Biomarkers of Exposure, Effect, and Susceptibility in Workers Exposed to Nitrotoluenes

Dinitrotoluenes (DNT) and mononitrotoluenes (NT) are important intermediates in the chemical industry. 2-Nitrotoluene (2NT), 2,4-dinitrotoluene (24DNT), 26DNT, and 2,4,6-trinitrotoluene (TNT) have been classified as carcinogenic in animals and as possibly carcinogenic in humans (1). The U.S. Environmental Protection Agency established a cancer slope factor (equal to the upper limit of the lifetime probability that a cancer-causing chemical will cause cancer at a dose of 1.0 mg/kg/d) of 0.23, 0.68, 0.68, and 0.03 (mg/kg/d)⁻¹, for 2NT, 24DNT, 26DNT, and TNT, respectively (www.epa.gov/iris). An excess of hepatobiliary cancer was found among munitions workers exposed to DNT (2). The incidence of urothelial and renal cancer cases found in miners exposed to explosives containing DNT was increased by a factor of 4.5 and 14.3, respectively (3). Morbidity of total malignant tumors in males was markedly higher in Chinese TNT factory workers than controls (relative risk, 2.3; ref. 4). Liver cancers constituted 31.9% of the total malignant tumor morbidity, and liver cancer mortality was 3.97 times higher than in the controls. Therefore, nitrotoluene exposure represents a major carcinogenic risk.

NT, DNT, and TNT are metabolized by reduction of the nitro group(s) or oxidation of the methyl group (reviewed in ref. 5). One or more nitro groups may be reduced to the corresponding aminonitrotoluenes, toluenediamines, or aminotoluenes, whereas the methyl group is oxidized to a benzylalcohol or benzoic acid. Aminonitrotoluenes, toluenediamines, and aminotoluenes can further be N-oxidized by cytochrome P450 1A2 (CYP1A2) and CYP3A4 to yield N-hydroxyarylamines. The N-hydroxyarylamines and benzylalcohols may undergo conjugation reactions with sulfuryl, glucuronide, or acetyl moieties. These metabolic steps are catalyzed in part by enzymes present in polymorphic forms in humans, e.g., the sulfotransferases SULT1A1 and SULT1A2 and the N-acetyltransferases NAT1 and NAT2 (6-9). Secondary products of N-oxidation or methyl oxidation are responsible for the genotoxic and cytotoxic effects of these compounds.

The initiation of chemical carcinogenesis generally involves the covalent binding of xenobiotics, or their reactive metabolites, with nucleophilic DNA centers (6, 7). Hemoglobin (Hb) in erythrocytes is a molecular target for reactive electrophilic species and has thus been used as a surrogate dosimeter to measure the proportion of exposure that attacks nucleophilic targets, such as DNA (refs. 10, 11, reviewed in ref. 12). Therefore, Hb-adducts are an excellent tool to biomonitor exposed workers.

Hb-adducts result from N-hydroxyarylamines that are oxidized to nitrosoarenes in erythrocytes and then form sulfinamide adducts with cysteine residues in Hb. 2NT, 24DNT, and 26DNT form Hb- and DNA-adducts in rats (13-16). Detoxification products are formed after N-acetylation catalyzed by N-acetyltransferase 2 (NAT2), N-sulfo-conjugation (catalyzed by SULT), and by reaction with glutathione [with and without glutathione S-transferase (GST) catalysis]. The specific GST forms involved in the detoxification process are presently unknown.

Following the paradigm of biomonitoring studies, the following variables were determined: (a) external exposure (air monitoring), (b) internal exposure (urine metabolites preshift and postshift), (c) biologically effective dose (Hb-adducts and urine mutagenicity), (d) individual susceptibility (genetic polymorphisms of nitrotoluene-metabolizing enzymes GSTT1, GSTM1, GSTP1, NAT1, NAT2, SULT1A1, SULT1A2), and (e) biological effects (chromosomal aberrations, clinical variables, splenomegaly, hepatomegaly, and health effects).

