Objective: To establish and validate a nomogram model for predicting the risk of microvascular invasion(MVI) in hepatocellular carcinoma. Methods: The clinical data of 210 patients with hepatocellular carcinoma who underwent hepatectomy at Department of Hepatobiliary and Pancreatic Surgery,the Affiliated Hospital of Qingdao University from January 2013 to October 2021 were retrospectively analyzed. There were 169 males and 41 females, aged(M(IQR)) 57(12)years(range:30 to 80 years). The patients were divided into model group(the first 170 cases) and validation group(the last 40 cases) according to visit time. Based on the clinical data of the model group,rank-sum test and multivariate Logistic regression analysis were used to screen out the independent related factors of MVI. R software was used to establish a nomogram model to predict the preoperative MVI risk of hepatocellular carcinoma,and the validation group data were used for external validation. Results: Based on the modeling group data,the receiver operating characteristic curve was used to determine that cut-off value of DeRitis ratio,γ-glutamyltransferase(GGT) concentration,the inverse number of activated peripheral blood T cell ratio (-aPBTLR) and the maximum tumor diameter for predicting MVI, which was 0.95((area under curve, AUC)=0.634, 95%CI: 0.549 to 0.719), 38.2 U/L(AUC=0.604, 95%CI: 0.518 to 0.689),-6.05%(AUC=0.660, 95%CI: 0.578 to 0.742),4 cm(AUC=0.618, 95%CI: 0.533 to 0.703), respectively. Univariate and multivariate Logistic regression analysis showed that DeRitis≥0.95,GGT concentration ≥38.2 U/L,-aPBTLR>-6.05% and the maximum tumor diameter ≥4 cm were independent related factors for MVI in hepatocellular carcinoma patients(all P<0.05). The nomogram prediction model based on the above four factors established by R software has good prediction efficiency. The C-index was 0.758 and 0.751 in the model group and the validation group,respectively. Decision curve analysis and clinical impact curve showed that the nomogram model had good clinical benefits. Conclusions: DeRitis ratio,serum GGT concentration,-aPBTLR and the maximum tumor diameter are valuable factors for preoperative prediction of hepatocellular carcinoma with MVI. A relatively reliable nomogram prediction model could be established on them.
目的： 建立并验证预测肝细胞癌患者微血管侵犯（MVI）发生风险的列线图模型。 方法： 回顾性收集2013年1月至2021年10月在青岛大学附属医院肝胆胰外科行肝切除术的210例肝细胞癌患者的临床资料，男性169例，女性41例，年龄［M（IQR）］57（12）岁（范围：30~80岁）。按就诊时间分为建模组（前170例）和验证组（后40例）。基于建模组临床数据，通过秩和检验及多因素Logistic回归分析筛选MVI的独立相关因素。应用R软件建立术前预测肝细胞癌伴MVI风险的列线图模型，利用验证组数据进行外部验证。 结果： 基于建模组数据，利用受试者工作特征曲线确定DeRitis比值、γ-谷氨酰转移酶（GGT）浓度、外周血活化T细胞比例的相反数（-aPBTLR）和肿瘤最大径预测MVI的截点值分别为0.95（曲线下面积为0.634，95%CI：0.549~0.719）、38.2 U/L（曲线下面积为0.604，95%CI：0.518~0.689）、-6.05%（曲线下面积为0.660，95%CI：0.578~0.742）、4 cm（曲线下面积为0.618，95%CI：0.533~0.703）。单因素和多因素Logistic回归分析结果显示，DeRitis≥0.95、GGT浓度≥38.2 U/L、-aPBTLR>-6.05%及肿瘤最大径≥4 cm是肝细胞癌患者发生MVI的独立相关因素（P值均<0.05）。应用R软件建立的基于以上4个指标的列线图预测模型具有良好的预测效能，建模组和验证组的C指数分别为0.758和0.751。决策曲线分析和临床影响曲线均显示列线图模型具有良好的临床效益。 结论： DeRitis比值、GGT浓度、-aPBTLR和肿瘤最大径是术前预测肝细胞癌合并MVI有价值的指标，基于此建立的的列线图预测模型有一定的临床实效性。.