Caspase activation is associated with spontaneous recovery from acute liver failure


  • Potential conflict of interest: Nothing to report.


Acute liver failure (ALF) has various causes and is characterized by rapid hepatocyte dysfunction with development of encephalopathy in the absence of preexisting liver disease. Whereas most patients require liver transplantation to prevent the high mortality, some patients recover spontaneously and show complete liver regeneration. Because of the low incidence of ALF, however, the molecular mechanisms of liver dysfunction and regeneration are largely unknown. In this study, we investigated the role of apoptosis and caspases in 70 ALF patients using novel biomarkers that allow the detection of caspase activation in serum samples. Compared with healthy individuals, activation of caspases was strongly enhanced in ALF patients. Interestingly, patients with spontaneous recovery from ALF revealed a significantly higher activation of caspases than patients that required transplantation or died, although in the latter patients extensive DNA fragmentation and signs of nonapoptotic cell death were observed. In the spontaneous survivors, increased caspase activation was accompanied by elevated levels of tumor necrosis factor alpha (TNF-α) and interleukin-6 (IL-6), important cytokines involved in liver regeneration. Conclusion: Our data suggest that caspase activation and apoptosis are involved in ALF of patients with spontaneous recovery, whereas caspase-independent cell death might be more relevant in irreversible forms of liver failure. These findings might be important for therapeutic options of ALF but also suggest that measurement of caspase activation might be of prognostic value to predict the outcome of acute liver failure. (HEPATOLOGY 2008.)

Acute liver failure (ALF) is a life-threatening syndrome characterized by sudden and massive liver cell dysfunction, resulting in the development of hepatic encephalopathy usually within 12 weeks of the onset of symptoms in individuals without previous history of liver disease.1, 2 ALF can be the consequence of various etiologies including intoxications, viral or autoimmune hepatitis, Wilson's disease, Budd-Chiari syndrome, metabolic disorders, ischemia, or unknown reasons.3–5 Despite advances in intensive care management, orthotopic liver transplantation is currently the only therapeutic option for patients who are unlikely to recover spontaneously.3, 6 However, liver transplantation bears the risk of complications and requires lifelong immunosuppression. Thus, selection of patients and timing of transplantation remain critical issues.

Unlike ALF patients that will not survive without transplantation, some patients recover spontaneously from ALF and show complete liver regeneration. The molecular mechanisms that determine the ALF outcome are not well characterized. There is increasing evidence that apoptotic liver cell death plays a role in ALF, although it remains almost unknown whether apoptosis and not other forms of cell death, such as necrosis, are involved in ALF and might influence the disease outcome. The key mediators of apoptosis are caspases, intracellular cysteine proteases that cleave various substrates including structural proteins such as cytokeratin-18.7 There is also increasing evidence that caspases, in addition to their function in cell death, participate in nonapoptotic processes, such as cell proliferation and differentiation.8

Caspases can be activated by two signaling routes, namely, the extrinsic and the intrinsic pathways.9 The intrinsic pathway is initiated by the mitochondrial release of cytochrome c, whereas the extrinsic pathway is triggered by ligand binding of death receptors such as CD95. Although hepatocyte apoptosis may occur by a variety of mechanisms, death receptor–mediated apoptosis is a particularly prominent process in the liver.10 CD95/CD95L expression has been shown to be up-regulated in chronic viral hepatitis and to correlate with disease activity.11, 12 Inappropriate activation of death receptors also might lead to ALF. This has been impressively demonstrated in mice that died rapidly of liver failure with massive hepatocyte apoptosis when agonistic anti-CD95 antibody was injected.13 Recent data implicate the CD95 system also in human ALF caused by viral hepatitis and Wilson's disease.14, 15 Moreover, caspase activation and apoptosis have been demonstrated in a small cohort of ALF patients.16, 17

