*Article* **Identification of the Potential Molecular Mechanisms Linking RUNX1 Activity with Nonalcoholic Fatty Liver Disease, by Means of Systems Biology**

**Laia Bertran <sup>1</sup> , Ailende Eigbefoh-Addeh <sup>1</sup> , Marta Portillo-Carrasquer <sup>1</sup> , Andrea Barrientos-Riosalido <sup>1</sup> , Jessica Binetti <sup>1</sup> , Carmen Aguilar <sup>1</sup> , Javier Ugarte Chicote <sup>1</sup> , Helena Bartra <sup>2</sup> , Laura Artigas <sup>2</sup> , Mireia Coma <sup>2</sup> , Cristóbal Richart <sup>1</sup> and Teresa Auguet 1,\***


**Abstract:** Nonalcoholic fatty liver disease (NAFLD) is the most prevalent chronic hepatic disease; nevertheless, no definitive diagnostic method exists yet, apart from invasive liver biopsy, and nor is there a specific approved treatment. Runt-related transcription factor 1 (RUNX1) plays a major role in angiogenesis and inflammation; however, its link with NAFLD is unclear as controversial results have been reported. Thus, the objective of this work was to determine the proteins involved in the molecular mechanisms between RUNX1 and NAFLD, by means of systems biology. First, a mathematical model that simulates NAFLD pathophysiology was generated by analyzing Anaxomics databases and reviewing available scientific literature. Artificial neural networks established NAFLD pathophysiological processes functionally related to RUNX1: hepatic insulin resistance, lipotoxicity, and hepatic injury-liver fibrosis. Our study indicated that RUNX1 might have a high relationship with hepatic injury-liver fibrosis, and a medium relationship with lipotoxicity and insulin resistance motives. Additionally, we found five RUNX1-regulated proteins with a direct involvement in NAFLD motives, which were NFκB1, NFκB2, TNF, ADIPOQ, and IL-6. In conclusion, we suggested a relationship between RUNX1 and NAFLD since RUNX1 seems to regulate NAFLD molecular pathways, posing it as a potential therapeutic target of NAFLD, although more studies in this field are needed.

**Keywords:** RUNX1; NAFLD; NASH; metabolism; systems biology

### **1. Introduction**

Nonalcoholic fatty liver disease (NAFLD) is a condition characterized by excess fat in the liver, without alcohol implication in the onset of the disease. The term NAFLD comprehends a substantial number of liver conditions, ranging from simple steatosis (SS) to the more aggressive form of nonalcoholic steatohepatitis (NASH), which may lead to cirrhosis and hepatocellular carcinoma [1]. SS is defined as the presence of ≥5% hepatic steatosis without evidence of hepatocellular injury in the form of hepatocyte ballooning, inflammation [2], and fibrosis, three remarkable events in NASH pathology. The progress of the disease may vary from individuals, depending on the accumulated fat to the immunological and the oxidant stress responses [3,4].

NAFLD is the most prevalent chronic liver disease, with a global prevalence in adults between 23–25% [5,6]. Nevertheless, nowadays there is no definitive diagnostic test apart

**Citation:** Bertran, L.; Eigbefoh-Addeh, A.; Portillo-Carrasquer, M.; Barrientos-Riosalido, A.; Binetti, J.; Aguilar, C.; Ugarte Chicote, J.; Bartra, H.; Artigas, L.; Coma, M.; et al. Identification of the Potential Molecular Mechanisms Linking RUNX1 Activity with Nonalcoholic Fatty Liver Disease, by Means of Systems Biology. *Biomedicines* **2022**, *10*, 1315. https://doi.org/10.3390/ biomedicines10061315

Academic Editors: Jinghua Wang and Albrecht Piiper

Received: 13 April 2022 Accepted: 1 June 2022 Published: 3 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

from invasive liver biopsy, and no specific approved treatment besides exercise and dietary interventions. Pharmacologic-based therapies for NAFLD are limited, but many clinical trials are in process [7]. For this reason, knowledge about NAFLD pathophysiology is continuously growing.

RUNX1 belongs to the runt-related transcription factor (RUNX) family of genes, and is also known as acute myeloid leukemia 1 [8]. RUNX1 regulates the differentiation of hematopoietic stem cells into mature blood cells [9,10]. It also plays a major role in the development of the neurons that transmit pain [11], and in angiogenesis and inflammation [12]. In addition, RUNX1 involvement in apoptotic processes has been reported on one hand to induce apoptosis and inhibit tumor progression in neuroblastoma [13] and leukaemia [14], while it contrarily seems to present an antiapoptotic effect in pancreatic and ovarian cancer [15,16].

Diseases associated with RUNX1 include platelet disorders with associated myeloid malignancy and blood platelet disease [17]. Related pathways include transport of glucose and other sugars, bile salts, organic acids, metal ions, and amine compounds, as well as transforming growth factor-beta (TGF-β) signaling pathways [18,19]. Recently, Kaur et al. reported a relationship between RUNX1 and NAFLD. Authors related its activity with the progression to NASH, since the interaction of RUNX1 and C-C motif chemokine 2 (CCL2), an important adhesion molecule, mediates the infiltration of pro-inflammatory and proangiogenic factors in NASH [20]. Thus, we previously wanted to study the role of RUNX1 mRNA and protein expression in NAFLD in a cohort of women with morbid obesity. We hypothesized that RUNX1 may play a protective role in NAFLD since its expression was enhanced in early stages of the disease and decrease along with the progression to NASH [21]. Given these controversies among our previous results and what was already known, the objective of the present work is to determine the proteins and the potential molecular mechanisms that could establish a link between the activity of RUNX1 and NAFLD pathogenesis by means of systems biology.

#### **2. Materials and Methods**

#### *2.1. Bibliographic and Metadata Analysis in Databases*

First, we built the molecular description of NAFLD pathophysiology through systematic searches and reviewing the most up-to-date scientific knowledge regarding this pathology (Supplementary Table S1). Accordingly, NAFLD was divided in specific pathophysiological processes–called motives–involved in SS, in NASH, or in both forms (Supplementary Table S1), and the corresponding molecular effectors (or key proteins) playing biological roles in these mechanisms were identified (Supplementary Table S2).

The interactome around RUNX1 was manually curated in order to better fit the mathematical models. Protein relationship databases including TRRUST database (Transcriptional Regulatory Relationships Unravelled by Sentence-based Text-mining) [22], BioGRID (The Biological General Repository for Interaction Datasets) [23], HPRD (Human Protein Reference Database) [24,25], INTACT (IntAct Molecular Interaction Database) [26], KEGG (Kyoto Encyclopedia of Genes and Genomes) [27], REACTOME (Reactome Pathway Database) [28], and available scientific literature were the sources used to identify and curate new direct interactors of RUNX1.

#### *2.2. Mechanistic Model Generation*

The compiled information was used to generate a mathematical model that simulate NAFLD pathophysiology by applying Therapeutic Performance Mapping System (TPMS) technology [29], which integrates all available biological, pharmacological, and medical knowledge to simulate human physiology in silico (Supplementary Table S3). Then, we used an artificial neural networks (ANNs) strategy [30,31] to analyze these models in order to establish the functional relationships between RUNX1 and NAFLD, considering the motives both together and individually. ANNs evaluate the relationship among protein sets or regions inside the Anaxomics network, providing a predictive score that quantifies

the probability of the existence of a functional relationship between the evaluated regions. Each score is associated with a *p*-value that describes the probability of the result being a true positive. The ranking score has been divided into five categories: very high (ANN score > 92; *p* < 0.01), high (ANN score = 78–92; *p* = 0.01–0.05), medium-high (ANN score 71–78; *p* = 0.05–0.1), medium (ANN score 37–71; *p* = 0.1–0.25), low (ANN score < 37; >0.25).

Sampling methods-based mathematical models were then generated to determine the potential molecular mechanisms that could justify our hypothesis:


TPMS sampling-based methods trace the most probable mechanisms of action (MoA) or paths, both in biological and mathematical terms, which lead from a stimulus (e.g., activation of RUNX1) to a response (e.g., activation of IR) through the biological human protein network. In this way, it identifies the set of possible MoA that achieve a response when the system is stimulated with the specific stimulus. A population of possible solutions was obtained, and this variability was exploited and analysed to obtain a representation with the most represented paths among the set of possible solutions. A detailed description of the applied methodology was described elsewhere [29,32] and in Appendix A.

#### **3. Results**

#### *3.1. Functional Relationship between RUNX1 and NAFLD: ANNs Analysis*

The possible functional relationship between RUNX1 and NAFLD, defined as the set of proteins included in its molecular characterization, has been evaluated by means of ANNs analysis. To deepen our insights, the analysis has also been performed individually for each pathophysiological motive included in NAFLD characterization: (1) increased body fat, (2) hepatic IR, (3) altered fatty acid metabolism, (4) lipotoxicity, and (5) hepatic injury and liver fibrosis. The first three pathophysiological processes occur in both SS and NASH, while the last two only happen in NASH pathophysiology or participate in the progression of NAFLD to NASH.

In this study, the relationship between RUNX1 and NAFLD or individual NAFLD motives has been evaluated, assuming that a possible functional relationship could indicate a participation of RUNX1 in NAFLD pathophysiology, either in promoting or reverting the process, since ANNs only indicate the existence of a possible relationship but not its direction. As shown in Table 1, the results obtained suggest a medium relationship of RUNX1 with the global NAFLD, considering all motives simultaneously.


**Table 1.** ANNs score of the relationship between RUNX1 and NAFLD, both globally and for each NAFLD motive.

When considering the motives separately, however, RUNX1 seems to show a high relationship with hepatic injury and liver fibrosis, and a medium relationship with both lipotoxicity and hepatic IR.

The different columns show the ANNs score obtained for NAFLD globally and for each individual pathophysiological motive, some involved in SS and NASH stages, while others are only implicated in NASH. Category splitting was based on *p*-value breaks. RUNX1, runt-related transcription factor 1; NAFLD, nonalcoholic fatty liver disease; SS,

simple steatosis; NASH, nonalcoholic steatohepatitis. A darker color indicates a higher ANN score.

The MoA of RUNX1 has been built specifically with regards to the pathophysiological motives–hepatic IR, lipotoxicity and hepatic injury & liver fibrosis–due to their high probability of relationship with RUNX1 and the previously known molecular information found in available scientific literature. Figure 1 shows the protein network of direct RUNX1 interactions with NAFLD effector proteins (the activity of which play a known role in the condition).

**Figure 1.** NAFLD effector proteins interacting with RUNX1 at distance 1 (direct link). RUNX1, runt-related protein 1; CEBPB, CCAAT/enhancer-binding protein beta; ATF-6, AMP-dependent transcription factor 6; JNK1, c-Jun N-terminal kinase 1; TNF, tumour necrosis factor; IL6, interleukin 6; SOCS3, suppressor of cytokine signaling 3; PKCE, protein kinase C epsilon type; IL17A, interleukin 17A; TLR4, toll-like receptor 4; NFKB1, nuclear factor kappa B 1; LMNA, lamin-A/C; TIMP1, metalloproteinase inhibitor 1; SPP1, secreted phosphoprotein 1; TFGB1, transforming growth factor beta-1 proprotein; IL1B, interleukin 1 beta; SMAD3, mothers against decapentaplegic homolog 3.

#### *3.2. Mechanisms of Action of RUNX1*

Then, TPMS sampling methods-based mathematical models were generated simulating NAFLD pathophysiology to identify the key proteins and the most probable paths that link the activation of RUNX1 with the most strongly-related motives according to ANNs analysis (hepatic IR, and lipotoxicity, hepatic injury, and liver fibrosis). To provide new insights on the different disease stages (SS or NASH), we studied two independent MoA, considering whether the motive occurs in early or later stages of the disease: (1) RUNX1 promoting IR and (2) RUNX1 promoting lipotoxicity and hepatic injury and fibrosis, respectively.

#### 3.2.1. Mechanism of Action of RUNX1 Promoting IR

Figure 2 summarizes some of the most interesting pathways that could be regulated by RUNX1 in the context of promotion of IR in NAFLD, including the modulation of genes such as CCAAT/enhancer-binding protein alpha (CEBPA), histone deacetylase 1 (HDAC1), the transcription factor c-JUN, nuclear factor kappa B (NFκB), and some types of protein kinase C (PKCβ and PKCε).

**Figure 2.** Most represented MoA of RUNX1 promoting IR in NAFLD in the population of TPMS model solutions. Gene names are used in the representations. RUNX1, runt-related protein 1; HDA1C, histone deacetylase 1; PTGS2, prostaglandin G/H synthase 2; c-Jun, protein encoded by JUN gene; TNF, tumour necrosis factor; NFκB, nuclear factor kappa B; IL6, interleukin 6; LCN2, neutrophil gelatinase-associated lipocalin 2; CEBPA, CCAAT/enhancer-binding protein beta; SOCS3, suppressor of cytokine signaling 3; PKC, protein kinase C; JNK1, c-Jun N-terminal kinase 1; MTOR, mammalian target of rapamycin serine/threonine-protein kinase; IRS, insulin receptor substrate. This picture was generated using Graphviz software.

Table 2 shows the IR effector proteins that are regulated by the activation of RUNX1 to promote this motive (considering that the activity values of the proteins in our models range from 1 to -1, only proteins with activation state > 0.1 are shown); the table contains all modulated proteins, not only those highlighted by the most represented paths. RUNX1 could be promoting IR through the regulation of 64.10% of the effector proteins involved in this motive. The IR effector proteins most activated by RUNX1-dependent downstream pathways are, in decreasing order: NFκB, JNK, PKCε, tumour necrosis factor (TNF), inhibitor of nuclear factor kappa B kinase subunit beta (IKBKB), and prostaglandin G/H synthase 2 (PTGS2); while the most inhibited ones are insulin receptor substrate (IRS)-1, phosphatase and tensin homolog (PTEN), IRS2 and sirtuin 1 (SIRT1).


**Table 2.** IR effector proteins modulated by RUNX1 activation. Causative effect indicates whether the protein is increased/overactivated (1) or reduced/inhibited (–1) in NAFLD.

NAFLD, nonalcoholic fatty liver disease; NAFLD, nonalcoholic fatty liver disease; RUNX1, runt-related transcription factor 1; MoA, mechanism of action. Green color indicates a positive interaction between the effector protein and RUNX1, while red color indicates a negative one. A more intense color indicates a higher intensity of activation/inhibition.

3.2.2. Mechanism of Action of RUNX1 Promoting Lipotoxicity and Hepatic Injury-Liver Fibrosis

As shown in Figure 3, most of the molecular pathways that may justify the potential role of RUNX1 promoting lipotoxicity and fibrosis-related processes are shared with those involved in the motive IR.

**Figure 3.** Most represented MoA of RUNX1 promoting lipotoxicity and hepatic injury and fibrosis in NAFLD in the population of TPMS model solutions. Gene names are used in the representations. RUNX1, runt-related protein 1; c-Jun, protein encoded by JUN gene; NFκB, nuclear factor kappa B; PKCε, protein kinase C epsilon; CEBPA, CCAAT/enhancer-binding protein alpha; SPP1, osteopontin; JNK1, c-Jun N-terminal kinase 1; BAX, BCL2 Associated X; CCL2, C-C motif chemokine 2; IL, interleukin; MMP2, matrix metalloproteinase-2; NLRP3, NACHT, LRR and PYD domains-containing protein 3; TLR, toll-like receptor; NOS2, inducible nitric oxide synthase; LCN2, neutrophil gelatinaseassociated lipocalin; PLIN1, perlipin; NOX, NADPH oxidase; IR, insulin resistance; LIPO, lipotoxicity; HILF, hepatic injury and liver fibrosis. This picture was generated using Graphviz software.

Table 3 describes the lipotoxicity and fibrosis effector proteins that are regulated by the activation of RUNX1 (considering that the activity values of the proteins in our models range from 1 to −1, only proteins with activation state >0.1 are shown). RUNX1 promotes lipotoxicity and fibrosis by the regulation of 50.88% and 62.07% of the effector proteins involved in these motives, respectively. In total, 17 proteins specific to lipotoxicity, 24 to fibrosis, and 12 involved in both motives are regulated by RUNX1. The proteins most regulated by RUNX1 involved in lipotoxicity-related processes are: JNK1, CEBPB, and IKBKB, and those involved in fibrosisrelated processes are mothers against decapentaplegic homolog 3 (SMAD3), angiopoietin-2 (ANGPT2), apoptosis regulator BAX, type-1 angiotensin II receptor (AGTR1), and TGF-β. Effector proteins with a role in both pathophysiological processes most activated by RUNX1 are NFκB, NADPH oxidase (NOX)-1, NOX4, CCL2, and TNF. The proteins most inhibited by RUNX1 are SIRT1 (lipotoxicity) and PTEN (fibrosis). Note that the list of proteins in Table 3 is not limited to those shown in the Figure 3.


