Document xrK3BOjZM63qDweJZppBQxXE

Chemico-Biological Interactions 153154 (2005) 103109 Protein adducts as biomarkers of human benzene metabolism Stephen M. Rappaport a,, Suramya Waidyanatha a, Karen Yeowell-O'Connell a, Nathaniel Rothman b, Martyn T. Smith c, Luoping Zhang c, Qingshan Qu d, Roy Shore d, Guilan Li e, Songnian Yin e a School of Public Health, University of North Carolina, Chapel Hill, NC 27599-7431, USA b Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, MD, USA c School of Public Health, University of California, Berkeley, CA, USA d Nelson Institute of Environmental Medicine, New York University School of Medicine, Tuxedo, NY, USA e Institute for Occupational Medicine, Beijing, China Available online 9 April 2005 Abstract We used cysteinyl adducts of serum albumin (Alb) to investigate the production of two reactive benzene metabolites, namely, benzene oxide (BO) and 1,4-benzoquinone (1,4-BQ) in workers exposed to benzene. Adducts were measured in 160 benzeneexposed workers who did not use respiratory protection (based upon individual geometric mean benzene exposure levels: median = 5.27 ppm, interquartile range = 2.1413.4 ppm, range = 0.074328 ppm) and 101 local controls, from populations in Shanghai and Tianjin, China. After isolation of Alb, these adducts (designated as BO-Alb and 1,4-BQ-Alb) were cleaved from the protein with methanesulfonic acid and trifluoroacetic anhydride and measured by gas chromatography-mass spectrometry. Although BO-Alb and 1,4-BQ-Alb were measured in all subjects, levels of both adducts were 2.4-fold greater (median value) in exposed subjects than in controls (interquartile-fold range = 1.634.05 for BO-Alb and 1.643.69 for 1,4-BQ-Alb). Loglog plots of the individual adduct levels versus exposure were quasi-linear with straight-line slopes of about 0.3 for both BO-Alb and 1,4-BQ-Alb. Since these log-space slopes were significantly less than one, we infer that adduct production was nonlinear, i.e., less-than proportional to benzene exposure, over the indicated range. This behavior points to saturation of CYP2E1 as a critical metabolic consequence of high exposure to benzene in humans. Thus, the biologically effective dose of BO and 1,4-BQ should be proportionally greater in persons exposed to low rather than high levels of benzene. 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Benzene; Protein adducts; Albumin; Benzene oxide; Benzoquinone; Nonlinear kinetics 1. Introduction Corresponding author. Tel.: +1 919 966 5017; fax: +1 919 966 0521. E-mail address: smr@unc.edu (S.M. Rappaport). Benzene is carcinogenic in both experimental animals and humans, presumably due to toxic effects in the bone marrow arising from its metabolism [1,2]. Although benzene is the simplest aromatic compound, 0009-2797/$ see front matter 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cbi.2005.03.014 104 S.M. Rappaport et al. / Chemico-Biological Interactions 153154 (2005) 103109 its metabolism is surprisingly complex. Following CYP2E1 oxidation of benzene to benzene oxide (BO) [3,4], a series of enzymatic and nonenzymatic steps lead to many other metabolites, notably, phenol, catechol, hydroquinone, and the ring-opened muconaldehydes (ultimately transformed to t,t-muconic acid). Portions of catechol and hydroquinone are oxidized to 1,2- and 1,4-benzoquinone, respectively, which are electrophilic and capable of reacting with DNA and other critical macromolecules, as are BO and the muconaldehydes. The toxicity of benzene is thought to involve one or more of these electrophilic metabolites (most notably 1,4-BQ) and/or reactive oxygen species, produced by redox cycling of catechol, hydroquinone and the respective benzoquinones [1,2]. Human metabolism of benzene has largely been inferred from in vitro experiments [510], by measurements of urinary metabolites in persons following occupational exposure [1116], and by physiologically based toxicokinetic models [1720]. While undoubtedly useful, these methods provide relatively little information about the biologically effective doses of the electrophilic benzene metabolites, whose residence times are very short in vivo (a few minutes at most). This motivated us to apply stable adducts of benzene's