Because of the rarity of ALF, so far the role of apoptosis in ALF has mainly been studied in animal models, whereas human data are sparse. Furthermore, no detailed study has investigated whether the extent of caspase activation might be of predictive value for the outcome of ALF patients. In the current study, we analyzed the role of apoptosis and caspase activation in 70 ALF patients who spontaneously recovered, underwent liver transplantation, or died. To this end, we employed two recently developed assays that allow the measurement of hepatic caspase activation in serum samples.18, 19 Our results show that caspase activation is strongly elevated in ALF patients compared with healthy controls. Interestingly, patients with spontaneous recovery revealed significantly higher caspase activation, accompanied by an elevated expression of the proregenerative cytokines interleukin-6 (IL-6) and tumor necrosis factor alpha (TNF-α), than nonrecovered individuals. Our results therefore demonstrate that apoptosis is not the predominant form of cell death in those critically ill ALF patients and, moreover, suggest that measurement of caspase activation might be of prognostic value to predict the outcome of acute liver failure.


AAP, acetaminophen; ALF, acute liver failure; ALT: alanine aminotransferase; AST: aspartate aminotransferase; ATP, adenosine triphosphate; CK, cytokeratin; ELISA, enzyme-linked immunosorbent assay; IL-6, interleukin-6; LDH, lactate dehydrogenase; NAAP, non-acetaminophen; RLU, relative light units; ROC, receiver operating characteristics; TNF-α, tumor necrosis factor alpha; TUNEL, terminal deoxynucleotidyl transferase-mediated nick-end labeling.

Patients and Methods


We investigated sera from 70 ALF patients (18 men, 52 women; mean age, 43 ± 1.8 years; range, 16–77 years). ALF was defined as acute-onset deterioration of liver function without signs of chronic liver disease resulting in jaundice, decrease in coagulation factors (international normalized ratio > 1.5) and development of encephalopathy within 8 weeks.20 Sera were collected at the day of hospital admission and stored at −20°C. The study was performed according to the guidelines of the Ethics Committee of Hannover Medical School. Causes included viral hepatitis (n = 11), toxic liver injury (n = 24), Budd-Chiari syndrome (n = 7), Wilson's disease (n = 5), autoimmune hepatitis (n = 1), and cryptogenic liver failure (n = 22). ALF patients with toxic liver injury included nine patients with acetaminophen (AAP), 12 patients with other drug or Amanita mushroom poisoning (non-acetaminophen, NAAP), and three patients with mixed poisoning. The 70 patients included 24 that recovered spontaneously and 46 that underwent liver transplantation (n = 31) or died (n = 15). Among the different causes, patient numbers with versus without spontaneous recovery were: viral hepatitis (6/5), toxic liver injury (14/10), Budd-Chiari syndrome (0/7), Wilson's disease (0/5), autoimmune hepatitis (0/1), and cryptogenic liver failure (4/18). Sera from 11 healthy volunteers (mean age, 24 ± 0.3 years) served as controls. King's College Hospital criteria and model of end-stage liver disease score were determined as previously defined.21, 22 The main demographic and clinical features of the patients are summarized in Table 1.

Table 1. Demographic and Clinical Features of Patients
 TotalSpontaneous RecoveryNo Spontaneous Recovery
  1. Abbreviations: INR, international normalized ratio; MELD, Model for End-Stage Liver Disease; KCH, King's College Hospital; and SEM, standard error of the mean.

No. of patients702446
Mean age ± SEM43 ± 1.842 ± 3.143 ± 2.2
Sex (% male)264217 n
MELD score33 ± 1.129 ± 1.935 ± 1.4
KCH criteria (% fulfillment)83.620.862.8
Mean bilirubin (μmol/L ± SEM)244 ± 25140 ± 28299 ± 32
Mean creatinine (μmol/L ± SEM)130 ± 14.3118 ± 22137 ± 19
Mean INR ± SEM4.5 ± 0.43.7 ± 0.44.9 ± 0.5
Mean lactate (mmol/L ± SEM)5.6 ± 0.63.7 ± 0.66.6 ± 0.9
Mean factor V (% ± SEM)28 ± 2.936.9 ± 7.123.1 ± 2.1
Mean ammonia (μmol/L ± SEM)91 ± 1259 ± 9.7108 ± 16.7

Histochemical Detection of CK-18 Cleavage, Caspase-3 Activation, and DNA Fragmentation.

