Journal of Pediatric Gastroenterology & Nutrition:
Original Articles: Hepatology and Nutrition
Pediatric End-stage Liver Disease Score in Acute Liver Failure to Assess Poor Prognosis
Sanchez, Maria C.; D’Agostino, Daniel E.
Pediatric Gastroenterology and Hepatology Division, Hospital Italiano Buenos Aires, Ciudad Autonoma Buenos Aires, Buenos Aires, Argentina.
Address correspondence and reprint requests to Maria C. Sanchez, MD, Hospital Italiano Buenos Aires, Ciudad Autonoma Buenos Aires, Buenos Aires, Argentina (e-mail: firstname.lastname@example.org).
Received 7 April, 2011
Accepted 25 August, 2011
The authors report no conflict of interest.
Background and Aim: Although establishing accurate prognosis in acute liver failure (ALF) is of paramount importance, prognostic scoring systems still fail to achieve success. The pediatric end-stage liver disease (PELD) score has been used as a predictor of mortality in children with chronic liver disease listed for liver transplantation (LT); however, experience with the PELD score in ALF is limited. The goal of the present study was to investigate the prognostic accuracy of the PELD score in children with ALF.
Patients and Methods: PELD score was calculated based on results of blood tests obtained at hospital admission from June 1999 to January 2009, in 40 consecutive patients younger than 18 years who presented with ALF. Poor outcome was defined as LT or death.
Results: Mean (±SD) age of patients was 5.3 ± 4.4 years (range 6 months–17 years); 52.5% were girls (n = 21). Etiologies of ALF were hepatitis A in 42.5% (17), indeterminate in 35% (14), autoimmune hepatitis in 17.5% (type 1 12.5% [n5], type 2 5% [n2]), and toxic in 5% (2). Mean PELD score was 34.92 ± 10.48 (range 6–55). PELD scores obtained on admission were significantly higher among nonsurvivors (39.8 ± 9.5) and recipients of an LT (39 ± 7.1) compared with those who survived without LT (31.3 ± 3) (P < 0.001). A cutoff of 33 in PELD score using receiver operating characteristic curves showed 81% specificity and 86% sensitivity for poor outcome (positive predictive value 92% and negative predictive value 69%; area under curve 0.88 95% confidence interval 0.77–1.0; P < 0.0001).
Conclusions: PELD score obtained upon admission may be of help to establish the optimal timing for LT evaluation and listing. Further validation in larger and more diverse populations is needed.
See “To Transplant or Not to Transplant: Certainly One of the Questions” by Shneider on page 164.
Acute liver failure (ALF) is characterized by an acute and diffuse liver lesion, with severe deterioration of liver function (1,2), responsible for a complex and multisystemic disorder. ALF is one of the most dramatic conditions in medicine occurring in healthy individuals and has a high risk of mortality. Coagulopathy and encephalopathy are classic indicators of liver insufficiency. In young children, encephalopathy can present with subtle mood changes, which may be difficult to detect. Encephalopathy is not considered essential for diagnosis (3).
Despite new therapies and support measures, survival continues to be low, ranging between 10% and 40% (4). After the introduction of liver transplantation (LT) as a therapeutic option for patients with ALF, survival rate reached 60% to 80% (5–7). The challenge consists in identifying early enough patients with ALF for whom chances of survival without LT are null to avoid developing severe complications that may contraindicate it. When to list a patient is critical for the success of LT. Knowledge of sensitive prognostic markers could also determine whether there is a possibility for recovery without LT. Transplantation implies surgical risks and permanent immunosuppression in a patient with a potentially reversible disease. LT must be indicated whenever chances for survival are lower without rather than with transplantation (8). Numerous prognostic models have been investigated mainly in adults with varying success. The goal of the present study was to investigate the prognostic accuracy of the PELD (pediatric end-stage liver disease)-MELD (model for end-stage liver disease) score in children with ALF admitted to our LT center.
PATIENTS AND METHODS
The retrospective analysis included a total of 50 consecutive patients younger than 18 years with ALF who were admitted to the pediatric intensive care unit of the Hospital Italiano, from June 1999 to January 2009. ALF was defined according to the Pediatric Acute Liver Failure Study Group criteria as the biochemical evidence of liver injury with absence of previously known history of chronic liver disease, coagulopathy not corrected by the use of vitamin K, and an international normalized ratio (INR) >1.5 in patients with encephalopathy or >2 in those without (9). Each PELD score was calculated based on blood tests results obtained at hospital admission. Poor outcome was defined as LT or death within 16 weeks of admission.
