ORIGINAL ARTICLES LIVER, PANCREAS, AND BILIARY TRACT

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6: ORIGINAL ARTICLES LIVER, PANCREAS, AND BILIARY TRACT Prognostic Implications of Lactate, Bilirubin, and Etiology in German Patients With Acute Liver Failure JOHANNES HADEM,* PENELOPE STIEFEL,* MATTHIAS J. BAHR,* HANS L. TILLMANN, KINAN RIFAI,* JÜRGEN KLEMPNAUER, HEINER WEDEMEYER,* MICHAEL P. MANNS,* and ANDREA S. SCHNEIDER* *Department of Gastroenterology, Hepatology and Endocrinology, and Department of Visceral and Transplant Surgery, Hannover Medical School, Hannover; and Medical Clinic and Policlinic II, University of Leipzig, Leipzig, Germany Background & Aims: Among the potentially helpful indicators of poor prognosis in acute liver failure (ALF) are etiology, encephalopathy grade, blood lactate, and King s College Criteria (KCC). The accuracy of these parameters in predicting transplantation or death shows significant variation in different countries. Methods: We retrospectively analyzed 102 patients with ALF treated at our institution between 1996 and Baseline parameters, simplified acute physiology score III (SAPS-III), KCC, Model for End- Stage Liver Disease (MELD) score, and a novel score of bilirubin, lactate, and etiology (BiLE score) were compared between transplant-free survivors and patients who required liver transplantation or died, by using multivariate linear regression analysis and receiver operating characteristics (ROC). Results: The most common causes of ALF were indeterminate liver failure (21%), acute hepatitis B (18%), acetaminophen ingestion (16%), and Budd-Chiari syndrome (9%). Transplantation-free survival was 38%, 44% of patients underwent liver transplantation, and 18% died without transplantation. Eight-week survival was 77%. The BiLE score was the best predictor of death or need of transplantation, with 79% sensitivity and 84% specificity. ROC analysis revealed a better performance of BiLE score when compared with bilirubin, lactate, MELD score, and SAPS-III (area under the curve: , , , , and , respectively). Conclusions: The simple, combined BiLE score emerged as the best predictor of poor outcome in our patient cohort and should be prospectively evaluated in other populations. In 1970, Trey and Davidson 1 defined acute liver failure (ALF) as an acute deterioration of liver function resulting in the development of encephalopathy within 8 weeks of the onset of symptoms in a patient with a previously healthy liver. According to the recent position paper of the American Association for the Study of Liver Diseases, ALF is defined as decline in liver function of less than 26 weeks duration, absence of cirrhosis, evidence of coagulation abnormality (usually indicated by an international normalized ratio [INR] 1.5), and any degree of encephalopathy. Patients with Wilson s disease, vertically acquired HBV, or autoimmune hepatitis might be included in spite of possible cirrhosis if the disease has only been recognized for less than 26 weeks. 2 The composition of etiologies varies among geographic regions, an issue that has been addressed by a number of studies. 3 King s College Hospital London reported 57% acetaminophen toxicity and 9% viral hepatitis among 1014 patients between 1973 and A large prospective study involving 17 sites in the United States (Acute Liver Failure Group) collected data of 308 patients regarding etiology, clinical and laboratory features, and outcome of patients presenting with ALF. The most common etiologies were acetaminophen overdose (39%), indeterminate (17%), idiosyncratic drug reactions (13%), and viral hepatitis A and B (12%). 5 Early reports from France indicated viral hepatitis to be the most important cause of ALF in central Europe, 6 with acute hepatitis B accounting for up to 45% of cases. 7 The diversity of etiologies was just recently assessed by an international survey of 6 liver transplantation centers summarizing 284 ALF cases. Although acetaminophen was the dominant etiology in London and Copenhagen with 55% and 74% of cases, respectively, hepatitis B (including acute exacerbations of chronic hepatitis B) was the most prominent etiology in Hong Kong (74%) and Karachi (38%). In addition, a high proportion of cases in Karachi (35%) were due to hepatitis E. 8 A major problem in the management of ALF is the prediction of spontaneous recovery. Of numerous proposed prognostic criteria for ALF, the King s College criteria (KCC) have been the most commonly used and most frequently evaluated. 9 KCC are probably the best validated tool currently available. 