11, 12, 14, 17 and 35 The prognostic impact of oncogenic KRAS in stage II and III colon cancers has been inconsistent, 9, 12, 14, 17, 46, 47 and 48 and BRAFV600E mutations have generally been associated with adverse outcomes, particularly

in metastatic CRCs. 12, 14, 15, 18, 47, 49 and 50 Importantly, we were able to validate the key findings for the prognostic impact of our subtype classifier in an independent cohort of stage III colon cancer patients treated with 5-FU–based adjuvant chemotherapy. This finding supports the robustness of our classifier to detect clinically significant prognostic differences. Patients in our study cohort were treated with the current standard adjuvant FOLFOX regimen, and only limited data are available for the prognostic impact of the biomarkers studied here in FOLFOX-treated patients.12 and 19 buy Fluorouracil CP868596 Important strengths of our study include the large size of our clinical trial cohort with uniform treatment, meticulous follow-up data, and an external validation cohort. Our subtype classifier capitalizes on common testing for KRAS and BRAF status in clinical practice

and the recommendation for universal MMR/MSI testing by the National Comprehensive Cancer Network. Limitations include the retrospective design and inability to examine the predictive potential of our subtype classifier with respect to treatment response. Although an effort was made to control for multiple comparisons during the study planning stage by utilizing well-established biomarkers whose classification was supported by the literature, pairwise comparisons with P values that are close to the .05 significance level should be interpreted with caution and their clinical significance considered. Thiamet G We acknowledge that other molecular events within the subtypes may indeed impact prognosis or chemosensitivity, which can contribute to the observed subtype-specific survival differences. A potential confounder is the use of aspirin or other nonsteroidal anti-inflammatory drugs

and using questionnaire data that were available from a subset of the study population (n = 1757), no evidence was found to indicate that use of these drugs modified the association between subtypes and DFS. In conclusion, we found that a biomarker-based classifier can identify prognostically distinct subtypes within stage III colon cancer patients that was externally validated. We identified a phenotype associated with BRAFV600E mutations and pMMR that was clinically aggressive as was the mutant KRAS subtype. The pMMR subtype without BRAF or KRAS mutations accounted for nearly half of our study cohort and had a favorable prognosis that did not differ significantly from dMMR cancers.