According to European Association for the Study

Transcription

1 LIVER FAILURE/CIRRHOSIS/PORTAL HYPERTENSION Acute-on-Chronic Liver Failure Precipitated by Hepatic Injury Is Distinct From That Precipitated by Extrahepatic Insults Yu Shi, 1 Ying Yang, 1 Yaoren Hu, 2 Wei Wu, 1 Qiao Yang, 3 Min Zheng, 1 Shun Zhang, 4 Zhaojun Xu, 5 Yihua Wu, 6 Huadong Yan, 2 and Zhi Chen 1 Patients with acute-on-chronic liver failure (ACLF) represent a heterogeneous population. The aim of the study is to identify distinct groups according to the etiologies of precipitating events. A total of 405 ACLF patients were identified from 1,361 patients with cirrhosis with acute decompensation and categorized according to the types of acute insults. Clinical characteristics and prognosis between the hepatic group and extrahepatic group were compared, and the performance of prognostic models was tested in different groups. Two distinct groups (hepatic-aclf and extrahepatic-aclf) were identified among the ACLF population. Hepatic-ACLF was precipitated by hepatic insults and had relatively wellcompensated cirrhosis with frequent liver and coagulation failure. In contrast, extrahepatic- ACLF was exclusively precipitated by extrahepatic insults, characterized by more severe underlying cirrhosis and high occurrence of extrahepatic organ failures (kidney, cerebral, circulation, and respiratory systems). Both groups had comparably high short-term mortality (28-day transplant-free mortality: 48.3% vs. 50.7%; P ); however, the extrahepatic-aclf group had significantly higher 90-day and 1-year mortality (90-day: 58.9% vs. 68.3%, P ; 1-year: 63.9% vs. 74.6%, P ). In hepatic-aclf group, the integrated Model for End-Stage Liver Disease (imeld) score had the highest area under the receiver operating characteristic curve (auroc ) among various prognostic models in predicting 28-day mortality, whereas CLIF-Consortium scores for ACLF patients (CLIF-C-ACLF) had the highest predictive value in the other group (auroc ). Conclusions: ACLF precipitated by hepatic insults is distinct from ACLF precipitated by extrahepatic insults in clinical presentation and prognosis. The imeld score may be a better predictor for hepatic-aclf short-term prognosis, whereas CLIF-C-ACLF may be better for extrahepatic-aclf patients. (HEPATOLOGY 2015;62: ) According to European Association for the Study of Liver Diseases/American Association for the Study of Liver Diseases consensus, the term acute-on-chronic liver failure (ACLF) describes patients hospitalized for acute deterioration of cirrhosis as the result of a precipitating event who develop organ failure(s) and have high short-term mortality. 1 In a recent prospective observational study, known as Chronic Liver Failure (CLIF) Acute-on-Chronic Liver Failure in Cirrhosis (CANONIC), provided an Abbreviations: ACLF, acute-on-chronic liver failure; AD, acute decompensation; ALT: alanine aminotransferase; AST, aspartate aminotransferase; auroc, area under the receiver operating curve; CANONIC, Chronic Liver Failure (CLIF) Acute-on-Chronic Liver Failure in Cirrhosis; CHB, chronic hepatitis B; CLIF-C- ACLFs, CLIF-Consortium scores for ACLF patients; CLIF-SOFA, Chronic Liver Failure Sequential Organ Failure Assessment; CTP, Child-Turcotte-Pugh; HAV, hepatitis A virus; HBV, hepatitis B virus; HE, hepatic encephalopathy; HEV, hepatitis E virus; HIV, human immunodeficiency virus; Ig, immunoglobulin; IQR, interquartile range; imeld, integrated MELD; INR, international normalized ratio; IV, intravenous; MAP, mean arterial pressure; MELD, Model for End-Stage Liver Disease; SBP, spontaneous bacterial peritonitis; SOFA, Sequential Organ Failure Assessment; SBP, spontaneous bacterial peritonitis; SD, standard deviation; TB, total bilirubin; UGIB, upper gastrointestinal bleeding; VB, variceal bleeding; WBC, white blood cell. From the 1 State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, Collaborative Innovation Center for Diagnosis and Treatment of Infectious Disease, The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China; 2 Department of Hepatology, Ningbo No. 2 Hospital, Ningbo, China; 3 Department of Infectious Diseases, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China; 4 Stem Cell Laboratory, Ningbo No. 2 Hospital, Ningbo, China; 5 Department of Intensive Care Unit, Ningbo No. 2 Hospital, Ningbo, China; and 6 Department of Epidemiology and Health Statistics, Zhejiang University School of Public Health, Hangzhou, China Received August 17, 2014; accepted March 15, Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/ /hep /suppinfo 232