10 However, the predictive value of KCC in the U.S. has been questioned 11 and has not been sufficiently evaluated in central Europe, where acetaminophen is not the leading cause of ALF. 12 Depending on etiology, KCC were shown to have a positive predictive value of 79% 88% and negative predictive value of Abbreviations used in this paper: ALF, acute liver failure; AUC, area under the curve; BiLE, bilirubin-lactate etiology; ICU, intensive care unit; INR, international normalized ratio; KCC, King s College Criteria; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; ROC, receiver operating characteristics; SAPS-III, simplified acute physiology score III by the AGA Institute /08/$34.00 doi: /j.cgh

2 340 HADEM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3 50% 65%. 13 In France, Clichy criteria are widely used to select ALF patients in need of orthotopic liver transplantation (OLT). Besides the presence of hepatic encephalopathy, they include patients age and factor V levels. 14 Only a few studies have attempted to directly compare the Clichy criteria and KCC, suggesting superiority of the KCC. 15,16 In acetaminophen-induced ALF, blood lactate levels above 3.0 mmol/l had a sensitivity of 76% and specificity of 97% in predicting death, 17 but results diverged in non acetaminophen-related ALF According to the large prospective U.S. multi-center study, encephalopathy grade at admission and etiology were the only predictors of outcome. 5 Recently, Yantorno et al 23 have suggested a cut-off value above 30 of the Model for End-Stage Liver Disease (MELD) score as predictor of poor outcome. The objectives of our retrospective analysis were to evaluate these prognostic parameters in a central European cohort. Methods Patients and Clinical Assessment We retrospectively identified 210 patients with acute hepatic dysfunction treated at the intensive care unit of our institution between 1996 and Of those, 102 patients (Figure 1) fulfilled the diagnostic criteria of ALF as initially published by Trey and Davidson 1 as well as those recently defined by the American Association for the Study of Liver Diseases (ie, hepatic encephalopathy, acute-onset increase of INR 1.5, and absence of signs of chronic liver disease in clinical and ultrasound examination). 2 The study was approved by the human research committee of the Medical School Hannover, thereby confirming that the protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. Informed consent for retrospective data acquisition was obtained from all patients whose current postal address was available or from patients next of kin in case of a fatal outcome. All parameters except the maximum encephalopathy grade were documented at admission to intensive care unit (ICU). Outcome end points included liver transplantation, death, and Figure 1. Outcome of 102 patients included in the retrospective analysis. Table 1. BiLE Score BiLE score Bilirubin ( mol/l)/100 lactate (mmol/l) 4 (in case of indeterminate ALF, Budd-Chiari syndrome, or phenprocoumon toxicity) 2 (in case of acetaminophen toxicity) 0 (in case of any other ALF etiology) NOTE. Sensitivity and specificity in predicting liver transplantation or death were calculated by using a BiLE score cut-off value 6.9. survival without transplantation for at least 8 weeks after admission at ICU. Hepatic encephalopathy was graded on a standard scale of I IV as described previously. 24 Simplified acute physiology score (SAPS) III, which is currently being evaluated as a predictive model in critical care patients, was calculated by using an EXCEL (Microsoft Corp, Redmond, WA) file provided by the SAPS-III Outcomes Research Group ( KCC were determined as described. 9 The calculation of MELD score was performed according to the equation: MELD score [9.57 log e creatinine (mg/dl) 3.78 log e bilirubin (mg/dl) log e INR 6.43]. 25 In addition to the established predictive tools, we introduced a simple combined bilirubin-lactate-etiology score (or BiLE score) that can be calculated right at the beginning of the ICU stay. The calculation of BiLE score, which was developed empirically, is BiLE score (baseline bilirubin [ mol/l]/100) baseline lactate [mmol/l] 4 [in case of indeterminate ALF, Budd-Chiari syndrome, or phenprocoumon toxicity] 2 [in case of acetaminophen toxicity] 0 [in case of any other ALF etiology] (Table 1). These parameters were then compared between the cohort of transplant-free survivors and patients who needed liver transplantation or died. Diagnoses were based on accepted diagnostic criteria that involved history, laboratory