1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Although rare instances of cardiac injury or arrhythmias have been reported in acute liver failure (ALF), overall, the heart is considered to be spared in this condition. Troponin I, a sensitive and specific marker of myocardial injury, may be elevated in patients with sepsis and acute stroke without underlying acute coronary syndrome, indicating unrecognized cardiac injury in these settings. We sought to determine whether subclinical cardiac injury might also occur in acute liver failure. Serum troponin I levels were measured in 187 patients enrolled in the US Acute Liver Failure Study Group registry, and correlated with clinical variables and outcomes. Diagnoses were representative of the larger group of >1000 patients thus far enrolled and included 80 with acetaminophen-related injury, 26 with viral hepatitis, 19 with ischemic injury, and 62 others. Overall, 74% of patients had elevated troponin I levels (>0.1 ng/ml). Patients with elevated troponin I levels were more likely to have advanced hepatic coma (grades III or IV) or to die (for troponin I levels >0.1 ng/ml, odds ratio 3.88 and 4.69 for advanced coma or death, respectively). Conclusion: In acute liver failure, subclinical myocardial injury appears to occur more commonly than has been recognized, and its pathogenesis in the context of acute liver failure is unclear. Elevated troponin levels are associated with a significant increase in morbidity and mortality. Measurement of troponin I levels may be helpful in patients with acute liver failure, to detect unrecognized myocardial damage and as a marker of unfavorable outcome. (HEPATOLOGY 2007;45:1489–1495.)

Acute liver failure (ALF) is a rare condition consisting of rapid-onset severe liver injury accompanied by coagulopathy and altered mental status. Approximately 2000 cases per year occur in the United States resulting in liver transplantation or death in more than 35% of these cases, frequently due to multiorgan failure (MOF).1

Acetaminophen (also known as APAP), a dose-related toxin that causes centrilobular hepatic necrosis and renal tubular damage, is currently the most common cause of ALF in the United States, accounting for nearly 50% of all patients.2 Cardiac problems are relatively infrequent in ALF except for a few case reports of histologic injury and arrhythmias in association with acetaminophen overdose.3–9 Although cardiac injury might result from a direct toxic effect of acetaminophen, subclinical heart disease in this setting could also represent a component of the MOF syndrome associated with ALF. Rarely have associations of cardiac injury been made in acute liver failure not related to acetaminophen.

Troponin I is a well-established, specific, and sensitive marker of myocardial injury, with both diagnostic and prognostic value. It permits early identification of patients at increased risk of death from acute coronary syndrome.10–12 However, recent evidence suggests that patients with certain conditions such as sepsis or acute stroke may also demonstrate elevated troponin I levels under certain circumstances in the absence of an acute coronary ischemic event, indicating that unrecognized myocardial injury may be occurring during other acute disease processes.12, 13 In such studies, patients demonstrating elevated troponin levels without acute coronary syndrome had a poorer prognosis than those with normal troponin levels.14

The US Acute Liver Failure Study Group has collected detailed prospective information and serum, DNA, and tissue samples from more than 1000 patients with ALF. To date, no studies have evaluated the prevalence of elevated troponin levels in a broad range of ALF patients. The aim of this exploratory study was to determine whether serum troponin I levels were elevated in patients with ALF and whether there appeared to be clinical implications related to these changes, such as new prognostic information that could guide transplantation decisions.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

The US Acute Liver Failure Study Group was established in 1997 as a consortium of liver centers interested in better defining the causes and outcomes of ALF. To date, 1038 patients have been enrolled at 25 tertiary centers around the United States, all but one of which has a liver transplantation program. All enrolled patients met standard entry criteria for ALF: 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 first symptoms in a patient without previous underlying liver disease).1 After informed consent was obtained from the patients' next of kin in accordance with guidelines of local institutional review boards, detailed demographic, clinical, laboratory, and outcome data as well as daily sera for 7 days, a DNA sample, and tissue (when available) were collected in prospective fashion. Detailed case report forms were completed at the site and reviewed by the central site with periodic monitoring visits. Logs of patients not meeting enrollment criteria were kept at the sites. Generally, these failures were due to inability to obtain consent from next of kin or failure to meet clinical criteria for severity of disease. Cardiac events were not specifically tracked except that the presence or absence of arrhythmias was noted on the case report forms. No patients were listed as having a primary cardiac cause of death, although nearly 40% of those who died were considered to have MOF as the cause.1

