Hepatic Encephalopathy Is Associated With Significantly Increased Mortality Among Patients Awaiting Liver Transplantation

Transcription

1 LIVER TRANSPLANTATION 20: , 2014 ORIGINAL ARTICLE Hepatic Encephalopathy Is Associated With Significantly Increased Mortality Among Patients Awaiting Liver Transplantation Robert J. Wong, 1,2 Robert G. Gish, 3,4 and Aijaz Ahmed 1 1 Division of Gastroenterology and Hepatology, Stanford University School of Medicine, Stanford, CA; 2 Division of Gastroenterology and Hepatology, Veterans Affairs Palo Alto Health Care System, Palo Alto, CA; 3 Robert G. Gish Consultants LLC, La Jolla, CA; and 4 Hepatitis B Foundation, Doylestown, PA The prioritization of liver transplantation (LT) for patients with end-stage liver disease uses the Model for End-Stage Liver Disease (MELD), which attempts to identify the sickest patients and thereby those who are in greatest need for LT. Hepatic encephalopathy (HE) is not included in MELD, and severity of liver disease and risk of wait-list removal or wait-list death may be underestimated by MELD in patients with HE. Using United Network for Organ Sharing registry data, we evaluated the impact of HE on 90-day wait-list survival among adult LT wait-list registrants in the United States from 2003 to Survival was stratified by HE severity (none, grade 1-2, grade 3-4) and MELD. There were 84,947 new LT wait-list registrants during the study period; 36.8% had no HE, 57.4% had grade 1-2 HE, and 5.9% had grade 3-4 HE. Ninety-day waitlist mortality was significantly higher among patients with grade 3-4 HE compared with patients with grade 1-2 HE or no HE (24.4% versus 6.8% versus 3.5%; P < 0.001). When stratified by MELD, patients with grade 3-4 HE had 90-day wait-list mortality similar to that of nonencephalopathic patients with MELD scores 6-7 points higher. With the multivariate Cox proportional hazards model, patients with grade 3-4 HE had 66% greater risk of 90-day mortality than patients without HE (hazard ratio , 95% CI ; P < 0.001). The inclusion of HE severity in MELD improved the area under receiver operating curve for predicting 90-day wait-list survival from to In conclusion, grade 3-4 HE at time of wait-list registration significantly increases 90-day wait-list mortality independent of MELD score. Incorporating HE in the assessment of LT priority may improve prognostication of liver disease severity and prioritization for LT. Liver Transpl 20: , VC 2014 AASLD. Received April 7, 2014; accepted August 10, Chronic liver disease is a leading cause of morbidity and mortality in the United States. 1 Progressive hepatic fibrosis among patients with chronic liver disease leads to cirrhosis and its complications, including hepatocellular carcinoma (HCC) and end-stage liver disease. 2-4 Liver transplantation (LT) is a curative option with 5-year post-lt survival greater than 80%. 5 However, the growing number of patients awaiting LT has far outpaced the availability of donor livers to be allocated for LT in the United States 6,7 The implementation of the Model for End-Stage Liver Disease (MELD) score in 2002 was an attempt to institute an objective system by which to prioritize patients for LT. 8,9 The MELD score incorporates objective measures of serum bilirubin, creatinine, and international normalized ratio (INR) to prioritize Additional Supporting Information may be found in the online version of this article. Abbreviations: AUROC, area under the receiver operating curve; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HE, hepatic encephalopathy; HR, hazard ratio; INR, international normalized ratio; ln, natural logarithm; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; TIPS, transjugular intrahepatic portosystemic shunt; UNOS/OPTN, United Network for Organ Sharing/Organ Procurement and Transplantation Network. Potential conflict of interest: Nothing to report. Address reprint requests to Robert J. Wong, M.D., M.S., Division of Gastroenterology and Hepatology, Stanford University School of Medicine, 300 Pasteur Drive, Always Building, Suite M-211, Stanford, CA Telephone: ; FAX: ; rwong123@stanford.edu DOI /lt View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI /lt. Published on behalf of the American Association for the Study of Liver Diseases VC 2014 American Association for the Study of Liver Diseases.