Laboratory test results collected at admission included serum bilirubin, albumin, creatinine, and INR. In addition, age, sex, weight, height, and etiology of ALF were recorded. The stage of encephalopathy was also registered at the time of admission, using the West-Hevan criteria (10). The PELD score was calculated from clinical parameters (failure to thrive: tables by sex based on weight and height for age) and the first laboratory results obtained upon hospital admission in children younger than 12 years. The MELD score was calculated based on first laboratory data in children older than 12 years, using the Internet calculator www.unos.org/resources/meld-PeldCalculator.asp(11). The King's College Hospital (KCH) criteria were calculated upon admission. All of the patients received standard therapy in the pediatric intensive care unit. Placing of an intracranial pressure monitor was indicated in those who progressed to stage 3 to 4 hepatic encephalopathy. Follow-up PELD scores were not established because several patients received fresh frozen plasma before placement of intracranial pressure monitors or undergoing other invasive procedures. Liver support devices were not used in the present study. LT was indicated in patients with stage 4 hepatic coma and in those with progression or lack of improvement of encephalopathy and/or coagulopathy during hospitalization. LT was contraindicated whenever irreversible brain damage, multiorgan failure, and uncontrolled infection developed (12,13). Medical care and criteria for listing and LT remained mostly unchanged throughout the last 10 years. The study was approved by the institutional review board of the Hospital Italiano.
Mean and standard deviation (range) were calculated for numeric variables and counts and percentages for categorical variables. The population was divided according to outcome during hospitalization into 2 groups: group I: survivors and group II: patients with poor prognosis, defined as LT or death. Comparisons between groups for numerical variables were performed by Student t test. P < 0.05 was considered statistically significant.
Receiver operating characteristic (ROC) curves were calculated for the PELD-MELD score results according to outcome. ROC curves are indicators of diagnostic precision and provide unifying criteria in test evaluation processes. The area under the ROC curve (AUC) >0.8 indicates an excellent diagnostic accuracy, and a model with an AUC >0.7 may be considered clinically useful (14). Variables calculated to describe the prognostic value in ALF included sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for both KCH and PELD-MELD score criteria. The Bayes theorem was applied to determine the posttest probability of LT or death. Statistical analyses were done with STATA software 8.0 (StataCorp, College Station, TX).
Of the total of 50 patients who were diagnosed as having ALF, 10 were excluded from analysis because of lack of complete laboratory data to calculate the PELD-MELD score. A total of 40 children with ALF were analyzed. Demographics, etiology, and clinical variants are shown in Table 1. Clinical variants of ALF, as defined by the interval between jaundice and encephalopathy and days in the transplant waiting list, did not differ between groups. Eleven (27.5%) survived with medical therapy, 6 (15%) died without LT, and 23 (57.5%) received an LT. Among the 11 patients who survived with medical treatment, only 7 were listed for LT. The remaining 4 patients significantly improved or resolved hepatic encephalopathy within 48 to 72 hours of hospitalization and therefore were not listed. Of the 6 patients who died without LT, 1 had a contraindication for the procedure on admission and 5 died while on the waiting list. Twelve (30%) developed stage 3 to 4 encephalopathy upon admission (Table 2). Mean PELD score was 34.92 ± 10.48 (range 6–55). PELD scores obtained on admission were significantly higher among nonsurvivors (39.8 ± 9.5) and patients receiving transplants (39 ± 7.1) compared with those who survived without LT (31.3 ± 3) (P < 0.001). Mean PELD score was significantly higher among patients with stage 0 to 2 encephalopathy who received LT or died within 16 weeks of admission (P < 0.0002) (Table 3).
A cutoff of 33 in PELD score using the ROC curve showed a specificity of 81% and a sensitivity of 86% for poor outcome (PPV 92% and NPV 69%, AUC 0.88 95%, confidence interval [CI] 0.77–1.0, P < 0.0001) (Fig. 1). The KCH criteria analysis provided a sensitivity of 88%, a specificity of 72%, a PPV of 88%, and an NPV of 61%, lower than what was calculated with the PELD-MELD score (Table 4). The pretest probability of a poor prognosis of 70% increased to 91% in those patients with a PELD score of 33 or higher (LR+ 4.52). In patients with a PELD score <33, the posttest probability decreased to 28% (LR− 0.17). The odds ratio between patients with positive and negative scores was 3.93 (CI 1.75–8.86, P = 0.000866).
