6%; Table 2) that did not differ significantly from dMMR tumors of the sporadic subtype or pMMR tumors lacking BRAFV600E and KRAS mutations ( Table 4). Of note, DFS for dMMR tumors of the familial subtype was poorer among distal vs proximal tumors ( Figure 2A and B). Among distal pMMR cancers, statistically significant differences in DFS were found only for KRAS-mutated tumors (vs those without KRAS and BRAF mutations), yet statistical

power was limited ( Table 4). A trend toward better DFS was found in distal vs proximal tumors with BRAFV600E mutations and tumors without BRAFV600E or KRAS mutations ( Table 2). Among patients with N1 tumors, the association of tumor subtypes with DFS did not differ significantly from the overall Talazoparib cohort (Figures 1B and 2C). Among patients with N2 tumors, however, poor DFS was observed for dMMR tumors of the sporadic subtype ( Table 2, Figure 2D) that did not differ significantly from DFS of pMMR subtypes with mutated KRAS (Padjusted = .9195) or mutated

BRAFV600E (Padjusted = .8231) ( Table 4). In contrast, N1 tumors of the dMMR sporadic subtype had DFS rates that were significantly improved find more compared with DFS of patients with pMMR mutated KRAS tumors (HR = 0.51; 95% CI: 0.31–0.82; Padjusted = .0054), or showed a strong trend vs the mutated BRAFV600E (HR = 0.50; 95% CI: 0.28–0.91; Padjusted = .0238) subtype ( Figure 2C). We attempted to validate the prognostic Dapagliflozin utility of our classifier in an independent

cohort of stage III colon cancer patients treated with 5-FU–based adjuvant chemotherapy. Patients from this external cohort were categorized into the same molecular subtypes as in our dataset, with the exception that dMMR tumors were divided based on BRAF status alone (see Materials and Methods). In this independent cohort, a statistically significant difference was seen among the 5 molecular subtypes (P = .014) as was demonstrated in the primary N0147 cohort ( Figure 3). A similarly favorable outcome for pMMR tumors lacking BRAFV600E or KRAS mutations and dMMR tumors was observed. In addition, poorer DFS among patients with BRAFV600E mutant or KRAS mutant pMMR cancers was observed as reflected in their 5-year DFS rates ( Figure 3, Table 2). Accordingly, the key prognostic findings of our biomarker classifier were validated. In patients undergoing surgical resection of CRC, prognosis and management are based entirely on the TNM staging system,24 despite considerable stage-independent variability in outcomes. Accordingly, prognostic classifiers that can be readily implemented into clinical practice are needed to enhance clinical decision making. In stage III colon cancers from a recent adjuvant chemotherapy trial,26 we classified tumors into 5 prespecified subtypes using a biomarker combination of BRAFV600E and KRAS mutations, MLH1 methylation, and MMR status.