The exposed (n = 104) and control (n = 72) workers were employed in a factory manufacturing DNT and TNT. The factory was situated in Liaoning (Liaoning Province, China). The industrial synthesis of DNT and TNT was done by continual batch nitration of NT and subsequently DNT with sulfuric acid and nitric acid. The workers were grouped according to their job description and work location as follows: group leader, NT-tank, DNT-tank, TNT-tank, laboratory of chemical analyses, transportation of TNT to packaging, packaging, control room, disposal of waste acid, and disposal of waste water. The control workers performed tasks that included no chemical exposure. The median age (range) was 34.5 years (22.4-54.7 years) for the exposed and 36.8 years (15.9-53.2 years) for the controls. The median number of work years was 10.5 years (3.6-38.0 years) in the exposed and 17.6 years (4.9-39.4 years) in the control group. Males constituted 71% of the exposed workers and 82% of the factory controls.

The study was done in accordance with the principles embodied in the declaration of Helsinki (www.wma.net/e/policy/b3.htm). Informed consent was obtained from each worker. Each participant was interviewed with a questionnaire about his general health status, exposure history, smoking and alcohol consuming habits, previous medical record, and present symptoms. Sample collection, medical examination, and the questionnaire were all done in the same week. The following examinations were done by the medical department of the Chinese Academy of Preventive Medicine: physical examinations involving blood pressure, nervous system, and heart rate; routine blood and urine tests; liver function test (alanine aminotransferase, alkaline phosphatase, total protein, albumin, and total bilirubin); electrocardiogram; ultrasonic type B examination for liver and spleen; and serologic assays of hepatitis B antigens and antibodies.

Urine metabolites of 8 controls and 80 exposed workers and air levels of 49 personal samplers were determined and have been published by Jones et al. (17). Urine metabolites were determined after β-glucuronidase treatment.

Hb-adducts of 2NT and DNT were determined as described by Jones et al. (14), from 61 controls and 99 exposed workers. Hb-adducts of TNT from 6 controls and 85 exposed workers were analyzed using the method published by Sabbioni et al. (18).

Organic extracts of urine samples from 24 exposed workers were prepared from unhydrolyzed, enzymatically hydrolyzed, or acid-hydrolyzed urine as described previously (19). Briefly, the enzymatic hydrolysis involved incubation of urines at 37°C for 3 hours with β-glucuronidase (Sigma, St. Louis, MO) and arylsulfatase type H-2 from Helix pomatia, E.C. 3.1.6.1 (Sigma). For the acid hydrolysis, urines were incubated at 70°C for 6 hours in 6 mol/L HCl and then neutralized by the addition of 6 mol/L NaOH and NaHCO₃. The organics from the unhydrolyzed or hydrolyzed urines were then extracted from the urines by passing the samples through C18 resin and eluting the organics with methanol. The organics were solvent exchanged into DMSO at a concentration of 150-fold (compared with urine). Urinary mutagenicity was assessed with the Salmonella plate-incorporation assay (20). The frameshift strain YG1041 (hisD3052, rfa, ΔuvrB, pKM101) was used, which also has elevated acetyltransferase and nitroreductase activities due to plasmid-mediated gene amplification (21). Extracts were evaluated at 0, 0.15, 0.75, 1.5, 3.0, and 7.5 mL-equivalents per plate in the absence of S9 mix. Mutagenic potencies (revertants/mL-equivalent) were calculated from the linear portion of the dose-response curves.

Genomic DNA from 101 exposed workers and 72 controls was extracted from lymphocytes by standard techniques (22-24). GSTM1- and GSTT1-specific primer pairs were used together with a third primer pair for β-globin in a multiplex PCR analysis. The absence of the GSTM1- or GSTT1-specific PCR product indicated the corresponding null genotype, whereas a β-globin-specific fragment confirmed proper functioning of the reaction (23, 24). Similarly, in the GSTP1-genotyping, the variant alleles containing a base substitution at nucleotide 313 (GSTP1*B and GSTP1*C) resulting in Ile¹⁰⁵Val amino acid change were differentiated from the wild-type allele (GSTP1*A) by SnaBI restriction enzyme digestion subsequent to a PCR amplification (24). Because this method did not differentiate between GSTP1*B and GSTP1*C alleles, the Val¹⁰⁵ alleles were designated as GSTP1 Val. The NAT1 alleles (*3, *4, *10, *11) were determined as previously described (25). The NAT2 alleles (*4, *5, *6, *7) were determined by the method of Bell et al. (26). To ensure laboratory quality control, two independent readers interpreted the results. Any sample with ambiguous results was retested and a random selection of 10% of all of the samples was repeated. No discrepancies were discovered upon replicate testing.

