Outcomes and potential surrogate markers for future clinical trials of non‐alcoholic steatohepatitis cirrhosis

Abstract Non‐alcoholic steatohepatitis has emerged as a major public health problem, and the burden of non‐alcoholic steatohepatitis cirrhosis is projected to increase by 64%‐156% by 2030. The threat is aggravated by the fact that are currently no approved drugs for the treatment of non‐alcoholic steatohepatitis. In this paper, we review the main challenges to drug development in patients with non‐alcoholic steatohepatitis cirrhosis, and describe the opportunities brought by the advances in the understanding of the clinical and pathophysiological nuances of cirrhosis. The design of therapeutic regimens for non‐alcoholic steatohepatitis cirrhosis will vary according to the specific cirrhosis substage (compensated vs decompensated), and the specific mechanistic basis of therapy, targeted either at improving aetiology‐specific pathways and/or at more general aetiology‐agnostic processes. The understanding of the probabilistic expectations for the whole range of potential outcomes, rooted at different mechanistic drivers at each specific substage, will be essential in order to choose adequate estimands and therapeutic strategies for clinical trials and individual patients with non‐alcoholic steatohepatitis cirrhosis. Finally, we provide a summary of the main pitfalls and uncertainties in the design of clinical trials for non‐alcoholic steatohepatitis cirrhosis and discuss potential biomarkers for use in trials and practice for these patients.


| THE BURDEN OF CIRRHOS IS DUE TO NON -ALCOHOLIC S TE ATOHEPATITIS (NA S H)
Non-alcoholic fatty liver disease (NAFLD) has emerged as a major public health threat, and the burden of end-stage liver disease due to NAFLD is projected to increase from 64% to 156% by 2030. 1 All longitudinal studies conducted on biopsy-proven NAFLD cohorts over the past 2 decades have highlighted the number of patients with cirrhosis. The prevalence of cirrhosis is currently estimated to range from 0.6% to 65%, with a median value of 9% (Table 1). [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] Overall mortality in patients with NAFLD is predicted to be approximately 20%, while liver-related mortality is approximately 4% across all stages of fibrosis, showing a stepwise progression from mild to severe fibrosis. [17][18][19] By 2030, cirrhosis and end-stage liver disease related to NASH are projected to increase worldwide. 1 Models of future disease burden for incident decompensated cirrhosis have shown the highest percentage in France (164%) followed by the USA (150%). 1 In addition, projections from one Canadian study showed that cirrhosis due to NASH will increase in all birth cohorts by 2040, accounting for 73% of all new diagnoses of cirrhosis. 20 As shown in Table 2, the median proportion of cirrhotic patients with liver decompensation in longitudinal studies of biopsy-proven NAFLD cohorts is 13.4%, ranging from 3% to 45% across the different cohorts. [4][5][6]8,10,[12][13][14][15][16] Clinical events related to portal hypertension are the most common findings: ascites (median 8.2%, min-max: 0.4%-70%), variceal bleeding (median 8.6%, min-max: 0%-66.4%) and hepatic encephalopathy (median 4%, min-max: 0%-31.6%). Accordingly, the development of liver decompensation constitutes a healthcare burden that requires high resource utilisation, and is related to inpatient and short-term mortality. 21 Onset of hepatocellular carcinoma (HCC) represents another relevant clinical outcome, as well as occurrence of liver transplantation (median 0.5%, minmax: 0%-34.2%), that underlines the progressiveness of NAFLD.
Otherwise, cardiovascular disease (CVD) involves a low rate of events in cirrhotic patients compared with liver-related outcomes and to the relative weight of morbimortality due to CVD in pre-cirrothic stages. 16 One recent study in Wales 22 conducted in nearly 70 000 individuals affected by cirrhosis has shown that the incidence of NAFLD has increased 10-fold over the past 10 years and has become the predominant cause of liver damage. As compared with other aetiologies, the course of NASH-related liver disease appears to be milder, with a smaller proportion of decompensated patients (8% of patients with NASH cirrhosis, vs 25% of patients with alcohol-related cirrhosis). However, clinical outcomes extracted from current longitudinal studies need to be related to the rapidly increasing burden of NASH cirrhosis, which will eventually cause a considerable increase in liver-related events and mortality.

