Plasma proteomic markers predict risk for bowel resection in inflammatory bowel disease: a retrospective cohort study.

Bowel resection is required in up to 27% of inflammatory bowel disease (IBD) patients due to severe complications. Identifying reliable predictors for future surgery is crucial for risk stratification and management. We retrospectively analyzed UK Biobank data of IBD patients with plasma proteomics measured using the Olink platform. A Random Survival Forest (RSF) model was used to evaluate the predictive performance of inflammatory proteins for surgical risk. The concordance index (C-index) compared Olink-based and conventional clinical models. A predicted risk score quantified individual surgery risk, and Kaplan-Meier analysis assessed differences in surgery-free survival between high- and low-risk groups. Cox proportional hazards models verified the independent predictive value of proteomics. Among 915 patients with IBD, 81 cases underwent bowel resection during follow-up. The RSF model using Olink proteins achieved a superior C-index compared to conventional markers (0.784 vs 0.543, P < 0.01). Key predictors included IL15, PAPPA, and PIK3AP1. Patients with a higher predicted risk score had a significantly higher surgery risk ( P < 0.0001). The predicted risk score remained an independent predictor for surgery risk [adjusted HR: 0.83, 95% confidence interval (CI): 0.81-0.86]. The net reclassification analysis demonstrated improved patient-level stratification by proteomic features beyond standard clinical predictors (net reclassification improvement: 0.096, 95% CI: 0.008-0.257, P = 0.027). Plasma proteomic markers demonstrated superior predictive performance over traditional markers in assessing future bowel resection risk in IBD. Incorporating proteomics into clinical practice could enhance risk stratification and inform early interventions to reduce surgical burdens.