values, ultrasound imaging, and in individual cases histopathologic examination. Indeterminate ALF was assumed when the above mentioned diagnostic procedures including toxicology screens and serology for hepatitis A, B, and C, herpes simplex virus, varicella zoster virus, cytomegalovirus, and Epstein-Barr virus as well as autoantibodies were inconclusive. Decision to enroll a patient in the high urgency liver transplantation program was made according to the guidelines of the Eurotransplant International Foundation (see for details). Statistical Analysis All calculations were performed with SPSS software (version 13.0; SPSS Inc, Chicago, IL). Results are expressed as medians and ranges unless otherwise stated. Differences in discrete variables were tested with the 2 test or the Fisher exact test, wherever appropriate. Continuous variables were compared with the t test for normally distributed variables and with the Mann-Whitney test for non-normally distributed variables. Multivariate linear regression analysis was performed to evaluate prognostic value. Receiver operating characteristic (ROC) curves were calculated for the most relevant parameters. Results Demographic Characteristics and Clinical Data Of the 102 patients with acute liver failure, 72 (71%) were women. The median age of the group was 38 years (range,

3 March 2008 PROGNOSTIC IMPLICATIONS OF LACTATE, BILIRUBIN, AND ETIOLOGY IN ALF 341 Table 2. Characteristics of Patients at Admission According to Outcome Variable Survival without transplantation (n 39) Liver transplantation or death (n 63) P value 95% Confidence interval of differences Age (y) 37 (16 74) 39 (17 74) NS Women, n (%) 23 (59) 49 (78) NS Maximum encephalopathy grade II (I IV) III (I IV).001 I(n) 15 8 II (n) III (n) IV (n) 3 19 Prothrombin time (% of normal) 24 (8 75) 20 (4 48) NS INR 3.5 ( ) 4.0 ( ).05 WBC (per L) 9.4 ( ) 11.2 ( ) Hemoglobin (g/dl) 13.6 ( ) 12.6 ( ) Factor V (% of normal) 29 (4 165) 20 (4 63) NS Sodium (mmol/l) 136 ( ) 137 ( ) NS Creatinine ( mol/l) 69 (33 416) 82 ( ) NS ALT (U/L) 2633 (55 16,726) 1062 (12 10,011) AST (U/L) 2154 (40 60,620) 769 ( ).05 Alkaline phosphatase (U/L) 151 (19 333) 197 (8 597) Bilirubin ( mol/l) 103 (10 624) 263 (20 848) Ammonia ( mol/l) 41 (6 185) 72 (16 634) Lactate (mmol/l) 2.9 ( ) 4.7 ( ) MELD score 28 (13 49) 34 (14 62) KCC fulfilled, n (%) 7 (18) 35 (58).001 SAPS-III 54 (39 98) 61 (42 103).004 BiLE score 4.3 (0 14.7) 9.7 ( ) BiLE score 6.9, n (%) 6 (16) 50 (79).001 NOTE. All variables are expressed as medians (and ranges), unless otherwise stated. Abbreviation: WBC, white blood count years). Exact data on the duration of jaundice before onset of encephalopathy were available for 54 patients. The median interval between jaundice and onset of encephalopathy was 6 days (range, 0 40 days) in the overall cohort, 1 day (range, 0 39 days) in patients surviving without OLT (referred to as OLT-free survival), and 9 days (0 40 days) in those who underwent transplantation or died (referred to as OLT or death) (P.01). Median value for individual maximum hepatic encephalopathy grade was III (range, I IV) for the whole group, II (range, I IV) in case of OLT-free survival, and III (range, I IV) in case of OLT or death (P.001). A serum creatinine of 2 mg/dl (177 mol/l) or greater was present in 6 of 39 (15%) patients with OLT-free survival versus 17 of 63 (27%) patients with OLT or death (not significant). Details are presented in Table 2. Etiology of Acute Liver Failure The distribution of etiologies of ALF was heterogeneous without predominance of one particular cause. Indeterminate ALF accounted for 21 patients (21%). Acute hepatitis B infection led to 18 cases (18%, 1 patient with hepatitis B/D coinfection), and 16 acetaminophen ingestions (16%) were observed. Another 9 patients (9%) had acute Budd-Chiari syndrome. Phenprocoumon toxicity and idiosyncratic drug reactions were deemed responsible for 7 (7%) and 5 (5%) cases of ALF, respectively. Diagnoses of Amanita phalloides ingestion, Wilson s disease, hepatitis A, ischemic hepatitis ( shock liver ), and halothane reaction accounted for 5 (5%), 5 (5%), 4 (4%), 4 (4%), and 3 (3%), respectively. Other causes were seen in 5 (5%) of ALF patients. Among them were one patient with acute lymphoblastic leukemia and another one with myelodysplastic syndrome who developed veno-occlusive disease after stem cell transplantation. An illustration of ALF etiologies is given in Figure 2. Figure 2. Causes and outcomes of ALF. Percentage of each etiology with regard to the whole study cohort is given on top of each bar.

