Abnormal concentrations of esterified carnitine in bile: A feature of pediatric acute liver failure with poor prognosis§


  • See Editorial on Page 696

  • Presented in part at the Meeting of the American Association for the Study of Liver Diseases, Boston, MA, October 27, 2003.

  • §

    Potential conflict of interest: Nothing to report.


The etiology of acute liver failure in children is unknown in a large number of cases. Defects in fatty acid oxidation have been shown to lead to severe liver injury. This retrospective analysis examined the bile acylcarnitine profiles of 27 children with acute liver failure who underwent liver transplantation or died. Results were compared with 758 postmortem samples from individuals without acute liver failure. Cumulative amounts of free carnitine, medium- or long-chain species in excess of the 95th percentile of the control group were considered abnormal. Fourteen samples had normal profiles. Three had markedly elevated concentrations of free carnitine, whereas ten showed elevations in medium- or long-chain species. The relative risk of death was 2.86 (95% confidence interval, 1.08-7.54, P = .01) in the 10 children with elevated concentrations of medium- or long-chain species compared with those with normal analyses. Overall, medium- and long-chain acylcarnitines were increased in those patients who died compared with survivors, (dead vs. alive; medium-chain, 187 ± 74 vs. 32 ± 12 μmol/L, P = .008; long-chain, 146 ± 74 vs. 15 ± 8 μmol/L, mean ± standard error of the mean, P = .018). These studies describe biliary free and esterified carnitine profiles in children with acute liver failure. In conclusion, the findings raise the hypothesis that abnormalities in fatty acid oxidation may predispose to a worse outcome in acute liver failure. (HEPATOLOGY 2005;41:717–721.)

Acute liver failure (ALF) in children is a clinical syndrome consisting of biochemical evidence of severe hepatic synthetic dysfunction with or without encephalopathy.1 In most circumstances, the etiology cannot be discerned. Well-characterized causes of ALF in children include acetaminophen toxicity, hepatitis A, hepatitis B, drug toxicity, autoimmune hepatitis, Wilson disease, and a range of rare metabolic diseases. Abnormalities in fatty acid metabolism have been associated with ALF in children.

Fatty acid oxidation (FAO) plays a major role in energy production during periods of fasting. More than 25 enzymes and transporters are involved in this pathway.2 Inherited FAO disorders represent a new and rapidly expanding class of metabolic diseases.3, 4 More than 20 FAO disorders have been described to date and present with a variety of clinical manifestations, including metabolic decompensation during fasting, hypoketotic hypoglycemia, abnormal function of fatty acid–dependent tissues, particularly liver and heart, and sudden and unexpected death. Several FAO disorders are described that present with Reye-like episodes or severe liver disease as the predominant clinical feature. Carnitine palmitoyltransferase I deficiency and medium-chain acyl-coenzyme A dehydrogenase deficiency may present with hepatomegaly, abnormal aminotransferases, steatosis, and encephalopathy. The clinical picture is not typically one of acute liver failure, although the encephalopathy is reminiscent of Reye's syndrome.5–8 More typical features of acute liver failure, including marked elevation in alanine aminotransferase, aspartate aminotransferase, and bilirubin, coupled with coagulopathy and encephalopathy, have been described in long-chain L-3-hydroxy acyl-coenzyme A dehydrogenase deficiency and short-chain L-3-hydroxy acyl-coenzyme A dehydrogenase deficiency and also in patients with defective processing of long-chain fatty acids into cultured skin fibroblasts.9–11

It is not surprising that defects in FAO could lead to acute liver injury. Intermediate metabolites in FAO are toxic and can lead to cellular necrosis and dysfunction.12 Because the liver is the site of a significant amount of FAO, it is a potential target for organ injury. Analysis of acylcarnitines in bile has been shown to be particularly useful in the identification of potential FAO defects.13, 14 The objective of this study was to evaluate FAO intermediates in bile samples obtained from children with ALF to determine whether FAO defects are present in pediatric ALF cases, especially those with indeterminant etiology.


ALF, acute liver failure; FAO, fatty acid oxidation.