2 LIVER TRANSPLANTATION, Vol. 20, No. 12, 2014 WONG, GISH, AND AHMED 1455 patients awaiting LT. In addition to the MELD score, hyponatremia has also been shown to be an important predictor of mortality among patients with cirrhosis With the concept of providing LT to patients who are the sickest first and thereby benefiting the sickest the most, LT is offered to patients with the highest MELD scores first. However, hepatic encephalopathy (HE), a marker of hepatic decompensation, is not included in the MELD scoring system, and several studies have raised concern that the MELD score underestimates the risk of mortality among patients with HE In addition, HE may not correlate well with MELD across the range of MELD scores, and patients with HE may not receive needed LT in a timely manner under the current MELD scoring system. 14,15 In an effort to assess the impact of HE on LT waitlist mortality, we performed a retrospective cohort study using 10 years of MELD-era data from a population-based registry of all adult LT wait-list registrants in the United States. We hypothesize that HE will be associated with significantly increased 90- day wait-list mortality independent of MELD scores. PATIENTS AND METHODS Study Population Adult men and women (age >18 years) who were registered on the wait-list for LT in the United States from January 1, 2003, to December 31, 2012, were evaluated with data from the United Network for Organ Sharing and Organ Procurement and Transplantation Network (UNOS/OPTN) registry. Our target population consisted of patients with chronic liver disease awaiting LT. Patients who were listed for LT secondary to acute liver failure were excluded. The prioritization of LT among patients with concurrent HCC in the United States incorporates an exception policy, such that patients with HCC within defined criteria are allocated additional points to their MELD score by regional transplant review boards, thereby increasing their priority and probability of receiving LT. The basis for this policy in part reflects the understanding that MELD score alone does not accurately predict 90-day mortality among patients with concurrent HCC. For this reason, patients with HCC were also excluded from our analyses. MELD scores at the time of wait-list registration and at the time of transplantation if LT was performed were calculated for each individual. MELD scores were calculated with standard formulae that incorporate the natural logarithms (ln) of INR, bilirubin, and creatinine: ln(inr) ln (creatinine, in milligrams per deciliter) ln (bilirubin, in milligrams per deciliter) , with a lower limit of 1 for all variables and an upper limit of 4 for creatinine. Patients on hemodialysis are given a creatinine score of 4. The grading of HE severity was based on West Haven Criteria (grade 1 5 trivial lack of awareness, euphoria or anxiety, shortened attention span, impaired performance of addition or subtraction; grade 2 5 lethargy or apathy, minimal disorientation for time or place, subtle personality change, inappropriate behavior; grade 3 5 somnolence to semistupor but responsive to verbal stimuli, confusion, gross disorientation; grade 4 5 coma (unresponsive to verbal or noxious stimuli). 16 Severity of HE (no HE, grade 1-2 HE, grade 3-4 HE) at time of wait-list registration and at time of LT among transplant recipients were documented in the UNOS registry and relied on review of clinical medical records to determine the most recently documented grade of HE prior to wait-list registration. Treatment for HE was not captured in the UNOS registry data. The outcome for patients on the LT wait list (ie, death, receipt of liver transplant, or removal from wait list for other reasons) allowed the calculation of our primary outcome, 90-day mortality among wait-listed patients. This outcome is commonly used to evaluate LT wait-list mortality. Statistical Analysis Descriptive statistics were stratified by severity of HE and presented as proportion (%) and frequency (N) for categorical variables, mean and standard deviation (SD) for normally distributed continuous variables, and median and range for nonnormally distributed variables. Comparisons between groups were performed with chi-square testing and analysis of variance methods. Kaplan-Meier methods were used to evaluate our primary outcome of 90-day wait-list mortality. Stratification by MELD score and degree of HE was applied in an attempt to determine whether the increased mortality associated