Paraffin-embedded liver sections from 13 ALF patients were obtained from the Department of Pathology of the Hannover Medical School. For detection of caspase-mediated cleavage of cytokeratin-18 (CK-18), sections were stained with the monoclonal antibody M30 essentially as described.18, 19 Double-staining for active caspase-3 and DNA fragmentation was performed as described previously using an anti-cleaved caspase-3 antibody and a terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) cell death detection kit.23

Serological Detection of Caspase Activity, CK-18 Release, TNF-α, and IL-6.

Caspase activity in serum samples was determined via two independent methods. For the quantitative measurement of the caspase-generated neoepitope of CK-18, we used the M30-Apoptosense enzyme-linked immunosorbent assay (ELISA) kit (Peviva, Bromma, Sweden) as described previously.18, 19 Samples with high values outside the standard curve were diluted, yielding satisfying linearity. In addition, we used a luminescent substrate assay (Caspase-Glo, Promega, Mannheim, Germany) for active caspase-3 and caspase-7.19 We further used the M65 ELISA (Peviva) that quantitates both uncleaved and cleaved CK-18. Serum levels of TNF-α and IL-6 were determined by ELISA (Bender MedSystem; Vienna, Austria).

Statistical Analysis.

Statistical analyses comparing the concentration of the different variables in the serum of ALF patients showing spontaneous and nonspontaneous recovery were performed using the analysis of variance and post hoc test for multiple comparisons as well as the two-tailed t test for equality of means (SPSS 15.0 software). The predictive value of the respective assays was determined by receiver operating characteristics (ROC) plot analysis. The statistical analysis was confirmed by a professional statistician. A P value of less than 0.05 was considered significant. All assays were performed in duplicate.


Detection of Caspase-Generated CK-18 Fragments in Liver Sections of ALF Patients.

In first experiments we investigated the role of apoptotic caspase activation in liver sections of ALF patients. For this reason, sections of patients with ALF (n = 13) were stained by immunohistochemistry with the antibody M30, which selectively detects a caspase-generated neoepitope of CK-18 that is constitutively expressed by hepatocytes.18, 24 Healthy liver tissue (n = 10) served as control and did not show any immunoreactivity for caspase-generated CK-18 fragments (Fig. 1A). In contrast, liver sections of representative patients with ALF caused by Wilson's disease (Fig. 1B) or toxic liver disease (Fig. 1C) showed hepatocytes with a cytoplasmic staining for caspase-generated CK-18 fragments. Similar results were obtained in sections of patients with ALF caused by viral hepatitis (data not shown). Areas with features of necrosis generally showed no immunoreactivity.

Figure 1.

Immunohistochemical detection of caspase-generated CK-18 cleavage fragments (A) in a representative sample of healthy liver and (B,C) in liver explants of representative patients with ALF caused by Wilson's disease and toxic liver injury, respectively. Liver sections were immunostained with the M30 antibody that selectively detects a caspase-generated neoepitope of CK-18 (normal magnification ×400). (A) Healthy control liver showed no immunoreactivity, whereas (B,C) cytoplasmic staining was detected in hepatocytes of both ALF patients.

Detection of Caspase-Generated CK-18 Fragments in Serum of ALF Patients.

Having demonstrated that CK-18 cleavage fragments were elevated in ALF sections, we investigated caspase activity in the serum of ALF patients using the M30 ELISA that detects caspase-generated CK-18 fragments.18, 19 We first analyzed the time course of caspase activity during the first 7 days of hospitalization. In most cases, caspase activation was maximal at the day of hospital admission and then continuously declined together with aminotransferase and lactate dehydrogenase (LDH) levels (Fig. 2A, left panel). In a few of the nonsurviving patients, caspase activity remained relatively constant during the first days after hospitalization and then suddenly increased at the time of death, which was paralleled by an increase of LDH levels (Fig. 2A, right panel). Aminotransferase levels decreased in these patients, however, suggesting that the increased caspase activity was presumably not caused by hepatocyte apoptosis but rather caused by severe circulation problems that might have secondarily affected other epithelial tissues. We therefore analyzed caspase activation in sera of ALF patients (n = 70) at day 1 of hospitalization. As shown in Fig. 2B, mean levels of caspase activity differed among different causes of ALF. However, statistical analysis by the analysis of variance and post hoc test for multiple comparisons did not reveal significant differences of CK-18 cleavage between the different etiological groups.

Figure 2.

Time course and influence of etiology on caspase activity in ALF patients. (A) Sera from ALF patients were measured for caspase-mediated CK-18 cleavage (M30 ELISA) and AST, ALT, and LDH levels each day within the first week of hospitalization. The enzyme activities are indicated as the percentage of day 1 values. The left panel shows a representative example of a spontaneously recovered patient with ALF of indeterminate reason showing maximal caspase activity and other markers of liver damage at day 1, followed by a continuous decline thereafter. The right panel shows one of the rare cases among ALF patients without spontaneous recovery, in which caspase activity remained relatively constant during the first days and then suddenly increased together with LDH at the time of death (day 4). The fact, however, that aminotransferases further decreased suggests that the increased caspase activity in this patient was not hepatocyte-derived but caused by cell death of other epithelial tissues. (B) Serum caspase activities (M30 ELISA) in healthy controls and different causes of ALF including viral hepatitis, acetaminophen (AAP), and non-acetaminophen (NAAP) induced liver injury, Budd-Chiari syndrome, Wilson's disease and unknown causes. The number of the patients in each category is indicated. Using the analysis of variance and post hoc test for multiple comparisons, no statistical difference was observed between the different etiological ALF groups.

ALF Patients with Spontaneous Recovery Showed Higher Serum Caspase Activity Than Patients Without Spontaneous Recovery.

To investigate a potential role of caspase activity for the ALF outcome, we separated the ALF group into patients who subsequently underwent liver transplantation or died (n = 46) and patients with spontaneous recovery (n = 24). As shown in Fig. 3A, ALF patients who required liver transplantation or died as well as patients with spontaneous recovery showed significantly elevated (P < 0.01) caspase-generated CK-18 cleavage fragments (5118 ± 923 U/L; range, 111–24,850 U/L and 10,558 ± 2437 U/L; range, 287–38,670 U/L, respectively) in the serum compared with healthy control individuals (n = 11, 87 ± 11 U/L; range, 50–163 U/L). Interestingly, patients with spontaneous liver reconstitution showed significantly (P < 0.05) higher serum levels of caspase-generated CK-18 fragments compared with patients who underwent liver transplantation or died (Fig. 3A).

Figure 3.

Serological detection of caspase activity in ALF patients who either showed spontaneous recovery (n = 24) or underwent liver transplantation or died (nonspontaneous recovery, n = 46). Serological caspase activation was assessed by two independent assays including (A) the M30 ELISA specific for a caspase-generated CK-18 neoepitope and (B) a luminescent substrate assay detecting caspase-3/caspase-7 activity. In both assays, caspase activation was significantly (P < 0.01) elevated in patients with ALF compared with healthy control individuals (n = 11). Moreover, caspase activation was significantly (P < 0.05) stronger in the serum of patients who spontaneously recovered (SR) compared with patients without spontaneous recovery (NSR). (C) Measurement of uncleaved and cleaved CK-18 as a marker of overall cell death using the M65 ELISA. Compared with healthy individuals ALF patients both with and without spontaneous recovery showed significantly elevated serum levels of caspase-cleaved and uncleaved CK-18. Patients with worse outcome showed significantly higher levels of total CK-18 than patients with spontaneous recovery. Data represent the mean values ± SEM. *P < 0.05; **P < 0.01. RLU, relative light units.

To verify these findings, we additionally employed a luminometric enzyme assay for the measurement of caspase activity.19 To this end, serum samples were incubated with the caspase substrate DEVD-aminoluciferin, which on cleavage by caspase-3 and caspase-7 generates a luminescent signal that can be measured in a luciferase reaction (Fig. 3B). In line with the results of the ELISA, ALF patients showed significantly higher (P < 0.01) caspase activity in serum compared with healthy control individuals [1151 ± 143 relative light units (RLU); range 893–1704 RLU). In the luminometric assay, patients with spontaneous recovery displayed significantly elevated (P < 0.05) caspase-3/-7 activity in the serum (16,319 ± 3467 RLU; range, 811–56,927 RLU) compared with patients without spontaneous recovery (8904 ± 1851 RLU; range, 343–61,718 RLU; Fig. 3B). Differences between both outcome groups were also found when ALF etiologies with low serum caspase activity (Budd-Chiari syndrome and Wilson's disease; see Fig. 2B) were excluded from the analysis. Also in this case, patients with spontaneous recovery revealed higher serum levels of caspase-mediated CK-18 fragments (10,560 ± 2437 U/L) than patients with worse outcome (6319 ± 1172 U/L). Thus, our results indicate that pronounced caspase activation is associated with spontaneous survival.

Patients Without Spontaneous Recovery Show Only Weak Caspase Activation but Extensive DNA Fragmentation.

In addition, we measured the serum levels of both uncleaved and cleaved CK-18 using the M65 ELISA as a marker of overall cell death. Patients without and with spontaneous recovery showed significantly (P < 0.01) elevated total CK-18 in serum (26,736 ± 5823 U/L; range, 1611–163,656 U/L; and 12,834 ± 2440 U/L; range, 1387–49,100 U/L) compared with healthy control individuals (1079 ± 128 U/L; 590–1890 U/L). Interestingly, patients with nonspontaneous recovery showed significantly (P < 0.05) increased CK-18 release in their sera compared with spontaneously recovered patients (Fig. 3C). The low amount of caspase-cleaved CK-18 and the high levels of total CK-18 in serum therefore suggest that in these patients hepatocyte death is mainly mediated by a nonapoptotic mechanism.

This assumption was further supported when we investigated caspase-3 activation and DNA fragmentation (TUNEL) in liver sections of ALF patients. Consistent with the results obtained in serum, liver sections of patients with spontaneous recovery showed considerably more caspase-3–positive than TUNEL-reactive hepatocytes (Fig. 4). This finding is in line with previous reports demonstrating that caspase activation, unlike DNA fragmentation, is a very early marker of apoptosis.25 In contrast, patients without spontaneous recovery showed extensive DNA fragmentation in the nuclei of hepatocytes but showed only a weak staining for active caspase-3 (Fig. 4). In these patients, most of the caspase-3–positive hepatocytes were also TUNEL-positive. Because TUNEL staining does not distinguish between necrotic and apoptotic cell death,26, 27 the absence of caspase activation in the presence of extensive DNA fragmentation therefore indicates that necrosis and activation of apoptosis-independent endonucleases might be predominantly involved in ALF patients without spontaneous recovery.

Figure 4.

Detection of caspase-3 activation and DNA fragmentation in patients without and with spontaneous recovery from ALF. Liver sections of patients (n = 13) were stained with antibodies specific for active caspase-3 (red) or with TUNEL (green) to detect DNA fragmentation. Patients with spontaneous recovery displayed apoptotic caspase-3 activation but considerably less DNA fragmentation. In contrast, patients without spontaneous liver reconstitution revealed extensive DNA fragmentation but only a weak activation of caspase-3. The results of a representative experiment are shown.

Comparison of Cell Death in AAP-Induced and NAAP-Induced ALF.

It is currently controversial whether ALF caused by an overdose of acetaminophen (AAP) is mediated by necrosis or by apoptosis. Therefore, we compared in more detail the serum levels of caspase-generated CK-18 fragments (M30) and total CK-18 (M65) in patients with AAP (n = 9) and with non-acetaminophen (NAAP; caused by Amanita mushroom or other drugs, n = 12)–induced toxic liver injury. As demonstrated in Fig. 5A, no significant differences were observed in caspase-mediated CK-18 cleavage between both groups (AAP: 8530 ± 2654 U/L; NAAP: 6856 ± 1800 U/L). However, AAP patients showed significantly (P < 0.05) higher total CK-18 serum levels (51,540 ± 19,860 U/L) compared with NAAP patients (11,470 ± 2364 U/L). This result suggests that nonapoptotic cell death plays presumably a major role in AAP-induced ALF (Fig. 5B). This assumption was supported by our finding of significantly higher serum levels of aspartate aminotransferase (AST; Fig. 5C) or alanine aminotransferase (ALT, data not shown) in AAP compared with NAAP patients (Fig. 5C). Importantly, however, no significant differences were obtained when caspase activities in patients with spontaneous recovery were compared between the AAP and NAAP cohorts (Fig. 5D).

