Future directions in acute liver failure.

Acute liver failure (ALF) describes a clinical syndrome of rapid hepatocyte injury leading to liver failure manifested by coagulopathy and encephalopathy in the absence of pre-existing cirrhosis. The hallmark diagnostic features are a prolonged prothrombin time (ie, an international normalized ratio of prothrombin time of ≥1.5) and any degree of mental status alteration (HE). As a rare, orphan disease, it seemed an obvious target for a multicenter network. The Acute Liver Failure Study Group (ALFSG) began in 1997 to more thoroughly study and understand the causes, natural history, and management of ALF. Over the course of 22 years, 3364 adult patients were enrolled in the study registry (2614 ALF and 857 acute liver injury-international normalized ratio 2.0 but no encephalopathy-ALI) and >150,000 biosamples collected, including serum, plasma, urine, DNA, and liver tissue. Within the Registry study sites, 4 prospective substudies were conducted and published, 2 interventional ( N -acetylcysteine and ornithine phenylacetate), 1 prognostic [ 13 C-methacetin breath test (MBT)], and 1 mechanistic (rotational thromboelastometry). To review ALFSG's accomplishments and consider next steps, a 2-day in-person conference was held at UT Southwestern Medical Center, Dallas, TX, entitled "Acute Liver Failure: Science and Practice," in May 2022. To summarize the important findings in the field, this review highlights the current state of understanding of ALF and, more importantly, asks what further studies are needed to improve our understanding of the pathogenesis, natural history, and management of this unique and dramatic condition.

Autoimmune hepatitis presenting as acute liver failure: A 20-year retrospective review of North America.

Autoimmune hepatitis is a common cause of acute liver failure. Treatment includes steroids for acute liver injury and liver transplantation in those who fail to respond or develop acute liver failure. The aim of this study is to further characterize acute liver failure secondary to autoimmune hepatitis and identify variables that predict 21-day transplant-free survival. This study included adults hospitalized with acute liver failure enrolled in the Acute Liver Failure Study Group Registry between 1998 and 2019 from 32 centers within the US. The etiology of all cases was reviewed by the Adjudication Committee, and all cases identified as autoimmune hepatitis were included. Acute liver injury was defined as an INR ≥2.0 without encephalopathy and acute liver failure as INR ≥ 1.5 with encephalopathy. Laboratory and clinical data were reviewed. Variables significantly associated with 21-day transplant-free survival were used to develop a multivariable logistic regression model.  A total of 193 cases of acute liver failure secondary to autoimmune hepatitis were identified and reviewed. There were 161 patients (83.4%) diagnosed with acute liver failure on enrollment, and 32 (16.6%) developed acute liver failure during hospitalization. At 21 days, 115 (59.6%) underwent liver transplantation, 28 (14.5%) had transplant-free survival, and 46 (23.8%) died before liver transplantation. Higher admission values of bilirubin, INR, and coma grade were associated with worse outcomes. A prognostic index incorporating bilirubin, INR, coma grade, and platelet count had a concordance statistic of 0.84. Acute liver failure secondary to autoimmune hepatitis is associated with a high short-term mortality. We developed a model specifically for autoimmune hepatitis that may be helpful in predicting 21-day transplant-free survival and early identification of patients in need of expedited liver transplant evaluation.

The Midazolam RAMPART Study Medical Records Project: A Unique Use of Real-World Data in a Complex Collaborative Partnership to Support a New Drug Application.

INTRODUCTION: This project aimed to retrospectively obtain, review, and extract key safety data from medical records of participants enrolled in RAMPART, the NIH-supported Rapid Anticonvulsant Medication Prior to ARrival Trial of intramuscular midazolam versus intravenous lorazepam for pre-hospital treatment of status epilepticus, to support a US new drug application (NDA) for intramuscular midazolam. METHODS: A collaborative partnership was established between the NDA sponsor, the RAMPART trial lead academic institution, US government agencies, and contract research organizations to retrieve, review, and extract relevant safety data from the medical records of RAMPART participants and summarize those data to include in an NDA submitted to the US Food and Drug Administration (FDA). RESULTS: Key data in the medical records of 890 RAMPART trial participants (1020 enrollments, including 130 repeat enrollments) were reviewed and extracted into a project database. Safety events occurred in 771 (86.6%) participants, and included additional information not collected in the RAMPART trial. This database also enabled subgroup analyses based on medical history and prior/concurrent medications, building upon previous analyses according to age, sex, and race. No previously unrecognized safety patterns were identified, and no association was observed between efficacy and medical history or medication usage. CONCLUSIONS: The use of unstructured real-world retrospective medical record data can effectively support an NDA submission in place of conducting another interventional clinical trial. This retrospective medical records review and extraction of additional safety data contributed to the FDA approval of intramuscular midazolam for the pre-hospital treatment of status epilepticus in 2018. CLINICALTRIALS: GOV: NCT00809146.

