Abstract 4722: A BRAF-mutated gene expression signature identifies BRAF wild type colon cancer patients with poor prognosis

Abstract

Abstract Background: We previously identified highly conserved gene expression patterns associated with BRAF V600E mutant MSS colon cancer (Tejpar et al, ASCO 2010). Since these tumors have very poor overall survival and are refractory to therapy, a better understanding of their molecular characteristics is relevant. In this study we explored if the observed BRAF-mutated characteristic gene expression patterns are also present in any non BRAF-mutant subgroup of colon cancer and what the prognostic implications might be. Methods: A representative set of 667 stage II and III FFPE colon cancers from the PETACC3 clinical trial with 7 years follow up data were used for this study. Samples were characterized for KRAS (39% of samples) and BRAF (6.5%) mutation status, Microsatellite instability (9.7%) (Roth et al, JCO 2009) and gene expression data was obtained with the ALMAC “Colorectal Cancer DSA(tm)” platform. A BRAF signature was built using top scoring pairs and was validated internally (split-sample and cross-validation) and externally (in two independent data sets). It was then applied to KRAS mutant and KRAS wild type (WT) MSI and MSS colon carcinomas. Prognostic value of the signature for overall survival (OS) and survival after relapse (SAR) was assessed by Cox proportional hazard models and survival distributions were compared using likelihood ratio tests. Results: A novel 78-gene BRAF signature was identified and validated, with independent validation performance measured by area under the ROC curve between 0.83 and 0.95. The BRAF signature had high sensitivity in predicting BRAF mutants (>88% internal and >75% external validation) and, in addition, it identified a subpopulation of ‘BRAF-like’ patients with poor prognosis, consisting of 37% of all KRAS mutants and of 13% of all double WT (WT2 – non BRAF and non KRAS mutants). In the whole population and in the MSS subpopulation the signature was prognostic for OS (whole population: hazard ratio HR=1.63, p=0.001; MSS HR=2.25, p<0.001) and for SAR (whole population: HR=2.33, p<0.001; MSS HR=2.64, p<0.001). In the WT2/MSS subpopulation, the signature was prognostic for OS (HR=2.37, p=0.018) and SAR (HR=4.22, p<0.001), but only for SAR in the whole WT2 subpopulation (HR=2.07, p=0.021). The signature was prognostic in the KRAS mutants subpopulation for SAR (HR=1.92, p=0.006), and for OS (HR=1.75, p=0.014) and SAR (HR=1.93, p=0.007) in KRAS/MSS. Conclusions: Using a BRAF mutated derived gene expression signature, we identified a new subgroup within the BRAF WT population that bears a similar gene expression pattern and displays a similar bad prognosis. The signature identifies patients for whom additional therapeutic interventions would be important and the genes included might provide new therapeutic targets. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4722. doi:10.1158/1538-7445.AM2011-4722