Unrecognized acetaminophen toxicity as a cause of indeterminate acute liver failure


  • Potential conflict of interest: Dr. Lee consults for Eli Lilly, Novartis, and Gilead. He receives grants from Bristol-Myers Squibb, Vertex, SPRI, Siemens, and GlobeImmune. Dr. Larson is on the speakers' bureau of HCV Nova and SNPA Meeting Solutions. Dr. James is part owner of the Acetaminophen Toxicity Diagnostics.


Despite extensive investigations, the cause of liver injury in 14% of patients with acute liver failure remains unknown (indeterminate). In a pilot study using a novel assay, highly specific acetaminophen-cysteine adducts were detected in 7 of 36 indeterminate patients (19%). To extend these observations, sera from 110 subjects enrolled in the Acute Liver Failure Study Group registry with indeterminate acute liver failure were analyzed with a similar but more efficient and sensitive adduct assay. As positive controls, another 199 patients with known or presumed acetaminophen-induced liver failure were assessed for the presence and quantity of adducts. Clinical, laboratory, and outcome data were compared for the two groups. On the basis of previous data for known therapeutic exposures and acetaminophen overdoses, an adduct concentration ≥1.0 nmol/mL of serum indicated a definite acetaminophen overdose. Among the 110 indeterminate cases, 18% had assay values ≥1.0 with a median level of 9.2 nmol/mL; 94.5% of the positive controls (known acetaminophen cases) had values ≥1.0 nmol/mL. Regardless of the initial diagnosis, subjects with elevated adduct levels demonstrated the clinical profile and hyperacute biochemical injury pattern associated with acetaminophen overdose: a predominance of female gender, very high aminotransferase levels, and low bilirubin levels. Conclusion: These data confirm and extend previous observations regarding the high (18%) prevalence of unrecognized or uncertain acetaminophen toxicity among subjects with indeterminate acute liver failure. N-Acetylcysteine use was limited in this group, presumably because of the lack of a specific diagnosis of acetaminophen toxicity. (HEPATOLOGY 2011;53:567-576.)

Acute liver failure (ALF) is a rare but life-threatening condition occurring in individuals without preexisting liver disease and is characterized by sudden severe liver dysfunction associated with coagulopathy (international normalized ratio >1.5) and hepatic encephalopathy.1, 2 Over the last decade, clinically defined acetaminophen (APAP) overdose has been the most common cause of adult ALF in the United States and has accounted for close to 50% of all cases (APAP is a dose-dependent hepatotoxin).2 The next most frequent cause of ALF, accounting for approximately 14% of the total, is unknown; that is, an identifiable etiology cannot be found.3 These indeterminate cases are classified as such after an extensive evaluation that includes a medical history, a physical examination, laboratory testing, and, specifically, the exclusion of clinically defined APAP overdose.4

Low doses of APAP are primarily metabolized by glucuronidation and sulfation. With higher doses, the conjugation pathways become saturated, and the metabolism of APAP is shunted to the cytochrome P450 system; this generates the highly reactive toxic metabolite N-acetyl-p-benzoquinone imine.5N-Acetyl-p-benzoquinone imine binds to hepatocellular proteins when glutathione stores, normally involved in detoxification, are depleted.6 The resulting acetaminophen-cysteine (APAP-CYS) protein adducts (APAP bound to cellular proteins via cysteine residues) can be quantified by a recently developed assay employing high-pressure liquid chromatography with electrochemical detection (HPLC-EC).7 Previous data from the mouse model of APAP toxicity by immunohistochemical approaches have demonstrated that adducts localize in the centrilobular hepatocytes of the liver and that these same cells subsequently lyse and release both adducts and hepatic aminotransferases in the serum.8, 9

Although the HPLC-EC assay is not yet available for clinical use, it allows for the detection of a highly specific biomarker of APAP hepatotoxicity in occult or late presentation ALF cases when the parent compound is at low or undetectable levels in the plasma.6 In the initial study, samples from 81 patients were analyzed in a blinded fashion: 36 cases with indeterminate ALF, 20 certain APAP cases, and 25 other disease controls for whom there was no history of APAP exposure. The assay correctly identified all cases of APAP toxicity and was negative for all patients with ALF due to other causes. In addition, adducts were detected in 7 of the 36 indeterminate cases (19%).6 Adducts have also been identified in 12.5% of pediatric patients with ALF of indeterminate etiology10 and may be detected for up to 12 days after clinically defined APAP overdose.11, 12

To confirm and extend our previous report regarding the detection of adducts in patients with ALF of indeterminate etiology,6 the following study was conducted in a larger cohort of patients with ALF.

