Development and validation of web-based prediction models for major low anterior resection syndrome at 3 and 6 months after sphincter-preserving surgery in patients with rectal cancer

OBJECTIVE: To analyse the factors influencing major low anterior resection syndrome (LARS) at 3 months and 6 months after surgery in rectal cancer patients undergoing sphincter-preserving procedures, and separately develop and validate risk prediction models for the 3-month and 6-month postoperative periods. METHODS: This study enrolled patients who underwent radical sphincter-preserving surgery for rectal cancer at the First Affiliated Hospital of the University of Science and Technology of China between August 2017 and September 2024. The LARS scale was used to assess bowel function at 3 months and 6 months postoperatively. In total, 794 patients at 3 months postoperatively and 749 patients at 6 months postoperatively were included and randomly allocated to a training set and a validation set in a 7:3 ratio. Least absolute shrinkage and selection operator (LASSO) regression and multivariable logistic regression were used to identify factors influencing major LARS at 3 months and 6 months postoperatively, and to separately develop risk prediction models for the two time points. Receiver operating characteristic (ROC) curves, calibration curves, and clinical decision curve analysis were applied to evaluate the discrimination, calibration, and clinical applicability of the two models. RESULTS: Multivariable logistic regression demonstrated that preoperative chemotherapy, preoperative radiotherapy, and tumour distance from the anal verge were common risk factors associated with major LARS at 3 months and 6 months postoperatively in rectal cancer patients undergoing sphincter-preserving surgery. Anastomotic leakage was identified as a risk factor specific to 3 months postoperatively, whereas prophylactic stoma was a risk factor specific to 6 months postoperatively. Based on these findings, risk prediction models for major LARS at 3 months and 6 months postoperatively were developed, and interactive web-based calculators were developed using the RShiny platform, accessible at https://zhang13579.shinyapps.io/Postoperative_3-Month/ and https://zhang13579.shinyapps.io/Postoperative_6-Month/, respectively. Both models demonstrated good predictive performance in the training cohort and the validation cohort, as assessed by discrimination, calibration, and clinical applicability. CONCLUSION: The risk prediction models for major LARS at 3 months and 6 months postoperatively in rectal cancer patients undergoing sphincter-preserving procedures developed in this study demonstrated good predictive performance and can assist clinical healthcare professionals in identifying patients at high risk of major LARS.