A Placebo-Controlled, Multicenter, Double- Blind, Phase 2 Randomised Trial of the Pan-Caspase Inhibitor Emricasan in Patients with Acutely Decompensated Cirrhosis

Background: Cirrhosis and acute-on-chronic liver failure (ACLF) are associated with systemic inflammation, and caspase-mediated hepatocyte cell death. Emricasan is a novel, pan-caspase inhibitor. Aims of this study were to assess the pharmacokinetics, pharmacodynamics, safety and clinical outcomes of emricasan in acute decompensation (AD) of cirrhosis. Methods: This was a phase 2, multicentre, double-blind, randomised trial. The primary objective was to evaluate the pharmacokinetics, pharmacodynamics and safety of emricasan in patients with cirrhosis presenting with AD and organ failure. AD was defined as an acute decompensating event ≤6 weeks’ duration. Patients were randomised proportionately to emricasan 5 mg bid, emricasan 25 mg bid, emricasan 50 mg bid or placebo. Treatment was continued to 28 days, or voluntary discontinuation. Results: Twenty-three subjects were randomised, of whom 21 were dosed (placebo n = 4; 5 mg n = 5; 25 mg n = 7; 50 mg n = 5). Pharmacokinetic data showed 5 mg dose was associated with low plasma levels (<50 ng/ml), and 25 mg and 50 mg doses showed comparable pharmacokinetic profiles. Therefore, for analysis of secondary endpoints, placebo and 5 mg groups were merged into a ‘placebo/low-dose’ group, and 25 mg and 50 mg groups were merged into a ‘high-dose’ group. Five deaths occurred amongst the 21 patients, all due to progression of liver disease (2 in placebo/low-dose, 3 in high-dose). No statistically significant changes from baseline MELD score or CLIF-C ACLF score were noted between placebo/low-dose and high-dose groups at day 7 (MELD —1 vs —1, CLIF-C ACLF 0.7 vs 0.8). An initial reduction in cleaved keratin M30 fragment was noted between placebo/low-dose and high-dose groups (percent relative change: day 2: —11.6 vs —42.6, P = 0.017, day 4: —3.5 vs —38.9 P = 0.017) although this did not persist to day 7 (—3.1 vs —20.8, P = 0.342). Conclusion: This study demonstrates that emricasan is safe and well tolerated in advanced liver disease. However, this study fails to provide proof-of-concept support for caspase inhibition as a treatment strategy for ACLF. Trial registration: EudraCT 2012-004245-33 ( J CLIN EXP HEPATOL 2017;XX:1–11)

The natural history of cirrhosis is characterised by progression to episodes of acute decompensation (AD) of liver function.1 Most patients with cirrho-sis and AD are treated successfully in the majority of cases.However, about 30% of these patients develop hepatic and/or extra-hepatic organ failure that progresses in about 20% to multi-organ failure and death.2 When this occurs rapidly, within a period of weeks, the condition is referred to as acute-on-chronic liver failure (ACLF). The 170,000 cirrhosis deaths in Europe each year are largely due to ACLF and the condition costs $3Bn in the USA and£50 K per survivor in the UK.3 There are as yet no specific therapies for ACLF.The pathobiology of ACLF is characterised by hepatic and systemic inflammation, and progressive, unrelenting hepatocyte injury and death.4,5 Approaches targeting sys- temic inflammation have been tried, such as anti-TNF therapies, although these have led to negative outcomes suggesting that alternative approaches are required.6Apoptosis is a highly regulated form of or programmed cell death. In response to injury or inflammation, hepa- tocytes can undergo apoptosis via an extrinsic pathwayactivated by death ligands, Fas, and tumour necrosis factor-related apoptosis-inducing ligand (TRAIL), or an intrinsic pathway activated by intracellular stress of mem- brane-bound organelles, such as lysosomes, endoplasmic reticulum and mitochondria.