2 HEPATOLOGY, Vol. 62, No. 1, 2015 SHI ET AL. 233 evidence-based definition of ACLF as: the presence of organ failure (defined by the Chronic Liver Failure Sequential Organ Failure Assessment [CLIF-SOFA] score) and high 28-day mortality (>15%). 2 From the CANONIC study, ACLF was identified as a new clinical entity distinct from the traditional decompensated cirrhosis, and the definition of ACLF has been validated by several studies However, according to the CANONIC definition, ACLF patients remain a heterogeneous population, which varies in etiologies of cirrhosis, types of precipitating events, and presentation of organ failures. 2,11 Therefore, it is likely that there are distinct subgroups of ACLF that had different natural history and prognosis. The CANONIC study revealed significantly different clinical characteristics between ACLF patients with alcoholic cirrhosis with or without recent active alcohol consumption, 2,11 suggesting that precipitating events may help identify distinct groups of ACLF. 1 Therefore, in the study, we retrospectively identified 405 ACLF patients from a prospective cohort of 1,361 patients with cirrhosis with acute decompensation in our center and explored the possibility of subcategorization of ACLF by etiologies of acute insults in these patients. Patients and Methods Patients. Patients with cirrhosis with a new episode of acute decompensation (AD) from a prospective cohort referred between January 1, 2009 and June 30, 2013 in Ningbo No. 2 Hospital (Ningbo, China) was screened. Events of acute decompensation included overt ascites, hepatic encephalopathy (HE), upper gastrointestinal bleeding (UGIB), bacterial infection, and acute severe hepatic damage (defined as total bilirubin [TB] 85 mmol/l and international normalized ratio [INR] 1.5 within 1 month after disease onset). Cirrhosis was diagnosed by liver biopsy, endoscopic signs of portal hypertension, radiological evidence of liver nodularity, or clinical evidence of previous hepatic decompensation (including ascites, HE, and UGIB) in patients with chronic liver diseases. 12 Overt ascites was diagnosed by clinical examination and confirmed by ultrasonography. 13 HE was defined and graded by West Haven criteria. 14 ACLF was diagnosed according to CLIF-SOFA score, which identified organ failures of liver, coagulation, kidney, circulation, lung, and cerebral systems. Patients who had more than two organ failures, single kidney failure (creatinine 2.0 mg/dl or 176 umol/l or need for renal replacement therapy), or one organ failure with presence of kidney dysfunction (1.5 mg/ dl or 132 umol/l creatinine <2.0 mg/dl or 176 umol/l) and/or mild-to-moderate HE were diagnosed as ACLF (liver: serum bilirubin 12 mg/dl or 204 umol/l; coagulation: INR 2.5 or platelet count /L; kidney: creatinine 2.0 mg/dl or 176 umol/l or need for renal replacement therapy; circulation: mean arterial pressure (MAP) <70 mmhg despite adequate fluid resuscitation and need for vasoactive agents; lung: PaO 2 /FiO or SpO 2 /FiO or need for mechanical ventilation; cerebral: HE grade III or IV). The CLIF-SOFA scale was used retrospectively. The exclusion criteria included: (1) age <18; (2) those who might be pregnant; (3) human immunodeficiency virus (HIV) infection; (4) those who had hepatocellular carcinoma outside Milan criteria or other types of tumors; (5) chronic renal diseases (estimated glomerular filtration rate [egfr] <90 ml/min) and other severe comorbidities, such as myocardial infarction, subarachnoid or cerebral hemorrhage, severe trauma, and chronic cor pulmonale; and (6) those who had received a liver transplant. The study fulfilled the principles of the Declaration of Helsinki and was approved by the ethics committee of Ningbo No. 2 Hospital. Written consent was retrospectively obtained from each participant or their legal representatives. Data Collection. We collected the following clinical and demographic information in a prespecified datasheet: age; sex; etiologies of cirrhosis and precipitating events; history (including previous episodes of hepatic decompensation); laboratory parameters; events of organ failures; and prognosis. Patients who were diagnosed with ACLF at hospital admission or This work was supported by the grants from the 12-5 State S&T Projects of China (2012ZX ), Chinese National Natural Science Foundation (Nos and ), China Foundation for Hepatitis Prevention and Control, Research Funds for Tian Qing Liver Diseases (TQGB ), and support from the Ningbo Science and Technology Innovation Team Project (2011B82016). Address reprint requests to: Huadong Yan, M.D., Department of Hepatology, Xibei Street No. 41, Ningbo No. 2 Hospital, Ningbo, China. huadongy588@163.com; fax: Copyright VC 2015 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /hep Potential conflict of interest: Nothing to report.

