Abstract
OBJECTIVE: To validate the predictive power of newly developed magnetic resonance (MR) morphological and clinicopathological risk models in predicting low anterior resection syndrome (LARS) 6 months after anterior resection of middle and low rectal cancer (MLRC). METHODS: From May 2018 to January 2021, 236 patients with MLRC admitted to two hospitals (internal and external validation) were included. MR images, clinicopathological data, and LARS scores (LARSS) were collected. Tumor morphology data included longitudinal involvement length, maximum tumor diameter, proportion of tumor to circumference of the intestinal wall, tumor mesorectal infiltration depth, circumferential margin status, and distance between the tumor and anal margins. Pelvic measurements included anorectal angle, mesenterial volume (MRV), and pelvic volume. Univariate and multivariate logistic regression was used to obtain independent risk factors of LARS after anterior resection Then, the prediction model was constructed, expressed as a nomogram, and its internal and external validity was assessed using receiver operating characteristic curves. RESULTS: The uni- and multivariate analysis revealed distance between the tumor and anal margins, MRV, pelvic volume, and body weight as significant independent risk factors for predicting LARS. From the nomogram, the area under the curve (AUC), sensitivity, and specificity were 0.835, 75.0 %, and 80.4 %, respectively. The AUC, sensitivity, and specificity in the external validation group were 0.874, 83.3 %, and 91.7 %, respectively. CONCLUSION: This study shows that MR imaging and clinicopathology presented by a nomogram can strongly predict LARSS, which can then individually predict LARS 6 months after anterior resection in patients with MLRC and facilitate clinical decision-making. CLINICAL RELEVANCE STATEMENT: We believe that our study makes a significant contribution to the literature. This method of predicting postoperative anorectal function by preoperative measurement of MRV provides a new tool for clinicians to study LARS.