基于炎症标志物构建肝细胞癌术后预后预测模型

Establishment of predictive model based on inflammatory markers for HCC patients after hepatectomy

目的构建基于炎症反应标志物预测肝细胞癌患者术后3和5年生存模型。方法回顾性分析2010-01-13-2015-10-23南通市肿瘤医院(279例)与南通大学附属医院(359例)首次行肝细胞癌切除术治疗的638例患者临床资料。收集患者术前实验室检查指标。采用Epidata经双人录入核查构建数据库,运用电子随访信息系统或电话方式进行随访。caret包createDataPartition函数随机分为训练集448例与验证集190例。受试者工作特征(ROC)曲线分析确定中性粒细胞与淋巴细胞比率(NLR)、γ-谷氨酰转移酶与丙氨酸氨基转移酶比率(GGT/ALT)的最佳截断值。LASSO回归和多因素Cox比例风险回归分析独立危险因素,构建肝细胞癌患者术后生存3和5年列线图模型。在验证集进行模型效能验证。结果ROC曲线分析确定术前NLR和GGT/ALT最佳截断值为2.72和1.44。患者术后1、3和5年累计生存率分别为88.0%、68.3%和58.8%。乙肝病毒e抗原(HBeAg)阳性(HR=1.988,P=0.011)、NLR>2.72(HR=1.814,P=0.005)、GGT/ALT>1.44(HR=1.508,P=0.022)、天门冬氨酸氨基转移酶与血小板比率指数(APRI)>2.00(HR=2.347,P<0.001)、甲胎蛋白(AFP)>400μg/L(HR=1.680,P=0.014)、乳酸脱氢酶(LDH)>195 U/L(HR=1.517,P=0.036)、D-二聚体>0.33 mg/L(HR=1.495,P=0.026)、淋巴细胞计数(HR=1.668,P=0.002)和住院天数(HR=1.032,P=0.027)是肝细胞癌患者3年总生存期(OS)的独立危险因素。HBeAg阳性(HR=1.971,P=0.006)、NLR>2.72(HR=1.496,P=0.010)、GGT/ALT>1.44(HR=1.482,P=0.013)、APRI>2.00(HR=2.236,P<0.010)、肿瘤直径≥5 cm(HR=1.543,P=0.030)和住院天数(HR=1.029,P=0.031)是患者5年OS的独立危险因素。将Cox回归分析P<0.100的变量构建列线图模型,验证集3和5年OS的ROC曲线下面积(AUC)为0.748(95%CI:0.679~0.817)和0.774(95%CI:0.708~0.840)。结论构建基于炎症标志物的预后列线图模型,确定NLR、GGT/ALT和APRI是肝细胞癌患者术后生存的危险因素,可应用于指导临床治疗方案,改善肝细胞癌术后患者的预后。

Objective To construct a nomogram based on inflammation markers to predict the 3-year and 5-year survival of patients with hepatocellular carcinoma receiving hepatectomy.Methods The clinical data of 638 HCC patients were retrospectively collected in Nantong Tumor Hospital(279 cases)and Affiliated Hospital of Nantong University(359 cases)from January 13 th 2010 to October 23 rd 2015.Preoperative clinical indicators of patients were collected.Epidata software was used to construct a database thru two-person entry,and an electronic tracking system or telephone was used for follow-up in this study.Patients were randomly divided into two groups by"createDataPartition"function:the training group(448cases)and the validation group(190cases).The receiver operating characteristic curve(ROC)was used to determine the best cutoff values for NLR and GGT/ALT.The LASSO regression and the Cox proportional risk model were used to analyze independent risk factors of overall survival(OS)in patients with HCC.Independent risk variables were used to create the 3-year and 5-year nomogram models.In the validation group,the nomogram’s effectiveness was verified.Results The best cutoff values for NLR and GGT/ALT,according to ROC curve analysis,were 2.72and 1.44,respectively.Patients’cumulative survival rates at 1,3and 5years after surgery were 88.0%,68.3%,and 58.8%,respectively.HBV e antigen positive(HR=1.988,P=0.011),NLR>2.72(HR=1.814,P=0.005),GGT/ALT>1.44(HR=1.508,P=0.022),APRI>2.00(HR=2.347,P<0.001),AFP>400μg/L(HR=1.680,P=0.014),LDH>195U/L(HR=1.517,P=0.036),D-dimer>0.33mg/L(HR=1.495,P=0.026),lymphocytes(HR=1.668,P=0.002)and hospitalization days(HR=1.032,P=0.027)were risk factors for the 3-year OS of HCC patients.HBV e antigen positive(HR=1.971,P=0.006),NLR>2.72(HR=1.496,P=0.010),GGT/ALT>1.44(HR=1.482,P=0.013),APRI>2.00(HR=2.236,P<0.010),tumor diameter≥5cm(HR=1.543,P=0.030),and hospitalization days(HR=1.029,P=0.031)were independent risk factors that affected the 5-year survival of HCC patients.Risk factors for OS by Cox regression were all contained into the nomogram(P<0.100).In validation cohort,the Area under Curve(AUC)of 3-year and 5-year OS was 0.748(95%CI:0.679-0.817)and 0.774(95%CI:0.708-0.840),respectively.Conclusion LASSO Cox regression is used to construct a nomogram model including inflammatory markers such as NLR,GGT/ALT,and APRI,which are found to be risk factors of OS for HCC patients,and are helpful to improve the prognosis of patients with HCC after hepatectomy.