Modification of Occupational Exposures on Bladder Cancer Risk by Common Genetic Polymorphisms

Abstract

Few studies have demonstrated gene/environment interactions in cancer research. Using data on high-risk occupations for 2258 case patients and 2410 control patients from two bladder cancer studies, we observed that three of 16 known or candidate bladder cancer susceptibility variants displayed statistically significant and consistent evidence of additive interactions; specifically, the GSTM1 deletion polymorphism (P_interaction ≤ .001), rs11892031 (UGT1A, P_interaction = .01), and rs798766 (TMEM129-TACC3-FGFR3, P_interaction = .03). There was limited evidence for multiplicative interactions. When we examined detailed data on a prevalent occupational exposure associated with increased bladder cancer risk, straight metalworking fluids, we also observed statistically significant additive interaction for rs798766 (TMEM129-TACC3-FGFR3, P_interaction = .02), with the interaction more apparent in patients with tumors positive for FGFR3 expression. All statistical tests were two-sided. The interaction we observed for rs798766 (TMEM129-TACC3-FGFR3) with specific exposure to straight metalworking fluids illustrates the value of integrating germline genetic variation, environmental exposures, and tumor marker data to provide insight into the mechanisms of bladder carcinogenesis.

Occupational exposures are a leading cause of bladder cancer, second only to smoking (1). Over 40 high-risk occupations for bladder cancer have been identified (1–3), yet few studies have examined whether the occupational risk is modified by genetic factors. We have demonstrated that the effect of smoking on absolute risk of bladder cancer, quantified using risk difference (RD) parameters, can vary by genetic susceptibility loci for bladder cancer (4–5). To further explore gene/environment interactions, we examined whether common single-nucleotide polymorphisms (SNPs) modify the association between employment in high-risk occupations and bladder cancer risk using data from 2258 case patients and 2410 control patients who participated in the New England Bladder Cancer Study (NEBCS) and the Spanish Bladder Cancer Study (6–7). All subjects gave informed consent, and each study was approved by the host institution’s internal review board.

Lifetime occupational histories were obtained and exposure-oriented questionnaire modules were administered to elicit information on selected exposures (2–3). High-risk occupations were identified separately by sex in each study (2–3), defined as those with bladder cancer odds ratios (ORs) of 1.5 or greater and with 10 or more employed individuals (Supplementary Table 1, available online), and were analyzed as a group hypothesizing a common mechanism of action from exposure to a mixture of putative carcinogens. We also assessed interactions with an a priori suspect occupational exposure, straight metalworking fluids (composed of mineral oils plus additives), which is common among the more prevalent high-risk occupations in both New England and Spain (Supplementary Table 1, available online). Based on detailed exposure data to quantify exposure to straight metalworking fluids in the NEBCS, we have shown a statistically significant association with increased bladder cancer risk (8). We used data on 16 SNPs identified as susceptibility variants for bladder cancer (4,9–17) to evaluate multiplicative and additive interactions. The Supplementary Materials (available online) contain additional details on genotyping, genome-wide interaction analysis, and the statistical methods used to assess interaction (5,18). All statistical tests were two-sided, and a P value of less than .05 was considered statistically significant.

For five loci we observed a statistically significant interaction with employment in a high-risk occupation (P_interaction < .05) (Table 1). Four loci had statistically significant differences in the RD for high-risk occupation by genotype (P_{additive interaction} < .05) (Table 1), the GSTM1 deletion polymorphism (P_interaction ≤ .001), rs11892031 (UGT1A, P_interaction = .01), rs907611 (LSP1- miRNA-4298, P_interaction = .01), and rs798766 (TMEM129-TACC3-FGFR3, P_interaction = .03). In sensitivity analyses based on higher odds ratio cutpoints to define high-risk occupation (OR ≥ 1.7 or ≥ 2.0), three of these four SNPs showed consistent evidence for differences in RD for high-risk occupation by genotype, specifically rs11892031 (UGT1A), rs798766 (TMEM129-TACC3-FGFR3), and GSTM1 present/null genotype (Supplementary Table 2, available online). Evidence of additive interactions for rs798766 (TMEM129-TACC3-FGFR3) and GSTM1 present/null genotype were also present when the genotypes were modeled as a three-level categorical variable rather than as a binary variable assuming a dominant model (Supplementary Table 3, available online). The evidence for additive interaction was strongest for GSTM1. We estimated that the difference in the 30-year absolute risk for men age 50 years living in the United States with and without a high-risk occupation was 3.0% for carriers of GSTM1-null and 1.9% GSTM1-present genotypes (Supplementary Table 4, available online) (5,19). There was limited evidence of multiplicative interaction, with only rs401681 (TERT-CPTML) showing statistically significant differences in the relative risk for high-risk occupation by genotype (P_{multiplicative interaction} = .03) (Table 1; Supplementary Table 2, available online).

