Establishment of a recurrence prediction model after endoscopic resection of colorectal serrated lesions

Colorectal serrated lesions are precursors to colorectal cancer, with a high recurrence risk after endoscopic resection. Identifying risk factors for recurrence and establishing a prediction model are crucial for optimizing surveillance strategies.

A single-center retrospective study included 164 patients with colorectal serrated lesions who had endoscopic follow-up information after the surgery. These cases were divided into a polyp recurrence group (79 cases) and a non-recurrence group (85 cases). Demographic data, medical histories, laboratory findings, endoscopic characteristics of the lesions, and follow-up durations were therefore collected. Cox regression identified independent risk factors, which were incorporated into a nomogram. Model performance was evaluated using C-index, calibration curves, and decision curve analysis.

Multivariate analysis identified non-alcoholic fatty liver disease (HR 1.813, 95%CI 1.080-3.045), atrophic gastritis (HR 1.733, 95%CI 1.087-2.760), history of gallbladder abnormalities (HR 1.791, 95%CI 1.107-2.898) and elevated total bile acid levels (HR 1.023, 95%CI 1.003-1.043) as independent predictors of recurrence. The nomogram demonstrated good discrimination (C-index: 0.703-0.820) and clinical applicability (decision curve analysis net benefit: 27-80% threshold).

This study established a validated prediction model for colorectal serrated lesions recurrence, highlighting metabolism-related factors as key determinants. Targeting these factors and applying the nomogram may enhance risk stratification and postoperative surveillance, potentially reducing colorectal cancer progression. Limitations include the retrospective design, single-center data, and lacking further validations.