目的

基于Meta 分析构建肝细胞癌（肝癌）微血管侵犯（MVI）的术前预测模型。

方法

检索国内外公开发表的有关肝癌MVI 预测模型构建研究。筛选检索结果、提取资料，并采用纽卡斯尔-渥太华量表（NOS）进行质量评价。对纳入研究进行Meta 分析，根据结果提取合并效应量明显的危险因素及合并危险值，构建预测模型，绘制列线图。选取2020 年1 月至2023 年1 月在赤峰市医院接受手术治疗的64 例肝癌术后患者为模型验证集，采用ROC 曲线下面积（AUC）评价模型的预测性能，采用校准曲线和临床决策曲线分析模型的准确性和临床实用性。

结果

共纳入20 项队列研究，共4 021 例肝癌患者。Meta 分析20 项研究，合并效应量最终筛选出7 个危险因素，分别为

AFP（OR=1.38，95%CI：1.22～1.54）、GGT（OR=1.49， 95%CI：1.14～1.83）、ALB（OR=0.90，95%CI：0.78～1.02）、瘤内动脉（OR=3.48， 95%CI：3.12～3.85）、瘤周强化（OR=2.49，95%CI：1.94～3.04）、肿瘤直径（OR=2.93，95%CI：2.59～3.27）和肝外生性生长（OR=2.08，95%CI：1.75～2.42）。将7 个危险因素分为影像特征组、实验室指标组和联合组，并赋值，构建3 个预测肝癌术后早期复发模型。联合组模型AUC 为0.877，敏感度和特异度分别为0.813 和0.875。该模型预测情况和实际情况大约一致，且有良好的临床收益。

结论

Meta 分析筛选的肝癌MVI 危险因素包括AFP、GGT、ALB、瘤内动脉、瘤周强化、肿瘤直径和肝外生长。基于Meta 分析构建的MVI 危险预测模型具有较好的预测性能，可作为MVI 危险评估工具。

Objective

To construct a preoperative prediction model for microvascular invasion(MVI) of hepatocellular carcinoma (HCC) based on Meta-analysis.

Methods

Studies related to prediction model for MVI of HCC published at home and abroad was retrieved.The retrieval results were screened and the data were extracted.The quality of included studies was assessed by Newcastle-Ottawa Scale (NOS).Meta-analysis was carried out for the included studies.The risk factors with significant combined effects were extracted and the risk values were combined according to the results.The prediction model was constructed and a nomogram was delineated.64 HCC patients undergoing surgery in Chifeng Municipal Hospital from January 2020 to January 2023 were selected into the model validation set.The prediction performance of this model was evaluated by the area under the ROC curve (AUC).The accuracy and clinical practicability of this model were analyzed by the calibration curve and decision curve analysis.

Results

A total of 20 cohort studies consisting of 4 021 HCC patients were included.In the Meta-analysis of these 20 studies, 7 risk factors were finally screened after combined effects, including AFP (OR=1.38, 95%CI: 1.22-1.54), GGT (OR=1.49, 95%CI: 1.14-1.83)and ALB (OR=0.90, 95%CI: 0.78-1.02), intratumoral artery (OR=3.48, 95%CI: 3.12-3.85), peritumoral enhancement (OR=2.49, 95%CI: 1.94-3.04), tumor diameter (OR=2.93, 95%CI: 2.59-3.27) and extrahepatic growth (OR= 2.08, 95%CI: 1.75-2.42), respectively.7 risk factors were divided into the imaging feature group,laboratory index group and combined group, and assigned with values, and 3 models were constructed to predict early postoperative recurrence of HCC.The AUC of the model in the combined group was 0.877, and the sensitivity and specificity were 0.813 and 0.875, respectively.The prediction performance of this model was almost consistent with the actual values, indicating high clinical benefits.

Conclusions

Meta-analysis demonstrates that the risk factors of MVI in HCC include AFP, GGT, ALB, intratumoral artery, peritumoral enhancement, tumor diameter and extrahepatic growth.The prediction model for MVI risk based on Metaanalysis yields favorable prediction performance, which can be used as an assessment tool for the risk of MVI.