Among 1038 patients enrolled in the US ALF registry, >80% had available serum specimens. For the present sub-study, we selected samples from 168 consecutively enrolled patients from May 1998 to August 2000. We supplemented this group with 14 additional nonconsecutive samples from patients with ischemic liver injury (to give a total of 19 ischemic cases) to evaluate whether this category might contain a higher number of patients with positive troponin I levels and thus serve as a positive control. With this one exception, the sample cohort was representative of the overall diagnostic categories observed in this study.1 Sample size in this study reflected the number of patient samples comprised in 2 ELISA kits for troponin I. Diagnoses for each patient had been established by site investigators who used standard criteria across sites as defined in the study's Manual of Operations; diagnoses were confirmed by review at the central site.1 Hepatic coma was graded on a standard scale of I to IV, as described.1 An indeterminate etiology was considered when no known cause could be found after extensive clinical, radiographic, and laboratory evaluation. Patient management was based on clinical guidelines at each institution. Liver transplant candidacy was determined at individual centers based on United Network of Organ Sharing guidelines. Outcomes were determined 3 weeks after study admission. Table 1 characterizes the etiologies, laboratory data, and outcomes observed.

Table 1. Clinical and Laboratory Characteristics of the Patient Groups
VariableOverallTroponin I <0.1Troponin I ≥0.1P value
  • *

    Mann-Whitney test.

  • χ2 test.

  • Fisher's Exact Test.

  • §

    Etiology of heart disease included 3 with atrial fibrillation, 3 with coronary artery disease, 3 with left ventricular dysfunction, and 2 unknown.

Age (years), Median (range)39 (15–81)39 (17–75)39 (15–81)0.3061*
Gender female (%)122 (65%)35 (71%)87 (63%)0.2897
Previous health and illnesses    
 Hypertension, n (%)19 (10%)1 (2%)18 (13%)0.0285
 §Heart disease, n (%)11 (6%)2 (4%)9 (7%)0.5329
 Renal disease, n (%)10 (5%)1 (2%)9 (7)0.2310
 Endocrine/diabetes, n (%)21 (11%)3 (6%)18 (13%)0.1875
 Chronic liver disease, n (%)3 (2%)1 (2%)2 (1%)>0.9999
 APAP802753 (66%)0.1947
 Ischemia/shock19316 (84%) 
 Viral hepatitis A12111 (92%) 
 Viral hepatitis B14212 (86%) 
 Indeterminate41932 (78%) 
 Other dx21714 (67%) 
 Admission MAP, median (range)86.7 (33.7–133.7)83.5 (62.0–114.7)88.3 (33.7–133.7)0.3292*
Admission Labs    
 INR, median (range)2.7 (1.2–26.1)2.5 (1.4–10.4)2.85 (1.2–26.1)0.1010*
 CK (IU/l), median (range)249 (0–21000)105 (2–2694)269 (0–21000)0.0906*
 HCO3 (mEq/dl), median (range)22 (6–35)24 (13–31)21 (6–35)0.0114*
 Creatinine (mg/dl), median (range)1.6 (0.4–10)1.15 (0.4–10)1.9 (0.4–7.3)0.0021*
 ALT, median (range)2227.5 (29–18079)2002.5 (29–12700)2341 (127–18079)0.7000*

Measurement of Serum Troponin I Levels.

Sera were collected, stored at −80°C, and shipped to the central site prior to use in the study. Troponin I levels were measured from sera collected on day of admission, via the 2-site immunoassay with direct chemi-luminometric technology (ADVIA Centaur, Bayer Diagnostics, Tarrytown, NY). Negative controls were included in each run. The European Society of Cardiology and The American College of Cardiology recommend using the 99th percentile as a cut-off value for troponin assays, above which any value is considered abnormal. For the ADVIA Centaur assay, a value ≥0.1 ng/ml is considered an elevated level, representing a positive test indicative of myocardial injury.10 The lower limit of detection for this assay is 0.02 ng/ml. The coefficient of variation was 0.05.