3 234 SHI ET AL. HEPATOLOGY, July 2015 developed ACLF during hospital stay were both entered into the analysis. For all patients, data were collected at diagnosis of ACLF and/or admission. Acute precipitating events were categorized into hepatic insults and extrahepatic insults. Hepatitis A virus (HAV) and hepatitis E virus (HEV) superimposed infection was diagnosed by the presence of anti- HAV/HEV serum immunoglobulin (Ig)M and IgG. 15 Active alcohol drinking was defined as more than 14 drinks per week in women and more than 21 drinks per week in men. 2 Use of hepatotoxic drugs should be within 3 months before admission. Acute exacerbation or flare-up of chronic hepatitis B (CHB) was defined as an upsurge of alanine aminotransferase (ALT) more than 5 times of the upper limit of normal value or more than twice of the baseline value with HBV-DNA positive within 1 month before admission. Events causing secondary acute exacerbation or flare-up of CHB included inappropriate withdrawal of nucleos (t)ide analogs and use of immunosuppressive drugs (such as glucocorticoid) or chemotherapy. Spontaneous exacerbation or flare-up of CHB was considered only when other types of hepatic insults were excluded in patients with hepatitis B virus (HBV)-related cirrhosis. Diagnosis criteria of different bacterial infections were the following, as previously described 12 : (1) spontaneous bacterial peritonitis (SBP): ascitic fluid polymorphonuclear cells >250/mL; (2) bacteremia: positive blood cultures without a source of infection; (3) pneumonia: new pulmonary infiltrate with fever (T >38 C), any respiratory symptoms (such as cough, sputum, dyspnea, or pleuritic pain), any findings on auscultation (rales or crepitation) or white blood cell (WBC) count >10,000/mm 3 or <4,000/mm 3 ; (4) urinary tract infection: urine WBC >10/high power field with positive culture and symptom of urinary irritation; and (5) other bacterial infections, including skin infection, intra-abdominal infection, and infection of unknown reason. Management of ACLF patients ranged from treatment against acute precipitating events/complications to therapy correcting organ dysfunction/failure. To eliminate or control precipitating events/complications, patients with HBV-DNA positive were immediately given nucleoside analogs (lamivudine alone 100 mg, telbivudine alone 600 mg, entecavir alone 0.5 mg, or lamivudine 100 mg plus adefovir 10 mg daily). Patients using hepatotoxic drugs or actively drinking alcohol were required to stop using or abstain from alcohol. Patients with bacterial infection were immediately treated with empirical antibiotic therapy, and adjustment of antibiotic therapy was based on bacteria culture and antibiotic sensitivity test. In addition, weight-based intravenous (IV) albumin was used, especially in patients with SBP; furthermore, all patients with acute variceal bleeding (VB) received IV somatostatin, proton pump inhibitors, and antibiotic prophylaxis. For those with uncontrolled hemorrhage resulting from pharmacological therapy, Sengstaken- Blackmore tube and/or urgent therapeutic endoscopy were performed. Diagnostic abdominocentesis was performed in patients with ascites to examine whether SBP was present. Those with moderate ascites were treated with restriction of sodium intake and/or diuretics (aldosterone antagonist and/or furosemide). Paracentesis combined with IV albumin was used in those with large or refractory ascites. Patients with renal failure were treated with IV albumin, vasoconstrictors (such as dopamine, noradrenaline, or terlipressine), or renal replacement therapy, and potential precipitating events were investigated. Patients with HE were given lactulose, antibiotics, and L-ornithine aspartate, and potential precipitating events were investigated. Fluid replacement was performed in patients with MAP <70 mmhg, and vasoactive agents were used when necessary. Oxygen therapy was performed in patients with decreased PaO 2 or SpO 2 ; nasal catheter, mask, or venturi mask oxygen inhalation or mechanical ventilation were chosen according to severity of respiratory dysfunction. Last, IV high glucose, albumin, or plasma and use of glutathione, adenosylmethionine, or branched-chain amino acids were used as nutritional support. Follow-up of patients started at onset of hospital admission. For patients discharged from the hospital, prognostic information was obtained from medical records, telephone contact, or personal visit. The primary endpoint of the study was 28-day, 90-day, and 1-year transplant-free mortality. Model Calculation. Prognostic models used in predicting the 28-day mortality of ACLF patients included: Child-Turcotte-Pugh (CTP); Model for End-Stage Liver Disease (MELD); the integrated MELD (imeld); Sequential Organ Failure Assessment (SOFA); CLIF- SOFA; and CLIF-Consortium-ACLF (CLIF-C-ACLF) score. CTP score is measured by ascites, HE, albumin, serum bilirubin, and INR, ranging from 0 to MELD score (range, 6-40) was calculated as follows: loge [creatinine (mg/dl)] loge [bilirubin (mg/dl)] loge (INR) (etiology: 0 if cholestatic or alcoholic, 1 otherwise). 17 imeld model for ACLF patients was recently proposed and calculated as follows: 0.030*age *(HE score) * MELD. 18 SOFA score (range, 0-24) is calculated by the