Both pathways of apo- ptosis converge on the caspases (cysteine aspartyl pro- teases), which play an essential role in the initiation, execution and regulation of apoptosis.11Previous work has demonstrated that apoptosis is a key pathway of cell death in patients with ACLF. Fragmented chromatin and caspase-dependent cleaved keratin 18 are both terminal end-products of the apoptotic pathway.12 Adebayo et al. have shown that serum levels of cleaved keratin 18 (M30 fragment) are significantly elevated in patients with ACLF compared to patients with AD alone, and correlate with disease severity.13 Immunohistochem- istry of liver tissue also demonstrates apoptotic M30-pos- itive hepatocytes in patients with severe ACLF. Similarly, Cao et al. have demonstrated elevated levels of fragmented chromatin in the ACLF compared with the AD group.14 These data support the central role of hepatocyte apopto- sis in the progression of ACLF, and provide the rationale for therapeutic targeting of apoptosis in ACLF.Caspases provide a druggable target for the inhibition of apoptosis in liver disease. As such, caspase inhibitors have been shown to decrease liver injury in rodent models of acute liver failure, fatty liver disease, cholestatic liver injury and alcohol-induced liver injury.15–20Emricasan (IDN-6556) is a novel, orally active, pan- caspase protease inhibitor. Emricasan has been studied in eight phase 1 studies and eight phase 2 studies involv- ing over 650 patients, providing initial safety data and supporting the dose range of 5–50 mg used in phase 2 studies in patients with chronic liver disease.

The aims of this study were to assess the pharmacokinetics, phar- macodynamics, safety and clinical outcomes of emricasan in patients with cirrhosis and a rapid deterioration of liver function associated with organ failure.Study IDN-6556-02 (ClinicalTrials.gov NCT01937130) was conducted in accordance with Good Clinical Practiceguidelines and the Declaration of Helsinki. Written informed consent was obtained from all patients before enrolment, in agreement with approved protocols from research ethics committees (North West-Haydock Research Ethics Committee, reference 13/NW/0464).Study DesignThis was a phase 2, multicentre, double-blind, randomised trial was conducted in 10 sites in the United States and United Kingdom. The primary objective was to evaluate the pharmacokinetics, pharmacodynamics and safety of emricasan, orally administered for 28 days in patients with cirrhosis and a rapid deterioration of liver function asso- ciated with organ failure (Figure 1).The relationship between emricasan dose and biomarker responses had been thoroughly characterised in subjects with active HCV hepatitis and normal hepatic func- tion.21,22 However, the dose-biomarker response relation- ship had not been characterised in subjects with impaired hepatic function. The initial studies in subjects with HCV hepatitis and normal function assessed the effect of emri- casan upon a panel of 4 biomarkers (caspase 3/7, M30, ALT and AST) at oral doses ranging from 0.5 mg BID up to 200 mg BID. Doses as low as 0.5 mg BID dose were pharmacodynamically active, decreasing M30 and ALT by nearly 50%, but had less effect upon AST and caspase 3/7. Doses greater than 50 mg BID did not appear to have any greater reduction in the 4 biomarkers compared to the 50 mg BID dose. Thus, doses of 5, 25 and 50 mg BID were selected for this study in patients with severe hepatic impairment.

The sample size was based around a simulation exercise of 1000 simulated outcomes, which were used to predict the dose required for a set target exposure. A sample size of 15 subjects per group (60 total) would have resulted in >80% of simulated outcomes recommending the correct dose totake forward to a follow-on ACLF efficacy study, if the exposure and pharmacodynamic effects in cirrhoticsubjects from this study were similar to that observed in subjects with active HCV hepatitis and normal hepatic function previously studied (data modelled from Pockros et al.21 and Shiffman et al.22).Eligible patients included those 18 years of age or older, with stable compensated or decompensated cirrhosis, presenting with an acute deterioration of liver function and associated organ failure. Cirrhosis was diagnosed by clinical, radiological and/or histological means. An acute deterioration of liver function was defined as an acute decompensating event or illness of ≤6 weeks’ duration, on the background of cirrhosis (see definitions below).Recent hospital admission (within 4 weeks) for a compli- cation of cirrhosis, greater than two-organ failure, HIV infection, uncontrolled bacterial infection, pre-existing chronic kidney disease, autoimmune liver disease, active malignancy aside from hepatocellular carcinoma, need for mechanical ventilation, inability to obtain consent, and haemodynamic instability (including use of inotropes, aside from terlipressin for hepatorenal syndrome).Criteria for organ failure used in this study were:Liver: bilirubin ≥ 12.0 mg/dL Kidney: creatinine ≥ 2.0 mg/dLCerebral: hepatic encephalopathy West Haven grade III or IV Coagulation: INR ≥ 1.7.

