Prediction of metachronous gastric neoplasms after curative endoscopic submucosal dissection for early gastric cancer

Abstract

Abstract Background: As more early gastric cancer (EGC) patients are being treated with endoscopic submucosal dissection (ESD), it is important to find the risk factors which contribute to the occurrence of metachronous gastric neoplasms (MGN). This study aimed to analyze the characteristics of MGN and to generate useful tools for the prediction of MGN following ESD. Methods: A total of 819 patients were retrospectively enrolled. The risk factors for MGN were analyzed using Cox hazard proportional model. Moreover, RStudiosoftware is used to build nomogram models, and evaluate the discrimination, consistency and decision curves. Results: A total of 5.0% (41/819) of patients had MGN after ESD for EGC. The median interval between the discovery of MGN and the initial EGC was 23 months. Cox multivariable analysis found that history of drinking [HR:2.486(1.082-5.710), p=0.032], location of lesion (middle 1/3) [HR: 2.613(1.192-5.725), p=0.016], node metastasis [HR: 3.152(1.046-9.502), p=0.041] and whole lesion removal(not) [HR: 2.526(1.029-6.200), p=0.043] were independent risk factors for the development of MGN. In addition, we were developed and internally validated nomogram to predict the 3-, 5-, and 7-year probability of developing MGN after ESD for EGC (C index= 0.665). The calibration chart showed that the predicted probability of MGN after ESD for EGC was the similar to the actual observed result and the DCA decision curve analysis showed strong clinical practicability. Conclusions: We constructed a nomogram for predicting the incidence of MGN after ESD in EGC patients and the performance of the model is excellent.