Glomerular filtration rate (GFR) was approximated by using estimated creatinine clearance calculated by using the Crockroft-Gault formula. A GFR below 60 ml/minute was considered indicative of moderate renal impairment. Acute Physiology, Age and Chronic Health Evaluation (APACHE II) score was calculated using 12 common physiological and laboratory values (temperature, mean arterial pressure, heart rate, respiratory rate, oxygenation, arterial pH, serum sodium, serum potassium, serum creatinine, hematocrit, white blood cell count, and Glasgow coma scale). The calculation of APACHE II and GFR was based on values recorded on day of admission.

Statistical Methods.

Using the standard guidelines and appropriate controls established for the assay, troponin I cutoff values were first considered positive when ≥0.1 ng/ml; we compared differences in clinical characteristics in patients with <0.1 ng/ml versus ≥0.1 ng/ml. Analysis was performed using the chi-squared or Fisher's exact tests, as appropriate, and the Mann-Whitney U test for continuous variables. A second set of cutoff values examined a “dose response” effect of troponin I with 5 groups defined by the following cutoff values (all as nanograms per milliliter): <0.1, 0.1 to <0.3, 0.3 to <1.0, 1.0 to <3.0, and 3.0 or greater. The Jonckheere-Terpstra Test was used to examine ordered differences among the different troponin I groups for death, nonspontaneous survival (death or transplant), advanced coma grade (III-IV) versus earlier grade (I-II), and the presence of cardiac arrhythmias within the first 7 study days. Admission creatinine levels for the 5 troponin I groups were compared using the Kruskal Wallis test, and if significant, pair-wise comparisons between average ranks were performed using the nonparametric multiple comparison tests (Dunn Method). All statistical analyses were performed using SPSS (version 14.0; SPSS Inc., Chicago, IL) and SAS 9.1 (SAS Institute Inc., Cary, NC).


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Among the 187 patients whose sera were tested were 80 with acetaminophen-related injury, 26 with viral hepatitis (VH), 19 with ischemic hepatic injury (IH), and 62 others. Overall, 74% of the admission serum samples tested were positive at a troponin I cut-off >0.1 ng/ml, with a range of 0 to >50 ng/ml. We initially evaluated the association between elevated troponin I level on clinical and laboratory parameters, as shown in Table 1.

Clinical Features in Relation to Elevated Troponin Levels.

There were no differences in age or in gender between the groups with normal and elevated troponin I levels, nor were there differences in any feature of past health or illnesses except that patients with hypertension were more likely to demonstrate elevated troponin I levels (P < 0.03). No differences were observed in mean arterial pressure (MAP) on admission between the groups, although it was noted the patients with elevated troponin I levels had higher initial MAP levels. Overall, 22.4% (11 of 49 patients) in the normal troponin group (< 0.1 ng/ml) versus 52.9% (73 of 138 patients) in the elevated troponin I (> 0.1 ng/ml) group had more advanced coma grades (III or IV) on admission to study (P = 0.0002; Table 2). This yielded an odds ratio (OR) of 3.88 (95% confidence interval [CI]: 1.83 to 8.21) for those in the elevated troponin group to also have a higher coma grade. In addition, patients with troponin I levels > 0.1 ng/ml had a trend toward more arrhythmias, with an odds ratio of 2.08 (95% CI: 0.96 to 4.53; P = 0.65).

