预测T4a期胃癌腹膜转移危险因素的列线图模型的建立与验证

目的探究T_4a期胃癌发生腹膜转移的危险因素，以此为训练组构建列线图模型，并进行内验证及外验证。 方法回顾性分析2011年1月至2017年12月394例行胃切除术+D2淋巴结清扫术影像学诊断为T_4a期胃癌患者的临床资料，其中将2011年1月至2015年12月224例患者作为训练集，2016年1月至2017年12月170例患者作为验证集。根据术前影像学表现判断有无腹膜转移并且最终经病理资料证实，采用SPSS 24.0软件通过t检验、秩和检验或卡方检验对发生腹膜转移的危险因素进行统计学分析，经单因素和多因素Logistic回归筛选T_4a期胃癌患者发生腹膜转移的潜在危险因素，采用R软件（版本4.0.2）建立列线图模型。采用Bootstrap法进行内验证，采用ROC曲线评价模型的符合度并计算95%CI，绘制校准曲线评价模型的符合度。绘制DCA曲线评价模型的临床获益度。 结果224例训练集患者中，共37（16.5%）例患者发生腹膜转移，验证集患者中23（13.5%）例患者发生腹膜转移，多因素分析显示糖类抗原125（CA125）、腹水、术前白蛋白（ALB）和肿瘤分化程度是胃癌患者发生腹膜转移的独立危险因素。绘制ROC曲线结果提示，内部训练集AUC曲线下面积为0.783（95%CI：0.699-0.867），外部验证集AUC曲线下面积为0.848（95%CI：0.763-0.932）；且以此建立的列线图模型具有良好的区分度、校准度和临床获益度。 结论基于4个独立危险因素的列线图模型对T_4a期胃癌患者腹膜转移的发生具有良好的区分度和校准度，可用于术前对T_4a期胃癌患者腹膜转移风险进行评估，具有一定的临床推广和参考价值。

Establishment and verification of a line graph model for predicting the risk factors of peritoneal metastasis in stage T4a gastric cancer

ObjectiveTo explore the risk factors of peritoneal metastasis in T_4a gastric cancer，and construct a line graph model for the training group，and conduct internal and external verification. MethodsClinical data of 394 patients with stage T_4a gastric cancer who underwent gastrectomy +D2 lymph node dissection from January 2011 to December 2017 were retrospectively analyzed，including 224 patients from January 2011 to December 2015 as the training set，and 170 patients from January 2016 to December 2017 as the verification set. Based on the preoperative imaging findings to determine presence of peritoneal metastasis and finally confirmed by pathological data，using SPSS 24.0 software by t test，rank and inspection or chi-square statistical analysis of risk factors for peritoneal metastases，by single factor and multiple factors Logistic regression filter T_4a period the potential risk factors of peritoneal metastasis in gastric cancer patients，R software（version 4.0.2）was used to build the line graph model. Bootstrap method was used for internal verification，ROC curve was used to evaluate the model's compliance，95%CI was calculated，and calibration curve was drawn to evaluate the model's compliance. DCA curve was drawn to evaluate the clinical benefit of the model. ResultAmong 224 patients in the training set，37（16.5%）patients developed peritoneal metastases，while 23（13.5%）patients in the verification set developed peritoneal metastases. Multivariate analysis showed that carbohydrate antigen 125（CA125），ascites，preoperative albumin（ALB）and tumor differentiation degree were independent risk factors for peritoneal metastases in gastric cancer patients. ROC curve drawing results indicated that the AUC area under the curve of internal training set was 0.783（95%CI：0.699-0.867），and the AUC area under the curve of external validation set was 0.848（95%CI：0.763-0.932）. And the model of column graph established by this method has good differentiation，calibration and clinical benefit. ConclusionThe line graph model based on four independent risk factors has a good degree of differentiation and calibration for the occurrence of peritoneal metastasis in patients with stage T_4a gastric cancer，which can be used for preoperative assessment of the risk of peritoneal metastasis in patients with stage T_4a gastric cancer，and has certain clinical promotion and reference value.