Association of the Collagen Signature in the Tumor Microenvironment With Recurrence and Survival of Patients With T4N0M0 Colon Cancer.

BACKGROUND The current clinicopathological risk factors do not accurately predict disease recurrence in patients with T4N0M0 colon cancer. We hypothesized that the collagen signature combined with clinicopathological risk factors (new model) had a better prognostic value than clinicopathological risk factors alone (clinicopathological model). OBJECTIVE This study aimed to establish a collagen signature in the tumor microenvironment and to validate its role in predicting the recurrence of T4N0M0 colon cancer. DESIGN This was a retrospective study. SETTINGS This study took place at a tertiary medical center. PATIENTS Patients with T4N0M0 colon cancer who underwent surgery at our center between 2009 and 2015 (n=416) were included. INTERVENTION A total of 142 collagen features were analyzed in the tumor microenvironment in specimens of colon cancer using second harmonic generation imaging. A collagen signature was constructed using a LASSO Cox regression model. MAIN OUTCOME MEASURES Disease-free survival and overall survival. RESULTS The training and testing cohorts consisted of 291 and 125 randomly assigned samples, with recurrence rates of 19.9% and 22.4%, respectively. A 3-feature-based collagen signature predicted the recurrence risk at 1, 3, and 5 years, with the area under the receiver-operating characteristic curves of 0.808, 0.832, and 0.791 in the training cohort and 0.836, 0.807, and 0.794 in the testing cohort, respectively. Multivariate analysis revealed that the collagen signature could independently predict the disease-free survival (HR=7.17, p<0.001) and overall survival rates (HR=5.03, p<0.001). The new model had a better prognostic value than the clinicopathological model, which included four clinicopathological risk factors: obstruction or perforation, lymphovascular invasion, tumor budding, and no chemotherapy. LIMITATIONS This study was limited by its retrospective design. CONCLUSIONS The collagen signature in the tumor microenvironment may be a new prognostic marker that can effectively predict the recurrence and survival of patients with T4N0M0 colon cancer. See Video Abstract at http://links.lww.com/DCR/B503 .