 Abbreviations: ALF, acute liver failure; ALT, alanine aminotransferases; APAP, acetaminophen; APAP-CYS, acetaminophen-cysteine; AST, aspartate aminotransferases; HPLC-EC, high-pressure liquid chromatography with electrochemical detection; MELD, Model for End-Stage Liver Disease; NAC, N-acetylcysteine; NS, not significant; OLT, orthotopic liver transplantation.

Patients and Methods

Study Population.

The US Acute Liver Failure Study Group was established in 1998 as a consortium of liver centers interested in better defining the causes and outcomes of ALF. To date, 1431 subjects have been enrolled prospectively at 23 tertiary centers within the United States, all of which have liver transplantation programs. All enrolled subjects have met the standard criteria for ALF: the presence of coagulopathy (prothrombin time >15 seconds or international normalized ratio ≥1.5) and any degree of hepatic encephalopathy occurring within 26 weeks of the onset of first symptoms in a patient without previous underlying liver disease.1, 3 Because the subjects were encephalopathic by definition, written, informed consent was obtained from their legal next of kin. Detailed demographic, clinical, laboratory, and outcome data as well as daily sera for 7 days, a DNA sample, and tissue (when available) were collected prospectively. All centers were in compliance with their local institutional review board requirements. A certificate of confidentiality was obtained from the National Institutes for Mental Health for the entire study.

Assessment of Indeterminate and APAP-Induced ALF.

Each site's principal investigator, an academic hepatologist, was responsible for ascertaining the etiology of ALF by a careful historical review and an extensive clinical, radiographic, laboratory, and pathological evaluation (as needed) according to standard criteria.3 The etiology was considered indeterminate when these evaluations failed to indicate a defined cause and specifically when no clear evidence for clinically defined APAP overdose was found. During the period between 1998 and 2006, 1115 patients were enrolled; 158 of these patients were considered to have indeterminate ALF.

For all patients with any suspicion of APAP ingestion, a careful ingestion history was obtained; this included the total dose ingested, the specific APAP product taken, and the duration of use. For the 275 APAP cases included in the previous article describing the clinical features of APAP toxicity, an additional audit was performed by one investigator (A.M.L.).2 The criteria used in the previously published article for assigning APAP as the cause of ALF were as follows: (1) a history of potentially toxic APAP ingestion (i.e., >4 g of APAP/day, which is the maximum dose approved for use by the Food and Drug Administration on the package) within 7 days of presentation; (2) the detection of any level of APAP in the serum; or (3) a serum alanine aminotransferase (ALT) level ≥1000 IU/L with a history of APAP ingestion, regardless of the APAP level.

Severity Markers.

Hepatic comas were graded on a standard scale of I to IV.3 The Model for End-Stage Liver Disease (MELD) score and the King's College Hospital criteria for ALF (King's criteria) were used to assess the overall severity of illness upon enrollment.13, 14 For MELD scores, the etiology was assumed to be nonalcoholic and noncholestatic, laboratory tests less than 1.0 were corrected to 1.0, and serum creatinine values were assumed to be 4.0 mg/dL for all values >4.0 mg/dL or if the patient received dialysis, as described elsewhere.15 Outcomes (survival, death, and transplantation) were determined 3 weeks after study admission with a standardized case report form. Spontaneous survival was defined as survival at 3 weeks without transplantation. Overall survival was defined as survival at 3 weeks regardless of transplantation status. Patient management was based on the local standard of care, although general guidelines were employed by the US Acute Liver Failure Study Group sites.4 Liver transplant candidacy was determined at individual centers according to United Network for Organ Sharing guidelines.16

Measurement of Serum APAP Adduct Concentrations.