Materials and Methods

ALF, as defined by the Pediatric ALF Study Group, is the constellation of acute liver injury coupled with either severe coagulopathy (international normalized ratio > 2.0 or prothrombin time > 20 seconds) or encephalopathy in the setting of moderate coagulopathy (INR > 1.5 or prothrombin time > 15 seconds). Disease severity in the particular cohort of patients in this analysis was high, as samples were obtained in some but not all patients who either underwent liver transplantation or autopsy; thus, the samples do not reflect all patients enrolled in the Pediatric ALF study. Clinical diagnoses were made by using standard criteria before the final result of the bile analysis and were determined by the clinical team caring for the patient. Bile samples were obtained from the gallbladder of the explanted liver at the time of liver transplantation or at the time of postmortem investigation. Bile acylcarnitine analyses were performed by electrospray ionization tandem mass spectrometry as previously described.15 Results were compared with normative data derived from 758 postmortem bile samples clinically tested in the context of a postmortem screening protocol.15 The median values for free carnitine, medium-chain acylcarnitine species, and long-chain species were 80.3, 5.2 and 2.9 μmol/L, respectively. Two cutoff levels were set to classify abnormal results: greater than the 99th percentile (365, 221, and 126 μmol/L) and greater than the 95th percentile (228, 69, and 41 μmol/L). Concentrations of individual species in bile were totaled to arrive at a cumulative amount of medium- (C6-C10) or long-chain (C14-C18) species. Statistical analyses were performed by analysis of variance using the Tukey-Kramer multiple comparison test and by Fisher's exact test using software by Graph Pad (San Diego, CA). Institutional review board approval was obtained for this retrospective analysis of clinical data and for prospectively collected data as part of the ALF Study Group.


A total of 27 bile samples were assessed (Table 1). The clinical diagnoses in the 27 patients were indeterminant (n = 17; 3 with aplastic anemia), autoimmune hepatitis (n = 3), acetaminophen overdose (n = 2), “metabolic” (n = 2), Wilson disease (n = 1), tyrosinemia type I (n = 1), and Herpes simplex infection (n = 1). Twenty-two patients underwent liver transplantation, and three died in the postoperative period. Five children died without liver transplantation.

Table 1. Clinical and Biochemical Analyses
 IDOLTDeathDiagnosisAge (years)Max ALT (IU/L)Max TB (mg/dL)Max PT (sec)MC (μmol/L)LC (μmol/L)FC (μmol/L)Steatosis
  1. NOTE. Subjects sorted by predominant abnormality if more than one present. Fourteen patients (1–14) had uninformative profiles. Three patients (15–17) showed marked elevation of free carnitine. Ten patients (18–27) had elevated concentrations of medium- or long-chain species. Three of the latter group (25–27) had increases greater than 99%.

  2. Abbreviations: OLT, orthotopic liver transplant; Max, maximum; ALT, alanine aminotransferase; TB, total bilirubin; PT, prothrombin time; MC, summation of medium-chain species; LC, summation of long-chain species; FC, free carnitine; y, yes; n, no; aa, aplastic anemia; AIH, autoimmune hepatitis.

Normaln = 14           
NormalMeann = 12n = 2 51,73921.853.5   n = 3
 SEM   1.54102.06.4    
FC > 95 percentilen = 0           
FC > 99 percentilen = 3           
  n = 3n = 0        n = 2
MC/LC > 95 percentilen = 7           
MC/LC > 99 percentilen = 3          n = 3
MC/LC > 95 percentileMeann = 6n = 7 9.63,89411.453.4   n = 2
 SEM   2.31,1382.38.6    
 p, MC/LC > 95 percentile compared to normal>.05.01 .09.053.002>.05   .203

Acylcarnitine analysis of 14 of the bile samples was unremarkable; of these, the diagnosis was indeterminant in 10 (including all three with aplastic anemia), and known in four (one each for autoimmune hepatitis, Wilson disease, herpes, and tyrosinemia). Thirteen of the 27 samples showed abnormal patterns of free and esterified carnitine species. The pattern of abnormalities could be divided into two categories (Fig. 1): (1) elevated free carnitine (508-1208 μmol/L, indeterminant [n = 3]) and (2) elevated medium- or long-chain acylcarnitines (3 > 99th percentile, 7 > 95th percentile, indeterminant [n = 4], metabolic [n =2], autoimmune hepatitis [n=2], and acetaminophen [n =2]). Comparison of the clinical characteristics of these patients with those with normal analyses of bile can be seen in Table 1. Maximum alanine aminotransferase was nearly twice as high in children with elevations in biliary medium- or long-chain acylcarnitines, whereas maximal total bilirubin was lower. The risk of death (RR, 2.86%-95%; confidence interval, 1.08-7.54) was higher in children with abnormal medium- or long-chain species in bile compared with those with normal bile (7/10 vs. 2/14, P = .01). All deaths in the children with abnormal bile profiles were directly attributable to complications resulting from acute liver failure (primarily brainstem herniation or subsequent irreversible neurologic injury, n = 2; multi-organ failure, n = 5). Five of the ten patients with elevations in biliary medium- or long-chain acylcarnitines had hepatic microvesicular steatosis noted on routine processing of the explanted liver or liver specimen obtained at autopsy (Fig. 2). Hepatic steatosis was more common in patients with abnormalities in bile compared with those without (7/13 vs. 3/14, P = .12). Too few patients had elevated free carnitine levels for meaningful statistical analysis. Overall, medium- and long-chain acylcarnitines were increased in those patients who died compared with survivors, whereas free carnitine levels were similar (dead vs. alive: medium-chain, 187 ± 74 vs. 32 ± 12, P = .008; long-chain, 146 ± 74 vs. 15 ± 8, P = .018; free carnitine, 140 ± 35 vs. 201 ± 80, mean ± SEM, P > .05).