**Table 3.** Lipotoxicity and fibrosis effector proteins modulated when RUNX1 is activated.


**Table 3.** *Cont.*

NAFLD, nonalcoholic fatty liver disease; RUNX1, runt-related transcription factor 1. Green color indicates a positive interaction between the effector protein and RUNX1, while red color indicates a negative one. A more intense color indicates a higher intensity of activation/inhibition.

#### *3.3. Overlap between the Mechanistic Pathways Modulated by RUNX1 Activation in IR and Lipotoxicity & Fibrosis Stimulation*

The NAFLD motives that have been studied for the generation of the two MoA in this project seem to be pathophysiologically related to each other since there is an overlap of effector proteins from the three motives, as described in Figure 4.

This high relationship prompted us to explore whether an overlap existed in the pathways regulated by the activation of RUNX1 in promoting these motives, and therefore, to be able to relate them. Thus, we have evaluated the similarities that interrelate the motives at the level of common effector proteins and/or pathways modulated by RUNX1 according to our models.

Common NAFLD effector proteins regulated by RUNX1 downstream mechanisms have been recognized by studying the overlap for the three motives together and studying pairs of motives separately. The proteins that we consider to be RUNX1-regulated with an activation value >0.1 are shown in Table 4. In this sense, cannabinoid receptor 1 (CNR1), which was found to be one of the common effector proteins with the three NAFLD analysed motives, presented an activation value lower than 0.1, and it is for this reason that we stop taking this protein into account from now on.

Values of protein activity in each MoA are displayed. "Causative effect in NAFLD" indicates whether the protein is increased/overactivated (1) or reduced/inhibited (−1) in NAFLD. NAFLD, nonalcoholic fatty liver disease; MoA, mechanism of action; IR, insulin resistance; L&F, lipotoxicity and fibrosis; RUNX1, runt-related transcription factor 1; NFκB, nuclear factor kappa B; TNF, tumour necrosis factor; IL6, interleukin 6; ADIPOQ, adiponectin; NOX, NADPH oxidase; CCL2, C-C motif chemokine 2; CYBB, cytochrome b-245 heavy chain; TLR, toll-like receptor; JNK1, c-Jun N-terminal kinase; IKBKB, inhibitor of nuclear factor kappa B subunit beta; MTOR, mammalian target of rapamycin serine/threonine-protein kinase; LCN2, neutrophil gelatinase-associated lipocalin 2; SIRT1, sirtuin 1; PTEN, phosphatase and tensin homolog. Protein codes were obtained from UniProt database.

As shown in Table 4, overlapping of RUNX1-regulated proteins is observed in all three motives and in each pair. Despite finding six effector proteins that share the three NAFLD motives, only five presented sufficient signal intensity to be considered downstream effector proteins of RUNX1 inducing NAFLD; these are NFκB1, NFκB2, TNF, ADIPOQ, and IL-6.


**Figure 4.** Overlap of effector proteins between the three NAFLD motives evaluated in the project: VENN diagram showing the number of effector proteins overlapping between the three indicated NAFLD motives. There are 39 proteins involved in IR mechanism, 57 in lipotoxicity, and 58 in fibrosis. Concretely, there are 9 proteins involved in IR and lipotoxicity, 6 in lipotoxicity and fibrosis, and only 2 in IR and fibrosis. In this regard, there are six proteins involved in the three motives of NAFLD pathogenesis: NFκB1, NFκB2, TNF, ADIPOQ, IL-6, CNR1. IR, insulin resistance; NFKB, nuclear factor kappa B; TNF, tumour necrosis factor; ADIPOQ, adiponectin; IL6, interleukin 6; CNR1, cannabinoid receptor 1; TLR, toll-like receptor; CYBB, cytochrome b-245 heavy chain; LEP, leptin; CCL2, C-C motif chemokine 2; NOX, NADPH oxidase; PTEN, phosphatase and tensin homolog; TGR5, G protein-coupled bile acid receptor-1; JNK1, c-Jun N-terminal kinase; SIRT1, sirtuin 1; IKBKB, inhibitor of nuclear factor kappa B kinase subunit beta; MTOR, mammalian target of rapamycin serine/threonine-protein kinase; LCN2, neutrophil gelatinase-associated lipocalin 2; RETN, resistin; SFRP5, secreted frizzled-related protein 5; PPARA, peroxisome proliferator-activated receptor alpha; DGAT2, diacylglycerol O-Acyltransferase 2.


**Table 4.** Effector proteins modulated by RUNX1 activation shared by the three motives: lipotoxicity and fibrosis; IR and lipotoxicity; and IR and fibrosis.

#### **4. Discussion**

The novelty of this work lies in the fact that we aimed to perform a high-throughput screening to determine the molecular mechanisms that could establish a link between the activity of RUNX1 and NAFLD pathogenesis.

Until now, the connection between RUNX1 and NAFLD remains uncertain. On one hand, Kaur et al. showed a significant correlation between RUNX1 expression and inflammation, fibrosis, and NASH activity score in patients presenting NASH; they also reported RUNX1 function as a pro-angiogenic factor in SS and NASH [20]. On the other hand, Liu et al. presented low levels of RUNX1 in hepatocellular carcinoma. In this sense, these authors suggested that RUNX1 is a tumour suppressing factor that inhibits angiogenesis [33]. Regarding our previous study, we reported that the mRNA and protein expression of RUNX1 in liver seems to be involved in first steps of NAFLD with a proangiogenicrepairing role; meanwhile, RUNX1 appears to be downregulated in the NASH stage [21]. Since these disagreements, an exhaustive study of the relationship between RUNX1 MoA and NAFLD/NASH pathogenesis need to be performed to clarify this issue. In addition, this study could help to recognize RUNX1 as a potential therapeutic target of NAFLD. Previous reports have suggested that RUNX1 could be a potential therapeutic target of cancers, such as acute myeloid leukaemia, since this protein is an important regulator of haematopoiesis in vertebrates [34,35]. The beneficial effect of the therapeutic amelioration of RUNX1 in patients with nonsmall-cell lung cancer has also been described, since the RUNX1 overexpression is correlated with enhanced metastasis [36]. In addition, RUNX1 have been suggested as a potential therapeutic target to limit the progression of adverse cardiac remodeling and heart failure [37,38]. In this regard, to analyze the potential use of RUNX1 as a therapeutic target of NAFLD should be thoroughly studied. For example, investigating liver targeting through liposomes or bile acids in liver cancer [39] could be possible future strategies to evaluate the role of RUNX1 in the pathogenesis of NAFLD.

In this sense, when we performed an ANN analysis concerning the probability of the relationship between RUNX1 protein and NAFLD motives, our first main finding is that RUNX1 seems to show a medium intensity relationship with both motives–hepatic IR and lipotoxicity–and a high intensity relationship with hepatic injury and liver fibrosis motives, suggesting that this protein probably plays a role in these processes. In this regard, this result matches with Kaur et al., who reported a relevant association between RUNX1 expression and fibrosis and inflammation, two of the main NASH parameters [20]. However, this result contradicts our previous reported hypothesis about the potential protective role of RUNX1 in early stages of NAFLD [21]; in contrast, our current result has shown a low or medium intensity relationship with SS-related parameters. Given that our ANN approach provides the probability of functional relationship–regardless of the activity status (up or downregulation)–and the current conflicting results in the literature, further studies in humans or in vivo are required to clarify these contradictions, although the current available evidence clearly supports an involvement, either by presence or absence, of RUNX1 in NAHLD and NASH.

In the current literature, no direct role of RUNX1 on IR has been described yet. However, as a novelty, we demonstrate in the present study that RUNX1 interacts with proteins involved in this pathophysiological process. IR can be defined as a reduced response of the liver to the effects of insulin, which triggers impaired glucose homeostasis (gluconeogenesis and glucose uptake). IR may exert multiple effects on hepatic metabolism such as increased lipogenesis, increased free fatty acids (FFA) uptake, impaired FFA export, and decreased FFA oxidation. Moreover, outside the liver, IR causes increased serum FFA levels because of failure of insulin to suppress hormone sensitive lipase-mediated lipolysis in adipose tissue [3,40–42]. In this situation, the PKCε, a downstream intermediate of RUNX1 signaling [20], is activated by the accumulation of diacylglycerol and participates in hepatic IR through impairing insulin signaling [43,44]. In addition, it is believed that RUNX1 could also be involved in IR through the transcription of IL-17 [45], a cytokine that leads to neutrophil and monocyte infiltration in the liver, thereby increasing IR [46]. In contrast, RUNX1 has been shown to inhibit the expression of Suppressor of cytokine signaling (SOCS)-3 [47]—an intracellular protein interfering with insulin signaling via ubiquitin-mediated degradation of IRS1 and IRS2 [48]—therefore ameliorating the IR. According to these facts, RUNX1 seems to have a dual role both promoting and/or preventing hepatic IR.

Another crucial event clearly involved in NAFLD progression is the lipotoxicity resulting from an excessive FFA influx to hepatocytes. Hepatic lipotoxicity occurs when the capacity of the hepatocytes to manage and export FFA as triglycerides is overwhelmed [49]. The molecular mechanisms responsible for lipotoxicity in NAFLD include endoplasmic reticulum and oxidative stress and impaired autophagy, processes that in turn activate apoptotic cascades, thus promoting tissue damage and inflammation [49].

Consequently, in conditions of hypoxia induced by steatosis [50] and inflammation, angiogenesis is triggered in chronic liver diseases [51]. It was demonstrated that proangiogenic factors have an early function in NAFLD progression from SS to NASH since proangiogenic treatments reduce not only inflammation but also steatosis [52]. In this regard, RUNX1, a pivotal regulator of hematopoiesis and angiogenesis [12,53], could be activated in order to repair the liver damage in early stages of NAFLD [21,54]. In contrast, RUNX1 activates target genes involved in lipotoxicity [55–58] such as CEBPB [59] and Cyclic AMP-dependent transcription factor (AT6) in a regulatory feed-back loop with the transcription factor AP-1 and JNK [60]. If exposure of hepatocytes to lipotoxicity and liver injury continues, it can induce apoptosis [61] and trigger inflammation by interacting with toll-like receptors (TLRs). Inflammation is a component of the wound healing process that leads to fibrosis, the deposition of extracellular matrix in liver parenchyma [62]. Additionally, RUNX1 may contribute to fibrosis and inflammation by modulation of proinflammatory cytokines (IL-1β, IL-6, TNF, etc.) [47,63], tissue inhibitor of metalloproteinase 1 (TIMP-1) [64], osteopontin [65] and TLRs [66], among others. Hence, RUNX1 seems to

play a dual role, inducing pro-inflammatory cytokines and triggering liver damage, but at the same time having a protective effect by trying to repair the hepatic damage via angiogenesis-related processes.

The second notable finding of this work was obtained because we performed the TPMS technology to identify the key proteins and the most probable paths that link the activation of RUNX1 with the most strongly related motives according to ANNs analysis. In this regard, we wanted to evaluate IR first, since it is one of the main parameters involved in the first stages of NAFLD [67]. Accordingly, IR effector proteins most activated by RUNX1-dependent downstream pathways are NFκB, JNK, PKCε, TNF, IKBKB, and PTGS2, while the most inhibited ones are IRS1, PTEN, IRS2, and SIRT1.

RUNX1 could activate the PTGS2/cyclooxygenase-2 (COX-2) through its interaction with HDAC1 [68,69]. When PTGS2/COX-2 signaling is activated during inflammation in adipose tissue, it can act as a crucial factor for the promotion of obesity-induced IR and fatty liver [70,71]. According to the inflammatory role of PTGS2/COX-2, this enzyme can be induced by growth factors and different cytokines, such as TNF-α, that play a feed-back regulation role [71]. The cytokine TNF-α, produced by adipocytes and macrophages, is highly activated by the downstream mechanisms of RUNX1, particularly via the interaction with the proto-oncogene c-Jun [60,72] or the activation of NFκB [73]. The IκB kinase (IKBK) complex is the master regulator for activation of the NFκB signaling pathway. The kinase complex comprises the two catalytic subunits, IKK1 (IKBKA) and IKK2 (IKBKB), and the regulatory subunit NEMO (IKBKG), which mediates NFκB activation in response to a number of different stimuli such as RUNX1, by phosphorylating IκB proteins [74]. NFκB plays an important role in the regulation of a wide range of proteins/molecular pathways involved in IR. Its activation can be induced by TNF-α and JNK mechanisms [75,76] and can lead to the up-regulation of TNF-α, IL-6 and neutrophil gelatinase-associated lipocalin (LCN-2), contributing to IR-related processes in NAFLD [48,73,77,78]. In addition, the transcription factor AP-1 aggravates IR by inflammation-related processes, inducing the expression of IL-6 [79,80] and TNF-α [72]. TNF-α could be importantly contributing to the development of IR by inhibition and degradation of the IRS mediated by a serine phosphorylation through different mechanisms: (1) SOCS3 is induced by the NFκB/JNK-mediated activation of TNF-α and IL-6 [81], or via CEBPA activation [82], inducing ubiquitin-mediated degradation of IRS1 and IRS2 [83]; (2) MTOR can be activated by TNF-α or PKCβ pathways [84,85] due to hyperglycemia, leading to phosphorylation of multiple serine residues in IRS1 and IRS2 with their further degradation; (3) JNK1 promotes IRS1 and IRS2 serine phosphorylation [86,87]. The inhibitory effects of JNK1 could be also stimulated by PKCβ and PKCε [88,89]. In this regard, some studies have identified associations of PKC activity with disruption of the insulin-induced signal transduction pathway [90–92].

In contrast, apart from the degradation/silencing of IRS induced by RUNX1, which was explained above, our analysis has also reported the negative effect of RUNX1 in phosphatase and tensin homologue (PTEN) signaling. Decreased PTEN activity would lead to excessive fat deposition in the liver [40]. PTEN physiological functions negatively regulate the activity of phosphatidylinositol 3-kinase (PI3K)/AKT pathway, which in normal conditions induces lipogenesis in hepatocytes, consequently triggering IR [93,94]. PTEN downregulation has been reported to be carried out by mechanisms involving the sequential activation of MTOR and NFκB [95]. On the other hand, we have also reported a strong repression of SIRT1 by RUNX1 action. SIRT1 is an essential negative regulator of pro-inflammatory pathways, mainly through down-modulating NFκB transcriptional activity, decreasing de novo lipogenesis, and increasing fatty acid β-oxidation [96]. Hence, RUNX1 mediated inhibition of SIRT1 interrupts the beneficial effect of this protein, thus promoting IR. In short, the action of all these effector proteins together gives rise to IR mechanisms.

Regarding the second main finding of this work, we wanted to focus the study of the most implicated motives in NASH stage [97]. In this sense, the effector proteins with a role in NASH (lipotoxicity and fibrosis related processes) most activated by RUNX1 are NFκB, TNF, CCL2, NOX1, and NOX4; the proteins most inhibited by RUNX1 are SIRT1 (lipotoxicity) and PTEN (fibrosis).

NFκB appear to be a relevant regulation core since several RUNX1 interactors regulate its expression [47]. NFκB might be activated by molecular mechanisms such as those explained above (TNF-α/AP-1/JNK pathways). The downstream effects of NFκB activation result in lipotoxicity, hepatocyte injury, inflammation, and fibrosis [40] through upregulated expression of the pro-inflammatory and/or pro-fibrogenic cytokines: CCL2 also called monocyte chemoattractant protein 1 (MCP-1) [98], IL-6 [77] and matrix metalloproteinase-2 (MMP-2) [99]. In particular, higher levels of CCL2 have been identified in NASH subjects in comparison with simple fatty liver [3,100].

Free fatty acids promote hepatic lipotoxicity by stimulating TNF-α expression via a lysosomal pathway, which could be stimulated by the RUNX1/c-Jun link [60,101–104]. JNK-1, also activated by RUNX1 regulated PKCε activation [20,89], leads to the induction of NFκB dependent pathways [105] and the proapoptotic protein BAX [106], resulting in hepatic tissue damage [107].

Additionally, the isoforms NOX1 and NOX4 seem to be upstream regulated by RUNX1. These proteins show a crucial role on both lipotoxicity and fibrosis-related processes, specially by regulating the activation of hepatic stellate cells and apoptosis, which are two important aspects in the fibrogenic process in NASH [108]. Oxidative biomolecular damage and dysregulated redox signaling induce high oxidative stress and thereby liver injury. Moreover, several studies have shown that the inhibition of NOX1 and NOX4 leads to decreased oxidative stress, lipid peroxidation, hepatic injury, inflammation, and fibrosis in NASH [108,109]. RUNX1 could induce NOX4 expression via PKCε [110,111] and NFκB dependent pathways [112], and induce NOX1 only through PKCε activation [111].

Conversely, RUNX1 have shown to inhibit SIRT1 and PTEN. Some studies have reported that liver-specific disruption of SIRT1 not only causes hepatic steatosis but also promotes the progression to an advanced metabolic disorder stage such as lipotoxicity [113]. Additionally, it seems that dysregulations of PTEN expression/activity in hepatocytes represents an important and recurrent molecular mechanism contributing to the development of liver disorders [114], given that further aberrant activation of hepatic stellate and Kupffer cells trigger the development of liver fibrosis and inflammation [95]. In summary, the pathway that constitute these effector proteins gives rise to processes of lipotoxicity and liver damage.

Accordingly, we have reported for the first-time specific MoA that RUNX1 could play a role in NAFLD pathogenesis motives, but this is only an in silico study and needs to be further validated in experimental research.

The last main objective of the present study is to analyze the overlapping proteins between the studied motives involved in NAFLD. In this sense, the shared proteins between IR and lipotoxicity most activated by RUNX1 are JNK1, IKBKB, MTOR, and LCN2, while the most inhibited by RUNX1 is SIRT1. On the other hand, the overlapping proteins observed in lipotoxicity and fibrosis motives that are the most positively modulated by RUNX1 mechanisms are NADPH oxidase NOX1 and NOX4, the chemokine CCL2, the cytochrome CYBB, and the TLRs 2, 4, and 9. The only effector that is shared between IR and fibrosis negatively modulated by RUNX1 is PTEN. Finally, the main contribution of this study is that we found five RUNX1-regulated proteins with a direct involvement in the three main NAFLD motives, which are NFκB 1, NFκB 2, TNF, ADIPOQ, and IL-6. These proteins are indicators of the relevance of their processes in terms of the relationship with RUNX1 mechanisms towards promoting NAFLD. NFκB1 and TNF present a high activation due to RUNX1 activity, as we explained above. In fact, NFκB-dependent pathways seem to definitely be a key element in these MoA due to its high number of up/downstream links, and for its important regulation of a lot of effector proteins of these motives, especially immune response-related proteins that trigger inflammation, fibrosis, or lipotoxicity [96].

On the other hand, in this study, NFκB 2, IL-6, and ADIPOQ present moderated values of activation, which differ from those of NFκB 1 and TNF. NFκB 2 is an important regulator

of RUNX1. It was shown that transcription levels of NFκB 2 were increased in RUNX1 deficient cells [115]. Furthermore, IL-6, as we already mentioned, is a pro-inflammatory cytokine that takes part in fibrosis and tissue damage induced by RUNX1 [47]. High TNF-α and IL-6 levels have been found in NAFLD patients, indicating an important role of these cytokines in the disease. In fact, IL-6 reduction was significantly correlated with both weight loss and insulin sensitivity [48]. Conversely, ADIPOQ seems to be downregulated by RUNX1 signaling. It has been shown that significantly up-regulated ADIPOQ expression in white adipose tissue leads to increased serum adiponectin concentrations. Low adiponectin levels are closely related to the severity of liver histology in NAFLD [116].

Our approach, as all modelling approaches, is subjected to limitations. First, it is limited by the current knowledge on the key studied elements, in this case RUNX1 functions and interactors and NAFLD molecular pathophysiology; thus, the models and conclusions are susceptible to being updated over time if prospective data and new information are generated. Nevertheless, TPMS models are built by considering the whole human protein network and a wide range of drug–pathology relationships (Supplementary Table S3); not only limited to the key studied elements, or even to hepatic involvement, they present accuracies against the training set above 80% in the case of ANN models, and above 90% in sampling methods-based models [32]. Thus, systems biology and artificial intelligence approaches allowed us to explore and present mechanistic hypotheses that are in agreement with current knowledge, providing a guide for further pre-clinical investigation in the advancement towards defining treatments for NAFLD. Further studies are needed for confirmation and advancement of these data.

#### **5. Conclusions**

NAFLD pathophysiological motives most functionally related to RUNX1, according to an ANNs-based analysis, are hepatic insulin resistance, lipotoxicity, and hepatic injury-liver fibrosis. These three pathophysiological processes are molecularly related, since they share NFκB1, NFκB2, TNF, ADIPOQ, and IL-6 as effector proteins. This connection suggested that RUNX1 could regulate molecular pathways involved in NAFLD pathogenesis, but more studies in this field are needed.

**Supplementary Materials:** The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/biomedicines10061315/s1, Table S1: Condition\_MOTIVES; Table S2: Condition\_PROTEINS; Table S3: S3 Training set Runx.

**Author Contributions:** Conceptualization, T.A. and L.B.; methodology, L.A. and M.C.; validation, T.A., L.B. and A.E.-A.; formal analysis, H.B., L.A. and M.C.; investigation, T.A., L.B., A.E.-A., M.P.-C., A.B.-R. and J.U.C.; resources, T.A., L.B., J.B., C.A., J.U.C. and C.R.; data curation, T.A., L.B. and H.B.; writing—original draft preparation, T.A., L.B. and A.E.-A.; writing—review and editing, T.A., L.B., A.E.-A., M.P.-C., A.B.-R., J.B., C.A., J.U.C., H.B., L.A., M.C. and C.R.; visualization, T.A., L.B.; supervision, T.A.; project administration, T.A., C.R.; funding acquisition, T.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was financed by own funding through the project PV20062N (Teresa Auguet).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

#### *The Sampling Methods*

TPMS sampling-based methods [29] generate models like a Multilayer Perceptron of an Artificial Neural Network over the human protein network (where neurons are the proteins, and the edges of the network are used to transfer the information). This methodology can be used for describing with high capability all plausible relationships between an input (or stimulus, in this case RUNX1) and an output (or response, in this case NAFLD motives: IR, lipotoxicity and fibrosis protein effectors).

Although this type of network would generate many possible mechanistic solutions, it can be limited by constraints and restrictions that must be respected: (a) the topology of the protein network, (b) the functional, medical, and biological information stored in Anaxomics' databases, and (c) the available data about the drug (known effects on the target and target biology). Various different approaches and optimization systems can be used for such a purpose, from those based on randomized systems (such as a Montecarlo-based system [30]) to those based on information derived from the topology of the network, in order to solve each parameter of the equation, i.e., the weights associated to the links between the nodes in the human protein network.

The algorithms construct and analyze the regularities of the sampling of different plausible solutions. This information is used to construct feature vectors descriptive of the most probable protein network interaction structure and network activation signal flow derived from the space of plausible protein interaction solutions. The feature vectors are further used as input to supervise machine learning methods as ensembles of classifiers that allow us to infer new clinical and protein level knowledge. K-Fold and leave-one-out cross-validation methods are employed to assess generalization capability.

The mathematical algorithm can be envisioned as an extremely complex multi-parametric function, where each parameter corresponds to the relative weight of a link connecting nodes (genes/proteins) in a graph (protein map). Mathematical models of biological systems have more variables than restrictions (e.g., the number of entries in the training set is always smaller than the number of parameters-link weights-required by the algorithm), so various sets of parameters are equally valid. Therefore, using TPMS sampling-based methods, we could generate populations of solutions that comply with the biological restrictions of the training set.

From this base set of valid mathematical solutions, this approach can be employed to trace back the biological effects on molecules by analyzing the different populations of solutions. This methodology traces the most probable path (in biological and mathematical terms) that leads from the stimulus to the response through the biological network. In other words, it identifies the most probable MoA that achieve a physiological response when the system is stimulated with a specific stimulus. Not all solutions are used for the analysis, as solutions that comply with the general knowledge collated in the training set are preferred. That is, only MoAs that are plausible from the standpoint of currently accepted scientific knowledge are considered. Accuracy is calculated considering the number of restrictions in the training set that the model complies with, and only models with accuracy above 90% are considered.

In this study, the predicted MoAs were aimed at the elucidation of the mechanisms of RUNX1 activity that leads to the promotion of NAFLD motives: (A) IR and (B) lipotoxicity and fibrosis mechanisms. A set of 250 biologically plausible solutions have been calculated, with a mean accuracy of 94%.

#### **References**


## *Article* **Comorbidities and Outcomes among Females with Non-Alcoholic Fatty Liver Disease Compared to Males**

**Naim Abu-Freha 1,2,\* ,†, Bracha Cohen 3,†, Sarah Weissmann 2,3, Reut Hizkiya 2,4, Reem Abu-Hammad <sup>2</sup> , Gadeer Taha <sup>5</sup> and Michal Gordon <sup>3</sup>**


**Abstract:** Sex-based medicine is an important emerging discipline within medicine. We investigated the clinical characteristics, complications, and outcomes of Nonalcoholic Fatty Liver Disease (NAFLD) in females compared to males. Demographics, comorbidities, malignancy, complications, outcomes, and all-cause mortality of NAFLD patients older than 18 years were analyzed. The data were extracted using the MDClone platform from "Clalit" in Israel. A total of 111,993 (52.8%) of the study subjects were females with an average age of 44.4 ± 14.7 years compared to 39.62 ± 14.9 years in males, *p* < 0.001. Significantly higher rates of hypertension, diabetes mellitus, obesity, dementia, and thyroid cancer and lower rates of ischemic heart disease (22.3% vs. 27.3%, *p* < 0.001) were found among females. Females had a higher rate of cirrhosis, 2.3% vs. 1.9%, *p* < 0.001, and a lower rate of hepatocellular carcinoma, 0.4% vs. 0.5%, *p* < 0.001. In the multivariate analysis, a relationship between age, diabetes mellitus, and cirrhosis development were found among males and females. A lower age-adjusted mortality rate was found among females, 94.5/1000 vs. 116/1000 among males. In conclusion, older age at diagnosis, higher rates of hypertension, diabetes mellitus, obesity, cirrhosis, and a lower age-adjusted all-cause mortality rate were found among females with NAFLD.

**Keywords:** fatty liver; cirrhosis; females; gender; liver

#### **1. Introduction**

Non-Alcoholic Fatty Liver Disease (NAFLD) is the most common liver disease, affecting around 25–30% of the population in some countries, with the highest prevalence in the Middle East and South America and the lowest in Africa [1–3]. The prevalence is increased among older people as well as those diagnosed with diabetes or obesity, possibly even reaching 60% of these populations [1]. As a common chronic liver disease, NAFLD frequently causes cirrhosis [4]. Moreover, a large part of chronic liver disease complications such as hepatocellular carcinoma (HCC), liver transplantation, and mortality result from NAFLD [5].

Sex-based medicine is a relatively new and important field of research that has emerged in the last decade. The impact of sex on illnesses can manifest as differences in prevalence, disease course, and outcomes. In the gastroenterology and hepatology field, significant sex-based differences have been found in colorectal cancer development and incidence, anatomical site, survival, indications and upper endoscopy findings [6–8]. Sex-related differences in epidemiology, disease progression, and treatment strategies of liver diseases have also been reported [9]. Drug toxicity and drug-dose gender gaps have been widely reported between males and females. Women have higher rates of autoimmune hepatitis (70–90% of cases are women), primary biliary cholangitis, and hepatocellular carcinoma. Women, however, have lower rates of primary sclerosing cholangitis, with a male:female ratio of 7:3 [9]. Sex

**Citation:** Abu-Freha, N.; Cohen, B.; Weissmann, S.; Hizkiya, R.; Abu-Hammad, R.; Taha, G.; Gordon, M. Comorbidities and Outcomes among Females with Non-Alcoholic Fatty Liver Disease Compared to Males. *Biomedicines* **2022**, *10*, 2908. https://doi.org/10.3390/ biomedicines10112908

Academic Editor: Jinghua Wang

Received: 7 October 2022 Accepted: 9 November 2022 Published: 12 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

differences have also been found regarding alcohol consumption and alcohol-associated liver disease: the prevalence of severe alcohol use disorder was reported in 18.3% of men and 9.7% of women in the USA, with women developing more severe alcohol-associated liver disease at lower levels of exposure compared to their male counterparts [10].

Hepatocellular carcinoma (HCC) is one of the feared complications of chronic liver diseases, and significant sex-related differences have been previously reported. HCC is a liver neoplasm with a multifaceted nature of causes, risk factors and genetic alterations [11,12]. Females present with HCC at an older age and with a higher number of HCC and hypertension cases in their family histories than males [13,14]. In addition, females with HCC were more likely to undergo HCC surveillance, have smaller tumor sizes at diagnosis, and have less vascular involvement [13,14].

Only scant data were published regarding sex-related disparities of NAFLD patients, and it is a relatively under-researched field [15]. On average, females make up a higher percentage of NAFLD cases than males [15]. Sex-related differences have been found in adolescents, with a higher prevalence of NAFLD (16.3% vs. 10.1%) and central obesity (33.2% vs. 9.9%) reported among females [16]. In general, the prevalence of NAFLD is higher among men and postmenopausal women than among women of reproductive age, possibly suggesting a hormonal protective role [17].

This study aimed to investigate and determine the disparities in comorbidities (particularly metabolic syndrome), laboratory data, liver-related outcomes, and mortality of female patients with NAFLD compared to males with NAFLD. Understanding these disparities is crucial for the diagnosis, follow-up, treatment, and surveillance of patients with NAFLD.

#### **2. Materials and Methods**

#### *2.1. The Materials Study Design and Patients*

This was a retrospective study that included patients aged 18 years or older diagnosed with NAFLD between the years 2000 and 2021. NAFLD patients were identified by having an ICD 10 code of K76.0 at any time in their chronic disease list (according to community data or hospital data). A total of 9353 patients with liver-related comorbidities including alcoholic liver disease, hepatitis B, and hepatitis C were excluded from our population. The sample of NAFLD patients was then subdivided according to sex.

#### *2.2. Data Collection*

The data was extracted from Clalit Health Services (CHS) using Clalit's datasharing platform powered by MDClone (https://www.mdclone.com (accessed on 7 October 2022)). CHS is the largest health maintenance organization in Israel, with about 4.7 million insured residents.

Demographics, laboratory data, complications, outcomes, and mortality data were retrospectively collected from NAFLD patients. Laboratory data included complete blood count, alanine transaminase (ALT), aspartate transaminase (AST), bilirubin, albumin, and international normalized ratio (INR), taken from blood samples at or nearest to the time of diagnosis. In addition, the Fib-4 and APRI scores were calculated at the time of diagnosis. Comorbidities including metabolic syndrome, cancer, and other common diseases were collected from computerized files according to the specific ICD-10 codes. Outcomes including cirrhosis, hepatocellular carcinoma, liver transplantation, and all-cause mortality were collected according to the ICD-10 codes as well. All collected data were compared between males and females.

#### *2.3. Statistical Analysis*

Data are presented as mean ± standard deviation (SD) for continuous variables and as a percentage (%) of the total for categorical variables. Univariate analyses were performed using independent T-tests for continuous variables and chi-square tests for categorical variables. We used logistic regression models to examine the multivariate relationships between risk factors and the odds of death. Before introducing the variables into the model, multicollinearity of

the variables was examined using the Variance Inflation Factor (VIF) statistic. The variables found to be significant in the univariate analysis were introduced into the multivariate model one after the other, and included age at diagnosis, gender, diabetes mellitus, cirrhosis, hepatocellular carcinoma, and esophageal varices. We calculated the all-cause mortality death rate among the groups, subdivided by age. The all-cause mortality death rate was age-adjusted using a general population control group from Clalit (452,012 people). All statistical analyses were performed using IBM SPSS version 26 (Chicago, IL, USA). *p*-values less than 0.05 were considered statistically significant. The study was carried out in accordance with the principles of the Helsinki Declaration. The study protocol was approved by the Institutional Helsinki Committee, approval number 198-21-SOR.

#### **3. Results**

#### *3.1. Patients*

The baseline characteristics, comorbidities, and malignancy rates among NAFLD patients are presented in Table 1. A higher percentage of our cohort was female (n = 111,993, 52.8%). Females were diagnosed at an older age, 44.4 ± 14.7 years, compared to males, 39.62 ± 14.9 years, *p* < 0.001. Higher rates of hypertension, diabetes mellitus, obesity, and dementia were observed among female NAFLD patients, 60.7% vs. 53.5%, 24.7% vs. 21.6%, 64.2% vs. 52.7%, 5.2% vs. 2.9%, respectively, *p* < 0.001. However, lower rates of ischemic heart disease and chronic renal failure were found among females, 22.3% vs. 27.3%, 11.2% vs. 14.8%, respectively, *p* < 0.001. A higher rate of thyroid carcinoma and a lower rate of kidney carcinoma were observed among females, 1% vs. 0.4%, *p* < 0.001, and 0.6% vs. 1% *p* < 0.001. No significant difference regarding other malignancies was found between the two populations. We found that NAFLD was diagnosed before most other metabolic syndrome-related diseases. A total of 99.5% of patients were diagnosed with diabetes mellitus after being diagnosed with NAFLD (0.5% of males and 0.5% of females were diagnosed with diabetes before NAFLD (*p* = 0.966). Only 8% of males and 8.5% of females were diagnosed with obesity before NAFLD, compared to 92% of males and 91.5% of females who were diagnosed after being diagnosed with NAFLD, *p* = 0.001. Hypertension, dyslipidemia, CIHD, and CVA were also diagnosed more commonly among males after NAFLD diagnosis compared to females.


**Table 1.** Baseline characteristics, comorbidities, and malignancy among the study groups.


#### **Table 1.** *Cont.*

BMI = Body Mass Index, CIHD = Chronic Ischemic, CVA = Cerebrovascular Accident, COPD = Chronic Obstructive Pulmonary Disease, CRF = Chronic Renal Failure, CRC = Colorectal cancer.

#### *3.2. Laboratory Results among the Study Groups*

The laboratory results are summarized in Table 2. Significant differences between females and males were found regarding several lab values: AST (30.7 ± 36 vs. 33.8 ± 39, *p* < 0.001), ALT (34.2 ± 38.6 vs. 47.14 ± 52.9, *p* < 0.001) GGT (51.89 ± 84 vs. 61.45 ± 100, *p* < 0.001) and albumin (4.22 ± 1.4 vs. 4.42 ± 1.76, *p* < 0.001). In addition, lower values of APRI (0.36 ± 0.66 vs. 0.44 ± 0.83, *p* < 0.001) but higher FIB-4 levels were found among females (1 ± 1 vs. 0.96 ± 1.1, *p* < 0.001). All statistical analyses were performed using IBM SPSS version 26 (Chicago, USA). *p*-values less than 0.05 were considered statistically significant.

**Table 2.** Laboratory values of females and males included in the study.



**Table 2.** *Cont.*

All values presented as mean ± SD. WBC = White Blood Cells, PLT = Platelets, ALT = Alanine Aminotransferase, AST = Aspartate Aminotransferase, GGT = Gamma-Glutamyl Transferase, INR = International Normalized Ratio, APRI = AST to Platelet Ratio Index.

#### *3.3. Liver-Related Outcomes and All Cause-Mortality*

The liver-related outcomes and all-cause mortality rates are summarized in Table 3. More females were diagnosed with cirrhosis (2.3% vs. 1.9%, *p* < 0.001), but at an older age compared to males (65.9 ± 12.3 years vs. 63.4 ± 13.7 years, *p* < 0.001, respectively). Lower rates of HCC and liver transplantation were found among females (0.4% vs. 0.5%, *p* < 0.001, 0.07% vs. 0.11%, *p* < 0.003, respectively). No statistical difference was found regarding esophageal varices, esophageal variceal bleeding, spontaneous bacterial peritonitis, and hepatorenal syndrome between males and females. There was a significantly higher rate of all-cause mortality among females compared to males (11.4% vs. 10.2%, *p* < 0.001). The age-adjusted mortality rate was calculated in our cohort using a reference control group of non-NAFLD patients. The all-cause age-adjusted mortality rate was lower among females compared to males (94.5 patients per 1000 female NAFLD patients compared to 116 patients per 1000 male NAFLD patients). The cirrhosis and all-cause mortality rates according to age group are presented in Tables 4 and 5. A lower rate of liver transplantation was performed in females compared to males (0.07% vs. 0.11%, *p* = 0.003).

**Table 3.** Liver-related outcomes and all-cause mortality rates among females and males with NAFLD.




**Table 4.** Cirrhosis rate according to age group among females and males with NAFLD.



#### *3.4. Factors Associated with Cirrhosis and All-Cause Mortality*

The multivariate analysis regarding cirrhosis development among males and females is presented in Table 6. A relationship between age, diabetes mellitus, and cirrhosis was found among males and females with NAFLD. A significant relationship between obesity and cirrhosis was found among males but not females.

A multivariate model for the risk of death among NAFLD patients included in our study is shown in Table 7. Age at diagnosis, gender, diabetes mellitus, cirrhosis hepatocellular carcinoma, and esophageal varices were found to be risk factors for death among NAFLD patients in the univariate and multivariate analyses, with odds ratios of 1.125, 1.382, 2.648, 4.016, 9.086 and 2.021, *p* < 0.001, respectively.


**Table 6.** Univariate and multivariate analyses of risk factors for cirrhosis among NAFLD patients.

**Table 7.** Univariate and multivariate analyses of risk factors for death among NAFLD patients.


#### **4. Discussion**

This study included more than 200,000 NAFLD patients (52.8% female). We found (1) females were diagnosed with NAFLD at an older mean age than males, (2) females had higher rates of comorbidities including metabolic syndrome, hypertension, diabetes mellitus, and obesity than their male counterparts, (3) females had a higher rate of thyroid carcinoma but no significant difference in rates of other cancers, (4) female patients had higher rates of cirrhosis than males and had higher all-cause mortality rates than males, (5) age and diabetes were found to be predictors for cirrhosis among males and females, but obesity was found to be a predictor for cirrhosis only among males, not females, and finally, (6) diabetes mellitus, cirrhosis, and HCC were found to be predictors of death among female NAFLD patients.

Sex-related differences in the context of NAFLD could be attributed to several factors: differences in body structure, behavioral risk factors, comorbidities, metabolic factors, genetics, and hormonal effects.

The body structures of females and males are inherently different. Differences in fat storage, fat metabolism, and health risks of obesity among females and males have been noted [18]. All of these differences could influence the prevalence of NAFLD among females and may have an impact on the clinical course and complications of the disease.

Behavioral risk factors such as smoking, alcohol and food consuming habits could also have an impact on the development of NAFLD. These differences in habits could be co-factors for NAFLD development and progression. Smoking, alcohol use, and fast food consumption are more common among males compared to females [19–24]. Despite these differences, the prevalence of NAFLD, cirrhosis development, and all-cause mortality are more common among females, possibly indicating other factors are more dominant influencers of NAFLD among females.

NAFLD is considered as the hepatic manifestation of metabolic syndrome and has a strong relationship with obesity. The chronological relationship between NAFLD and comorbidities is still unclear. In particular, the impact each has on the other, and the causal relationship between the two are still unknown. In our study, the rates of diabetes mellitus, hypertension, and obesity were higher among females than males. Most likely, diabetes mellitus and obesity influence the rate of disease progression of cirrhosis and all-cause mortality rates. In our study, diabetes mellitus was found to be a predictor for cirrhosis among both males and females, while obesity was found to be a predictor for cirrhosis among males only.

Several animal studies have demonstrated sexually dimorphic hepatic genes associated with NAFLD. These genes, related to lipid metabolism, drug metabolism, and glucose homeostasis, impact the severity of cirrhosis and inflammation and are risk factors for the onset, progression, and treatment response of NAFLD [17].

Another critical factor that could contribute to sex differences in NAFLD is the hormonal differences between males and females. Estrogen is a vital sex hormone that not only regulates the female reproductive system but also contributes to several biological functions and protection from different diseases.

In a rodent model, the peak serum tumor necrosis factor-alpha (TNF-a), a proinflammatory cytokine, in the liver was twice as high in rodents who received estrogen compared to controls. This study concluded that estrogen sensitizes Kupffer cells to lipopolysaccharide (LPS), resulting in increased toxic mediator production [25]. This pro-inflammatory and toxic mediator production could also affect the progression of liver diseases such as NAFLD. Hormonal, inflammatory, and oxidative stress factors are part of a complex cascade of NAFLD pathogenesis with sex-related differences [17].

Our results show females are diagnosed with NAFLD about five years later than their male counterparts. This could be explained by the protective estrogen effect from NAFLD, which is lost in postmenopausal women. This is consistent with increasing NAFLD rates with age in women [26,27]. Our findings supported this theory: 34.7% of our female patients were diagnosed with NAFLD at age fifty or older, compared to 25.7% of males.

With regard to comorbidities, we found higher rates of diabetes mellitus, hypertension, and obesity among female NAFLD patients but a lower rate of ischemic heart disease. This finding could be related to the protective effect of estrogen on cardiovascular disease incidence among women [27].

Our study found a higher rate of thyroid malignancy and a lower rate of HCC among females compared to males. Previous studies showed disparities in HCC among females compared to males in terms of undergoing HCC surveillance, tumor size at diagnosis, and vascular involvement [13,14]. Previous studies showed that older age, male sex, the severity of compensated cirrhosis at presentation, and sustained activity of liver disease are important predictors of HCC [28–30].

Our study demonstrated that higher rates of cirrhosis development in females, despite an older age at diagnosis and shorter exposure to the steatosis process in females. Hormonal effects and comorbidities such as diabetes and obesity may influence the progression of fibrosis. Whether or not sex is a risk factor for the progression of fibrosis is a controversial issue with conflicting findings across differently designed studies [27]. However, adjusting the cirrhosis rate according to the different age groups, we found a slightly lower rate of cirrhosis among most of the female age groups.

The all-cause age-adjusted mortality rate was lower among females in our study. In addition, a lower rate of HCC was found among females, though there was no significant difference in other complications such as esophageal varices and hepatorenal syndrome. Lower rates of HCC in females may account for the decreased rate of the all-cause mortality.

One of this study's limitations is the lack of availability of data on liver-specific causes of mortality. This makes it difficult to understand the difference in mortality rate, as it is possibly related to other comorbidities. Nevertheless, in the multivariate analysis, the factors with a significant impact on death were age at diagnosis, gender, diabetes mellitus, cirrhosis, and HCC.

To summarize, significant differences were found between females and males in terms of comorbidities, liver-related outcomes, and all-cause mortality rates. Understanding these differences in depth is crucial for prevention, early diagnosis, interventions, and treatment of NAFLD. Special consideration may be required for females in order to decrease the rate of cirrhosis and all-cause mortality. Additional studies are needed before specific interventions can be carried out; however, the practical implication of the present study lie in increasing awareness about the disparities between NAFLD development and outcomes in males and females.

This study is further strengthened by the use of national-based cohort data with a large number of included patients. However, some limitations should be mentioned. The retrospective design of the study design based on an electronic health file database prevented our ability to differentiate between NAFLD and NASH and there was no data regarding liver biopsy or fibrosis grade available.

#### **5. Conclusions**

In conclusion, significant differences were found between males and females with NAFLD regarding the age of diagnosis, comorbidities, liver-related complications and all-cause mortality.

**Author Contributions:** Conceptualization, N.A.-F., M.G. and G.T.; Methodology, N.A.-F., R.H., R.A.-H. and G.T.; Software, B.C.; S.W. and M.G., Validation, B.C., S.W. and M.G.; Formal Analysis, N.A.-F., B.C. and S.W.; Investigation, S.W., R.H., R.A.-H. and G.T., Resources, B.C., S.W. and M.G.; Data Curation, B.C., S.W., R.H. and R.A.-H.; Writing—Original Draft Preparation, N.A.-F.; Writing—Review and Editing, B.C., S.W., R.H., R.A.-H., G.T. and M.G.; Supervision, N.A.-F.; Project Administration, N.A.-F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The protocol for this research has been approved by the local Helsinki committee, the Soroka Helsinki committee, and it conforms to the provisions of the Declaration of Helsinki, approval number 198-21-SOR.

**Informed Consent Statement:** Patient consent was waived due to the retrospective design of the study.

**Data Availability Statement:** No additional data are available.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Review* **Gene Therapy for Acquired and Genetic Cholestasis**

**Javier Martínez-García 1 , Angie Molina <sup>1</sup> , Gloria González-Aseguinolaza 1,2,3 , Nicholas D. Weber 3,\* and Cristian Smerdou 1,2,\***


**Abstract:** Cholestatic diseases can be caused by the dysfunction of transporters involved in hepatobiliary circulation. Although pharmacological treatments constitute the current standard of care for these diseases, none are curative, with liver transplantation being the only long-term solution for severe cholestasis, albeit with many disadvantages. Liver-directed gene therapy has shown promising results in clinical trials for genetic diseases, and it could constitute a potential new therapeutic approach for cholestatic diseases. Many preclinical gene therapy studies have shown positive results in animal models of both acquired and genetic cholestasis. The delivery of genes that reduce apoptosis or fibrosis or improve bile flow has shown therapeutic effects in rodents in which cholestasis was induced by drugs or bile duct ligation. Most studies targeting inherited cholestasis, such as progressive familial intrahepatic cholestasis (PFIC), have focused on supplementing a correct version of a mutated gene to the liver using viral or non-viral vectors in order to achieve expression of the therapeutic protein. These strategies have generated promising results in treating PFIC3 in mouse models of the disease. However, important challenges remain in translating this therapy to the clinic, as well as in developing gene therapy strategies for other types of acquired and genetic cholestasis.

**Keywords:** cholestatic diseases; gene therapy; AAV; PFIC

## **1. Cholestatic Diseases**

Cholestatic diseases are based on bile dysfunction due to defects affecting bile synthesis or secretion. These processes involve a wide range of enzymes and membrane transporters involved in hepatobiliary circulation. According to its origin, cholestasis can be classified into two main groups: acquired cholestasis and genetic cholestasis [1].

## *1.1. Acquired Cholestasis*

Most cholestatic diseases are acquired, presenting a dysregulation of the hepatobiliary transporters as a consequence of an adaptive and protective response to bile acid (BA) accumulation in the liver. This regulation is multifactorial, involving different elements such as hormones, BAs, proinflammatory cytokines, and drugs. These different factors mediate the activation of transcription factors that regulate the expression of export pumps, which promote the reduction of intracellular BAs by their excretion in the urine, resulting in the detoxification of the liver [2]. Acquired cholestatic diseases include primary biliary cholangitis (PBC), primary sclerosing cholangitis (PSC), intrahepatic cholestasis of pregnancy (ICP), biliary atresia, drug-induced cholestasis, and inflammation-mediated cholestasis [1,3].

PBC and PSC are classified as autoimmune diseases of the hepatobiliary system, characterized by the presence of antimitochondrial antibodies, portal inflammation, and an

**Citation:** Martínez-García, J.; Molina, A.; González-Aseguinolaza, G.; Weber, N.D.; Smerdou, C. Gene Therapy for Acquired and Genetic Cholestasis. *Biomedicines* **2022**, *10*, 1238. https://doi.org/10.3390/ biomedicines10061238

Academic Editor: Jinghua Wang

Received: 29 April 2022 Accepted: 24 May 2022 Published: 26 May 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

immune-mediated destruction of intra- and extra-hepatic bile ducts [4,5]. Clinical manifestations vary widely, from asymptomatic to end-stage biliary cirrhosis. The pathogenesis of the disease is multifactorial, involving genetic, epigenetic, and environmental factors [4,6].

ICP, which is the most common disorder of the hepatobiliary system, is characterized by high serum BA levels in the third trimester of pregnancy that cause severe pruritus. In the development of this cholestatic disorder, high levels of gestational hormones, such as estrogen and progesterone, play a major causative role, while genetic factors may also be involved. Although symptoms disappear after childbirth, the biliary disorder can often recur during future pregnancies [7].

Biliary atresia is a rare liver disease affecting the bile ducts, resulting in the main cause of neonatal cholestasis. The etiology of this biliary disorder is unknown. In some cases, the origin is thought to be due to an exacerbated autoimmune response in the bile duct epithelium as a consequence of a viral infection or due to toxin-induced injury after birth [8]. In other cases, it is thought to be due to a malformation of the bile ducts during gestation. However, it is known that an early diagnosis allows for better outcomes after surgery [9].

Finally, drug- and inflammation-induced cholestasis are closely related. Both drugs and proinflammatory agents can induce cholestasis following inhibition of hepatobiliary transporters but rarely result in severe liver injury. These types of cholestasis have an immunological origin mediated by proinflammatory cytokines directed against the bile duct epithelium that can alter BA secretion [10].

#### *1.2. Inherited Cholestasis*

Genetic cholestasis, which represents a minority of all cholestatic disorders, includes different types of progressive familial intrahepatic cholestasis (PFIC) associated with mutations in relevant channel transporters of the hepatobiliary system. PFIC is a heterogeneous group of autosomal recessively inherited monogenic disorders with a low incidence of 1:50,000–100,000 births worldwide, representing approximately 15% of all cases of neonatal cholestasis [11]. These cholestatic syndromes are characterized by an early onset of the disease, usually in infancy, associated with clinical manifestations such as pruritus, jaundice, malabsorption of fat and fat-soluble vitamins, and hepatomegaly [11]. PFIC is associated with several liver complications, such as portal hypertension and cirrhosis, and can progress to end-stage liver disease and liver failure between childhood and adulthood. Depending on the type of PFIC, extrahepatic clinical manifestations or hepatocellular carcinoma (HCC) may occur [12]. The most common biochemical features of this group of hepatobiliary diseases are increased serum BAs and bilirubin [11]. Depending on their genetic origin, PFICs can be classified into six types. Mutations in *ATP8B1*, *ABCB11*, *ABCB4*, tight junction protein 2 (*TJP2*), *NR1H4*, and Myosin VB (*MYO5B*) genes are known to be the cause of PFIC 1-6 types, respectively (Figure 1). In PFIC1, mutations in the familial intrahepatic cholestasis 1 (*FIC1*) gene cause the loss of the asymmetric distribution of phospholipid content in the canalicular membrane, leading to membrane destabilization and reduced BA transport, resulting in their accumulation in hepatocytes, causing cholestasis. Mutations in the *ABCB11* gene can result in PFIC2 due to the absence of a functional bile salt export pump (BSEP) protein, which also leads to toxic accumulation of BA in hepatocytes. In PFIC3, mutations in *ABCB4* cause multidrug resistance protein 3 (MDR3, ABCB4) deficiency, which results in low levels of phosphatidylcholine (PC) in the bile, which is needed to form micelles and neutralize the toxicity of hydrophobic BAs, resulting in damage to the epithelium of bile canaliculi. Mutations in *TJP2* lead to the misdistribution of claudin tight junction in canaliculi, resulting in bile leakage and subsequently in PFIC4. PFIC5 is due to mutations in the *NR1H4* gene that cause deficiency in farnesoid X receptor (FXR), resulting in a reduction of BSEP and ABCB4 expression and the accumulation of toxic BAs in the hepatocytes. Finally, mutations in *MYO5B* interfere with the processing of normal intracellular trafficking of BSEP, reducing its expression and activity at the canalicular membrane, which results in the accumulation of toxic BAs in hepatocytes, giving rise to PFIC6 [13].

Different disease characteristics such as the age of onset, severity, and the manifestation of specific complications and serum markers vary between PFIC types [12,13]. severity, and the manifestation of specific complications and serum markers vary between PFIC types [12,13].

processing of normal intracellular trafficking of BSEP, reducing its expression and activity at the canalicular membrane, which results in the accumulation of toxic BAs in hepatocytes, giving rise to PFIC6 [13]. Different disease characteristics such as the age of onset,

*Biomedicines* **2022**, *10*, x FOR PEER REVIEW 3 of 21

**Figure 1.** Genetic classification and pathogenesis of PFIC. The diagrams show the genes and functions altered in each type of PFIC. The main deficient proteins for each type of PFIC are indicated by red crosses, while derived alterations in other proteins or pathways are indicated by blue crosses. Damage due to the abnormal accumulation of BAs is shown as yellow circles with orange lightnings. **Figure 1.** Genetic classification and pathogenesis of PFIC. The diagrams show the genes and functions altered in each type of PFIC. The main deficient proteins for each type of PFIC are indicated by red crosses, while derived alterations in other proteins or pathways are indicated by blue crosses. Damage due to the abnormal accumulation of BAs is shown as yellow circles with orange lightnings.

The role of BSEP in the functioning of the hepatobiliary system is very important, as mutations in different genes involved in BA metabolism and transport, such as *ABCB11*, *NR1H4*, and *MYO5B* causing its deficiency, cause PFIC [14–16]. In addition, depending on the severity of the disease, inherited intrahepatic cholestasis resulting from mutations in *ATP8B1* or *ABCB11* can be classified as either PFIC1 or 2, respectively, or benign recurrent intrahepatic cholestasis (BRIC) 1 or 2, respectively. Sometimes it is clinically difficult to discern between PFIC and BRIC because, in both cases, patients may present mild cholestasis with long-term complications [17]. In addition, some missense mutations in less conserved regions of the *ABCB11* and *ABCB4* genes promote the development of more moderate variants of cholestasis such as BRIC2, ICP, cholesterol cholelithiasis, drug-induced cholestasis, adult biliary cirrhosis, transient neonatal cholestasis, and others [18,19]. In addition, mutations in cholangiocyte transporter genes (e.g., the cystic fibrosis transmembrane conductance regulator (*CFTR*) gene) can cause cholestasis. In fact, a direct association between cystic fibrosis and cholestatic conditions, such as bile duct complications, gallstones, and primary sclerosing cholangitis, has been observed due to mutations in *CFTR* [20]. Other genetic multisystemic diseases associated with cholestatic disorders include Alagille syndrome (ALGS) and cerebrotendinous xanthomatosis (CTX). ALGS arises due to mutations in genes involved in the Notch signaling pathway, such as *JAG1* and *NOTCH2*, and the majority of patients present cholestasis and a deficiency of bile ducts [21]. CTX is caused by mutations in the *CYP27A1* gene, resulting in impaired BA biosynthesis The role of BSEP in the functioning of the hepatobiliary system is very important, as mutations in different genes involved in BA metabolism and transport, such as *ABCB11*, *NR1H4*, and *MYO5B* causing its deficiency, cause PFIC [14–16]. In addition, depending on the severity of the disease, inherited intrahepatic cholestasis resulting from mutations in *ATP8B1* or *ABCB11* can be classified as either PFIC1 or 2, respectively, or benign recurrent intrahepatic cholestasis (BRIC) 1 or 2, respectively. Sometimes it is clinically difficult to discern between PFIC and BRIC because, in both cases, patients may present mild cholestasis with long-term complications [17]. In addition, some missense mutations in less conserved regions of the *ABCB11* and *ABCB4* genes promote the development of more moderate variants of cholestasis such as BRIC2, ICP, cholesterol cholelithiasis, drug-induced cholestasis, adult biliary cirrhosis, transient neonatal cholestasis, and others [18,19]. In addition, mutations in cholangiocyte transporter genes (e.g., the cystic fibrosis transmembrane conductance regulator (*CFTR*) gene) can cause cholestasis. In fact, a direct association between cystic fibrosis and cholestatic conditions, such as bile duct complications, gallstones, and primary sclerosing cholangitis, has been observed due to mutations in *CFTR* [20]. Other genetic multisystemic diseases associated with cholestatic disorders include Alagille syndrome (ALGS) and cerebrotendinous xanthomatosis (CTX). ALGS arises due to mutations in genes involved in the Notch signaling pathway, such as *JAG1* and *NOTCH2*, and the majority of patients present cholestasis and a deficiency of bile ducts [21]. CTX is caused by mutations in the *CYP27A1* gene, resulting in impaired BA biosynthesis and the accumulation of toxic metabolites. Although liver damage is not common in all CTX patients, some cases of severe infantile cholestasis have been reported [22].

#### **2. Current Treatments**

#### *2.1. Surgical Procedures: Hurdles and Limitations*

Currently, therapeutic approaches for cholestatic disorders are limited, with liver transplantation being the only curative strategy for the more severe syndromes [23,24]. However, liver transplantation has numerous limitations, such as organ failure, donor shortage, limited organ viability, the requirement of life-long immunosuppression, and immunological rejection [25]. For inherited diseases, such as some types of PFIC, liver transplantation is considered for end-stage patients with severe complications, such as hepatocellular carcinoma (HCC), hepatic steatosis, and liver cirrhosis. Orthotopic transplantation successfully improves cholestasis and related symptoms in 3–5 years [12,26]. However, liver transplant has been shown to be associated with the development of circulating anti-BSEP antibodies in a small fraction of transplanted PFIC2 patients, resulting in the rejection of the transplanted organ [27,28]. Moreover, this approach is only partially effective for cholestatic diseases with extrahepatic manifestations, such as PFIC1.

A therapeutic alternative prior to liver transplantation is surgical treatment aiming to interrupt the enterohepatic circulation, including procedures, such as partial internal biliary diversion (PIBD), ileal exclusion, and partial external biliary diversion (PEBD), that lead to lower BA levels, less pruritus, and even reversal of hepatic fibrosis [29,30]. However, complications such as stoma bag-associated difficulties (e.g., dehydration or leakage) have been reported [30]. For treatment of hereditary cholestatic diseases, biliary diversion has been found to be more effective in PFIC2 patients with residual BSEP activity, while for PFIC3 patients it is usually done late in the disease process, making it hard to prevent disease progression [31,32]. Therefore, there is an urgent need to seek alternative therapeutic approaches to liver transplants and surgical approaches. However, there is room for hope since the increased understanding of the mechanisms leading to genetic and acquired cholestatic diseases has opened the window to develop new drug and gene therapies for the treatment of these disorders.

#### *2.2. Pharmacological Therapies*

Drug therapies are considered first-line treatments for cholestatic diseases. The main strategies in the pipeline are based on FXR agonists and inhibitors of BA uptake transporters in the enterohepatic circulation [33,34].

#### 2.2.1. FXR Agonists

In recent years, the use of selective FXR agonists, such as ursodeoxycholic acid (UDCA), has been the first option to treat cholestatic disorders. UDCA, a hydrophilic BA, reduces the hydrophobic pool of toxic BAs in hepatocytes as well as the detergent properties of bile in the bile canaliculi (Figure 2A). Currently, beneficial effects of UDCA have been reported in patients with ICP, PBC, and PFIC3, especially at the early stages of these diseases [35,36], although approximately 50% of the PFIC3 and PBC patients did not respond or had an incomplete response [19,37]. It has also been observed that PFIC3 patients with milder forms of ABCB4 deficiency respond better to UDCA treatment [38]. In contrast, this treatment fails to offer any symptomatic improvement for the majority of patients with PFIC2 or PSC [39,40]. On the other hand, UDCA-derived BAs such as 24-norursodeoxycholic acid (Nor-UDCA) or its taurine conjugate (TUDCA) have also shown potential as therapeutic agents for these liver diseases [35]. Nor-UDCA has shown improvement in serum disease biomarkers such as transaminases and alkaline phosphatase (ALP) levels in patients with PSC [41], although larger studies are needed to establish its real efficacy [42]. Currently, there is one clinical trial evaluating its use in PSC patients (NCT01755507). A recent study has shown that TUDCA was able to normalize serum ALP values in PBC patients [43]. Another FXR agonist with therapeutic potential in the treatment of cholestatic diseases is the semi-synthetic BA, obeticholic acid (OCA). Two phase II studies in PBC and PSC patients demonstrated the safety and beneficial effect of OCA in reducing serum ALP levels [44,45] and, in fact, OCA has been approved as an alternative treatment

for patients with PBC who do not respond to UDCA [46]. In addition, a recent study showed that OCA was able to reduce liver damage in a mouse model of PFIC2 [47]. Despite these promising results, its use in cholestatic patients has been associated with severe pruritus, which would make it difficult to be approved as a therapy for PFIC, in which pruritus is one of the main symptoms of concern [48]. Similarly, the non-steroidal FXR agonist cilofexor, which has been reported to lead to significant improvements in cholestasis markers in PSC patients [49], may cause pruritus in a dose-dependent manner as a side effect and is not recommended for certain cholestatic disorders [50]. tive treatment for patients with PBC who do not respond to UDCA [46]. In addition, a recent study showed that OCA was able to reduce liver damage in a mouse model of PFIC2 [47]. Despite these promising results, its use in cholestatic patients has been associated with severe pruritus, which would make it difficult to be approved as a therapy for PFIC, in which pruritus is one of the main symptoms of concern [48]. Similarly, the nonsteroidal FXR agonist cilofexor, which has been reported to lead to significant improvements in cholestasis markers in PSC patients [49], may cause pruritus in a dose-dependent manner as a side effect and is not recommended for certain cholestatic disorders [50].

values in PBC patients [43]. Another FXR agonist with therapeutic potential in the treatment of cholestatic diseases is the semi-synthetic BA, obeticholic acid (OCA). Two phase II studies in PBC and PSC patients demonstrated the safety and beneficial effect of OCA in reducing serum ALP levels [44,45] and, in fact, OCA has been approved as an alterna-

**Figure 2.** Pharmacological treatments for cholestatic diseases. (**A**) Mechanisms of action of UDCA, which favors the presence of hydrophilic BAs over hydrophobic BAs in bile, decreasing the toxic effect of "detergent bile" in cholestatic patients. (**B**) NTCP transporter inhibitors block the entry of BAs into hepatocytes. (**C**) ASBT inhibitors prevent the reabsorption of BAs in enterocytes, decreasing their entrance into the enterohepatic recirculation. Inhibitions are indicated with blue crosses. BA, bile acid (yellow circles). **Figure 2.** Pharmacological treatments for cholestatic diseases. (**A**) Mechanisms of action of UDCA, which favors the presence of hydrophilic BAs over hydrophobic BAs in bile, decreasing the toxic effect of "detergent bile" in cholestatic patients. (**B**) NTCP transporter inhibitors block the entry of BAs into hepatocytes. (**C**) ASBT inhibitors prevent the reabsorption of BAs in enterocytes, decreasing their entrance into the enterohepatic recirculation. Inhibitions are indicated with blue crosses. BA, bile acid (yellow circles).

Altogether, these data indicate that the identification and development of new and more efficient FXR agonists represents a very interesting area of investigation for the improved clinical management of cholestatic diseases (Table 1) [51,52]. Altogether, these data indicate that the identification and development of new and more efficient FXR agonists represents a very interesting area of investigation for the improved clinical management of cholestatic diseases (Table 1) [51,52].

#### 2.2.2. Inhibitors of Bile Acid Uptake Transporters 2.2.2. Inhibitors of Bile Acid Uptake Transporters

*Biomedicines* **2022**, *10*, x FOR PEER REVIEW 5 of 21

Recently, there has been great interest in developing drugs that are able to interrupt the enterohepatic circulation in a non-invasive manner for cholestatic disorders. The four transporters that allow circulation of BAs between the liver and intestine are the apical bile salt transporter (ASBT, also known as IBAT for ileal bile acid transporter), BSEP, the sodium-taurocholate cotransporter polypeptide (NTCP) and the basolateral organic solute transporter (OST) [1]. The inhibition of BSEP and OST transporters is not an option as this would result in toxic accumulation of BAs in hepatocytes and enterocytes, respectively [53,54]. In contrast, pharmacological inhibition of the hepatic transporter NTCP results in a well-tolerated increase of BAs in plasma and a subsequent decrease in the liver (Figure 2B) [55]. In fact, recent studies have shown the hepatoprotective effect of NTCP Recently, there has been great interest in developing drugs that are able to interrupt the enterohepatic circulation in a non-invasive manner for cholestatic disorders. The four transporters that allow circulation of BAs between the liver and intestine are the apical bile salt transporter (ASBT, also known as IBAT for ileal bile acid transporter), BSEP, the sodium-taurocholate cotransporter polypeptide (NTCP) and the basolateral organic solute transporter (OST) [1]. The inhibition of BSEP and OST transporters is not an option as this would result in toxic accumulation of BAs in hepatocytes and enterocytes, respectively [53,54]. In contrast, pharmacological inhibition of the hepatic transporter NTCP results in a well-tolerated increase of BAs in plasma and a subsequent decrease in the liver (Figure 2B) [55]. In fact, recent studies have shown the hepatoprotective effect of NTCP inhibition, resulting in attenuation of cholestasis [56]. ASBT inhibitors prevent the reabsorption of BAs in enterocytes and their recirculation to the liver, favoring their excretion in feces (Figure 2C). ASBT antagonists currently being tested in clinical trials include odevixibat (A4250, Albireo, Boston, MA, USA), maralixibat (LUM001, Mirum Pharmaceuticals, Foster City, CA, USA), elobixibat (A3309, Albireo), linerixibat (GSK2330672, Glaxo

Smith Kline, Brentford, United Kingdom) and volixibat (SHP626, Mirum Pharmaceuticals) (Table 1) [57,58]. Several preclinical studies and clinical trials have shown high safety profiles for all these compounds with limited adverse effects outside the gastrointestinal tract and a high specificity for ASBT when orally administered. The observed therapeutic effects include a decrease of BAs in the liver and serum, reduction in pruritus, liver inflammation, and liver fibrosis [57,58]. In 2021, odevixibat was approved for clinical use in PFIC patients by the US Food and Drug Administration (FDA) and European Medicines Agency (EMA). Moreover, its safety and efficacy for treatment of other cholestatic diseases, such as ALGS, are being evaluated [59]. Maralixibat has also been evaluated in PBC and PSC, but clinical trials were discontinued because this treatment did not improve pruritus compared to placebo [60]. Recently, maralixibat was approved for clinical use for ALGS patients by the FDA [61]. However, its use for other cholestatic diseases, such as PFIC1-4, is currently under evaluation by the EMA [62].

#### 2.2.3. Other Pharmacotherapeutic Agents

Further additional pharmacotherapeutic approaches for the treatment of cholestatic disorders are being explored. Peroxisome proliferator-activated receptor (PPAR) agonists and fibroblast growth factor (FGF) analogues have been shown to be effective for diseases such as PBC and PSC [63]. Activators of FXR transcriptional regulators, such as sirtuin 1, have been shown to alleviate cholestatic liver injury in mice with BA-induced cholestasis by increasing the hydrophilic character of the hepatic BA composition and decreasing plasma BA concentration [64]. The use of antifibrogenic and anti-inflammatory therapeutic agents, such as inhibitors of histone deacetylases and phosphodiesterase 5, led to reduced fibrosis and liver damage in a PFIC3 mouse model [65]. Finally, ABC transporter enhancers, such as ivacaftor, may rescue the functionality of canalicular membrane transporters implicated in cholestatic disorders, including BSEP. Thus, PFIC2 patients may benefit from this type of pharmacological treatment [66]. The use of fibrates, such as the PPAR agonists bezafibrate, fenofibrate, and elafibranor (Table 1), could also be beneficial for the treatment of PBC patients who do not respond to UDCA [67].


**Table 1.** Drug therapy for cholestatic diseases in clinical trials.


**Table 1.** *Cont*.

Although the pharmacological strategies mentioned above significantly improved the pathology of cholestatic diseases and the quality of life of the patients [63], they do not represent a definitive cure for hepatobiliary dysfunction. For this reason, the development of new strategies, such as cell and gene therapy, that allow stable, long-term correction of these diseases is highly desired. In the following section, we will focus on gene therapy strategies tested in preclinical models of cholestatic diseases.

#### **3. Gene Therapy**

Gene therapy involves the addition, removal, or modification of the genetic material of an individual in order to treat a disease [83]. Its efficacy depends on successful delivery to target cells, for which vectors (viral and non-viral) are utilized. Viral vectors are based on modified viruses, such as adenoviruses (Adv), adeno-associated viruses (AAV), retroviruses, and lentiviruses, among others, which have proven to be very effective for gene delivery, although they present some drawbacks such as immunogenicity and limitations in cargo size. Non-viral vectors, such as polymeric or lipid nanoparticles (LNP), unlike viral vectors, do not achieve delivery to the cell nucleus and induce much more transient transgene expression, but have a better safety profile, are not limited by packaging restrictions, and offer several advantages in manufacturability and shelf-life. Recently, non-viral vectors have shown a high degree of efficacy as demonstrated by the COVID-19 vaccines based on messenger RNA (mRNA)-containing LNPs [84].

Gene therapy has emerged as a promising approach to achieve safe, stable, and efficient long-term correction for a wide range of genetic diseases [85], including monogenic liver disorders, for which liver transplantation remains the only cure [86], as well as acquired liver diseases [87]. Viral and non-viral vectors have shown promising therapeutic results in numerous clinically relevant animal models, as well as in a large number of clinical trials [88,89]. The fact that more than a dozen gene therapy products have been approved by the FDA and EMA, albeit only three for liver gene therapy, is a promising sign for the future application of this technology for liver disorders [90,91].

#### *3.1. Gene Therapy for Acquired Cholestasis*

Since no definitive treatment has yet been developed for some acquired hepatic cholestasis, such as PBC and PSC, there is a great need to identify novel therapeutic alternatives that can reduce fibrogenesis and potentially prevent the development of chronic liver injury, making genetic-based treatments an attractive strategy to achieve sustained long-term therapeutic effects. injury, making genetic-based treatments an attractive strategy to achieve sustained longterm therapeutic effects. To generate animal models of acquired cholestatic disorders, interventions including

approved by the FDA and EMA, albeit only three for liver gene therapy, is a promising

Since no definitive treatment has yet been developed for some acquired hepatic cholestasis, such as PBC and PSC, there is a great need to identify novel therapeutic alternatives that can reduce fibrogenesis and potentially prevent the development of chronic liver

sign for the future application of this technology for liver disorders [90,91].

*Biomedicines* **2022**, *10*, x FOR PEER REVIEW 8 of 21

*3.1. Gene Therapy for Acquired Cholestasis*

To generate animal models of acquired cholestatic disorders, interventions including bile duct ligation (BDL) and the induction of cholestasis by drugs, such as estrogens and carbon tetrachloride (CCL4), have been utilized [92]. The development of cholestasis involves several processes including: cellular apoptosis, production of proinflammatory cytokines, and fibrogenesis that ultimately leads to biliary impairment [93]. bile duct ligation (BDL) and the induction of cholestasis by drugs, such as estrogens and carbon tetrachloride (CCL4), has been utilized [92]. The development of cholestasis involves several processes including: cellular apoptosis, production of proinflammatory cytokines, and fibrogenesis that ultimately leads to biliary impairment [93]. Gene therapy approaches for acquired cholestasis have been addressed to mitigate

Gene therapy approaches for acquired cholestasis have been addressed to mitigate liver damage by reducing apoptosis and fibrosis and improving bile formation (Figure 3). Next, we will describe the most relevant gene therapy strategies described so far. liver damage by reducing apoptosis and fibrosis and improving bile formation (Figure 3). Next, we will describe the most relevant gene therapy strategies described so far.

**Figure 3.** Gene therapy approaches for acquired cholestatic diseases. Different gene therapy strategies have resulted in an alleviation of liver disorders according to their anti-apoptotic, anti-inflammatory, and anti-fibrotic properties, respectively. Adv, adenoviral vector; AAV8, adeno-associated vector with serotype 8; ACE2, angiotensin-converting enzyme; AQP-1, aquaporin; Cthrc-1, collagen triple helix repeat containing-1; HNF4a, hepatocyte nuclear factor 4 alpha; IGF, insulin-like growth factor; SOD, superoxide dismutase; uPA, urokinase-plasminogen activator. This figure was created using BioRender.com. **Figure 3.** Gene therapy approaches for acquired cholestatic diseases. Different gene therapy strategies have resulted in an alleviation of liver disorders according to their anti-apoptotic, anti-inflammatory, and anti-fibrotic properties, respectively. Adv, adenoviral vector; AAV8, adeno-associated vector with serotype 8; ACE2, angiotensin-converting enzyme; AQP-1, aquaporin; Cthrc-1, collagen triple helix repeat containing-1; HNF4a, hepatocyte nuclear factor 4 alpha; IGF, insulin-like growth factor; SOD, superoxide dismutase; uPA, urokinase-plasminogen activator. This figure was created using BioRender.com.

#### 3.1.1. Apoptosis Attenuation

One of the main targets for gene therapy of acquired liver disorders is the reduction of hepatocyte apoptosis. Hydrodynamic-based gene delivery to the liver of an insulin-like growth factor 1 (IGF-1)-expressing plasmid has demonstrated attenuation of hepatocellular apoptosis and liver injury in rats with BDL. IGF-1 promotes amelioration of cholestatic disease through activation of the phosphatidylinositol-3-kinase pathway, the inhibition of glycogen synthase kinase-3 beta, and the blockade of caspase-9 cleavage. Additionally, inactivation of hepatic stellate cells has been observed, which may explain the notable improvement in the degree of liver fibrosis [94].

#### 3.1.2. Reduction of Mitochondrial Oxidative Stress

Reducing oxidative stress has been shown to be a therapeutic target for acquired liver cholestasis. For example, Adv-mediated mitochondrial superoxide dismutase (SOD) gene delivery leads to a reduction in liver injury by avoiding the formation of oxygen free radicals derived from the accumulation of hydrophobic BAs and preventing the release of proinflammatory cytokines, such as TNFα and TGF-β, in mice with BDL [95]. Similarly, administration of Adv vectors expressing an inhibitor gene of proinflammatory cytokine signaling like collagen triple helix repeat containing-1 (Cthrc-1) has shown a reduction of liver fibrosis in mice subjected to BDL and drug-mediated cholestasis through the inhibition of TGF-β signaling caused by the accelerating degradation of phospho-Smad3 [96].

#### 3.1.3. Anti-Fibrotic Therapies

Anti-fibrotic therapies for cholestatic disorders via reducing pro-inflammatory factors tend to promote collagen degradation and thus reduce the degree of liver fibrosis. Adv vectors expressing the urokinase-plasminogen activator (uPA) gene resulted in a slight reduction of liver fibrosis, leading to a partial improvement of liver histology in rats with BDL associated with the activation of metalloproteinases that trigger collagen degradation [97,98]. Additionally, AAV vectors that allow hepatic expression of angiotensinconverting enzyme (ACE2) provided a sustained anti-fibrotic effect in different animal models of BDL and drug-induced cholestasis [99]. A different strategy to fight fibrosis is based on the gene delivery of human hepatocyte nuclear factor 4 alpha (HNF4A) via AAV vectors or mRNA containing LNP. This type of gene therapy was able to decrease the expression of genes involved in profibrogenic activity and revert fibrosis in several mouse models with induced or genetic cholestasis [100].

#### 3.1.4. Amelioration of Bile Flow

Finally, Adv-mediated hepatic delivery of aquaporin-1 (AQP1) has shown an improvement in the bile flow of estrogen-induced cholestatic rats [101]. In fact, this approach resulted in a marked reduction of serum ALP, as well as serum and biliary concentrations of bile salts. Moreover, AQP1 gene transfer increased biliary output as mediated by a significant increase in BSEP transport activity [102].

Thus, gene therapy approaches may offer a new avenue for the development of novel treatments for acquired cholestatic disorders.

#### *3.2. Gene Therapy for Inherited Cholestasis*

Gene therapy for the treatment of inherited hepatic diseases has garnered a great deal of attention after demonstrating that AAV vectors expressing human coagulation factors IX and VIII in the livers of patients with hemophilia B and A, respectively, resulted in a sustained therapeutic effect for more than three years [103]. In fact, a large number of gene therapy products have demonstrated promising therapeutic effects in clinically relevant animal models, leading to clinical trials for inherited liver disorders, such as phenylketonuria, familial hypercholesterolemia, ornithine transcarbamylase deficiency, acute intermittent porphyria, methylmalonic acidemia, and Wilson's disease, among others [88]. In the next sections of the review, we will focus on the use of gene therapy for inherited cholestatic diseases, which include genetic disorders with associated cholestasis and the different forms of PFIC.

#### 3.2.1. Gene Therapy of Genetic Disorders with Associated Cholestasis

Preclinical studies have shown promising results in animal models of Cerebrotendinous xanthomatosis (CTX) and Crigler-Najjar syndrome type 1. In the first case, the administration of an AAV8 vector expressing CYP27A was able to restore BA metabolism

and normalize the concentration of most BAs in plasma in a mouse model of CTX [104]. Interestingly, this therapeutic effect was achieved with only 20% of transduced hepatocytes, which could greatly facilitate the clinical translation of this approach. Secondly, treatment of Crigler–Najjar syndrome type 1 with an AAV8 vector expressing UDPglucuronosyltransferase family 1-member A1 (UGT1A) showed normalization of total serum bilirubin levels in two animal models of the disease, Gunn rats and *Ugt1a1-/* mice [105]. In this last model, a therapeutic effect was also demonstrated in newborn mice, although high doses of vector were required to maintain the effect [106]. These preclinical results led to a phase I/II clinical trial sponsored by Genethon (Évry, France), which is currently ongoing (NCT03466463).

The results observed in preclinical studies of Crigler–Najjar syndrome showed that one of the main limitations for gene therapy of genetic cholestatic diseases could be related to the loss of viral genomes associated with hepatocyte proliferation occurring in young patients [107].

#### 3.2.2. Gene Therapy for PFIC Diseases

Gene therapy approaches for PFIC can be based on gene supplementation or gene editing strategies to modify and repair the affected genes. The implementation of gene therapy for the different types of PFIC has some limitations. Firstly, in some types of PFIC in order to achieve stable and long-term therapeutic efficacy, it could be necessary to transduce most of the hepatocytes, which may require the use of high doses of the viral vector with the concomitant safety concerns [107,108]. Secondly, some types of PFIC have extrahepatic clinical manifestations hampering the liver-targeted treatment [109]. Finally, PFIC diseases requiring therapy are generally diagnosed in pediatric patients, and gene therapy based on non-integrative vectors, such as AAV, may be inefficient due to the loss of viral genomes associated with hepatocyte proliferation in a growing liver [107]. The decision to undergo gene therapy for PFIC, as well as the outcome of the therapy, will likely be influenced by the type of mutations present in the affected gene. For example, patients with missense mutations leading to decreased protein activity will probably respond better than those with a complete deficiency.

Although the loss of viral genomes could be a problem for most inherited cholestasis, ABCB4 deficiency, which causes PFIC3, has certain advantages over other PFIC types for liver gene therapy. For example, previous results using hepatocyte transplantation in a mouse model of PFIC3 showed that engraftment of 12% of healthy hepatocytes was enough to achieve therapeutic efficacy [110]. This evidence led to four preclinical studies examining the feasibility of gene therapy for PFIC3 in three different *Abcb4-/-* mouse models with a range of phenotypes depending on the mouse strain [111].

#### Gene Therapy for PFIC3 Based on ABCB4 Supplementation

The first study tested gene therapy in C57BL/6 *Abcb4-/-* mice that were challenged with a BA-enriched diet to increase liver toxicity due to their mild phenotype. Treatment with an AAV8 vector expressing ABCB4 demonstrated long-term efficacy by preventing the increase of serum transaminases and the loss of biliary PC levels after BA challenge [112]. In a second study, performed by our group, we evaluated PFIC3 AAV-based gene therapy in FVB *Abcb4-/-* mice, which have a clinically relevant phenotype characterized by high serum levels of bile salts and transaminases, hepatosplenomegaly, and liver fibrosis [113]. In this model, we demonstrated that an AAV8 vector containing a codon-optimized *ABCB4* sequence downstream of the liver specific alpha-1 antitrypsin (AAT) promoter resulted in stable and long-term correction of PFIC3 by improving all disease markers. Interestingly, this therapy was not only able to prevent disease progression in young mice (two-weekold), in which symptoms had not yet developed, but also in older mice with an established phenotype (five-week-old and sixteen-week-old mice). The therapeutic effect was dose dependent, and it was observed that restoration of biliary PC levels above 12–13% (over 4000 µM) of wild-type levels was enough to have a curative effect. This indicates that

PFIC3 could be treated even if only a small fraction of hepatocytes were transduced, in this way resembling gene therapy of other diseases like hemophilia B, in which therapeutic effects can be obtained with a small percentage of transduced hepatocytes. In our study, the therapeutic threshold was achieved with as little as 2–3% of wild-type ABCB4 expression levels [113]. Interestingly, this therapy was more efficacious in male mice compared to females, although a sustained therapeutic effect could be obtained in females by the administration of a second vector dose [113].

Recently, a preclinical study based on LNP-encapsulated mRNA therapy was able to transiently reverse the disease phenotype in BALB/c *Abcb4-/-* mice [114]. BALB/c *Abcb4-/-* show similar levels of serum biomarkers as the FVB *Abcb4-/-* mice, but with a faster progression of liver fibrosis, leading to early development of primary liver cancers as well as an earlier onset of other complications, such as portal hypertension [111]. Five repeat *ABCB4* mRNA-LNP injections were able to restore ABCB4 expression and biliary PC levels (~42% of wild-type levels), as well as improve serum biomarker levels, liver fibrosis, and hepatomegaly [114,115]. However, these previously described non-integrative vectorbased gene therapy strategies may have important limitations, such as loss of transgene expression, either because of loss of viral genomes due to hepatocyte division or because the short half-life of mRNA requires periodic administration to maintain the therapeutic effect. An alternative strategy to solve this hurdle is gene delivery mediated by an integrative vector.

Using this type of approach, Siew et al. tested PFIC3 correction by the use of an integrative hybrid vector based on the expression of a piggyBac transposase and an AAV8 vector containing a piggyBac ABCB4 expression cassette in FVB *Abcb4-/-* mice. A single dose of the hybrid vector in neonates demonstrated the recovery of biliary PC levels and normalization of serum biomarkers. Additionally, the hybrid AAV-piggyBac treatment prevented biliary cirrhosis and reduced tumorigenesis [116]. However, the possibility of this vector integrating into oncogenic sites represents a high risk for clinical application. Results from these preclinical studies have led to orphan drug designation of an AAV vector harboring a codon optimized version of ABCB4 (VTX-803) developed by Vivet Therapeutics (Paris, France), opening a promising pathway for the treatment of patients with this cholestatic disorder (Table 2).

#### Gene Therapy for PFIC3 Targeting Mechanisms of Disease

Although gene supplementation or correction of the affected gene is the most straightforward gene therapy strategy for PFIC3, several studies have shown that it is also possible to treat this disease by altering the expression of other genes that are involved in this pathology. One example is the delivery of vectors that express genes that contribute to the attenuation of liver fibrosis, such as ACE2 and HNF4A, as described in Section 3.1.3. In this sense, an AAV8 vector expressing ACE2 was able to reduce liver fibrosis in earlyand late-stage FVB *Abcb4-/-* mice [117]. Moreover, hepatocyte-targeted administration of *HNF4A* mRNA encapsulated with a biodegradable lipid restored the metabolic activity of hepatocytes in FVB *Abcb4-/-* mice, leading to a robust inhibition of fibrogenesis [100].

