Risk factors and prediction model for post-polypectomy metachronous colorectal adenoma

Endoscopic resection of colorectal adenomas helps reduce colorectal cancer mortality. However, post-polypectomy metachronous adenomas may reduce the sufficiency of colonoscopy scanning.

To explore the risk factors for post-polypectomy metachronous colorectal adenoma and establish a risk prediction model.

This retrospective cohort study included patients who underwent colonoscopy at Beijing Friendship Hospital from January 2013 to January 2023. Data on patients' demographics, laboratory results, colonoscopy findings, and pathology reports were collected. The enrolled patients were randomly divided into a training set and a validation set in a 7:3 ratio. LASSO analysis was employed to identify risk factors for metachronous adenomas. Based on these risk factors, a Cox regression model was used to create a risk prediction model. The C-index was calculated to assess the model's prediction accuracy, while time-dependent ROC (td-ROC) curves and calibration curves evaluated model predictive performance. A Decision Curve Analysis (DCA) was conducted to assess clinical utility.

A total of 523 patients meeting the inclusion and exclusion criteria were enrolled and randomly divided into a training set (n = 366) and a validation set (n = 157). Twenty-one clinical and pathological features were included in the LASSO regression analysis. Among these, age, gender, baseline adenoma size, baseline adenoma location, baseline pathological grade, history of hypertension, and serum LDL level were included in the multivariable Cox regression analysis, leading to the establishment of a visualized nomogram model. The C-index of the predictive model was 0.729 (95% CI: 0.686, 0.772) in the training set and 0.724 (95% CI: 0.667, 0.781) in the validation set. The time-dependent ROC curve and calibration curve indicated good reliability of the model, and the DCA curve suggested satisfactory clinical utility. An online webserver was also constructed to visualize the model and facilitate the calculation of metachronous adenoma risk for clinicians (URL: https://colorectal-metachronous-adenoma-prediction.shinyapps.io/DynNomapp/ ).

A total of 7 risk factors for post-polypectomy metachronous adenoma were identified. A risk prediction model that possesses good prediction accuracy and good clinical utility was established, providing a reliable tool for patient risk stratification.