Table 2. Odds Ratios (OR) for Outcome Variables, Coma Grade 3-4 and Arrhythmia
VariablesTroponin IOdds Ratio (OR)P value95.0% C.I. for OR
<0.1 (n = 49)≥0.1 (n = 138)LowerUpper
Dead5 (10.2%)48 (34.8%)4.690.00221.7512.62
Transplanted8 (16.3%)27 (19.6%)1.250.61800.522.97
Coma grade III-IV at admission to study11 (22.4%)73 (52.9%)3.880.00041.838.21
Arrhythmia during the first 7 days after admission to study10 (20.4%)48 (34.8%)2.080.06510.964.53

Laboratory Features in Relation to Troponin I Levels.

Patients with elevated troponin I levels were found to have significantly higher creatinine levels (average ranks for creatinine 100.7 versus 72.9, P = 0.002 in the 2 groups) and significantly lower HCO3 levels (average ranks for HCO3 75.3 versus 97.1, P = 0.012) than patients without elevated troponin I levels. Other admission laboratory analyses [INR, creatinine kinase (CK), ALT, and MAP] were found to be similar between the 2 groups; however, there was a trend toward elevated troponin levels being associated with higher CK levels with the median rank CK level for the elevated troponin I group greater than for the normal troponin I group (269 versus 105 IU/l, P = 0.091). GFR calculations were completed on 173 patients due to missing information on 14 patients. The mean GFR was 108.15 ml/minute in the troponin I group with <0.1 ng/ml, and 80.43 ml/minute in the troponin I group with >0.1 ng/ml (P = 0.0020). Sixty-eight percent (28 of 41) of the patients in the troponin I group with <0.1 ng/ml had GFRs > 60 ml/minute, whereas 36% (48 of 132) of the patients with troponin I levels > 0.1 ng/ml had GFRs > 60 ml/minute.

Role of Etiology.

A similar prevalence of positive values (troponin I > 0.1 ng/ml) was observed in all etiologic categories: 84% with ischemic liver injury, 66% with acetaminophen injury, 88% with viral hepatitis, and 74% of those in the other causes group. Similarly, there was not one etiologic category (such as ischemia) overrepresented at very high troponin levels: 14 patients had values >10 ng/ml, representing a 100-fold increase above the upper limit of normal: 8 acetaminophen cases, 4 ischemia cases, 1 with HBV infection, and 1 indeterminate case.


Mortality was assessed in all patients as outcome at 3 weeks. A total of 134 patients survived and 53 patients died. One-hundred-six patients (56.6%) recovered without orthotopic liver transplant (spontaneous survival). Forty-six (24.6%) patients died without transplant. Mortality was 34.4% at 3 weeks among patients with a positive troponin value and 10.2% among those with normal troponin values (OR 4.69; 95% CI: 1.75 to 12.62, P = 0.001). Of the 53 deaths, 5 patients had troponin I levels <0.1 ng/ml, 2 died from cerebral edema, and 3 died of MOF. The causes of death in the remainder of patients included MOF (6); cerebral infarct, bleed, or edema (10); sepsis (7); myocardial infarction (6); cardiopulmonary failure (4); renal failure (3); hypotension (1); aortic dissection (1); bleeding (1); pulmonary embolism (1); and unknown (7).

A total of 35 patients (18.8%) underwent orthotopic liver transplantation of whom 7 (3.8%) died. Twenty-six (74%) of the patients who underwent transplantation had a troponin I level >0.1 ng/ml. All 7 patients who died after orthotopic liver transplantation had a troponin I level > 0.1 ng/ml. Causes of death included 4 with cerebral edema or herniation, 1 with sepsis, 1 with myocardial infarction, and 1 unknown.

Dose Response Relationships.

To further evaluate the association of high troponin I levels with other clinical parameters, 5 groups were identified with the following cutoff values (all as nanograms per milliliter): <0.1, 0.1 to <0.3, 0.3 to <1.0, 1.0 to <3.0, and 3.0 or greater, with only the first representing normal troponin levels (Table 3; Fig. 1). For all parameters examined, a significant relationship was evident between higher troponin I levels and coma grade, presence of arrhythmias, death and transplant, or death after transplant. For example, those with normal troponin I levels had a 22.4% incidence of advanced coma, a 20.4% incidence of arrhythmias, and only 10.2% likelihood of dying. In comparison, those with troponin I levels > 3.0 ng/ml had a 71.4% advanced coma grade, 47.6% arrhythmias, and a 33.3% likelihood of dying. Similarly, GFR decreased across increasing categories of troponin I (Table 3).