Figure 5.

Comparison of caspase activity and other serum markers of liver injury in ALF patients with acetaminophen (AAP)-induced and non–acetaminophen (NAAP)-induced toxic liver injury. (A) When AAP (n = 9) and NAAP (n = 12) patients were compared for caspase-mediated CK-18 cleavage (M30 ELISA), no significant differences were observed between both causes. (B,C) AAP patients showed significantly elevated serum levels of cleaved and uncleaved CK-18 (M65 ELISA) and AST. (D) No significant difference in serum levels of caspase-generated CK-18 fragments could be detected when patients with spontaneous recovery (SR) in the AAP (n = 7) and NAAP (n = 5) groups were compared. *P < 0.05; n.s., nonsignificant.

Caspase Activity Correlates with Aminotransferase and LDH Levels in ALF Patients Without Spontaneous Recovery but Not in Patients with Spontaneous Survival.

To further investigate the mode of cell death and its role on ALF outcome, we compared caspase activity with the serum levels of AST, ALT, and LDH that are traditionally considered as necrotic markers of cell death. As shown by regression analysis (Fig. 6), in patients without spontaneous recovery serum concentrations of CK-18 cleavage products significantly (P < 0.01) correlated with serum levels of AST (r = 0.64), ALT (r = 0.58), and LDH (r = 0.41). In contrast, in patients with spontaneous liver reconstitution no significant correlations were detected between serum CK-18 fragments and AST (r = 0.26), ALT (r = 0.06), or LDH (r = 0.39).

Figure 6.

Regression analysis correlating serum levels of caspase-generated CK-18 fragments with aspartate aminotransferase (AST), alanine aminotransferase (ALT), and lactate dehydrogenase (LDH) in patients with (n = 24) or without (n = 46) spontaneous recovery from ALF. Patients without spontaneous liver reconstitution showed a significant (P < 0.01) correlation of caspase-generated CK-18 fragments in the serum with (A) AST, (C) ALT, and (E) LDH. In contrast, in patients with spontaneous recovery no significant correlation could be observed between CK-18 cleavage fragments and (B) AST, (D) ALT, or (F) LDH. **P < 0.01; n.s., nonsignificant.

ALF Patients with Spontaneous Recovery Show Higher Serum Levels of IL-6 and TNF-α Compared with Patients Without Spontaneous Recovery.

Cytokines, such as TNF-α and IL-6, can promote hepatic survival by stimulating liver regeneration and providing hepatoprotection in a variety of liver injury models.28 In addition to caspase activation, we therefore measured serum levels of TNF-α and IL-6 in the two patient cohorts. Compared with healthy individuals, both cytokines were significantly (P < 0.01) increased in ALF patients (Fig. 7A, B). Interestingly, patients who spontaneously recovered showed significantly (P < 0.05) elevated IL-6 levels (316 ± 89 pg/mL) compared with ALF patients who underwent transplantation or died (110 ± 18 pg/mL). Similarly to IL-6, TNF-α levels were also significantly increased (P < 0.01) in the spontaneous survivors (2.6 ± 0.4 pg/mL) compared with patients without spontaneous recovery (1.5 ± 0.2 pg/mL, Fig. 7B). Regression analysis showed a significant (P < 0.01) correlation of CK-18 fragments with TNF-α in serum of ALF patients (r = 0.44, Fig. 7C).

Figure 7.

Serological detection of (A) IL-6 and (B) TNF-α in patients with (n = 24) or without (n = 46) spontaneous recovery from ALF. Healthy individuals (n = 11) served as controls. Serum levels of TNF-α and IL-6 were assessed by ELISA. Compared with healthy individuals, (A) IL-6 and (B) TNF-α serum levels were significantly (P < 0.01) elevated in patients with ALF. Patients with spontaneous recovery (SR) showed significantly (P < 0.05) higher (A) IL-6 and (B) TNF-α levels compared with patients that underwent transplantation or died (NSR). Regression analysis revealed a significant correlation (R = 0.44; P < 0.01) of caspase-generated CK-18 fragments in the serum with TNF-α (C). Data represent the mean values ± SEM. *P < 0.05; **P < 0.01.

Determination of the Cutoff Value of Serological Caspase Activity to Predict the ALF Outcome.

We then calculated the cutoff value of serum caspase activity that correctly predicts ALF outcome with the best-compromise sensitivity/specificity (predictive discriminating value). To this end, we performed a receiver operating characteristics (ROC) plot analysis comparing patients with and without spontaneous recovery. Caspase-generated CK-18 fragments above or below 6712 U/L correctly predicted the ALF outcome with a sensitivity of 52% and a specificity of 76% (area under the curve, 0.67; confidence interval, 95%; Fig. 8A). Compared with the ELISA, similar results were obtained with the luminometric substrate test. Caspase-3/caspase-7 activity above or below 9276 RLU correctly predicted the ALF outcome with a sensitivity of 60% and a specificity of 74% (area under the curve, 0.75; confidence interval, 95%; Fig. 8B). Thus, these data suggest that measurement of caspase activity might be valuable as a noninvasive marker for prediction of ALF outcome.

Figure 8.

Predictive discrimination of (A) caspase-generated CK-18 fragments and (B) caspase-3/7 activity as determined by receiver operating characteristics (ROC) plot analysis. The ROC analysis indicates the caspase activity threshold for the best compromise sensitivity/specificity to predict ALF outcome (AUC, area under the curve; CI, confidence interval; RLU, relative light units).


Although ALF remains associated with high mortality, a portion of patients show spontaneous liver regeneration. The underlying mechanisms of liver damage and regeneration are incompletely understood, and no reliable marker exists that predicts survival from ALF. A variety of prognostic variables of ALF have been described, including age, creatinine, bilirubin, phosphate, international normalized ratio, encephalopathy, alpha-fetoprotein, arterial pH, and lactate levels.21, 22, 29 Several of these markers merely reflect the loss of liver function, and their discriminatory power increases while the disease progresses and the general condition of the patient deteriorates. Because the decision for liver transplantation should be made as early as possible, there is a strong need for early markers that predict spontaneous recovery from ALF. However, performance of clinical studies in ALF patients presents serious problems because of the varied causes, the small number of cases, and difficulties in predicting the patient's outcome without transplantation.

In this study, we investigated the activation of caspases in ALF patients with respect to the ability of spontaneous survival. As in many other diseases, it is still unclear whether apoptosis or necrosis plays the predominant role in ALF.30 Although a distinction between both processes might be artificial, determination of the mode of cell death might be of therapeutic relevance, in particular in view of the recent development of therapeutic caspase inhibitors.31 In models of AAP-induced liver damage, which is a major cause of human ALF, some studies showed apoptosis after drug exposure of mice and mouse hepatocytes, whereas other reports indicated that necrosis is the principal mode of liver damage.32, 33 These observations might be related to our findings. Using two independent methods, we demonstrate for the first time that ALF patients have considerable caspase activation, which was unexpectedly higher in spontaneous survivors than in patients that required transplantation or died. Nevertheless, despite a weaker activation of caspases, patients without spontaneous recovery revealed extensive TUNEL reactivity in the liver. Moreover, sera from these patients contained increased levels of total CK-18 but reduced levels of its caspase-generated fragments as compared with patients with spontaneous recovery. These findings strongly suggest that necrosis but not apoptosis is predominant in those critically ill ALF patients. It cannot be excluded that also the cause of ALF impacts on caspase activation and disease outcome. We did not find a statistically significant difference between caspase activation among the different etiological groups. Although our study certainly presents one of the largest efforts to characterize cell death processes in ALF, we were still limited by the relatively small patient numbers with different disease outcome within each category of ALF etiology.