In terms of organ failure, eligibility for this study was on the basis of one of the following: (i) renal failure (defined as creatinine ≥2.0 to ≤ 3.4 mg/dL); (ii) other single organ failure with (a) renal impairment (defined asan increase in creatinine of >0.3 mg/dL from an estab- lished baseline level) and/or (b) hepatic encephalopathygrade I or II; (iii) two organ failures. As noted above, patients with respiratory failure requiring mechanical ventilation, and circulatory failure requiring inotropic support, were excluded. Additionally, MELD score and CLIFC-ACLF score were calculated as previously described.23,24The Sponsor, investigational staff and subjects were all blinded to the treatment group assignment. Patients were randomised proportionately into four groups: emricasan 5 mg bid, emricasan 25 mg bid, emricasan 50 mg bid and matching placebo bid through the study’s Interactive Web Response System. The randomisation schedule was gen- erated using a validated randomisation program and veri- fied for accuracy using quality control procedures.Treatment was continued for 28 days in all four groups. Randomisation was stratified on the basis of a positive alcohol history, and also prior steroid use. A positive alcohol history was defined as ongoing alcohol use up until the time of entry into the study. Treatment was continued for 28 days unless the patient voluntarily dis- continued treatment.

Routine haematology and biochemistry parameters were monitored, as well as blood tests for pharmacokinetic and biomarker evaluations. Pharmacokinetic studies were performed at day 1 and day 4 following initiation, and serum biomarkers of apoptosis (cleaved keratin 18 (M30 Apoptosense1; Peviva, West Chester, OH, USA) and full-length keratin 18 (M651; Peviva, West Chester OH, USA)) were also collected at days 1, 2, 4 and 7. Follow-up visit was scheduled at 4 weeks following the end of treatment, and telephone contact at 12 and 24 weeks.Statistical analyses of the change from Baseline in M30, M65, M30/M65 ratio, MELD and CLIF-C ACLF score were conducted using an analysis of covariance (ANCOVA) adjusted for Baseline. Given the distribution of data, M30 and M65 were log-transformed prior to analysis and results were back-transformed to provide a relative percent change from Baseline. No adjustments for multi- ple comparisons were made. Associations between changes in cell death biomarkers and efficacy endpoints were assessed using Pearson’s correlation coefficient.The decision to pool the placebo and 5 mg emricasan for purposes of efficacy analyses was made after the pre- mature termination of the study and unblinding. The goal of pooling those groups was to increase the number of subjects for summary purposes. The justification for pool- ing was that the 5 mg emricasan dose had no detectable pharmacodynamic effect on the biomarkers that were measured (see “Results” section).

RESULTS
Between 22 January 2014 and 17 September 2014, 23 subjects were randomised into this study, of whom 21 were dosed (placebo n = 4; emricasan 5 mg bid n = 5; emricasan 25 mg bid n = 7; emricasan 50 mg bid n = 5) at 10 sites. The study was stopped once adequate phar- macokinetic samples had been obtained since the study was slow to recruit. The CONSORT diagram is presented in Figure 2. Baseline characteristics of these patients by individual randomised treatment group is presented in Supplementary Table 1.A summary of the key pharmacokinetic data is presented in Table 1. The geometric means of the emricasan AUC0–8, Cmax, and AUC0–last increased in an approximately dose-proportional manner between the 5 mg and 50 mg doses on day 1 and day 4. No plasma accumulation was apparent in any of the treatment arms on day 4 compared to day 1. Generally, lower between-subject variability was observedin all PK parameters in the lower dose treatment arms than in the 50 mg arm, with coefficients of variation (CVs) ranging from 28% to 48% in the IDN-6556 5 mg and 25 mg groups, and from 99% to 258% in the IDN-6556 50 mg group.Changes in the keratin fragment M30 (mean SD) and caspase 3/7 (mean SD) are shown in Figure 3. In this study, changes in caspase 3/7 activity (Figure 3b), rather than M30 levels (Figure 3a), were a more sensitive indica- tor of emricasan pharmacodynamic activity. The placebo and 5 mg groups had no significant effect upon M30 levels on any study day, although there was a trend towards reduction in M30 with the 25 mg and 50 mg doses between days 2 and 4.