4 342 HADEM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3 Table 3. Characteristics of Patients at Admission According to the 3 Most Common Etiologies Variable Indeterminate (n 21) Fulminant hepatitis B (n 18) Acetaminophen (n 16) Age (y) 43 (28 73) 34 (20 57) 37 (20 52) Women, n (%) 18 (86) 11 (61) 9 (56) Maximum encephalopathy grade III (I IV) II (I IV) II (I IV) Prothrombin time (% of normal) 22 (8 43) 19 (4 39) 24 (8 48) WBC (per L) 11.0 ( ) 8.6 ( ) 11.6 ( ) Hemoglobin (g/dl) 12.9 ( ) 13.3 ( ) 13.0 ( ) Coagulation factor V (% of normal) 21 (7 54) 28 (4 165) 19 (7 42) Sodium (mmol/l) 138 ( ) 137 ( ) 138 ( ) Creatinine ( mol/l) 72 (28 485) 65 (3 482) 87 ( ) ALT (U/L) 1062 ( ) 3345 ( ) 3497 ( ) AST (U/L) 803 ( ) 1435 ( ) 2474 ( ) Alkaline phosphatase (U/L) 212 ( ) 193 (19 333) 179 (45 331) Bilirubin ( mol/l) 310 (73 660) 214 (67 624) 58 ( ) Ammonia ( mol/l) 106 (24 634) 94 (6 218) 83 (17 496) Lactate (mmol/l) 4.1 ( ) 4.3 ( ) 3.2 ( ) C-reactive protein (mg/l) 9 (4 40) 22 (4 148) 38 (2 208) MELD score 34 (23 50) 33 (20 44) 27 (13 62) KCC fulfilled, n (%) 18 (86) 2 (11) 4 (25) BiLE score 11.3 ( ) 6.9 ( ) 1.9 (0 25.2) BiLE score 6.9, n (%) 21 (100) 8 (44) 3 (19) Received transplant, n (%) 13 (62) 9 (50) 2 (13) Died, n (%) 6 (29) 4 (22) 3 (19) NOTE. All variables are expressed as medians (and ranges), unless otherwise stated. Abbreviation: WBC, white blood count. Comparison of Clinical and Laboratory Features by Etiology A comparison of the 3 most common etiologies of ALF in our cohort, ie, indeterminate ALF, hepatitis B, and acetaminophen ingestion, is given in Table 3. There is a notably high proportion of women (18 of 21, 86%) in indeterminate ALF. Patients with indeterminate ALF also had a higher maximum encephalopathy grade (median III) and higher bilirubin level, but ALT and C-reactive protein were markedly lower than in the total cohort, respectively. In contrast, acetaminophen toxicity affected men and women equally and was associated with higher ALT values. Although information on jaundice duration was available for 54 patients only, time interval between jaundice and onset of encephalopathy appeared to vary between groups. Outcome of Acute Liver Failure As shown in Figure 1, overall survival 8 weeks after admission at ICU was 77% (79 of 102 patients). Thirty-nine patients (38%) survived without transplantation, and 45 (44%) received a liver graft. Five patients (5%) died after transplantation, 5 (5%) died awaiting a graft, and 13 (13%) died without being listed for transplantation. In the latter group, causes of death were multiorgan failure, septicemia, malignancies, and respiratory failure, among others. Eight-week outcome was strongly influenced by the cause of ALF. Survival without transplantation was high in hepatitis A (4 of 4 patients, 100%), A phalloides toxicity (4 of 5, 80%), idiosyncratic drug reaction other than phenprocoumon (4 of 5, 80%), and acetaminophen ingestion (12 of 16, 75%). Low rates of survival without OLT were seen in Wilson s disease (0 of 5 patients, 0%), indeterminate ALF (1 of 21, 5%), Budd-Chiari syndrome (1 of 9, 11%), and phenprocoumon toxicity (1 of 7, 14%). Transplantation was performed in only 12% of the acetaminophen group as compared with 62% of the indeterminate ALF group. Women were more likely to require liver transplantation or die than men (Figure 1). Predictive Factors for a Fatal Outcome Several clinical and laboratory parameters as well as prognostic models emerged as discriminators between survival without OLT on the one hand and OLT or death on the other (Table 2). Lactate (cutoff 3.5 mmol/l), MELD score (cutoff 32), and KCC achieved a sensitivity and specificity of 59% and 66%, 65% and 69%, and 58% and 82%, respectively (Table 4). No Table 4. Predicting Death or Need of Transplantation Parameter Cut-off value Sensitivity (%) Specificity (%) Positive predictive value (%) Negative predictive value (%) Lactate 3.5 mmol/l Bilirubin 140 mol/l MELD score KCC BiLE score

