Cilofexor

Bile acid modulators for the treatment of nonalcoholic steatohepatitis (NASH)

KEYWORDS : Bile acids; Farnesoid-x-receptor; non-alcoholic steatohepatitis; bile acid modulators; FXR-based therapies; GPBAR1- based therapies; FXR ligand; obeticholic acid; non- alcoholic fatty liver disease; NAFLD

1. Introduction

Nonalcoholic fatty liver disease (NAFLD) is a spectrum of highly prevalent human disorders characterized by an excessive accu- mulation of lipids in hepatocytes (hepatic steatosis) in subjects that were not exposed to significant alcohol consumption and drugs [1]. NAFLD is categorized histologically into nonalcoholic fatty liver (NAFL) and nonalcoholic steatohepatitis (NASH). While NAFL is a simple steatosis with no evidence of hepatocellular injury, NASH is characterized by the presence of steatosis, inflam- mation with hepatocyte injury (ballooning) with fibrosis. In con- trast to NAFL, NASH is a progressive disease, that, in parallel with the development of diabetes and obesity pandemics, has become the leading cause of liver disease on a global scale and is projected to increase further over the next 10–15 years, with a higher proportion of patients developing fibrosis, the main determinant of both liver-related and overall mortality [2]. In Western countries, NAFLD development is essentially linked to a sedentary lifestyle and dietary habits and is strongly associated with metabolic syndrome (insulin resistance, increased body mass index, hypertension, and dyslipidemia) and its clinical cor- relates, obesity and type 2 diabetes mellitus [3]. For this reason, it has been suggested that renaming NAFLD as MAFLD (metabolic associated fatty liver disease) would lead to a better classifica- tion/understanding of this condition [4]. The idea that MAFLD is the liver counterpart of the metabolic syndrome does not explain why NAFLD can develop in the absence of obesity, the so-called lean-NAFLD, but has the merit to set the frame of therapeutic requirements that would be needed to treat this condition. Indeed, while NAFL, a disease with mild insulin resistance, can be approached essentially by life-style modifications and is not considered as a suitable pharmacological target, NASH, repre- sents an unmet clinical need, and, similarly to type II diabetes, a pharmacological treatment should be considered when benefit from lifestyle modifications are not achieved or sustained. The therapeutic approaches to the NASH patients will be grounded by the presence and severity of metabolic syndrome and the staging of liver fibrosis [5].

Currently, there are no approved drugs for the treatment of NASH. Several pathogenic mechanisms have been identified in NASH, and animal studies have provided evidence that disease progression/reversion could be obtained by a variety of approaches targeting very different biochemical mechanisms [6,7]. In the real life, however, the goal of treatment is to prevent the progression to cirrhosis and reduce the need for liver transplantation and improve survival [8]. Because in addition to metabolic syndrome, the liver fibrosis is the main determinant of prognosis in NASH patients, medical treatments are indicated in patients with clinically relevant fibrosis [5–8]. However, liver fibrosis is a slowly progressive condi- tion and development of liver cirrhosis and its complication requires decades, making the process of drug discovery and devel- opment for NASH extremely challenging [8]. Development of vali- dated surrogated endpoints is therefore essential.

2. Bile acids

Bile acids are a family of atypical steroids generated in the liver and intestine by the coordinated activity of mammalian and bacterial genes [9]. Two main families of bile acids can be identi- fied in humans. The primary bile acids, i.e. cholic acid (CA) and chenodeoxycholic acid (CDCA) are generated in the liver from cholesterol by a chain of reaction that involves the products of 27 genes [10]. These primary bile acids are then secreted in the biliary tract and transported to the gallbladder to be released into the duodenum in response to food ingestion. In the intes- tine, CA and CDCA undergo a series of enzymatic modifications that are catalyzed by bacterial enzymes, to generate a second group of steroids, known as secondary bile acids (or degener- ated), i.e.lithocholic acid (LCA) from CA, and deoxycholic acid (DCA) from CDCA (Table 1). In addition, to these four species, other bile acid species can be found in physiological states in humans. These are, the ursodeoxycholic acid (UDCA) that is a tertiary bile acid in human and is generated by the 7OH epimerization of CDCA. Additionally, CA could be hydroxylated in the liver to generate the hyocholic (HCA) which is then dehy- droxylated in the intestine to hyo-deoxycholic acid (HDCA) [9]. The various bile acid species can be found in humans tissues in several chemical arrangements, i.e. as free bile acids (also indi- cated as non-conjugated) or conjugated bile acids, also indicated as bile acid salts, that are generated in the liver by conjugation (side chain amidation) of primary and secondary bile acids with glycine and, to a minor extent, with taurine [9,10]. This step leads to the formation of two amidated derivatives for each one of the various bile acids (Table 1). In specific settings, bile acids can be conjugated with the glucuronide, a reaction mediated by the UDP glucoronosyltransferases (UGT1A1, 2B4, and 2B7), or can be sulfated by the sulfotransferases (SULT2A1 and SULT2A8) at position C3 and C7 [9,10]. In general, however, the amidated bile salts represents the chemical arrangement of the large majority of bile acids found in the blood, liver, and small intestine. In contrast, the largest proportion of free bile acids is found in the colon and feces. Other bile acids could be found in specific liver diseases (such as the oxo-bile acids) but their role is limited to cholestatic disorders and will not be discussed here [9]. Of relevance rodents have specific bile acids, i.e. the α- and β- muricholic acids (MCA), that are primary bile acids, generated in the liver from CDCA, and ω-MCA, a secondary bile acid generated in the intestine by the 7α-dehydroxylation of the α- and β-MCA, that are not found in humans (Table 1 and Figure 1). Accordingly, while the bile acid pool in humans consists of CA (∼40%), CDCA (∼40%), DCA (∼20%), with a glycine and taurine conjugation ratio of 3 to 1; in mice, the bile acid pool consists of TCA (∼60%) and Tα- MCA and Tβ-MCA (∼40%) and >90% of bile acids are taurine conjugated [9].

2.1. Synthesis and recycling of endogenous bile acids

Bile acids are produced in hepatocytes by two main metabolic pathways known as ‘classic’ and ‘alternative’ [9,10]. The cho- lesterol-7α-hydroxylase (CYP7A1) is the first and rate-limiting enzyme in the classic pathway and converts cholesterol to 7α- OH-cholesterol; while the alternative pathway initiates with the C27 hydroxylation of cholesterol by the sterol 27- hydro- xylase (CYP27A1) [10]. In the intestine, several bacterial species operate additional biotransformations of primary bile acids [11–14]. The first biotransformation consists in the enzymatic hydrolysis of the C-24 N-acyl amide (deconjugation) and is operated by bile salt hydrolases (BSH)-expressing bacteria (Bacteroides, Lactobacillus, Bifidobacteria, and Clostridium among others) [11]. This step is followed by the 7α- dehydroxylation by 7α-dehydroxylase expressing bacteria (mainly Clostridium and Eubacterium) giving rise to 3α-mono- hydroxylated bile acids (i.e. LCA from CDCA), and 3α-12α- dihydroxylated bile acids (i.e. DCA from CA) [11,13,14]. Additionally, the C7 β-epimerization by Bacteroides, Clostridium, Escherichia, Eubacerium (and others) generates the 3α,7β-dihydroxy-5β-cholanoic acid, UDCA (Table 1) [11]. The intestinal bile acids are reabsorbed in the terminal ileum and transported back to the liver through the portal vein, completing a cycle in the so-called ‘entero-hepatic circulation’, a highly integrated process that allows the conservation and recycling of bile acid species, maintaining the stability of the bile acid pool and its regulation [9].

2.2. Bile acids receptors

Bile acids are ligands for various receptors, collectively known as bile acid activated receptors (BAR), a family of cell surface and nuclear receptors [reviewed in [15]]. The two best characterized BARs are the Farnesoid-x-receptors (FXR), discovered in 1995 and de-orphaned in 1999 as a nuclear transcription factor activated by primary bile acids [16–18] and the G Protein Bile Acid Receptor (GPBAR)-1, also known as TGR5, a seven-transmembrane G-protein coupled receptor discovered in 2002 [19,20] (Table 2). In addition, bile acids can activate other nuclear receptors including the Pregnane-x-receptor (PXR)], the Constitutive androstane receptor (CAR), the vitamin D receptor (VDR) and the liver-x-receptors (LXRα and β) and other G-protein coupled receptors including the sphyn- gosine 1-phosphate receptor (SP1), and others (we will not discuss these receptors in this review) [15].

Figure 1. Role of FXR and GPBAR1 ligands in NASH.Hepatic FXR-regulated lipid and glucose homeostasis. Preclinical studies have shown that activation of hepatic FXR represses the synthesis of triglycerides and fatty acid (FA) synthesis as a result of SHP-dependent inhibition of SREBP-1 c, which is a positive regulator of genes involved in lipogenesis, acetyl-CoA carboxylase (ACC), fatty acid synthase (FAS), and stearoyl CoA desaturase (SCD). Conversely, FXR activation induces PPARα thus promoting FA β-oxidation. Furthermore, FXR promotes the clearance of TG-rich lipoproteins inducing VLDL receptor and human syndecan-1. Additionally, activation of hepatic FXR represses the expression of PEPCK and G6Pase while increases the phosphorylation of GSK3β, promoting the activity glycogen synthase (GS). Overall, these changes result in decreased hepatic gluconeogenesis, decreased plasmatic glucose levels, and increased hepatic glycogen synthesis.
Gpbar1 in not expressed in hepatocytes, but its activation in the enterocytes promotes the release of glucagon-like peptide 1 (GLP-1) from intestinal ‘L’ endocrine cells, which in its turn, regulates insulin secretion by pancreatic β-cells. Moreover, Gpbar1 promotes energy expenditure positively regulating PGC1α via CREB-activation.

Additionally, FXR exerts anti-inflammatory effects through SHP-dependent, inhibition of the recruitment of NF-kB on the promoter of several pro-inflammatory genes, and SHP-independent by stabilization of the binding of NCor1 complex on the promoter of pro-inflammatory genes. GPBAR1 promotes a tolerogenic phenotype by activating a GPBAR1-PKA-CREB axis in macrophages thus inducing the release of IL-10 of an anti-inflammatory cytokine. Moreover, GPBAR1 increasing the phosphorylation of FOXO1 and negatively regulates the expression of CCL2. Furthermore, in hepatic stellate cells (HSCs) FXR activation negatively regulates collagen deposition at least in part by SHP-dependent repression of TGF-β/SMAD3 and JunD/AP-1 pathway (reviewed in 6).

In human CDCA is the most potent ligand for FXR, while in mice this role is exerted by CA (Table 2). Secondary bile acids, LCA, and DCA, are preferential ligands for GPBAR1, but there is a significant overlap of activity of various bile acids toward the two receptors, suggesting that a functional selectivity is obtained essentially by spatial-temporal gradients, in addition to receptors affinity. The receptors are differentially expressed through the intestine, with both reaching the higher expres- sion in the ileum (Table 2). Further on, while GPBAR1 is not expressed by parenchymal liver cells, FXR is mainly expressed by hepatocytes (Table1). Both receptors are expressed by adipocytes. GPBAR1 is also expressed by immune cells, macro- phages, and dendritic cells and non-parenchymal liver cells such as Kupffer cells and sinusoidal cells [21,22]. In mice, the two main bile acids Tα- and Tβ-MCA, are FXR antagonists [23,24]. In contrast, UDCA is a weak GPBAR1 agonist [25] and neutral toward FXR (Table 2).

3. Role for FXR and GPBAR1 in metabolic syndrome and NASH

At least two decades of research have shown that FRX and GPBAR1 regulate essential steps of lipid and glucose metabo- lism in liver, intestine and adipocytes, either directly or by promoting the release of potent regulatory factors such as Fibroblasts growth factors (FGF)-15 (FXR) and GLP-1 (GPBAR1) [26,27]. These activities have been reviewed exten- sively recently by us and others, and are summarized in the Figure 1

4. Current status of FXR and GPBAR1 ligands in NASH

There are several FXR and GPBAR1 agonists that are currently investigated for their potential in treating NASH. According to their selectivity toward the two receptors they can be classi- fied in the tree main groups: GPBAR1 selective agonists, FXR selective agonists, and GPBAR1/FXR dual agonists (Table 2). In addition, FXR antagonists are under evaluation. From the chemical standpoint they can be further sub-divided in ster- oidal (bile acid and non-bile acid derivatives) and non- steroidal agonists [reviewed in [28,29]].

4.1. GPBAR1 agonists

GPBAR1 deficient mice, have a relatively neutral phenotype in term of their metabolic activities. Despite the fact that GPBAR1 has been suggested as a potent regulator of energy balance [30], GPBAR1−/- mice do no gain weight spontaneously in comparison to their congenic littermates [31]. One potential bias of this phenotype however is that these mice are hyper- active and appear to use more energy because a more intense activity [32]. We and other have shown that GPBAR1 activation in mice fed high-fat diet ameliorates vascular function, reduce atherosclerosis, and attenuates liver fat deposition [33–35]. Preclinical studies have shown that several GPBAR1 ligands might hold utility in the treatment of NASH or metabolic syndrome, including INT777 [33] and INT767 (dual FXR and GPBAR1 ligand) both from Intercept Pharmaceuticals (one of the authors of this article, Prof. Fiorucci is a joint inventor of INT767). Additionally, BAR501 (a selective GPBAR1 ligand) and BAR502 (a dual FXR and GPBAR1 ligand) both from BAR Pharmaceuticals (Prof. Fiorucci and Prof. Zampella are co- inventors of these agents) attenuates liver fat deposition and fibrosis in mice fed a high-fat diet [34–38]. In addition, there are numerous natural GPBAR1 ligands (see Table 2) including oleanolic acid, betulinic acid, ursolic acid, and others [28]. There are no GPBAR1 agonists in clinical development.

4.2. Selective FXR agonists

Several FXR agonists are currently developed for treating NASH. The list includes both steroidal and non-steroidal ligands (See Table 2) [28,29]. Additionally, also FXR antagonists have shown beneficial effects in preclinical models of NASH [23,24]. The Figure 2 illustrates some of the potential beneficial effects of FXR agonism in NASH [reviewed in [7]].

4.2.1. Steroidal FXR agonists

The most advanced FXR agonist is obeticholic acid, a semi- synthetic derivative of CDCA originally developed at the University of Perugia as 6-ethyl-CDCA and INT-747 [39] by investigators that later co-founded Intercept Pharmaceuticals (Roberto Pellicciari and Stefano Fiorucci). The obeticholic acid is the first in a class of FXR ligands that has reached a clinical stage [40–42] and was originally approved in 2016 as a second-line treatment for primary biliary cholangitis (PBC). This agent has recently completed a Phase 3 trial the REGENERATE study (Randomized Global Phase 3 Study to Evaluate the Impact on NASH with Fibrosis of Obeticholic Acid Treatment) for the treatment of patients with NASH and biopsy-proven fibrosis [43]. The results of this trial were that 23.1% of patients with NASH and F2 or F3 fibrosis who received 25-mg obeticholic acid and 11.9% of subjects who received placebo (P = 0.0002) demonstrated a ≥ 1 stage
improvement in fibrosis without worsening of NASH. Pruritus, a common side effect of obeticholic acid, occurred in 51% of subjects receiving the active ingredient and 18% of subjects receiving placebo. Treatment discontinuation due to pruritus occurred in 9% of patients treated with obeticholic acid and 1% in the placebo group. Other side effects such gallstones and acute cholecystitis were more frequent in the obeticholic acid group than in the placebo (3% and <1%, respectively). Treatment with obeticholic acid associated with an increase in LDL cholesterol. The later results were consis- tent with the results of the Phase 2 FLINT study [44]. EDP-305 (Enanta Pharmaceuticals) is a steroidal, non–bile acid, FXR agonist (Table 2) that is being currently evaluated [44,45] in a Phase 2a study to assess the safety, tolerability, pharmacokinetics, and efficacy in NASH patients (NCT03421431). In a Phase 1 multiple ascending dose study, EDP-305 was shown to activate FXR has demonstrated by increased levels of FGF19 and decreases bile acid synthesis [45]. Increased incidence of pruritus was observed in subjects who received the 20-mg dose of EDP-305 relative to placebo. This dose was also associated with reductions in HDL and total cholesterol, but not increases in LDL, in subjects with pre- sumptive NASH. 4.2.2. Non-steroidal FXR agonists Several second-generation non-steroidal FXR agonists are cur- rently under investigation in NASH (see Table 2) [28,29]. Cilofexor (GILEAD), previously known as GS-9674, originally developed by Phenex Pharmaceuticals as Px-201 [29], is a potent (EC50 43 nM) non-steroidal FXR agonist currently advanced through pre-approval trials by Gilead Sciences. In phase 2b trial cilofexor 30 mg and 100 mg obtained a 3.7% and a 24.6% relative median reduction in liver fat compared to placebo (P < 0.001) [46]. These effects were associated with trends toward a reduction in ALT, though these did not reach statistical significance. Trends toward increased total choles- terol and LDL cholesterol were also observed. Moderate to severe pruritus was observed in 14.3% of subjects who received 100 mg cilofexor, compared with 3.6% of placebo. Because some lack of efficacy and a dose-dependent develop- ment of pruritus, cilofexor is now advanced as a part of a combination strategy together with selonsortib and firsoco- stat in an ongoing phase 2 trial: Safety and Efficacy of Selonsertib, Firsocostat, Cilofexor, and Combinations in Participants With Bridging Fibrosis or Compensated Cirrhosis Due to Nonalcoholic Steatohepatitis (NASH). This trial (the ATLAS trial) was designed to investigate the safety, tolerability, and efficacy of monotherapy and dual combination regimens of cilofexor 30 mg, firsocostat 20 mg, an acetyl-CoA carbox- ylase inhibitor, and selonsertib 18 mg in patients with advanced fibrosis (F3-F4) due to NASH (NCT02781584). Selonsortib was discontinued because of lack of efficacy. However, the results of the combination of cilofexor and firsocostat were considered sufficient to promote a phase 3 trial: https://www.gilead.com/news-and-press/press-room/press- releases/2019/12/gilead-announces-topline-results-from-phase-2-atlas-study-in-patients-with-bridging-fibrosis-f3-and- compensated-cirrhosis-f4-due-to-nonalcoholic-steatohepatitis). In the meantime, a study published in the November 2019, reported no increase of pruritus in PBC patients treated with 100 mg and 30 mg of cilofexor in comparison to placebo (figures were 14%, 20%, and 40% respectively) [47]. The NOVARTIIS FXR agonist, tropifexor, previously known as LJN452 [29], is currently investigated in combination with cenicriviroc, a CCR2 antagonist in patients with NASH and fibrosis (The Tandem study). Results from a stand-alone ther- apy in patients with NASH, the FLIGHT-FXR Phase 2 study (NCT02855164), have shown efficacy in reducing the liver fat content by 5.4% and 10.7% at 60 μg and 90 μg doses, respectively (ttps://www.novartis.com/news/media-releases/novartis-data-show-tropifexor-ljn452-significantly-improves- several-key-biomarkers-nash-patients-moderate-severe-fibrosis- released on 11 November 2019). Similarly, ALT was reduced from baseline by 8.2% and 11.4% (placebo adjusted). Tropifexor administration was associated with a mild increase in LDL and a decrease in HDL and pruritus was reported in 14% and 8% of subjects at the 60-μg and 90-μg doses, respectively, compared to 7% in the placebo group [48].Finally, TERN-101 (Lilly) is another non-steroidal agonist of FXR that similarly to AGN 242,268 (Allergan) and MET409 (Metacrine) is currently undergoing a phase 1 evaluation [29]. While Px-102 from Phenex has been advanced to a Phase 2a in NAFLD patients. Similarly, in Phase 2, are nidu- fexor (Novartis) and EYP001 (Enyo Pharma) [29]. 4.2.3. Clinical results with FXR agonists The results of phase 2 and 3 with obeticholic acid (the Flint study and the REGENERATE) study [43,44] have shown that this agent holds some potential in the treatment of NASH, as it reduced the steatosis scores and the slowing down the progression or rever- sing the liver fibrosis [43]. In the REGENERATE trial the severity fibrosis was reduced by at least 1 point in a significant proportion of patient. The conclusions of the study were: ‘(the) fibrosis improvement endpoint was achieved by 37 (12%) patients in the placebo group, 55 (18%) in the obeticholic acid 10 mg group (p = 0.045), and 71 (23%) in the obeticholic acid 25 mg group (p = 0.0002)’ [43]. In contrast, obeticholic acid failed to improve the NASH features [43]. These results were somewhat in contrast with a previous Phase 2 trial, the FLINT study was an improve- ment of some NASH parameters was, in contrast, achieved [44]. Side effects, such pruritus occurred in 51% of patients treated with 25 mg [43]. The results of the REGENERATE study were in agreement with another Phase 2 trial, carried out in Japan and reported only in a concise form by Intercept Pharmaceuticals [Intercept Pharmaceuticals Press Release. Intercept pharmaceuticals announces results of the phase 2 trial of OCA in NASH patients in Japan. GLOBE NEWSWIRE, 28 October 2015], showing a substantial lack of efficacy of obeticholic acid at the dose of 10 and 20 mg in treating patients with NASH. A dose of 40 mg achieved statistical significance versus the placebo treated harm with 38% versus 20% of subjects showing at least 2 points improvement in NASH score. However, significant differences most of the histology endpoints were not meet, including inflammation improvement, ballooning resolution, and NASH resolution or fibrosis reduction. A dose-dependent increase in pruritus and changes in LDL, HDL, and cholesterol were observed and were consistent with data reported in the FLINT trial [44]. In short, amelioration of steato- hepatitis scores has been obtained in the phase 2 FLINT trial but not in the Japanese study (where only patients treated with 40 mg showed some improvement) and not in the phase 3 trial [43]. Side effects in contrast, occurred at the same levels in all these trials, were dose-dependent, and increases with doses greater than 10 mg. Results from phase 2 trials with other FXR agonists have provided signals that while FXR agonism could be effective in treating steatosis and liver inflammation in NASH, but side effects including pruritus have manifested in all these trials. 4.2.4. Side effects of FXR agonists in NASH The results of Phase 2 and 3 trials, and the results of studies carried out in PBC patients indicate that the use of obeticholic acid associates with several side effects [40–42]. In all the three trials carried out in NASH patients side effects occurred in a dose-dependent manner [43,44]. Pruritus was a common side effect occurring in up to 51% of patients treated with 25 mg in the REGENERATE trial. The rate of itching in patients with PBC exposed to obeticholic acid is even higher [reviewed in ref [49]]. Additional side effects observed with obeticholic acid were changes in metabolic parameters including increased levels of cholesterol and LDLc and reduction of HDL. Results of phase 2 trials with other FXR agonists have confirmed this trend. In a phase 2 trial, cilofexor (GS-9674) was shown effective in reducing liver steatosis at the dose of 100 mg, but side effects (pruritus) also manifest at this dose. A dose of 30 mg cilofexor was found safer but also less effective. Cilofexor will likely be developed as a combination therapy and used at the dose of 30 mg in combination with other agents as described above. Similarly, tropifexor or and EDP-305 have shown positive results in phase 2 trials, but also pruritus and changes in metabolic parameters, i.e. increase in LDL-c, manifested in a dose-dependent manner. These data suggest that pruritus, along with an increase in LDL-c and reduction in HDL, are class-related effect of FXR agonists. In addition to these side effects, in 2017 a cluster of major side effects, including liver failure requiring intensive care therapy and liver transplantation were reported in cirrho- tic PBC patients administered obeticholic acid. The searching of the FDA adverse Events Reporting System (FAERS) on February 2020, retrieves over 4000 side effects linked to the use of obeticholic acid in the United States in the years 2106–2019, with 920 cases considered serious side effects including 163 deaths https://fis.fda.gov/sense/app/d10be6bb- 494e-4cd2-82e4-0135608ddc13/sheet/45beeb74-30ab-46be- 8267-5756582633b4/state/analysis). The severity of these effects has to lead the FDA to issue a drug safety communica- tion on February 1, 2018, [source: FDA – Drug safety commu- nication 02/01/2019] and a warning was added to the Ocalive stating that ‘Health care professionals should follow the Ocaliva dosing regimen in the drug label, which is based on calculating a Child-Pugh score in PBC patients with suspected liver cirrhosis before treatment to determine their specific classification and starting dosage’. Further on: ‘Dosing higher than recommended in the drug label can increase the risk for liver decompensating, liver failure, and sometimes deaths.’ The warning highlights that in patients with liver cirrhosis the initial dose of obeti- cholic acid should not exceed 5 mg once a week. However, the most likely dose that will be used in NASH is 25 mg/day. How safety concerns and this dose-related adjustments will impact on the development of obeticholic acid in NASH remains unclear. In addition to pruritus, several metabolic side effects have been linked to FXR agonism in NASH patients. These include an increase in cholesterol and LDLc and a reduction of circu- lating levels of HDLc [49]. These metabolic side effects were identified earlier in animal studies with obeticholic acid and have been confirmed by results of the FLINT study, a phase 2 clinical trial [50]. In a post hoc analysis of this study, however, it was found that metabolic changes were reversed in a subgroup of patients that were treated with atorvastatin [50]. An ad hoc study designed to confirm the role of statin in controlling cholesterol metabolism in NASH patients receiv- ing obeticholic acid, the CONTROL study [51], has confirmed this view. The CONTROL study was a small size (84 patients), phase 2 study carried out in NASH patients receiving placebo or 5, 10, or 25 mg obeticholic acid for 16-week and atorvasta- tin (10 mg/day) in combination with atorvastatin starting from the 4 weeks. Results of this study shows that, while any dose of obeticholic acid increases LDLc and LDL particle concentra- tion (LDLpc) after 4 weeks, these effects were reversed by 10 mg/day atorvastatin, while no further benefit were observed by higher dose of this statin. Again, pruritus occurred in 55% of patients treated with 25 mg obeticholic acid [51]. Several mechanisms have been linked to LDLc increase in response to FXR activation, including reduced liver expression of LDL receptor, leading to reduced LDLc clearance [52]. Additionally, FXR agonists might increase cho- lesterol plasma levels because their repressive effects on bile acid synthesis. 5. Tissue restricted, intestinal FXR agonist One approach that has been taken to limit potential side effects linked to generalized activation of FXR/GPBAR1, is to target the receptors in a tissue-specific manner. The intestinal FXR (iFXR) has been identified as one of these targets, for several reasons [53]. First of all, the iFXR is involved in regulat- ing intestinal and liver metabolism by inducing the release of Fibroblasts Growth Factor (FGF) 15 (mouse)/19 (human) from enterocytes [54]. FGF19 exerts a number of regulatory effects on lipid and glucose homeostasis and enhances energy expenditure [26,55,56]. FGF19 analogues have been proven effective in treating NASH. In addition, poorly absorbable FXR ligands, such as fexaramine [53], have been shown effec- tive in proof-of-concept studies in rodent models of obesity [53]. With some surprise, the beneficial effects of fexaramine have been linked to a shift in bile acid composition leading to GPBAR1 activation that manifests as increased energy expen- diture due to brown adipose tissue activation, browning of white adipose tissues. These effects were abrogated by GPBAR1 gene ablation. 6. FXR antagonists Of relevance also iFXR antagonism by glycine-β-MCA, or bile acid sequestrants, reduces blood glucose and ameliorates insulin sensitivity [23,24], and have been proposed for the treatment of the metabolic syndrome. Intestinal FXR agonists are of interest because of the close interplay between bile acids and the intestinal microbiota [57]. Furthermore, obeti- cholic acid, has been shown effective in shaping the intestinal microbiota [58]. 7. Dual GPBAR1 and FXR agonists Dual GPBAR1 and FXR knockout mice show a profound dysregu- lation of bile acid homeostasis and are prone to develop liver fibrosis [59]. BAR502 [28], is a dual GPBAR1/FXR agonist devel- oped by BAR Pharmaceuticals, that is slightly preferential for GPBAR1 (Table 1). Preclinical studies have shown that this agent exerts beneficial effects in models of NASH [60,61]. A phase I trial assessing the safety and pharmacokinetic of BAR502 in healthy volunteers is due by 2020. Another dual FXR and GPBAR1 agonist is INT767, that has been mentioned before. 8. Conclusion Development of FXR agonists for treating NASH has been proven challenging. Nevertheless, several FXR agonists are currently under development. According to early studies, the main benefit of obeticholic acid seems to be its ability to reduce the severity of liver fibrosis [62], while the metabolic impact remains unclear. In contrast, the effect of other FXR agonists on fibrosis needs to be firmly established. Several dose-dependent, class related, side effects have been observed in NASH patients administered an FXR agonist. In addition to pruritus, increased cholesterol plasma levels and LDLc along with the reduction of HDL have been reported. These metabolic effects might be mitigated/reversed by co-treating patients with statins. Furthermore, although stains are currently recommended in the treatment of NASH patients with cardiovascular risk [63] the clinical impact of FXR agonists per se on NASH patients with known cardiovascular risk remains unknown. Liver toxicity related to statins should also be consid- ered. An important information obtained from these studies is that the side effects of FXR agonists are dose-dependent. This has shifted the focus on the use of lower doses of FXR agonists in combination with agents that target different molecular mechanisms, including ACC or CCR2 inhibitors and statins. It appears that the most likely future scenario of FXR agonists in the NASH arena will be their combination with existing or cur- rently developed drugs directed toward different molecular mechanisms, tailoring these therapies to specific patient subsets. 9. Expert opinion Development of FXR agonists for treating liver diseases has proven to be challenging. Historically the prototype FXR agonists have been the GW4064, a trisubstituted isoxazole originally developed by investigators at GSK in early years of this century [64]. However, GW4064 has served only as a tool compound because its limited in vivo bioavailability and short half-life [29]. Further on, GW4064 harbors a stilbene moiety in its chemical structure that is photolabile and potentially endowed with potent estrogenic effects due to the activation of the Estrogen- Related Receptor alpha (ERRα) [29]. Working on this isoxazole ‘hammerhead’ trisubstituted, Phenex Pharmaceuticals has generated several FXR agonists, and two of them (cilofexor and Px-103) are currently undergoing phase 2 and 3 trials [29]. Using another approach, i.e. development of semisynthetic derivatives of natural bile acids, we have reported in 2002 the generation of a semisynthetic derivative of CDCA, the 6-ethyl-CDCA [39], later renamed INT747 [62] which has been then christened as obeti- cholic acid by Intercept Pharmaceuticals in 2009. Obeticholic acid has been the first in a class of FXR ligand reaching the clinical stage, and has been approved for clinical use in 2016, as a second-line treatment for UDCA-resistant PBC patients. Obeticholic acid has been advanced for NASH and is the first FXR agonist completing a phase 3 trial in NASH patients [49,65]. Following these leads, several other FXR agonists have been developed and advance to clinical trials (Table 2). The results of these investigations, however, have shown that FXR activation in vivo associates with side effects. The burden of these side effects, pruritus and negative impact on cholesterol, LDLc and HDLc metabolism, has promoted a reconsideration of the role of FXR agonists in the field of NASH. There are several questions, however, that need to be answered. First of all, are the side effects observed with various FXR ligands, linked to FXR activation? There some doubts that this is the case, since it is well known that chemically different FXR ligands might bind the receptor in a different manner, leading to a variable modulation of FXR-regulated genes. For some of the currently available FXR ligands, such as obeticholic acid, specific PK properties impact on the safety profile. Obeticholic acid is a CDCA derivative and its glyco- and tauro-conjugates have a very high rate of intestinal absorption (>95%) leading to recy- cling in the entero-hepatic circulation and liver-body accumula- tion. In contrast, glyco- and taurine-conjugated of isoxazole tri- substitutes such as cilofexor have very poor intestinal uptake and perhaps these agents might be used to selectively modulate liver FXR. If so, would intestine restricted FXR agonists be better tolerated than ‘full’ or liver restricted FXR ligands? There might be a chance that a second/third generation of iFXR agonists could be developed with an improved safety profile? Also, will an iFXR agonist more effective than currently available agents in modulating the fibroblast growth factor (FGF)19 pathway [65]?
Second question: could available FXR agonists proposed as a stand-alone therapy in the treatment of NASH? Based on available information on efficacy and side effects, we believe that the answer is mostly negative. The various FXR agonists currently investigated exert variable effects on lipid metabo- lism, which suggests that association of an FXR agonist with lipid-lowering agents [63] or insulin and insulin/sensitizers would be required to overcome the relatively modest efficacy of these agents on lipid and glucose metabolism. This approach will also allow the use of lower doses of FXR ago- nists to limit side effects.