Caspase 3/7 activity also showed a non-significant trend in reduction with the 25 mg and 50 mg doses by the day 2 visit and throughout the initial 7 days of the study, while the placebo and 5 mg dose had no effect (Figure 3b). Since the 5 mg emricasan group had no detectable pharmacodynamic effect, the placebo and 5 mg groups were merged into a ‘placebo/low-dose’ group, and the 25 mg and 50 mg groups were merged into a ‘high- dose’ group. A statistically significant decrease in log- transformed M30 at day 2 and day 4 was noted between the placebo/low-dose and high-dose groups. These results were back-transformed to provide percent relative change from Baseline (Day 2: —11.6 vs —42.6, P = 0.017, Day 4:—3.5 vs —38.9, P = 0.017). The direction of treatmenteffect remained in favour of the high-dose group at day 7; however this was not statistically significant (—3.17 vs—20.8, P = 0.342). No statistically significant differences between the high-dose and placebo/low dose were seen for the change from baseline in log-transformed M65 at any time point up to day 7. No consistent correlations were seen across timepoints in associations between changes in cell death biomarkers, caspase 3/7 activity and efficacy endpoints.As evident in the CONSORT diagram (Figure 2), of the 21 subjects dosed, 7 completed 28 days of administration. Of the remaining 14 patients, 3 died as a consequence of their liver disease and 6 were lost to follow-up. The patient baseline characteristics, summarised by the pla- cebo/low-dose and high-dose groups, are presented ininfection was identified as a further precipitant in 29% of cases. No significant differences in severity of liver disease at baseline were noted, as assessed by MELD score or CLIF- C ACLF score (where ACLF was present).Patient disposition and mortality are presented in Figure 2.

A total of 5 deaths occurred amongst the 21 patients dosed during the study period, although 2 deaths were after completion of the 28-day treatment period. All deaths were due to progression of liver disease – 2 in the placebo/low-dose group and 3 in the high-dose group. For the remaining secondary efficacy endpoints, data was analysed up to the day 7 time point, in view of the attrition of patient numbers at day 28.Changes in MELD score and CLIF-C ACLF score are presented in Table 4. No statistically significant differ- ences in these parameters were noted between the placebo/ low-dose and high-dose groups at day 7 (MELD —1 vs —1, CLIF-C ACLF 0.7 vs 0.8). Similarly, no specific changes in liver, kidney, brain or coagulation function were noted at day 7, by CLIF-C ACLF score parameter.As demonstrated in Figure 2, there were a total of 10 deaths across all treatment groups. There were 5 on-study deaths, 2 of which were during the one-month follow-up period following the full course of treatment, and a further 5 were registered via serious adverse event (SAE) reporting follow- ing discontinuation or study completion. All of these deaths were attributed to progressive liver disease.Adverse events (AEs) were reported by 17 of the 21 patients, of whom 13 patients reported SAEs. None of the SAEs was determined to be treatment-related. The only AEs deemed to be treatment-related were nausea and vomiting which were reported by one placebo subject. The AEs and SAEs are presented in Table 5.

DISCUSSION
Excessive cell death has been demonstrated to play a fundamental role in the progression of ACLF. HepatocyteTable 4 Changes from Baseline in M30, M65, M30/65, MELD, and CLIF-C ACLF Score.Placebo/5 mg (n = 9) 25 mg/50 mg (n = 12) P-valuen Median (range) LSMean n Median (range) LSMean%Relative change from baseline (95% CI) %Relative change from baseline (95% CI)cCK18 (M30) Day 27425 (284–1999) —11.6 (—32.0, +14.9) 11560 clinical models, apoptosis has been shown to promote inflammation and fibrogenesis.25–27 Caspase inhibitors have been shown to protect from liver injury in a rodent model of acute liver failure, and to attenuate inflammation and fibrosis in rodent models of cholestatic, alcohol and fatty liver dis- ease.15,16,19 This study represents the first use of the cas- pase inhibitor emricasan in patients with decompensated cirrhosis.The primary objective of this study was to evaluate the pharmacokinetics, pharmacodynamics and safety of emri- casan in this population. All emricasan doses led to >10- fold higher levels of exposure than observed in previous studies of either healthy volunteers or non-cirrhotic sub-jects with active HCV hepatitis, at the same doses, likely as consequence of porto-systemic shunting and associated bypass of first-pass hepatic uptake. Despite this, none of the reported adverse events were deemed to be treatment related, although, as discussed above, there was no attri- tion of subjects at later time points.

