Document mBrROp34qpgLdMNbQBrM1gBnJ

Chemico-Biological Interactions 153154 (2005) 231237 Non-Hodgkin lymphoma and benzene exposure: A systematic literature review S.H. Lamm a,b,, A. Engel a, D.M. Byrd c a Consultants in Epidemiology and Occupational Health, LLC, Washington, DC, USA b Department of Health Policy and Management, Johns Hopkins University-Bloomberg School of Public Health, Baltimore, MD, USA c Consultants in Toxicology, Risk Assessment, and Product Safety, Washington, DC, USA Available online 10 May 2005 Abstract Epidemiological papers that address an association between benzene exposure and non-Hodgkin lymphoma (NHL) were identified and separated by study design. Eighteen studies contained 21 study groups--11 population-based casecontrol study groups, 3 nested occupation-based casecontrol study groups and 7 occupational benzene cohort study groups. Petroleum industry studies were not included. Only two of these 21 study groups showed statistically significant associations. However, these were among workers with multiple exposures across industries. Eleven of the 21 study groups presented ratios less than one, two equaled one, and eight were greater than one. Over all, about as many cases were observed (404) as expected (390.0) for an observed to expected ratio of 1.04 (0.941.14). After removal of the studies with multiple chemical exposures problems, the observed was 359 cases with 373.2 cases expected, yielding an odds ratio of 0.96 (0.861.06). Further assessment of an association with NHL should document the benzene exposure and separate out the contribution of non-benzene exposures. 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Benzene; Non-Hodgkin lymphoma; Casecontrol; Cohort; Pooled studies; Meta-analysis 1. Introduction A systematic review of the literature on benzene exposure and non-Hodgkin lymphoma (NHL) has been conducted. A search of the National Library of Medicine Pub-Med file has been regularly updated through April 15, 2005. Newly identified studies, and those in our previous review [1], were searched to Corresponding author. E-mail address: Steve@CEOH.com (S.H. Lamm). find additional references. There is a large literature of petroleum industry studies that have previously received extensive analysis [2] and within which no increased risk of NHL has been observed. In order to more clearly evaluate the rest of the literature, the petroleum industry studies have not been included in this analysis. Care has been taken to include only those studies that specified benzene exposure rather than exposure to solvents including benzene. Eighteen studies with 21 study groups met this initial criterion for inclusion. 0009-2797/$ see front matter 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cbi.2005.03.027 232 S.H. Lamm et al. / Chemico-Biological Interactions 153154 (2005) 231237 Table 1 Study characteristics of population-based casecontrol studies of benzene exposure and non-Hodgkin lymphoma Authors Site Population/ Dates industry Exposure source Bernard et al. [3] Blair et al. [4] Fabbro-Peray et al. [5] Gerin et al. [6], Fritschi and Siemiatycki [7] Mao et al. [8] Persson and Fredrikson [9] Scherr et al. [10] Yorkshire, United Kingdom Iowa, Minnesota agriculture Languedoc-Roussillon, France Montreal, Canada Eight Provinces, Canada Orebro and Linkoping, Sweden Metropolitan Boston Population Population (WM) Population Population (Males) Population Population Population 19791981 19801983 19921996 19791986 19941997 19641986, 19751984 19801982 Interviews Interviews Interviews Interviews Mailed Questionnaires Questionnaires Interviews Outcome ascertainment Registry, physician/lab surveillance Registry (IA), hospital surveillance Hospital records Hospital surveillance Provincial cancer registries Hospital registry Hospital records Schumacher and Delzell North Carolina [11] Population 1968970, 19751977, Death 19801982 certificates Death certificates 2. Population-based casecontrol studies Table 1 shows the eight population-based casecontrol studies [311] by study characteristics and Table 2 shows them by study results. They contain odds ratios yielding a pooled odds ratio of 1.00 (95% confidence limits, 