4 HEPATOLOGY, Vol. 62, No. 1, 2015 SHI ET AL. 235 sum of scores for six organ systems, including respiratory, cardiovascular, renal, neurological systems, liver, and coagulation. 19 CLIF-SOFA score (range, 0-24) comprises the same six organ systems as SOFA and is proposed to evaluate organ failures in ACLF patients. 2 CLIF-C-ACLF score is modified based on the CLIF- SOFA scale and calculated as follows: 10*[0.33*CLIF- OFs *Age *Ln(white-cell count) 2]. 5 Statistical Analysis. Continuous variables were expressed as mean 6 standard deviation (SD) or median with interquartile range (IQR). Continuous data between two groups were compared by Student t test or Mann-Whitney s U test. Nominal variables were expressed as number/percentage and compared using chi-square test. Survival curves of different groups were drawn and the overall or pair-wise comparison of cumulative survival was performed by the log-rank test. A Cox s proportional hazard model was used to explore predictors of time-dependent death of ACLF patients. Candidate variables (P < 0.10) after a bivariate analysis were entered into a multivariate Cox s model by a backward-forward approach. In addition, a multivariate logistic regression analysis was performed to identify risk factors associated with development of ACLF after admission. For multivariate analysis, the entry and removal probability for step-wise method was set as 0.05 and 0.10, respectively, and variables with P < 0.05 were kept in the final model. To compare the predictive value of different prognostic scoring systems, area under the receiver operating curve (auroc) was calculated and compared by Z test (Delong s method). Statistical analyses were performed using SPSS (16.0; SPSS, Inc., Chicago, IL), GraphPad Prism (5; GraphPad Software Inc., San Diego, CA), or MedCalc software (MedCalc Software, Belgium). Results Categorization of ACLF Subgroups. In a total of 1,361 patients with cirrhosis with AD, 266 with ACLF were identified at admission (Fig. 1). Clinical characteristics of ACLF and no-aclf patients at admission are shown in Supporting Tables 1-5. In addition, 139 patients developed ACLF during their hospital stay (see Supporting Fig. 1). Overall, a total of 405 ACLF patients were identified. Types of acute insults among 405 patients were listed in Table.1. Among hepatic insults, flare-up or exacerbation of HBV (35.8%) was the most common, followed by superimposed HAV or HEV infection (6.4%). A total of 6.9% patients had more than one hepatic insults. Furthermore, bacterial infection was the most frequent extrahepatic insult (27.9%). Two percent of patients had more than one extrahepatic insults. A total of 7.9% patients had mixed hepatic and extrahepatic insults. In addition, 20.5% patients had no definite acute insult. We categorized 322 ACLF patients with definite precipitating events into two groups according to types of acute insults: patients precipitated by hepatic insults (hepatic-aclf) and those precipitated exclusively by extrahepatic insults (extrahepatic-aclf). [Correction added on May 6, 2015, after first online publication: In preceding sentence, 332 ACLF patients was changed to 322 ACLF patients. ] Comparison of Clinical Characteristics Between Hepatic-ACLF and Extrahepatic-ACLF Patients. At the time point of ACLF diagnosis, patients with hepatic-aclf and extrahepatic-aclf differed from one another by various aspects (Table 2). Hepatic-ACLF patients were younger than those with extrahepatic- ACLF. Most hepatic-aclf patients had HBV-related cirrhosis (90.0%), with a history of previous hepatic decompensation in only 29.4%. In contrast, patients with extrahepatic-aclf had a higher proportion of non-hbv cirrhosis, and underlying liver diseases were more severe with a history of hepatic decompensation in 87.3%. Hepatic injury-related parameters, such as ALT, aspartate aminotransferase (AST), serum bilirubin, and INR, were significantly higher in the hepatic-aclf group, whereas extrahepatic-aclf patients had higher levels of WBC count. With regard to organ failures, liver and coagulation failures were highly prevalent in hepatic-aclf patients, whereas those with extrahepatic- ACLF developed higher frequency of extrahepatic organ failures (kidney, circulation, respiratory, and cerebral systems). Furthermore, the extrahepatic-aclf group had significantly higher occurrence of renal dysfunction than those with hepatic-aclf (19.0% vs. 9.4%; P ). Though cerebral failure was more frequent in the extrahepatic-aclf group (19.0% vs. 8.3%; P ), both groups had high occurrence of mildto-moderate hepatic encephalopathy (28.3% vs. 30.3%; P ). Comparison of Prognosis and Mortality Predictors Between Hepatic-ACLF and Extrahepatic- ACLF. Both groups had high 28-day mortality (48.3% vs. 50.7%; P ; Table 2; Fig. 2). However, 90-day (58.9% vs. 68.3%; P ) and 1-year mortality (63.9% vs. 74.6%; P ) was significantly higher in patients in the extrahepatic-aclf group than those in the hepatic-aclf group. Whether the two ACLF groups had differential predictors of 28-day mortality was then examined by a

5 236 SHI ET AL. HEPATOLOGY, July 2015 Fig. 1. Flow chart of ACLF patient admission. Abbreviation: LT, liver transplantation. multivariate Cox s proportional hazard model (candidate variables were selected by a bivariate analysis; see Supporting Table 6). Older age, high WBC count, high levels of INR, creatinine and TB, and presence of HE were independently associated with poor prognosis in patients precipitated by hepatic-related insults, whereas high WBC count, high creatinine level, presence of HE, and respiratory and circulation failure were predictive of 28-day mortality in ACLF patients precipitated exclusively by extrahepatic insults (see Table 3). Comparison of Prognostic Models in Hepatic- ACLF and Extrahepatic-ACLF. Seven prognostic scoring systems were tested in predicting 28-day, 90- day, and 1-year mortality of ACLF patients in different groups. Among hepatic-aclf patients, we found that imeld model had the highest auroc (28-day: 0.787; 90-day: 0.785; 1-year: 0.784). Furthermore, for patients with extrahepatic-aclf, CLIF-C-ACLFs had the highest predictive value, with an auroc of (28-day), (90-day) and (1-year) (see Table 4 and Fig. 3). Discussion ACLF is a new clinical entity defined as acute deterioration of preexisting, chronic liver disease due to a precipitating event and associated with high short-term mortality. 20 Though an evidence-based diagnostic criterion was proposed by the CANONIC study, ACLF patients remain a heterogeneous population, with variation in etiologies of acute insults and underlying cirrhosis, as well as other clinical traits. 11 The present study categorized ACLF patients according to the types