with HE was dependent on MELD score at time of wait-list registration. In other words, for each MELD score (6-40), we calculated 90-day wait-list mortality for patients with no HE, grade 1-2 HE, and grade 3-4 HE. To determine whether there was a statistically significant nonlinear relationship between severity of HE and 90-day waitlist mortality, we used a second-order polynomial (quadratic) model to compare the graphic function represented by each category of HE. The resulting smooth curves depicted the relationship between severity of HE and 90-day wait-list mortality for individual MELD scores. Multivariate Cox proportional hazards models were used to evaluate the association between HE and 90-day wait-list mortality. Forward stepwise regression methods included variables that satisfied biological priori (eg, age, sex) and those that demonstrated significant associations in the univariate models (P < 0.10). The final multivariate model was adjusted for sex, age, race/ethnicity, etiology of liver disease [hepatitis C virus (HCV) versus non- HCV], MELD score, ascites, serum albumin, serum sodium, year of wait-list registration, and HE. Statistical significance was met with a two-tailed P value <0.05. We also performed additional regression analyses to compare the area under the receiver operating curve (AUROC) for MELD alone versus the inclusion of HE into the MELD score at predicting 90-day wait-

3 1456 WONG, GISH, AND AHMED LIVER TRANSPLANTATION, December 2014 TABLE 1. Demographic and Clinical Characteristics of Patients on the LT Wait List No HE Grade 1-2 HE Grade 3-4 HE Percentage N Percentage N Percentage N P Value Male sex , , ,142 <0.001 Race/ethnicity <0.001 Non-Hispanic white , , Black Hispanic , Asian Age (years; mean 6 SD) <0.001 MELD (mean 6 SD) <0.001 HCV , , <0.001 Ascites , , <0.001 Albumin (g/dl; mean 6 SD) <0.001 Sodium (meq/l; mean 6 SD) TABLE 2. Distribution of HE Severity at Time of Wait-List Registration and at Time of LT Wait-List Registration Liver Transplantation HE severity Percentage N Percentage N No HE , ,694 Grade , ,160 Grade list survival. All statistical analyses were performed in the Stata statistical package (version 10; Stata Corporation, College Station, TX) and Prism statistical software (version 6.0; GraphPad Software, Inc., La Jolla, CA). RESULTS Characteristics of the Study Cohort From January 1, 2003, to December 31, 2012, there were 84,947 new LT wait-list registrants, among whom 36.8% (n 5 31,239) had no HE, 57.4% (n 5 48,722) had grade 1-2 HE, and 5.9% (n ) had grade 3-4 HE. Although the majority of wait-list registrants were men, there were significantly fewer men in the no-he group compared with grade 1-2 HE or grade 3-4 HE cohorts (Table 1). Non-Hispanic whites and Hispanics were more likely to have grade 1-2 HE at time of wait-list registration, whereas blacks and Asians were more likely to have no HE or grade 3-4 HE. The mean age was significantly higher for patients with HE compared with those with no HE (grade 3-4 HE years versus grade 1-2 HE years versus no HE years; P < 0.001). In addition, patients with more severe HE at time of wait-list registration had higher mean MELD scores (grade 3-4 HE versus grade 1-2 HE versus no HE ; P < 0.001). Patients with more severe HE also had significantly higher rates of ascites and lower serum albumin (Table 1). Although the severity of HE among LT wait-list patients was provided at time of wait-list registration, additional time points documenting progression or improvement in HE were not available in the registry. However, among patients who underwent LT, HE severity at time of LT was documented. We performed a subanalysis to determine the change in HE severity from time of wait-list registration to time of LT (Tables 2 and 3). Among patients with no HE at registration, 32.1% (n ) had increased HE severity at time of LT, whereas the remainder continued to have no HE (Table 3). Among patients with grade 1-2 HE at registration, 8.9% (n ) had increased HE severity, 11.2% (n5 2707) had decreased HE severity, and 79.9% (n 5 19,314) had unchanged HE severity at time of LT. Among patients with grade 3-4 HE at registration, 35.9% (n5 1041) had decreased HE severity, and the remainder had unchanged HE severity at time of LT (Table 3). Liver Transplantation Wait-List Mortality Overall 90-day mortality among LT wait-list patients was 6.3%. However, patients with grade 3-4 HE at time of wait-list registration had significantly higher 90-day wait-list mortality compared with patients with grade 1-2 HE or patients with no HE (24.4% versus 6.8% versus 3.5%; P < 0.001; Fig. 1). This impact of HE on 90-day wait-list mortality was similar between men and women. Similar trends were seen when evaluating 30-day and 60-day wait-list mortality