Among the 1115 cases in the registry, we identified 158 patients (14%) as having indeterminate ALF. Sera from study day 1 or 2 were available for 110 subjects (69.6%; groups 1 and 2; Fig. 1). From our previous clinically defined APAP overdose patient cohort,2 sera were available from study day 1 or 2 for 199 of 275 patients (72.4%; groups 3 and 4; Fig. 1). Sera collected at each site were aliquoted promptly, stored at −80°C, and shipped to a central storage site and then to the analytical laboratory of one investigator (L.P.J.). Serum samples (100-500 μL each) from 309 subjects (110 + 199; Fig. 1) were analyzed in a blinded fashion for APAP protein adducts with a previously reported method.7, 12 In brief, serum samples were treated with gel filtration to remove small molecules, including APAP and APAP metabolites. The resulting samples were treated with protease, and the resulting peptides were analyzed for APAP-CYS adducts by HPLC-EC.7, 12 The range of linearity for the method was 0.03 to 30.0 nmol of APAP-CYS adducts/mL of serum. The coefficients of variation for the assay were consistently <15% at APAP-CYS adduct concentrations of 0.03, 1.0, 6.0, and 30.0 nmol/mL. On the basis of the coefficients of variation of the standard curve for the assay, the lower limit of quantitation was defined as 0.03 μM APAP-CYS. In a recent study, a receiver operator curve analysis found that an adduct cutoff point ≥1.1 nmol/mL of serum in patients with a clinically defined APAP overdose and an ALT value >1000 IU/L had a sensitivity of 96.8% and a specificity of 95%.12

Figure 1.

Flow diagram of testing for patients diagnosed initially with either indeterminate or APAP ALF by assay results.

Statistical Methods.

Using the standard guidelines and appropriate controls established for the assay, we compared differences in the clinical characteristics of subjects with negative and positive levels of APAP adducts. Positive was used to designate samples with levels ≥1 nmol/mL, whereas negative was used to designate samples with adduct levels <1 nmol/mL. To compare continuous variables among the groups, we used the one-way analysis of variance unless Bartlett's test showed unequal variances; then, the Kruskal-Wallis test was used. To compare categorical variables among the groups, we used the χ2 test or Fisher's exact test. Once a global P value was ascertained (which indicated that a difference existed), comparisons were made between two groups with the Student t test, Mann-Whitney U test, or χ2 test as appropriate. For all analyses, a two-tailed P value ≤0.05 was considered significant and was adjusted for multiple two-way comparisons.


Study Subjects.

The overall results of the APAP-CYS adduct assay are presented in Fig. 1 and Table 1. Twenty of the 110 subjects (18%) with indeterminate ALF had levels of APAP-CYS adducts suggesting that ALF was secondary to an APAP overdose. For the purposes of the present study, results of adduct assays ≥1.0 nmol/mL were defined as positive, whereas adduct assay results <1.0 nmol/mL were defined as negative; this signified that the adduct values were below levels previously identified as being associated with a clinically defined APAP overdose and the development of an ALT level >1000 IU/L.12 Using the previously defined cutoff, we divided the overall indeterminate group into two groups: 90 patients with levels <1.0 nmol/mL (group 1) and 20 patients demonstrating significant quantities of APAP adducts (group 2). The median APAP-CYS concentration was 9.2 nmol/mL in group 2 versus 0.0 nmol/mL in group 1; for the latter, all values were below the threshold (Table 1). Among the 199 patients with a clinically defined APAP overdose whose sera were analyzed, 188 had values above the threshold for a toxic exposure (group 3), whereas 11 had values below the toxicity threshold (group 4). Thus, 94.5% of the patients with a clinically defined APAP overdose had APAP-CYS concentrations that were above the threshold for toxicity. The median APAP-CYS concentration was 11.1 nmol/mL for the 188 APAP-CYS–positive patients (group 3) and 0.3 nmol/mL for the 11 APAP-CYS–negative patients (group 4).

Table 1. Baseline Features of Patients With Indeterminate ALF and Patients With Clinically Defined APAP Overdose ALF Stratified by Their Assay Status
CharacteristicGroup 1: Indeterminate, Assay-Negative (n = 90)Group 2: Indeterminate, Assay-Positive (n = 20)*Group 3: APAP, Assay-Positive (n = 188)*Group 4: APAP, Assay-Negative (n = 11)Global P ValueGroup Comparison
  • Abbreviation: NS, not significant (P > 0.05)

  • *

    Assay-positive is defined as a level ≥ 1.0 nmol/mL.

  • Statistically significant two-way group comparisons are shown with adjusted P values < 0.05; the adjusted P value is the actual P value times the number of comparisons made (six).