The patient deaths and adverse events reported generally reflected the very severe liver disease in the study population.From a pharmacodynamic perspective, the 5 mg dose had negligible biologic effects upon the biomarkers mea- sured (M30, M65 and caspase 3/7) in this study, and as a consequence, the placebo and 5 mg dose groups were merged for the analysis of secondary efficacy endpoints. However, it is important to note that in non-cirrhotic subjects with active hepatitis C infection and presumably normal hepatic function, the 5 mg dose resulted in Cmax concentrations that were over 10-fold lower than observed in this study and yet decreased M30 levels ~50% (Study A8491010, unpublished data). The 5 mg dose has also been shown to significantly decrease ALT levels.21,22 The failure to see significant pharmacodynamic effects of the 5 mg dose in this study was not related to inade- quate circulating drug concentrations but likely due to porto-systemic shunting and severe hepatocellular func- tional impairment. These factors, either separately or col- lectively, could result in low, sub-therapeutic hepatocyte levels of emricasan particularly since emricasan is a high first-pass extraction drug.Statistically significant reductions in the keratin frag-ments M30 were noted at day 2 and day 4 following therapy when the 25 mg and 50 mg doses were analysed together, compared with the combined placebo and 5 mg group. However, this reduction was not sustained to 7 days or beyond.

Potential contributing reasons for the failure to achieve statistical significance include insuffi- cient power of the study due to the low recruitment of subjects, high attrition rates, poor compliance, and con- sequent shortened duration of follow-up for analysis.Despite the reduction in apoptosis markers, the results of this study failed to provide proof-of-concept supportfor the hypothesis of targeting apoptosis to prevent pro- gression of ACLF. Although both the 25 and 50 mg doses of emricasan decreased caspase 3/7 activity, proving that although emricasan was able to inhibit its biological tar- get, only the 50 mg dose clearly decreased M30 levels, a marker of apoptosis, and there were no significant changes in MELD score, CLIF-C ACLF score or CLIF-C organ function score at 7 days. There are several possible reasons for this, including the above-noted comments of an underpowered study population, shortened duration of analysis and sub-therapeutic hepatic drug levels due to porto-systemic shunting. However, a further possibility is the complexity of modes of cell death involved in the progression of liver injury in ACLF. Aside from apoptosis, other forms of hepatocyte cell death have been described in ACLF, including autophagy, necroptosis and pyropto- sis.28 The relative contribution of these, or predominant mode of cell death with different mechanisms of liver injury, remains to be established.

Moreover, it remains unclear as to whether apoptosis is causal in the progres- sion of ACLF or is an adaptive response to liver injury. In an open-label study, Frenette et al. have recently reported that emricasan significantly decreased cleaved keratin lev- els and serum ALT levels in a population of patients with cirrhosis after six months of treatment at the 25 mg bid dose.29 Moreover, the subgroup of patients with a baseline MELD scores ≥15 had a significant improvement in MELD score, bilirubin and INR over the 3-month study period. Therefore, further, studies are warranted to address the efficacy of emricasan in advanced cirrhosis and ACLF. Amongst patients with hepatitis B related ACLF, levels of the M30 apoptosis marker were higher in survivors than non-survivors, suggesting that apoptosis may also be an adaptive response to liver injury, possibly involved in liver regeneration.30There are several weaknesses with this study. Only 23patients were recruited from the planned study popula- tion of 60 subjects on the basis of the difficult enrolment. Additionally, the population was predominantly alcohol- related liver disease, thus limiting the applicability of the findings of this study to other aetiologies of cirrhosis. The follow-up data from this cohort was also limited, and less than half of the subjects completed 28 days of follow-up, hence limited conclusions can be drawn regarding longer- term secondary endpoints addressing the efficacy of emri- casan in ACLF. Despite these limitations, the study con- clusions are likely to be valid.

In conclusion, this study demonstrates that emricasan is safe and well-tolerated in advanced liver disease, although hepatic drug levels may be sub-therapeutic due to porto-systemic shunting. Although reductions in caspase 3/7 activity and M30 component of the cyto- keratin fragments were detected, this was not associated with any clinical benefit in this small study. The efficacy Caspase inhibitor of emricasan in advanced cirrhosis and ACLF was not confirmed. Further mechanistic work is required to delineate the role of other modes of cell death, such as necroptosis and pyroptosis, in the progression of ACLF.