Outcomes of patients with acute liver failure listed for liver transplantation: A multicenter prospective cohort analysis.

Liver transplantation (LT) is a life-saving treatment for patients with acute liver failure (ALF). Currently, there are few detailed data regarding long-term outcomes after LT for ALF. We combined prospective data from the Acute Liver Failure Study Group (ALFSG) Registry with those of the Scientific Registry of Transplant Recipients (SRTR) to assess outcomes among consecutive patients with ALF listed for LT. Cohort analysis of detailed pretransplantation data for patients listed for LT for ALF in the ALFSG Registry between January 1998 and October 2018 matched with transplantation-related data from the SRTR. Primary outcomes were 1- and 3-year post-LT patient survival. Secondary outcome was receipt of LT; independent associations with successful receipt of LT were determined using multivariable logistic regression. Of 624 patients with ALF listed for LT, 398 (64%) underwent LT, 100 (16%) died without LT, and 126 (20%) recovered spontaneously. Among LT recipients, etiologies included seronegative/indeterminate (22%), drug-induced liver injury (18%), acetaminophen overdose (APAP; 16%), and viral hepatitis (15%). The 1- and 3-year post-LT patient survival rates were 91% and 90%, respectively. Comparing those dying on the waiting list versus with those who received LT, the former had more severe multiorgan failure, reflected by increased vasopressor use (65% vs. 22%), mechanical ventilation (84% vs. 57%), and renal replacement therapy (57% vs. 30%; p < 0.0001 for all). After adjusting for relevant covariates, age (adjusted odds ratio [aOR] 1.02, 95% confidence interval [CI] 1.00-1.04), APAP etiology (aOR 2.72, 95% CI 1.42-5.23), requirement for vasopressors (aOR 4.19, 95% CI 2.44-7.20), Grade III/IV hepatic encephalopathy (aOR 2.47, 95% CI 1.29-4.72), and Model for End-Stage Liver Disease (MELD) scores (aOR 1.05, 95% CI 1.02-1.09; p < 0.05 for all) were independently associated with death without receipt of LT. Post-LT outcomes for ALF are excellent in this cohort of very ill patients. The development of multiorgan failure while on the transplantation list and APAP ALF etiology were associated with a lower likelihood of successful receipt of LT.

Comparison of Bayesian vs Frequentist Adaptive Trial Design in the Stroke Hyperglycemia Insulin Network Effort Trial.

Importance: Bayesian adaptive trial design has the potential to create more efficient clinical trials. However, a barrier to the uptake of bayesian adaptive designs for confirmatory trials is limited experience with how they may perform compared with a frequentist design. Objective: To compare the performance of a bayesian and a frequentist adaptive clinical trial design. Design, Setting, and Participants: This prospective cohort study compared 2 trial designs for a completed multicenter acute stroke trial conducted within a National Institutes of Health neurologic emergencies clinical trials network, with individual patient-level data, including the timing and order of enrollments and outcome ascertainment, from 1151 patients with acute stroke and hyperglycemia randomized to receive intensive or standard insulin therapy. The implemented frequentist design had group sequential boundaries for efficacy and futility interim analyses at 90 days after randomization for 500, 700, 900, and 1100 patients. The bayesian alternative used predictive probability of trial success to govern early termination for efficacy and futility with a first interim analysis at 500 randomized patients and subsequent interims after every 100 randomizations. Main Outcomes and Measures: The main outcome was the sample size at end of study, which was defined as the sample size at which each of the studies stopped accrual of patients. Results: Data were collected from 1151 patients. As conducted, the frequentist design passed the futility boundary after 936 participants were randomized. Using the same sequence and timing of randomization and outcome data, the bayesian alternative crossed the futility boundary approximately 3 months earlier after 800 participants were randomized. Conclusions and Relevance: Both trial designs stopped for futility before reaching the planned maximum sample size. In both cases, the clinical community and patients would benefit from learning the answer to the trial's primary question earlier. The common feature across the 2 designs was frequent interim analyses to stop early for efficacy or for futility. Differences between how these analyses were implemented between the 2 trials resulted in the differences in early stopping.