6 HEPATOLOGY, Vol. 62, No. 1, 2015 SHI ET AL. 237 Items Table 1. Etiology of Precipitating Events for All Diagnosed ACLF Patients Frequency With hepatic insults (N 5 180) Hepatic insults alone Flare-up or exacerbation of HBV, no. (%) 145 (35.8) Superimposed HAV or HEV infection, no. (%) 26 (6.4) Hepatotoxic drugs, no. (%) 10 (2.5) Active drinking, no. (%) 25 (6.2) Flare-up of AIH or Wilson s disease, no. (%) 6 (1.5) More than one hepatic insult, no. (%) 28 (6.9) Mixed with extrahepatic insults Bacterial infection, no. (%) 28 (6.9) UGIB, no. (%) 4 (1.0) Extrahepatic insults alone (N 5 142) Bacterial infection, no. (%) 113 (27.9) UGIB, no. (%) 36 (8.9) Surgery, no. (%) 1 (0.2) More than one extrahepatic insult, no. (%) 8 (2.0) Unknown (N 5 83) 83 (20.5) Data are expressed as number (percent). Abbreviation: AIH, autoimmune hepatitis; of precipitating events and found that patients precipitated by hepatic insults and those by extrahepatic insults had distinct clinical phenotypes and prognosis. The driving difference between the hepatic-aclf and extrahepatic-aclf groups was the distinction in etiology of precipitating events. The former was induced by hepatic insults, such as acute exacerbation or flare-up of CHB, superimposed infection of HAV or HEV, active drinking of alcohol, use of hepatotoxic drugs, and so on, whereas the latter was exclusively precipitated by extrahepatic insults, such as bacterial infection and UGIB. It should be noted that some patients with hepatic-aclf were complicated with bacterial infection. Though bacterial infection could directly induce ACLF, 2 it was more likely that bacterial infection serves as a second hit, but not another initial precipitating event of ACLF, in those patients with definite hepatic insults. Susceptibility to bacterial infection of ACLF patients might be associated with immunosuppression status. 21 This study also noted that the constitution of different types of acute insults in ACLF was closely related to the severity of cirrhosis in these patients. ACLF patients exclusively precipitated by extrahepatic insults had significantly higher frequency of previous decompensation than those precipitated by hepatic insults. The finding suggested that extrahepatic insults alone could induce ACLF in patients with cirrhosis with previous decompensation, but hardly in those with well-compensated cirrhosis. This is in line with the hypothesis that the decrease in the tolerance of liver could sensitize it to less-intensive damage. 1 Also, it was supported by the CANONIC study, which showed that patients with previous decompensation developed a less-severe form of ACLF than those without previous decompensation. 2 Presentation of organ failures was distinct between the two groups as well. Liver and coagulation failures were much more frequent in the hepatic-aclf group. Because level of INR is a sensitive marker reflecting the synthesizing capacity of the liver, 22 coagulation failure in CLIF-SOFA scale may also reflect liver failure in patients with hepatic insults, provided that severe sepsis was excluded. Thus, we defined ACLF patients in the hepatic-aclf group as hepatic inflammation predominant. In contrast, patients with extrahepatic-aclf were characterized by high occurrence of extrahepatic organ failures (such as kidney, cerebral, circulation, or respiratory failures), which we could define as extrahepatic organ failure predominant. These findings supported the concept that liver failure is not indispensable for the diagnosis of ACLF 2,23 in patients precipitated by extrahepatic insults. Both groups had high occurrence of HE, though the mechanisms may be different. 24,25 We hypothesized that HE in hepatic-aclf patients were likely to be precipitated by systemic inflammation and dysfunction because the ACLF subgroup had similar clinical presentation to acute liver failure patients who previously had normal livers. In patients with decompensated cirrhosis, HE may be complicated with portal-systemic shunting. 26 Further studies are needed to compare the occurrence of brain edema and intracranial hypertension, which is a typical feature secondary to HE in acute liver failure between the two groups. 27 Though the incidence of cerebral failure (III- IV grade HE) was found to be higher in patients with extrahepatic-aclf, a considerable proportion of patients in the hepatic-aclf group developed mildto-moderate HE. Because grouping analysis was performed at an early stage of ACLF, occurrence of cerebral failure may be underestimated among hepatic- ACLF patients. The comparison of survival curves between ACLF groups revealed that both subpopulations had comparably high short-term mortality (28-day). However, the short-term prognosis was distinct between hepatic- ACLF and extrahepatic-aclf. For patients precipitated by hepatic insults, the imeld score incorporating age and HE into MELD score best predicted 28- day prognosis. There are several potential advantages of this novel scoring system in hepatic-aclf: First, age was an independent predictor of prognosis in hepatic-aclf patients, but not in those with extrahepatic-aclf. We hypothesized that older age

