经肝动脉化疗栓塞治疗中期肝癌预后的预测模型的建立和验证

The establishment and validation of nomogram model used for predicting the prognosis of patients with mid-stage HCC after receiving TACE

目的建立并验证对中期肝癌(BCLC B期)患者行TACE治疗的生存期列线图预测模型。方法收集2009年1月至2017年12月在3所单位接受TACE治疗的首诊首治中期肝癌患者。其中常州市第一人民医院和常州市第二人民医院作为训练集,苏州大学附属第一医院作为测试集。收集患者的基线资料。计算患者的生存期,采用Kaplan-Meier法绘制生存曲线,采用log-rank检验进行生存分析。采用Cox回归模型分析影响生存的预后因素并建立列线图。采用AUROC及C指数比较每个模型预测的准确性及效能。结果训练集158例,测试集141例,共计299例中期肝癌患者。训练集和测试集的生存时间分别为22.7(19.9~30.1)和23.8(19.9~29.0)个月(P=0.96)。单因素分析中,性别、AFP、肿瘤大小、白蛋白、AST、Child-Pugh分期、up-to-7标准、up-to-11标准及肿瘤应答均与预后相关(P<0.05)。在多因素分析中,AFP、up-to-11标准及肿瘤应答为独立预后因素(P<0.05)。基于独立预后因素建立的列线图预测模型在训练集和测试集的C指数分别为0.700(95%CI:0.693~0.707)和0.647(95%CI:0.637~0.657)。列线图模型预测1年、2年生存率的AUROC值和C指数都高于HAP、BCLC B sub-classification及new BCLC B subclassification等模型。结论列线图模型能够很好地预测BCLC B期肝癌行TACE治疗的生存期,模型的准确性及效能都高于其他模型,有助于筛选获益患者及预后分层。

Objective To establish and validate a nomogram model which is used for predicting the survival time in patients with mid-stage hepatocellular carcinoma(HCC)who are treated with transcatheter arterial chemoembolization(TACE).Methods The clinical materials of 299 patients with mid-stage HCC,who received initial TACE at three hospitals in Changzhou City of China during the period from January 2009 to December 2017,were collected.The clinical data obtained from Changzhou Municipal First People’s Hospital and Changzhou Municipal Second People’s Hospital were used as the training set for model development,and the clinical data obtained from the First Affiliated Hospital of Soochow University were used as the testing set(validation set).The baseline data of patients,including epidemiological information(age,gender),laboratory results(liver and kidney functions,alpha-fetoprotein levels),and imaging findings(tumor size,number),were collected.The patient’s survival time was calculated.By using Kaplan-Meier method the survival curve was drawn,and log-rank test was used to analyze the survival data.Cox regression model was adopted to analyze the prognostic factors affecting the survival time,and the nomogram was established.The area under the receiver operator characteristic curve(AUROC)and C-index were used to compare the prediction accuracy and effectiveness between different models.Results Of the 299 patients,158 were included in the training set group and 141 in the testing set group.The mean survival time in the training set group and the validation set group was 22.7 months(19.9-30.1 months)and 23.8 months(19.9-29.0 months)respectively(P=0.96).Univariate analysis indicated that gender,AFP,tumor size,albumin,AST,Child-Pugh stage,up-to-7 criteria,up-to-11 criteria and tumor response were well correlated with the prognosis(P<0.05).Multivariate analysis revealed that AFP,up-to-11 criteria and tumor response were independent prognostic factors(P<0.05).The C-index scores obtained by nomogram prediction model,which was established on the basis of independent prognostic factors in the training set group and the validation set group,were 0.700(95%CI:0.693-0.707)and 0.647(95%CI:0.637-0.657)respectively.The AUROC values and C index scores of one-year and 2-year survival rate obtained by nomogram prediction model were higher than those obtained by other models including HAP,BCLC B sub-classification,new BCLC B sub-classification,etc.Conclusion The nomogram prediction model can well predict the survival time in patients with HCC of BCLC B stage after receiving TACE.Both the prediction accuracy and effectiveness of this nomogram prediction model are much better than other models.The use of this model is beneficial to patients and is helpful in making stratification of prognosis.