Intensive Versus Standard Treatment of Hyperglycemia in Acute Ischemic Stroke Patient: A Randomized Clinical Trial Subgroups Analysis.

BACKGROUND: Benefit from blood glucose (BG) control during acute ischemic stroke may depend on glycemic parameters. We evaluated for associations between the SHINE (Stroke Hyperglycemia Insulin Network Effort) randomized treatment group and the SHINE predefined 90-day functional outcome, within-patient subgroups defined by various glycemic parameters. METHODS: The SHINE Trial randomized 1151 patients within 12 hours with acute ischemic stroke and hyperglycemia to standard (target BG 80-179 mg/dL) or intensive (target BG 80-130 mg/dL) BG control for 72 hours. We predefined 6 glycemic parameters: acute BG level, absence versus presence of diagnosed and undiagnosed diabetes, hemoglobin A1c, glycemic gap (acute BG-average daily hemoglobin A1c based BG), stress hyperglycemia ratio (acute BG/average daily hemoglobin A1c based BG), and BG variability (SD). Favorable functional outcome was defined by the SHINE Trial and based on the modified Rankin Scale score at 90 days, adjusted for stroke severity. We computed relative risks adjusted for baseline stroke severity and thrombolysis use. RESULTS: Likelihood for favorable outcome was lowest among patients with undiagnosed diabetes compared to patients with true nondiabetes (adjusted relative risk, 0.42 [99% CI, 0.19-0.94]). We did not find any relationship between the favorable outcome rate and baseline BG or any of the glycemic parameters. No differences between SHINE treatment groups were identified among any of these patient subgroups. CONCLUSIONS: In this exploratory subgroup analysis, intensive versus standard insulin treatment of hyperglycemia in acute ischemic stroke patient subgroups, did not influence the 90-day functional outcomes, nor did we identify associations between these glycemic parameters and 90-day functional outcomes.

Generation of neutrophil extracellular traps in patients with acute liver failure is associated with poor outcome.

BACKGROUND AND AIMS: Acute liver failure (ALF) is characterized by significant changes in the hemostatic system and by systemic inflammation. The formation of neutrophil extracellular traps (NETs), in which an activated neutrophil expels its DNA, histones, and granular enzymes, such as myeloperoxidase (MPO), has been associated with immune-mediated and thrombotic diseases. We hypothesized that formation of NETs in patients with ALF contributes to progression of disease. APPROACH AND RESULTS: A total of 676 patients with ALF (international normalized ratio [INR], ≥1.5) or severe acute liver injury (ALI; INR, ≥2.0) were recruited from the U.S. ALF Study Group Registry between 2011 and 2018, of whom 308 patients (45.6%) had acetaminophen-induced ALF. Up to 21 days after admission, 483 patients (71.5%) survived without liver transplantation (LT). Levels of cell-free DNA (cfDNA) and the specific NET marker MPO-DNA complexes were measured in plasma samples obtained on admission and compared to levels in healthy controls. In addition, liver tissue obtained at transplantation of 20 ALF patients was stained for NETs. Levels of cfDNA were 7.1-fold, and MPO-DNA complexes 2.5-fold, higher in patients with ALF compared to healthy controls. cfDNA levels were not associated with 21-day transplant-free survival, but were higher in those patients with more-severe disease on admission, as reflected by various laboratory and clinical parameters. MPO-DNA levels were 30% higher in patients with ALF who died or required urgent LT. Liver tissue of ALF patients stained positive for NETs in 12 of 18 evaluable patients. CONCLUSIONS: Here, we provide evidence for NET formation in patients with ALF. Elevated plasma levels of MPO-DNA complexes in patients with ALF were associated with poor outcome, which suggests that NET formation contributes to disease progression.

Prognostic Value of the 13 C-Methacetin Breath Test in Adults with Acute Liver Failure and Non-acetaminophen Acute Liver Injury.

