Document gZ032M6wo5d6yNKrdjZLYgDq

;l' The Relationship between Low-level Benzene Exposure and leukemia in Canadian aim of this article is to study this lower portion of the: dosc-response curve for benzene and leukemia. Petn~leum Distribution Wor~ers Methods We identified cases from all workers in a ALo. rRrioebSer.tTShcohmnaptstoenr,,'2TMhoamrkaJs.WNi.cAolricmhs,'trAornngo,ld M. Katz, Wendy W. Huebner,' and Eileen D. Pearlman 1Exxon Biomedical Sciences, Inc., East Millstone, New Jersey; previously conducted cohort study (13) meeting the following crimia: a) died with an underlying cause of death of Ie:ukemia (1nr~rnarional Classification of DiSl:as~, codes 204-207); b) ~er worked in either 21mperial Oil Limited, Toronto, Ontario This stug.y was conducted to evaluate the relationship between leukemia occurrence and longterm, low-level benzene exposures in petroleum distribution workers. Fourteen cases were identified among a previously stUdied cohort (Schnaner et aI., Environ Health Perspect 101 (Suppl 6):85-99 (1993)1. Four controls per case were seleCted from the same cchort. controlling for birth year and time at rislr- Industrial hygienists estimated workplace exposures for benzene, without knowledge of case-control status. Average benzene concentrations ranged from 0.01 to 6.2 porn. Company medical records were used to abstract information on other potential confounders such as cigarene smoking. Odds ratios were calculated for several exposure metrics. Conditional 10\;lstic regression modeling was used to control for potential confounders. The risk of leukemia w::!s not associated with increasing C:Jmuiative exposure to benzene for these exposure levels. Duration of belizene exposure was more closely associated WITh leukemia risk than other the marketing/distribution. marine. or pipdine segments; and c) died betWeen 1964 and 1983, the stUdy end dare. Statistics Canada coded all de:ath certifi- cates for underlying cause of death. The: criteria resulted in 16 leuke..'nias. We were: unable to obtain reliable information on leukemia cdl types for the cases. We: sele:cted four controls for each case rrom records in the same cohort. Controls were restriCted to males. frequency matched by decade of birth. and were alive on or aITer the 'case s date: of death. After e..'\:cIudin~ exposure metrics, although results were not statistically significant. A family history of cancer and ';1arene smoking were ...xposure showing no the twO strongest riSk fa:::1:ors for leukemia. with cumulative additional risk when considered in the same models. This benzene study is cases and controls with inadequate work histories. there were 14 leukemia ca$es and 55 controls. consistent witn other data in tha;: and long-term. low-ievel benzene it was unable to exposures. The demonstrate power of the a relationship betWeen leukemia stUdy was limited. .Thus, further st:Jdy on celizene exposures in this concentration range are warranted. Environ Health The derails of the :xposure assessment strategy are described elsewh ere (J 4). Briefly. work histories were abstraCted ITom Perspect 104ISuppI6):1375-1379 (1996= hard copy personnel records for each case: Kev words: benzene, leukemia, petroleum and control and fOl"\varded. without easel control StatUS to industrial hygienistS. The industrial hygienists derived workplace Introduction c:x.posure estimates for benzene and tOtal hydrocarbons for every jobllocation/era Benzene has been classified as a known human carcinogen (1). Most investigators (2-5) base risk predictions for occupa- tional and environmental" exposures on a cohort of rubber hvdrochloride workers . (6). However, this ~ohort was exposed to high concli:ntrations of benzene. which sometimes exccede:d'time-weighted average concentrations of 50 ppm (,7,8). Petroleum distribution workers are expose:d to benzene while transferring gasoline and other petro- leum productS. These exposure levels are generally less than 1 ppm on an 8-he cirne- weighted average basis. Previous StUdies of these workers (9-12) have not e:'l:amined leukemia risk by benze:ne exposure. The combination. The process Started with site charaCterizations ror the 89 stUdy locations. including loading/unloading technology present at the sires. the types of materials handled. the typical tasks performed by workers, and typical environmental condi- tions such as average ambient temperatUres. Surveys were available: for some the sites and were supplemente:d with data