0.891.11) based on 325 cases observed and 326.1 cases expected. These odds ratios range between 0.5 and 2.0 with seven of the eleven being between 0.8 and 1.2. The set of 11 population-based study groups from the 8 studies did not show evidence of an associated with benzene exposure (Fig. 1). Statistical significance is seen in only one association between benzene exposure and NHL [5], with an odds ratio = 2.0 and 95% confidence limits of 1.13.9. However, this association related to an aggregated benzene exposure group of workers in chemical or rubber production and of workers employed as printers, painters, and shoemakers. There is no analysis of Table 2 Study results of population-based casecontrol studies of benzene exposure and non-Hodgkin lymphoma Authors Nn Expected OR Bernard et al. [3] Blair et al. [4] Fabbro-Peray et al. [5] Gerin et al. [6], Fritschi and Siemiatycki [7] Mao et al. [8] Persson and Fredrikson [9] Scherr et al. [10] Schumacher and Delzell [11] Sum Excluding Fabbro-Peray [5] 158 662 445 215 764 (M) 705 (F) 199 303 433 (W) 68 (B) 153 22 9 (M/H) 19 (L) 187 (none) 36 (M) 5 (F) 3 12 56 (W) 10 (B) 325 303 139.1 11 11.3 (M/H) 31.7 (L) 187 (none) 30 (M) 6.3 (F) 3.8 10 72.3 (W) 10.6 (B) 326.1 315.1 0.49 1.1 2 0.8 (M/H) 0.6 (L) 1.0 (none) 1.2 (M) 0.8 (F) 0.8 1.2 0.77 (W) 0.94 (B) 1.00 0.96 95% Confidence limits 0.212.00 0.91.4 1.13.9 0.41.6 (M/H) 0.41.0 (L) 0.81.9 (M) 0.21.8 (F) 0.13.8 0.52.6 0.561.07 (W) 0.582.30 (B) 0.891.11 0.851.07 S.H. Lamm et al. / Chemico-Biological Interactions 153154 (2005) 231237 233 Fig. 1. Population-based casecontrol studies odds ratios of non-Hodgkin lymphoma for benzene exposure (OR and 95% confidence limits; individually and pooled). whether the risks are similar in the various industrial groups or whether they cluster. This study contributed only 7% of the observed cases and had little effect on the over-all risk estimate. The pooled odd ratio with its exclusion was 0.96 with 95% confidence limits of 0.851.07. Many studies that do not show an association with benzene had identified other significant occupational or non-occupational exposure factors. 3. Occupation-based casecontrol studies None of three nested occupation-based casecontrol studies (Table 3) have a significant odds ratio with benzene exposure (Table 4). The Greenland et al. [12,13] and Ott et al. [14] studies find as many exposed cases as expected. The Wilcosky et al. [15] study of rubber workers found significant associations for lymphosarcoma with carbon tetrachloride, carbon disulfide, hexane and xylenes but not for benzene, a common co-exposure in solvent mixtures. These confounding exposures lend difficulty to attributing the observations to benzene exposure. 4. Occupation-based cohort studies Five of the seven occupational benzene cohorts (Table 3) have standardized mortality ratios (SMR) of 0.891.06 (Table 4), one [20] has no exposed case and the seventh [27] (the Chinese benzene cohort) has an SMR of 4.5 (1.38.4). Controversy exists about the variety and levels of exposures in that cohort as well as underascertainment of NHL cases in the comparison group [28]. Unlike various studies of rubber workers, most of the Chinese benzene cohort exposures have not been reported with examinations of co-exposures and other exposures. In the Chinese benzene cohort, the industrial sectors with the greatest number of cases [29] were chemical, other/mixed exposure and coatings (painters and other coating application workers). Analysis assessing similarity of risk across industrial groups has not been done. Until the contribution of benzene exposure, or the risk to persons with significant exposure only to benzene, is separately reported, the import of these data for the assessment of risk from benzene exposure cannot be determined. Fur- 234 S.H. Lamm et al. / Chemico-Biological Interactions 153154 (2005) 231237 Table 3 Study characteristics of occupation-based casecontrol and cohort studies of benzene exposure and non-Hodgkin lymphoma Authors Site Population/industry Dates Exposure source Outcome sourcea Occupation-based casecontrol studies Greenland et al. [12], Greenland Pittsfield, [13] Massachusetts Ott et al. [14] Union