electrophilic intermediates, i.e., BO and 1,4benzoquinone (1,4-BQ) with Hb and/or albumin (Alb), in the blood of benzene-exposed workers as measures of the biologically effective dose [2124]. Our studies of protein adducts relied upon populations of workers, exposed to a wide range of airborne benzene levels, and concurrent controls in Shanghai [2123] and Tianjin, China [24]. We previously reported exposureadduct relationships in these two populations, focusing primarily upon the Alb adducts, which were more abundant than those of Hb in our samples. In both cases, significant departures from linear kinetics were observed, suggesting saturation of benzene metabolism [23,24]. In the current investigation we combine results of the Alb adducts from the two studies to further scrutinize 160 subjects who were exposed to benzene from less than 1 ppm to greater than 300 ppm. Since significant and varying background levels of BO-Alb and 1,4-BQ-Alb were observed among 101 control subjects in these studies, we express adduct levels of exposed workers as fold increases above concurrent control values prior to examining the exposureadduct relationships. Results in- dicate that subjects were exposed over the full range of kinetic processes from linear (first-order) metabolism at low exposures to essentially saturated metabolism at high exposures. 2. Measurements of benzene exposure and adducts The recruitment of subjects and analytical methods have been described in detail elsewhere [2126]. Briefly, the first study was sponsored by the National Cancer Institute (hereafter the `NCI study'). From the NCI study, adduct measurements were available from 43 exposed workers and 50 controls of both sexes from Shanghai, China. Of the exposed workers, 16 wore respirators and were excluded from the current analyses. Each exposed worker had 5 or 6 personal full-shift air measurements of benzene during the week prior to blood collection (using organic vapor monitors and gas chromatography). The geometric mean (GM) of these individual personal measurements was used as the measure of individual benzene exposure. The median level of benzene exposure (individual GM value) among the 27 workers from the NCI study (who did not use respirators) was 17.2 ppm (interquartile range: 13.592.9 ppm, range: 1.65328 ppm). The second study was sponsored by the Health Effects Institute (hereafter the `HEI study'). From the HEI study, adduct measurements were available from 133 exposed workers and 51 controls of both sexes from Tianjin, China. Each exposed subject had between 1 and 6 personal fullshift measurements (111 subjects had 4 measurements) of benzene exposure, obtained weekly during the month prior to blood collection (also measured with organic vapor monitors and gas chromatography). Again, the individual GM was used as the measure of benzene exposure. The median level of benzene exposure among the 133 exposed workers from the HEI study was 3.70 ppm (interquartile range: 2.0310.0 ppm, range: 0.07446.6 ppm). Combining results from the NCI and HEI studies, the median level of benzene exposure among the 160 exposed workers was 5.27 ppm (interquartile range: 2.1413.4 ppm, range: 0.074328 ppm). Adducts of serum Alb were measured with an assay, developed in our laboratory, which selectively cleaved S.M. Rappaport et al. / Chemico-Biological Interactions 153154 (2005) 103109 105 Fig. 1. Scheme for analysis of albumin adducts of benzene oxide (BO-Alb) and 1,4-benzoquinone (1,4-BQ-Alb). (GCNICIMS refers to gas chromatographymass spectrometry with negative ion chemical ionization). cysteinyl adducts from the protein and converted BOAlb and 1,4-BQ-Alb into volatile derivatives, suitable for gas chromatography-mass spectrometry. As shown in Fig. 1, these analytes were generated by reacting 35 mg of dried Alb with trifluoroacetic anhydride and methanesulfonic acid. All adduct levels were based upon a single blood specimen from each subject. The precision of the assay was estimated from coefficients of variation of 0.30 for BO-Alb and 0.20 for 1,4-BQAlb [24]. Table 1 summarizes the adduct levels for exposed and control subjects in the two studies. Among control subjects, levels of BO-Alb were significantly greater in the HEI