| CURRENTLY APPROVED ENDP OINTS IN NA S H CIRRHOS IS TRIAL S
There are currently no Food and Drug Administration (FDA)-approved drugs for compensated NASH cirrhosis. Patients with cirrhosis are the most likely to develop hard outcomes (eg death, HCC, liver transplantation). Current trials in patients with cirrhosis capture outcomes related to cirrhosis, such as hepatic encephalopathy, ascites and variceal haemorrhage, as well as the laboratory components of the Child-Pugh score. Improvement in histological fibrosis stage or hepatic collagen content have been also used as primary endpoints of outcome events in trials focused on NASH cirrhosis. However, the reversal of cirrhosis to lesser degrees of fibrosis is an ambitious goal, and the relationship between histological changes in cirrhosis and clinical outcomes has not been characterised. Improvements in fibrosis stage may reflect either true regression or sampling bias.
Currently, the only endpoints recommended by the FDA to support marketing approval in compensated NASH cirrhosis are clinical outcomes. 23 According to the FDA, Phase III trials in patients with compensated NASH cirrhosis should evaluate the effect of the investigational drug relative to placebo on the composite endpoint of time from randomisation to the first of any one of the following outcome events:

Key points
• Non-alcoholic steatohepatitis (NASH) represents a major public health problem, with the burden of NASH cirrhosis projected to increase 64%-156% by 2030.
• There are no Food and Drug Administration-approved drugs for NASH cirrhosis, and clinical outcomes are the only recommended endpoints for market approval.
• Therapeutic regimens will vary according to cirrhosis stage (compensated vs decompensated) and the mechanistic basis of therapy (aetiology specific vs symptomatic).
• A clear definition of NASH cirrhosis, understanding of liver disease biology and detailed patient risk stratification are required for future clinical trials.
• Non-invasive tests, eg enhanced liver fibrosis and liver stiffness, are promising biomarkers.