Table 3. Evidence for a Dose Response Effect with Higher Troponin I Levels
Troponin I RangeTroponin I Groups*  
  • *

    Patients were grouped into 5 groups, based on troponin levels as shown.

n (%)49 (26%)59 (32%)35 (19%)23 (12%)21 (11%)  
Variables     Jonckheere-Terpstra P value 
Coma III-IV n (%)11 (22.4%)22 (37.3%)21 (60%)15 (65.2%)15 (71.4%)<0.0001 
Arrhythmia, n (%)10 (20.4%)13 (22%)12 (34.3%)13 (56.5%)10 (47.6%)0.0007 
Death, n (%)5 (10.2%)16 (27.1%)12 (34.3%)13 (56.5%)7 (33.3%)0.0003 
Transplant, n (%)8 (16.3%)11 (18.6%)5 (8.5%)2 (3.4%)2 (3.4%)0.7833 
APACHE II Score n (Mean)27 (11.11)38 (13.32)31 (17.94)19 (17.63)18 (20.67)  
95% Confidence Interval8.92–13.3011.48–15.1515.43–20.4415.03–20.2317.13–24.20  
Creatinine (mg/dl)     Kruskal WallisSignificant Groups (P < 0.05), Dunn Method
Median1.151.402.202.503.500.00011 vs. 4
Range0.40–10.000.50–7.000.40–7.300.50–6.800.50–7.20 1 vs. 5
       2 vs. 5
GFR (Mean)108.1593.3186.1269.0050.620.0002
thumbnail image

Figure 1. Apparent dose effect of troponin levels in relation to coma grade, presence of arrhythmias, and outcomes. Higher troponin levels were associated in a general way with greater degrees of coma, greater likelihood of arrhythmias and death. Lower likelihood of transplantation was observed with higher troponin levels.

Download figure to PowerPoint

APACHE II scores were calculated for 133 patients (54 patients were missing one or more variables necessary to calculate the score). The mean APACHE II score among patients with troponin I levels < 0.1 ng/ml (n = 27) was 11.11 ± 5.54 versus 16.69 ± 6.71 among those with troponin I levels > 0.1 ng/ml (n = 106; P < 0.0001). In addition, patients with higher troponin I levels also had significantly higher APACHE II scores (Table 3).


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Although general experience suggests that cardiac damage in ALF has rarely been clinically significant, subclinical cardiac injury as part of MOF may contribute to the ultimate demise of many patients with this life-threatening condition.8 We confirmed that elevations of serum troponin I levels are found with remarkable frequency in ALF (74% overall), among all etiologies, suggesting that mild to moderate degrees of myocardial injury commonly accompany ALF. As expected, very high levels were found in patients with hepatic ischemia who might be expected to have concomitant myocardial ischemia secondary to cardiovascular collapse. The fact that a comparable number of patients with APAP injury or viral hepatitis showed similarly high levels of troponin I suggests that more fundamental mechanisms common to all ALF patients may be operative.

An increase in troponin I reflects myocardial cell damage.12 Elevated values have been reported in other conditions including myocarditis, pulmonary embolism, use of chemotherapy, acute stroke, chronic obstructive pulmonary disease, and septic shock without underlying acute coronary syndrome, indicating that unrecognized myocardial injury may be associated with a variety of disease processes.11, 12 Cardiac troponins have also been reported to predict mortality in these settings.12–14 Of the patients who died in our study population, 91% had elevated troponin I levels. Our study also showed a clear relationship between the level of troponin I observed and the likelihood of a poor outcome; patients with elevated troponin levels had at least a 2-fold greater likelihood of advanced hepatic coma, arrhythmias, death, or transplantation. In addition, elevated APACHE II scores, an independent global measure of disease severity and outcome, were correlated with higher troponin I levels. Thus, the finding of elevated troponin I correlated with overall disease severity, suggesting a generalized pathogenic process.

