P0322Superior Performance of a Fecal Calprotectin and Intestinal Ultrasound Model for Predicting Endoscopic Activity in Ulcerative Colitis

Background The management of ulcerative colitis (UC) requires simple, reliable, non-endoscopic tools for assessing disease activity. Combining biomarkers like fecal calprotectin (FC) with intestinal ultrasound (IUS) represents a promising composite strategy to enhance predictive accuracy. Therefore, this study aimed to determine which of two modeling strategies offers superior performance for predicting endoscopic activity: one using FC alone versus another combining FC with IUS, with both incorporating clinical parameters. Methods We performed a retrospective, single-center study including consecutive UC patients (May 2023 - Sep 2025). Diagnostic performance for moderate-to-severe endoscopic activity (Mayo score ≥2) was calculated. Agreement was assessed with Kappa statistics, and AUCs were compared via DeLong’s test. Predictive models were developed using multivariate logistic regression (backward selection on variables with p < 0.10 from univariate analysis) and evaluated for discrimination (C-index) and calibration (Hosmer-Lemeshow test) following internal validation (bootstrapping) and external validation. Results A total of 171 UC patients were analyzed. The combination of FC and IUS (AUC: 0.719, Kappa: 0.433) demonstrated superior diagnostic performance over FC alone (AUC: 0.692, Kappa: 0.391) for predicting moderate-to-severe endoscopic activity. This composite strategy was then integrated into predictive models. In the derivation (n = 129) and validation (n = 43) cohorts, all four models based on FC combined with IUS consistently achieved significantly higher discriminatory power (AUCs: 0.747-0.876) compared to the corresponding FC-only models (AUCs: 0.649-0.857), alongside a consistent trend toward higher sensitivity while maintaining comparable specificity. All models were well-calibrated (all p < 0.05 by Hosmer-Lemeshow test). The predictive nomogram derived from the optimal model successfully validated these findings externally. Conclusion This study validates that a composite biomarker strategy integrating FC and IUS outperforms FC-only assessments. The developed and validated nomogram translates this finding into a practical clinical tool, which can aid in non-invasive disease monitoring and optimize management strategies for UC patients. Conflict of interest: Ms. Tang, Xiaoyan: No conflict of interest Yang, Hong: No conflict of interest