study (median = 174 pmol/g Alb) than in the NCI study (median = 115 pmol/g Alb) while levels of 1,4-BQ-Alb were significantly greater in the NCI study (median = 2270 pmol/g Alb) than in the HEI study (median = 994 pmol/g Alb) (p < 0.0001 in both cases by the Wilcoxon test). While the reasons for these differences have not been fully elucidated, the much larger control values of 1,4-BQ-Alb in the NCI study can be largely attributed to collection of blood in EDTA rather than in citrate, as for the HEI study (based upon unpublished experiments in our laboratory). Other differences in background adducts were probably not greatly influenced by environmental exposures to benzene, but rather to smoking and unexplained dietary and endogenous factors that contributed phenol, hydroquinone, and other precursor molecules in vivo [21,22,24,27]. To adjust for differences in background adduct levels across studies, we express each exposed subject's adduct level as a fold increase above the control value. That is, the Relative Adduct Level (RAL) of the ith exposed worker in the hth study is given by RALhi = Ahi/Ch, where Ahi is the observed adduct level for that worker and Ch is the estimated GM adduct level of control workers in that study. In our samples, Ch = 115 and 169 pmol/g Alb for BO-Alb, and Ch = 2490 and 1030 pmol/g Alb for 1,4-BQ-Alb in studies 1 (NCI) and 2 (HEI), respectively. The last three columns of Table 1 show the median RALhi values, along with the interquartile ranges and ranges for the NCI and HEI studies. Interestingly, while the median level of BO-Alb was significantly greater for the NCI study (median = 3.82-fold) than the HEI study (median = 2.19-fold) (p = 0.0003, Wilcoxon test) that for 1,4-BQ-Alb was not (median = 2.34-fold for the NCI study versus 2.42 for the HEI study) (p = 0.933, Wilcoxon test). With data combined from both the NCI and HEI studies, the median value of RALhi was 2.4fold for both adducts (interquartile range = 1.634.05 for BO-Alb and 1.643.69 for 1,4-BQ-Alb). 3. Exposureadduct relationships Scatter plots showing RALhi values for BO-Alb and BQ-Alb versus benzene exposure are given in Fig. 2 1.3919.5 0.9696.41 0.59144.1 0.25810.1 Range 2.806.57 1.643.34 1.573.57 1.553.89 Median IQRa Exposed workers Relative adduct level (fold increase above controls) 15904000 26510400 2.42 1007465 2.19 41008320 241016000 2.34 1602240 3.82 Range 266605 323757 Median IQRa 7091540 4272570 133 2490 82.0544 133 371 27 5830 16303610 9609410 91.9158 44.3250 27 440 Exposed workers n Table 1 Levels of albumin adducts of benzene oxide (BO-Alb) and 1,4-benzoquinone (1,4-BQ-Alb) in the serum of workers from the NCI and HEI studies 106 S.M. Rappaport et al. / Chemico-Biological Interactions 153154 (2005) 103109 with different symbols representing the NCI and HEI subjects. The data appear to be consistent across studies, as indicated by the superposition of observations in the region of overlapping benzene exposure, between about 2 and 30 ppm. The scatter plots also indicate that the loglog relationships between relative adduct levels and benzene exposure are reasonably linear over the indicated range of exposures. After combining data from both studies, the (natural) logged RALhi values for the two adducts were regressed upon the logged benzene exposure levels as shown in Fig. 3A. Since the outcome variable RALhi = Ahi/Ch is the ratio of the adduct level of the ith subject from group h to the GM adduct level for group h, note that these straight-line models were regressing [ln(Ahi) - ln(Ch)], representing the logged individual adduct level minus the mean of the logged control adduct levels, on ln(Benzene Exposure). For both adducts, the fits to the straight-line model were reason- Fig. 2. Scatter plots of relative adduct levels (RAL, fold increase above control values) vs. benzene exposure for 160 workers from the NCI () and HEI (open symbol) studies. (A) Benzene oxidealbumin adducts; (B) 1,4-benzoquinonealbumin adducts. Range 124204 Median IQRa 1,4-BQ-Alb 51 994 51 174 1,4-BQ-Alb 50 2270 50 115 Control workers Adduct level (pmol/g Alb) n a Interquartile range. BO-Alb BO-Alb Study Adduct NCI NCI HEI HEI S.M. Rappaport et al. / Chemico-Biological Interactions 153154 (2005) 103109 