7 238 SHI ET AL. HEPATOLOGY, July 2015 Table 2. Comparison of Demographic and Clinical Characteristics of Patients at Diagnosis of ACLF According to Etiology of Acute Insults Variables Total(N 5 322) Hepatic-ACLF(N 5 180) Extrahepatic-ACLF(N 5 142) P Value Age, years Male, no. (%) 232 (72.0) 130 (72.2) 102 (71.8) 0.94 Etiology of cirrhosis HBV, no. (%) 207 (64.3) 127 (70.6) 80 (56.3) Alcohol, no. (%) 36 (11.2) 7 (3.9) 29 (20.4) HBV plus alcohol, no. (%) 46 (14.3) 35 (19.4) 11 (7.7) <0.001 Others, no. (%) 18 (5.6) 7 (3.9) 11 (7.7) 0.13 Cryptogenic, no. (%) 15 (4.7) 4 (2.2) 11 (7.7) Previous decompensation Total 177 (55.0) 53 (29.4) 124 (87.3) <0.001 Ascites 160 (49.7) 46 (25.6) 114 (80.3) <0.001 HE 45 (14.0) 15 (8.3) 30 (21.1) <0.001 UGIB 42 (13.0) 4 (2.2) 38 (26.8) <0.001 SBP 33 (10.2) 4 (2.2) 29 (20.4) <0.001 Laboratory parameters ALT, IU/L 121 (353) 317 (579) 50 (81) <0.001 AST, IU/L 184 (369) 312 (501) 102 (120) <0.001 Albumin, g/dl <0.001 TB, mmol/l 304 (242) 368 (217) 208 (291) <0.001 Hemoglobin, g/l <0.001 INR 2.8 (1.2) 3.0 (1.1) 2.4 (1.3) <0.001 Serum sodium, mmol/l 134 (8) 136 (6) 132 (9) WBC count, 10 9 /L 7.0 (6.3) 6.6 (4.8) 7.6 (9.0) Creatinine, lmol/l 88 (95) 75 (39) 145 (122) <0.001 Platelet count, 10 9 /L 66 (52) 70 (53) 55 (45) <0.001 MAP, mmhg <0.001 Complications Ascites, no. (%) 288 (89.4) 160 (88.9) 124 (87.3) 0.67 UD ascites, no. (%) 59 (18.3) 44 (24.4) 15 (10.6) Overt ascites, no. (%) 225 (69.9) 116 (64.4) 109 (76.8) Bacterial infection, no. (%) 141 (43.8) 28 (15.6) 113 (79.6) <0.001 UGIB, no. (%) 40 (12.4) 4 (2.2) 36 (25.4) <0.001 Organ failure Liver, no. (%) 246 (76.4) 171 (95.0) 75 (52.8) <0.001 Kidney, no. (%) 66 (20.5) 9 (5.0) 57 (40.1) <0.001 Cerebral, no. (%) 42 (13.0) 15 (8.3) 27 (19.0) Coagulation, no. (%) 229 (71.1) 155 (86.1) 74 (52.1) <0.001 Lung, no. (%) 31 (9.6) 5 (2.8) 26 (18.3) <0.001 Circulation, no. (%) 36 (11.2) 4 (2.2) 32 (22.5) <0.001 Mild-to-moderate HE, no. (%) 94 (29.2) 51 (28.3) 43 (30.3) 0.70 Renal dysfunction, no. (%) 44 (13.7) 17 (9.4) 27(19.0) Severity score CTP 13.0 (1) 12.5 (1) 13.0 (2) MELD 28.2 (8.6) 28.9 (7.5) 27.1 (11.4) MELD-Na 29.2 (14.3) 28.1 (16.6) 32.1 (21.1) 0.60 imeld 5.2 (2.2) 5.1 (1.8) 5.5 (2.4) SOFA 8.0 (3) 7.0 (2) 9.0 (4) <0.001 CLIF-SOFA 9.0 (2) 8.0 (2) 10.0 (4) <0.001 CLIF-C-ACLFs 43.7 (14.6) 40.0 (11.5) 48.2 (16.7) <0.001 Cumulative transplant-free mortality 28-day, no. (%) 159 (49.4) 87 (48.3) 72 (50.7) day, no. (%) 203 (63.0) 106 (58.9) 97 (68.3) year, no. (%) 221 (68.6) 115 (63.9) 106 (74.6) Data are expressed as mean 6 SD, median (IQR), or number (percent). Comparisons between hepatic and extrahepatic groups were performed by Student t test, Mann-Whitney s U test, or a chi-square test. Pair-wise comparison of cumulative transplant-free mortality was performed by log-rank test. Abbreviation: UD, ultrasound-detected. correlates with more-advanced cirrhosis in the former group, but not the latter group. Second, HE was the strongest predictor of death in hepatic-aclf, and the presence and severity of HE was emphasized in this model. Third, serum bilirubin, creatinine, and INR, reflecting hepatic, coagulation, and renal failure (or dysfunction), were appropriately weighted in the established MELD model. But for extrahepatic-aclf

8 HEPATOLOGY, Vol. 62, No. 1, 2015 SHI ET AL. 239 Fig. 2. Comparison of 28- day, 90-day, and 1-year survival between ACLF subgroups. Cumulative survival across groups was compared using the log-rank test. patients, we found CLIF-C-ACLFs as the optimal prognostic models. One major advantage of CLIF-C- ACLFs was that it takes the extrahepatic organ, such as cerebral, renal, lung, and circulation failure/dysfunction, into account, each of which had a considerable occurrence in those patients and was associated with death. Another advantage was that CLIF-C-ACLFs incorporated WBC count, which measures the degree of inflammatory response and is an independent risk factor of poor prognosis in extrahepatic-aclf. Table 3. Risk Factors Associated With Transplant-Free 28-Day Mortality by a Multivariable Cox s Proportional Hazard Model in Hepatic-ACLF and Extrahepatic-ACLF Patients Variables RegressionCoefficient Hazard Ratio 95% Confidence Interval P Value Hepatic-ACLF (N 5 180) Serum sodium Age <0.001 WBC count TB Creatinine INR <0.001 HE grading <0.001 (0/I-II/III-IV) Extrahepatic-ACLF (N 5 142) AST Ascite grading (none/ud/overt) INR Presence of bacterial infection WBC count Creatinine HE grading (0/I-II/III-IV) Presence of lung failure* <0.001 Presence of circulation failure Statistical analysis was performed using a multivariable Cox s proportional hazard model. For ACLF patients with hepatic insults, variables entering multivariate analysis were age, WBC count, serum sodium, creatinine, TB, INR, and HE. For those with exclusive extrahepatic insults, variables entering multivariate analysis were WBC count, AST, creatinine, INR, HE, ascites, presence of bacteria infection, and respiratory and circulation failure (preliminary screening by a binary analysis; see Supporting Table 6). *Lung failure was defined as PaO 2 /FiO or SpO 2 /FiO or need for mechanical ventilation. Circulation failure was defined as need for vasoactive agents. Abbreviation: UD, ultrasound-detected.

