Calibrated Absolute Risk of Lymph-Node-Metastasis After Non-curative Endoscopic Resection of pT1 Colorectal Cancer

After endoscopic resection of T1 colorectal cancer (CRC), the decision to recommend completion radical surgery is primarily driven by pathologic "high-risk" features. However, these binary criteria identify broad heterogeneous groups and do not provide individualized estimates of lymph-node-metastasis (LNM) risk.

This study analyzed patients with pathologic T1 CRC who underwent radical resection after endoscopic resection from 2004 to 2024 at a single tertiary center. Five histologic features (lymphatic/venous/perineural invasion, tumor budding, poor differentiation, submucosal invasion ≥2000 μm, and positive resection margin) were assigned one point each to construct a composite pathologic score (0-5).

Among 1162 patients, LNM occurred for 148 patients (12.7 %). The composite score showed a stepwise gradient in LNM risk as follows: 6.6 % for score 0, 12.0 % for score 1, 29.2 % for score 2, and 66.7 % for scores 3 to 4. A threshold of ≥2 identified a high-risk group with substantially higher LNM prevalence than scores 0 to 1 (35.1 % vs 9.5 %; odds ratio [OR], 4.79), corresponding to an absolute risk difference of 24 % and number-needed-to-surgery (NNS) of 4.1. The score demonstrated acceptable discrimination (area under the curve, 0.673; 95 % confidence interval, 0.624-0.722) and good calibration, with close agreement between predicted and observed probabilities across all strata (29 % for score 2, 60 % for score 3, and 100 % for score 4).

A simple composite pathologic score integrating five adverse features provides a calibrated, clinically interpretable estimate of LNM risk after endoscopic resection of T1 CRC. By translating routine pathology into absolute risk and NNS, this model offers a practical framework to support individualized recommendations for completion radical surgery.