4 LIVER TRANSPLANTATION, Vol. 20, No. 12, 2014 WONG, GISH, AND AHMED 1457 TABLE 3. Change in Severity of HE From Time of Wait-List Registration to Time of Transplantation Among LT Recipients Wait-List Registration Increased Decreased Unchanged HE severity Percentage N Percentage N Percentage N Percentage N No HE , Grade , ,314 Grade ,857 Figure 1. Overall 30-day, 60-day, and 90-day survival among patients awaiting LT. (Fig. 1). On multivariate Cox proportional hazards modeling of 90-day wait-list mortality, patients with grade 3-4 HE had significantly greater risk of death compared with patients with no HE at time of wait-list registration (hazard ratio [HR] , 95% CI ; P < 0.001; Fig. 2). Compared with women, men had lower risk of death (HR , 95% CI ; P ). Compared with non-hispanic whites, Hispanics had a significantly lower risk of wait-list mortality (HR , 95% CI ; P ), but no significant difference in mortality was seen among blacks or Asians (Fig. 2). HCV diagnosis, compared with non-hcv, was not associated with a significant difference in 90-day wait-list mortality (HR , 95% CI ; P ). Although our multivariate regression analysis demonstrated a 66% increased risk of 90-day wait-list mortality among patients with grade 3-4 HE compared with patients with no HE at time of wait-list registration, we hypothesized that the degree of increased risk was not uniform across the MELD score spectrum. To test this hypothesis, 90-day wait-list survival was stratified by both degree of HE (no HE versus grade 1-2 HE versus grade 3-4 HE) and MELD score at time of wait-list registration (Fig. 3). A line of best fit was determined by using polynomial modeling, and a second-order polynomial (quadratic) demonstrated the most accurate statistical fit. The resulting smooth curve depicts the relationship between 90-day wait-list survival and MELD score at time of wait-list registration stratified by severity of HE (Fig. 3). Overall, 90-day survival was significantly different among the three curves, with significantly lower 90-day wait-list survival among patients with grade 3-4 HE. However, the survival difference was not uniform across MELD scores, and a greater survival difference was seen with higher MELD scores (Fig. 3). For example, among patients with MELD score of 20 at time of wait-list registration, those with grade 3-4 HE had a 90-day wait-list survival of 84.8% compared with 93.3% among patients with no HE. Among patients with MELD score of 30 at time of wait-list registration, 90-day wait-list survival was 61.1% among those with grade 3-4 HE compared with 78.1% among those with no HE (Fig. 3). Using regression methods, we calculated the AUROC for predicting 90-day wait-list survival

5 1458 WONG, GISH, AND AHMED LIVER TRANSPLANTATION, December 2014 Figure 2. Multivariate Cox proportional hazards model of 90-day probability of death among patients on the LT wait list. associated with MELD score alone and MELD score with inclusion of HE severity (Fig. 4). By MELD score alone, the AUROC for predicting 90-day wait-list survival was , and when HE severity was included the AUROC improved to (Fig. 4). To evaluate further the impact of including HE severity with MELD score on improved reclassification of 90-day wait-list survival, additional analyses were performed to determine the overall 90-day wait-list survival with and without HE stratification. This was further substratified by MELD score to determine the more specific impact of HE inclusion at each MELD score (Supporting Information Table A). DISCUSSION Chronic liver disease is one of the leading causes of morbidity and mortality in the United States. 1 Although LT offers a curative option for patients with end-stage liver disease, the numbers of patients awaiting LT far outweigh the availability of organs to be allocated for LT. 5-7 The MELD score has been used to prioritize patients on the LT wait list. However, HE is currently not factored into the prioritization for LT. 8,9 Our current study evaluated the impact of HE at the time of waitlist registration on 90-day wait-list mortality. Patients with grade 3-4 HE had 66% higher risk of 90-day waitlist mortality compared with patients without HE.