BACKGROUND AND AIMS: The 13 C-methacetin breath test (MBT) is a noninvasive, quantitative hepatic metabolic function test. The aim of this prospective, multicenter study was to determine the utility of initial and serial 13 C-MBT in predicting 21-day outcomes in adults with acute liver failure (ALF) and non-acetaminophen acute liver injury (ALI). APPROACH AND RESULTS: The 13 C-MBT BreathID device (Exalenz Biosciences, Ltd.) provided the percent dose recovery (PDR) for a duration of 60 minutes after administration of 13 C-methacetin solution as the change in exhaled 13 CO2 /12 CO2 compared with pre-ingestion ratio on study days 1, 2, 3, 5, and 7. Results were correlated with 21-day transplant-free survival and other prognostic indices. A total of 280 subjects were screened for enrollment between May 2016 and August 2019. Median age of the 62 enrolled patients with adequate data was 43 years, 79% were Caucasian, 76% had ALF with the remaining 24% having ALI. The mean PDR peak on day 1 or day 2 was significantly lower in nonsurvivors compared with transplant-free survivors (2.3%/hour vs. 9.1%/hour; P < 0.0001). In addition, serial PDR peaks were consistently lower in nonsurvivors versus survivors (P < 0.0001). The area under the receiver operating characteristic curve (AUROC) of the 13 C-MBT in the combined cohort was 0.88 (95% CI: 0.79-0.97) and higher than that provided by King's College (AUROC = 0.70) and Model for End-Stage Liver Disease scores (AUROC = 0.83). The 13 C-MBT was well tolerated with only two gastrointestinal adverse events reported. CONCLUSIONS: The 13 C-MBT is a promising tool to estimate the likelihood of hepatic recovery in patients with ALF and ALI. Use of the PDR peak data from the 13 C-MBT point-of-care test may assist with medical decision making and help avoid unnecessary transplantation in critically ill patients with ALF and ALI.

VWF/ADAMTS13 Imbalance, But Not Global Coagulation or Fibrinolysis, Is Associated With Outcome and Bleeding in Acute Liver Failure.

BACKGROUND AND AIMS: Recent studies of acute liver failure (ALF) in man and animals have suggested that rebalanced hemostasis occurs, with distinct hypercoagulable features clinically evidenced by a low risk of bleeding. Rodent models have shown a link between intrahepatic microthrombus formation and progression of ALF. We sought to confirm these earlier findings in a large series of patients with well-characterized ALF from the Acute Liver Failure Study Group. APPROACH AND RESULTS: Citrated plasma samples taken on admission from 676 patients with ALF or acute liver injury (international normalized ratio ≥2.0 without hepatic encephalopathy) were used to determine levels of von Willebrand factor (VWF), a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13 (ADAMTS13) activity, thrombomodulin-modified thrombin generation, and clot lysis time (CLT) and compared with the levels in 40 healthy controls. Patients had 3-fold increased VWF levels, 4-fold decreased ADAMTS13 activity, similar thrombin generating capacity, and 2.4-fold increased CLT, compared with controls. Increasing disease severity was associated with progressively more elevated VWF levels as well as hypofibrinolysis. Patients who died or underwent liver transplantation within 21 days of admission had higher VWF levels, lower ADAMTS13 activity, but similar thrombin generation and a similar proportion of patients with severe hypofibrinolysis, when compared with transplant-free survivors. Likewise, patients with bleeding complications had higher VWF levels and lower ADAMTS13 activity compared to those without bleeding. Thrombin generation and CLT did not differ significantly between bleeding and nonbleeding patients. CONCLUSIONS: Rebalanced hemostatic status was confirmed in a large cohort of patients with acute liver injury/ALF, demonstrating that VWF/ADAMTS13 imbalance is associated with poor outcome and bleeding. The association between VWF/ADAMTS13 imbalance and bleeding suggests that bleeding in ALF relates more to systemic inflammation than a primary coagulopathy.

An Unsuccessful Randomized Trial of Percutaneous vs Endoscopic Drainage of Suspected Malignant Hilar Obstruction.

Percutaneous transhepatic biliary drainage (PTBD) and endoscopic retrograde cholangiopancreatography (ERCP) are widely accepted but competing approaches for the management of malignant obstruction at the hilum of the liver. ERCP is favored in the United States on the basis of high success rates for non-hilar indications, the perceived safety and superior tissue sampling capability of ERCP relative to PTBD, and the avoidance of external drains that are undesirable to patients. A recent randomized controlled trial (RCT) comparing the 2 modalities in patients with resectable hilar cholangiocarcinoma was terminated prematurely because of higher mortality in the PTBD group.1 In contrast, most observational data suggest that PTBD is superior for achieving complete drainage.2-6 Because the preferred procedure remains uncertain, we aimed to compare PTBD and ERCP as the primary intervention in patients with cholestasis due to malignant hilar obstruction (MHO).

Role of Hepatitis E Virus Infection in North American Patients With Severe Acute Liver Injury.

