OBJECTIVE: To examine germline single nucleotide polymorphisms (SNPs) as markers of response to gemcitabine platinum (GP) combination chemotherapy in urothelial carcinoma (UC). METHODS: Saliva or blood was prospectively collected from 216 patients treated with GP for UC of the bladder between 1991 and 2011. Based on reported associations with gemcitabine and cisplatin response or putative mechanisms of gemcitabine or cisplatin/carboplatin activity, we selected SNPs of interest and were able to genotype 59 SNPs (using the SequenomMass ARRAYiPLEX platform) in 261 patients randomly split 2/3 into a training set (n = 174) and 1/3 into a test set (n = 87). Logistic regression was used to test the association between response to GP and SNPs. RESULTS: The median age at diagnosis was 64 years (range: 28 - 85) for the discovery set and 67 years (range: 30 - 84) for the validation set. Males composed 76% and 69%, and white non-Hispanics composed 88% and 91% of the training and test validation sets, respectively. Three SNPs on GALNTL4 (rs7937567, rs12278731, and rs9988868) and one intergenic SNP (rs1321391) were significantly associated with response to GP in the training set and were used to build a SNP score. However, when assessed in the test set, the SNP score was not significantly associated with response. CONCLUSION: Multiple SNPs selected from previous studies failed to predict response to GP in this cohort. Larger studies capable of accounting for population-based allele frequency heterogeneity may be required for replication of genetic alterations important to pharmacogenomics. .
OBJECTIVE: To examine germline single nucleotide polymorphisms (SNPs) as markers of response to gemcitabine platinum (GP) combination chemotherapy in urothelial carcinoma (UC). METHODS: Saliva or blood was prospectively collected from 216 patients treated with GP for UC of the bladder between 1991 and 2011. Based on reported associations with gemcitabine and cisplatin response or putative mechanisms of gemcitabine or cisplatin/carboplatin activity, we selected SNPs of interest and were able to genotype 59 SNPs (using the SequenomMass ARRAYiPLEX platform) in 261 patients randomly split 2/3 into a training set (n = 174) and 1/3 into a test set (n = 87). Logistic regression was used to test the association between response to GP and SNPs. RESULTS: The median age at diagnosis was 64 years (range: 28 - 85) for the discovery set and 67 years (range: 30 - 84) for the validation set. Males composed 76% and 69%, and white non-Hispanics composed 88% and 91% of the training and test validation sets, respectively. Three SNPs on GALNTL4 (rs7937567, rs12278731, and rs9988868) and one intergenic SNP (rs1321391) were significantly associated with response to GP in the training set and were used to build a SNP score. However, when assessed in the test set, the SNP score was not significantly associated with response. CONCLUSION: Multiple SNPs selected from previous studies failed to predict response to GP in this cohort. Larger studies capable of accounting for population-based allele frequency heterogeneity may be required for replication of genetic alterations important to pharmacogenomics. .