目的 探讨2018版肝脏影像报告和数据系统(Liver Imaging Reporting and Data System version 2018,LI-RADS v2018)对肝细胞癌(hepatocellular carcinoma,HCC)细胞角蛋白 19(cytokeratin 19,CK19)表达的术前预测及预后评估的价值.材料与方法 回顾性分析220例术前接受MRI检查并经病理证实为HCC患者的临床、病理及影像资料,包括CK19阳性组59例,CK19阴性组161例.将患者按7∶3比例分为训练集和验证集.通过单因素与多因素logistic回归分析确定CK19阳性表达HCC的独立预测因素并构建列线图评分模型.采用受试者工作特征(receiver operating characteristic,ROC)曲线分析模型诊断效能,绘制校准曲线、决策曲线评价模型的校准性能和临床适用性.计算患者的列线图得分并进行高低风险分组,采用Kaplan-Meier生存曲线分析比较不同亚组患者的总体、早期及晚期无复发生存率.结果 晕状强化(OR=3.432,P=0.045)、环形动脉期高强化(OR=32.073,P=0.017)、靶样扩散受限(OR=12.941,P=0.006)、不光滑肿瘤边缘(OR=4.590,P=0.014)及肝胆期肿瘤-肝实质相对增强比(the relative enhancement ratio,RER)(OR=0.014,P=0.023)是CK19阳性表达HCC的独立预测因素.预测模型在训练集和验证集的曲线下面积(area under the curve,AUC)分别为0.884(95%CI:0.823～0.930)、0.748(95%CI:0.625～0.846),校准曲线、决策曲线显示模型的校准性能和临床适用性较好.CK19阳性与阴性组的总体无复发生存率、高与低风险组的总体、早期及晚期无复发生存率之间均存在显著差异(P<0.05).结论 晕状强化、环形动脉期高强化、靶样扩散受限结合不光滑肿瘤边缘、肝胆期增强定量参数可对HCC的CK19表达进行术前风险预测,并有助于评估HCC术后复发.
Risk prediction and prognostic evaluation of hepatocellular carcinoma CK19 expression based on LI-RADS v2018 and other MRI features
Objective:To investigate the value of the Liver Imaging Reporting and Data System version 2018(LI-RADS v2018)in preoperative predicting the expression of cytokeratin 19(CK19)and assessing postoperative prognosis in hepatocellular carcinoma(HCC).Matirials and Methods:Two hundred and twenty patients with pathologically-confirmed HCC were retrospectively included who underwent preoperative MRI examination,including 59 cases with CK19-positive expression and 161 cases with CK19-negative expression.All patients were divided into training and validation sets in a 7∶3 ratio,and the clinical,pathological,and imaging data of the patients were analyzed.Independent predictors for CK19 expression in HCC were determined by univariate-and multivariate logistic regression analysis and a nomogram scoring model was constructed.The diagnostic performance of the model was assessed using receiver operating characteristic(ROC)curve analysis.Calibration curves and decision curves were plotted to evaluate the calibration performance and clinical applicability of the model.Nomogram scores for patients were calculated and used for high and low risk stratification.The Kaplan-Meier survival curve was employed to compare the overall,early,and late recurrence-free survival rates among different subgroups.Results:Corona enhancement(OR=3.432,P=0.045),rim arterial phase hyperenhancement(OR=32.073,P=0.017),targetoid diffusion restriction(OR=12.941,P=0.006),irregular tumor margin(OR=4.590,P=0.014),and relative enhancement ratio(RER)in the hepatobiliary phase(OR=0.014,P=0.023)were independent predictors for CK19 expression in HCC.The AUC of the prediction model in the training set and validation set were 0.884(95%CI:0.823-0.930)and 0.748(95%CI:0.625-0.846),respectively.The calibration curve and decision curve showed that the calibration performance and clinical applicability of the model were good.Significant differences were observed in overall recurrence-free survival rates between the CK19-positive and CK19-negative groups,as well as in overall,early and late recurrence free survival rates between high and low-risk groups(P<0.05).Conclusions:Corona enhancement,rim arterial phase hyperenhancement,and targetoid diffusion restriction,combined with irregular tumor margin and enhancement quantitative parameter in hepatobiliary phase can achieve preoperative risk prediction of CK19 expression in HCC,which also aids in assessing HCC postoperative recurrence.
primary hepatocellular carcinomacytokeratin 19prediction modelLiver Imaging Reporting and Data Systemmagnetic resonance imaging
国家科技期刊平台
NSTL
万方数据
