Development and Validation of an Ileocolonoscopy-Based Nomogram for Predicting Perianal Penetrating Complications in Patients with Crohn’s Disease

Siyuan Zhang,1 Jianmin Wang,1 Qing Lin,2 Yuqin Sheng,2 Ming Li1,3 1Department of Anorectal, The First Affiliated Hospital of Anhui University of Chinese Medicine, Hefei, Anhui, 230031, People’s Republic of China; 2Department of Nursing, Anhui Branch of Shuguang Hospital Affiliated to Shanghai University of Chinese Medicine, Hefei, Anhui, 230031, People’s Republic of China; 3Department of Anorectal, Anhui Branch of Shuguang Hospital Affiliated to Shanghai University of Chinese Medicine, Hefei, Anhui, 230031, People’s Republic of ChinaCorrespondence: Ming Li, Department of Anorectal, The First Affiliated Hospital of Anhui University of Chinese Medicine, Hefei, Anhui, 230031, People’s Republic of China, Email [email protected]: Perianal penetrating complications (PPC) in Crohn’s disease (CD) are inadequately predicted. PPC risk correlates with ileocolonoscopic scores, however, the association between specific ileocolonoscopic features and it remains unclear. This study aimed to identify predictive ileocolonoscopic features and develop a dedicated PPC nomogram, thereby enabling proactive management of high-risk patients. This tool is designed as a prediction model, not a clinical decision-making instrument.Methods: CD patients from two centers between January 1, 2012 and July 31, 2025 are enrolled: the First Affiliated Hospital of Anhui University of Chinese Medicine (FAHAUTCM, Center 1) and Anhui Branch of Shuguang Hospital Affiliated to Shanghai University of Chinese Medicine (ABSHASUTCM, Center 2). Their demographic and ileocolonoscopic data were collected. Center 1 enrolled patients (n=431) were randomized to training (n=301) and internal validation (n=130) sets at a 7:3 ratio; Center 2 enrolled patients (n=127) served as the external validation set. The Boruta algorithm and Least Absolute Shrinkage and Selection Operator (LASSO) regression identified the most predictive features, which were incorporated into a multivariable logistic regression. Developed model was appraised via Receiver Operating Characteristic (ROC) curves, calibration curves, Decision Curve Analysis (DCA) and nomogram score distribution.Results: Multivariate logistic regression confirmed five independent predictors: the largest ulcer diameter (OR=1.504, 95% CI=1.095– 2.097, P< 0.05), ulcer area in rectum (OR=2.900, 95% CI=2.188– 3.955, P< 0.001), ulcer area in descending colon (OR=1.402, 95% CI=1.005– 1.965, P< 0.05), nodular lesions (OR=1.976, 95% CI=1.451– 2.751, P< 0.001), and stenosis (OR=2.544, 95% CI=1.765– 3.789, P< 0.001). The model achieved AUCs of 0.857 (internal validation) and 0.847 (external validation), with favorable calibration (P=0.240 and 0.498 for Hosmer–Lemeshow tests, respectively). DCA and nomogram score distribution further verified the model’s clinical utility.Conclusion: We identified several ileocolonoscopic predictors and developed a nomogram which showed good predictive accuracy. This nomogram overcomes the limitations of common scoring systems, which assign equal weight to all intestinal segmental lesions. It enables rapid clinical risk stratification without complex calculations and helps clinicians consider personalized surveillance.Keywords: Crohn’s disease, ileocolonoscopy, prediction nomogram model, complications