基于肝细胞癌磁共振成像强化模式的术前微血管侵犯预测研究

Prediction of microvascular invasion based on enhanced mode magnetic resonance imaging for patients with hepatocellular carcinoma

目的探讨肝细胞癌(HCC)患者术前肿瘤MRI影像学特征与强化模式在预测微血管侵犯(MVI)中的价值。方法收集2017年1月至2019年6月在宁夏医科大学总医院行MRI检查并在肝胆外科行手术切除的孤立性HCC患者临床资料,按照术后病理诊断分为MVI(+)与MVI(-)两组,分析MVI与HCC的扩散加权成像(DWI)信号、强化模式、强化类型等影像学特征之间的关系。结果最终纳入HCC患者84例,其中男性65例,女性19例,年龄(54.94±11.51)岁。MVI(+)组46例,MVI(-)组38例,两组患者的肿瘤最大径分别为(7.08±3.45)cm和(4.28±2.47)cm,差异有统计学意义(P<0.01)。单因素分析表明,HCC MRI影像学特征指标DWI信号、包膜是否完整、强化模式、肿瘤边缘是否光整、是否存在瘤周异常强化和瘤内动脉,两组患者间差异均有统计学意义(均P<0.05);而T1加权成像信号、T2加权成像信号、病灶是否含脂,两组间差异无统计学意义(均P>0.05)。logistic回归分析表明肿瘤最大径>6.33 cm、3/4型强化模式、肿瘤边缘不光整为MVI(+)的独立危险因素(均P<0.05)。ROC曲线分析表明联合诊断的截断值为0.53、曲线下面积为0.881、敏感性0.870、特异性0.789、约登指数0.659。结论利用多因素logistic回归模型联合ROC曲线分析可以根据HCC MRI影像学特征进行术前HCC MVI预测。

Objective To study preoperative MRI imaging and its enhanced mode on tumor features in predicting microvascular invasion(MVI)in patients with hepatocellular carcinoma(HCC).Methods The clinical data of patients with a solitary HCC who underwent MRI examination followed by surgical resection at the General Hospital of Ningxia Medical University from January 2017 to June 2019 were studied.The patients were divided into the MVI(+)and MVI(-)groups according to the findings on postoperative pathological diagnosis.The relationship between the rates of MVI and MRI tumor features including diffusion weighted imaging(DWI)signal,enhancement mode,enhancement type and other imaging characteristics were analysed.Results Of 84 patients with HCC enrolled into this study,there were 65 males and 19 females.Their age(Mean±SD)was(54.94±11.51)years.MVI(+)was found in 46 patients and MVI(-)in 38 patients.The maximum tumor diameters(Mean±SD)of the two groups were(7.08±3.45)cm and(4.28±2.47)cm(P<0.01).Single-factor analysis and comparison of imaging characteristics of the two groups of patients showed tumor DWI signal,tumor encapsulation,enhancement mode,tumor edge smoothness,abnormal enhancement around tumors,and intratumoral arteries were significantly different(P<0.05);There were no significant differences in T1WI signals,T2WI signals,tumor periphery,and enhancement types between groups.After inputting MVI(+)as a risk factor into the logistic regression model,tumor maximum diameters>6.33 cm,type 3/4 enhancement mode,and unsmoothness of tumor edge were independent risk factors(all P<0.05).Through combined diagnosis using ROC curve analysis with a cut-off value of 0.53,the area under the curve was 0.881,the sensitivity 0.870,specificity 0.789,and the Youden index 0.659.Conclusion The multivariate logistic regression model and combined diagnosis using ROC curve analysis improved the diagnostic efficacy of MVI in its prediction of HCC on imaging studies.The risk predictors were easy to use and to promote in clinical practice.