INTRODUCTION: The aim of this study was to determine the role of hepatitis E virus (HEV) infection in a large cohort of prospectively enrolled patients with severe acute liver injury (ALI). METHODS: Serum samples from 594 consecutive adults enrolled between 2008 and 2018 in the US Acute Liver Failure Study Group ALI registry were tested for anti-HEV IgM and anti-HEV IgG levels. Those with detectable anti-HEV IgM underwent further testing for HEV RNA using real-time polymerase chain reaction. RESULTS: The median age of patients was 38 years; 41% were men and 72% Caucasian. Etiologies of ALI included acetaminophen hepatotoxicity (50%), autoimmune hepatitis (8.9%), hepatitis B virus (8.9%), and idiosyncratic drug-induced liver injury (7.9%). Overall, 62 patients (10.4%) were negative for anti-HEV IgM but positive for IgG, whereas only 3 men (0.5%) were positive for both anti-HEV IgM and IgG. These 3 cases were initially diagnosed as having indeterminate, HEV, and hepatitis B virus-related ALI. One of these patients had detectable HEV RNA genotype 3, and another anti-HEV IgM+ patient had detectable HEV antigens by immunohistochemistry on liver biopsy. On multivariate modeling, older (odds ratio: 1.99) and non-Caucasian subjects (odds ratio: 2.92) were significantly more likely to have detectable anti-HEV IgG (P < 0.0001). DISCUSSION: Acute HEV infection is an infrequent cause of ALI in hospitalized North American adults. The anti-HEV IgG+ patients were significantly older and more likely to be non-Caucasian. These data are consistent with other population-based studies that indicate exposure to HEV in the general US population is declining over time and might reflect a cohort effect.

Correction to: Rectal indomethacin alone versus indomethacin and prophylactic pancreatic stent placement for preventing pancreatitis after ERCP: study protocol for a randomized controlled trial.

An amendment to this paper has been published and can be accessed via the original article.

BiMM tree: A decision tree method for modeling clustered and longitudinal binary outcomes.

Clustered binary outcomes are frequently encountered in clinical research (e.g. longitudinal studies). Generalized linear mixed models (GLMMs) for clustered endpoints have challenges for some scenarios (e.g. data with multi-way interactions and nonlinear predictors unknown a priori). We develop an alternative, data-driven method called Binary Mixed Model (BiMM) tree, which combines decision tree and GLMM within a unified framework. Simulation studies show that BiMM tree achieves slightly higher or similar accuracy compared to standard methods. The method is applied to a real dataset from the Acute Liver Failure Study Group.

Sleep for Stroke Management and Recovery Trial (Sleep SMART): Rationale and methods.

RATIONALE: Obstructive sleep apnea is common among patients with acute ischemic stroke and is associated with reduced functional recovery and an increased risk for recurrent vascular events. AIMS AND/OR HYPOTHESIS: The Sleep for Stroke Management and Recovery Trial (Sleep SMART) aims to determine whether automatically adjusting continuous positive airway pressure (aCPAP) treatment for obstructive sleep apnea improves clinical outcomes after acute ischemic stroke or high-risk transient ischemic attack. SAMPLE SIZE ESTIMATE: A total of 3062 randomized subjects for the prevention of recurrent serious vascular events, and among these, 1362 stroke survivors for the recovery outcome. METHODS AND DESIGN: Sleep SMART is a phase III, multicenter, prospective randomized, open, blinded outcome event assessed controlled trial. Adults with recent acute ischemic stroke/transient ischemic attack and no contraindication to aCPAP are screened for obstructive sleep apnea with a portable sleep apnea test. Subjects with confirmed obstructive sleep apnea but without predominant central sleep apnea proceed to a run-in night of aCPAP. Subjects with use (≥4 h) of aCPAP and without development of significant central apneas are randomized to aCPAP plus usual care or care-as-usual for six months. Telemedicine is used to monitor and facilitate aCPAP adherence remotely. STUDY OUTCOMES: Two separate primary outcomes: (1) the composite of recurrent acute ischemic stroke, acute coronary syndrome, and all-cause mortality (prevention) and (2) the modified Rankin scale scores (recovery) at six- and three-month post-randomization, respectively. DISCUSSION: Sleep SMART represents the first large trial to test whether aCPAP for obstructive sleep apnea after stroke/transient ischemic attack reduces recurrent vascular events or death, and improves functional recovery.

Alcohol consumption is associated with the severity and outcome of acute liver injury/failure.

