Prognostic and predictive value of BRAF mutation alone and in combination with microsatellite instability in stage III colon cancer

Abstract

Purpose The prognostic and predictive role of biomarkers in colorectal cancer is still being defined. The aim of this study was to determine the prognostic value of BRAF mutation alone and in combination with microsatellite instability (MSI), and to determine the interaction between BRAF mutation and MSI status in determining survival benefit after adjuvant 5-FU (5- fluorouracil) based chemotherapy in stage III colon cancer. Methods We performed a retrospective cohort study including all curatively resected stage III colon cancer cases over a 33-year period. A clinicopathologic database was collated (adjuvant chemotherapy, age, gender, obstruction, perforation, tumour location, grade, mucin, nodal stage, extramural vascular and perineural invasion). BRAF (V600E) mutation testing was performed and MSI status established by immunohistochemistry for mismatch repair proteins and molecular testing for National Cancer Institute (NCI) panel markers. Patients were categorised into four groups for comparison: MSS and BRAF-ve (termed “traditional”), MSI and BRAF-ve (termed “presumed Lynch”), MSI and BRAF+ve (termed “sporadic MSI”) and MSS and BRAF+ve (termed “other BRAF”). The primary endpoint was cancer specific survival. Interaction testing was conducted to determine whether there was differential benefit from chemotherapy between groups. Results A total of 686 unselected cases met our inclusion criteria and had tissue available, of which 15.7% had BRAF mutation and 13.8% had MSI. In the adjusted analysis, neither BRAF mutation nor MSI mutation were independently prognostic (HR 1.78, 95% CI 0.89-1.79, P = 0.18, and HR 0.49, 95% CI 0.75-1.83, P = 0.48, respectively). On univariate analysis, survival of patients with presumed Lynch cancers was similar to those with traditional cancers (5-year survival, 62 and 61%, respectively), and while there was no difference in cancer specific survival between sporadic MSI and other BRAF, these tumours had poorer outcome when compared to traditional or presumed Lynch cancers. Adjusted analysis of the four groups, however, showed that none of the subgroups were independently prognostic. Thirty-nine percent received chemotherapy. Overall, adjuvant chemotherapy produced a cancer specific survival benefit (chemotherapy: HR 0.66, 95% CI 0.49-0.88, P < 0.01). On adjusted analysis, neither BRAF nor MSI status were individually predictive of survival benefit. On adjusted analysis specifically of the chemotherapy effect in each subgroup, only patients in the presumed Lynch group (HR 0.260, 95% CI 0.09-0.80, P < 0.01) and other BRAF groups (HR 0.45, 95% CI 0.23-0.87, P 0.01) had a significant survival benefit from chemotherapy. On interaction testing of subgroups, adjusting for all the clinicopathological parameters, patients in the presumed Lynch group (HR 0.28, 95% CI 0.10-0.75, P < 0.01) gained a differentially greater benefit from chemotherapy than other groups. Conclusions BRAF mutated cancers demonstrated a trend towards poorer outcomes, however, after adjusting for clinicopathological factors, MSI and chemotherapy, BRAF mutation was not found to be an independent prognostic biomarker in stage III colon cancer, even when combined with MSI. In this historical cohort, MSI testing was predictive of response to adjuvant chemotherapy in stage III colon cancer, but only when results are interpreted in combination with BRAF. This supports the role of routine testing for these biomarkers.