Carbide Wilcosky et al. [15] Akron, Ohio Occupational cohort studies Bloemen et al. [16], Bond et al. [17], Ott et al. [18] Dow Chemical Michigan Collins et al. [19] Decouffle et al. [20] Monsanto Sauget, Illinois Conoco plant Rinsky et al. [21], Paxton et al. [22] Seniori Costantini et al. [23,24], Fu et al. [25] Sorahan et al. [26] Goodyear plants, OH, USA Florence, Italy UK Yin et al. [27] China Transformer assembly plant (WM) Chemical Plant Research and Development Rubber manufacturing plant 19691984 19401978 19641973 Plant manager interviews, work history records Work history records Work history records R&D, chemical manufacturing plants (WM) Chemical plant Chemical manufacturing Pliofilm manufacture Shoemakers Benzene factories Benzene industries 19401996 Work history records 19401997 19501977 19401996 19501999 19682002 19721987 Work history records Hire-fire work records Work history records Work history records Inspectory of factories registry Salary and administrative records Compensation records, DCs DCs DCs DCs DCs DCs DCs DCs Office of national statistics Employee medical record, DCs a DC, death certificate. ther, the reported doseresponse curves for benzene in units of either ppm or ppm-yrs are U-shaped, not monotonically increasing, which also needs an explanation. The data from the Chinese study is further aberrant, since its mortality data for malignant lymphoma includes both NHL and Hodgkin lymphoma. In contrast, the UK study [26] has clearly distinguished the two. Our analysis of their NHL cases by industrial group (personal communication) indicates similar risks across the various industrial groups without clustering in certain industries or among the industries where most measured plants had the highest exposures (i.e., 25 ppm). 5. Discussion Most studies of benzene exposure and NHL found no association. Occasional study groups (2 of 21) exhibit an association between benzene and NHL, but these present analyses of association with benzene without considering concurrent solvent exposures and were in populations with mixed chemical exposures. Across all study designs, we found that investigators observed a total of 404 cases with 390.0 cases expected for a pooled odd ratio of 1.04 (0.941.14) (Table 5). The observed/expected ratio is 1.04 with 95% confidence limits of 0.931.14. Exclusion of the multiexposure study groups yield a set of 17 study groups with 359 cases with an expected of 373.2 cases, for a pooled odds ratio of 0.96 (0.861.07). In toto, the studies do not demonstrate an increased risk of NHL from benzene exposure. Clarification of the contribution of non-benzene exposures to the risk estimates are necessary before a causal inference can be considered. Refined analysis of any association between benzene exposure and NHL will require validation of the benzene exposure and careful assessment of alternative and concurrent exposures before a different judgment can be rendered. Studies from the rubber industry in the 1980s had initially pointed out an association between benzene and chronic lymphocytic leukemia. However, S.H. Lamm et al. / Chemico-Biological Interactions 153154 (2005) 231237 235 Table 4 Study results of occupation-based casecontrol and cohort studies of benzene exposure and non-Hodgkin lymphoma Authors Cases Observed Expected OR Occupation-based casecontrol studies Greenland et al. [12], Greenland [13] Ott et al. [14] Wilcosky et al. [15] Sum Excluding Wilcosky [15] 15 LS and RCS 52 NHL deaths among 29,139 male employees 9 NHL in WMs of 6678 total employees 19641973 6 5 6 17 11 6 1.00 5.0 1.0 2.0 3.0 13 1.31 11 1.00 95% Confidence limits 0.224.53 0.22.3 ns 0.762.09 0.501.79 Occupational cohort studies Bloemen et al. [16], Bond et al. [17], Ott et al. [18] Collins et al. [19] Decouffle et al. [20] Rinsky et al. [21], Paxton et al. [22] Seniori Costantini et al. [23,24], Fu et al. [25] Sorahan et al. [26] Yin et al. [27] Sum Excluding Yin [27] a Includes Hodgkin lymphoma. 