Of prostateGWAS SNPs, Pesticides and Prostate CancerTable 3. Stratified odds ratios and 95 CI, adjusted for age and state, for associations between pesticides and prostate cancer.Pesticide Use None SNP/Region EHBP1 rs2710647 TET2 rs7679673 17q24 rs1859962 SR3029 site PDLIM5 rs17021918 TERBUFOS TERBUFOS ALDRIN Pesticide MALATHION Genotype TT CT+CC AA AC+CC TT GT+GG CC CT+TT Ca/Co 9/50 95/192 22/82 204/444 65/194 242/486 121/290 185/392 REF REF REF REF REF REF REF REF Low Ca/Co 24/65 99/211 10/21 39/111 28/55 78/146 48/85 60/116 OR (95 CI) 2.17 (0.91, 5.14) 0.96 (0.68, 1.36) 1.86 (0.73, 4.75) 0.79 (0.52, 1.20) 1.72 (0.98, 3.03) 1.06 (0.77, 3.03) 1.38 (0.91, 2.11) 1.09 (0.75, 1.58) High Ca/Co 28/50 91/223 13/14 51/117 28/47 70/151 46/71 53/129 OR (95 CI) 3.43 (1.44, 8.15) 0.80 (0.56, 1.15) 3.67 (1.43, 9.41) 0.97 (0.67, 1.42) 25033180 2.05 (1.16, 3.64) 0.92 (0.66, 1.28) 1.59 (1.03, 2.45) 0.87 (0.60, 1.26) 0.042 0.037 0.006* P-interaction 0.003**Noteworthy at an FDR = 0.20 level. doi:10.1371/journal.pone.0058195.tcancer [17,18,22?7,36,37] (Table 2), except for rs12500426 (PDLIM5) for which the opposite allele was observed to be the risk allele compared to the initial report [17]. Among the 30 genotyped SNPs, the strongest association was with the MSMB SNP rs10993994 (p-trend = 0.0002, Table 2). Additionally, there were eight loci with 0.001,P-trend ,0.01 (rs1859962, rs5759167, rs2710647, rs4430796, rs7841060, rs902774, rs17632542, rs16901979) and three loci with 0.01,P-trend ,0.05 (rs10896449, rs266849, rs10486567). Stratified odds ratios for the association between pesticide use and prostate cancer for interactions ,0.05 and a significant increased risk of prostate cancer following a monotonic pattern are presented in Table 3. Among men carrying two T alleles at rs2710647 in EH domain binding protein 1 (EHBP1), the risk 23977191 of prostate cancer in those with low malathion use was 2.17 times those with no use (95 CI: 0.91, 5.14) and in those with high malathion use was 3.43 times those with no use (95 CI: 1.44?8.15) (P-interaction = 0.003). Among men carrying two A alleles at rs7679673 in TET2, the risk of prostate cancer associated with low aldrin use was 1.86 times those with no use (95 CI: 0.73, 4.75) and for high aldrin use was 3.67 times those with no use (95 CI: 1.43, 9.41) (P-interaction = 0.006). In contrast, associations were null for other genotypes. After BIBS39 correction for multiple tests, both of these interactions remained noteworthy at the FDR = 0.20 level. Among men carrying the variant allele at the PDLIM5 SNPs rs1859962 or rs17021918 increased prostate cancer risk was observed with high compared to no terbufos use (OR = 2.05, 95 CI: 1.16?.64, P-interaction = 0.037), (OR 
= 1.59, 95 CI: 1.03?2.45, P-interaction = 0.042), respectively (Table 3). Although nominally significant without adjustment for multiple testing, these interactions were not noteworthy after adjustment using 
the FDR method. No interactions were observed between cumulative genetic score and pesticide use in relation to prostate cancer risk (data not shown).DiscussionWe observed four quantitative interactions between GWAS loci and select pesticide use and risk of prostate cancer. Two of these, malathion-rs2710647 and aldrin-rs7679673, were noteworthy at the FDR = 0.20 level after correction for multiple testing. Additional interactions with terbufos were also observed with a lesser level of significance. Interestingly, all of the observed interactions are with pesticides