6 LIVER TRANSPLANTATION, Vol. 20, No. 12, 2014 WONG, GISH, AND AHMED 1459 Figure 3. of HE. Overall 90-day wait-list survival stratified by degree Since the introduction of the MELD score in 2003 as an objective measure for prioritizing patients with end-stage liver disease for LT, several studies have Figure 4. Receiver operating curve analysis of MELD alone versus MELD with HE severity at predicting 90-day LT wait-list survival. attempted to improve upon its performance in predicting wait-list mortality. In the original multivariate model used to establish the MELD score for the prioritization of LT, the addition of HE was not shown to impact the c-statistic of the MELD model significantly. 9 However, other studies prior to the implementation of the MELD score did demonstrate a significantly increased mortality associated with HE. 19,20 Subsequent authors have reported similar findings, raising concern that the lack of considering HE in prioritization of patients for LT may significantly underestimate liver disease severity and mortality risk among patients with HE An early study by Yoo et al. 14 evaluated 66 patients undergoing LT evaluation to determine the correlation of MELD score with severity of HE. HE was assessed based on electroencephalogram and a series of 6 neuropsychiatric tests, including Trail Making Tests A and B, Rey- Osterreith Complex Figure Test including Reycopy and Reyrecall, Hopkins Verbal Learning Test, and Mini-Mental State Examination. The study indicated no significant difference in MELD scores across severity of HE and demonstrated poor correlation of MELD score with severity of HE, raising concern that patients with severe HE are disadvantaged in the current MELD model and may not receive LT in a timely manner. 14 Said et al. 18 evaluated 1611 consecutive patients with chronic liver disease seen at a single university hospital health care system. Their analyses demonstrated that HE was a strong and significant predictor of mortality among patients with chronic liver disease and that the addition of HE to the MELD score would improve its prognostic value. However, the inherent subjectivity of grading HE severity, compounded by effect of patient-specific factors (eg, medication noncompliance), continued to raise concerns about the incorporation of HE into current LT prioritization discussions. Stewart et al. 15 performed a retrospective evaluation of 494 patients with cirrhosis that included 223 patients undergoing transjugular intrahepatic portosystemic shunt (TIPS) placement and 271 patients hospitalized for decompensated cirrhosis. On univariate analyses, grade >3 HE was associated with significantly increased risk of mortality (HR ; 95% CI ; P < 0.001) among patients undergoing TIPS, and grade >2 HE was associated with significantly increased mortality among hospitalized patients. Furthermore, when HE and MELD were considered together, HE remained strongly predictive of mortality among hospitalized patients, demonstrating that HE may provide additional prognostic information independent of MELD score. 15 Another important factor to consider is the evolution of and improvement in diagnosis and treatment of HE over time; better recognition and treatment of HE in the latter part of the study may lead to less significant impact of grade 3-4 HE on 90-day wait-list survival. We performed a subanalysis separating our cohort into two periods, and Although patients with grade 3-4 HE continued to demonstrate significantly lower 90-day

7 1460 WONG, GISH, AND AHMED LIVER TRANSPLANTATION, December 2014 wait-list survival compared with patients with no HE in both the earlier and the later periods, patients with grade 3-4 HE in the later period ( ) did in fact have improved survival compared with patients with grade 3-4 HE in the earlier period ( ; 80.7% versus 69.8%; P<0.01). Our current study uses 10 years of UNOS registry data that includes all adult LT registrants in the United States We demonstrate that patients with grade 3-4 HE at time of waitlist registration have 66% increased risk of 90-day wait-list mortality independent of MELD score. Furthermore, MELD score does not accurately predict 90- day wait-list mortality among patients with grade 3-4 HE, such that these patients have 90-day mortality rates significantly higher than nonencephalopathic patients with similar MELD scores (Fig. 3). The incorporation of HE into the discussion regarding LT priority not only will improve prognostication but will help in prioritizing the sickest patients for LT. The impact of HE on survival among patients with cirrhosis is well established. 2-4,17,21 Itisamarkerof decompensated disease and is a component of the Child-Pugh-Turcotte scoring system. 