Fibrosis signature of anastomotic margins for predicting anastomotic stenosis in rectal cancer with neoadjuvant chemoradiotherapy and sphincter-preserving surgery.

Abstract

Radiation-induced colorectal fibrosis (RICF) is a common pathological alteration among patients with rectal cancer undergoing neoadjuvant chemoradiotherapy (nCRT). Anastomotic stenosis (AS) causes symptoms and negatively impacts patients' quality of life and long-term survival. In this study, we aimed to evaluate the fibrosis signature of RICF and develop a nomogram to predict the risk of AS in patients with rectal cancer undergoing nCRT. Overall, 335 pairs of proximal and distal margins were collected and randomly assigned at a 7:3 ratio to the training and testing cohorts. The RICF score was established to evaluate the fibrosis signature in the anastomotic margins. A nomogram based on the RICF score for AS was developed and evaluated by using the area under the curve, decision curve analysis, and the DeLong test. The training cohort included 235 patients (161 males [68.51%]; mean age, 59.61 years) with an occurrence rate of AS of 17.4%, whereas the testing cohort included 100 patients (72 males [72.00%]; mean age, 57.17 years) with an occurrence rate of AS of 11%. The RICF total score of proximal and distal margins was significantly associated with AS (odds ratio, 3.064; 95% confidence interval [CI], 2.200-4.268; P < 0.001). Multivariable analysis revealed that the RICF total score, neoadjuvant radiotherapy, and surgical approach were independent predictors for AS. The nomogram demonstrated good discrimination in the training cohort (area under the receiver-operating characteristic curve, 0.876; 95% CI, 0.816-0.937), with a sensitivity of 68.3% (95% CI, 51.9%-81.9%) and a specificity of 85.5% (95% CI, 78.7%-89.3%). Similar results were observed in the testing cohort. This study results suggest that the RICF total score of anastomotic margins is an independent predictor for AS. The prediction model developed based on the RICF total score may be useful for individualized AS risk prediction in patients with rectal cancer undergoing nCRT and sphincter-preserving surgery.