5 March 2008 PROGNOSTIC IMPLICATIONS OF LACTATE, BILIRUBIN, AND ETIOLOGY IN ALF 343 Figure 3. Lactate, MELD score, and BiLE score discriminate between survivors without liver transplantation and liver transplantation or death. Survivors without liver transplantation (white boxes) are compared with patients who required transplantation or died (grey boxes). Box plots show medians as well as interquartile ranges. Error bars show ranges. statistical difference was observed for sodium, creatinine, prothrombin time, and factor V. Multivariate linear regression analysis revealed that bilirubin and lactate were the most predictive individual laboratory parameters in our patients. We attempted to create a prognostic score simple enough for bedside use that could be calculated right after ICU admission. Therefore, baseline bilirubin ( mol/l) was divided by 100, thereby transferring the values of this parameter to the value range of baseline lactate (mmol/l). To take into account the predictive importance of ALF etiology, the sum of bilirubin/100 and lactate was then modified by adding positive or negative summands for certain etiologic entities. The novel combined BiLE score (calculation formula given in Table 1) achieved highest significance in predicting liver transplantation or death (Figure 3; Table 4). With a cut-off value of 6.9, its sensitivity, specificity, and positive and negative predicitive values were 79%, 84%, 89%, and 71%, respectively. Sensitivity of BiLE score was particularly high (100%) in patients with indeterminate ALF. In addition, the performance of the BiLE score in comparison to other parameters was measured by ROC analysis. Area under the curve (AUC) values revealed a better performance of the BiLE score in comparison with bilirubin, lactate, KCC, MELD score, and the SAPS III score (Figure 4, Table 5). Discussion ALF is estimated to cause approximately 3.5 deaths per million people per year in industrialized countries. 26 The currently largest prospective study on ALF, which enrolled 308 patients in the United States, demonstrated a 43% chance of survival without OLT and an overall survival of 67%. 5 Our study represented a large cohort of patients with acetaminophen and non acetaminophen-induced ALF in central Europe. Of 102 patients, 38% survived without transplantation. The overall survival 8 weeks after ICU admission in our patient group was 77%, slightly higher than in the recently published U.S. study. This might be due to differences between the European (Eurotransplant) and U.S. (United Network for Organ Sharing) liver allocation systems (mortality on the waiting list was 5% in our Figure 4. ROC curves for prediction of death or need of liver transplantation. (A) Comparison of lactate and bilirubin with BiLE score. (B) Comparison of MELD score and SAPS-III with BiLE score. Abbreviations are shown in Table 2. cohort [data not shown] versus 10% in the U.S. study). 5 Obviously, our data might be subject to referral bias. Analyzing patients from a single center might on the other hand imply a more homogeneous patient cohort as a result of low interindividual variability regarding diagnostic criteria and standard of care. Finally, we did not include patients with acute exacerbations of chronic hepatitis B, although arguably this scenario might be difficult to distinguish from and overlap with fulminant hepatitis B induced ALF. 8 Recent studies have shown a remarkable geographic diversity of ALF etiologies. 3 5,7 9,14,17 Our data found indeterminate ALF Table 5. Predicting Death or Need of Transplantation: ROC Parameter n AUC SE (95% CI) P value Lactate ( ).001 Bilirubin ( ).001 MELD score ( ).001 BiLE score ( ).001 SAPS-III ( ).005 Abbreviations: SE, standard error; 95% CI, 95% confidence interval.

6 344 HADEM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3 (21%) and acute hepatitis B (18%) as the most common etiologic ALF entities, whereas acetaminophen ingestion and Budd- Chiari syndrome accounted for only 16% and 9%, respectively. In that sense, overall survival rates of greater than 75% are even more surprising, considering the low survival rate without OLT (17% in the U.S. study, only 5% in our study) in patients with ALF of indeterminate origin. 5 The high incidence of indeterminate ALF in Germany is one of the most interesting results of this study. As in the U.S. ALF study, 71% of patients were women. But interestingly, women accounted for 86% of patients with indeterminate ALF. Whether women are innately more susceptible to ALF or are more commonly exposed to different kinds of drugs remains to be determined. 27 Other statistically significant features of indeterminate ALF were the low levels of liver enzymes, C-reactive protein, the relatively high bilirubin levels, and a higher encephalopathy grade. One can speculate that a high proportion of cases of indeterminate ALF take a subacute course with late ICU admission, which has been shown to be associated with poor prognosis in other reports 7 as well as in this study. Furthermore, it would be interesting to know whether infectious complications contribute to encephalopathy in these patients. 28 The survival rate without OLT of 38% in acute hepatitis B might be an underestimation of today s prognosis of these patients. Although several case reports have suggested that early treatment with lamivudine in patients with subfulminant or fulminant hepatitis B might reduce the need for life-saving liver transplantation, 29 there is no agreement whether antiviral therapy should be instituted in this setting. In fact, a randomized placebo-controlled trial from India did not support lamivudine treatment in acute hepatitis B. 30 In our center, patients with ALF have been treated with lamivudine since 2000 (ie, 12 of 18 patients with acute hepatitis B in the present cohort). Patients with ALF caused by phenprocoumon toxicity represented a unique feature of this German patient cohort. Phenprocoumon, the most widely used coumarin anticoagulant in Germany, has been associated with hepatitis and subacute liver failure (2% and 0.2%, respectively, of phenprocoumon-associated adverse events in Germany). 31 The poor outcome in this group was probably influenced by contraindications to transplantation imposed by the multimorbidity of these patients. The major aim of our study was to evaluate commonly used prognostic models in ALF, comparing transplant-free survivors with patients who underwent transplantation or died. Limitations come from the retrospective data analysis. A number of patients had to be excluded from the study because of incomplete documentation of hepatic encephalopathy at ICU admission. Our study was also limited by the fact that treatment and outcome might have changed during the 9-year study period; one example is the treatment of fulminant hepatitis B with lamivudine as discussed above. 29 Finally, defining poor prognosis as death or liver transplantation might have implicated a bias related to the decision whether to perform transplantation in an individual patient. We found that in German ALF patients with a high proportion of indeterminate ALF and acute viral hepatitis maximum encephalopathy grade, bilirubin, lactate, KCC, and MELD score could all discriminate between transplant-free survivors and patients who needed liver transplantation or died. Our observation is in line with a recent study on 144 patients with ALF caused by acute viral hepatitis from India. The authors found that the presence of 3 of 6 clinical parameters (among them, encephalopathy grade and jaundice-encephalopathy interval) and to a lesser extent also MELD and KCC discriminated survivors from nonsurvivors. 32 The prognostic value of MELD in acetaminophen-induced liver injury has recently been investigated prospectively in 460 patients. MELD was not found to be superior over KCC or INR in predicting death. 