1560 1597 1695 1706 1706 1710 1710 1710 1710 1710 1710 1710 1710 1710 1710 1710 1710 Person years 19252. 21232. 6 , 24737. 24794. 29534i87. ~ 39166. 44819.2 45830. 49610. 51648. 51648. 52453. 52453. 52453. 52453. 52453. 52453. 52453. 52453. 52453. 52453. 52453. 52453. Crump Nat Person risk years 267 480 554 636 993 1056 1204 1222 1264 1293 1293 1419 1465 1493 1568 1568 1568 1609 1609 1609 1609 1609 1616 1710 4894. 9326.3 10303. 12405. 20382. 23379. 27246. 28082.1 32482.4 34963. 34963. 38398. 40984.4 42012. 44593. 44593. 44593. 47064. 47064. 47064. 47064. 47064. 47516.3 52453. Paustenbach Person risk years 105 188 210 704 730 771 910 1021 1087 1088 1166 1267 1389 1457 1494 1548 1610 1610 1673 1681 1695 1710 795. 857.3 1073. 1878.1 2754. 10955. 12377.3 14995. 18196. 21433. 24950. 24976.3 26398. 29245.3 34725. 38417. 40672.4 43258. 45915. 45933. 46993. 50465.1 51483. 52453. related to risk can be examined. This method is useful for identifying " empirical thresholds " i. , critical con- centrations which are not associated with any excess risk in the database. 3. RESULTS Table I displays the aggregate number of workers and person-years exposed at or below selected cutpoints. Table I clearly shows that the estimates developed by Rinsky(l) are, on average, lower than those developed by Crump and Allen (9) which in turn are slightly lower than those developed by Paustenbach (IS) et al. Table II displays SMRs for total leukemia in Plio- film workers exposed at or below specific concentra- tions. For all sets of exposure estimates, Table II shows total leukemia SMRs are not different from 1.0 among those workers always exposed to benzene concentrations less than 20 ppm. However, for workers exposed to concentrations between 20 and 40 ppm, a nonstatistically significant elevation of total leukemia mortality is evi- dent for all estimates , except those of Paustenbach al. (IS) Table III displays the more relevantSMRs for AMML 4 by exposure concentration. These results indi- cate no AMMLs (except for the Crump and Allen(9) es- timates) in workers only exposed to concentrations of20 ppm and below. The Rinsky (1) estimates support et al. the notion of a critical concentration which must be at- tained for increased risk, in that exposures between 20- 25 ppm show a high SMR, while those below 20 ppm show no cases (SMR = 0, 95% C.I. = 0, 4.53). The median exposure estimates and the Crum~and Allen(9) exposure estimates are also indicative of this critical concentration in that exposures below 50 ppm do not result in excess risk (SMR = 0. , 95% C.I. = 0. 80), while those between 55 and 60 ppm do. ThePaus- tenbach(1S) exposure estimates, also show an increasing SMR trend for higher concentrations, but the estimates do not show markedly excess risks, until concentrations of 140 ppm are reached. Figure 1 graphically depicts the results for both the median exposure estimates and the Rinsky (I) ex- et at. posure estimates. This plot suggests that exposures under 4 In the Pliofi1m workers, there were no forms of ANLL other than acute myelocytic leukemia and acute monocytic leukemia. Following the CrumpCS) terminology, these cell types are referred to as AMML. GARABRANT/BERRYMAN 00334 GARABRANT/BERRYMAN 00335 GARABRANT/BERRYMAN 00336 GARABRANT/BERRYMAN 00337 GARABRANT/BERRYMAN 00338 GARABRANT/BERRYMAN 00339