9 240 SHI ET AL. HEPATOLOGY, July 2015 Table 4. Comparison of Predictive Value of Prognostic Scoring Systems in Hepatic-ACLF and Extrahepatic-ACLF Patients 28-day 90-day 1-year Models auroc (95%CI) Z Value P Value auroc (95%CI) Z value P Value auroc (95%CI) Z Value P Value Hepatic-ACLF (N 5 180) imeld ( ) CTP ( ) MELD ( ) MELD-Na ( ) SOFA ( ) CLIF-SOFA ( ) CLIF-C ACLFs ( ) Extrahepatic-ACLF (N 5 142) CLIF-C ACLFs ( ) CTP ( ) MELD ( ) MELD-Na ( ) imeld ( ) SOFA ( ) CLIF-SOFA ( ) ( ) 5.65 < ( ) ( ) ( ) 3.45 < ( ) ( ) ( ) ( ) 4.33 < ( ) ( ) ( ) ( ) ( ) ( ) ( ) 4.29 < ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) aurocs of different models were calculated and compared by Z test (Delong s method). Abbreviation: CI, confidence interval. Though hepatic-aclf and extrahepatic patients had similar 28-day mortality, our results showed that the 90-day and 1-year survival of the extrahepatic group were significantly higher than the hepatic group, suggesting a considerable delayed death in the extrahepatic group. It was possible that the recovery of organ failure would be delayed in those with more severe underlying cirrhosis, during which second insults, including Fig. 3. Receiver operating characteristic curves of prognostic models in predicting 28-day, 90-day, and 1-year mortality in ACLF groups.

10 HEPATOLOGY, Vol. 62, No. 1, 2015 SHI ET AL. 241 bacterial infections, may exacerbate the disease. Further studies were needed to address the issue. It should be noted that the composition of ACLF patients in the present study was different from those in Western countries, given that the etiologies of precipitating events and cirrhosis are distinct by geographical location across the world. 28 ACLF patients in Western societies typically have alcoholic or HCVrelated cirrhosis and precipitated mostly by nonviral insults, such as bacterial infection and VB, or hepatic insults, such as active drinking of alcohol. 29 In this study, those with HBV-related cirrhosis and precipitated by HBV flare-up or superimposed infection of other hepatitis viruses were the most common. Thus, it was not surprising that a considerable proportion of ACLF patients in the study did not have a history of hepatic decompensation. Another caution is that the grouping of ACLF patients was performed at disease diagnosis, which may underestimate the occurrence of organ failures in the natural history of ACLF. In summary, this study shed light on the subcategorization of the complex ACLF population. The two groups had distinct etiologies and natural history, and the grouping of ACLF patients would be a helpful tool for optimization of treatment and prognosis. Acknowledgments: The authors thank Dr. Jiawei Li (Royal Free Hospital, London, UK) for proofreading of the manuscript and Wenjie Sun for technical assistance (Zhejiang University, Hangzhou, China) in statistical analysis. References 1. Olson JC, Kamath PS. Acute-on-chronic liver failure: concept, natural history, and prognosis. Curr Opin Crit Care 2011;17: Moreau R, Jalan R, Gines P, Pavesi M, Angeli P, Cordoba J, et al.; CANONIC Study Investigators of the EASL CLIF Consortium. Acute-on-chronic liver failure is a distinct syndrome that develops in patients with acute decompensation of cirrhosis. Gastroenterology 2013;144: e Silva PE, Fayad L, Lazzarotto C, Ronsoni MF, Bazzo ML, Colombo BS, et al. Single-centre validation of the EASL-CLIF Consortium definition of acute-on-chronic liver failure and CLIF-SOFA for prediction of mortality in cirrhosis. Liver Int 2014 May 20. doi: / liv [Epub ahead of print]. 4. Bajaj JS, O Leary JG, Reddy KR, Wong F, Biggins SW, Patton H, et al. Survival in infection-related acute-on-chronic liver failure is defined by extrahepatic organ failures. HEPATOLOGY 2014;60: Jalan R, Saliba F, Pavesi M, Amoros A, Moreau R, Gines P, et al. Development and validation of a prognostic score to predict mortality in patients with acute on chronic liver failure. J Hepatol 2014;61: Lee M, Lee JH, Oh S, Jang Y, Lee W, Lee HJ, et al. CLIF-SOFA scoring system accurately predicts short-term mortality in acutely decompensated patients with alcoholic cirrhosis: a retrospective analysis. Liver Int 2014;35: McPhail MJ, Shawcross DL, Abeles RD, Chang A, Patel V, Lee GH, et al. Increased Survival for patients with cirrhosis and organ failure in liver intensive care and validation of the Chronic Liver Failure- Sequential Organ Failure scoring system. Clin Gastroenterol Hepatol 2014 Sep 21. pii: S (14) doi: / j.cgh [Epub ahead of print]. 