Table 1.

Association between employment in a high-risk occupation and bladder cancer risk by susceptibility allele status in the New England Bladder Cancer Study and the Spanish Bladder Cancer Study

			No risk allele				One or two risk alleles
Gene region	SNP	Employed in a high-risk occupation* (Yes/No)	Case patients, No.	Control patients, No.	OR† (95% CI)	P	Case patients	Control patients	OR† (95% CI)	P	P additive interaction‡	P multiplicative interaction§
UGT1A	rs11892031	No	170	280	Ref		979	1272	Ref
		Yes	117	97	2.14 (1.50 to 3.05)	2.76×10-5	813	518	2.20 (1.90 to 2.54)	2.69×10-26	.01	.15
TP63	rs710521	No	70	123	Ref		1060	1397	Ref
		Yes	51	43	2.81 (1.59 to 4.96)	3.91×10-4	860	556	2.19 (1.90 to 2.52)	6.97×10-28	.12	.61
TMEM129- TACC3-FGFR3	rs798766	No	721	1025	Ref		426	530	Ref
		Yes	565	421	2.09 (1.77 to 2.48)	5.31×10-18	366	196	2.48 (1.98 to 3.11)	4.37×10-15	.03	.33
TERT-CLPTML	rs401681	No	185	322	Ref		949	1208	Ref
		Yes	182	122	2.96 (2.16 to 4.06)	1.61×10-11	729	478	2.09 (1.80 to 2.43)	9.81×10-22	.39	.03
NAT2	rs1495741	No	416	631	Ref		733	922	Ref
		Yes	337	258	2.09 (1.68 to 2.59)	2.20×10-11	595	358	2.31 (1.94 to 2.74)	1.98×10-21	.16	.99
	rs9642880	No	318	470	Ref		815	1057	Ref
		Yes	248	201	1.96 (1.53 to 2.51)	8.59×10-8	661	398	2.36 (2.00 to 2.77)	7.22×10-25	.12	.72
PSCA	rs2294008	No	311	443	Ref		836	1104	Ref
		Yes	255	198	2.08 (1.62 to 2.67)	1.18×10-8	676	415	2.32 (1.98 to 2.72)	7.58×10-25	.57	.90
CCNE1	rs8102137	No	476	731	Ref		674	822	Ref
		Yes	393	258	2.52 (2.05 to 3.10)	3.24×10-18	540	359	2.02 (1.70 to 2.41)	5.13×10-15	.44	.81
CBX6-APOBEC3A	rs1014971	No	98	176	Ref		1052	1377	Ref
		Yes	76	67	2.46 (1.57 to 3.86)	8.74×10-5	857	550	2.19 (1.90 to 2.52)	1.10×10-27	.18	.85
UGT1A	rs17863783	No	39	79	Ref		1105	1473	Ref
		Yes	28	28	1.67 (0.78 to 3.59)	1.85×10-1	903	585	2.23 (1.95 to 2.56)	1.53×10-30	.07	.60
SLC14A2	rs10775480/ rs10853535	No	319	485	Ref		827	1066	Ref
		Yes	284	188	2.47 (1.93 to 3.16)	5.85×10-13	647	427	2.16 (1.83 to 2.53)	8.46×10-21	1.00	.94
GSTM1	null vs present	No	457	777	Ref		735	894	Ref
		Yes	364	311	2.11 (1.72 to 2.58)	7.10×10-13	624	351	2.36 (1.99 to 2.81)	7.66×10-23	<.001	.42
TERC	rs10936599	No	44	78	Ref		1086	1443	Ref
		Yes	46	31	2.61 (1.35 to 5.06)	4.36×10-3	859	564	2.19 (1.91 to 2.52)	1.99×10-28	.46	.20
LSP1- miRNA-4298	rs907611	No	474	706	Ref		643	794	Ref
		Yes	358	277	2.13 (1.73 to 2.63)	8.18×10-13	533	309	2.31 (1.92 to 2.78)	4.37×10-19	.01	.25
	rs6104690	No	162	275	Ref		967	1237	Ref
		Yes	149	104	2.77 (1.96 to 3.90)	6.63×10-9	751	488	2.11 (1.82 to 2.45)	9.17×10-23	.81	.19
CDKAL1	rs4510656	No	227	318	Ref		904	1204	Ref
		Yes	176	128	2.21 (1.63 to 2.99)	3.23×10-7	729	467	2.23 (1.91 to 2.60)	6.80×10-25	.28	.73

*A high-risk occupation was defined to include all occupations where the odds ratio was 1.5 or higher and 10 or more individuals were employed. CI = confidence interval; OR = odds ratio; SNP = single-nucleotide polymorphism; Ref = reference.