BACKGROUND & AIMS: Non-medical factors which contribute to the severity of acute liver failure (ALF) remain poorly defined. The association of alcohol consumption on the severity of presentation and outcome were determined in patients with ALF and acute liver injury (ALI) in a large, multicentre registry. METHODS: Alcohol consumption during the 6 months prior to study entry was analysed in 1170 patients enrolled in the ALF Study Group Registry. Consumption was categorized as none/minimal (<3 alcoholic beverages/week) or at least moderate (≥3/week). Clinical characteristics, the severity of liver injury at presentation (ALI or ALF) and outcome were compared. RESULTS: In patients with acetaminophen (APAP) overdose, at least moderate alcohol consumption was associated with higher peak aminotransferases, bilirubin, creatinine and INR on admission, compared to no/minimal consumption. In patients with non-APAP ALI/ALF, at least moderate alcohol consumption was associated with higher peak aminotransferases and creatinine. In APAP, non-APAP or all aetiologies, at least moderate alcohol consumption was associated with a 75%, 89% and 82% higher odds, respectively, of presenting as ALF rather than ALI (all P < .005). At least moderate alcohol consumption increased the odds of death by 45% (P = .01) across all aetiologies. In multivariate analysis, older age, non-Caucasian race, peak INR, peak bilirubin and at least moderate alcohol consumption were significantly associated with death. Finally, in Kaplan-Meier analysis of patients with all aetiologies, at least moderate alcohol consumption was associated with decreased time-dependent survival (P = .002). CONCLUSION: Alcohol consumption adversely affects the presentation and outcome of both APAP- and non-APAP-induced ALI/ALF.

BiMM forest: A random forest method for modeling clustered and longitudinal binary outcomes.

Clustered binary outcomes and datasets with many predictor variables are frequently encountered in clinical research (e.g. longitudinal studies). Generalized linear mixed models (GLMMs) typically employed for clustered endpoints have challenges for some scenarios, particularly for complex datasets which contain many interactions among predictors and nonlinear predictors of outcome. We propose a new method called Binary Mixed Model (BiMM) forest, which combines random forest and GLMM methodology. BiMM forest offers a flexible and stable method which naturally models interactions among predictors and can be employed in the setting of clustered data. Simulation studies show that BiMM forest achieves similar or superior prediction accuracy compared to standard random forest, GLMMs and its tree counterpart (BiMM tree) for clustered binary outcomes. The method is applied to a real dataset from the Acute Liver Failure Study Group. BiMM forest offers an alternative method for modeling clustered binary outcomes which may be applied in myriad research settings.

Predicting daily outcomes in acetaminophen-induced acute liver failure patients with machine learning techniques.

BACKGROUND/OBJECTIVE: Assessing prognosis for acetaminophen-induced acute liver failure (APAP-ALF) patients during the first week of hospitalization often presents significant challenges. Current models such as the King's College Criteria (KCC) and the Acute Liver Failure Study Group (ALFSG) Prognostic Index are developed to predict outcome using only a single time point on hospital admission. Models using longitudinal data are not currently available for APAP-ALF patients. We aim to develop and compare performance of prediction models for outcomes during the first week of hospitalization for APAP-ALF patients. METHODS: Models are developed for the ALFSG registry data to predict longitudinal outcomes for 1042 APAP-ALF patients enrolled 01/1998-02/2016. The primary outcome is defined as daily low versus high coma grade. Accuracy in prediction of outcome (AC), sensitivity (SN), specificity (SP) and area under the receiver operating curve (AUC) are compared between the following models: classification and regression tree, random forest, frequentist generalized linear mixed model (GLMM), Bayesian GLMM, BiMM tree, and BiMM forest using original and imputed datasets. RESULTS: BiMM tree offers predictive (test set) 63% AC, 72% SP and 53% SN for the original dataset, whereas BiMM forest offers predictive (test set) 69% AC, 63% SP and 74% SN for the imputed dataset. BiMM tree has the highest AUC for the original testing dataset (0.697), whereas BiMM forest and standard random forest have the highest AUC for the imputed testing dataset (0.749). The three most important predictors of daily outcome for the BiMM tree are pressor use, bilirubin and creatinine. The BiMM forest model identifies lactate, ammonia and ALT as the three most important predictors of outcome. CONCLUSIONS: BiMM tree offers a prognostic tool for APAP-ALF patients, which has good accuracy and simple interpretation of predictors which are consistent with clinical observations. BiMM tree and forest models are developed using the first week of in-patient data and are appropriate for predicting outcome over time. While the BiMM forest has slightly higher predictive AC, the BiMM tree model is simpler to use at the bedside.