 Age, years    <0.025NS
  25th and 75th percentiles27, 5123.5, 38.530, 4537, 43  
 Female, n (%)43 (47.8)16 (80.0)140 (74.5)6 (54.5)<0.0011 versus 3
Laboratory markers
 ALT, IU/L    <0.0011 versus 2
  Median811515640251810 1 versus 3
  25th and 75th percentiles335, 14053463, 81402422, 6680621, 9386  
 AST, IU/L    <0.0011 versus 2
  Median691812643331808 1 versus 3
  25th and 75th percentiles342, 13145162, 12,2952027, 8192277, 10,860  
 Bilirubin, mg/dL    <0.0011 versus 2
  Median24. 1 versus 3
  25th and 75th percentiles17.6, 31.72.5, 6.53.0, 5.91.4, 11.1 1 versus 4
 APAP assay, nmol/mL    <0.0011 versus 2
  Median09.211.10.25 1 versus 3
  25th and 75th percentiles0, 0.44.4, 14.36.7, 18.60.2, 0.3 1 versus 4
      2 versus 4
      3 versus 4
Severity markers
 MELD score    0.0141 versus 3
  Standard deviation9.26.89.611.0  
 APAP King's criteria met, n (%)9 (10.0)5 (25.0)29 (15.4)1 (9.1)NS 
 Hepatic coma grade, n (%)    NS 
  119 (22.6)3 (16.7)46 (25.3)2 (20.0)  
  226 (31.0)3 (16.7)43 (23.6)1 (10.0)  
  314 (16.7)6 (33.3)42 (23.1)4 (40.0)  
  425 (29.8)6 (33.3)51 (28.0)3 (30.0)  
Any NAC treatment, n (%)16 (17.8)8 (40.0)177 (94.1)9 (81.8)<0.0011 versus 31 versus 42 versus 3
Outcomes, n (%)
 Spontaneous survival19 (21.1)11 (55.0)120 (63.8)5 (45.5)<0.0011 versus 21 versus 3
 Liver transplantation39 (42.2)2 (10.0)15 (8.0)2 (18.2)<0.0011 versus 21 versus 3

Baseline Characteristics: Comparison of the Four Groups.

The 20 APAP-CYS–positive patients with indeterminate ALF (group 2) were remarkably similar to the 188 APAP-CYS–positive control patients (group 3) in demographic, clinical, and laboratory markers of disease. The APAP-CYS–positive patients were younger than their APAP-CYS–negative counterparts in both the indeterminate cohort (median age of 32.5 years for group 1 versus 39 years for group 2) and the APAP control cohort (median age of 37 years for group 3 versus 42 years for group 4), although these differences were not statistically significant. The APAP-CYS–positive patients were predominantly female (group 2, 80%; group 3, 74.5%) in comparison with their APAP-CYS–negative counterparts in both the indeterminate cohort (48%; group 1) and control cohort (54.5%; group 4). Groups 2 and 3 had median ALT values >4000 IU/L (4025 and 5156 IU/L, respectively), whereas the median ALT value for group 1 was much lower (811 IU/L, P< 0.05). The same was true for aspartate aminotransferase (AST) values, with very high levels of AST observed in groups 2 and 3 and significantly lower levels observed in group 1. Bilirubin concentrations were much lower in both APAP-CYS–positive groups (group 2, median = 5.05 mg/dL; group 3, median = 4.1 mg/dL) versus group 1 (median = 24.3 mg/dL, P< 0.05). Group 4, the APAP-CYS–negative control group, also had a lower median bilirubin concentration (7.3 mg/dL, P< 0.05) in comparison with group 1, but it was similar to the values for groups 2 and 3. Of the 20 APAP-CYS–positive patients who were indeterminate (group 2), 19 (95%) had an ALT level >1000 IU/L [17 (85%) had an ALT level >3000 IU/L], and 18 (90%) had a bilirubin level <10 mg/dL. Similarly, of the 188 APAP-CYS–positive controls (group 3), 175 (93%) had an ALT level >1000 IU/L [117 (62%) had an ALT level >3000 IU/L], and 175 (93%) had a bilirubin level <10 mg/dL.

A retrospective review of the 20 adduct-positive indeterminate cases disclosed that the circumstantial evidence favored APAP in some cases. Twelve of the 20 patients had this evidence suggesting APAP as the cause, so they were considered ill-defined APAP overdose patients by the site's principal investigator. In hindsight, a clinically defined APAP overdose seemed a possible or even likely diagnosis. In each of these cases, there were high aminotransferase levels (all >1500 IU/L) and either a low but measurable APAP level or a history of some APAP-containing medications. Interestingly, all but 2 of the 12 patients were graded as coma grade II or higher. Seven of the eight remaining patients had negative APAP levels; in five patients, there was no history supporting any APAP ingestion. Thus, these eight were clinically unrecognized APAP patients. In two of the eight patients, there was retrospective support for an APAP overdose (a previous APAP overdose denied by the patient or a report of an APAP overdose in the autopsy report but not in the case report form).