that have.Of prostateGWAS SNPs, Pesticides and Prostate CancerTable 3. Stratified odds ratios and 95 CI, adjusted for age and state, for associations between pesticides and prostate cancer.Pesticide Use None SNP/Region EHBP1 rs2710647 TET2 rs7679673 17q24 rs1859962 PDLIM5 rs17021918 TERBUFOS TERBUFOS ALDRIN Pesticide MALATHION Genotype TT CT+CC AA AC+CC TT GT+GG CC CT+TT Ca/Co 9/50 95/192 22/82 204/444 65/194 242/486 121/290 185/392 REF REF REF REF REF REF REF REF Low Ca/Co 24/65 99/211 10/21 39/111 28/55 78/146 48/85 60/116 OR (95 CI) 2.17 (0.91, 5.14) 0.96 (0.68, 1.36) 1.86 (0.73, 4.75) 0.79 (0.52, 1.20) 1.72 (0.98, 3.03) 1.06 (0.77, 3.03) 1.38 (0.91, 2.11) 1.09 (0.75, 1.58) High Ca/Co 28/50 91/223 13/14 51/117 28/47 70/151 46/71 53/129 OR (95 CI) 3.43 (1.44, 8.15) 0.80 (0.56, 1.15) 3.67 (1.43, 9.41) 0.97 (0.67, 1.42) 25033180 2.05 (1.16, 3.64) 0.92 (0.66, 1.28) 1.59 (1.03, 2.45) 0.87 (0.60, 1.26) 0.042 0.037 0.006* P-interaction 0.003**Noteworthy at an FDR = 0.20 level. doi:10.1371/journal.pone.0058195.tcancer [17,18,22?7,36,37] (Table 2), except for rs12500426 (PDLIM5) for which the opposite allele was observed to be the risk allele compared to the initial report [17]. Among the 30 genotyped SNPs, the strongest association was with the MSMB SNP rs10993994 (p-trend = 0.0002, Table 2). Additionally, there were eight loci with 0.001,P-trend ,0.01 (rs1859962, rs5759167, rs2710647, rs4430796, rs7841060, rs902774, rs17632542, rs16901979) and three loci with 0.01,P-trend ,0.05 (rs10896449, rs266849, rs10486567). Stratified odds ratios for the association between pesticide use and prostate cancer for interactions ,0.05 and a significant increased risk of prostate cancer following a monotonic pattern are presented in Table 3. Among men carrying two T alleles at rs2710647 in EH domain binding protein 1 (EHBP1), the risk 23977191 of prostate cancer in those with low malathion use was 2.17 times those with no use (95 CI: 0.91, 5.14) and in those with high malathion use was 3.43 times those with no use (95 CI: 1.44?8.15) (P-interaction = 0.003). Among men carrying two A alleles at rs7679673 in TET2, the risk of prostate cancer associated with low aldrin use was 1.86 times those with no use (95 CI: 0.73, 4.75) and for high aldrin use was 3.67 times those with no use (95 CI: 1.43, 9.41) (P-interaction = 0.006). In contrast, associations were null for other genotypes. After correction for multiple tests, both of these interactions remained noteworthy at the FDR = 0.20 level. Among men carrying the variant allele at the PDLIM5 SNPs rs1859962 or rs17021918 increased prostate cancer risk was observed with high compared to no terbufos use (OR = 2.05, 95 CI: 1.16?.64, P-interaction = 0.037), (OR = 1.59, 95 CI: 1.03?2.45, P-interaction = 0.042), respectively (Table 3). Although nominally significant without adjustment for multiple testing, these interactions were not noteworthy after adjustment using the FDR method. No interactions were observed between cumulative genetic score and pesticide use in relation to prostate cancer risk (data not shown).DiscussionWe observed four quantitative interactions between GWAS loci and select pesticide use and risk of prostate cancer. Two of these, malathion-rs2710647 and aldrin-rs7679673, were noteworthy at the FDR = 0.20 level after correction for multiple testing. Additional interactions with terbufos were also observed with a lesser level of significance. Interestingly, all of the observed interactions are with pesticides that have.