22 However, concerns have been raised over the potential subjectivity that affects the grading of HE severity, and this subjectivity might have influenced its exclusion from the original MELD scoring model. Although there are several scoring systems for grading severity of HE, the West Haven Criteria are commonly used Ham et al. 16 recently proposed the standardization of the West Haven Criteria for the assessment of HE among patients with chronic liver disease. Other objective tools for assessing for HE include the psychometric hepatic encephalopathy score and the critical flicker frequency Although the West Haven Criteria are the most commonly used tool for assessing HE severity, they too are susceptible to some degree of subjectivity and intraobserver variability. Although no studies have clearly made a detailed assessment of intraobserver variability when using the West Haven Criteria, it is conceivable that variability exists in accurately identifying different grades of HE severity, especially distinguishing grade 1 from grade 2 HE. Thus, although the importance of implementing a single scoring system will help improve consistency of HE assessment across patient populations and practice environments, it is even more important to develop a more objective tool that will improve accuracy and reproducibility of HE severity assessment and thereby lead to a tool that harbors greater clinical significance and greater prognostic ability. Prospective studies are needed to pilot the incorporation of such a tool into the MELD score to assess liver transplant wait-list mortality more accurately. The strengths of the current study include the inclusion of 10 years of population-based data inclusive of all adult liver transplant registrants in the United States. The large cohort from a single registry source allows for consistency of data reporting, and improves the generalizability of the study findings. However, the utilization of registry data has several limitations. The available data are subject to coding and data entry errors that are inherent in databasebased studies. Specifically, the coding of HE severity may not follow uniform reporting measures and is not necessarily subject to cross-checking for accuracy. Furthermore, it is not clear whether all reporting centers used similar systems for grading HE severity. The grading of HE severity has been criticized for its potential for subjective bias, and this might have influenced its exclusion from the original MELD score and the gravitation away from Child- Pugh score to MELD for LT allocation. This limitation is even greater in a retrospective assessment. Furthermore, data regarding treatment for HE and how well patients responded to therapy were not available for inclusion in the study. In addition, disparities in access to HE treatment might also have impacted the severity of HE at time of listing as well as progression of HE while on the LT wait list, but surrogate markers of disparities in medication access were not available in the registry. It is also well known that severity of HE is a dynamic process, and, with progression of disease or initiation of treatment, severity of HE may change over time. Although the current database does not provide severity of HE at several time points, our findings are still clinically applicable, such that reassessment of HE along with MELD score among wait-list patients can be used to provide an updated estimate of 90-day wait-list mortality. This concept of reassessing a patient s clinical situation to provide an updated assessment of prognosis is similar to the current model used with MELD scores. However, it is likely not the HE itself that is directly causative of increased mortality; rather, it is a marker for more severe liver disease and risk of death that are not completely captured by MELD alone. The UNOS region where the patient was listed for LT may be an additional potentially confounding variable. Disparate average MELD scores at time of LT lead to disparate length of time waiting for LT in different UNOS regions, and this may impact the effect of HE severity of 90-day wait-list survival. We performed a subanalysis to reassess the impact of HE severity of 90-day wait-list survival stratified by UNOS region. Although some variation was seen in the magnitude of impact of HE severity on 90-day survival in different UNOS regions, the significantly lower 90-day wait-list survival among patients with grade 3-4 HE compared with patients with no clinical evidence of HE was consistent across all regions. Nevertheless, our large cohort study demonstrates a significant association between HE and 90-day wait-list mortality. In conclusion, the results of the current study demonstrate that greater severity of HE at time of LT waitlist registration significantly increases 90-day wait-list mortality independent of MELD score. Incorporating HE in the decision regarding LT wait-list priority will improve prognostication and thereby improve the prioritization of patients awaiting LT.