Related to the previous issue, there is a third important ques- tion: if a multi-therapy should be developed, what will be the ideal combination? This question remains largely un-answered now and there is a need to investigate efficacy and side effects of various FXR agonists in combinations with either existing drugs (i.e. statins, PPARs, fibrates) and drugs which are currently under development (ACC inhibitor, CCR2 antagonists, among others). This is an uncharted territory because the different PK and PD properties of various FXR agonists do not allow precise predictions. Extensive preclinical studies in validated models of NASH are needed to test alternative approaches and limit the risk of drug failure in clinical trials.

Fourth. There is an urgent need for standardized and pre- dictive preclinical models of NASH.Fifth. NASH is a heterogeneous disease with a very robust metabolic component. In most patients, agents that target insulin signaling are needed. It is unclear to what extent the combination of FXR agonism with insulin, insulin sensitizers or GLP-1 analogues will improve treatment efficacy. Identification of patient subsets based on predictive biomarkers is needed. Sixth. Use of validated, noninvasive, biomarkers for liver fibrosis is another point that will speed up clinical trials in NASH.

Finally, will the development of dual FXR and GPBAR1 ligands helpful in reducing side effects linked to FXR activation? GPBAR1 agonists improve energy expenditure in rodents and their use might help to reduce the dose of FXR agonist. Side effects of these agents, however, are poorly described and only a few agents are currently under development. If dual GPBAR1/FXR ligands are to be developed what is the ideal profile of these agents in terms of GPBAR1 and FXR specificity?

Additionally, GPBAR1 agonism might exert a beneficial effect on the vascular component of the disease [34,66] and might be useful to balance the negative effects of FXR activa- tion on lipid profile.