Development and Internal Validation of a LASSO-Based Prediction Model for Colorectal Adenoma Recurrence After Polypectomy.

BACKGROUND: Colorectal adenoma recurrence after polypectomy remains an important clinical concern, with current surveillance strategies based primarily on index adenoma characteristics. This study aimed to develop and validate a prediction model integrating clinical and metabolic factors for personalized recurrence risk assessment. METHODS: We conducted a retrospective cohort study of 328 patients with colorectal adenomas confirmed by pathology between January 2018 and December 2021. Variable selection was performed using LASSO-penalized Cox regression with 10-fold cross-validation. Model performance was assessed through bootstrap validation (1000 resamples) with calculation of Harrell's C-index, calibration curves, and decision curve analysis. RESULTS: The final model included age (HR 1.28, 95% CI 1.12-1.47), alcohol consumption history (HR 2.12, 95% CI 1.68-2.67), and bile acid levels (mean 3.14 ± 0.85 μmol/L in the recurrence group vs 2.73 ± 0.78 μmol/L in the non-recurrence group). The model demonstrated good discrimination (bootstrap-corrected C-index 0.878, 95% CI 0.843-0.912) and calibration (slope 0.894, 95% CI 0.851-0.937). CONCLUSION: The developed prediction model integrating age, alcohol history, and bile acid levels provides a practical tool for stratifying recurrence risk after polypectomy, with potential to guide personalized surveillance strategies. External validation is warranted to confirm these findings.