107 about 50 ppm of benzene. Although the initial point of departure from linear kinetics cannot be inferred from Fig. 3B, we previously estimated this point to be about 1 ppm, based upon the HEI data, which provide most of the low-exposed subjects in the current analysis [24]. Fig. 3. Relative adduct levels (RAL, fold increase above control values) vs. benzene exposure for benzene oxide-albumin adducts () and 1,4-benzoquinonealbumin adducts (open symbol) from 160 workers combined from the NCI and HEI studies. Lines represent the predictions from least-squares regression of the natural logarithm of RAL on the natural logarithm of benzene exposure. (A) Log-space relationships; (B) natural-space relationships. able with slopes of about 0.3 (BO-Alb: slope = 0.289, S.E. = 0.024, R2 = 0.480; 1,4-BQ-Alb: slope = 0.312, S.E. = 0.026, R2 = 0.485). Since these logspace slopes were significantly less than one (upper 95% confidence limits were 0.337 and 0.362 for BO-Alb and 1,4-BQAlb, respectively), we infer that the corresponding natural-space relationships were substantially nonlinear with concave-downwards shapes. This is illustrated in Fig. 3B, which shows the RALhi values plotted versus benzene exposure in natural-space along with the predicted values from the log-space regressions. Despite the large variability, apparent for both adducts at a given benzene exposure, the exposureadduct relationships reached plateaus at higher benzene levels, suggesting that CYP2E1 was substantially saturated at 4. Implications for health risk As noted in the introduction, protein adducts offer the opportunity to study the disposition of reactive electrophiles that cannot readily be measured in vivo. In this application, we used cysteinyl adducts of serum Alb to evaluate the production of two transient benzene metabolites, i.e., BO and 1,4-BQ over a wide range of benzene exposures (0.07328 ppm). We previously reported that BO-Alb was chemically stable in vivo, being eliminated with a presumed half time of 21 days in humans, and that 1,4-BQ-Alb was moderately stable, with an estimated elimination half time of 13.5 days in humans [24]. Thus, these adducts reflect the integration of exposure over 3 or 4 weeks prior to blood collection and are relatively insensitive to day-to-day variability in exposure that can obscure exposurebiomarker relationships [28]. The relevance of BO-Alb and 1,4-BQ-Alb to human health risk should be considered separately. Since BO is formed during the initial CYP2E1 oxidation of benzene, BO-Alb represents a global measure of benzene metabolism and, thus, should be unequivocally relevant to health risk, insofar as metabolism appears to be a necessary condition for benzene to cause leukemia and toxic effects of the bone marrow. Benzoquinone, on the other hand, is formed following nonenzymatic rearrangement of BO to phenol, CYP2E1 oxidation of phenol (to hydroquinone), and oxidation of hydroquinone to 1,4-BQ, either spontaneously or via peroxidases [1,2]. Thus, formation of 1,4-BQ is subject to several factors involving the dispositions of BO, phenol and hydroquinone (e.g., reactions with glutathione and other nucleophiles, oxidizing and reducing enzymes, conjugation with sulfate, etc.) as well as the competitive inhibition of CYP2E1 by BO and phenol. As such, 1,4-BQ-Alb is less useful than BO-Alb as a global measure of benzene metabolism but is arguably more relevant to the health risk of benzene, insofar as 1,4-BQ is widely thought to be an important toxic metabo- 108 S.M. Rappaport et al. / Chemico-Biological Interactions 153154 (2005) 103109 lite of benzene [1,2]. But in either case, the concave downwards production of BO-Alb and 1,4-BQ-Alb, observed at increasing exposure to benzene, point to the initial CYP2E1 oxidation as a key factor in modulating human health risk of benzene. Finally, it is worth mentioning that the concave downwards shapes of the exposureadduct relationships shown in Fig. 3B, suggest that metabolism per unit of benzene exposure should be much more effective at low air levels than at high air levels of benzene [24]. This has two important implications for risk assessment. First, in occupational studies where fullshift benzene levels can be very high, one