The impact of population pharmacogenomics and risk allele frequencies on cisplatininduced peripheral sensory neuropathy (PSN)

12092 Background: Taxane-treated breast cancer patients with genetically African ancestry have worse PSN than other groups. However, no study has examined the association between ancestry and PSN after cisplatin-based chemotherapy. Increased risk could be partially explained by differing risk allele frequencies across populations for alleles increasing the general vulnerability to PSN or altering drug metabolism. Methods: The Platinum Study enrolled cisplatin-treated testicular cancer survivors (TCS) who completed clinical exams and surveys. A PSN score was derived from the mean of 8 sensory items (using EORTC-CIPN20), assigning severity on a 0-2 scale. Multidimensional scaling scores for each TCS were calculated, plotted and anchored by data from the 1000 Genomes Reference population to determine genetic ancestry. Multinomial logistic regression assessed the association between genetic ancestry and PSN. To determine risk alleles for PSN and allele frequencies across populations, risk allele panels were created, including ancestry-informative markers (AIMs) determined by the AncestrySNPMiner tool. Allele frequencies were calculated in each group; SNPs with frequency differences > 0.3 in the African (AFRAFR) population vs. others were included. For filtering the AIMs, GTEx data was used to identify expression quantitative trait loci (eQTL) or splicing quantitative trait loci (sQTL) in nerve/brain tissue. Multinomial logistic regression assessed associations between SNP genotype and PSN phenotype for SNPs with differing allele frequencies across populations. Results: Despite small numbers of non-Europeans, TCS with African ancestry had increased incidence and severity of PSN vs. TCS with European ancestry. In a subset analysis of EUR (n = 681) and AFRAFR (n = 13) patients who received 400-450 mg/m 2 of cisplatin, the relative risk ratio (RRR) in the AFRAFR vs. EUR TCS of severe neuropathy vs. none was 7.96 (P = 0.074) and the RRR for any neuropathy vs. none was 7.79 (P = 0.049). There were 394 independent AIMs with significant ( > 0.3) allele frequency differences between the AFRAFR and other populations that were eQTLs and/or sQTLs in nerve/brain tissue. Using multinominal logistic regression between genotype and phenotype for all TCS (n = 1513) with covariates for age at survey and 10 genetic principal components: 16 SNPs had P-values < 0.05 for severe PSN vs. none and/or any PSN vs. none. Although not statistically significant with multiple testing corrections, these suggestively significant SNPs could be potentially validated in additional populations. Conclusions: These results are preliminary evidence for the potential importance of differing risk allele frequencies across populations in explaining some disparities in cisplatin-related PSN. If confirmed, genotyping for risk variants could impact treatment decisions and enable monitoring to mitigate PSN.