摘要目的 探讨能有效预测肝衰竭发生风险的乙型肝炎病毒(HBV)相关慢加急性肝衰竭前期(pre-ACLF)的诊断标准. 方法 收集重度黄疸的慢性乙型肝炎(CHB)和(或)CHB严重急性加重患者1 279例,分析不同血清水平的丙氨酸氨基转移酶(ALT)、天冬氨酸氨基转移酶(AST)、总胆红素(TBil)、血浆凝血酶原时间国际化比率(INR)及性别与年龄对患者慢加急性肝衰竭(ACLF)发生率的影响,并制订出pre-ACLF的诊断标准和ACLF发生预测模型.分类变量的比较用x2检验,连续变量采用独立样本t检验;多因素logistic回归分析用于评估患者的肝衰竭发生风险. 结果 基线ALT、AST、TBil与INR水平和患者年龄是影响肝衰竭发生的独立危险因素,P＜ 0.05.pre-ACLF的诊断标准是:(1)INR≥1.30;(2) AST≥10倍正常值上限(ULN)且有明显黄疸(TBil≥51.3 μ mol/L),或者TBil≥342.0μmol/L.该标准的阳性预测值与阴性预测值分别为45.9％和89.8％,灵敏度与特异度分别为69.1％和76.9％.ACLF发生风险的预测模型为:PY_1=eX/(1+ex) (PY_1表示logistic回归分析阳性),X=-10.245+ 0.026×AST (ULN)-0.025 ×AST(ULN)+0.046×TBil (mg/dl,1 mg/dl=17.1 μmol/L)+4.642×INR+0.049×年龄(岁).PY值越高者的ACLF发生率也越高,PY≥0.60患者ACLF发生率高达75.3％;0.40 ～ 0.59患者ACLF发生率超过50％;＜0.10患者ACLF发生率仅为1.8％,P＜0.01. 结论 本pre-ACLF诊断标准与预测模型能有效评估HBV相关ACLF的发生风险.

abstractsObjective To investigate the diagnostic criteria for HBV-related acute-on-chronic pre-liver failure (pre-ACLF) which can effectively predict the risk of liver failure.Methods A total of 1279 patients with severe icteric chronic hepatitis B (CHB) and/or severe acute exacerbation of CHB were enrolled.The influence of serum levels of alanine aminotransferase (ALT),aspartate aminotransferase (AST),and total bilirubin (TBil),international normalized ratio (INR) of prothrombin time,sex,and age on the incidence rate of acute-on-chronic liver failure (ACLF) was analyzed,the diagnostic criteria for pre-ACLF and predictive model for ACLF were developed.The chi-square test was used for comparison of categorical variables,and the independent samples t-test was used for continuous data;multivariate logistic regression analysis was performed to evaluate the risk of liver failure.Results The baseline serum levels ofALT,AST,and TBil,and INR were independent risk factors for liver failure (P ＜ 0.05).The diagnostic criteria for pre-ACLF were as follows:(1) INR ≥ 1.30;(2) AST ≥ 10×upper limit of normal (ULN) and obvious jaundice (TBil ≥ 51.3 μmol/L),or TBil ≥ 342.0 μmol/L.These criteria had a positive predictive value of 45.9％,a negative predictive value of 89.8％,a sensitivity of 69.1％,and a specificity of 76.9％.The predictive model for the risk of ACLF was PY =1=eX/(1+ex) (PY represented positive results of logistic regression analysis),X =-10.245+0.026×AST(ULN)-0.025×AST(ULN)+0.046×TBil(mg/d1) + 4.642×INR+0.049×age(years).The patients with higher PY values tended to have a higher incidence rate of ACLE The incidence rate of ACLF was 75.3％ in patients with PY ≥ 0.60,more than 50％ in patients with a PY value of 0.40-0.59,and 1.8％ in patients with PY ＜ 0.10 (P ＜ 0.01).Conclusion The diagnostic criteria for pre-ACLF and predictive model can effectively evaluate the risk of HBV-related ACLF.