Significant associations of mismatch repair gene polymorphisms with clinical outcome of pancreatic cancer.

PURPOSE: DNA mismatch repair (MMR) is critical in maintaining genomic stability and may modulate the cellular response to gemcitabine. We hypothesized that genetic variations in MMR may affect the clinical outcome of patients with pancreatic cancer. PATIENTS AND METHODS: We evaluated 15 single-nucleotide polymorphisms (SNPs) of eight MMR genes in 154 patients with potentially resectable pancreatic adenocarcinoma who were enrolled onto phase II clinical trials for preoperative gemcitabine-based chemoradiotherapy from 1999 to 2006. Associations of genotypes with tumor response to therapy (change of tumor size by radiologic evaluation at restaging), margin-negative tumor resection, and overall survival were evaluated using logistic regression and Cox proportional regression models. RESULTS: Five, six, and 10 genotypes were significantly associated with tumor response to preoperative chemoradiotherapy, tumor resectability, and overall survival, respectively, in univariable analysis. TREX1 EX14-460C>T and TP73 Ex2+4G>A genotypes remained as significant predictors for tumor response, MLH1 IVS12-169C>T and TP73 remained as significant predictors for tumor resectability, and EXO1 R354H, TREX1, and TP73 remained as significant predictors for overall survival in multivariable models that included all clinical factors and genotypes examined. A strong combined genotype effect on each clinical end point was observed. For example, 20 of the 25 patients with zero to one adverse genotypes were alive, those with two, three, four, five, and six to seven adverse genotypes had median survival times of 36.2, 23.9, 16.3, 13.0, and 8.3 months, respectively (P < .001). CONCLUSION: SNPs of MMR genes have a potential value as predictors for clinical response to chemoradiotherapy and as prognostic markers for tumor resectability and overall survival of patients with resectable pancreatic cancer.

Duke Scholars

Author James Abbruzzese Medicine, Medical Oncology