8 LIVER TRANSPLANTATION, Vol. 20, No. 12, 2014 WONG, GISH, AND AHMED 1461 REFERENCES 1. Hoyert DL, Xu J. Deaths: Preliminary data for Natl Vital Stat Rep 2012;61(6): D Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: a systematic review of 118 studies. J Hepatol 2006;44: Gines P, Quintero E, Arroyo V, Teres J, Bruguera M, Rimola A, et al. Compensated cirrhosis: natural history and prognostic factors. Hepatology 1987;7: Merkel C, Bolognesi M, Angeli P, Noventa F, Caregaro L, Sacerdoti D, Gatta A. Prognostic indicators of survival in patients with cirrhosis and esophageal varices, without previous bleeding. Am J Gastroenterol 1989;84: Taniguchi M. Liver transplantation in the MELD era analysis of the OPTN/UNOS registry. Clin Transpl 2012; 2012: Freeman RB, Edwards EB, Harper AM. Waiting list removal rates among patients with chronic and malignant liver diseases. Am J Transplant 2006;6: Goldberg D, French B, Trotter J, Shetty K, Schiano T, Reddy KR, Halpern SD. Underreporting of liver transplant waitlist removals due to death or clinical deterioration: results at four major centers. Transplantation 2013;96: Wiesner R, Edwards E, Freeman R, Harper A, Kim R, Kamath P, et al.; for United Network for Organ Sharing Liver Disease Severity Score Committee. Model for End- Stage Liver Disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001;33: Kim WR, Biggins SW, Kremers WK, Wiesner RH, Kamath PS, Benson JT, et al. Hyponatremia and mortality among patients on the liver-transplant waiting list. N Engl J Med 2008;359: Heuman DM, Abou-Assi SG, Habib A, Williams LM, Stravitz RT, Sanyal AJ, et al. Persistent ascites and low serum sodium identify patients with cirrhosis and low MELD scores who are at high risk for early death. Hepatology 2004;40: Biggins SW, Rodriguez HJ, Bacchetti P, Bass NM, Roberts JP, Terrault NA. Serum sodium predicts mortality in patients listed for liver transplantation. Hepatology 2005;41: Biggins SW, Kim WR, Terrault NA, Saab S, Balan V, Schiano T, et al. Evidence-based incorporation of serum sodium concentration into MELD. Gastroenterology 2006;130: Yoo HY, Edwin D, Thuluvath PJ. Relationship of the Model for End-Stage Liver Disease (MELD) scale to hepatic encephalopathy, as defined by electroencephalography and neuropsychometric testing, and ascites. Am J Gastroenterol 2003;98: Stewart CA, Malinchoc M, Kim WR, Kamath PS. Hepatic encephalopathy as a predictor of survival in patients with end-stage liver disease. Liver Transpl 2007;13: Ham J, Gish RG, Mullen K. Model for end-stage liver disease (MELD) exception for hepatic encephalopathy. Liver Transpl 2006;12(suppl 3):S102-S Milani A, Marra L, Siciliano M, Rossi L. Prognostic significance of clinical and laboratory parameters in liver cirrhosis. A multivariate statistical approach. Hepatogastroenterology 1985;32: Said A, Williams J, Holden J, Remington P, Gangnon R, Musat A, Lucey MR. Model for end-stage liver disease score predicts mortality across a broad spectrum of liver disease. J Hepatol 2004;40: Cooper GS, Bellamy P, Dawson NV, Desbiens N, Fulkerson WJ Jr, Goldman L, et al. A prognostic model for patients with end-stage liver disease. Gastroenterology 1997;113: del Olmo JA, Pe~na A, Serra MA, Wassel AH, Benages A, Rodrigo JM. Predictors of morbidity and mortality after the first episode of upper gastrointestinal bleeding in liver cirrhosis. J Hepatol 2000;32: Sanyal AJ, Banas C, Sargeant C, Luketic VA, Sterling RK, Stravitz RT, et al. Similarities and differences in outcomes of cirrhosis due to nonalcoholic steatohepatitis and hepatitis C. Hepatology 2006;43: Child CG, Turcotte JG. Surgery and portal hypertension. In: Child CG, ed. The Liver and Portal Hypertension. Philadelphia, PA: W.B. Saunders; Cash WJ, McConville P, McDermott E, McCormick PA, Callender ME, McDougall NI. Current concepts in the assessment and treatment of hepatic encephalopathy. QJM 2010;103: Atterbury CE, Maddrey WC, Conn HO. Neomycin-sorbitol and lactulose in the treatment of acute portal-systemic encephalopathy. A controlled, double-blind clinical trial. Am J Dig Dis 1978;23: 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: Weissenborn K, Ennen JC, Schomerus H, Ruckert N, Hecker H. Neuropsychological characterization of hepatic encephalopathy. J Hepatol 2001;34: Romero-Gomez M, Cordoba J, Jover R, del Olmo JA, Ramırez M, Rey R, et al. Value of the critical flicker frequency in patients with minimal hepatic encephalopathy. Hepatology 2007;45: