DNA Single Nucleotide Polymorphisms as Predictors of Toxicity

The majority of ovarian cancer patients are treated with a combination of platinum- and taxane-based chemotherapy. While initial response rates are high (\>90%), some patients experience severe AEs which can lead to discontinuation of therapy. However, both GOG and the paclitaxel package insert include treatment guidelines that involve a decreased dose regimen if AEs are encountered \[1, 2\]. The clinical validation of the genetic differences in these genes as biomarkers for severe AEs would allow the treating physician to alter dosing, thus increasing the time a patient could remain on the drug while decreasing side effects and unnecessary morbidity. Another clinical utility of these genetic differences in ovarian cancer patient care is in the identification of which patients may not benefit from intraperitoneal (IP) chemotherapy or dose dense chemotherapy. While IP chemotherapy has been shown to improve patient outcome, the side effects are much more frequent and severe due to the high dose \[3\]. Testing the patients prior to treatment for predictive genotypes may factor into a doctor's decision to forego IP chemotherapy in favor of standard intravenous delivery. Dose dense chemotherapy has demonstrated improvements in outcome but also some increases in side effects \[4, 5\]. In both therapy regimens, the ability to stratify patients based on the risks of toxicities associated with treatment may lead to a greater benefit of IP or dose dense therapy while minimizing side effects and associated health care costs. Previous studies in ovarian cancer have demonstrated that expression of the proteins ERCC1, GST, and p53 can affect the response to platinum based therapies in ovarian cancer patients \[6-13\]. When SNPs in the genes that encode these proteins were evaluated, a correlation to AEs in response to platinum-based therapies was made \[14, 15\]. Mutations that affected activity of ERCC1 were associated with nephrotoxicity. Mutations in GST family members were associated with neutropenia (GSTA1), neuropathy (GSTM3 and GSTP1), or anemia and thrombocytopenia (GSTM3). Mutations in p53 were associated with neutropenia. In these studies, mutations in XPD and XRCC1 were also found to be predictive of neutropenia (both XPD and XRCC1) and anemia (XPD only). More recently, a study done by the Scottish Gynecological Clinical Trials Group identified SNPs in BCL2, Current Version Date: 04/01/15 Previous Version Date: N/A, Initial version Page 6 of 23 OPRM1 SOX10 and TRPV1 that were associated with neurotoxicity \[16\]. Both the individual and combined value of these biomarkers needs to be assessed. Reviewing studies done in other tumor types has identified SNPs associated with toxicities that may also be applicable in ovarian cancer. In breast cancer, SNPs in the CYP2C8, CYP17A1, and ABCG1 genes have been associated with an increased risk of grade 2+ peripheral neuropathy in patients treated with taxane-based therapies. \[17, 18\]. A different SNP in CYP17A1 (rs619824) has also been previously reported to be associated with bortezomib-induced neuropathy in multiple myeloma patients, potentially supporting the role of this protein in the onset of neuropathy \[19\]. Since these SNPs are germ line differences, there is reason to believe they may be valuable biomarkers in ovarian cancer also. In multiple myeloma, it has been suggested that identification and closer monitoring of patients at risk for neuropathy could be beneficial \[20\]. A similar strategy could be used with ovarian cancer patients to reduce severe toxicities that results in discontinuation of an effective drug. The detection of these polymorphisms by QPCR has been confirmed by direct sequencing of cell lines and tumor tissue. The protocol has been tested in our research lab and has been validated in our CLIA regulated lab for both formalin fixed paraffin embedded samples and blood. In brief, DNA is purified over a spin column, quantitated, and mixed with a specific SNP assay and Master Mix (Life Technologies). Results were analyzed in the Life Technologies TaqMan Genotyper software version 1.3 to determine genotype. The SNPs were detected in blood samples collected from ovarian cancer patients and FFPE ovarian cancer specimens with similar frequencies as reported in the papers referenced above. SNPs from other tumor types were detected in the same blood samples from ovarian cancer patients or FFPE ovarian cancer specimens as well as in blood and FFPE samples from endometrial cancer patients.