Collagen score in the tumor microenvironment predicts the prognosis of rectal cancer patients after neoadjuvant chemoradiotherapy

Background and purpose Little is known about the relationship between collagen and the prognosis of patients with locally advanced rectal cancer (LARC) after neoadjuvant chemoradiotherapy (nCRT). This study aimed to quantitatively analyze collagen alterations, establish a collagen score (CS) in the tumor microenvironment, and evaluate and validate the relationship of the CS with prognosis in these patients. Materials and methods A total of 365 primary patients diagnosed with LARC after nCRT between 2011 and 2018 were retrospectively reviewed (training cohort: 210; independent validation cohort: 155). Multiple collagen features of two fields in the tumor microenvironment, the core of the tumor (CT) and the invasive margin (IM), were derived from multiphoton imaging, and the CSIM-CT was generated using least absolute shrinkage and selection operator (LASSO) Cox regression analysis. Results The CSIM-CT was created based on 3 features: collagen area, number of collagen fibers and a Gabor textural feature. In the training cohort, the CSIM-CT predicted 3-year disease-free survival (DFS) with an area under the receiver operating characteristic curve (AUROC) of 0.765 (0.675–0.854) and an overall survival (OS) with AUROC of 0.822 (0.734–0.909). Additionally, the CSIM-CT was significantly associated with DFS and OS in the two cohorts. A nomogram with the CSIM-CT was developed and showed good prognostic value predicting a 3-year DFS with an AUROC of 0.826 (0.748–0.905) and an OS with AUROC of 0.882 (0.803–0.960). Conclusions The CSIM-CT is an effective prognostic marker in patients with LARC after nCRT, and the nomogram with the CSIM-CT can be used to accurately predict the individual prognosis of these patients.