| WHI CH PATIENTS WITH NA S H CIRRHOS IS ARE ELI G IB LE FOR PHARMACOLOG IC AL THER APY AND HOW TO S ELEC T THEM: THE D IFFERENT S TAG E S OF NA S H CIRRHOS IS
The structure and focus of therapeutic regimens for NASH cirrhosis will vary according to 2 main factors: (i) the specific substage of cirrhosis at which the therapeutic strategy is directed, and (ii) the specific mechanistic basis of therapy-either targeted at improving aetiology-specific pathways (such as the metabolic derangements typical of NASH) or at more general aetiology-agnostic processes (such as the intrahepatic or extrahepatic vascular dysregulation common to all forms of portal hypertension). [24][25][26] The classical concept of cirrhosis is histological in nature, and has been traditionally regarded as a static, non-reversible last stage for all forms of progressive liver diseases. 24 However, both the success of antiviral therapies in cirrhosis, as well as the integration of clinical and haemodynamic knowledge generated in the past 2 decades, have helped to develop a more comprehensive, dynamic and TA B L E 1 Prevalence of cirrhosis, and overall and liver-related mortality in longitudinal studies of biopsy-proven NAFLD cohorts [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]  nuanced view of this last phase of chronic liver disease. 24 Nowadays, there are at least 4 distinct, well-differentiated substages of cirrhosis, classified according to the mechanisms driving progression and potential for regression of disease at each stage and the expected probabilities of a different range of outcomes. 24,26 Table 3  The compensated stage is the longest phase and can go undiagnosed given its asymptomatic nature. Nonetheless, early identification of patients at this stage would be of paramount importance, since this is precisely the phases at which therapies might be still able to prevent the transition to the decompensated stage, or even lead to a regression to pre-cirrhotic stages. Within the compensated stage, the presence and degree of portal hypertension is a key predictor of outcome. By integrating haemodynamic data, further distinction of prognostic relevance can be made into an early compensated stage and a late compensated stage. 24,26,27,30 The early compensated stage is characterised by a mild degree of portal hypertension (defined by an hepatic pressure venous gradient (HVPG) <10 mm Hg), and a low risk of CD (<10% at 4 years). 31,32 The late compensated stage is characterised by the presence of clinically significant portal hypertension (CSPH), defined by an HVPG ≥10 mm Hg, the threshold above which CD is up to 4 times more likely to occur. 31,32 The distinction between the early compensated stage and CSPH is not merely prognostic but also carries pathophysiological and therapeutic implications. In the early compensated stage, the main drivers of progression and outcomes are aetiology-specific. 25,26 There seem to be intrahepatic vascular adaptations, both structural and functional-the latter being characterised by a moderate increase in intrahepatic vascular resistance (IHVR)-all of which translates into mild portal hypertension. 30,33 Thus, within this stage, therapeutic efforts should be directed towards arresting the driving mechanism of NASH. The underlying assumption (extrapolated from the lessons learnt from antiviral therapies [34][35][36][37], is that the improvement of those aetiology-specific mechanisms will carry over indirectly into the arrest (or even regression) of fibrosis and the functional improvement of IHVR, leading to a decrease of portal pressure and the risk of CD.
Within this clinical and pathophysiological framework, the choice of outcomes in this specific early stage should focus on histological improvement of NASH features and fibrosis as the main endpoint (as in current Phase III trials). Progression to CSPH and/or resolution of portal hypertension could be considered as secondary outcomes or for earlier Phase II trials. 25 In order to assess the impact of hard clinical endpoints in this early NASH cirrhosis population, large sample sizes and a longer duration of follow up are likely to be required, as seen in current Phase III/IV trials. 32,38 Once the critical CSPH threshold is reached, the risk of CD increases considerably, 27,31,32 and therefore the prevention of CD should become the main therapeutic goal and the ideal endpoint for trials. 25 During this late compensated stage, structural intrahepatic changes are characterised by the presence of thicker fibrotic septa, thus making the regression of fibrosis unlikely. 24 Also, extrahepatic adaptations start to develop, with an increase in portal blood flow, the development of collateral circulation and systemic adaptations leading to the initial phases of a mild hyperdynamic circulatory state. 39,40 Once these adaptations appear, splanchnic vasoconstrictors start to become effective at halting the progression of portal hypertension and reducing the risk of decompensation. 40,41 Extrapolating from experience with antiviral therapies, it can be assumed that for NASH cirrhosis, there might also be a 'point of no return' somewhere after CSPH is established, for which the effects of aetiology-specific therapies are no longer able to reverse CSPH and eliminate the risk of CD without concomitant medication. [34][35][36][37] Under these pathophysiological assumptions, therapeutic efforts in this late compensated phase should aim to target the same aetiology-specific mechanistic pathways as well as both intrahepatic and extrahepatic vascular tone. As a matter of fact, in patients already harbouring oesophageal varices, the use of non-selective beta-blockers is the standard of care 25 and cannot be avoided as a comparator for trials. 26 The transition of HVPG values to <10 mm Hg could be considered as a potentially relevant surrogate endpoint in these patients, 25,26 but histological changes in NASH and fibrosis become less likely 32 and are thus not ideal endpoints.
Finally, the occurrence of CD signals that intrahepatic and extrahepatic structural and functional adaptations are insufficient and therefore that the patient has progressed to NASH cirrhosis. 24,30 The risk of further decompensation, progression to end-stage liver disease and death increases exponentially as decompensating events accumulate. 26  The recent proposal to substitute the term NASH by Metabolic

| Biology of liver disease in cirrhosis
The drugs currently in development for NASH can be divided into 3 broad categories: metabolic, anti-inflammatory and antifibrotic.
Favourable metabolic effects from a drug are desirable in patients with NASH, but to what extent the partial correction of these metabolic abnormalities leads to a more favourable outcome in cirrhosis is unknown. The biology of clinical decompensation in patients with previously compensated cirrhosis is not fully understood. For example, diabetes arising in liver cirrhosis (so-called hepatogenous diabetes) has a profound impact on the pathology and natural history of the liver disease. 54 Although peripheral insulin resistance and impairment of the hepatocellular function are two potential major causes, the beta-cell capacity plays a critical role. Recent evidence that the failing liver exerts an independent 'toxic' effect on betacells suggests that individuals with hepatogenous diabetes might benefit from interventions aimed at improving beta-cell function, such as thiazolidinediones and incretins. 55 On the other hand, cirrhosis is a condition of intense muscle protein wasting leading to sarcopenia even in individuals affected by obesity (so-called sarcopenic obesity). 43 Malnutrition and sarcopenia are associated with higher complication rates in patients with cirrhosis. 43 Moreover, they are associated with increased mortality in hospitalised patients with cirrhosis and those waiting for liver transplantation. 43 In this setting, drugs promoting weight loss may have a detrimental impact on the progression of cirrhosis. Therefore, it remains to be determined whether treatment of the underlying steatohepatitis as opposed to the fibrosis will prevent clinical decompensation. These possibilities are currently under active investigation.