A retrospective review of the 11 patients considered to be APAP overdose cases but not confirmed by elevated adduct levels (group 4) disclosed that in 6 cases, there was late presentation to the medical facility (5-14 days after ingestion or the onset of symptoms). Three of these six late presenters had low ALT levels (<1000 IU/L), whereas one of the five early presenters had low ALT levels. Among the remaining four, three had high aminotransferase levels, strong histories, and measurable or high APAP levels, whereas the final patient was probably diagnosed in error on the basis of high enzyme levels. He had no obtainable history and had a known seizure disorder but had been found unconscious. There was no APAP level, aminotransferase levels were higher than 3000 IU/L (AST level = 10,860 IU/L, ALT level = 3779 IU/L), and he was more likely suffering from status epilepticus with ischemic injury and myonecrosis. Repeat adduct analysis was performed in these 11 cases with separate serum samples, and all repeat assays showed levels <1.0 nmol/mL.

The clinical and laboratory data for the 48 indeterminate patients who were excluded because their sera were not available for APAP-CYS adduct testing were analyzed, and they resembled the remaining group 1 patients. Briefly, 67% were female, the median age was 34 years, and the biochemical profile showed a high median bilirubin concentration (24.6 mg/dL) and low median levels for AST (780 IU/L) and ALT (649 IU/L).

Severity Markers and Clinical Outcomes.

Further analysis was conducted to compare physiological scores and patient outcomes between the four groups. The average MELD score, an established marker of liver failure severity, was statistically similar between groups 2 and 3 (P = 0.56). The hepatic coma grade on admission and the percentage of patients who met King's criteria were similar across all four groups. Outcomes were similar between the two APAP-CYS–positive groups (Table 1), with similar short-term spontaneous survival rates (55% versus 63.8%) and liver transplantation rates (10% versus 8%).

The mean MELD score for the 90 non-APAP group 1 patients (those who were assay-negative and below the threshold) was higher than the score for group 3, the clinically defined APAP overdose control group (35.5 versus 31.7, P< 0.014), largely because of the differences in bilirubin concentrations between the groups (higher in group 1; Table 1). This difference was not due to larger sample sizes because the standard deviations (or interquartile ranges) were larger than group 2.

Likewise, group 1 had a 4- to 5-fold higher liver transplantation rate (42%) in comparison with groups 2 and 3 (P< 0.05 for both comparisons). The spontaneous (transplant-free) survival rate for group 1 was significantly lower (21%) than the rates for groups 2 and 3 (63% and 55%, respectively, P< 0.05 for both comparisons).

N-Acetylcysteine (NAC) Treatment.

The percentage of patients who received NAC differed across the four patient groups (Table 2). Patients from group 3 with a clinically defined APAP overdose received NAC 94.1% of the time versus only 40% for the adduct-positive indeterminate patients (group 2). Still, the 40% level of NAC use for this group was considerably higher than the 17.8% level of use recorded for group 1, the non-APAP or adduct-negative, indeterminate group. The level of NAC use for group 4 was 81.8%, as befits patients with a clinically defined APAP overdose. Although 100% use of NAC would be ideal, this series represents patients developing liver failure and not the overall clinical spectrum of APAP overdoses. Thus, the patients in this group represent the most severe cases and, in some cases, failure of the initial treating physician to begin NAC in a timely fashion. Spontaneous survival was slightly lower in group 2 than group 3 (55% versus 63.8%), but this was not significantly different. In the present patient series, 24 indeterminate patients were enrolled in a double-blind, placebo-controlled trial of its use in non-APAP ALF patients, with 10 receiving NAC and 14 receiving a placebo (Table 2). Another 14 patients had received NAC at some point in their care, though not within the aegis of the clinical trial, because the trial excluded participation if there had been previous NAC use. Previous NAC treatment outside the trial took place in 5 of 14 members of group 2 (35.7%) and in 9 of 72 members of group 1 (12.5%; Table 2); NAC-treated patients, regardless of how treatment was provided, showed roughly comparable outcomes (data not shown).

Table 2. Comparison of Severity Markers and Outcomes Between the Four Groups Stratified by the Use of NAC
CharacteristicGroup 1: Indeterminate, Assay-Negative (n = 90)Group 2: Indeterminate, Assay-Positive (n = 20)*Group 3: APAP, Assay-Positive (n = 188)*Group 4: APAP, Assay-Negative (n = 11)
  • *

    Assay-positive is defined as a level ≥ 1.0 nmol/mL.

