Xiehe Yixue Zazhi (Jan 2024)

Establishment and Validation of Prediction Models for Non-curative Resection After ESD for Early Gastric Cancer

  • DONG Na,
  • MA Ganqing,
  • WANG Lulu,
  • SHI Ronghui,
  • FENG Jie,
  • HUANG Xiaojun

DOI
https://doi.org/10.12290/xhyxzz.2023-0442
Journal volume & issue
Vol. 15, no. 1
pp. 109 – 116

Abstract

Read online

Objective To establish a prediction model for non-curative resection in patients with early gastric cancer (EGC) who underwent endoscopic submucosal dissection (ESD), and to evaluate its predictive value. Methods Clinical data of EGC patients in the Second Hospital & Clinical Medical School, Lanzhou University from January 2014 to July 2023 were retrospectively collected. According to the postoperative pathological results of ESD, the patients were divided into curative resection group and non-curative resection group. Multifactorial Logistic regression analysis was used to screen the risk factors for non-curative resection after ESD surgery and establish a prediction model, and the model was evaluated using receiver operating characteristic(ROC) curves, calibration curves and clinical decision curve analysis. Results A total of 479 EGC patients who underwent ESD were included, with 60 cases in the non-curative resection group and 419 cases in the curative resection group. The results of multifactorial Logistic regression analysis showed that the lesion diameter > 2 cm (OR=3.017, 95% CI: 1.483-6.136, P=0.002), flat lesion morphology (OR=2.712, 95% CI: 0.774-9.497, P=0.043), undifferentiated/mixed histologic type (OR= 4.199, 95% CI: 1.621-10.872, P=0.003), and submucosal infiltration (OR=30.329, 95% CI: 13.059-70.436, P < 0.001) were independent risk factors for non-curative resection after ESD in EGC patients. The area under the curve of ROC validated within the column-line graph prediction model constructed accordingly was 0.867 (95% CI: 0.811-0.923), the calibration curve showed that the model had good calibration, and decision curve analysis showed the model had a good clinical usefulness. Conclusions The prediction model constructed based on lesion diameter, lesion morphology, histologic type, and depth of mucosal infiltration has good differentiation, calibration, and clinical utility. This model is expected to assist in the early clinical screening of the population at high risk for noncurative resection after ESD in patients with EGC, and to provide a basis for the development of optimal clinical decisions.

Keywords