| Risk stratification
As described above, cirrhosis represents a broader spectrum of disease compared with its usual histological classification (F4), and the main challenge is the stratification of NASH patients with compensated cirrhosis to identify strata for decompensation that represent  Abbreviations: NASH: non-alcoholic steatohepatitis; T2DM: type 2 diabetes mellitus.

| Placebo reponse and Hawthorne effect
with specific recommendations. 59 Trial duration may also affect the impact of the Hawthorne effect. Long trail durations, as currently required for phase 3 trials in NASH cirrhosis, might mitigate the potential impact of a Hawthorne effect (since lifestyle modifications are harder to maintain in the long term), but this is yet to be proven.
Other factors such as sample size or geographic location must be also considered when interpreting the results of placebo arms in RCTs.

| Improved histological readings
In an attempt to mitigate the inter and intra-observer variability in NASH histology readings discussed in the first section, improved machine-driven methods to quantify basal amount of fibrosis and changes in time are being increasingly explored in the last few years.
A recent paper evaluating a machine-learning (ML) based-approach with paired biopsies from 3 large NASH RCTs including patients with advanced fibrosis (STELLAR-3, STELLAR-4 and ATLAS) 60 showed that these artificial intelligence-driven techniques are sensitive and reliable, and represent promising approaches to correlate dynamic changes in fibrosis (even in the F4 stage) with clinical outcomes, although the number of events in those clinical trials was very small and further studies are required to increase the confidence in this novel and exciting techniques. The demonstration that ELF score and liver stiffness had the ability to identify those most likely to progress from F3 to cirrhosis, in addition to predicting which patients with cirrhosis at baseline were most likely to have a liver-related event, 61 was of particular interest.

| New biomarkers in NASH cirrhosis
The authors defined a change in ELF score of 0.5 points as the value that correlated with clinical liver events. This value also correlated with other non-invasive tests, liver biochemistry tests, glycaemic parameters, CK-18 values, serum bile acid values and body weight, but not with histological features. These findings give hope that in the future, reliance on histological endpoints will be a historical concept for NASH clinical trials. Validation from such biomarkers will hopefully be a reality as clinical events accumulate in ongoing Phase III trial efforts.
Functional testing is a novel approach to assess the severity of liver disease and changes in actual liver function in the context of treatment trials, especially in patients with cirrhosis. The HepQuant test simultaneously measure clearance from portal and systemic circulation as well as portal-systemic shunting. 62 Preliminary data indicate that this assessment of liver impairment correlates with hard endpoints, but more extensive validation is needed. Another test is the methacetin breath test, which is being evaluated in several ongoing clinical trials for NASH, 63,64 and longitudinal data that correlate this assessment with histological changes may be forthcoming. How the HepQuant, methacetin breath test and other tests of specific metabolic functions of the liver compare with clinical outcomes remains to be determined.

| CON CLUS IONS
NASH cirrhosis has emerged as a major healthcare problem and finding effective therapies for patients with NASH has become one of the main unmet clinical needs in hepatology. Drug development in this field will face important challenges, due to the time needed to gather hard outcomes and the unreliable nature of fibrosis as an endpoint. A deeper understanding of the clinical nuances of NASH cirrhosis, and the mechanistic processes driving disease progression at each substage, will hopefully guide therapeutic efforts and appropriate patient selection, informative biomarkers and clinically meaningful endpoints.

ACK N OWLED G EM ENTS
Funding was provided by Boehringer Ingelheim. Boehringer Ingelheim did not have a role in the study design, analysis or interpretation of the data, the writing of the manuscript or the decision to submit the manuscript for publication.

CO N FLI C T S O F I NTE R E S T
Jesús Rivera-Esteban reports no conflicts of interest. Angelo Armandi reports no conflicts of interest. Salvador Augustin is now an employee of Boehringer Ingelheim but was not when the manuscript was written. Elisabetta Bugianesi reports advisory board activity for Boehringer Ingelheim, Bristol Myers Squibb, Gilead, Innova, Intercept, Novo Nordisk.

R E S E A RCH E TH I C S A N D PATI E NT CO N S E NT
Not applicable.