  • Received a placebo.

Any NAC treatmentYesNoYesNoYesNoYesNo
Overall NAC treatment, n (%)16 (17.8)74 (82.2)8 (40.0)12 (60.0)177 (94.1)11 (5.9)9 (81.8)2 (18.2)
 Controlled treatment (within a randomized controlled trial), n711330000
 Before NAC treatment, n963591771192
Severity markers
 MELD score
  Standard deviation10.
 ALT ≥ 1000 IU/L, n (%)9 (56.3)26 (35.1)8 (100)11 (91.7)164 (92.7)11 (100)6 (66.7)1 (50.0)
Outcomes, n (%)
 Spontaneous survival6 (37.5)13 (17.6)6 (75.0)5 (41.7)113 (63.8)7 (63.6)4 (44.4)1 (50.0)
 Liver transplantation6 (37.5)33 (44.6)02 (16.7)14 (7.9)1 (9.1)2 (22.2)0
 Overall 3-week survival11 (68.8)45 (60.8)6 (75.0)5 (41.7)122 (68.9)8 (72.7)6 (66.7)1 (50.0)


The present study confirms and extends the previous findings of our earlier report6 regarding the use of the APAP protein adduct assay in the diagnostic evaluation of patients with ALF. In this larger data set with a refined assay method, APAP protein adducts were detected in 95% of patients with a clinically defined APAP overdose. In addition, nearly one-fifth of patients with ALF classified as indeterminate showed evidence of APAP toxicity not identified by experienced investigators using current diagnostic techniques. Reasons for failing to diagnose cases included the presence of hepatic encephalopathy on admission, possible deception by patients, or their failure to recognize and report excessive dosing of this readily available product. In some instances, a vague history of APAP ingestion had been provided by the patient or family but could not be confirmed. The clinical features observed in the ill-defined APAP overdose patients and the clinically unrecognized APAP patients (collectively called group 2) were remarkably similar in demographics, laboratory values, and outcomes to the ALF cases with a clinically defined APAP overdose, and perhaps they should have been recognized on that basis. Using the clinical characteristics as criteria for APAP toxicity in patients with unclear APAP histories may allow earlier identification of these cases in the absence of an available APAP adduct assay. This study was not intended to validate the previously published criteria assigning APAP as the cause of ALF. Using those criteria in an emergency setting may lead to an overdiagnosis of APAP-induced hepatotoxicity and, therefore, APAP-induced ALF. However, the aim of those criteria is to be more sensitive than specific in the detection of APAP-induced hepatotoxicity.

Patients with unrecognized APAP hepatotoxicity received NAC treatment much less often than patients with a clinically defined APAP overdose, as might be expected. This withholding of care would likely have affected outcomes if larger numbers of patients had been included. Withholding NAC in the future may be less likely to occur because of recent evidence that NAC is of value in the setting of non-APAP ALF.17 Recognition of APAP overdosing is important for other reasons: this knowledge may affect the decision to continue NAC and decisions concerning transplant candidacy, prognosis, referral for psychiatric counseling, and educational intervention for unintentional cases. Although NAC is being increasingly used for non-APAP cases, the practice is not necessarily widespread among all practitioners. Finally, having a better diagnostic test for this condition should provide clinicians with better epidemiological data and enhance future education and prevention efforts.

The pattern of hyperacute liver injury (low bilirubin and high ALT) was almost exclusively confined to the two adduct-positive groups, and this supports the short-duration, hyperacute illness pattern associated with APAP toxicity. However, this hyperacute liver injury pattern also occurs in ischemic liver injury after a decrease in cardiac output for any reason. In the absence of a readily available adduct assay, clinicians should consider the hyperacute pattern with short-duration illness, high aminotransferase levels, and low bilirubin levels as likely indicating APAP toxicity (or possible ischemic hepatitis). In contrast, the patients of group 4, who suffered a clinically defined APAP overdose but had adducts below the toxicity threshold, had somewhat atypical APAP features with a heterogeneous demographic, clinical, biochemical, and outcome profile (Table 3). A dose history was not available for 6 of the 11 patients; 6 of the 11 patients had a history of ethanol use, but it was unclear. Six of the 11 patients had late presentations to medical centers (more than 7 days after the onset of illness). Although every attempt was made to use day 1 or 2 study samples for this analysis, approximately 30% of patients in the overall analysis had study samples from day 3 or later. Because of the limited sample size, no clear pattern could best identify this subgroup, its prognosis, or the reason that these 11 patients were negative (or below the threshold) in the adduct assay, although clearance of adducts is possible in patients presenting late.12 Another less likely possibility is that NAC treatment could have affected the assay results, although nearly all members of group 3 with a clinically defined APAP overdose had NAC treatment (95%) and were positive for adducts. Further studies are needed to truly understand the impact of NAC on the interpretation of the APAP adduct assay via a prospective analysis or possibly animal models. It is important to note that the previous receiver operating characteristic analysis (and the generation of sensitivity and specificity parameters) of adduct levels in patients with a clinically defined APAP overdose was anchored to patients with an ALT level >1000 IU/L. Thus, nontoxic levels of adducts or lower levels of adducts in patients with an ALT level <1000 IU/L are not unexpected.12

