Exposure to diisocyanates is well known to cause asthma and symptoms such as cough, wheezing, and dyspnoea 1. Moreover, exposure to diisocyanates also often causes symptoms from the upper airways such as rhinitis and conjunctival irritation 2. In a recent study we have shown that TDI exposure can affect risk of symptoms from the eyes, nose bleed, and symptoms of the lower airways (odds ratios between 1.8–3.7) 3.

Because diisocyanates cause asthma in only a minor fraction of exposed workers, genetic factors probably modify disease susceptibility 4. The metabolism of TDI is not yet understood, but there is evidence that several different metabolic pathways are involved in its biotransformation. Isocyanates may react with intracellular glutathione, either directly or through catalysis by the glutathione transferase (GST) system. In the group of GST enzymes that were analyzed, GSTP1 and GSTM1 were found to be the most efficient catalysts of glutathione conjugation of several different isothiocyanates 5. Moreover, GSTs are involved in the protection against reactive oxygen, which is a key component of inflammation.

Non-enzymatic hydrolysis converts isocyanates to their corresponding amines. There is ample proof that aromatic and heterocyclic amines are activated by oxidation of the primary amine by cytochrome P450 enzymes to aryl hydroxylamines. The most important oxygenase is CYP1A2, but also CYP1A1 and CYP1B1 can activate a range of arylamines to reactive intermediates 6. Another frequent step in arylamine metabolism is the generation of acetoxy esters by N-acetyltransferases (NAT1 and NAT2). NAT1 is more abundant in the lung as well as in the eye, but low levels of NAT2 have also been reported in these tissues 78. An additional step in the biotransformation of many aromatic amines is sulfatation by cytosolic sulfotransferases (SULTs). A peroxidase-catalysed mechanism, possibly through myeloperoxidase (MPO) in neutrophiles, to activate arylamines has been suggested 9. Proof for neutrophilic activation in occupational asthma has been reported 10. Functional polymorphisms in such metabolic genes affecting the metabolism of diisocyanates have been studied for their impact on isocyanate-related symptoms 11121314.

There are also several immune-related genes of relevance for the susceptibility to respiratory diseases. The chemokine CCL5 (also labeled RANTES) is responsible for the recruitment of inflammatory cells, such as eosinophils and T-lymphocytes. The HLA class II molecules are involved in the immune responses as they bind antigen-derived peptides and present them to lymphocytes via the T-cell receptor. Mapp et al. have shown overrepresentation of the DQB1*0503 allele among subjects with TDI-induced asthma 15. Tumor necrosis factor alfa (TNF) is chronically expressed in TDI-induced asthma 16 and a study in TNF receptor knockout mice has shown that these animals fail to develop immunological effects from TDI 17.

The objective of the present study was to investigate whether the associations of TDI and symptoms of the eyes and airways were modified by relevant functional polymorphisms in metabolizing enzymes and/or in proteins involved in airway inflammation. Thus, a group of workers exposed to isocyanates, mainly TDI, and a non-exposed group, not exposed to any isocyanates, were analyzed for genotype and health effects.

There were statistically significant differences for sex (p-value < 0.001), genotype distributions for MPO (p = 0.008) and HLADQB1*0501 (p = 0.038) between exposed and non-exposed groups (Table 1). Also there were statistically significant or near significant differences in prevalence of symptoms between the two groups (symptoms of the eyes p < 0.001; wheezing p = 0.060; dry cough p < 0.001; cough with mucus p = 0.069; and nose bleed p < 0.001), which has been reported elsewhere 3. All genes were found to be in Hardy Weinberg equilibrium in the entire study population. This was also true when the subjects were divided into exposed and non-exposed groups, except for SULT1A1, which among exposed workers displayed more variant heterozygotes than expected.

In Table 2, all results that demonstrated statistical significance with respect to genotype-associated risk in either exposure-strata, or interaction between genotype and exposure status, are presented. For the majority of analyses, the unadjusted results were very similar to the adjusted. In the cases where the statistical significance differed, the unadjusted results are presented in the text below. For the other genotypes, no statistically significant associations were found.

