Microvascular invasion (MVI) is an aggressive behavior of hepatocellular carcinoma (HCC) that being an independent predictor of tumor recurrence after hepatectomy or liver transplantation. Preoperative prediction of MVI has important clinical significance.

To use non-invasive imaging and serological indicators to construct a nomogram for predicting MVI in order to provide a clinical reference.

The relevant clinical data of 284 patients who underwent radical resection of HCC in the Second Affiliated Hospital of Kunming Medical University from 2016 to 2021 were retrospectively analyzed, and HCC patients admitted from 2016 to 2020 were categorized as the model group (n=208) according to the admission time, and HCC patients admitted in 2021 were categorized as the validation group (n=76) . LASSO regression and multivariate Logistic regression analysis were performed to determine the independent risk factors for MVI in HCC, and R software was used to establish a nomogram model for predicting the risk of MVI in HCC before operation. The internal validation of the model was performed by Bootstrap method, external validation of the model was performed by using the validation group. The consistency index, calibration curve and decision curve analysis (DCA) were used to evaluate the discrimination, calibration ability and clinical application value of the nomogram.

Multivariate Logistic regression analysis showed that WBC>7.1×10⁹/L〔OR=3.144, 95%CI (1.301, 7.598) , P=0.011〕, tumor diameter>7.05 cm〔OR=3.836, 95%CI (1.758, 8.372) , P=0.001〕, S-Index>0.097〔OR=3.165, 95%CI (1.024, 9.779) , P=0.040〕, AAR>0.879〔OR=2.146, 95%CI (1.062, 4.337) , P=0.030〕, ANRI>24.074〔OR=2.769, 95%CI (1.175, 6.526) , P=0.020〕 were independent predictors of MVI in HCC. The consistency index of the nomogram model established using the five variables combined with AFP was 0.800〔95%CI (0.739, 0.861) 〕 and 0.755〔95%CI (0.641, 0.868) 〕 in the model group and validation group, respectively. The model fitted well with the calibration prediction curve. The optimal critical value of the nomogram calculated by Youden index was 174 points. The sensitivity, specificity, positive predictive value and negative predictive value at the critical value were 90%, 61%, 71%, and 85% in the model group, and 78%, 71%, 76% and 74% in the validation set, respectively.

The nomogram constructed with AFP>45 ng/mL, WBC>7.1×10⁹/L, tumor diameter>7.05 cm, S-Index>0.097, AAR>0.879, ANRI>24.074 can better predict the risk of preoperative MVI in HCC, the use of this nomogram can conveniently guide the clinical treatment of HCC patients by routine examination test indicators.

微血管侵犯（MVI）是肝细胞癌（HCC）的一种侵袭性行为，是肝切除或肝移植术后肿瘤复发的独立预测因子，术前预测MVI有重要临床意义。

使用无创的影像学、血清学指标构建预测MVI的列线图，以期为临床提供参考。

回顾性分析2016—2021年在昆明医科大学第二附属医院行HCC根治术的284例患者的相关临床资料，按入院时间将2016—2020年收治的HCC患者归为模型组（208例），2021年收治的HCC患者归为验证组（76例）。进行LASSO回归和多因素Logistic回归分析确定HCC发生MVI的独立危险因素，应用R软件建立术前预测HCC发生MVI风险的列线图模型，用Bootstrap法进行模型的内部验证，用验证组进行模型的外部验证，用一致性指数、校准曲线和决策曲线分析（DCA）评价列线图的区分度、校准能力和临床应用价值。

多因素Logistic回归分析结果显示，白细胞计数（WBC）>7.1×10⁹/L〔OR=3.144，95%CI（1.301，7.598），P=0.011〕、肿瘤直径>7.05 cm〔OR=3.836，95%CI（1.758，8.372），P=0.001〕、S-Index>0.097〔OR=3.165，95%CI（1.024，9.779），P=0.040〕、AAR>0.879〔OR=2.146，95%CI（1.062，4.337），P=0.030〕、ANRI>24.074〔OR=2.769，95%CI（1.175，6.526），P=0.020〕是HCC发生MVI的独立预测因素。使用该5个变量结合甲胎蛋白（AFP）建立的列线图模型一致性指数在模型组和验证组分别为0.800〔95%CI（0.739，0.861）〕、0.755〔95%CI（0.641，0.868）〕，模型与校准预测曲线贴合良好，通过Youden指数计算出列线图的最佳截断值为174分，截断值下的灵敏度、特异度、阳性预测值和阴性预测值在模型组中分别为90%、61%、71%和85%，在验证组中分别为78%、71%、76%和74%。

以AFP>45 ng/mL、WBC>7.1×10⁹/L、肿瘤直径>7.05 cm、S-Index>0.097、AAR>0.879、ANRI>24.074构建的列线图可较好地预测术前HCC发生MVI的风险，使用该列线图可通过常规检查检验指标方便地指导HCC患者的临床治疗。