Table 3. Summary Features for 11 Group 4 Patients (Negative APAP Adduct Assay)
  • Those patients who did not die or undergo transplantation survived without transplantation. Dashes indicate that information was not available.

  • Abbreviation: OLT, orthotopic liver transplantation.

  • *

    The subsequent assessment was probable autoimmune hepatitis leading to ALF.

Age, years2876414043524333433742
Overdose typeSuicidalAccidental/ unintentionalSuicidalUnknownAccidental/ unintentionalAccidental/ unintentionalSuicidalSuicidalSuicidalAccidental/ unintentionalSuicidal
APAP dose, g408431.57519.5
APAP level, μg/mL01861000310232
Alcohol useNoNoYesYesYesNoYesYesNoYes
Symptom-to-coma duration, days5431114148550
Late presenterYesNoNoNoNoYesYesYesYesYesNo
ALT, IU/L115457613,11037799386629136181062140309810
AST, IU/L485180818,71010,86017,78086212917027738996910
Bilirubin, mg/dL7.81.4100.70.734.622.911.17.364.8
APAP adducts, nmol/mL0.2070.6320.2460.3420.2940.1570.0100.3350.4460.0190.244
MELD score30.210.641.934.929.146.233.538.119.244.941.6
APAP King's criteria metNoNoNoNoNoNoNoNoNoNoYes
Hepatic coma grade4214331334
Any NAC treatmentYesYesYesYesYesYesNoNoYesYesYes
OLT, days     3    5
Death, days5289   5    

The significant differences in the magnitude of APAP protein adduct levels between clinically defined APAP overdoses and regular therapeutic use of APAP were recognized in two previous studies. We previously reported the pharmacokinetic profile of adducts in 18 adults with a clinically defined APAP overdose and found the mean elimination half-life to be 1.73 days.12 In this analysis, adduct values were plotted with respect to the day of overdose, so adduct levels on day 3 of the overdose were in the 7 to 9 nmol/mL range; on days 8 to 10 of the overdose, the levels were closer to the 1.0 nmol/mL cutoff point. In some patients, levels were <1.0 nmol/mL on days 8 to 10. In a further study, adduct formation was examined in healthy adults receiving APAP (4 g/day) in an inpatient clinical study setting.18 Low levels of adducts were detected in these patients; however, the mean maximum plasma concentration for adducts was 0.3 nmol/mL, a value that was approximately 2 orders of magnitude below the levels observed in the early stages of very serious cases of clinically defined APAP overdose. Thus, very low levels in the late stages of toxicity cannot be distinguished from levels that would be detected in patients receiving therapeutic doses, but the peak of symptoms and the accompanying hepatic aminotransferase elevation occur early (2-4 days) in overdose patients.

Outcomes for those unrecognized adduct-positive patients who did not receive the NAC antidote were poorer (5/12, 42%) than the outcomes for those who received NAC (6/8, 75%; Table 2); both groups, however, were relatively small. Taken together, our results indicate that NAC improved short-term spontaneous survival in both adduct-positive and adduct-negative indeterminate groups.

The present study suggests that all patients presenting with rapid-onset ALF for which an etiology cannot be determined should be considered potential APAP cases. In the absence of a readily available adduct assay, clinicians should consider the hyperacute pattern with short-duration illness, high aminotransferase levels, and low bilirubin levels as likely indicating APAP toxicity. Rapid institution of NAC is indicated in most circumstances because its toxicity is low, its value is well established in APAP toxicity, and it appears useful for non-APAP cases as well.

