Analysis of Molecular Markers by Anatomic Tumor Site in Stage III Colon Carcinomas from Adjuvant Chemotherapy Trial NCCTG N0147 (Alliance)

PURPOSE: To determine the frequency and prognostic association of molecular markers by anatomic tumor site in patients with stage III colon carcinomas. EXPERIMENTAL DESIGN: In a randomized trial of adjuvant FOLFOX ± cetuximab, BRAF(V600E) and KRAS (exon 2) mutations and DNA mismatch repair (MMR) proteins were analyzed in tumors (N = 3,018) in relationship to tumor location, including subsite. Cox models were used to assess clinical outcome, including overall survival (OS). RESULTS: KRAS codon 12 mutations were most frequent at the splenic flexure and cecum; codon 13 mutations were evenly distributed. BRAF mutation frequency sharply increased from transverse colon to cecum in parallel with deficient (d) MMR. Nonmutated BRAF and KRAS tumors progressively decreased from sigmoid to transverse (all P < 0.0001). Significantly, poorer OS was found for mutant KRAS in distal HR, 1.98; 95% confidence interval (CI), 1.49-2.63; P < 0.0001 versus proximal (1.25; 95% CI, 0.97-1.60; P = 0.079) cancers. BRAF status and outcome were not significantly associated with tumor site. Proximal versus distal dMMR tumors had significantly better outcome. An interaction test was significant for tumor site by KRAS (P(adjusted) = 0.043) and MMR (P(adjusted) = 0.010) for OS. Significant prognostic differences for biomarkers by tumor site were maintained in the FOLFOX arm. Tumor site was independently prognostic with a stepwise improvement from cecum to sigmoid (OS: P(adjusted) = 0.001). CONCLUSIONS: Mutation in BRAF or KRAS codon 12 was enriched in proximal cancers whereas nonmutated BRAF/KRAS was increased in distal tumors. Significant differences in outcome for KRAS mutations and dMMR were found by tumor site, indicating that their interpretation should occur in the context of tumor location.