8. Pan HC, Jenq CC, Tsai MH, Fan PC, Chang CH, Chang MY, et al. Scoring systems for 6-month mortality in critically ill cirrhotic patients: a prospective analysis of Chronic Liver Failure-Sequential Organ Failure Assessment score (CLIF-SOFA). Aliment Pharmacol Ther 2014;40: Theocharidou E, Pieri G, Mohammad AO, Cheung M, Cholongitas E, Agarwal B, Burroughs AK. The Royal Free Hospital score: a calibrated prognostic model for patients with cirrhosis admitted to intensive care unit. Comparison with current models and CLIF-SOFA score. Am J Gastroenterol 2014;109: Agrawal S, Duseja A, Gupta T, Dhiman RK, Chawla Y. Simple organ failure count versus CANONIC grading system for predicting mortality in acute on chronic liver failure. J Gastroenterol Hepatol 2015;30: Moreau R, Arroyo V. Acute on chronic liver failure: a new clinical entity. Clin Gastroenterol Hepatol 2014 Feb 28. pii: S (14) doi: /j.cgh [Epub ahead of print]. 12. Bajaj JS, O Leary JG, Reddy KR, Wong F, Olson JC, Subramanian RM, et al. Second infections independently increase mortality in hospitalized patients with cirrhosis: the North American consortium for the study of end-stage liver disease (NACSELD) experience. HEPATOLOGY 2012;56: European Association for the Study of the Liver. EASL clinical practice guidelines on the management of ascites, spontaneous bacterial peritonitis, and hepatorenal syndrome in cirrhosis. J Hepatol 2010;53: Xia Q, Dai X, Zhang Y, Guo Y, Xu X, Yang Q, et al. A modified MELD model for Chinese Pre-ACLF and ACLF patients and it reveals poor prognosis in Pre-ACLF patients. PLoS One 2013;8:e Hamid SS, Atiq M, Shehzad F, Yasmeen A, Nissa T, Salam A, et al. Hepatitis E virus superinfection in patients with chronic liver disease. HEPATOLOGY 2002;36: Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973;60: Wiesner R, Edwards E, Freeman R, Harper A, Kim R, Kamath P, et al. Model for End-Stage Liver Disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: Yan H, Wu W, Yang Y, Wu Y, Yang Q, Shi Y. A novel integrated Model for End-Stage Liver Disease model predicts short-term prognosis of hepatitis B virus-related acute-on-chronic liver failure patients. Hepatol Res 2014 May 22. doi: /hepr [Epub ahead of print]. 19. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996;22: Jalan R, Gines P, Olson JC, Mookerjee RP, Moreau R, Garcia-Tsao G, et al. Acute-on chronic liver failure. J Hepatol 2012;57: Wasmuth HE, Kunz D, Yagmur E, Timmer-Stranghoner A, Vidacek D, Siewert E, et al. Patients with acute on chronic liver failure display sepsis-like immune paralysis. J Hepatol 2005;42: Wlodzimirow KA, Eslami S, Abu-Hanna A, Nieuwoudt M, Chamuleau RA. A systematic review on prognostic indicators of acute on chronic liver failure and their predictive value for mortality. Liver Int 2013;33: Jalan R, Yurdaydin C, Bajaj JS, Acharya SK, Arroyo V, Lin HC, et al. Toward an improved definition of acute-on-chronic liver failure. Gastroenterology 2014;147:4-10.

11 242 SHI ET AL. HEPATOLOGY, July Cordoba J, Ventura-Cots M, Simon-Talero M, Amoros A, Pavesi M, Vilstrup H, et al. Characteristics, risk factors, and mortality of cirrhotic patients hospitalized for hepatic encephalopathy with and without acute-on-chronic liver failure (ACLF). J Hepatol 2014;60: Romero-Gomez M, Montagnese S, Jalan R. Hepatic encephalopathy in patients with acute decompensation of cirrhosis and acute-on-chronic liver failure. J Hepatol 2015;62: Ferenci P, Lockwood A, Mullen K, Tarter R, Weissenborn K, Blei AT. Hepatic encephalopathy definition, nomenclature, diagnosis, and quantification: final report of the working party at the 11th World Congresses of Gastroenterology, Vienna, HEPATOLOGY 2002;35: Bernal W, Wendon J. Acute liver failure. N Engl J Med 2013;369: Bajaj JS. Defining acute-on-chronic liver failure: will East and West ever meet? Gastroenterology 2013;144: Wang FS, Zhang Z. Liver: how can acute-on-chronic liver failure be accurately identified? Nat Rev Gastroenterol Hepatol 2013;10: Author names in bold designate shared co-first authorship. Supporting Information Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/ /hep /suppinfo