Polymorphisms in codon 213 (resulting in an Arg/His exchange) of SULT1A1 and codon 235 of SULT1A2 (resulting in an Asn/Thr exchange) were determined by restriction fragment length analyses as described by Engelke at al. (27).

The lack of GSTM1 or GSTT1 activity is due to a homozygous deletion (null genotype) of the GSTM1 and GSTT1 gene, respectively. GSTP1*A and GSTP1*D encode proteins containing valine in position 105, whereas GSTP1*B and GSTP1*C encode isoleucine in this position. This substitution affects the substrate specificity. Compared with the Val forms, the Ile forms show enhanced activity toward the broad-spectrum GST substrate 1-chloro-2,4-dinitrobenzene and decreased activity toward dihydrodiol-epoxides derived from polycyclic aromatic hydrocarbons (28).

For NAT1, the NAT1*10 and NAT1*11 alleles were classified as rapid alleles. The wild-type-like allele NAT1*3 and the wild-type NAT1*4 alleles were considered comparable and classified as normal acetylation alleles. Two groups were formed: homozygous normal acetylators versus the rapid acetylators, which included the homozygous rapid acetylators (individuals with two rapid alleles) and the heterozygous rapid acetylators (individuals with one rapid and one slow allele).

For NAT2, the NAT2*4 allele was considered as the rapid allele; NAT2*5, NAT2*6, and the NAT2*7 were considered as the slow alleles. The genotyping method used (26) did not differentiate between the NAT2*5A and NAT2*5B, NAT2*6A and NAT2*6B, and NAT2*7A and NAT2*7B alleles, respectively. NAT2 genotypes were divided in two groups: the homozygous slow acetylators (individuals having two slow alleles) versus the homozygous rapid acetylators and the heterozygotes.

Functional genetic polymorphisms are known for SULT1A1 and SULT1A2. SULT1A1 His variant shows lower enzyme activity and thermostability than SULT1A1 Arg in platelets (29). cDNA-expressed SULT1A2 Thr enzyme is substantially less active than SULT1A2 Asn (30, 31). SULT1A1 and SULT1A2 are expressed in livers of humans (32). SULT1A1 is also expressed in many other tissues and cells, including lymphocytes, in which chromosomal aberrations were determined in the present study.

Five-milliliter samples of heparinized peripheral blood were collected from the exposed workers and controls for the chromosome aberration assay. Two lymphocyte cultures per sample were established in 20 mL vials within 24 hours after the sampling. Each culture contained 0.3 mL whole blood and 6 mL culture medium consisting of (v/v) 97% of RPMI 1640 (Life Technologies, Glasgow, United Kingdom), 1% phytohemagglutinin (Murex, Dartford, United Kingdom), 1% of 200 mmol/L l-glutamine solution (Life Technologies), and 1% penicillin-streptomycin solution (100 units/mL penicillin, and 100 μg/mL streptomycin; Life Technologies; ref. 33). The cultures were incubated at 37°C for 44 hours. Colcemid solution (70 μL, 10 μg/mL) was added to the vials 2.5 hours before the harvest to arrest mitotic cells in metaphase. The cells were harvested by centrifugation, treated with a hypotonic solution (0.075 mol/L KCl) at 37°C for 8 minutes, and fixed thrice in methanol-glacial acetic acid (3:1). The duplicate cell suspensions of each sample were united after the second fixation. From each tube, six to eight microscope slides were prepared by dropping a few drops of the cell suspension on wet glass slides. The slides were air dried, stained in 10% Giemsa [4%, in Sørensen buffer (pH 7.0), 5 minutes], and coded for the analysis.

Whenever possible, 100 metaphases from each individual were analyzed for chromosomal aberrations using Cytogenetics Image Analysis System CS2 metaphase finder (Cytoscan; Image Recognition Systems, Warrington, United Kingdom). One laboratory technician with a 20-year experience on cytogenetic analysis did all the scoring. The analysis of chromosomal aberrations was done according to International System for Human Cytogenetic Nomenclature (34). When the analyses were completed, the code was broken. The samples of 60 factory controls and 91 exposed workers were analyzed.

Statistical analyses were done with the program SPSS 10.0. The results of the questionnaire and of the medical examination were not known to the scientists performing the analyses of the biomarkers. All results were disclosed at the end of the analyses. All correlation coefficients given in the text were determined with the Spearman rank-order test. For the comparison of one dichotomous-dependent variable with a continuous independent variable, the Mann-Whitney test was used. For the comparison of two sets of dichotomous variable, contingency tables were used.

