摘要
目的 基于超声造影和临床资料构建联合模型,探讨其预测小肝癌患者发生微血管侵犯(MVI)的临床价值.方法 选取于我院行手术治疗的小肝癌患者423例,按照7∶3比例随机分为训练集296例和验证集127例;以术后病理结果为标准,将其分为MVI阳性组171例和MVI阴性组252例.基于训练集数据,比较MVI阳性组与MVI阴性组临床资料和超声造影参数的差异;采用多因素Logistic回归分析筛选预测MVI的独立影响因素.基于单独临床资料、超声造影参数及二者联合分别构建临床模型、超声模型及联合模型.绘制受试者工作特征曲线(ROC)分析各模型预测小肝癌患者发生MVI的诊断效能;绘制校准曲线和临床决策曲线评估各模型的校准度和临床收益.结果 训练集中MVI阳性组与MVI阴性组LI-RADS分类、廓清时间、廓清类型、增强后肿瘤面积增大百分比、性别比、甲胎蛋白(AFP)比较差异均有统计学意义(均P<0.05).多因素Logistic回归分析显示,LI-RADS分类、增强后肿瘤面积增大百分比、AFP均为预测MVI的独立影响因素(OR=1.68、1.29、1.59,均P<0.05).ROC曲线分析显示,临床模型、超声模型、联合模型预测训练集患者发生MVI的曲线下面积(AUC)分别为0.624、0.669、0.851,预测验证集患者发生MVI的AUC分别为0.663、0.611、0.838,以联合模型的AUC最高,与其他模型比较差异均有统计学意义(均P<0.05).校准曲线和临床决策曲线分析显示,联合模型校准度最好,可获得更大的临床净收益.结论 基于超声造影和临床资料构建的联合模型在预测小肝癌患者发生MVI方面具有较好的临床价值,可为临床制定个性化方案提供参考.
Abstract
Objective To construct a combined model based on contrast-enhanced ultrasound(CEUS)and clinical data,and to explore the clinical value of the model in predicting the occurrence of microvascular invasion(MVI)in patients with small hepatocellular carcinoma(HCC).Methods A total of 423 patients with small HCC who underwent surgical treatment in our hospital were selected and randomly divided into a training set(296 cases)and a validation set(127 cases)in a 7∶3 ratio.Postoperative pathological results were used as the standard,all patients were divided into the MVI-positive group(171 cases)and MVI-negative group(252 cases).Based on the data of training set,the clinical data and CEUS parameters were compared between the MVI-positive group and MVI-negative group,and multivariate Logistic regression analysis was used screen the influencing factors for predicting MVI.Clinical,ultrasound and combined models were constructed based on clinical data alone,CEUS parameters and the combination of the two,respetively.Receiver operating characteristic(ROC)curve was drawn to analyze the diagnostic efficacy of each model for predicting the occurvence of MVI in patients with small HCC.Calibration curves and clinical decision curves were drawn to assess the calibration and clinical utility of each model.Results In the training set,there were significant differences in LI-RADS classification,clearance time,clearance type,post-contrast tumor area increase ratio,gender and alpha-fetoprotein(AFP)between the MVI-positive and MVI-negative groups(all P<0.05).Multivariate Logistic regression analysis showed that LI-RADS classification,post-contrast tumor area increase ratio and AFP were independent factors influencing the occurrence of MVI(OR=1.68,1.29,1.59,all P<0.05).ROC curve analysis showed that the area under the curve(AUC)for predicting the occurrence of MVI in the training set by the clinical,ultrasound,and combined models were 0.624,0.669 and 0.851,respectively,and the AUC of those in validation set were 0.663,0.611 and 0.838,respectively,with the combined model had the highest AUC,which was significantly different from the other models(all P<0.05).Calibration curve and clinical decision curve analysis indicated that the combined model had the best calibration and could achieve greater clinical net benefit.Conclusion The combined model based on CEUS and clinical data has good clinical value in predicting the occurrence of MVI in patients with small HCC,and may serve as a reference for developing personalized clinical treatment plans.
维普
万方数据