In the present retrospective study, we evaluated the prognostic accuracy of the PELD-MELD score upon hospital admission in a cohort of pediatric patients with ALF. Using the ROC curve, it was determined that a PELD-MELD cutoff value of 33 or higher presented an accurate specificity and sensitivity to discriminate patients with poor outcome, a value identical to that found by Dhiman et al (8) in a study that compared the prognostic value of the MELD score versus the KCH criteria in adult patients. These results suggest that the PELD-MELD score obtained at admission in pediatric patients could be helpful in establishing the optimal moment for LT evaluation and inclusion in the emergency waiting list because it can significantly discriminate patients with poor prognosis. When patients with stage 0 to 2 encephalopathy are evaluated, the use of the PELD score may contribute to identify those patients with poor outcome despite absence of encephalopathy.
KCH criteria reported by O’Grady et al (15) and the Clichy criteria reported by Bernuau and Benhamou (16) and Bernuau et al (17) are considered to be the most valuable tools to assess prognosis. Several studies have reported a PPV of 70% to 100% and a NPV of 25% to 94%, with acceptable specificity but low sensitivity. Yantorno et al (18) concluded that Clichy criteria were more useful in children and that the MELD-PELD score was superior to that of the KCH and Clichy, with a diagnostic accuracy of 95% (C-statistic 0.96). Compared to the KCH and Clichy criteria, the MELD score was a better death predictor (KCH: PPV 71%, NPV 84%; Clichy: PPV 91%, NPV 72%; MELD >30: PPV 91%, NPV 100%). Renal failure developed in 45% of those who died while on the waiting list. This represents an additional advantage of MELD over KCH and Clichy because its calculation does not include serum creatinine as a prognostic variable.
The Pediatric Acute Liver Failure Study Group was created in 1999 and developed a database with the purpose of facilitating the comprehension of all of the aspects of ALF in children. By logistic regression analysis, it was shown that serum bilirubin values ≥5 mg/dL, INR ≥2.5, and hepatic encephalopathy were predictive factors of death or LT (9), both variables which are included in the PELD-MELD score. A study of 210 patients made in 2 pediatric LT centers in Argentina concluded that the most significant indicators of death or need for LT were prothrombin time and stage 3 to 4 encephalopathy (19). During the last 2 decades, many static and dynamic scores have been proposed to assess prognosis in patients with ALF such as the liver injury units score (20) and the acute physiology and chronic health evaluation (21); however, none of these models offered advantages over the existing ones. Several other studies suggested different clinical, biochemical, radiological, and histological prognostic markers (prothrombin time (22), factor V and factor VIII (23), serum bilirubin (24), α-fetoprotein (25), serum phosphate (26,27), lactate (28), galactose elimination capacity (29), ammonia (30), and vitamin D–binding protein (31)), images (residual liver value <700 cm3), necrosis extension obtained by transjugular biopsy (32,33), and others such as age, etiology, and encephalopathy stage (16,20). None of the aforementioned has been validated in prospective studies including a large enough cohort of pediatric patients.
Combining in the same group of analysis patients with LT with those that died could introduce a bias because it implies that every patient with LT would have otherwise died under medical treatment. Another limitation was the fact that 10 patients were excluded from the analysis because of the lack of complete laboratory reports to calculate the PELD-MELD score. Clinical characteristics of these 10 patients did not differ from the rest of the studied group. As opposed to chronic patients, the accessory variables of PELD over MELD in terms of weight and height by age would have no value in the acute presentation in which they would not be modified. Age could be of importance, however, because (as seen with the KCH criteria) young patients could present a higher risk of death. In addition, hepatitis A is an infrequent cause of ALF in most countries. Although the predominant cause of ALF in the present study was hepatitis A, we found a high percentage of indeterminate hepatitis (35%), similar to what was described in other geographical areas, and a surprisingly high percentage of autoimmune hepatitis (AIH) (17.5%). Finally, the absence of acetaminophen toxicity limits the generalization of the results.
Determining prognosis in ALF is critical for both LT indication and organ assignment. Future studies to improve these score systems will allow for a better assessment and evaluation, and optimize the organs and resources in adults and children.
The authors thank Dr Fernando Alvarez (CHU, Sainte-Justine, Université de Montréal) and Dr Carlos Lifschitz (Hospital Italiano, Buenos Aires, Argentina) for critical review of and corrections to the manuscript. We also thank the Department of Biostatistics of the Hospital Italiano for support for statistical analysis.
1. Caraceni P, Van Thiel D. Acute liver failure. Lancet
2. Mas A, Rodes J. Fulminant hepatic failure. Lancet
3. Tissieres P, Devictor D. Fulminant hepatic failure and transplantation. In: Roger's Textbook of Pediatric Intensive Care
, 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2008:1535–49.