† Odds ratios are adjusted for smoking status, region/state, age, and sex. All tests for gene/environment interactions were conducted using categorical variables (each single-nucleotide polymorphism was coded as a dichotomous variable indicating the presence or absence of any risk allele).

‡ Test for additive interaction based on likelihood ratio test assuming gene-environment independence; all statistical tests were two-sided.

§ Test for multiplicative interaction based on likelihood ratio test assuming gene-environment independence; all statistical tests were two-sided.

Recognizing that employment in high-risk occupations reflects heterogeneous exposures, we further explored interactions using detailed data on a specific exposure, straight metalworking fluids. We observed statistically significant differences in the RD for straight metalworking fluids by rs798766 genotype (P_{additive interaction} = .02) (Table 2; Supplementary Table 4, available online).

Table 2.

Association between straight metalworking fluids exposure and bladder cancer risk by susceptibility alleles in the New England Bladder Cancer Study

			No risk allele				One or two risk alleles
Gene region	SNP	Direct/probable straight metalworking fluids exposure (Yes/No)	Case patients	Control patients	OR* (95% CI)	P	Case patients	Control patients	OR* (95% CI)	P	P additive interaction†‡	P multiplicative interaction§
UGT1A	rs11892031	No	32	46	Ref		190	258	Ref
		Yes	8	4	3.86 (0.89 to 16.85)	.072	44	33	1.77 (1.06 to 2.94)	.028	-†	-†
TP63	rs710521	No	7	15	Ref		213	275	Ref
		Yes	1	2	1.92 (0.04 to 94.7)	.742	47	33	1.85 1.12 to 3.05)	.017	-†	-†
TMEM129-TACC3-FGFR3	rs798766	No	141	202	Ref		84	103	Ref
		Yes	25	26	1.44 (0.77 to 2.7)	.252	27	11	3.25(1.45 to 7.27)	.004	.02	.06
TERT-CLPTML	rs401681	No	30	61	Ref		191	235	Ref
		Yes	12	7	4.11 (1.22 to 13.85)	.022	36	28	1.62 (0.93 to 2.83)	.089	.19	.06
NAT2	rs1495741	No	83	118	Ref		142	185	Ref
		Yes	23	17	1.86 (0.9 to 3.86)	.093	28	20	1.84 (0.96 to 3.50)	.064	.33	.28
	rs9642880	No	70	94	Ref		149	198	Ref
		Yes	15	14	1.11 (0.48 to 2.58)	.802	33	21	2.33 (1.24 to 4.36)	.008	.84	.92
PSCA	rs2294008	No	57	98	Ref		168	206	Ref
		Yes	14	9	2.66 (1.01 to 6.96)	.047	38	28	1.65 (0.95 to 2.87)	.078	.88	.81
CCNE1	rs8102137	No	95	131	Ref		130	174	Ref
		Yes	20	14	1.68 (0.77 to 3.67)	.191	32	23	1.98 (1.07 to 3.68)	.031	.62	.62
CBX6-APOBEC3A	rs1014971	No	21	41	Ref		204	263	Ref
		Yes	3	9	0.65 (0.12 to 3.36)	.603	49	28	2.17 (1.29 to 3.67)	.004	-†	-†
UGT1A	rs17863783	No	6	14	Ref		218	291	Ref
		Yes	2	3	7.93 (0.17 to 377.01)	.293	50	34	1.92 (1.18 to 3.13)	.009	-†	-†
SLC14A2	rs10775480/ rs10853535	No	66	90	Ref		159	214	Ref
		Yes	14	6	2.92 (1 to 8.51)	.050	38	31	1.62 (0.94 to 2.79)	.080	.89	.78
GSTM1	null vs present	No	83	133	Ref		151	182	Ref
		Yes	21	23	1.50 (0.75 to 3.01)	.248	35	21	1.96 (1.07 to 3.62)	.031	.79	.78
TERC	rs10936599	No	11	13	Ref		207	280	Ref
		Yes	2	3	0.43 (0.04 to 4.81)	.496	46	32	1.97 (1.18 to 3.27)	.009	-†	-†
LSP1- miRNA-4298	rs907611	No	88	131	Ref		127	154	Ref
		Yes	25	16	2.40 (1.15 to 5.04)	.020	22	18	1.47 (0.73 to 2.96)	.276	.15	.91
	rs6104690	No	40	55	Ref		176	235	Ref
		Yes	3	7	0.76 (0.16 to 3.69)	.734	44	27	2.16 (1.26 to 3.70)	.005	-†	-†
CDKAL1	rs4510656	No	40	61	Ref		179	232	Ref
		Yes	8	5	3.61 (0.95 to 13.8)	.060	40	30	1.70 (0.99 to 2.91)	.055	.72	.79

* Odds ratios are adjusted for smoking status, region/state, age, and sex. All tests for gene/environment interactions were conducted using categorical variables (each single-nucleotide polymorphism was coded as a dichotomous variable indicating the presence or absence of any risk allele). CI = confidence interval; OR = odds ratio; SNP = single-nucleotide polymorphism; Ref = reference.

† Interaction P values were not calculated for rs11892031, rs710521, rs1014971, rs17863783, rs10936599, and rs6104690 because of cell counts containing fewer than five subjects.

‡ Test for additive interaction based on likelihood ratio test assuming gene-environment independence; all statistical tests were two-sided.

§ Test for multiplicative interaction based on likelihood ratio test assuming gene-environment independence; all statistical tests were two-sided.

Because rs798766 is located within the FGFR3 region and somatic changes in FGFR3 have been described in bladder tumors, we explored the relationship between tumor FGFR3 expression, the rs798766 genotypes, and straight metalworking fluids exposure (see the Supplementary Material, available online). Using FGFR3 immunohistochemistry data on 483 tumors from New England, we observed a statistically significant association between increased FGFR3 expression and the rs798766 risk allele (P = .01) (see the Supplementary Materials and Supplementary Figures 1–3, available online), consistent with a previous report (10). Data on mRNA expression from 197 bladder tumors also showed a trend of higher levels of FGFR3 expression with the rs798766 risk allele (Supplementary Figure 4, available online). Interestingly, the interaction between straight metalworking fluids and rs798766 was more apparent in the subset of patients with tumors displaying strong/intermediate FGFR3 expression (P_{additive interaction} = .02) compared with those with weak/no FGFR3 expression (P_{additive interaction} = .80) (Supplementary Table 5, available online). We found no association between the rs798766 and exposure to straight metalworking fluids in control patients, confirming that these factors are independent. Cumulatively, these data provide support that FGFR3 expression may play a role in defining the underlying mechanistic interaction between the rs798766 and straight metalworking fluid exposure.

It is of interest to note that the three polymorphisms with consistent statistically significant additive interactions with high-risk occupation (GSTM1 null, rs11892031 [UGT1A], rs798766 [TMEM129-TACC3-FGFR3]) have also been observed to have statistically significant additive interactions with exposure to tobacco smoke (4,5,20). Given that GSTM1 and UGT1A are known carcinogen-metabolizing genes, our findings suggest that these genes may be important in the detoxification of one or more bladder chemical carcinogens in both tobacco and some high-risk occupations. Interestingly, a German study also found evidence of interaction among case patients who were both carriers of the GSTM1-null genotype and were employed in occupations exposed to known bladder carcinogens (21,22). Further, we present evidence linking exposure to straight metalworking fluids with bladder cancer risk using FGFR3 tumor expression data. Given that rs798766 also interacts with tobacco smoke (4), these data support the hypothesis that there may be pathways that are common to genotoxic exposures in tobacco smoke and straight metalworking fluids. Lastly, our data support the rs798766 (TMEM129-TACC3-FGFR3) results to be a biologically plausible interaction and also add to the weight of the evidence linking exposure to straight metalworking fluids to bladder cancer risk.

Strengths of our study include high-quality detailed occupational exposure data in two large, well-designed case-control studies. Limitations of our study include limited power to broadly explore gene/environment interactions using the entire genome-wide association study data. Larger studies are needed to replicate and extend the observed findings. Only the P value of less than .001 for the additive interaction test of GSTM1and high-risk occupation remained statistically significant after Bonferroni correction; however, our study revealed additional loci with suggestive interactions with high-risk occupation (eg, UGT1A and LSP1) that will require further study. In addition, although we were able to follow up on our observation of gene/environment interactions with high-risk occupation to a specific prevalent occupational exposure, straight metalworking fluids, we were unable to explore interactions with less prevalent exposures in our study populations because of small numbers of exposed subjects.

In summary, our study provides evidence of gene/environment interactions for occupational exposures and susceptibility loci for bladder cancer. In particular, the interaction we observed for rs798766 (TMEM129-TACC3-FGFR3) with specific exposure to straight metalworking fluids illustrates the value of integrating germline genetic variation, environmental exposures, and tumor marker data to provide insight into the mechanisms of bladder carcinogenesis.

Funding

This work was supported by the Intramural Research Program of the National Institutes of Health and the Division of Cancer Epidemiology and Genetics, National Cancer Institute.

The funders did not play a role in the study design, data collection analysis, writing, or submission of the manuscript.