Acute liver failure is associated with elevated liver stiffness and hepatic stellate cell activation

Abstract

Acute liver failure (ALF) is associated with massive short-term cell death, whereas chronic liver injury is accompanied by continuous cell death. Hepatic stellate cells (HSCs) contribute to tissue repair and liver fibrosis in chronic liver injury, although their role in ALF remains unexplained. Twenty-nine patients (median age = 43 years, 17 females and 12 males) with ALF according to the Acute Liver Failure Study Group criteria were included. Upon the diagnosis of ALF and after 7 days, we determined liver stiffness (LS) with FibroScan, standard laboratory parameters, and serum levels of matrix metalloproteinase 1 (MMP-1), MMP-2, MMP-9, tissue inhibitor of metalloproteinases 1 (TIMP-1), TIMP-2, hyaluronic acid, and markers of overall cell death (M65) and apoptosis (M30). Stellate cell activation and progenitor response were analyzed immunohistochemically in biopsy samples of 12 patients with α-smooth muscle actin (α-SMA), keratin-17, and keratin-19 staining, respectively. Cell death markers (M30 level = 2243 ± 559.6 U/L, M65 level = 3732 ± 839.9 U/L) and fibrosis markers (TIMP-1 level = 629.9 ± 69.4 U/mL, MMP-2 level = 264 ± 32.5 U/mL, hyaluronic acid level = 438.5 ± 69.3 μg/mL) were significantly increased in patients versus healthy controls. This was paralleled by collagen deposition, elevated α-SMA expression, and higher LS (25.6 ± 3.0 kPa). ALF was associated with ductular progenitor proliferation. Conclusion: Our results demonstrate HSC activation and a progenitor response in ALF. Positive correlations between LS, the degree of liver cell damage, and the intensity of HSC activation suggest that fibrosis is a response to ALF in an attempt to repair damaged tissue. (Hepatology 2010)

Acute liver failure (ALF) is a devastating clinical syndrome associated with high mortality in the absence of immediate state-of-the-art intensive care or liver transplantation.1 ALF can occur as a result of various etiologies (overdosing with acetaminophen or other drugs, viral hepatitis, ischemia, and other causes1); in approximately 15% of adult cases and 50% of pediatric cases, the reason is unidentified.2 When ALF is not fatal, the liver has a unique capacity to recover completely, although reliably predicting mortality remains challenging. ALF is associated with massive cell death and, consequently, impairment of liver function.3, 4 In contrast, chronic liver injury is associated with progressive hepatocyte apoptosis over years to decades, which promotes liver fibrogenesis.5 During chronic liver injury, hepatic stellate cells (HSCs) differentiate into contractile myofibroblasts, which contribute to tissue repair alongside regions of cellular demise.6, 7 However, it is not clear whether HSCs play a role in ALF.

Although the pathophysiological events leading to ALF remain poorly characterized, they consistently provoke excessive hepatocyte apoptosis and necrosis.8, 9 The consequential loss of functional liver mass, ultimately culminating in organ failure,3, 4 may be physiologically counteracted by increased formation of extracellular matrix, which forms a scaffold for optimal support of the regenerative processes. Indeed, profibrogenic cells (e.g., HSCs and myofibroblasts) are activated early after liver injury to produce extracellular matrix components and hyaluronic acid (an indirect marker for collagen formation).10 This is paralleled by up-regulation of tissue inhibitor of metalloproteinases 1 (TIMP-1) and TIMP-2, which serve as markers of HSC activation.11–13 It remains controversial whether and how these HSC activity markers contribute to changes in liver architecture and fibrogenesis in ALF. In contrast, we recently found that the engulfment of hepatocyte-derived apoptotic bodies activates the expression of profibrotic genes in HSCs14 (R. K. Gieseler et al., unpublished data, 2010). Moreover, recent results have revealed a correlation between liver injury and organ stiffness as measured by transient elastography,15 and this suggests that this technology indirectly monitors early fibrosis as well as edema and inflammation.16

Our present study was thus aimed at elucidating whether (1) hepatocellular death activates HSCs and fibrogenesis in ALF, (2) ongoing fibrogenesis is associated with increasing liver stiffness (LS) during the course of ALF, (3) remodeling of extracellular matrix is indicative of a positive outcome in ALF, and (4) our findings support the concept of excess hepatocyte death as a trigger of regenerative fibrosis in ALF (which is resolved during regeneration).