For three polymorphisms: CYP1A1*2A, CYP1A1*2B, and TNF -308 there was a pattern consistent with interaction between genotype and exposure status for the majority of symptoms investigated (Fig 1a–c). Among TDI-exposed subjects, the carriers of the CYP1A1*2A variant (crude: exposed workers OR = 1.9, 95% CI 0.62–5.8; non-exposed workers OR = 0.18, 95% 0.021–1.5; interaction p = 0.048 for symptoms of the eyes) and carriers of the CYP1A1*2B variant, had an increased risk for symptoms as compared to the non-exposed subjects. For TNF-308, the variant A allele was associated with a higher risk of symptoms among the non-exposed group, whereas the opposite was found to be true for exposed subjects. Among exposed subjects, the A variant allele was found to protect particularly against wheeze etc., but also against dry cough and cough with mucus (Fig 1c, wheeze etc.: crude OR = 0.54, 95% CI 0.22–1.3, interaction p = 0.130). For SULT1A1 R213H, there was no genotype exposure interaction, but the AG+AA carriers had reduced risk as compared to GG, both among exposed and non-exposed subjects (Fig 1d).

For other polymorphisms, associations were restricted to certain symptoms (Table 2). The rapid genotype of NAT1 was associated with a higher risk of symptoms of the eyes among exposed and interaction was indicated but not statistically significant. CCL5 AG+AA carriers had up to three times higher risk of wheeze etc. and cough with mucus among exposed workers. Carrying HLA-DQB1*05 was associated with an increased risk of dry cough among exposed subjects, and interaction was statistically significant. After subtyping for different HLA-DQ05 alleles, it was found that HLA-DQB1*0502 carriers, but not *0501, had an elevated risk of dry cough (*0502 OR = 4.9, 95% CI 1.1–21) among exposed workers. There were too few individuals with DQB1*0503 or *0504 for any association analysis.

Individuals with GSTM3 AB+BB were protected from cough with mucus among non-exposed subjects, with statistical significance for interaction. In the non-exposed group,GSTM1 null as well as GSTP1*105I/V+V/V individuals displayed significantly higher risk of cough with mucus. For GSTP1, evidence of genotype-exposure interaction was found as well. Among exposed subjects, TNF-863 CA+AA carriers had higher risk of cough with mucus than CC (unadjusted analysis OR = 2.5, 95% CI 1.1–5.6; adjusted analysis OR = 2.1 95% CI 0.91–5.0). Variant allele carriers of MPO were protected against nose bleed among exposed workers.

The gene-gene combinations analysis resulted in statistically significant associations between gene-gene combinations and symptoms. However, the results did not differ much from the results in the single gene analysis: e.g. for the combination NAT1+SULT1A1 and symptoms of the eyes, carriers of NAT1 IM+S and SULT1A1 AG+AA had an OR = 0.19 95% CI 0.066–0.53; for the combination of TNF-308 and SULT1A1 and wheeze etc., carriers of TNF-308 AG+AA and SULT1A1 AG+AA displayed an OR = 0.27 95% CI 0.081–0.91. There was no strong evidence for interaction apart from the combination of GSTM1 and GSTP1 for cough with mucus among non-exposed workers, where carriers of GSTM1 null and GSTP1*105 Val- displayed OR = 22, 95% CI 2.5–200, compared to GSTM1 present and GSTP1*105 Ile/Ile.

Gene-exposure interaction was observed for symptoms of the eyes and lower airways, whereas there were no strong indications of interaction for nose bleed. The CYP1A1 genotypes appeared to modify the effect of TDI. Similar pattern for the MspI (*2A) variant as the Val (*2B) variant was found, which probably is due to linkage between these two polymorphisms 31. However, the effect was strongest for *2B, possibly reflecting the higher biological relevance of this polymorphism. The difference between the wild type Ile and the Val variant appears to depend on the substrate: when the enzyme contained Val, ethoxyresorufin exhibited a mildly elevated activity, and estrogen exhibited a substantially elevated activity. No such effects were observed for benzo(a)pyrene 3233. There are no indications of effect of the MspI variant alone on the CYP1A1 activity 34. The role of CYP1A1 for TDI-related effects of the eyes and airways is at present largely unknown. In a murine model of TDI asthma, challenging with TDI resulted in an increased expression of CYP1A1 in the lung 35. However, the role of CYP1A1 for TDI-asthma was unclear, since TDI seemed to partly inhibit the activity of CYP1A1.

Elevated levels of TNF are frequently observed in bronchoalveolar fluid of asthmatic subjects 36. The -308G/A variant has been shown to be associated with elevated plasma TNF levels and with higher amounts of TNF upon stimulation 37. The -863C/A variant is associated with reduced circulating levels of TNF 38. In our study an increased risk of symptoms was indicated for the variant allele among the non-exposed group, which is in line with what has been reported in other studies for TNF -308 and non-occupational asthma 39. However, we found that among the exposed subjects, the variant allele of TNF -308 was associated with reduced risk among the exposed whereas the variant of -863 was associated with increased risk. Thus, our results indicate that genotypes associated with low TNF circulating levels (-308 G and -863 A) may be linked to symptoms in the TDI-exposed. The impact of TNF polymorphism (-308) on TDI-asthma has been analyzed in a recent study by Beghé et al 40, but no associations were found in that study.

In SULT1A1, the variant allele of R213H (G/A), carries in vitro only 10–25% of the activity of the more common allele 4142. In our study, SULT1A1 showed significant effect on symptoms from the eyes and on wheezing, and a similar non-significant pattern was observed for all symptoms analyzed. With regards to eye symptoms, homozygotes for the more common allele G had up to three times higher risk than individuals heterozygous or homozygous for the low-activity A allele. However, the data did not suggest a gene-exposure interaction due to the fact that a protective effect was seen for exposed as well as for referents.

Higher NAT1 and NAT2 activity could hypothetically contribute to the formation of reactive damaging metabolites, by O-acetylation, in the eyes and airways. Hence, the rapid NAT1 and NAT2 genotypes could increase an individual's risk for TDI-associated symptoms. Gene-TDI interaction was suggested for NAT1 only for eye symptoms, where the rapid genotype showed approximately two times higher risk for symptoms among exposed workers. For NAT2 rapid acetylators, an increased risk was suggested for this symptom as well (OR = 2.17, 95%CI 0.99–4.79, p = 0.054). For the lower airways, an association between NAT1 slow genotype and increased risk for dry cough was found (data not shown), but it was only among non-exposed and there were no symptomatic exposed individuals with slow genotype, making it very difficult to evaluate the effect of this genotype. Thus, in contrast to Wikman and colleagues (14), we did not find any reliable associations with neither NAT1 nor NAT2 with the symptoms of the lower airways. The difference between the studies on isocyanate-exposed workers could be due to differences in selection of study subjects: in contrast to the study by Wikman et al., this is a cross-sectional study and the selection criterion was exposure and not disease. Moreover, we studied predominantly TDI-exposed individuals and did not include study subjects with a mixed diisocyanate exposure. Furthermore, in contrast to Wikman, we did not select asthma cases but rather analyzed symptoms which may be early indicators of asthma-related changes.

The -403G/A variant of CCL5 results in a new consensus element for the GATA transcription factor family 43, which suggests a role in the regulation of gene expression. In the present study, the CCL5 genotype appeared to have an effect on the risk for TDI-associated airway symptoms such as wheeze etc. and cough with mucus: the variant allele was associated with approximately three times higher risk among exposed, suggesting a CCL5-TDI interaction. An association for CCL5 in TDI-associated airway symptoms is also suggested from a recent study showing increased expression of CCL5 in vitro in bronchial epithelial cells when stimulated with TDI 44.

There was also evidence for association of HLA-DQB1*05, in particular the *0502 allele, and isocyanate-related symptoms of the lower airways. Compared to the *0501 allele, the *0502 has an exchange of a valine to a serine at amino acid position 57 in the beta chain of HLA class II molecules. Mapp et al. 15 have found an increased risk of asthma for *0503 carriers (in *0503, valine is substituted with aspartic acid at position 57), which we were unable to analyze, due to low number of study subjects carrying this allele. In contrast to Mapp, we did not find a protective effect for carriers of DQB1*0501 15.