on similar operations from outside the This paper was presented at Benzene '95: An Intemational CDnference on the Toxicity, Carcinogenesis. and Epidemiology of Benzene held 17-20 June 1995 in PiscaTaWay; New Jersey. Manuscript received 16 January 14June 1996., 1996: manuscript accepted Neill The author:; wish to thank G. Jorgensen and C. Miiano for computer programming assistance. A. O' . Vodarsik, and L Imperial Oil limited, Mackenzie provided data including A. Staynes and entrY and derical support. The S. Sotoudeh, were instrumental Oinclcoucpaatitniogn, aalbHsteraaCltntinDgi,vaisnidonpno:fr cmoamtespfat nfoyr tjoobd/leo:craivteio"nb/earsaesecxepnoasriuorseinesthtiework histories. Tho: industrial hygienists then applied adjustment faCtors. based on differences in environmental. operational. vidlng information from work. history and medical records. N. Murray and H. Siegel coorolnated Imperial Oil Limited Industrial Hygiene C!ssistance. P. Lalonde and M. Fair of Statistics Canada lcindly provided access to records from our previous study. allowing us to Obtain further information on cases and controls in a confiden- tial manner. We are particularly grateful to the Science Advisory Board who advised us on study methods throughout the project. including R. Shore of New York University (chair). R. Herrick and R. Ainsley of the ~ationallnstitutes of Occupational SafetY and Health, G. Sween of the University of Limburg (The Netherlands!. and G. Theriault of McGill University IQuebect. Finally, we are deeply indebted to W. Thar whose support and gCuNidAa2dn3dc5ree0ms,sEadcaeostrthrMeissisplltoUstndodynepeno,csNesJibto0le8.D8r7. 5A-.2R3.5S0.chTnealettpehro, nEex:x1o9n0B8)io8m73e-d8ic0a1l6S. cFiaexn:c(e9s0, 8I)nc8.7,3M-6e0tt0l9er.sER-moaadil,: ro be rt. schnatt e~ ere. e xxo n. s print. com taSk. and worksite conditions. The: values for the adjustment factors were estimated through ph~'Sical-chemical first principles. or empirical data. The validity of the estimating method was tested by comparing exposure c:stimates from the esrimacing procedure with resultS outfrom industrial hygiene surveys crried 1 u-.,."'" D..... GARABRANT/BERRYMAN 00340 . Vnl 104. C;unnlement 6 . December 1996 1375 .~,".~,;..,,;:.'. .' ..J SCH NA ITER ET AL.. during the relevant time period. On Tabla 1. Comparison of attributes for leukemia cases T abl a J. Leukemia risk !rf amdative eJCpJSIJI'e to average, estimates were within 22% o( the ~cnt.measured data. This was considered lI:3Sonable Eight- hour rime-weighted average exposure intensity estimates wen: assigned to each line in every worker's job/loClcion history. We subtraCted absentee informa- and c:antTOls. C1arac:teristic Age, at case s death Age, at first exposure Years exposed Cases 68.5 2B. 30.3 Contrals 68. 31. 2B. Benzene exposure, ppm-years O-il.17. 0.18-i1. 50- 9.- 219. No. exposed Odds cases ratio 9S%Q 1.00 5.06 . 0. 2.11 34-295 91-18.2 0.10-138 cion from time at work, and then multiplied the intensity estimates by length of rime in a job. These results were summed to arrive at a ppm-year estimate for r:very worker s career. Exposures for controls were only summed up tC the corresponding case s date of death. We also lagged exposures by 5, 10, and 15 years (15). This Strategy does not count exposures received 5, 10, or 15 years immediatdy prior to the ". case s date of death. The industrial hygien- ists also provided a ranked estimate for each line indicating the probability of der- \a1 e.'tposure to hydrocarbons. This was Tabla 2. Leukemia risk by potential c:anfounders. No. exposed Odds cases ratio 95%CI Socioeconomic job type Managerial/professional Clerk/technician Operator/driver Smoking status Never Ever Familial cancer Yes 1.00 03-3.10 0.41 07-z.33 1.00 1.00 2.51 51-13.3 G-O12b 01350-219. O-il.49c 50- 19. 20. 219. O-il.45d )-0.45-4 )-4 5-45 )-45 O-il.90d )-0.90- )-9. 99. )-99. 0.72-4g. 0.10-11.2 1.00 0-1.82 01-3. 09- 1.00 01- 0.16 0-1.32 1.47 16-13. 1.00 04-2.36 0.48 01- 1.03 02-20. kept as a separate index &om the estimated inhalation concentrations. Potential confounders were abstracted from company medical records, and included information on smoking habits, No. chest X-rays 10-14 15-19 20+ "Categorized according to quartiles. IlCategorized accord1.00 18-111 ing to tertiles. 'Categorized according to median . 75th 0.75 01-18. and 90th percentiles. 'tategorized according to regula1.73 02-156 tory considerations. hobbies, previous e.'tposures and occupa- tions, diagnostic radiation e.'tposure. and family history of cancer. The dependent CI. confidence interval. Cases were also exposed for a similar Table 4. Leukemia risk by altemate benzene exposure metrics. variable in all analyses was case/control number of years as controls. No. status. The primary independent variable Table 2 displays matched odds ratios of interest is cumulative benzene e.'tposure (ORs) for the 14 leukemia cases and 55 exposed Odds cases ratio 95%CI me:!sured in ppm-years. Other exposure controls according to potentially confound- Benzene intensity, characterizations. such as dermal exposure and average intensity of e.'tposure during the: entire work history. were also analyzed. We also categorized cumulative exposure in various ways to guard against a curpoint effeCt (16). We e:umined results according to the following schemes using the distrib- ution of exposures in the controls: the quartile distribution the tertile distribution ing variables. The tWO strongest risk rnCtors are a rnmily history of cancer (OR = 2. 51). and smoking (OR=-), although both have wide or noncalculable confidence intervals. The number of chest X-rays documented in medical records is not strongly related to leukemia risk. while the risk of leukemia is highest in managerial and professional job designations. Table 3 shows the risk of leukemia mean ppm )-0.0120-il.49 50- Maximum benzene Intensity, ppm " .c::0. 5-i1. 1.0+ Maximum probability 1.00 OB1.55 09- 1.00 1.02 21 -416 four categories split at the median, 75th. and 90th percentiles . ppm-years split at 0.45, 4. 5. and 45 ppm-years (the category boundaries according to cumulative exposure to ben- zene. None of the categorizations shows a monotonic trend for leukemia risk by cumulative exposure. although there are a of dermal exposure Law Medium High 1.00 0.12-2.51 04- correspond to 0. 01. 0. 1 ppm. and 1 small number of cases and controls in each ppm for 45 years) category. For cumulative benzene exposure, . ppm- years split at 0. 9, 9. 9, and 99 the highest leukemia risks are observed in e.'tposure group, although the confidence ppm-years (the category midpoints the second quartile (OR=5.06) and mid- intervals are e:mcmdy wide. t ; correspond to 0.01. 0. 1. and 1.0 ppm dle tertile (0 R = 4. 37). but the 0 Rs . Leukemia risk according to other expo- for 45 years). decre:a.se in the highest quarciles and terciIe. sure metries is shown in Table 4. Risk did ~ i Results When e."C:1mining the highest exposure not increase in a consistent way for the me:II1 categories in Table 3, cumulative benzene intensity over a worker s career, for workers The mean ages (at first exposure and last exposures greater than 5.5. 8, 20, 45, and ever exposed betWeen 0. 5 and 1 ppm or follow up) and number of years exposed. 99.9 (up to 220 ppm-years) result in ORs over 1 ppm. nor by a worker s highc:st are displayed in Table 1 for leukemi:! cases of 0.92, 2. 11, 0.96, 1.47. and 1.03, respec- ranked probability of dermal exposure. and controls. On aver:lge. the leukemia cases evely. All five of the categorizaeons suggest 0 Rs and p-values for coefficientS in the were three years younger than controls. risks consistent with unity for the highest logistic models were very similar when JI ii I~ 1376 Environmental Health Perspectives. Vol 104. Supplement 6 . December 1996 GARABRANT/BERRYMAN 00341 ~. (p=p= p:& p= (p.(.p LOW-U'JEL BENZENE AND LEUKEMIA exposures were lagged for 0, 5, 10, or 15 Therefore, we will report results only for no lag period. Cumulative benzene exposure did not show a Strong relationship with leukemia when regressed separately (OR= 1.002/ ppm-year. 77). The p-value for the score Statistic 76) indicates th;;.t this model does not fit the data well (fable 4, model 1). We also added into the model separate terms for an employee s mean exposure intensity and total exposure duration. This maneuver produced a noninter- pretable result; exposure intc:nsity (OR= 27/ppm) and duration (OR= 1.07/y=) Tabl. 50 Concfrtionallogisrlc regression modeling results for leukemia and benzene exposure. Model Model p-value 0.76 D.28 Variable Cumulative benzene exposure Cumulative benzene exposure Mean intensity Duration Intensity Duration Years at 0.5+ ppm Years at 1.0+ ppm Years at low dermal Years at medium dermal Years at high dermal Odds ratio 1.002 980 1271 1.069 171 1.061 1.015 004 1.061 1.054 022 95% Lower limit 0.989 933 426 990 729 985 940 921 984 954 895 95% Upper limit 015 1.030 12.098 1.155 1.880 1.144 1.095 1.094 1.144 1.164 1.167 Variable p-value 0.34 0.12 0.14 0.75 showed a positive relationship, yet the OR for cumulative exposure fell below 1.0 (Table 5). A model with only exposure duration showed a coefficient of 1.06/year exposed with a 95% confidence inter..aJ of 99 to 1.14 (Table 5, model 4) and resulted in a reasonable overall model value (p=0. 10). Thus, for these data, the . Based on score statistic. bBased on Wald chi-square statistic. Table 6. Conditional logistic modeling results for leukemia for cases and conuols with known values for potential confounders. Model Model p-value 0.11 Variable Family history of cancer Odds ratio 95'10 Lower limit . 95% Upper limit 56. Variable p-value simple measure of exposure duration was most closely associated with leukc:mia, Family history of cancer Ever smoked cigarettes 14. 1.04 188. 119. while cumulative benzene exposure and 1 intensity of benzene exposure did I. _'(plain leukemia risk. Next we examined whether exposure above a certain level was related to leukemia risk. by using the number of years worked above either 0. 5 or 1 ppm as independent variables. Neither of these variables e..,plained leukemia risk adequately nor fit the data well (fable 5. models 5. 6). The coefficient for years above 0. 5 ppm (1.02) was slighdy greater than the coefficient for years above 1 ppm (1.00). Family history of cancer Ever smoked cigarettes Cumulative benzene exposure Family history of cancer Ever smoked cigarettes Intensity Family history of cancer Ever smoked cigarettes Duration Family history of cancer Ever smoked cigarenes Duration Duration squared 11. 112 18. 15. 16. 13. 1.01 0.48 1.33 1.1 1 0.37 0.43 160. 100. 1.15 119. 115. 26. 265. 376. 1.22 244. 455. 1.57 1.02 0.49 The number of years spent in jobs .Based on score statistic. bBased on Wald chi-square statistic. ranked as having a low. medium, or high probability of dermal exposure was explored in one model. The risks were higher (and closer to significance) for a year spent in a low-probability job (OR = 1.06/year, 14) versus a high probability job (OR= 1.02/year, 76). Since the Mantel-Haenszel analyses showed that leukemia. risk did not increase for increasing catego ries of cumulative eJ!:posure, we constructed models with square terms for duration and intensity of exposure. This allows for nonexponential increases in risk per unit exposure. For all combinations of duration and intensity of exposure, and the squares of these variables, ~ -.odel with only duration of exposure ired fit the data best 05). The UK for this model (1.001/year ) was not quite Staristically significant = 0.07). Since both cigarette smoking and ;;. family history of cancer were related to leukemia in the Mantd-Hacnszd analvses. we con- StrUCted a series of models am"ong cases and controls for which we had known values for these variables. We then added cumulative exposure, exposure intensity, and years exposed to these models. Since these models are performed on a different set of cases and controls, the re:sulrs should not be com- pared to those in Table 5. Table 6 shows results from the model with only the pOten- ti2l confounders (an employee s family his- tory of cancer and whether he ever smoked cigarettes). This model produced high but unstable ORs (OR= 14.0 and 8. 9, respec- tively) for each variable. The score Statistic indicated that the model fit was Statistically significant 02). Next, we added different combinations of cumulative ben- zene exposure, mean intensity of benzene exposure, duration of exposure, and the square of the latter tWO variables. None the expanded models resulted in a better fit to the data, as measured by the score statis- tic. Adding cumulative bc:nzcne c:x:posure to this model resulted in an OR for b=ene of 97/ppm-yc:ir. and reduced the score test significance from 0. 02 to 0.06 (fable 6). Only when duration of exposure squared was added did the model again achieve sta- tistical significance (score test p-value = 04). All of these models resulted in ORs for cumulative benzene c:xposure and mean benzene intensity ofless than 1.0, and ORs for duration of exposure of greater than 1.0, but none of the c:x:posure variables were Statistically significant 025). Thus, these results show diaia family history of cancer' and cigarette smoking are the tWO srrongc::sr risk faCtors in these dara. Cumulative benzene exposure did not GARABRANT/BERRYMAN 00342 GARABRANT/BERRYMAN 00343 GARABRANT/BERRYMAN 00344