Analysis of tumor microenvironmental features to refine prognosis by T, N risk group in patients with stage III colon cancer (NCCTG N0147) (Alliance)

Background: Tumor-infiltrating lymphocytes (TILs), tumor budding, and micropapillary architecture may influence tumor growth and metastatic potential, thereby enhancing prognostic stratification. We analyzed these features and their relative contribution to overall outcome and in low (T1–3 N1) and high (T4 and/or N2) risk groups that are used to inform the duration of adjuvant chemotherapy in patients with resected stage III colon cancers. Patients and methods: Among 1532 patients treated in a phase III adjuvant trial of FOLFOX-based therapy, intraepithelial TIL densities, tumor budding, and micropapillary features were analyzed and quantified in routine histopathological sections with light microscopy. Optimal cut-points were determined in association with disease-free survival (DFS) in training and validation sets. Associations or relative contributions of individual features or combined variables with DFS were determined using multivariable Cox regression models. Results: TILs, tumor budding, and micropapillary features were shown to differ significantly by T, N risk groups and by mismatch repair (MMR) status. Low TILs, high budding, and their combined variable [hazard ratio = 2.07 (95% CI, 1.50% to 2.88%); P_adj < 0.0001], but not micropapillary features, were each significantly associated with poorer DFS in a training data set and confirmed in a validation set. TILs were prognostic in proficient mismatch repair (pMMR) and deficient mismatch repair (dMMR) tumors; budding was prognostic only in pMMR tumors. The percentage relative contribution of budding/TILs to DFS was second only to nodal status overall, was second (24.4%) after KRAS in low-risk patients, and was the most important contributor (45.4%) in high-risk patients. Conclusions: TIL density and tumor budding were each validated as significant prognostic variables and their combined variable provided robust prognostic stratification by T, N risk groups, being the strongest predictor of DFS among high-risk stage III patients. ClinicalTrials.gov Identifier: NCT00079274.