早期胃癌浸润深度的主要影响因素分析及预测模型的构建

To explore the factors that affect the depth of invasion in early gastric cancer (EGC) and establish a nomogram model to predict the depth of invasion. Methods: A total of 451 patients who underwent endoscopic submucosal dissection (ESD) or surgery and were pathologically diagnosed with early gastric cancer at The Second Affiliated Hospital of Xi'an Jiaotong University from January 2011 to December 2019 were selected as the study subjects. The clinicopathological features of the mucosal group and submucosal group were compared. Multivariate Logistic regression analysis was used to select variables to construct a predictive model, and internal tests were carried out to evaluate the feasibility and accuracy of the model. Results: A total of 409 cases of EGC were included in the study, of these, the cancer was confined to the mucous membrane in 240 cases and in 169 cases the submucosa had been affected. Multivariate Logistic analysis revealed that lesion length >20 mm, marginal eminence, ulcer, non-intestinal metaplasia, mixed histology, and undifferentiated type were all independent risk factors for submucosal infiltration. Compared with the degree of differentiation, the predictive model based on histological characteristics combined with other risk factors had a better predictive performance. The nomogram model established on this basis to predict the depth of invasion of EGC was internally verified using the Bootstrap method, and its consistency index was 0.776. The correction curve showed that there was a good correlation between the predicted depth of invasion and the actual depth of invasion. Conclusions: The nomogram model based on lesion size, histology, marginal eminence, intestinal metaplasia, and ulcer is efficient at predicting the depth of invasion of EGC and has a certain clinical reference value.