Troponin I has been reported to be elevated in approximately 5% of patients with end-stage renal disease even in the absence of cardiac signs or symptoms.15, 16 However, numerous reports found that troponin I retains prognostic value in end-stage renal disease by predicting early mortality.17, 18 In addition, minor elevations in troponin levels were associated with pathological evidence of myocardial damage in patients with and without renal failure.19 Although elevated troponin I was associated with a lower GFR than in our study, the mean GFR in the group with elevated troponin I was only 80 ml/minutes, which is indicative of minimal renal impairment. Thus, it is unlikely that renal insufficiency explains the association between troponin I and outcomes in our study.

The exact mechanism for cardiac injury in ALF is not clear. Acetaminophen may cause direct cardiac injury under certain circumstances, and this finding correlates with previous histologic findings in the literature.2–6 Diffuse fatty infiltration of myocardial cells has been reported in patients who died of viral hepatitis as well as acetaminophen overdose.7 There is evidence in acetaminophen-induced ALF that depletion of sulfhydryl groups interferes with endothelial nitric oxide, leading to a functional coronary insufficiency.8, 9 However, the widespread elevations of troponin I within all etiologies of ALF suggest that a specific toxic effect of acetaminophen is unlikely. Some as-yet undetermined mechanism such as an effect of the high cytokine levels or other factors leading to MOF seems more likely.13

Critically ill patients are at increased risk of myocardial cell injury because they are exposed to many stresses that increase myocardial oxygen demands; at the same time, the myocardial oxygen supply can be limited by shock, anemia, tachycardia, hypoxemia, and impaired tissue perfusion. Such events can result in release of troponin I from cardiomyocytes into the serum.13 Ammann et al. found that troponin-positive status was associated with increased risk of mortality in critically ill patients without acute coronary syndrome.20 Wu et al. showed similar results in that elevated troponin I level upon admission was associated with increased morbidity and mortality in critically ill noncardiac patients.13 Elevated troponin I level is associated with elevated levels of TNF-alpha, C-reactive protein, and IL-6 in critically ill patients without acute coronary syndrome.20 In addition, depressed myocardial function itself is thought to induce myocyte apoptosis, resulting in reduced coronary artery flow and decreased ejection fraction, which leads to further necrosis with troponin I release.13, 20 It is unclear whether elevated troponin I reflects reversible or irreversible myocardial injury in the setting of ALF.

Several limitations of our study need to be considered when interpreting these data. The optimal cutoff for abnormal troponin I level has not been identified for patients who do not have traditional clinical manifestations of an acute coronary syndrome; others have suggested that a different threshold be considered in such patients.14 The troponin assay also has some limitations. Heterophilic antibodies in human serum can react with reagent immunoglobulins, interfering with in vitro immunoassays.10 However, this is rare with current troponin assays, and there is no rationale to expect this to be a specific problem in ALF. Additional information may be required for diagnosis in specimens that have icterus. These specimens can demonstrate a 5% or smaller change in troponin results with levels of conjugated bilirubin of 20 mg/dl or 40 mg/dl unconjugated bilirubin.10 There is less than 10% interference from several drugs, including acetaminophen in drug concentrations of 30 ng/ml.10

Although nearly all patients in the study would have been in an intensive care unit, their primary problem was considered to be hepatic; data concerning cardiac output and specific tests of cardiac function such as routine echocardiograms were not performed. The ubiquity of the abnormal values observed still supports the concept that an underlying cardiac injury is observed widely in ALF. Further studies have been planned to explore whether other markers of abnormal cardiac function such as brain natriuretic peptide (BNP) are elevated in this severely ill group of patients.

In summary, ALF is a rare but progressive and often fatal condition in which many patients succumb to MOF. In this setting, subclinical cardiac injury appears to be common, based on our findings in this study; higher troponin levels are associated with more severe clinical features and poorer outcomes. Understanding the role of cardiac injury in ALF and in other forms of MOF might shed light on both pathogenesis and prognosis in these conditions.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References