33 Two other studies in patients with non-acetaminophen ALF and seronegative hepatitis suggested that KCC might be of similar predictive strength as MELD, and that its application results in high patient survival rates. 34,35 Multivariate linear regression analysis of our data revealed that bilirubin and lactate were the most predictive of all parameters evaluated. Although the prognostic value of lactate is well-recognized in acetaminophen-induced ALF, 17,36 its clinical utility in non acetaminophen-related ALF is less well-established because of the small patient number examined and etiologic heterogeneity of this group of ALF patients. Dabos et al 20 studied 59 patients with non acetaminophen-related ALF, of whom 34 underwent liver transplantation or died and had significantly higher lactate levels. MacQuillan et al 22 presented probability curves of nonsurvival on the basis of blood lactate levels in 27 patients with non acetaminophen-related ALF. Baseline and 8-hour lactate levels did not differ significantly between 7 spontaneous survivors and 20 patients who needed liver transplantation or died. Finally, hyperlactatemia was not considered to be of prognostic value in 63 patients with non acetaminophen-related ALF. However, the interpretation of this study by Taura et al 21 is complicated by the fact that only liver transplant candidates were included. Because 86 ALF cases (84%) of our patient cohort were non acetaminophen-related, our study added evidence to the assumption that lactate is among the best predictors of a fatal outcome even in non acetaminophen-related ALF. Bilirubin and lactate in combination with etiology were subsequently integrated into a novel, empirically developed prognostic score that can be calculated at bedside right after ICU admission. This combined BiLE score achieved the highest sensitivity and specificity in predicting poor outcome and was revealed to be superior over other parameters in the ROC analysis. The particular advantage of BiLE score might be its very high sensitivity in predicting death or transplantation in indeterminate ALF. It would therefore be interesting to learn more about the implications of this parameter in areas of the world in which non-acetaminophen ALF is predominant. In our opinion, BiLE score deserves to be validated in other centers with other patient cohorts. References 1. Trey C, Davidson CS. The management of fulminant hepatic failure. In: Popper H, Schaffner F, eds. Progress in liver diseases. New York: Grune and Stratton, 1970: Polson J, Lee WM. AASLD position paper: the management of acute liver failure. Hepatology 2005;41: Goldberg E, Chopra S. Fulminant hepatic failure: definition, etiology, and prognostic indicators. In: Uptodate 2004, version Available at: Accessed on May 15, O Grady JG, Portmann B, Williams R. Fulminant hepatic failure. In: Schiff L, Schiff R, eds. Diseases of the liver. Philadelphia: Lippincott, 1993: Ostapowicz G, Fontana RJ, Schiødt FV, et al. Results of a pro-

7 March 2008 PROGNOSTIC IMPLICATIONS OF LACTATE, BILIRUBIN, AND ETIOLOGY IN ALF 345 spective study of acute liver failure at 17 tertiary care centers in the United States. Ann Intern Med 2002;137: Bernuau J, Rueff B, Benhamou JP. Fulminant and subfulminant liver failure: definitions and causes. Semin Liver Dis 1986;6: O Grady JG. Acute liver failure. In: O Grady JG, Lake JR, Howdle PD, eds. Comprehensive clinical hepatology. London: Mosby, 2000: Bernal W, Schiødt FV, Hamid S, et al. Etiologies and outcomes for acute liver failure at 6 sites around the world. Dig Dis Sci 2005; 24:A O Grady JG, Alexander GJM, Hayllar KM, et al. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology 1989; 97: Renner EL. How to decide when to list a patient with acute liver failure for liver transplantation? Clichy or King s College criteria, or something else? J Hepatol 2007;46: Lee WM. Acute liver failure in the United States. Semin Liver Dis 2003;23: Escorsell A, Mas A, de la Mata M. Acute liver failure in Spain: analysis of 267 cases. Liver Transplant 2007;13: Anand AC, Nightingale P, Neuberger JM. Early indicators of prognosis in fulminant hepatic failure: an assessment of the King s criteria. J Hepatol 1997;26: Bismuth H, Samuel D, Castaing D, et al. Orthotopic liver transplantation in fulminant and subfulminant hepatitis: the Paul Brousse experience. Ann Surg 1995;222: Izumi S, Langley PG, Wendon J, et al. Coagulation factor V levels as a prognostic indicator of fulminant hepatic failure. Hepatology 1996;23: Pauwels A, Mostefa-Kara N, Florent C, et al. Emergency liver transplantation for acute liver failure. J Hepatol 1993;17: Bernal W, Donaldson N, Wyncoll D, et al. Blood lactate as an early predictor of outcome in paracetamol-induced acute liver failure: a cohort study. Lancet 2002;359: Bihari D, Gimson AE, Lindridge J, et al. Lactic acidosis in fulminant hepatic failure. Some aspects of pathogenesis and prognosis. J Hepatol 1985;1: Murphy ND, Kodakat SK, Wendon JA, et al. Liver and intestinal lactate metabolism in patients with acute hepatic failure undergoing liver transplantation. Crit Care Med 2001;29: Dabos KJ, Newsome PN, Parkinson JA, et al. Biochemical prognostic markers of outcome in non-paracetamol-induced fulminant hepatic failure. Transplantation 2004;27: Taura P, Martinez-Palli G, Martinez-Ocon J, et al. Hyperlactatemia in patients with non-acetaminophen-related acute liver failure. World J Gastroenterol 2006;12: MacQuillan GC, Seyam MS, Nightingale P, et al. Blood lactate, but not serum phosphate levels can predict patient outcome in fulminant hepatic failure. Liver Transpl 2005;9: Yantorno SET, Trentadue JJ, Ruf AE. The model for end stage liver disease (MELD): a useful tool to assess prognosis in fulminant hepatic failure. Liver Transl 2004;10:C Lee WM. Acute liver failure. N Engl J Med 1993;329: Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology 2003;124: Hoofnagle JH, Carithers RL, Shapiro C, et al. Fulminant hepatic failure: summary of a workshop. Hepatology 1995;21: Third National Health and Nutrition Examination Survey , NHANES III Second Laboratory Data File (CD-ROM, series 11, no 2A), US Department of Health and Human Services, National Center for Health Statistics. Hyattsville, MD. Centers for Disease Control and Prevention, Vaquero J, Polson J, Chung C, et al. Infection and the progression of encephalopathy in ALF. Gastroenterology 2003;125: Tillmann HL, Hadem J, Leifeld L, et al. Safety and efficacy of lamivudine in patients with severe acute or fulminant hepatitis B, a multicenter experience. J Viral Hepat 2006;13: Kumar M, Satapathy S, Monga R, et al. A randomized controlled trial of lamivudine to treat acute hepatitis B. Hepatology 2007; 45: Schimanski CC, Burg J, Möhler M, et al. Phenprocoumon-induced liver disease ranges from mild acute hepatitis to (sub-) acute liver failure. J Hepatol 2004;41: Dhiman RK, Jain S, Maheshwari U, et al. Early indicators of prognosis in fulminant hepatic failure: an assessment of the Model for End-Stage Liver Disease (MELD) and King s College Hospital Criteria. Liver Transpl 2007; Mar 16 [epub ahead of print]. 33. Schmidt LE, Larsen FS. MELD score as a predictor of liver failure and death in patients with acetaminophen-induced liver injury. Hepatology 2007;45: Katoonizadeh A, Decaestecker J, Wilmer A, et al. MELD score to predict outcome in adult patients with non-acetaminophen-induced acute liver failure. Liver Int 2007;27: Wigg AJ, Gunson BK, Mutimer DJ. Outcomes following liver transplantation for seronegative acute liver failure: experience during a 12-year period with more than 100 patients. Liver Transpl 2005;1: Schmidt LE, Larsen FS. Prognostic implications of hyperlactatemia, muliple organ failure, and systemic inflammatory response syndrome in patients with acetaminophen-induced acute liver failure. Crit Care Med 2006;34:337. Address requests for reprints to: Johannes Hadem, MD, Department of Gastroenterology, Hepatology, and Endocrinology, Medical School Hannover, Carl-Neuberg-Strasse 1, Hannover, Germany. hadem.johannes@mh.hannover.de; fax: