Sonika Ujjwal, Jadaun Shekhar, Ranjan Gyan, Rout Gyanranjan, Gunjan Deepak, Kedia Saurabh, Nayak Baibaswata
Department of Gastroenterology and Human Nutrition, All India Institute of Medical Sciences, Ansari Nagar, New Delhi, 110 029, India.
Indian J Gastroenterol. 2018 Jan;37(1):50-57. doi: 10.1007/s12664-018-0827-z. Epub 2018 Feb 23.
Various prognostic scores are available for predicting outcome in acute-on-chronic liver failure (ACLF). We compared the available prognostic models as predictors of outcome in alcohol-related ACLF patients.
All consecutive patients with alcohol-related ACLF were included. At admission, prognostic indices-acute physiology and chronic health evaluation score (APACHE II), model for end-stage liver disease (MELD), MELD-Na, Maddrey's discriminant function (DF), age-bilirubin-INR-creatinine (ABIC), and Chronic Liver Failure Consortium (CLIF-C) ACLF score (CLIF-C ACLF) score were calculated. Receiver operator characteristic (ROC) curves were plotted for all prognostic scores with in-hospital, 90-day, and 1-year mortality as outcome.
Of the 171 patients, 170 were males, and grade 1 ACLF in 20 (11.7%), grade 2 in 52 (30.4%), and grade 3 in 99 (57.9%) patients. One hundred and nineteen (69.6%) died in-hospital. The median (IQR) Maddrey's score, MELD, MELD-Na, ABIC, APACHE II, and CLIF-C ACLF were 87.8 (66.5-123.0), 33.1 (27.6-40.0), 34.4 (29.5-40.0), 8.5 (7.3-9.6), 15 (12-21), and 51.1 (44.1-56.4), respectively. On multivariate Cox regression analysis, independent predictors of in-hospital outcome were presence of hepatic encephalopathy (early HR, 2.078; 95%CI, 1.173-3.682, p = 0.012 and advanced, HR, 2.330; 95% CI, 1.270-4.276, p = 0.006), elevated serum creatinine (HR, 1.140; 95% CI, 1.023-1.270, p = 0.018), and infection at admission (HR, 1.874; 95% CI, 1.160-23.029, p = 0.010). On comparison of ROC curves, APACHE II and CLIF-C ACLF AUROC were significantly higher than MELD, MELD-Na, DF, and ABIC (p < 0.05) for predicting in-hospital, 90-day, and 1-year mortality. The AUROC was highest for APACHE II followed by CLIF-C ACLF (Hanley and McNeil, p = 0.660).
Alcohol-related ACLF has high in-hospital mortality. Among the available prognostic scores, CLIF-C ACLF and APACHE II perform best.
有多种预后评分可用于预测慢加急性肝衰竭（ACLF）的结局。我们比较了现有的预后模型作为酒精性ACLF患者结局预测指标的情况。
纳入所有连续性酒精性ACLF患者。入院时计算预后指标——急性生理与慢性健康状况评分系统Ⅱ（APACHEⅡ）、终末期肝病模型（MELD）、MELD-Na、马德雷判别函数（DF）、年龄-胆红素-国际标准化比值-肌酐（ABIC）以及慢性肝衰竭协作组（CLIF-C）ACLF评分（CLIF-C ACLF）。以院内、90天和1年死亡率为结局指标，绘制所有预后评分的受试者工作特征（ROC）曲线。
171例患者中，170例为男性，20例（11.7%）为1级ACLF，52例（30.4%）为2级，99例（57.9%）为3级。119例（69.6%）患者院内死亡。马德雷评分、MELD、MELD-Na、ABIC、APACHEⅡ和CLIF-C ACLF的中位数（四分位间距）分别为87.8（66.5 - 123.0）、33.1（27.6 - 40.0）、34.4（29.5 - 40.0）、8.5（7.3 - 9.6）、15（12 - 21）和51.1（44.1 - 56.4）。多因素Cox回归分析显示，院内结局的独立预测因素为肝性脑病（早期HR，2.078；95%CI，1.173 - 3.682，p = 0.012；晚期HR，2.330；95%CI，1.270 - 4.276，p = 0.006）、血清肌酐升高（HR，1.140；95%CI，1.023 - 1.270，p = 0.018）以及入院时感染（HR，1.874；95%CI，1.160 - 23.029，p = 0.010）。比较ROC曲线发现，APACHEⅡ和CLIF-C ACLF的曲线下面积（AUROC）在预测院内、90天和1年死亡率方面显著高于MELD、MELD-Na、DF和ABIC（p < 0.05）。APACHEⅡ的AUROC最高，其次是CLIF-C ACLF（Hanley和McNeil检验，p = 0.660）。
酒精性ACLF院内死亡率高。在现有的预后评分中，CLIF-C ACLF和APACHEⅡ表现最佳。
