基于多模态超声参数特征构建肝细胞肝癌微血管侵犯的风险列线图模型

Constructing a Risk Line Graph Model for Microvascular Invasion in Hepatocellular Carcinoma Based on Multimodal Ultrasound Parameter Features

目的:基于多模态超声参数特征构建肝细胞肝癌(HCC)微血管侵犯(MVI)的风险列线图模型。方法:回顾性选取2021年5月至2022年8月在医院诊断后接受治疗的172例HCC患者作为研究对象,根据病理学诊断分为MVI未发组和MVI组。使用多因素Logistic回归分析MVI发生的影响因素并构建相应的风险预测模型。结果:其中MVI未发组70例(41.70%),MVI组102例(59.30%);两组患者病灶尺寸、AP、PP、PVP是影响HCC患者MVI是否发生的独立影响因素(P<0.05)。模型公式:Logit(P)=12.151+0.067×病灶尺寸-2.253×AP-2.553×PP-3.565×PVP。其ROC曲线下面积为0.971,95%CI:0.951~0.991,敏感度为0.892,特异度为0.943。结论:HCC患者MVI风险列线图预测模型具有较好的预测效能,其中病灶尺寸、AP、PP、PVP是患者MVI是否发生的主要影响因素。

Objective:Based on the characteristics of multimodal ultrasound parameters,a risk nomogram model of microvascular invasion(MVI)in hepatocellular carcinoma(HCC)was constructed.Methods:A total of 172 patients with HCC who were treated after diagnosis in the hospital from May 2021 to August 2022 were retrospectively selected as the research objects.According to the pathological diagnosis,they were divided into MVI non-occurrence group and MVI group.Multivariate Logistic regression was used to analyze the influencing factors of MVI and construct the corresponding risk prediction model.Results:Among them,70 cases(41.70%)in the MVI non-onset group and 102 cases(59.30%)in the MVI group;the lesion size,AP,PP and PVP of the two groups were independent factors affecting the occurrence of MVI in HCC patients(P<0.05).Model formula:Logit(P)=12.151+0.067×lesion size-2.253×AP-2.553×PP-3.565×PVP.The area under the ROC curve was 0.971,95%CI:0.951-0.991,the sensitivity was 0.892,and the specificity was 0.943.Conclusion:The predictive model of MVI risk column chart for HCC patients has good predictive performance,with lesion size,AP,PP,and PVP being the main influencing factors for the occurrence of MVI in patients.