2266 employees > 1 month in benzene 4417 employees hired 19401977 259 men in plant 19471960 1845 employees in plants, 19361996 2008 employees after 1938 and during 19501999 5514 workers exposed 19661967 or earlier 74,828 employees, 19721987 10 11 0 5 4 15 17a 62 45 9.4 12.4 0.36 5.19 3.84 15.9 3.78 50.9 47.1 1.06 0.511.79 0.89 0.441.59 0 0.96 0.312.25 1.04 0.282.67 0.94 0.531.56 4.5 1.328.4 1.22 0.931.56 0.96 0.701.28 when the work histories were developed to show the exposures to each of a wide range of solvents, significant associations were found for carbon tetrachloride and carbon disulfide, but not for benzene [30]. Similar analyses should be conducted for other studies in which an association between benzene exposure and NHL is suggested. A particular powerful observation is found in the report by Rinsky et al. (2002), which is the latest followup report of the Ohio Pliofilm cohort study. This study has been the primary basis for most quantitative risk analyses of the carcinogenic potency for benzene. It has had the most intensive examination and re-examination of its exposure database and is the study for which most Table 5 Pooled summary analysis of casecontrol and cohort studies of benzene exposure and non-Hodgkin lymphoma across study designs, with and without multi-exposure exclusions Source Design Observed Expected OR 95% Confidence limits All studies Population-based Occupation-based Occupation-based Casecontrol Casecontrol Cohort 325 17 62 326.1 13 50.89 1.00 0.891.11 1.31 0.762.09 1.22 0.941.56 Sum 404 390.0 1.04 0.941.14 With exclusions Population-based Occupation-based Occupation-based Casecontrol Casecontrol Cohort Sum 303 11 45 359 315.1 11 47.1 373.2 0.96 0.851.07 1.00 0.501.79 0.96 0.701.29 0.96 0.861.06 236 S.H. Lamm et al. / Chemico-Biological Interactions 153154 (2005) 231237 observers accept that benzene was the only significant solvent exposure. While there is considerable disagreement as to how best to handle the exposure and population assumptions, there is little disagreement about the outcome data in this study. By the time of this report, all living cohort members had had at least 20 years of follow-up. The observation period has now been sufficient for the expectation of more than five NHL deaths, and five have been observed. In isolation, this study [21] yields a risk estimate of 0.96 (0.312.25) and the UK study [26] yields a risk estimate of 0.94 (0.531.56). When the results of the 18 study groups without known multi-exposure/diagnostic issues are brought into the analysis, the pooled odds ratio is 0.96 (0.861.06). Pooling of the results from all 18 of these study groups has not changed the point estimate but has yielded a marked narrowing of the confidence limits of that point estimate from 0.312.25 to 0.861.06. This procedure demonstrates that the inclusion of the data from the additional relevant study groups has greatly reduced the range of uncertainty in the risk estimate assessing the association hypothesis. References [1] S.H. Lamm, G.D. Kirk, Benzene and non-Hodgkin's lymphoma--an analysis of the literature, 1998, in: Annual Meeting of the American Association of Cancer Research, New Orleans, LA, 1998. [2] O. Wong, G.K. Raabe, Non-Hodgkin's lymphoma and exposure to benzene in a multinational cohort of more than 308,000 petroleum workers, 19371996, J. Occup. Environ. Med. 42 (2000) 554568. [3] S.M. Bernard, R.A. Cartwright, C.C. Bird, I.D. Richards, I. Lauder, B.E. Roberts, Aetiologic factors in lymphoid malignancies: a casecontrol epidemiological study, Leuk. Res. 8 (1984) 681689. [4] A. Blair, A. Linos, P.A. Stewart, L.F. Burmeister, R. Gibson, G. Everett, L. Schuman, K.P. Cantor, Evaluation of risks for nonHodgkin's lymphoma by occupation and industry exposures from a casecontrol study, Am. J. Ind. Med. 23 (1993) 301312. [5] P. Fabbro-Peray, J.P. Daures, J.F. Rossi, Environmental risk factors for non-Hodgkin's lymphoma: a population-based casecontrol study in Languedoc-Roussillon, France, Cancer Causes Control 12 (2001) 201212. [6] M. Gerin, J. Siemiatycki, M. Desy, D. Krewski, Associations between several sites of cancer and occupational exposure to benzene, toluene, xylene, and styrene: results of a casecontrol study in Montreal, Am. J. Ind. Med. 34 (1998) 144 156. [7] L. Fritschi, J. Siemiatycki, Lymphoma, myeloma and occupation: results of a casecontrol study, Int. J. Cancer 67 (1996) 498503. [8] Y. Mao, A.M. Hu, K. Ugnat, White, for the Canadian Cancer Registries Epidemiology Research Group, Non-Hodgkin's lymphoma and occupational exposure to chemicals in Canada, Ann. Oncol. 