4. Lee WM. Medical progress: acute liver failure. N Engl J Med
5. O’Grady JG, Alexander GJ, Thick M, et al. Outcome of orthotopic liver transplantation in the etiological and clinical variants of acute liver failure. Q J Med
6. Vickers C, Neuberger J, Buckels J, et al. Transplantation of the liver in adults and children with fulminant hepatic failure. J Hepat
7. Rhee C, Narsinh K, Venick RS, et al. Predictors of clinical outcome in children undergoing orthotopic liver transplantation for acute and chronic liver disease. Liver Transpl
8. Dhiman R, Jain S, Masheshawri 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
9. Squires RH JR, Shneider BL, Bucuvalas J, et al. Acute liver failure in children: the first 348 patients in the pediatric acute liver failure study group. J Pediatr
10. Ferenci P, Lockwood A, Mullen K, et al. Hepatic encephalopathy: definition, nomenclature, diagnosis and quantification, final report of the working party at the 11th World Congress of Gastroenterology Vienna, 1998. Hepatology
12. Cochran JB, Losek JB. Acute liver failure in children. Pediatr Emerg Care
13. Nadalin S, Heuer M, Wallot M, et al. Paediatric acute liver failure and transplantation: the University of Essen experience. Transpl Int
14. Wiesner RH, Edwards F, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology
15. O’Grady JG, Graeme JM, Hayllar KM, et al. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology
16. Bernuau J, Benhamou JP. Fulminant and subfulminant liver failure. In: McIntyre N, Benhamou JP, Bircher J, et al, eds. Oxford Textbook of Clinical Hepatology
. Oxford New York: University Press; 1991: 923–42.
17. Bernuau J, Goudeau A, Poynard T, et al. Multivariate analysis of prognostic factors in fulminant hepatitis B. Hepatology
18. Yantorno S, Kremers W, Ruf A, et al. MELD is superior to King's College and Clichy's criteria to assess prognosis in fulminant hepatic failure. Liver Transpl
19. Ciocca M, Ramonet M, Cuarterolo M, et al. Prognostic factors in paediatric acute liver failure. Arch Dis Child
20. Lu B, Gralla J, Liu E, et al. Evaluation of a scoring system for assessing prognosis in pediatric acute liver failure. Clin Gastroenterol Hepatol
21. Mitchell I, Bihari D, Chang R, et al. Earlier identification of patients at risk from acetaminophen-induced acute liver failure. Crit Care Med
22. Harrison PM, O’Grady JG, Keays RT, et al. Serial prothrombin time as prognostic indicator in paracetamol induced fulminant hepatic failure. Br Med J
23. Pereira LM, Langley PG, Hayllar KM, et al. Coagulation factor V and VIII/V ratio as predictors of outcome in paracetamol induced fulminant hepatic failure: relation to other prognostic indicators. Gut
24. Hoofnagle JH, Carithers RL, Shapiro C, et al. Fulminant hepatic failure: summary of a workshop. Hepatology
25. Murray-Lyon IM, Orr AH, Gazzard B, et al. Prognostic value of serum alpha-fetoprotein in fulminant hepatic failure including patients treated by charcoal haemoperfusion. Gut
26. 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
27. Schmidt LE, Dalhoff K. Serum phosphate is an early predictor of outcome in severe acetaminophen-induced hepatotoxicity. Hepatology
28. Bernal W, Donaldson N, Wyncoll D, et al. Blood lactate as an early predictor of outcome in paracetamolinduced acute liver failure: a cohort study. Lancet
29. Ranek L, Andreasen PB, Tygstrup N. Galactose elimination capacity as a prognostic index in patients with fulminant liver failure. Gut
30. Bhatia V, Singh R, Acharya SK. Predictive value of arterial ammonia for complications and outcome in acute liver failure. Gut
31. Lee WM, Galbraith RM, Watt GH, et al. Predicting survival in fulminant hepatic failure using serum Gc protein concentrations. Hepatology
32. Donaldson BW, Gopinath R, Wanless IR, et al. The role of transjugular liver biopsy in fulminant liver failure: relation to other prognostic indicators. Hepatology
33. Malinchoc M, Kamath PS, Gordon FD, et al. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology
This article has been cited 3 time(s).
Pediatric TransplantationPediatric acute liver failure: Etiology, outcomes, and the role of serial pediatric end-stage liver disease scoresPediatric Transplantation
Pediatric TransplantationArtificial neural network is highly predictive of outcome in paediatric acute liver failurePediatric Transplantation
Journal of Pediatric Gastroenterology and NutritionTo Transplant or Not to Transplant: Certainly One of the QuestionsJournal of Pediatric Gastroenterology and Nutrition
acute liver failure; children; PELD score
Copyright 2012 by ESPGHAN and NASPGHAN
Highlight selected keywords in the article text.