The air measurements of the present worker group have been published recently (17). Briefly, as the exposure occurred as a mixture, all the possible DNT and NT isomers were investigated (17). 2NT and 4NT were the major component in the air (Table 1). The mean 8-hour time weighted average (in μg/m³) exposure is listed in Table 1. The occupational exposure limit set by the U.S. National Institute of Occupational Safety and Health is 0.15 mg/m³ for nitrotoluenes. The sum of the NT air levels was above the exposure limit for 90% of the workers. The sum of the DNT air levels was above the exposure limit for 8% of the workers.

Results of urine metabolites derived from exposure to 2NT, 4NT, 24DNT, and 26DNT have been described in detail by Jones et al (17). In relation to 2NT and 4NT exposure, the levels of 2- and 4-nitrobenzoic acid (2NBA and 4NBA) and 2- and 4-nitrobenzylalcohol (2NBAlc and 4NBAlc) were determined (Table 1; ref. 17). The nitrobenzoic acids were the major metabolites resulting from exposure to 2NT and 4NT and were present in 96% and 73% of the exposed workers, respectively. The metabolites 4-amino-2-nitrobenzoic acid (4A2NBA), 2,4-dinitrobenzylalcohol (24DNBAlc), 2-amino-4-nitrobenzoic acid (2A4NBA), and 2,4-dinitrobenzoic acid (24DNBA) resulting from exposure to 24DNT were found in 91%, 89%, 88%, and 78% of the exposed workers, respectively. The metabolites 26DNBAlc and 26DNBA, resulting from 26DNT exposure, were found in 99% and 86% of the exposed workers, respectively. Quantitatively, 2A4NBA, 4A2NBA, and 26DNBAlc were the major metabolites.

Results of Hb-adducts deriving from 2NT, 4NT, 24DNT, and 26DNT have been published recently (14). Hb-adducts resulting from exposure to 2NT and 4NT were 2-methylaniline (2MA) and 4MA, respectively (Table 1). Hb-adducts resulting from exposure to 24DNT were 4-amino-2-nitrotoluene (4A2NT) and 2,4-toluenediamine (24TDA). Hb-adducts from 26DNT exposure were 2-amino-6-nitrotoluene (2A6NT) and 26TDA. With respect to 24DNT, 4A2NT was the predominant adduct, and its amount was ∼24-fold higher than that of 24TDA. With respect to 26DNT, 2A6NT was the predominant adduct, and its amount was ∼20-fold higher than that of 26TDA. 2MA and 4MA were found in all exposed and factory controls. 4A2NT, 2A6NT, 24TDA, and 26TDA were found in 99%, 96%, 100%, and 85% of the exposed workers, respectively. In the factory controls, 4A2NT, 2A6NT, 24TDA, and 26TDA were present in 62%, 31%, 64%, and 62% of the workers. Hb-adducts resulting from exposure to TNT were analyzed in 85 of 99 exposed workers. TNT formed two adducts: 4-amino-2,6-dinitrotoluene (4ADNT) and 2-amino-4,6-dinitrotoluene (2ADNT). The levels of 4ADNT adducts were nearly 10-fold higher than the levels of 2ADNT adducts.

The mean Hb-adduct levels decreased in the following order 24DNT > TNT > 2NT > 26DNT > 4NT. This order was found in all job categories except in packers (n = 6), where TNT formed the major adducts. The levels of adducts were, on average, four times lower for TNT than for 24DNT. Taking the values generated by the same laboratory (35), Hb binding of TNT in rats was 10 times higher than Hb binding of 24DNT. Assuming that the same percentage of the dose binds in rats and humans, it seems that the workers of this factory were exposed to ∼40 times more 24DNT than TNT. TNT adducts have recently been determined in workers used in a munitions factory (18). The mean TNT adduct levels found in the exposed workers of the present study were 26 times lower than the mean adduct levels found in the munitions factory workers (476 pmol/g Hb; ref. 18). The health effects—splenomegaly, hepatomegaly, and cataract—were observed in 13%, 22%, and 60% of the ammunition factory workers (18). In the present group of workers exposed mainly to DNT and NT, splenomegaly and hepatomegaly were found in <2% of the workers. Cataract is not prevalent in DNT workers and was not investigated in the present group of workers.

Urine mutagenicity was determined in a subset of exposed workers (n = 24). The mean mutagenic potencies (revertants/mL-equivalent) of the unhydrolyzed, enzymatically hydrolyzed, and acid-hydrolyzed urines were 46.2 (21.5, 33.3, and 74.9 for the 25th, 50th, and 75th percentiles), 127 (40.3, 116.2, 141.6), and 354.2 (98.9, 251.9, 568.9), respectively. The mutagenicity of the unhydrolyzed urine correlated with that of the enzymatically hydrolyzed urine (r = 0.74, P < 0.001) and acid-hydrolyzed urine (r = 0.78, P < 0.001). The mutagenicity of the enzymatically and acid-hydrolyzed urines correlated with each other (r = 0.84, P < 0.001).

Spearman rank correlations between the external, internal, and biologically effective dose are summarized in Table 2. The different markers were not available for all workers. Within the same category, the biomarkers correlated well among each other, e.g., Hb-adducts of 24DNT versus Hb-adducts of 26DNT. In contrast, the air levels did not correlate with the urine metabolites nor with the Hb-adducts. The urine levels of 2NT correlated poorly with the Hb-adduct levels. Moderate correlations were found between the urine metabolites of 24DNT or 26DNT and the respective Hb-adducts. The levels of the urinary metabolite 26DNBAlc correlated (P < 0.01) with the mutagenicity of unhydrolyzed (r = 0.58), enzymatically hydrolyzed (r = 0.67), and acid-hydrolyzed urine (r = 0.67). The urinary levels of 24DNBAlc (P < 0.01) correlated with the mutagenicity of unhydrolyzed (r = 0.66), enzymatically hydrolyzed (r = 0.74), and acid-hydrolyzed urine (r = 0.71). Poor correlations were found between urinary mutagenicity and urinary levels of benzylalcohol metabolites from mononitrotoluenes (r < 0.30, P > 0.05). Thus, metabolites of the benzylalcohols of the dinitrotoluenes seem to be important for the observed urinary mutagenicity.

Hb-adduct levels of 24DNT, 26DNT, and TNT correlated with urine mutagenicity. Therefore, Hb-adducts are a good marker for the mutagenic metabolites present in urine. Urine mutagenicity was a consequence of DNT exposure, because in munitions workers exposed only to TNT, investigated for another study (data not shown)⁸

The genotype distributions found in the workers are presented in Table 3. The genotypes are not evenly distributed between control and exposed workers for GSTM1, NAT1, SULT1A1, and SULT1A2. In the exposed group, the GSTM1-null genotype was significantly less prevalent (P < 0.05, Fisher's exact test) than in the control group. This was also noted in workers exposed to TNT in an ammunition factory.⁸ The GSTM1-null genotype was present in 71% of the controls and in 55% of the exposed workers. Thus, it is possible that adverse effects deter GSTM1-deficient subjects from working with nitrotoluenes. The polymorphism of SULT1A1 and SULT1A2 showed strong genetic linkage. Among the 132 subjects homozygous for the Arg form (wild type) of SULT1A1, 131 were also homozygous for the Asn form (wild type) of SULT1A2 and the remaining subject was heterozygous SULT1A2. A single subject was homozygous for the His variant forms of SULT1A1 and also for the Thr variant form of SULT1A2. The remaining 35 subjects were heterozygous for heterozygous for SULT1A1. Thirty-four of them were also heterozygous for SULT1A2, and the other one was homozygous for its wild-type allele. A similar genetic linkage between SULT1A1 and SULT1A2 alleles was previously found in Caucasians (27).

In a first analysis, the genotypes of the exposed workers were correlated with biomarker levels without taking into account personal exposures (Table 4). The sum of the adduct levels deriving from 2NT, 24DNT, 26DNT, and TNT (see Tables 1 and 4) and the single adduct levels were compared among the different genotypes. Most differences were not statistically significant. There was a significant increase for the major adduct 4ADNT deriving from exposure to TNT (P < 0.05) of TNT-adduct levels in individuals who carried the GSTP1 Val variant alleles. An increase of borderline significance (P < 0.1) was seen for TNT-Hb-adduct levels in GSTM1-null subjects and for the levels of 24DNT and 26DNT adducts in workers with the rapid NAT1 genotype. Surprisingly, no significant differences were noticed between people with the rapid and slow NAT2 genotype. Maybe at high doses, the N-acetylation step was saturated as postulated in a study of 4-aminobiphenyl Hb-adducts in heavy smokers (36).

The levels of one of the 24DNT adducts (24TDA) was significantly lower (P < 0.05) in carriers of the SULT1A2 Thr variant allele, and a borderline effect was also seen for the 2NT adduct. The variant SULT1A2 enzyme is less active than the wild-type form (30, 31). Moreover, the SULT1A2 variant Thr allele is genetically linked with SULT1A1 His variant allele whose expression is reduced in vivo, as observed in platelets (29). Therefore, it seems that N-hydroxyarylamine sulfonation rather than N-sulfonation was an important step for Hb-adduct levels. The high-activity genotype, resulting in high amounts of the sulfuric acid ester, may result in a majority of the metabolite solubilizing back to the N-hydroxy derivative, thereby giving an additional opportunity for Hb-adduct formation. Poor metabolizers, on the other hand, are more likely to form N-O-glucuronides, which, although hydrolyzable, are largely excreted.

Pairs of genotypes were combined and compared with the Hb-adduct levels. For example, for carriers of the GSTM1-positive gene, the levels of the 24TDA adduct (P < 0.05) increased in the presence of the GSTP1 Val variant alleles instead of the GSTP1 Ile/Ile. In workers with the GSTP1 homozygous wild-type genotype, the increase of the 24DNT-Hb-adducts among subjects with the NAT1 rapid genotype was significant (P < 0.05) compared with the workers with the NAT1 slow genotype. In workers with the GSTM1-null genotype, all adduct levels were significantly higher (P < 0.05) in subjects with the SULT1A2 homozygous wild-type gene instead of the SULT1A2 Thr variant allele. A similar result was obtained for SULT1A1. Other combinations did not yield significant differences in the Hb-adduct levels.

In a further analysis, Hb-adduct levels were correlated to air levels by taking into account the genotypes. We expected that correlations would improve when genotypes were considered (Table 2). In workers (n = 26) with the GSTM1 positive genotype, correlations for 26DNT and 2NT adducts improved to r = 0.36 (P = 0.07) and r = 0.44 (P < 0.05), respectively. In workers with the NAT1 rapid genotype (n = 33), the correlation with 2NT increased to r = 0.51 (P < 0.01). In the workers with the NAT2 slow genotype (n = 9), the correlation between Hb-adduct and air levels for 2NT, 24DNT, and 26DNT increased to r = 0.97 (P < 0.01), r = 0.71 (P < 0.05), and r = 0.59 (P < 0.05), respectively. Except for the NAT2 genotype, the other genotypes did not influence substantially the correlation between air levels and Hb-adduct levels.

We found several statistically significant associations between Hb-adducts and clinical blood or urine variables (Table 5). The Hb-adduct levels were significantly higher in workers who were positive for glucose and/or protein in the urine. The Hb-adduct levels were also higher in workers who were positive for urobilinogen. In serum, bilirubin levels and urea levels were significantly lower in workers with high Hb-adduct levels. Alkaline phosphatase levels increased, but alanine aminotransferase levels decreased with nitrotoluene exposure. Albumin and total protein decreased, but Hb concentration and RBC count increased with exposure to the nitrotoluenes. The WBC count was lower in workers with high Hb-adduct levels.

In summary, the positive test of protein and glucose in urine could be caused from tubular or glomerular damage by the nitrotoluenes. The nephrotoxicity of nitrotoluenes has been documented in an earlier study in miners exposed to DNT (3) and in rodents dosed 2NT (37) and 24DNT (38). The high levels of alkaline phosphatase and the low level of serum proteins could be a sign of hepatotoxic effects. Hb concentration and RBC counts were higher in highly exposed workers. This is in contrast to the findings in animals dosed with 2NT (37) and DNT (refs. 39, 40; www.epa.gov/iris/subst/0524.htm).

In the present study, inertia, nausea, insomnia, somnolence, dizziness, and headache were more prevalent in the exposed workers than in control workers: 31.5% versus 1.4%, 23.5% versus 0%, 21.4% versus 8.3%, 8.2% versus 2.8%, and 8.2% versus 2.8%, respectively. The differences in the prevalences were all statistically significant (Fisher's exact test, P < 0.05), except for headache (14).

The health effects of all workers were compared with the Hb-adduct levels using logistic regression analysis (14). The odds ratio (OR) of suffering from inertia were 3.2 times higher [95% confidence interval (95% CI), 1.8-5.8] when the level of 4A2NT Hb-adducts increased by 1 log unit. Similar ORs were observed with somnolence (OR, 3.1; 95% CI, 1.4-6.9), nausea (OR, 2.4; 95% CI, 1.3-4.3), and dizziness (OR, 5.5; 95% CI, 1.3-24.2). Similar relationships were found for adducts resulting from 26DNT exposure using 2A6NT-Hb as a marker for the biologically effective dose. These results were tested for confounding factors like age, smoker status, and gender using stepwise forward logistic regression analysis. In the case of nausea, age was a borderline confounder (14).

The genotypes are not evenly distributed between control and exposed workers for GSTM1, NAT1, SULT1A1, and SULT1A2. Therefore, the comparison of the genotypes with health effects was done only for the exposed worker group using contingency tables. The genotypes found in the exposed workers were compared with the different health effects. In the following section, only the most significant findings are mentioned. It seems that workers carrying the GSTM1-null genotype were more susceptible to nausea and dizziness, noticed particularly in DNT-exposed workers. In the exposed group, 20 of 51 people who were GSTM1 null, and 3 of 41 carrying a functional GSTM1 gene suffered from nausea (OR, 8.8; 95% CI, 2.4-32.2). Workers with the GSTM1-null genotype also more often suffered from dizziness (OR, 3.3; 95% CI, 0.7-17.0). Workers with a functional GSTM1 gene had a higher risk of suffering from headache (OR, 9.5; 95% CI, 1.1-80.2) and insomnia (OR, 2.9; 95% CI, 1.1-8.1). Workers with the slow NAT2 gene suffered from nausea (OR, 2.4; 95% CI, 0.8-7.3), inertia (OR, 2.8; 95% CI, 0.99-8.2), and somnolence (OR, 2.6; 95% CI, 0.8-8.9). Workers with the SULT1A1 His or SULT1A2 Thr variant alleles suffered more from headache (OR, 5.9; 95% CI, 1.3-27.2). Workers with the slow NAT1 genotype suffered from dermatosis (OR, 5.3; 95% CI, 1.1-24.1).

Therefore, it seems that workers with the GSTM1-null genotype were transferred from the highly exposed sectors to other sectors of the factory because they experienced nausea more often than GSTM1-positive workers. This would explain the uneven distribution of the GSTM1 genotype among the exposed and control workers from the same factory.

The results of the chromosome aberration analyses of the workers exposed to nitrotoluenes and of the respective controls are presented in Table 6. The results suggest a clastogenic effect for exposure to nitrotoluenes. The total frequency of chromatid-type aberrations (breaks, exchanges, and gaps) and the total chromosomal aberration frequency (chromatid type + chromosome type + gaps) were significantly higher in the exposed group than the controls (P < 0.05, Mann-Whitney-test, independent t test). Total chromosomal aberration frequency was significantly higher in the exposed workers compared with a group of workers exposed only to TNT (2.67 ± 1.77).⁸ As the age structure and the prevalence of smokers were comparable in the three groups of workers, the cytogenetic effects were probably due to exposure to dinitrotoluenes and mononitrotoluenes.

The main confounders for cytogenetic analysis are age and smoking habits. Therefore, in a further analysis, the data of all workers were evaluated with contingency tables. Total chromosomal aberration frequencies and Hb-adduct levels were categorized in two groups. Smoking was categorized in smokers and nonsmokers. The OR for chromosomal aberrations between the groups were as follows: (a) exposed versus control workers (OR, 2.1; 95% CI, 1.1-4.2); (b) smokers versus nonsmokers (OR, 2.0; 95% CI, 1.0-3.9); and (c) high versus low Hb-adduct levels of 24DNT (OR, 1.5; 95% CI, 0.75-2.8). Chromosomal aberrations were then analyzed after stratification of the subjects into three age groups. For the youngest workers (15.9-31.0 years), a significant relationship was found between the sum of the Hb-adducts of 24DNT and chromosomal aberration frequency (OR, 6.5; 95% CI, 1.3-33.3). Similar results were obtained by comparison with 4A2NT, the major adduct of 24DNT. Hb-adducts showed no significant associations with chromosomal aberrations in the other age groups. Therefore, the level of chromosomal aberrations in young people significantly depended on exposure to nitrotoluenes.

The effect of the nitrotoluene exposure on chromosomal aberrations was primarily seen in SULT1A1 and SULT1A2 wild-type homozygotes but not in carriers of SULT1A1 and SULT1A2 variant alleles. Exposed workers with the SULT1A1 Arg/Arg or SULT1A2 Asn/Asn genotypes showed (a) a significantly higher frequency of total aberrations with (P < 0.01; Mann-Whitney test) or without gaps (P < 0.05) and chromatid-type aberrations with gaps (P < 0.05) than controls of the same genotypes, and (b) a borderline increase in the frequency of total aberrations, including gaps, in comparison with carriers of the respective variant alleles of SULT1A1 (P = 0.07) and SULT1A2 (P = 0.06; Table 6). These findings are in accordance with our results on Hb-adducts, further supporting the importance of N-hydroxarylamine sulfonation as a pathway resulting in a higher level of reactive metabolites.

For NAT1 genotype, the effect of the nitrotoluene exposure on chromosomal aberrations was restricted to NAT1 fast acetylators. Exposed workers who were NAT1 fast acetylators showed, in comparison with the same genotype in the controls, a statistically significant increase in total chromatid-type aberrations including gaps (P = 0.01), and total aberrations with and without gaps (P < 0.05; Table 6). No significant differences between the exposed and controls were seen for NAT1 slow acetylators. The effect of nitrotoluene exposure on chromosome aberrations was different in GSTT1-null than GSTT1-positive subjects. GSTT1-null exposed workers showed a significantly higher (P < 0.05) frequency of (a) chromatid-type aberrations including gaps (Table 6) than GSTT1-null subjects among the controls and (b) chromatid-type breaks than GSTT1-positive exposed workers (mean 2.68 ± 1.67 versus 1.96 ± 1.49). These findings concerned nonsmokers but not smokers (data not shown). GSTT1-positive exposed workers had a significant increase (P < 0.05) in (c) chromosome-type breaks (mean 0.53 ± 0.79 versus 0.17 ± 0.38) and total chromosome-type aberrations (Table 6) in comparison with GSTT1-positive controls and (d) chromosome-type breaks in comparison with GSTT1 null exposed (mean 0.21 ± 0.47); both (c) and (d) were restricted to smokers (data not shown). Among the controls, GSTT1-null subjects had an elevated frequency of chromosome-type aberrations (Table 6); this effect was separately seen in smokers (P < 0.05) but not in nonsmokers (data not shown), which agrees with our earlier results suggesting an effect of GSTT1-null genotype on the level of chromosome-type aberrations in smokers (41).

Among the exposed workers, chromosome-type exchanges were more frequent in carriers of GSTP1 Val variant allele (Ile/Val and Val/Val; mean ± SD 0.59 ± 0.75, n = 27) than in Ile/Ile homozygotes (0.21 ± 0.42, n = 61; P = 0.01). These data agree with our results on TNT Hb-adducts and suggest that GSTP1 variant allele may be associated with a lowered protection against reactive metabolites of TNT.

Chinese workers of the present factory were exposed to high levels of animal carcinogens. Air levels did not reflect the internal exposure. Other routes of exposure seem to be more important. Urine of the exposed workers was mutagenic. In general, there was a moderate correlation among urine metabolites and Hb-adducts. Genotypes did not significantly influence the Hb-adduct levels except for SULT1A2. This is possibly explained by the fact that the substrate specificity of the different enzyme forms is generally unknown for the compounds of the present study and their various metabolites. The genotype distribution of GSTM1 was significantly different between exposed and control workers. This may be due to nausea, which was particularly experienced by workers with the GSTM1 null-genotype. The unspecific health effects, which are typical of workers exposed to nitrotoluenes, such as inertia, somnolence, nausea, and dizziness, were related to the level of Hb-adducts. Some clinical blood and urine variables were significantly associated with the Hb-adduct levels. Chromosomal aberrations, a biomarker of cancer risk, were increased in the exposed workers, and in young people the level of chromosomal aberrations was significantly associated with nitrotoluene exposure. Therefore, the exposure in the present worker population should be reduced. Yan et al. (4) have shown that workers of Chinese TNT factories have a 2.3 higher risk to die from cancer compared with workers employed elsewhere. The effect of nitrotoluene exposure on the level of chromosomal aberrations depended on the SULT1A1, SULT1A2, NAT1, GSTT1, and GSTP1 genotypes of the subjects, suggesting that these polymorphisms affect the genotoxic effects of nitrotoluenes.

We thank Renate Heilmair for technical assistance.