构建原发性肝细胞癌(HCC)微血管侵犯(MVI)的术前预测模型并验证其准确性。

回顾性分析2017年1月至2019年6月行肝切除术的160例HCC患者的临床病理资料，观察患者MVI情况。采用SPSS20.0软件对数据进行处理分析，计数资料采用χ²检验；计量资料采用t检验；采用单因素和多因素Logistic回归分析影响MVI的独立危险因素，并构建HCC患者MVI的术前预测模型，通过描绘受试者工作特征曲线(ROC)并计算曲线下面积(AUC)从而来评估模型的预测能力，并以术后病理诊断结果为金标准对预测模型进行验证。

在160例患者中，有MVI者86例，无MVI者74例。对单因素分析有统计学意义的资料进行Logistic多因素分析，结果显示：肿瘤直径、瘤周低回声晕环、甲胎蛋白(AFP)水平、血小板与淋巴细胞比值(PLR)水平、循环肿瘤DNA(ctDNA)浓度是HCC的MVI独立危险因素。根据Logistic回归分析各变量的回归系数构建预测模型，通过绘制ROC曲线，计算出AUC值为0.914(95%CI 0.820～0.962)，当最佳临界值为0.069时对HCC患者MVI具有预测价值，灵敏度为86.5%，特异度为87.9%，约登指数为0.74。以术后病理诊断为金标准，验证预测模型，灵敏度为88.4%，特异度为93.2%，两者灵敏度和特异度无统计学差异(P>0.05)。

基于Logistic多因素回归分析建立预测模型具有较高的灵敏度和特异性，对HCC微血管侵犯的患者具有较高的预测价值，可为HCC患者的术前治疗方案、手术规划提供参考。

To construct a preoperative predictive model of primary hepatocellular carcinoma (HCC) microvascular invasion (MVI) and to verify its accuracy.

The clinicopathological data of 160 HCC patients underwent hepatectomy from January 2017 to June 2019 were analyzed retrospectively to observe the MVI status of the patients. SPSS20.0 software was used to process and to analyze the data. Count data were examined by using χ² test; measurement data were examined by using independent t test; Univariate and multivariate Logistic regression analysis were used to analyze the independent risk factors affecting MVI, and to construct a preoperative predictive model of primary hepatocellular carcinoma MVI. To evaluate the predictive ability of the model and verify the predictive model with the postoperative pathological diagnosis results as the gold standard.

Among 160 patients, 86 had MVI and 74 had no MVI. Logistic multivariate regression analysis of statistically significant clinical pathology data from single factor analysis showed that tumor diameter, peritumoral hypoechoic halo, alpha-fetoprotein (AFP) level, platelet-to-lymphocyte ratio (PLR) level, and circulating tumor DNA (ctDNA) concentration were independent risk factors for primary hepatocellular carcinoma MVI. According to the logistic regression analysis of the regression coefficients of each variable to construct a prediction model, by drawing the ROC curve, the AUC value was calculated to be 0.914 (95% CI 0.820-0.962). When the optimal cut-off value is 0.069, it has predictive value for primary hepatocellular carcinoma MVI. The sensitivity is 86.5%, the specificity is 87.9%, and the Youden index is 0.74. Using the postoperative pathological diagnosis as the gold standard to verify the prediction model, the sensitivity was 88.4% and the specificity was 93.2%. There was no statistical difference in sensitivity and specificity between them (P>0.05).

The establishment of a prediction model based on Logistic multi-factor regression analysis has a high sensitivity and specificity, and has a high predictive value for patients with primary hepatocellular carcinoma microvascular invasion, It could provide reference for the preoperative treatment plan and surgical planning of hepatocellular carcinoma patients.