would anticipate that transient days of exposure greater than about 50 ppm would have diminished impact upon the health risk, since benzene metabolism is substantially saturated in this region (Fig. 3B). This would lead to a concave downwards shape in the exposureleukemia incidence for benzene, as reported by Hayes et al. [29], and would reduce the impact of high transient (peak) exposures to benzene upon leukemia risk, as observed by Glass et al. [30]. Second, our results indicate that environmental exposure to benzene, which occurs at air levels less than 1 ppm [31], would produce the maximum possible mass of metabolites per unit of benzene exposure. Since essentially every human is exposed to low levels of benzene from use of hydrocarbon fuels and tobacco [32], the potential for benzene to induce leukemia following long-term environmental exposure should not be dismissed out of hand. Acknowledgements We appreciate the assistance of Larry Kupper for helpful discussions and of the following persons who contributed to the original studies from which data were provided: Mustafa Dosemici and Richard Hayes (National Cancer Institute), Ximei Jin, Beverley Cohen and Lung-Chi Chen (New York University), Assieh Melikian (American Health Foundation), Huifang Yan, Bohong Xu, Ruidong Mu, Yuying Li, Xiaoling Zhang, and Keqi Li (Tianjin, China). This work was supported by Grants P42ES05948 and P30ES10126 (S.M.R.) and P42ES04705 (M.T.S.) from the National Institute for Environmental Health Sciences and by contract HEI 96-5 from the Health Effects Institute. References [1] R. Snyder, Benzene and leukemia, Crit. Rev. Toxicol. 32 (2002) 155210. [2] R. Snyder, Overview of the toxicology of benzene, J. Toxicol. Environ. Health A 61 (2000) 339346. [3] A.B. Lindstrom, K. Yeowell-O'Connell, S. Waidyanatha, B.T. Golding, R. Tornero-Velez, S.M. Rappaport, Measurement of benzene oxide in the blood of rats following administration of benzene, Carcinogenesis 18 (1997) 16371641. [4] M.R. Lovern, M.J. Turner, M. Myer, G.L. Kedderis, W.E. Bechtold, P.M. Schlosser, Identification of benzene oxide as a product of benzene metabolism by mouse, rat, and human liver microsomes, Carcinogenesis 18 (1997) 16951700. [5] D.A. Eastmond, D.S. Rupa, L.S. Hasegawa, Detection of hyperdiploidy and chromosome breakage in interphase human lymphocytes following exposure to the benzene metabolite hydroquinone using multicolor fluorescence in situ hybridization with DNA probes, Mutat. Res. 322 (1994) 920. [6] C.E. Frantz, H. Chen, D.A. Eastmond, Inhibition of human topoisomerase II in vitro by bioactive benzene metabolites, Environ. Health Perspect 104 (Suppl. 6) (1996) 13191323. [7] I. Gut, V. Nedelcheva, P. Soucek, P. Stopka, B. Tichavska, Cytochromes P450 in benzene metabolism and involvement of their metabolites and reactive oxygen species in toxicity, Environ. Health Perspect 104 (Suppl. 6) (1996) 1211 1218. [8] R.D. Irons, W.S. Stillman, Impact of benzene metabolites on differentiation of bone marrow progenitor cells, Environ. Health Perspect 104 (Suppl. 6) (1996) 12471250. [9] P. Kolachana, V.V. Subrahmanyam, K.B. Meyer, L. Zhang, M. Smith, Benzene and its phenolic metabolites produce oxidative DNA damage in HL60 cells in vitro and in the bone marrow in vivo, Cancer Res. 53 (1993) 10231026. [10] J.W. Yager, D.A. Eastmond, M.L. Robertson, W.M. Paradisin, M.T. Smith, Characterization of micronuclei induced in human lymphocytes by benzene metabolites, Cancer Res. 50 (1990) 393399. [11] N.J. van Sittert, P.J. Boogaard, G.D.J. Beulink, Application of the urinary S-phenylmercapturic acid test as a biomarker for low levels of exposure to benzene in industry, Br. J. Ind. Med. 50 (1993) 460469. [12] N. Rothman, W.E. Bechtold, S.N. Yin, M. Dosemeci, G.L. Li, Y.Z. Wang, W.C. Griffith, M.T. Smith, R.B. Hayes, Urinary excretion of phenol, catechol, hydroquinone, and muconic acid by workers occupationally exposed to benzene, Occup. Environ. Med. 55 (1998) 705711. [13] A.A. Melikian, R. O'Connor, A.K. Prahalad, P. Hu, H. Li, M. Kagan, S. Thompson, Determination of the urinary benzene metabolites S-phenylmercapturic acid and trans,trans-muconic acid by liquid chromatography-tandem mass spectrometry, Carcinogenesis 20 (1999) 