The induction of apoptosis or necrosis is presumably determined by the magnitude and duration of liver injury.34 In contrast to necrosis, apoptosis is strictly adenosine triphosphate (ATP) dependent.35 In mouse models, it has been shown that an AAP overdose leads to massive hepatocellular necrosis characterized by the depletion of glutathione, reduced nicotinamide adenine dinucleotide, and ATP.32, 33 When ATP depletion was prevented, however, necrosis was blocked, whereas caspase-dependent apoptosis increased.36 In models of AAP-induced cytotoxicity it was also demonstrated that nonapoptotic calcium-dependent endonucleases, such as DNase-1, mediate TUNEL reactivity and DNA fragmentation.33 These data clearly suggest that TUNEL reactivity, which is often used to identify apoptosis but also occurs in necrosis, certainly overestimates the role of apoptosis in models of ALF.26, 27 Moreover, ATP depletion–induced necrosis resulting from severe mitochondrial dysfunction is consistent with the high lactate levels that develop in critically ill ALF patients and are associated with poor outcome.37

Our finding of particularly strong caspase activation in patients with spontaneous liver reconstitution could simply reflect weaker liver damage compared with those patients that do not recover from ALF. Nevertheless, it might be speculated that caspases also participate in regeneration processes of the liver. An important regulator in this context is nuclear factor kappa B, which is activated by TNF-α and involved in transcriptional activation of a huge number of cytokines and growth-promoting target genes.38 Several reports have shown that the death receptor–associated caspase-8 can induce nuclear factor kappa B activation and its target gene IL-6.38 Similarly to TNF-α, IL-6 plays a pivotal role in liver regeneration and hepatoprotection in a variety of liver injury as well as resection models.28 For instance, administration of IL-6 potently reverses hepatocellular injury.39 IL-6 also has been reported to restore metabolic function and to improve ATP levels after reperfusion injury and extreme hepatectomy.40 The observations are therefore in line with our finding of higher TNF-α and IL-6 levels in patients recovering spontaneously from ALF compared with those who failed to recover.

There is evidence that caspases also participate in nonapoptotic functions, including cellular proliferation and differentiation. Caspases can cleave different cytokine precursors to generate active cytokines and thus create an environment that could be essential for liver regeneration. Experiments in mice showed that caspase activation is associated with chemokine production and inflammation in the liver, whereas caspase-3 inhibition strongly reduced activity of proinflammatory transcription factors and chemokines.41 A role of caspases in liver regeneration is best exemplified by a recent report showing that the hepatocyte-specific knockout of caspase-8 attenuates hepatocyte proliferation after partial hepatectomy.42 Together with the finding that CD95 promotes liver regeneration,43 death receptor–mediated caspase activation therefore might be not only involved in liver damage, but also required for liver regeneration.

One could envision that in certain liver areas hepatocytes might die by caspase activation, whereas distantly located cells may be triggered to proliferate in the presence of limited caspase activation and additional survival factors. Whether traditional death regulators mediate differentiation of hepatic progenitor cells in ALF remains unexplored. Furthermore, death receptor signaling can induce differentiation of stellate cells and possibly other nonparenchymal liver cells,44 which might influence ALF outcome. Whether caspases are indeed directly involved in recovery from ALF certainly requires further investigation. Nevertheless, this possibility should not be ignored when considering the therapeutic use of caspase inhibitors in ALF treatment.

In conclusion, we demonstrate that apoptotic caspase activation is involved in ALF, in particular in patients with spontaneous liver reconstitution. In contrast, in patients that do not recover necrosis predominates, presumably because of massive liver damage, resulting in depletion of ATP. Our data therefore suggest that serological detection of caspase activity might be an early marker for prediction of spontaneous recovery from ALF. Although we are aware that biomarkers cannot replace clinical judgment, inclusion of new prognostic markers may allow refinement of and lead to a more precise selection of patients requiring liver transplantation. Discrimination between apoptosis or other forms of cell death also might be important for development of effective interventions to prevent hepatocellular death. Further studies of the prognostic value of apoptosis markers in ALF are therefore warranted.