Endoscopic prediction model for detecting advanced rectal polyps after ileorectal anastomosis in familial adenomatous polyposis: a dual-center retrospective study.

Familial adenomatous polyposis (FAP) patients who undergo ileorectal anastomosis (IRA) remain at risk for advanced rectal neoplasia. Early identification of polyps < 10 mm is crucial to prevent metachronous rectal cancer. This study aimed to develop an endoscopic prediction model using mucosal phenotypic features to identify advanced diminutive rectal polyps in FAP patients post-IRA. This dual-center retrospective study included 48 FAP patients who underwent IRA and endoscopic surveillance from 2018 to 2024. A total of 452 rectal polyps < 10 mm were evaluated using high-definition white-light endoscopy, NBI, dye-based chromoendoscopy, and magnifying endoscopy. Data from Center 1 were split into a training set (80%) and validation set (20%), while Center 2 served as an external validation cohort. Significant variables (P < 0.05) from univariate analysis were entered into multivariable logistic regression. In the overall cohort and the APC pathogenic mutation subgroup, nomograms were developed and evaluated using ROC curves, calibration curves, and decision curve analysis (DCA). Independent predictors of advanced polyps included chicken-skin mucosa (OR = 2.50, P < 0.01), JNET type 2 A-2B low (OR = 3.39, P <  0.01), erosion/ulceration (OR = 1.68, P = 0.02), and pit pattern III-L/IV (OR = 2.47, P < 0.01). The nomogram showed good discriminative ability with AUCs of 0.822 (training), 0.806 (internal validation), and 0.815 (external validation). Separate prediction models were developed for all IRA-postoperative FAP patients and the APC pathogenic mutation subgroup, integrating four endoscopic features to accurately identify advanced diminutive rectal polyps, thereby reducing the risk of malignancy in patients.