α-SMA, α-smooth muscle actin; ALF, acute liver failure; ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; ELISA, enzyme-linked immunosorbent assay; FasL, Fas ligand; γGT, gamma glutamyl transferase; HSC, hepatic stellate cell; IL-6, interleukin-6; INR, international normalized ratio; LS, liver stiffness; MELD, Model for End-Stage Liver Disease; MMP, matrix metalloproteinase; r_s, Spearman's rank correlation coefficient; TIMP, tissue inhibitor of metalloproteinases.

Patients and Methods

Patients.

Twenty-nine adult patients (17 females and 12 males), 41.1 ± 13.9 years old, were enrolled in our liver transplant center between February 2008 and June 2009. All patients presented with ALF according to the criteria established by the US Acute Liver Study Group within 4 weeks of onset of symptoms and without apparent or known preexisting liver disease at admission. The international normalized ratio (INR) was ≥1.5 because of acute severe liver injury with encephalopathy of grade 1 or higher.1, 17

Diagnosis, Outcome, and Etiology.

ALF was verified by characteristic laboratory parameters (discussed later), and other causes of liver dysfunction, such as undiagnosed cirrhosis, were excluded. ALF etiologies were established by serological testing for viral hepatitides and antinuclear antibodies, a recent history of drug exposure and the subsequent onset of ALF, cardiac function (transesophageal echocardiography), or the recent consumption of self-collected mushrooms. Ethanol consumption of >30 g per day was considered contributory to the development of ALF, and such cases were excluded.18 Upon admission and/or first examination, the presence of ascites (resulting in an inability to undergo transient elastography) was excluded ultrasonographically (for a detailed distribution of the etiologies and patient demographics, see Table 1).

Table 1. General Characteristics of Patients with ALF and Its Etiologies

	Controls	Total	Group 1: Increase in LS	Group 2: Decrease in LS
* Data are presented as means and standard deviations. † Patients who underwent transplantation were excluded as second LS measurements could not be taken.
Age (years)*	26 ± 3	41 ± 14	47 ± 14.5	36 ± 11.9
Females (%)	41	58	54	63
Males (%)	59	42	46	37
		Total	Group 1	Group 2
ALF etiology (n)
Viral hepatitis		8	5	3
Drug toxicity		9	3	6
Autoimmune hepatitis		3	0	3
Acute liver injury due to heart failure		2	2	0
Mycotoxicosis		3	0	3
Unknown origin†		4	1	1
Total		29	11	16
Outcome (n)
Recovery		25	9	16
Transplantation		2	0	0
Death		2	2	0

Liver Injury, Fibrogenesis, and Apoptosis.

Sera were collected at the times of first and second elastography (n = 27) or immediately before surgery (n = 2). All samples were stored at −20°C within 2 hours until analyses were performed. In addition to clinical and laboratory data, we determined parameters of hepatic fibrogenesis (TIMP-1, TIMP-2, and hyaluronic acid) as well as inhibitors of fibrogenesis [matrix metalloproteinase 2 (MMP-2) and MMP-9]. These markers were analyzed by enzyme-linked immunosorbent assays (ELISAs) for TIMP-1, TIMP-2, MMP-2, and MMP-9 (all from R&D Systems GmbH, Wiesbaden, Germany) and hyaluronic acid (Corgenix, Bloomfield, CO). Liver injury and regeneration were assessed with human sFas ligand ELISAs for Fas ligand (FasL; Diaclone Cell Sciences, Canton, MA). Interleukin-6 (IL-6) was determined at the Department of Clinical Chemistry and Laboratory Diagnostics, University Hospital Essen (Essen, Germany), with a sequential solid-phase chemiluminescence immunoassay (Immulite, Siemens, Eschborn, Germany). Overall cell death and apoptosis were determined with M65 and M30 Apoptosense ELISAs (Peviva, Bromma, Sweden), as described previously.8

Histological Sampling.

Tissue samples were retrieved from 13 patients (5 patients in group 1 and 8 patients in group 2). Because of the impact of liver biopsy on patients already suffering from ALF, it was not possible to retrieve comprehensive tissue samples from all enrolled patients.

Sirius Red Staining for Liver Fibrosis.

Staining of hepatic fibrosis was carried out as previously described.19 Direct Red 80 and Fast Green FCF (color index 42053) were obtained from Sigma-Aldrich (Taufkirchen, Germany). Red-stained collagen fibers were quantified in liver sections with digital image analysis as described before.

Immunohistochemistry for α-Smooth Muscle Actin (α-SMA) and Caspase-3.

Paraffin-embedded tissue sections were deparaffinized and rehydrated, and this was followed by antigen retrieval with a 10 mM sodium citrate buffer (pH 6.0) containing 0.05% Tween 20 for 10 minutes at 80°C. Blocking was performed with normal serum and hydrogen peroxidase. Sections were incubated overnight with monoclonal mouse anti-human α-SMA primary antibody (Dako, Hamburg, Germany) or rabbit polyclonal active caspase-3 (Abcam, Cambridge, MA) at 4°C. Visualization was performed with the HISTAR detection system (AbD Serotec, Duesseldorf, Germany) according to the manufacturer's protocol. For double staining, detection was performed with EnVision Detection Systems peroxidase/3,3′-diaminobenzidine (rabbit; Dako) and with alkaline phosphatase (AP) and Fast Red. Sections were counterstained with hematoxylin and coverslipped with a mounting medium for light microscopy.

Activity Score for HSCs.

Histological scoring for HSCs according to necrosis, infiltrates, inflammation, and cholestasis was performed as described previously by Feldstein et al.20

Detection of LS and Group Formation.

Upon the diagnosis of ALF, LS was evaluated with transient elastography (FibroScan, Echosens, Paris, France). After a mean duration of 7 days, transient elastography was repeated. Two groups of patients were defined according to the course of LS: group 1 included patients with an increase in LS between the first and second examinations, whereas group 2 comprised patients in whom LS had decreased.

Data Analyses and Statistics.

Data are presented as means and standard errors of the mean for numerical variables or as counts and percentages for categorical variables, unless stated otherwise. Univariate analyses were performed to determine maximum values during the course of disease for clinical parameters associated with death or transplantation [i.e., age, INR, serum creatinine, gamma glutamyl transferase (γGT), AP, bilirubin, liver enzymes, and Model for End-Stage Liver Disease (MELD) score]. Multivariate analyses also included the gender. For statistical evaluations, we performed the following: (1) two-sided unpaired t tests; (2) two-sided t tests with Welch's correction in case the variances' F test results differed with statistical significance for the body mass index, aspartate aminotransferase (AST; glutamic oxaloacetic transaminase), alanine aminotransferase (ALT; glutamic pyruvic transaminase), INR, serum creatinine, γGT, AP, or bilirubin; and (3) the Whitney-Mann test for the MELD score and age. Calculations were carried out with GraphPad Prism software (GraphPad Software, San Diego, CA) or, in the case of multivariate analyses, with SPSS version 12.0.1 (SPSS, Inc., Chicago, IL).

Results

ALF Etiologies and Outcomes.

The demographic parameters, etiologies, and outcomes of all enrolled patients are listed in Table 1. Drug toxicity was the leading cause of ALF (n = 9, 31%), and it was closely followed by viral hepatitis (n = 8, 28%). Other etiologies of ALF were mycotoxicosis (n = 3, 10%) and acute liver injury due to right-sided heart failure (n = 2, 7%). Three patients (10%) had antinuclear autoantibodies and thus exhibited autoimmune hepatitis, and in four cases (14%), the etiology of ALF remained indeterminate.

Of the 29 patients enrolled, 2 (7%) died within 1 month after the diagnosis of ALF, and 2 (7%) underwent urgent liver transplantation because conservative therapy failed, which made a second LS measurement impossible. All 25 remaining patients (86%) showed complete recovery with antiviral, immunosuppressant, and/or supportive medical therapy. Interestingly, both fatalities occurred in group 1, which was defined by an increase in LS over time.

Laboratory Parameters and Clinical Features.

All standard laboratory parameters (ALT, AST, γGT, AP, bilirubin, and INR) were dramatically increased above normal values. These elevated liver enzymes were expected as a general feature of ALF and demonstrated massive tissue injury. The distribution of patients into groups with increasing or decreasing LS did not reveal any significant differences in the clinical parameters between these groups (for detailed results, see Table 2).

Table 2. Selected Functional Liver and Kidney Parameters in the Course of ALF

	Reference Value	Total	Group 1: Increase in FibroScan or Transplantation	Group 2: Decrease in FibroScan
Data are presented as means of peak values and standard deviations.
Bilirubin (mg/dL)	0.3-1.2	14.8 ± 8.7	20 ± 7.7	10 ± 6.8
INR	1.0	2.7 ± 2.4	3 ± 3.4	2 ± 0.6
AST (IU/L)	0-34	4169 ± 5038	4043 ± 3637	4273 ± 6065
ALT (IU/L)	0-34	3372 ± 3201	3589 ± 2383	3196 ± 1679
γGT (IU/L)	0-34	264 ± 258	312 ± 250	225 ± 267
AP (IU/L)	25-100	224 ± 202	348 ± 271	223 ± 252
Creatinine (mg/Dl)	0.6-1.1	1.8 ± 1.9	2 ± 2.7	1 ± 0.7
MELD	—	25 ± 9	25 ± 11	24 ± 5

ALF Is Associated with Significantly Increased Apoptosis and Necrosis and Elevated Markers of Regeneration.

Sera of ALF patients revealed significantly increased levels of cytokeratin-18 and cytokeratin-18 neoepitope (i.e., M65 level for controls = 234.4 ± 36.6 U/L, M65 level for ALF = 3732 ± 839.9 U/L, P = 0.0005; M30 level for controls = 99.2 ± 16.4 U/L, M30 level for ALF = 2243 ± 559.6 U/L, P = 0.0012; Fig. 1A). These data indicate massive cell death due to apoptosis and necrosis in ALF. During hospitalization, markers of overall and apoptotic cell death diminished over time (Supporting Fig. 1), although differences between the groups were not observed (not shown).

Figure 1. Markers of cell death, regeneration, and fibrosis are significantly elevated in ALF. (A) Serum M65 and M30 levels, reflecting overall cell death and apoptotic cell death, respectively, were increased in ALF patients versus controls. (B) The serum concentration of IL-6 as a proinflammatory and proregenerative cytokine was raised markedly (the dotted line marks reference values of the standard method in healthy adults; the data are presented as means and standard errors of the mean). (C) FasL, a proapoptotic and proregenerative signaling ligand, was elevated in patients' sera in comparison with healthy controls. (D,E) TIMPs are established markers for ongoing liver fibrosis; they protect activated HSCs from apoptosis. MMPs are essential enzymes for extracellular matrix remodeling. Serum concentrations of (D) TIMPs and (E) MMPs in ALF patients versus healthy controls are presented as medians with 25th and 75th percentiles and ranges. (F) Hyaluronic acid, a marker of liver fibrogenesis, was significantly increased in patients with ALF. Data are presented as medians with 25th and 75th percentiles and ranges.

To check whether cell death might be associated with liver regeneration, we assayed regeneration-related serum markers in the ALF patients. IL-6 and FasL are well-established parameters of liver regeneration.21–23 As expected, both markers of liver regeneration were increased in the early phase of acute liver injury. On admission, serum values of IL-6 and FasL were clearly elevated in comparison with control and normal values, respectively (normal IL-6 level < 15 pg/mL, IL-6 level for ALF = 105.5 ± 55.65 pg/mL; FasL level for controls = 464.6 ± 27.59 pg/mL, FasL level for ALF = 913.8 ± 341.7 pg/mL, not significant; Fig. 1B,C). In addition, we prepared immunohistochemistry for the progenitor cell markers keratin-7 and keratin-19 and found progenitor cells within the liver tissue, especially around the portal areas (Supporting Fig. 2A,B). These data indicate a regenerative response of the liver tissue to abundant apoptosis, as shown previously.21

Serum and Intrahepatic Markers Demonstrate Fibrogenesis in ALF.

Although MMPs are involved in tissue remodeling, TIMP-1 and TIMP-2 promote fibrosis because they inhibit the action of MMPs and, moreover, protect activated HSCs from cell death. Both TIMPs were significantly increased in the sera of ALF patients in comparison with control individuals (TIMP-1 level for controls = 149.4 ± 8.3 ng/mL, TIMP-1 level for ALF = 629.9 ± 69.41 ng/mL, P < 0.0001; TIMP-2 level for controls = 124.6 ± 7.2 ng/mL, TIMP-2 level for ALF = 202.0 ± 14.8 ng/mL, P < 0.0001; Fig. 1D). These findings thus indicate ongoing fibrosis and HSC activation. In parallel, the ALF cohort revealed up-regulated MMP-1 and MMP-2 concentrations (MMP-1 level for controls = 2.47 ± 0.3 ng/mL, MMP-1 level for ALF = 5.95 ± 1.02 ng/mL, P = 0.0026; MMP-2 level for controls = 164.4 ± 7.89 ng/mL, MMP-2 level for ALF = 264 ± 32.52 ng/mL, P = 0.0057), whereas MMP-9 was only insignificantly increased in ALF (Fig. 1E). In addition, in ALF patients, there were significantly increased concentrations of hyaluronic acid as an indirect marker for collagen production (Fig. 1F). This overall elevation of profibrotic serum markers was paralleled by intrahepatic fibrosis. Collagen fibers specifically surrounded necrotic and/or apoptotic tissue and hepatocytes, as shown by Sirius Red staining. Collagen deposition was also observable in areas of ductular proliferation (Fig. 2A,B). In addition, we found abundant HSC activation, as marked by α-SMA expression, in areas of liver tissue in which massive cell death or proliferation had occurred (Fig. 2C,D). Costaining of caspase-3 as a marker of apoptosis and α-SMA revealed apoptotic cells around portal fields. α-SMA–positive cells were found abundantly in the tissue and also within close proximity of apoptotic cells (Fig. 2E,F). Scoring of HSC activity by histological parameters revealed an increased activity score in those patients with continuously increasing LS values (discussed later for group formation; Fig. 3A).

Figure 2. HSCs neighboring areas of massive cell death or proliferation are fully activated. Representative pictures originating from different patients are shown. (A,B) Sirius Red staining revealed collagen deposition, especially adjacent to necrotic areas and areas of ductular proliferation. (C,D) Immunohistochemistry for α-SMA, a marker of stellate cell activity, revealed widespread HSC activity (brown cells) in necrotic and proliferating tissue areas. (E,F) Double staining of caspase-3 (activated form) and α-SMA showed apoptotic cell death mainly in portal areas and in proximity to α-SMA–positive areas.

Figure 3. Cell death is associated with increasing LS and fibrosis. (A) The histological scoring of HSC activity demonstrated a higher rate of activated HSCs in group 1 patients (increasing LS between the day of admission and day 7). LS, measured by FibroScan, was used to discriminate between patients in group 1 in whom LS increased from admission with ALF to 7 days after admission and patients in group 2 with decreasing LS values during the same monitoring period. Depicted are cell death and fibrosis markers for both time points (admission with ALF and 7 days after admission). (B,C) Increasing LS was associated with continuously elevated serum M30 and M65 levels as markers of overall and apoptotic cell death, respectively (medians, 25th and 75th percentiles, and ranges are shown). Reductions in M30 and M65 paralleled decreases in LS, but the difference was not significant. (D,E) In addition, we found continuously elevated serum concentrations of TIMP-1 as a marker of HSC activity and fibrosis and increased ratios of MMP-2/TIMP-1 as a possible marker of fibrosis, respectively.

LS Is Increased in ALF.

We further clarified whether the findings in the serum and liver correlated with LS as measured by FibroScan. LS results from a combination of hepatocyte edema, bilirubin elevation, and intrahepatic collagen deposition. The degree of LS was indeed significantly increased in all ALF patients (Table 3). The dynamics of fibrosis were tested by comparative measurements over an interval of 1 week, whereupon the patients were divided into two groups according to increasing LS (n = 11) or decreasing LS (n = 16). Importantly, increasing FibroScan values were found in those patients who also had higher cell death and TIMP-1 values (Fig. 3B-D). During the observation period, cell death markers diminished in both patient groups, with a stronger effect in group 2, although the differences did not reach significance (Fig. 3B,C). Previous studies24, 25 had found that MMPs and TIMPs were insufficient as single prognostic markers of liver fibrosis, although TIMP-1 and MMP-2 exhibited correlations to fibrosis or cirrhosis. Thus, the ratio of MMP-2 to TIMP-1 has been suggested as a noninvasive marker of fibrosis.26 In our cohort of ALF patients, the MMP-2/TIMP-1 ratio was significantly decreased in group 1 individuals versus the controls and was still below that of the patients in group 2 on admission (1.19 ± 0.08 for the controls and 0.37 ± 0.07 for group 1, P <0.0001; 0.65 ± 0.19 for group 2, P = 0.02 versus controls; Fig. 3E). Strikingly, the ratio had not changed in group 1 7 days after the first examination, whereas a slight but insignificant increase was observed in group 2 (0.36 ± 0.06 for group 1 and 0.8 ± 0.17 for group 2, P = 0.03).

Table 3. LS as Determined by FibroScan Transient Elastography

	Reference for healthy individuals*	Total (n = 29)	Group 1: Increase in FibroScan (n = 11)	Group 2: Decrease in FibroScan (n = 16)
Data are presented as means and standard errors of the mean. * The data were taken from Foucher et al.33
LS at ALF diagnosis (kPa)	Up to 6	25.6 ± 3.0	22.2 ± 4.8	32.4 ± 5.9
LS 7 days after diagnosis (kPa)		20.9 ± 3.2	26.3 ± 4.3	12.9 ± 2.1

Fibrosis Markers and LS Are Correlated.

To determine if LS measured by FibroScan can predict the development of fibrosis in patients with ALF, we correlated serum fibrosis markers and LS values on admission and 7 days after admission. Although we found no significant correlations on the date of admission, on day 7, a strong correlation between LS values and serum levels of TIMP-1 [Spearman's rank correlation coefficient (r_s) = 0.526, P = 0.008], MMP-2 (r_s = 0.412, P = 0.04), and hyaluronic acid (r_s = 0.613, P = 0.002) was evident. Moreover, LS values and M65 as a marker of ongoing cell death were correlated significantly (r_s = 0.408, P = 0.04). Strikingly, the MMP-1 serum concentrations on admission were negatively correlated with LS on day 7 after admission (r_s = −0.528, P = 0.006), and this may indicate an early influence of MMP-1 expression on recovery from fibrosis. Altogether, the correlation of LS with three independent markers of fibrosis in serum clearly indicates ongoing liver fibrosis in ALF.

Discussion

Liver fibrosis is widely acknowledged as a damaging process in long-term liver disease with potential progression to cirrhosis and further sequelae that include liver cancer and hepatic failure. Recent evidence has indicated that fibrosis may be reversed upon the removal of the damaging agent.27–29 Such an understanding suggests that fibrosis is not an injurious process per se. Only if protracted over long periods of time (e.g., viral hepatitides, chronic ethanol intoxication, drug abuse, or obesity) are the excessive overproduction and deposition of collagen fibers within liver tissue harmful to the liver architecture. We thus argue that short-term production of collagen and the occurrence of fibrosis in acute liver damage may be a physiological and possibly beneficial response by the liver.

To the best of our knowledge, this is the first evidence of hepatic fibrogenesis during ALF. Our ALF patient cohort consistently revealed significant increases in serum TIMP and MMP levels as well as histological evidence for collagen formation and α-SMA expression. These findings clearly demonstrate ongoing profibrotic processes together with increased HSC activity. In addition, the use of transient elastography revealed increased stiffness in ALF patients. Earlier, others had interpreted similar findings15 as artifacts due to hepatocellular extension by liquid accumulation or liver swelling upon bile obstruction. In these cases, fibrosis had been excluded by biopsy. However, although individual variances among the pathologists' staging and grading criteria are disregarded, the fact that fibrosis is inhomogeneously distributed in the liver suggests that locally confined biopsy cannot reliably determine fibrosis.10, 30–32

The marked overexpression of TIMPs in ALF patients indicates potent MMP inhibition within the extracellular matrix.11, 12 Moreover, the compensatory increase in MMP expression is overcome by a parallel decrease in the MMP/TIMP ratio both during the onset and 7 days after the patients first presented with ALF. Strikingly, patients with decreasing LS also revealed significantly decreased serum TIMP-1 concentrations, whereas patients with increasing stiffness values exhibited continuously increasing TIMP-1 levels.

In addition, ALF patients showed excessive overall and apoptotic cell death, as verified by the M65 and M30 cytokeratin-18 markers. Consequently, the liver tissue must compensate for massive cellular loss. It is possible that a collagen matrix is synthesized and deposited as a structural framework to preserve the liver architecture. In acute circumstances such as ALF, fibrosis may serve as part of a beneficial wound healing process by transiently conserving the organ's structure until defective tissue areas are replaced by functional hepatocytes and accessory cells. This assessment is supported by the fact that initial elastography values in ALF patients are highly elevated in comparison with intact liver tissue and the levels of stiffness resemble those measured in liver cirrhosis.33, 34 Ultrasonographically and laparoscopically, the liver during ALF indeed presents as an inhomogeneous organ because of the formation of regenerative nodes whose surface resembles a cirrhotic liver. However, in stark contrast to cirrhosis, these nodes are completely remodeled into healthy liver tissue after ALF resolution.

It is also important that LS was reduced in parallel to an increasing MMP-2/TIMP-1 ratio in one group of ALF patients. This alteration can be ascribed to decreasing TIMP-1 levels rather than increasing MMP-2 levels. By promoting the survival of activated HSCs and dampening the expression and function of tissue MMPs, TIMP-1 serves as a crucial regulator of fibrosis. MMP-2, which was significantly increased in our patient cohort, was recently shown to promote HSC apoptosis by cleavage of N-cadherin as an essential HSC survival factor.35

Resolution of fibrosis is linked to apoptosis of activated HSCs, partly in a signal transducer and activator of transcription 1–dependent and natural killer group 2 member D–dependent manner triggered by natural killer cells.36, 37 Thus, the primary source of TIMP-1 is eliminated, and this allows for the degradation of fibrotic scar tissue and the regeneration of functional liver tissue.

The group with decreasing LS values mainly comprised patients with short-term liver damage due to drug toxicity or mycotoxicosis, with the exception of autoimmune hepatitis. In contrast, the patients showing increasing LS values had developed ALF mostly as a result of continuous liver injury by viral hepatitides or heart failure. This discrimination may thus be indicative of different damaging processes or may be a result of etiology-dependent time courses with respect to the damage and diagnosis. In addition, the two fatal outcomes occurred within group 1 (increasing LS values). Although this number appears low, recent data from Scandinavia suggest increasing patient survival after ALF.38 We believe the apparent differences in spontaneous survival among liver centers could be related to the various etiologies underlying ALF; for example, spontaneous survival may be more common after acetaminophen intoxication. In addition, intensive care has been greatly improved over the past years, and this has led to further reductions in ALF mortality. Surprisingly, the severity of ALF in the two groups did not differ significantly on presentation to our clinic.

In summary, our results suggest that ALF is accompanied by active hepatic fibrogenesis. Further investigations with larger cohorts of patients will be needed to clarify if the absence of LS increases could constitute a survival marker for ALF. Importantly, understanding the physiological roles of fibrogenic responses during ALF (i.e., nonspecific responses to organ demise versus active attempts at regeneration) could reveal novel opportunities for diagnosis and perhaps even treatment.