列线图预测乙型肝炎病毒相关慢加急性肝衰竭死亡风险的临床研究

Nomogram prediction of individual prognosis of patients with hepatitis B virus-associated acute on chronic liver

目的:构建可用于预测乙型肝炎病毒相关慢加急性肝衰竭患者结局的风险模型。方法:以国家“十二五”科技重大专项课题“慢加急性肝衰竭中西医结合治疗方案优化研究”入组的乙型肝炎病毒相关慢加急性肝衰竭患者为研究对象,按照2∶1分为训练集、测试集。在训练集人群中建立死亡风险模型并通过Cox回归分析,在测试集中验证模型的区分度与一致性。结果:年龄(HR=1.02,95%CI 1.01~1.04)、总胆红素(HR=1.04,95%CI 1.02~1.07)、国际标准化比值(HR=2.58,95%CI 1.98~3.37)、肝性脑病(Ⅰ/Ⅱ期:HR=2.20,95%CI 1.25~3.87;Ⅲ/Ⅳ期:HR=23.67,95%CI 7.74~72.37)、肝肾综合征(HR=3.64,95%CI 1.69~7.82)、低钠血症(HR=1.86,95%CI 1.22~2.84)、血小板计数(20~60:HR=1.42,95%CI 1.89~2.27;>60:HR=2.42,95%CI 1.94~6.20)是预后的独立影响因素;基于上述7项因素建立的列线图可以准确预测乙型肝炎病毒相关慢加急性肝衰竭人群结局;列线图在预测28、336 d结局方面优于终末期肝病模型(MELD)及MELD-Na评分(0.87 vs 0.79/0.80,0.82 vs 0.73/0.75);在预测90 d结局时优于MELD评分(0.86 vs 0.79,P=0.042),但与MELD-Na评分差异无统计学意义(0.86 vs 0.82,P=0.140)。结论:以年龄、总胆红素、国际标准化比值、肝性脑病、肝肾综合征、低钠血症、血小板计数等7个关键因素构建而成的列线图,在预测乙型肝炎病毒相关慢加急性肝衰竭患者死亡风险方面具有一定的价值。

Objective:To identify the independent prognostic factors,then establish a nomogram to predict the outcome for HBV ACLF patients.Methods:We performed an analysis of the“Study on HBV ACLF treated with integrated Traditional Chinese Medicine and western medicine”.HBV ACLF patients were divided into the test cohort and the train cohort.Screened the independent prognostic factors,established a nomogram to predict the outcome of HBV ACLF patients.The performance of the nomogram was identified in the test cohort.Results:Old age(HR=1.02,95%CI:1.01-1.04),high level of total bilirubin(TBil)(HR=1.04,95%CI:1.02-1.07),high level of international standardized ratio(INR)(HR=2.58,95%CI:1.98-3.37),the presence of HE(stageⅠ/Ⅱ:HR=2.20,95%CI:1.25-3.87;stageⅢ/Ⅳ:HR=23.67,95%CI:7.74-72.37),or hepatorenal syndrome(HRS)(HR=3.64,95%CI:1.69-7.82),or hyponatremia(HR=1.86,95%CI:1.22-2.84),and low level of platelet count(>20/≤60:HR=1.42,95%CI:1.89-2.27;>60:HR=2.42,95%CI:1.94-6.20)increased the mortality risk of HBV ACLF patients.The performance of nomogram using the area under the receiver operating characteristic curves(AUROC)was better than MELD and MELD-Na scores in predicting the outcomes of HBV ACLF patients at 28 days and 336 days(AUROC:0.87 vs 0.79/0.80,0.82 vs 0.73/0.75),and was better than MELD score(AUROC:0.86 vs 0.79,P=0.042)in predicting outcomes at 90 day,but was not different from MELD-Na score(AUROC:0.86 vs 0.82,P=0.140)at 90 day.Conclusion:The nomogram performed well in predicting the outcome of HBV ACLF patients.