肝细胞癌根治性切除后极早期复发的危险因素与预测模型

A novel model in predicting very early recurrence after surgical resection of hepatocellular carcinoma

目的探讨肝细胞癌根治性切除后极早期(术后3个月)复发的相关危险因素,并建立相应的临床预测模型。方法纳入2008年1月至12月在复旦大学附属中山医院行根治性切除,且在2年内(早期)复发的427例肝细胞癌患者为研究队列,其中男性368例,女性59例,年龄(52.7±12.1)岁;同时纳入2009年1月至6月接受根治性切除的所有肝细胞癌患者590例为验证队列,其中男性525例,女性65例,年龄(54.7±11.0)岁。研究极早期复发的危险因素,并建立预测模型。结果在早期复发的研究队列中,126例(29.5%)患者表现为极早期复发,预后极差。血清甲胎蛋白(AFP)>827μg/L、多发肿瘤、微血管侵犯(MVI)以及肿瘤数目是肝细胞癌根治性切除后极早期复发的独立危险因素。通过logistic回归方程定义新变量VR(0.809AFP+1.262肿瘤数目+0.983MVI)可预测极早期复发,受试者工作特性曲线显示VR预测极早期复发的曲线下面积为0.722(95%CI:0.669~0.774,P<0.001)。在验证队列中,此模型的曲线下面积为0.785(95%CI:0.715~0.855,P<0.001)。结论术前高AFP水平、多发肿瘤以及MVI都是肝细胞癌根治术后极早期复发的高危因素,通过对它们的综合评估建立的预测模型显示了较好的稳定性与潜在临床应用可能。

Objective To study the risk factors of very early recurrence(VER,within 3 months)after R0 resection of hepatocellular carcinoma(HCC),and to establish a predictive model.Methods Of 427 HCC patients[with 368 males,59 females,aged(52.7±12.1)years]who developed early recurrence(within 2 years)after R0 resection from January to December 2008 at Zhongshan Hospital,Fudan University were enrolled in the test cohort.Another 590 patients[with 525 males,65 females,aged(54.7±11.0)years]who underwent R0 resection from January to June 2009 were enrolled in the validation cohort.Risk factors were investigated and a predictive model was established.Results In the test cohort,126 patients(29.5%)developed VER and their survival outcomes were extremely poor.Serumα-fetoprotein(AFP)level>827μg/L,multiple tumors,microvascular invasion(MVI)and tumor number were independent risk factors for VER.A new predictive model(0.809·AFP+1.262·tumor number+0.983·MVI)was established by logistic regression in predicting VER after surgery.The receiver operating characteristic curve showed that the area under the curve(AUC)in predicting VER was 0.722(95%CI:0.669-0.774,P<0.001).In the validation cohort,the AUC of this model was 0.785(95%CI:0.715-0.855,P<0.001).Conclusions A high AFP level,multiple tumors,and MVI were independent risk factors for VER of HCC after R0 resection.The prediction model consisting of these three factors demonstrated robustness and it has the potential in clinical application.