719726. [14] S. Waidyanatha, N. Rothman, S. Fustinoni, M.T. Smith, R.B. Hayes, W. Bechtold, M. Dosemeci, G. Li, S. Yin, S.M. Rappaport, Urinary benzene as a biomarker of exposure among occupationally exposed and unexposed subjects, Carcinogenesis 22 (2001) 279286. S.M. Rappaport et al. / Chemico-Biological Interactions 153154 (2005) 103109 109 [15] B.L. Lee, A.L. New, P.W. Kok, H.Y. Ong, C.Y. Shi, C.N. Ong, Urinary trans,trans-muconic acid determined by liquid chromatography: application in biological monitoring of benzene exposure, Clin. Chem. 39 (1993) 17881792. [16] S. Waidyanatha, N. Rothman, G. Li, M.T. Smith, S. Yin, S.M. Rappaport, Rapid determination of six urinary benzene metabolites in occupationally exposed and unexposed subjects, Anal. Biochem. 327 (2004) 184199. [17] F.Y. Bois, T.J. Woodruff, R.C. Spear, Comparison of three physiologically based pharmacokinetic models of benzene disposition, Toxicol. Appl. Pharmacol. 110 (1991) 7988. [18] F.Y. Bois, M.T. Smith, R.C. Spear, Mechanisms of benzene carcinogenesis: application of a physiological model of benzene pharmacokinetics and metabolism, Toxicol. Lett. 56 (1991) 283298. [19] F.Y. Bois, E.T. Jackson, K. Pekari, M.T. Smith, Population toxicokinetics of benzene, Environ. Health Perspect 104 (Suppl. 6) (1996) 14051411. [20] C.C. Travis, J.L. Quillen, A.D. Arms, Pharmacokinetics of benzene, Toxicol. Appl. Pharmacol 102 (1990) 400420. [21] K. Yeowell-O'Connell, N. Rothman, M.T. Smith, R.B. Hayes, G. Li, S. Waidyanatha, M. Dosemeci, L. Zhang, S. Yin, N. Titenko-Holland, S.M. Rappaport, Hemoglobin and albumin adducts of benzene oxide among workers exposed to high levels of benzene, Carcinogenesis 19 (1998) 15651571. [22] K. Yeowell-O'Connell, N. Rothman, S. Waidyanatha, M.T. Smith, R.B. Hayes, G. Li, W.E. Bechtold, M. Dosemeci, L. Zhang, S. Yin, S.M. Rappaport, Protein adducts of 1,4benzoquinone and benzene oxide among smokers and nonsmokers exposed to benzene in China, Cancer Epidemiol. Biomarkers Prev. 10 (2001) 831838. [23] S.M. Rappaport, K. Yeowell-O'Connell, M.T. Smith, M. Dosemeci, R.B. Hayes, L. Zhang, G. Li, S. Yin, N. Rothman, Non-linear production of benzene oxide-albumin adducts with human exposure to benzene, J. Chromatogr. B 778 (2002) 367374. [24] S.M. Rappaport, S. Waidyanatha, Q. Qu, R. Shore, X. Jin, B. Cohen, L.C. Chen, A.A. Melikian, G. Li, S. Yin, H. Yan, B. Xu, R. Mu, Y. Li, X. Zhang, K. Li, Albumin adducts of benzene oxide and 1,4-benzoquinone as measures of human benzene metabolism, Cancer Res. 62 (2002) 13301337. [25] N. Rothman, G.L. Li, M. Dosemeci, W. Bechtold, G.E. Marti, Y.Z. Wang, M. Linet, L.Q. Xi, W. Lu, M.T. Smith, N. TitenkoHolland, L.P. Zhang, W. Blot, S.N. Yin, R.B. Hayes, Hematotoxocity among Chinese workers heavily exposed to benzene [see comments], Am. J. Ind. Med. 29 (1996) 236246. [26] Q. Qu, A.A. Melikian, G. Li, R. Shore, L. Chen, B. Cohen, S. Yin, M.R. Kagan, H. Li, M. Meng, X. Jin, W. Winnik, Y. Li, R. Mu, K. Li, Validation of biomarkers in humans exposed to benzene: urine metabolites, Am. J. Ind. Med. 37 (2000) 522531. [27] T.A. McDonald, N.T. Holland, C. Skibola, P. Duramad, M.T. Smith, Hypothesis: phenol and hydroquinone derived mainly from diet and gastrointestinal flora activity are causal factors in leukemia, Leukemia 15 (2001) 1020. [28] S.M. Rappaport, Biological considerations in assessing exposures to genotoxic and carcinogenic agents, Int. Arch. Occup. Environ. Health 65 (1993) S29S35. [29] R.B. Hayes, S.N. Yin, M. Dosemeci, G.L. Li, S. Wacholder, L.B. Travis, C.Y. Li, N. Rothman, R.N. Hoover, M.S. Linet, Chinese Academy of Preventive Medicine, National Cancer Institute Benzene Study Group, Benzene and the dose-related incidence of hematologic neoplasms in China, J. Natl. Cancer Inst. 89 (1997) 10651071. [30] D.C. Glass, C.N. Gray, D.J. Jolley, C. Gibbons, M.R. Sim, L. Fritschi, G.G. Adams, J.A. Bisby, R. Manuell, Leukemia risk associated with low-level benzene exposure, Epidemiology 14 (2003) 569577. [31] S.M. Rappaport, L.L. Kupper, Variability of environmental exposures to volatile organic compounds, J. Expo. Anal. Environ. Epidemiol. 14 (2004) 92107. [32] L. Wallace, Environmental exposure to benzene: an update, Environ. Health Perspect. 104 (Suppl. 6) (1996) 11291136.