Figure 1.

Pattern of biliary acylcarnitines and carnitine. A scatter plot of the concentration of biliary acylcarnitines and carnitines for the study subjects is shown. Each point represents the result of the analysis of a single subject. There are 27 points in each group. Patients 15, 16, 17, 25, 26, and 27 are highlighted in view of their measurements that are greater than the 99th percentile. FC, free carnitine; MC, medium chain (C6-C10); LC, long chain (C14-C18).

Figure 2.

Liver histology. Representative histology of the explanted liver from patient 26 is shown. (A) Severe, diffuse, centrilobular necrosis of the liver. The arrows indicate portal tracts surrounded by residual viable hepatocytes in zone 1 (hematoxylin-eosin stain; original magnification ×100). (B) Higher magnification shows macrovesicular and microvesicular steatosis of the residual hepatocytes. The arrow indicates a bile duct within the adjacent portal area (original magnification ×200).


This study evaluates biliary acylcarnitine metabolites in patients with ALF. The findings support several conclusions and raise important issues regarding the pathophysiology of ALF in children. First, the pattern of acylcarnitines in bile can be unremarkable despite the presence of severe ALF. The control group of samples for this analysis derives exclusively from postmortem samples. Given the stability of acylcarnitines in bile, this is likely to be an adequate control for bile samples obtained at the time of transplantation. The finding of these unremarkable profiles strongly suggests that the observation of an abnormal pattern of acylcarnitines is not simply a nonspecific feature of severe liver disease. Nearly half of the patients with normal analyses of bile had distinct and well-characterized liver diseases (tyrosinemia, Wilson disease, herpes simplex, and indeterminant disease with associated aplastic anemia). The relatively clear-cut etiologyof the liver disease in many of these patients makes an underlying defect in FAO less likely as an explanation for their liver injury.

Two different patterns of abnormal free and esterified carnitine in bile emerged from these analyses. The current analysis does not permit a complete understanding of these abnormal patterns, although several distinct possibilities exist. The first potential explanation is that these patients have specific inborn errors in FAO that have led to ALF. It seems reasonable to assume that patients with free and esterified carnitine levels greater than the 99th percentile are those most likely to have specific inborn errors. Unfortunately, many of these patients died, and follow-up genetic investigations have not been possible. The marked elevation in free carnitine in three patients is suggestive of a mitochondrial carnitine transport defect, as might be seen in carnitine palmitoyltransferase I deficiency.4 Those with elevations in medium- and long-chain acylcarnitines may have inborn errors in specific steps in the oxidation of medium- or long-chain fatty acids. A second explanation is that these patients have a genetic predisposition to abnormal FAO. The predisposition could be on the basis of single or compound synergistic heterozygosity for genetic defects or polymorphisms in FAO enzymes.16 Heterozygosity for FAO defects may be relatively common, with estimates of frequencies in excess of 1:40.4 Predisposition for an FAO defect could be unmasked by acute liver injury of any etiology. Accumulation of toxic fatty acid metabolites and interruption of oxidative phosphorylation could induce hepatocellular apoptosis or necrosis, thus worsening the original hepatic injury. The unexpectedly poor outcome (death) in the two patients with reported acetaminophen overdose might be explained by this potential modifier effect of an FAO gene heterozygote defect or polymorphism. The finding of steatosis in the livers of both of these patients supports the hypothesis of an underlying abnormality in FAO.14 Microvesicular steatosis is a common although not universal finding in FAO defects, but it is distinctly uncommon in acetaminophen toxicity.14, 17, 18 Interestingly, in one of the few other reported cases of steatosis in acetaminophen toxicity, the outcome was unexpectedly fatal.18 A third alternative is that these patterns in bile are markers of a specific type of liver injury possibly related to an unknown toxin.

In summary, the current analyses indicate that patients with abnormal acylcarnitine profiles have a worse overall prognosis. The three proposed pathophysiologies could explain the poorer prognosis. Future studies should be directed at prospectively confirming these findings in children and adults with ALF. Those studies should include acute analyses of serum and urinary acylcarnitines, assessment of fatty acid oxidation in cultured skin fibroblasts, and molecular analyses for genetic defects in fatty acid oxidation enzymes and transporters.