11 (Suppl. 1) (2000) S69S73. [9] B. Persson, M. Fredrikson, Some risk factors for non-Hodgkin's lymphoma, Int. J. Occup. Environ. Health 12 (1999) 135 142. [10] P.A. Scherr, G.B. Hutchison, R.S. Neiman, Non-Hodgkin's lymphoma and occupational exposure, Cancer Res. 52 (1992) 5425s5574s. [11] M.C. Schumacher, E. Delzell, A death-certificate casecontrol study of non-Hodgkin's lymphoma and occupation in men in North Carolina, Am. J. Ind. Med. 13 (1988) 317330. [12] S. Greenland, A. Salvan, D.H. Wegman, M.F. Hallock, T.J. Smith, A casecontrol study of cancer mortality at a transformer-assembly facility, Int. Arch. Occup. Environ. Health 66 (1994) 4954. [13] S. Greenland, A semi-Bayes approach to an analysis of correlated multiple associations, and an application to an occupational mortality study, Stat. Med. 11 (1992) 219230. [14] G.M. Ott, M.J. Teta, H.L. Greenberg, Lymphatic and hematopoietic tissue cancer in a chemical manufacturing environment, Am. J. Ind. Med. 16 (1989) 631643. [15] T.C. Wilcosky, H. Checkoway, E.G. Marshall, H.A. Tyroler, Cancer mortality and solvent exposures in the rubber industry, Am. Ind. Hyg. Assoc. J. 45 (1984) 809811. [16] L.J. Bloemen, A. Youk, T.D. Bradley, K.M. Bodner, G. Marsh, Lymphohaematopoietic cancer risk among chemical workers exposed to benzene, Occup. Environ. Med. 61 (2004) 270274. [17] G.G. Bond, E.A. McLaren, C.L. Baldwin, R.R. Cook, An update of mortality among chemical workers exposed to benzene, Br. J. Ind. Med. 43 (1986) 685691. [18] M.G. Ott, J.C. Townsend, W.A. Fishbeck, R.A. Langner, Mortality among individuals occupationally exposed to benzene, Arch. Environ. Health 33 (1978) 310. [19] J.J. Collins, B. Ireland, C.F. Buckley, D. Shepperly, Lymphohaematopoietic cancer mortality among workers with benzene exposure, Occup. Environ. Med. 60 (2003) 676679. [20] P. Decoufle, W.A. Blattner, A. Blair, Mortality among chemical workers exposed to benzene and other agents, Environ. Res. 30 (1983) 1625. [21] R.A. Rinsky, R.W. Hornung, S.R. Silver, C.Y. Tseng, Benzene exposure and hematopoietic mortality: a long-term epidemiological risk assessment, Am. J. Ind. Med. 42 (2002) 474480. [22] M.B. Paxton, Leukemia risk associated with benzene exposure in the pliofilm cohort, Environ. Health Perspect. 104 (1996) 14311436. [23] A. Seniori Costantini, M. Quinn, D. Consonni, M. Zappa, Exposure to benzene and risk of leukemia among shoe factory workers, Scand. J. Work Environ. Health 29 (2003) 5159. [24] A. Seniori Costantini, personal communication, September 22, 2004. [25] H. Fu, P.A. Demers, A.S. Costantini, P. Winter, D. Colin, M. Kogevinas, P. Boffeta, Cancer mortality among shoe manufac- S.H. Lamm et al. / Chemico-Biological Interactions 153154 (2005) 231237 237 turing workers: an analysis of two cohorts, Occup. Environ. Med. 53 (1996) 394398. [26] T. Sorahan, L.J. Kinlen, R. Doll, Cancer risks in a historical UK cohort of benzene exposed workers, Occup. Environ. Med. 62 (4) (April 2005) 231236. [27] S.N. Yin, R.B. Hayes, M.S. Linet, G.L. Li, M. Dosemeci, L.B. Travis, C.Y. Li, Z.N. Zhang, D.G. Li, W.H. Chow, S. Wacholder, Y.Z. Wang, Z.L. Jiang, T.R. Lai, W.Y. Zhang, X.J. Chao, P.Z. Ye, Q.R. Kou, X.C. Zhang, X.F. Lin, J.F. Meng, C.Y. Ding, J.S. Zho, W.J. Blot, A cohort study of cancer among benzeneexposed workers in China: overall results, Am. J. Ind. Med. 29 (1996) 227235. [28] O. Wong, A critique of the exposure assessment in the epidemiologic study of benzene-exposed workers in China conducted by the Chinese Academy of Preventive Medicine and the US National Cancer Institute, Reg. Toxicol. Pharmacol. 30 (1999) 259267. [29] R.B. Hayes, S.N. Yin, M. Dosemeci, L.G. Li, S. Wacholder, L.B. Travis, C.Y. Li, N. Rothman, R.N. Hoover, M.S. Linet, Benzene and the dose-related incidence of hematologic neoplasms in China, J. Natl. Cancer Inst. 89 (1997) 10651071. [30] H. Checkoway, T. Wilcosky, P. Wolf, H. Tyroler, An evaluation of the associations of leukemia and rubber industry solvent exposures, Am. J. Ind. Med. 5 (1984) 239249.