Glutathione S-transferases showed some effect on symptoms of the lower airways, in particular on cough with mucus. However, the effect was not observed among the exposed workers, but rather among the non-exposed subjects. One possibility is that these symptoms are caused by different mechanisms among exposed and non-exposed individuals, i.e. GSTs are not involved in TDI-related airway obstruction. Piirilä and co-workers 12, found an increased risk of diisocyanate-asthma among GSTM1*O subjects. We found only a non-significant higher risk of cough with mucus among the exposed subjects, and the strongest effect was found among non-exposed subjects. The differences in study design as discussed above for NAT1 and NAT2 could also explain differences in results for GSTM1 compared to Piirilä et al., since both Wikman's and Piirilä's reports are based on the same study population. Other studies have reported higher risk of non-occupational asthma for GSTM1*O subjects 4546. Piirilä also studied GSTM3 in relation to diisocyanate-asthma and found later reactions in the bronchial provocation test among GSTM3*AA individuals. We also found that GSTM3 was associated with airway symptom: there was a protective effect of GSTM3*B on cough with mucus. However, this was only evident among the non-exposed individuals. For GSTP1, the exposed I/V+V/V displayed a tendency towards a protective effect, which is in line with the study by Mapp et al. (13) on TDI-asthma and GSTP1. On the other hand, GSTP1 I/V+V/V subjects had an increased risk of cough with mucus among the non-exposed subjects.

Gene-gene interactions are of relevance to study, considering the complex metabolism of isocyanates as well as responses of the immune system. However, since the results differed for most polymorphisms for exposed and non-exposed, we had to analyse the gene-gene interactions in each group. The gene-gene combination analysis did not indicate any evidence for departure from an additive effect. The exception was the results from the combination of GSTM1 and GSTP1 for cough with mucus among non-exposed workers. Nevertheless, there may still be significant gene-gene interactions present for the symptoms analysed, however, for detection of such effects a larger study population had been required.

The low prevalence of certain risk genotypes are reflected in the large confidence interval for ORs obtained. Thus, the data should at this point be cautiously interpreted and new studies are needed for replication of the results. Another aspect to consider is the issue of multiple interferences. We examined fifteen polymorphic loci, and five symptoms for both exposed and non-exposed workers which might result in a large number of false positives. Nevertheless, no correction for multiple testing has been performed. Adjustments such as the Bonferroni correction are not applicable as they cannot take linkage between polymorphisms into account, which occurs among some genes investigated in our study. We do not stress individual significant findings, but rather believe that consistent association patterns between genotypes and symptoms indicate true findings.

By studying TDI-exposed as well as non-exposed individuals, specific gene-TDI interactions were indicated for symptoms of the eyes and lower airways. The finding that certain genotypes modified the effect of symptoms arising in different tissues suggests that these symptoms are associated with each other and have common mechanisms.

We did not expect to find individuals diagnosed with asthma among the exposed workers, since asthmatics should not be employed in isocyanate workplaces in Sweden today, and if workers acquire asthmatic symptoms after employment, they should be transferred to non-exposed jobs. Surprisingly, increased risks of severe symptoms of the airways, e.g., wheeze and/or chest tightness, which are associated with asthma (46), were recorded in the present study. Thus, bearing in mind that our study is a cross-sectional one, with potential for a healthy-worker-selection, our findings may be an underestimation of the true risk. Furthermore, gene-exposure interactions appear to be present in early signs of health effects.

The authors declare that they have no competing interests.

KB carried out part of the molecular genetic analyses, performed part of the statistical analyses and drafted the manuscript. HT carried out the exposure assessments and participated in the design of the study. AA performed part of the statistical analyses. MW and AR carried out part of the molecular analyses and the design of the study. ML conceived of the study, participated in its design and coordination, and performed the health investigations. All authors read and approved the final manuscript.