近端胃癌上切缘阳性术前列线图预测模型的建立和验证

Development and validation of a preoperative nomogram predictive model for proximal gastric cancer with microscopic positive margin

目的分析影响近端胃癌上切缘阳性(R1切除)的术前因素,建立预测模型并进行内部验证。方法回顾性分析187例接受胃肠外科手术的近端胃癌患者资料,根据上切缘状态(R0为完全切除,R1为镜下残留肿瘤)分为R0组(n=15)和R1组(n=172)。收集可能影响近端胃癌上切缘阳性的术前因素,包括患者年龄、性别、肿瘤长度、肿瘤位置、Borrmann分型、肿瘤分化、Lauren分型、cT分期及cN分期。应用受试者工作特征(ROC)曲线计算肿瘤长度预测近端胃癌上切缘阳性的最佳界值,对2组间单因素分析中存在统计学差异的变量进行多因素Logistic回归,筛出独立预测因素并构建预测模型;运用ROC曲线和Bootstrap法对模型预测准确性进行内部验证。结果肿瘤长度预测近端胃癌上切缘阳性的最佳界值为4.85 cm。单因素分析显示2组间肿瘤长度、肿瘤位置、Borrmann分型、Lauren分型、cT分期及cN分期差异均有统计学意义(均P<0.05)。多因素Logistic回归分析显示肿瘤长度>4.85 cm(OR=4.000,95%CI:1.039~15.399)、肿瘤位于食管胃结合部(OR=7.108,95%CI:1.604~31.494)、BorrmannⅢ-Ⅳ型(OR=6.991,95%CI:1.538~31.782)、Lauren分型为弥漫型或混合型(OR=7.583,95%CI:1.814~31.701)及cT分期为cT4(OR=8.249,95%CI:1.890~36.007)是近端胃癌上切缘阳性的术前独立预测因素,基于多因素分析结果建立列线图预测模型。经内部验证,列线图模型ROC曲线下面积为0.862。校准曲线显示模型预测近端胃癌上切缘阳性发生的概率与实际发生的概率具有较好一致性(Hosmer-Lemeshowχ^(2)=6.145,P=0.523)。结论建立的列线图预测模型可在术前预测近端胃癌上切缘阳性的概率,为制定手术策略提供临床指导。

Objective To explore the preoperative predictive factors influencing microscopic positive proximal margin in upper gastric cancer,and to establish a nomogram prediction model and to validate it internally.Methods Retrospective analysis of 187 patients with upper gastric cancer operated in the Department of Gastrointestinal Surgery of Hengshui People′s Hospital from January 2018 to October 2022 were included in this study.Patients were divided into the microscopic positive proximal margin(the R0 group,n=15)and the negative microscopic proximal margin group(the R1 group,n=172)according to histopathological diagnosis.Preoperative factors that may influence positive upper margin of proximal gastric cancer were collected,including patient age,gender,tumor size,tumor location,Borrmann staging,tumor differentiation,Lauren staging,cT stage and cN stage.Receiver operating characteristic(ROC)curve was used to figure out the optimal cut-off value for predicting positive margin of proximal gastric cancer by tumor length.Multivariate Logistic regression was used to analyze the variables with statistical difference between the two groups,and independent risk factors were screened out,and prediction mode was constructed.The prediction accuracy of the model was verified internally using Bootstrap method.Results The best threshold for predicting positive margin of proximal gastric cancer by tumor length was 4.85 cm.Univariate analysis showed that there were significant differences in tumor length,tumor location,Borrmann staging,Lauren staging,cT staging and cN staging between the two groups(all P<0.05).Multivariate Logistic regression analysis showed that tumor length>4.85 cm(OR=4.000,95%CI:1.039-15.399),tumor located in esophagogastric junction(OR=7.108,95%CI:1.604-31.494),Borrmann stagingⅢ—Ⅳ(OR=6.991,95%CI:1.538-31.782),Lauren staging as diffuse or mixed(OR=7.583,95%CI:1.814-31.701)and cT staging as cT4(OR=8.249,95%CI:1.890-36.007)were independent predictors of microscopic positive proximal margin of advanced upper gastric cancer before surgery,and a prediction model was established based on results of multivariate analysis.The area under ROC curve(AUC)value for subjects with the model was 0.862 after internal validation.The calibration curve showed that the model predicted the probability of microscopic positive proximal margin occurrence in good agreement with the probability of actual microscopic positive proximal margin occurrence(Hosmer-Lemeshowχ^(2)=6.145,P=0.523).Conclusion The established nomogram prediction model can predict the probability of positive upper incisal margin of proximal gastric cancer before operation,and provide clinical guidance for formulating surgical strategy.