The cause of the remaining 82% of indeterminate cases remains obscure. Other studies from our group have thus far failed to identify either new viruses or toxins involved in the indeterminate group.19-22 Nevertheless, use of the adduct assay, when it becomes available, should lead to the earlier detection of APAP hepatotoxicity and thus facilitate more aggressive NAC treatment: the use of NAC on a presumptive basis in the absence of confirmed toxicity should not be discouraged.

In conclusion, indeterminate ALF accounts for approximately 14% of all ALF cases referred to tertiary centers. However, this group includes nearly 18% suffering from unrecognized APAP toxicity, as demonstrated by the presence of high levels of APAP-CYS adducts in serum and by clinical and biochemical profiles virtually identical to those of known APAP cases. The fact that 18% of APAP hepatotoxicity cases were missed by experienced clinicians is of concern for point-of-care physicians and tertiary academic hepatologists. A lack of early recognition of APAP hepatotoxicity may indicate that many physicians as well as patients are uninformed about the ubiquity of the APAP problem. Because of its frequency, practitioners should consider readily the diagnosis of APAP hepatotoxicity in the proper setting even when an adequate history of ingestion is unavailable. Moreover, clinicians should have a low threshold for using NAC in the setting of hyperacute liver failure because adducts have been found in significant levels even in the absence of obvious clinically defined APAP toxicity. For the remaining 80% of indeterminate patients with no evidence of APAP hepatotoxicity, spontaneous survival is considerably less than that of APAP cases. The subacute disease pattern includes lower aminotransferase concentrations, higher bilirubin concentrations, and a higher rate of transplantation; this group may benefit from NAC as well. Prompt referral of all cases of potential ALF, regardless of etiology, should be made to centers at which liver transplantation is available.


The Acute Liver Failure Study Group (1998-2006) comprised the following investigators and coordinators, who worked tirelessly in support of this study: William M. Lee (principal investigator), Anne M. Larson, Carla Pezzia, Kruti Joshi, Nahid Attar, and Corron Sanders (University of Texas Southwestern Medical Center, Dallas, TX); Oren K. Fix (University of Washington, Seattle, WA); Timothy J. Davern and Kristine Partovi (University of California San Francisco, San Francisco, CA); Lawrence U. Liu and Manuela Tiangco-Zuniga (Mount Sinai Medical Center, New York, NY); Timothy M. McCashland and Tamara Bernard (University of Nebraska, Omaha, NE); J. Eileen Hay and Cindy Groettum (Mayo Clinic, Rochester, MN); Natalie G. Murray and Sonnya Coultrup (Baylor University Medical Center, Dallas, TX); A. Obaid S. Shaikh, Linda Gooch, and Diane Morton (University of Pittsburgh Medical Center, Pittsburgh, PA); Andres T. Blei, Daniel R. Ganger, and Jeanne Gottstein (Northwestern University Medical School, Chicago, IL); Atif Zaman, Jonathan M. Schwartz, Kenneth Ingram, Willscott E. Naugler, Harlene Finn, and Suni Wilson (Oregon Health and Science University, Portland, OR); Steven Han and Val Peacock (University of California Los Angeles, Los Angeles, CA); Robert J. Fontana and Suzanne Welch (University of Michigan Medical Center, Ann Arbor, MI); Michael Schilsky and Noelle Sowers (Yale University School of Medicine, New Haven, CT); Brendan M. McGuire, Stacy Eddleman, and Dorothy Faulk (University of Alabama, Birmingham, AL); Raymond T. Chung, Anna Rutherford, and Michael Chen (Massachusetts General Hospital, Boston, MA); Robert S. Brown Jr, Jonathan Kim, and Rudi Odeh-Ramadan (Columbia–Presbyterian Medical Center/Cornell–New York Hospital, New York, NY); Adrian Reuben and Stacey Minshall (Medical University of South Carolina, Charleston, SC); Santiago Munoz, Andres Riera, Stacey Carmody, and Victoria Rudzik (Albert Einstein Medical Center, Philadelphia, PA); K. Rajender Reddy, Mical Campbell, Mary Hammond, Wojciech Blonksi, and Kimberley Kime (University of Pennsylvania, Philadelphia, PA); Todd Stravitz and Melanie White (Virginia Commonwealth University, Richmond, VA); Lorenzo Rossaro, Laura Lester, Monica Ruiz, and Yulia Suprun (University of California Davis, Sacramento, CA); Raj Satyanarayana, David Kramer, and Dana Kontras (Mayo Clinic, Jacksonville, FL); and Tarek Hassenein, Fatma Barakat, and Lita Petcharaporn (University of California San Diego, San Diego, CA).