A novel approach that could be used to treat cholestatic diseases is based on the regulation of BA synthesis and homeostasis. It has recently been described that Limb expression 1-like protein (LIX1L) is increased in the liver of patients with cholestatic diseases and that the normalization of its expression alleviates cholestatic liver injury in different cholestatic mouse models, including FVB *Abcb4-/-* mice. LIX1L regulates the levels of miR-191-3p, a microRNA that downregulates transcription factor liver receptor homolog-1 (LRH-1), thereby inhibiting Cyp7a1 and Cyp8b1 expression, two enzymes required for BA synthesis. Based on these data, Li et al. [118], recently showed that an AAV vector overexpressing miR-191-3p was able to ameliorate cholestasis in FVB *Abcb4-/-* mice by direct repression of LRH-1 expression, thereby reducing de novo BA synthesis [118]. Another potential target for reducing liver fibrosis through gene therapy of cholestatic disorders is the suppression of the neurokinin 1 receptor (NK1R) axis as well as transforming growth

factor-β1 (TGF-β1)/miR-31 signaling. In FVB *Abcb4-/-* mice, knock-out of NK1R has been shown to decrease the levels of miR-31 and of proinflammatory molecules such as TFGβ1, resulting in the reduction of liver inflammation and fibrosis [119]. These therapeutic approaches could be very useful for either acquired cholestatic disorders or PFIC. *Biomedicines* **2022**, *10*, x FOR PEER REVIEW 12 of 21

> **Table 2.** Gene therapy approaches for PFIC3. **Table 2.** Gene therapy approaches for PFIC3.


dose. AAT, alpha-1 antitrypsin; AAV, adeno-associated vector; ALP, alkaline phosphatase; ALT, alanine aminotransferase; BW, body weight; LNP, lipid nanoparticles; LP1, liver-specific transcriptional AAT, alpha-1 antitrypsin; AAV, adeno-associated vector; ALP, alkaline phosphatase; ALT, alanine aminotransferase; BW, body weight; LNP, lipid nanoparticles; LP1, liver-specific transcriptional control unit; PC, phosphatidylcholine; TRsh, short piggyBac terminal repeats; VG, viral genomes; WT, wild-type.

control unit; PC, phosphatidylcholine; TRsh, short piggyBac terminal repeats; VG, viral genomes; Gene Therapy for Other Types of PFIC

WT, wild-type. Gene Therapy for PFIC3 Targeting Mechanisms of Disease Although gene supplementation or correction of the affected gene is the most straightforward gene therapy strategy for PFIC3, several studies have shown that it is also possible to treat this disease by altering the expression of other genes that are involved in For other types of PFIC, although gene supplementation using vectors expressing the specific mutated gene is also an option, there are certain barriers that make the development of these treatments more challenging than for PFIC3. For example, patients with PFIC1, PFIC4, PFIC5, and PFIC6 have extrahepatic manifestations that cannot be rescued by livertargeted gene therapy [109,120]. In addition, in contrast to gene therapy for PFIC3, where

this pathology. One example is the delivery of vectors that express genes that contribute

liver toxicity arises in the bile canaliculi and transgene delivery to a fraction of hepatocytes leads to sufficient ABCB4 protein to reverse toxicity, in other types of PFIC where toxicity occurs in hepatocytes, it is likely that correction of a high percentage of these cells will be required to achieve a therapeutic effect [110,121]. One additional problem to develop gene therapy strategies for some types of PFIC is the lack of suitable animal models that adequately recapitulate the phenotype of patients. Currently, there are no *TJP2*-deficient animal models available to test the feasibility of gene therapy for PFIC4 [121]. Likewise, the existing animal model for PFIC6 is not suitable, because it has a complete knock-out of the MYO5B protein, which is not an appropriate model for this cholestatic disease. For that, it is necessary to develop an animal model with missense mutations of the *MYO5B* gene that affect the motor domain but do not result in complete deficiency of the protein [122]. In the case of PFIC2, there are several animal models that show a varying degree of pathology depending on the genetic background. *Abcb11-/-* mice in a C57BL/6 background represent the closest model to the patient disease phenotype, showing a drastic decrease in bile salt content in the bile that leads to increased levels of serum transaminases, liver fibrosis, and hepatomegaly, with these changes being more severe in females than in males [123]. However, unlike PFIC2 patients, these mice only show a mild elevation of serum bile salts, which is one of the main biomarkers of the disease.

Finally, the loss of transgene expression by hepatocyte cell division is a drawback for the use of non-integrative vectors, such as AAV, in gene therapy of these inherited cholestatic disorders that need to be treated at very early ages, as only a few hepatocytes will maintain episomal AAV genomes [124]. Unlike PFIC3, for which partial gene therapy supplementation or correction of the affected gene is feasible, other types of PFIC may benefit from other gene therapy strategies aimed at reversing liver damage at several levels.

#### **4. Future Directions**

Due to the growing success of liver-targeted gene therapies and preclinical studies showing therapeutic efficacy against cholestatic diseases, such as PFIC3, the need to overcome challenges involved in taking these products from bench to bedside is even more critical.

One of the main challenges that gene therapy of cholestatic disorders faces is the potential loss of therapeutic effect in pediatric patients. This could be due to a decrease of viral genomes as a result of hepatocyte divisions in a growing liver in the case of AAV-based therapies, or to the transient expression of non-viral vector-mediated mRNA delivery [107,125]. Other challenges include immune responses to treatment (vector or transgene) and vector-mediated toxicities, particularly as a result of using very high vector doses. Strategies for addressing these challenges will guide the possible directions for present and future research.

First, the administration of repeated doses of the vector could allow the maintenance of the therapeutic effect. However, this is only straightforward for non-viral vectors, such as mRNA-loaded nanoparticles, although it will greatly increase the cost of this therapy [115,125]. For viral vectors, such as AAV, the induction of vector-neutralizing antibodies after the first dose prevents the use of the same vector for additional administrations. However, several strategies have been proposed to allow vector re-administrations, which include the use of alternative AAV serotypes without cross-reactivity [126], the elimination of neutralizing antibodies using IgG-degrading endopeptidases [127], and the prevention of humoral and cellular responses against the virus via co-administration of the vector with rapamycin encapsulated in LNPs [128].

Second, the combination of gene therapy vectors with pharmacological therapies, such as UDCA, could provide synergistic therapeutic effects, especially in PFIC3 patients with more severe pathology who do not respond to UDCA treatment [19]. The use of pharmacological therapies in some pediatric patients could lead to a healthier liver status, improving the vector transduction efficiency and/or allowing the administration of the

gene therapy product at an older age, at which vector genomes could be maintained for longer periods of time.

Third, the sequential therapy of non-viral vectors such as mRNA-loaded nanoparticles in pediatric patients with growing livers followed by the administration of a viral vector that allows safe and stable long-term expression of the transgene at an older age, or the combination of vectors that, after reducing liver injury, facilitate the long-term efficacy of gene therapy could be of interest.

Fourth, the improvement of gene therapy vectors by codon optimization or incorporating promoters that allow a more potent expression of the transgene with the aim of reducing the viral dose required to achieve a therapeutic effect could function to reduce the risk of toxicity from high doses [129,130]. Alternatively, inducible promoters could allow a safe, precise, and controlled expression of the transgene with physiological transgene regulation [131,132], thus avoiding unwanted effects of transgene overexpression, such as silencing, exacerbated immune responses, or cytotoxicity that could result in the elimination of the transduced hepatocytes [133,134]. The modification of the transgene via codon optimization with a reduced number of CpG motifs may also mitigate the risk of activating the Toll-like receptor 9 pathway [135], which has been theorized to result in loss of transduced hepatocytes [136].

And finally, for those cholestatic disorders in which correction of the majority of hepatocytes for a long-term therapeutic effect is likely necessary, as in the case of some PFIC subtypes [108,109], a promising alternative is the use of CRISPR/Cas9 to achieve specific gene correction by non-homologous end-joining, base editing, or prime editing. The high efficiency of liver-targeted gene delivery makes it an ideal organ for the application of gene editing strategies in animal models of PFIC [88]. However, there are still many barriers hampering the use of gene editing techniques in humans, such as reduced specificity of targeted integration leading to safety concerns, as well as the low efficacy of nonhomologous end-joining [137]. However, for most patients with more severe extrahepatic pathologies, liver-targeted gene therapy by itself will not be sufficient [109,122]. In this sense, the combination of gene therapy products targeting the liver with other therapies that allow the alleviation of extrahepatic damage could show a beneficial effect in these patients.

#### **5. Conclusions**

Although pharmacological therapies can be used to treat cholestatic diseases with milder phenotypes, they are less efficient in patients with a more severe pathology. As addressed in this review, alternative approaches, such as gene therapy, could represent a promising novel approach. To date, many preclinical studies using liver-directed gene therapy in clinically relevant animal models of both inherited and induced cholestasis have shown promising results. Although there are still many challenges for the implementation of these emerging treatments in the clinic, it is likely that some of these therapies will be approved in the near future, giving new hope for many cholestatic patients.

**Author Contributions:** J.M.-G., N.D.W. and C.S., writing—original draft preparation; J.M.-G., A.M., G.G.-A., N.D.W. and C.S., writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the following grant: Instituto Salud Carlos III financed with Feder Funds PI20/00415 ("A way to make Europe").

**Conflicts of Interest:** N.D.W. and G.G.-A. are Vivet Therapeutics employees and hold stock in the company.

## **References**


## *Review* **Update on the Pharmacological Treatment of Primary Biliary Cholangitis**

**Annarosa Floreani 1,2, Daniela Gabbia <sup>3</sup> and Sara De Martin 3,\***


**Abstract:** Ursodeoxycholic acid (UDCA) is the first-line therapy used for the treatment of PBC. In recent years, new pharmacological agents have been proposed for PBC therapy to cure UDCA-nonresponders. Obeticholic acid (OCA) is registered in many countries for PBC, and fibrates also seem to be effective in ameliorating biochemistry alteration and symptoms typical of PBC. Moreover, a variety of new agents, acting with different mechanisms of action, are under clinical evaluation for PBC treatment, including PPAR agonists, anti-NOX agents, immunomodulators, and mesenchymal stem cell transplantation. Since an insufficient amount of data is currently available about the effect of these novel approaches on robust clinical endpoints, such as transplant-free survival, their clinical approval needs to be supported by the consistent improvement of these parameters. The intensive research in this field will hopefully lead to a novel treatment landscape for PBC in the near future, with innovative therapies based on the combination of multiple agents acting on different pathogenetic mechanisms.

**Keywords:** PBC; ursodeoxycholic acid (UDCA); obeticholic acid (OCA); fibrates; FXR agonists; PPAR agonists; budesonide

## **1. Introduction**

Primary biliary cholangitis (PBC) is a chronic liver disease characterized by autoimmune responses, in which the small interlobular bile ducts are progressively disrupted, causing cholestasis. As in other chronic liver diseases, PBC can evolve into hepatic fibrosis and cirrhosis, causing the need for liver transplantation to prevent liver failure and death [1]. In regard to its geographical distribution, the highest number of patients is diagnosed in Northern Europe and North America, even though this disease is also quite common in Europe (mainly in the Southern countries), Asia, and Australia. Its global prevalence is 14.6 per 100.000, and the global incidence is 1.76 per 100,000 per year [2,3]. A gender difference can be observed in PBC patients, with a female predominancy. A F/M ratio of 9:1 was reported in a cohort series analyzing epidemiology, natural history, and clinical characteristics of PBC patients [1].

The clinical features and natural evolution of PBC may vary greatly between patients, who can experience either asymptomatic, slowly progressive, symptomatic, or rapidly evolving disease. Over the last 30 years, a modification of PBC symptoms was observed, which has changed from a disease with evident clinical manifestations, such as portal hypertension, to a milder condition, characterized by a long outcome [4]. The etiology of PBC is complex, and some mechanistic issues remain to be solved. Nevertheless, there is a general consensus indicating a predisposing genetic background that could lead to the onset of the disease in combination with infective, immunological, and/or environmental triggers [5–7]. The therapeutic management of PBC is a fascinating challenge, and several drugs with different mechanisms of action are either approved or under development (Figure 1).

**Citation:** Floreani, A.; Gabbia, D.; De Martin, S. Update on the Pharmacological Treatment of Primary Biliary Cholangitis. *Biomedicines* **2022**, *10*, 2033. https://doi.org/10.3390/ biomedicines10082033

Academic Editor: Jinghua Wang

Received: 22 July 2022 Accepted: 18 August 2022 Published: 20 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

drugs with different mechanisms of action are either approved or under development

**Figure 1.** Drugs approved (in blue) or under evaluation (in yellow) for the treatment of PBC. **Figure 1.** Drugs approved (in blue) or under evaluation (in yellow) for the treatment of PBC.

#### **2. Ursodeoxycholic Acid (UDCA) as First-Line PBC Therapy**

**2. Ursodeoxycholic Acid (UDCA) as First-Line PBC Therapy**  The standard therapy for PBC is currently ursodeoxycholic acid (UDCA), a natural hydrophilic tertiary bile acid with choleretic properties, used in clinical practice at the dose of 13–15 mg/Kg per day, according to the European guidelines. The mechanism of action of UDCA is complex and involves several molecular pathways, which have been extensively studied in preclinical settings. There is a general consensus that its therapeutic effect on PBC is mainly due to: (1) the stimulation of hepatocellular secretion and (2) the stimulation of cholangiocellular secretion, both resulting in a choleretic effect; (3) antiapoptotic effects on hepatocytes; (4) the reduction of bile acid toxicity. UDCA exerts its mechanisms of action by interacting with a panel nuclear receptors, i.e., retinoid X receptor (RXR), peroxisome proliferator-activated receptor α (PPARα), and pregnane X receptor (PXR), all of which transcriptionally modulate the synthesis and homeostasis of bile components [8]. The drug is given as a single dose, or divided into multiple doses, due to tolerability issues [9]. Observational studies evaluating PBC patients treated with UDCA helped to define an achievement of biochemical response in therapy-responding patients, evaluating also the prolongation of liver transplant (LT)-free survival, with respect to nonresponders. Altogether, the data collected with these clinical studies helped to define a population of PBC patients in which UCDA therapy is beneficial [10–15]. A multicentric study evaluating PBC patients treated with UDCA or placebo demonstrated that plasma bilirubin values below the current upper limits of normal (ULN) are predictive of survival, and a threshold of 0.6 x ULN was selected for assessing the increased risk of LT or death [16]. Furthermore, a study observed that a relevant proportion of PBC patients has an incomplete biochemical response to UDCA therapy, according to the Paris II criteria, and the presence of cirrhosis, elevated GGT, and alkaline phosphatase (ALP) at diagnosis could represent predictive factors for an incomplete UDCA response [17]. However, UDCA therapy has been demonstrated to improve LT-free survival in all PBC patients, irrespective of diseases severity and whether or not they meet the accepted criteria for the definition of a UDCA responder [18]. Thus, these observations are likely to prove that the improvement of cholestatic biochemical parameters in PBC patients, even of modest entity, can generate long-term benefits. However, despite all these efforts, the correlation between the lack of UDCA efficacy and survival in PBC patients still needs to be defined The standard therapy for PBC is currently ursodeoxycholic acid (UDCA), a natural hydrophilic tertiary bile acid with choleretic properties, used in clinical practice at the dose of 13–15 mg/Kg per day, according to the European guidelines. The mechanism of action of UDCA is complex and involves several molecular pathways, which have been extensively studied in preclinical settings. There is a general consensus that its therapeutic effect on PBC is mainly due to: (1) the stimulation of hepatocellular secretion and (2) the stimulation of cholangiocellular secretion, both resulting in a choleretic effect; (3) antiapoptotic effects on hepatocytes; (4) the reduction of bile acid toxicity. UDCA exerts its mechanisms of action by interacting with a panel nuclear receptors, i.e., retinoid X receptor (RXR), peroxisome proliferator-activated receptor α (PPARα), and pregnane X receptor (PXR), all of which transcriptionally modulate the synthesis and homeostasis of bile components [8]. The drug is given as a single dose, or divided into multiple doses, due to tolerability issues [9]. Observational studies evaluating PBC patients treated with UDCA helped to define an achievement of biochemical response in therapy-responding patients, evaluating also the prolongation of liver transplant (LT)-free survival, with respect to non-responders. Altogether, the data collected with these clinical studies helped to define a population of PBC patients in which UCDA therapy is beneficial [10–15]. A multicentric study evaluating PBC patients treated with UDCA or placebo demonstrated that plasma bilirubin values below the current upper limits of normal (ULN) are predictive of survival, and a threshold of 0.6 x ULN was selected for assessing the increased risk of LT or death [16]. Furthermore, a study observed that a relevant proportion of PBC patients has an incomplete biochemical response to UDCA therapy, according to the Paris II criteria, and the presence of cirrhosis, elevated GGT, and alkaline phosphatase (ALP) at diagnosis could represent predictive factors for an incomplete UDCA response [17]. However, UDCA therapy has been demonstrated to improve LT-free survival in all PBC patients, irrespective of diseases severity and whether or not they meet the accepted criteria for the definition of a UDCA responder [18]. Thus, these observations are likely to prove that the improvement of cholestatic biochemical parameters in PBC patients, even of modest entity, can generate long-term benefits. However, despite all these efforts, the correlation between the lack of UDCA efficacy and survival in PBC patients still needs to be defined in detail. Two groups, i.e., the Global PBC Study Group and the United Kingdom (UK)-PBC Consortium, have been created with the aim of setting up a prognostic model for disease progression in UDCA-treated patients. These two groups independently developed and evaluated the risk of PBC progression. In 2015, the first score, called the GLOBE score, was introduced to assess PBC risk progression. The setting up of this score accounted for a

wide derivation cohort (accounting for 2488 patients) and a validation cohort accounting for 1634 UDCA-treated patients. In the same years, another score was proposed by the United Kingdom (UK)-PBC Consortium, called the UK-PBC risk score (www.uk.pbc.com (accessed on 15 July 2022)), based on a nation-wide cohort of 1916 patients and validated in a cohort of 1249 UDCA-treated patients. These two predictive models have also been validated in PBC subjects not treated with UDCA, providing indications of disease activity and stage, based on biochemical liver function markers. The two main differences between the two models rely on the different endpoints used for calculating the scores. First, the GLOBE-PBC score takes into account all-cause mortality, in addition to LT-related mortality, whereas the UK-PBC score considers only liver-related death. Interestingly, in the study population of the Obeticholic Acid International Study of Efficacy (PBC POISE), both models demonstrated a potential usefulness in individualizing risk prediction, both in clinical practice and therapeutic trials for PBC [19]. It should be noted that the UDCA non-responder patients are around 30–40% of all UDCA-treated patients. Since they have a poorer prognosis due to a higher risk of disease progression, and the plausibility to require liver transplantation, as well as a greater mortality risk [20], the identification of novel effective treatments still represents an urgent medical need.

#### **3. Other Therapeutic Agents for PBC**

To overcome the problem of the incomplete response to UDCA and/or toxicity issues, several alternative therapeutic approaches have been proposed, and many clinical trials are currently ongoing to assess the possibility of repositioning approved drugs after the demonstration of their efficacy in PBC patients. Furthermore, a variety of candidate drugs are under evaluation in clinical trials for PBC patients because of their promising mechanisms of action, i.e., bile acid modulation, immunomodulation, and antifibrotic and anti-inflammatory effects. Table 1 summarizes the ongoing clinical trials.


**Table 1.** Ongoing controlled trials with experimental agents in PBC.


#### **Table 1.** *Cont*.

Abbreviations: OCA = obeticholic acid; OL = open label; RCT = randomized controlled trial; MSCs = mesenchymal stem cells; N/A = not available.

#### *3.1. Obeticholic Acid*

The only second-line drug approved for the treatment of PBC is obeticholic acid (OCA), which is indicated for patients who are non-tolerant or non-responding to UDCA after 12 months of treatment. OCA is a chemically modified derivative of BA chenodeoxycholic acid. Its mechanism relies on an agonistic activity on the farnesoid X receptor (FXR). Thanks to its affinity to FXR, OCA regulates the synthesis and export of bile acids (BAs), thereby preventing hepatic toxicity due to their toxic accumulation [21]. Beside the regulation of BA homeostasis, its complex and multifaced mechanism of action comprises anti-inflammatory and antifibrotic effects, as demonstrated by preclinical and clinical studies [21,22], thereby targeting a panel of pathological processes involved in PBC development.

The first clinical indication for the use of OCA in monotherapy came from an international randomized, double blind, placebo-controlled phase 2 study investigating the benefit of treating PBC patients with OCA in monotherapy [23]. In this study, patients were randomized into three groups, i.e., 23 PBC patients treated with placebo, 20 with

OCA (10 mg dose), and 16 patients with OCA (50 mg dose) for 3 months. As a primary endpoint, the ALP percentage change from baseline was considered. OCA significantly reduced ALP levels in patients treated at both dosage with respect to placebo. This study also reported an improvement in many biochemical parameters, among which were GGT, alanine aminotransferase, conjugated bilirubin, and immunoglobulins. The most common adverse drug reactions (ADRs) observed after OCA therapy was pruritus, which was reported in this study in patients treated with both OCA 10 mg (15%) and 50 mg (38%).

The FDA approved OCA in 2016 after the results of the phase 3 international trial POISE, with a multicentric randomized controlled design, enrolling more than 200 PBC patients [24]. Interestingly, it should be emphasized that more than 50% of UDCA-nonresponders had a beneficial effect by receiving the combination therapy of OCA plus UDCA for 12 months, as indicated by the achievement of the clinical endpoint, which was an ALP level lower than 1.67 times ULN, reduced by at least 15% from baseline. After 12 months, all patients received OCA therapy in the extension phase [25]. In the following 3-year interim analysis, OCA obtained good results on both efficacy and safety, demonstrating a stable therapeutic performance, even associated with a significant reduction in total and direct bilirubin, more evident in patients with high baseline direct bilirubin [26], and good tolerability. The most common adverse drug reactions (ADRs) reported in the POISE trial were pruritus and fatigue, which were experienced by 77% and 33% of OCA-treated patients, respectively [26]. Pruritus received the score "mild-to-moderate" by the visual analogue scale (VAS), and 8% of patients (n = 16) had to withdraw due to this ADR. However, most patients experiencing severe pruritus have been treated with specific drugs. The histological analysis of a subgroup of 17 patients recruited in the POISE trial who underwent liver biopsy at the time of enrollment and after 3 years of treatment, showed that the chronic therapy with OCA led to an improvement or at least a stabilization of the histology of PBC patients, assessed by evaluating ductular injury, fibrosis, and collagen morphometry [27]. Another sub-analysis of patients enrolled in the POISE trial investigated the beneficial effects of OCA on AST to platelet ratio (APRI) and transient elastography (TE), which are both non-invasive markers of liver fibrosis [28]. A significant APRI reduction from the baseline could be observed in OCA-treated patients and during the open-label extension phase with respect to placebo-treated group. Furthermore, the treatment with OCA 10 mg caused a decreasing tendency toward liver stiffness, while both patients treated with lower dosages of OCA or placebo showed a mean increase in liver stiffness [28]. Despite the small sample size, this study can be considered as a milestone in PBC therapy, since it demonstrated that most patients who respond inadequately to UDCA ameliorated or stabilized multiple histological PBC features when treated with OCA.

The decision to implement PBC pharmacological therapy with OCA deserves consideration if at least one of the following conditions is met: (i) ALP ≥ 1.67 x ULN (in Italy, the ALP threshold is 1.5 ULN); (ii) total bilirubin > ULN, but < 2 x ULN.

Three clinical studies analyzing real-world cohorts of OCA patients have been published so far, all reporting results for 12 months of OCA treatment [29–31]. The first real-world analysis on the effectiveness of OCA treatment was conducted on 64 Canadian PBC patients experiencing incomplete UDCA response, or who were intolerant to UDCA [29]. Among the 44 patients meeting the inclusion criteria of POISE, 39% (n = 17) underwent a 1-year biochemical evaluation. While only 18% of these patients (3 out of 17) reached the POISE primary endpoint after 12 months of treatment, 43% of patients (9 out of 21) achieved this target after a 19-month observation period. Overall, a significant ALP, GGT, transaminases, and IgM reduction was reported in the whole cohort. As regarding pruritus, either new onset or exacerbation was reported in 26 patients (41%), and 5 of them had to discontinue the drug for this reason. Other reasons for therapy discontinuation reported in this cohort were skin rash (n = 2), liver toxicity (n = 2), and incomplete response after 12 months of treatment (n = 2). In the Iberian cohort [30], 120 patients were enrolled (21.7% of them had cirrhosis and 26.7% received or were taking concomitant treatment with fibrates). A total of 78 patients completed at least 1 year of treatment. The GLOBE-PBC

score significantly decreased to 0.17 (*p* = 0.005), whereas the UK-PBC score decreased to 0.17, without reaching any significant difference (*p* = 0.11). According to the POISE criteria, 29.5% of patients achieved a response. In the Italian cohort recruited into the Italian PBC registry, 191 patients were analyzed [31], and 43% of them responded to OCA, according to the POISE criteria. Patients with cirrhosis showed lower efficacy (29.5%). Patients with AIH/PBC overlap syndrome showed a comparable efficacy to classical PBC, with a higher ALT reduction at 6 months. A further analysis was conducted in 100 cirrhotic patients from the Italian cohort (De Vincentis A, unpublished). The response to treatment, according to the POISE criteria, was obtained in 41% of cases, also confirming the efficacy of the drug in cirrhotic stage. Of note, by applying the normal range criteria, 11.5% of the cirrhotic patients reached the endpoint. A total of 22 patients (22%) discontinued the treatment due to severe side effects (5 patients with jaundice and/or ascitic decompensation and 4 with upper digestive bleeding. One patient died after TIPS placement).

#### *3.2. Non-Bile Acid FXR Agonists*

Three compounds without the classical bile acid structure, but able to bind and activate FXR, have been proposed to treat PBC patients, i.e., tropifexor, cilofexor, and EDP-305.

Tropifexor is a highly potent FXR agonist with a positive effect in treating both cholestasis and steatosis in animal models, mainly by reducing fibrosis [32]. A phase 2 study investigated tropifexor efficacy in PBC patients characterized by an inadequate UDCA response. Patients were randomized in arms, receiving once daily doses of 30 µg, 60 µg, or 90 µg of tropifexor or placebo for 4 weeks [33]. Moreover, an interim analysis was conducted in the cohort of patients treated with 90 µg. In this group of patients, a rapid decrease in the levels of GGT (72% reduction), ALP, ALT, and AST could be observed at day 28, as well as a good tolerability of tropifexor. HDL was reduced by 33% and 26% at the doses of 60 and 90 µg, respectively, and restored to physiological levels by the end of the study. No increase was observed in total or LDL cholesterol. The results of this trial suggested that this FXR agonist is a candidate drug for PBC therapy [33].

Another non-steroidal FXR agonist, called cilofexor, was tested in a trial (NCT02943447) enrolling 71 UDCA non-responders with PBC. They were randomized into three groups treated with 30 or 100 mg cilofexor or placebo once a day for 12 weeks. Patients treated with 100 mg cilofexor achieved a significant median reduction in GGT (8–47.7%, *p* < 0.001), ALT (8–13.8%, *p* = 0.05), C-reactive protein (8–33.6%, *p* = 0.03), and primary bile acids (−30.5%, *p* = 0.008). The reduction in ALP was greater than 25% in 17% of the patients treated with the dose of 100 mg and in 18% of those treated with 30 mg cilofexor, in comparison with 0% obtained in the placebo group. The major ADR observed after cilofexor treatment was pruritus, particularly common in patients treated with the higher dose. Moreover, promising results were obtained from a phase 3 trial in patients with PSC, thus suggesting the potential benefit of using this new non-bile acid FXR agonist [34].

EDP-305 is a potent FXR agonist with a "mixed" structure, containing steroid and non-steroid moieties, without the classical carboxylic acid group of the other FXR agonists or natural bile acids. The INTREPID study (NCT03394924) evaluated its safety, tolerability, and efficacy in PBC patients with inadequate response to UDCA. A total of 68 subjects were randomized to receive either EDP-305 2.5 mg, 1 mg, or placebo for 42 weeks [35]. The primary endpoint was the proportion of patients with at least 20% ALP reduction from the pre-treatment value, or normalization of ALP at week 12. The intention-to-treat analysis showed that EDP-305 resulted in ALP reduction of 45% and 46% in the 1 mg and 2.5 mg treatment groups, respectively, whereas this reduction was only 11% in the placebo group. Five patients treated with 2.5 mg EDP-305 had severe pruritus. Pruritus was present in 51% of the 2.5 mg-treated patients, whereas less than 10% of patients treated with 1 mg experienced it. In general, the other most common ADRs were gastrointestinal-related symptoms, e.g., nausea, vomiting, diarrhea, or headache, and dizziness.

#### *3.3. PPAR Agonists: Fibrates*

Fibric acid derivatives, also called fibrates, are an old class of lipid-lowering agents proposed as a second-line PBC therapy in the late 1990s. The first drug belonging to this class was clofibrate, discovered in 1962 [36]. These drugs attracted great attention for treating PBC patients because they showed efficacy against cholestasis, inflammation, and fibrosis. Their mechanism of action relies on their agonist effect on peroxisome proliferatoractivated receptors (PPARs), a family of nuclear receptors (NRs). Three main isoforms of PPARs have been described, i.e., α, β/δ, and γ, each encoded by distinct genes and characterized by a peculiar tissue distribution. Each fibric acid derivate displays a peculiar pattern of affinity towards these three PPAR isoforms, thus differently modulating PPARrelated pathways. Fenofibrate, by binding to PPARα, stimulates the transcription of the multidrug resistance protein 3 (MDR3) transporter, increasing the biliary secretion of phosphatidylcholine and improving cholestasis biomarkers [37]. At variance, bezafibrate, beside binding to PPARα and γ, is also an agonist of pregnane X receptor (PXR) [38], another transcription factor implicated in cholestatic liver disease [39]. The first placebocontrolled trial investigating the efficacy of fibrates in PBC treatment was the BEZURSO trial, a phase 3 study proposing a combination therapy with bezafibrate (BEZA) and UDCA. This study demonstrated that the addition of BEZA to the previous monotherapy of UDCA induced a significantly higher biochemical response with respect to patients of the placebo/UDCA arm [40]. This result was also associated with an improvement in PBC symptoms and surrogate markers of fibrosis. The main ADRs associated with the use of fibrates were linked to creatinine and transaminase increase and heartburn. In addition, gallstone formation, perhaps as consequence of the reduction in BA synthesis, and a paradoxical increase in cholesterol, have also been reported in PBC patients treated with clofibrate, even though the same ADRs have not been observed in patients treated with fenofibrate (FENO) or bezafibrate [41].

To compare the efficacy of OCA and fibrates as second-line therapies, a multicentric retrospective study including PBC patients from 30 centers has been undertaken in Spain [42]. A total of 86 patients receiving OCA (5 mg), 250 patients receiving fibrates (81% BEZA 400 mg, 19% FENO 200 mg), and 15 receiving OCA plus fibrates were enrolled. Both treatments decreased GGT and transaminases and improved the GLOBE score. ALP decrease was higher in patients treated with fibrates, whereas alanine aminotransferase was lower in OCA-treated patients. Discontinuation was more frequent in fenofibrate treatment due to low tolerability or the onset of ADRs. In summary, neither OCA nor fibrates emerged as a significantly better second-line treatment for PBC. Caution should be recommended, in any case.

At the AASLD meeting in Boston in 2019 [43], the results of another trial assessing the comparative efficacy of BEZA or OCA in 59 patients was presented. This study did not reveal significant differences in the incidence of severe hepatic impairment manifestations, such as varices, variceal bleeding, ascites, and LT list insertion between patients treated with OCA or bezafibrate. However, ALP reduction was more evident in bezafibrate-treated patients with respect to those treated with OCA (*p* < 0.001). A higher percentage of BEZAtreated patients experienced an elevation of bilirubin. These two studies offer great insight by presenting real-world data regarding the use of OCA and fibrates in PBC patients, paving the way for the design of future trials.

The additive effects of the combination of fibrates and OCA were investigated in a multicenter retrospective cohort of 58 patients with PBC [44]. A total of 50 of them were treated with a combination of OCA (5–10 mg/day), fibrates (BEZA 400 mg/day or FENO 200 mg/day), and UDCA (13–15 mg/day). Triple therapy was associated with a significant decrease in ALP levels with respect to dual therapy, and with an odds ratio for ALP normalization of 5.5 (95% CI: 1.8–17.1, *p* = 0.003).

Regarding the effect of fibrates on pruritus, this deserves a separate discussion. The benefit of fibrates in improving this symptom is well documented. The Fibrates for Cholestatic Itch (FITCH) trial was designed to investigate the effects of BEZA on pruritus in 70 patients with PBC, primary sclerosing cholangitis (PSC), or secondary sclerosing cholangitis who reported pruritus scored as "moderate to severe" [45]. The primary endpoint of this trial was the achievement of a reduction of more than 50% of VAS-assessed pruritus. BEZA (400 mg/day) led to this achievement in 45% of patients (41% PSC, 55% PBC), whereas only 11% reached the primary endpoint in the placebo group (*p* = 0.003). This effect in relieving cholestasis-associated pruritus occurs via an autotaxin-independent mechanism [46]. This improving effect on pruritus ensures that fibrates should be employed as a second-line option for PBC therapy in patients experiencing moderate to severe pruritus.

Since fibric acid derivates reduce cholesterol levels, they should be considered for the treatment of PBC patients with hypercholesterolemia associated with low levels of high-density lipoprotein [HDL], in whom these agents are protective against cardiovascular events.

#### *3.4. Other PPAR Agonists*

The efficacy of elafibranor (ELA), an agonist of PPAR α and δ, has been recently investigated in PBC patients enrolled in a phase 2, double-blind, placebo-controlled study [47]. A total of 45 PBC patients with inadequate UDCA response were randomized into three groups, receiving either ELA 80 mg or ELA 120 mg four times a day, or placebo four times a day for 12 weeks (NCT03124108). ELA significantly decreased mean ALP at week 12 in both groups (−48% in 80 mg-treated group and −40.6% in 120 mg-treated group, *p* < 0.001). The endpoint (ALP < 1.67 x ULN, ALP decrease >15%, and total bilirubin < ULN) was reached in most (67% and 79%) patients treated with the 80 or 120 mg doses, respectively. Moreover, in ELA caused an improvement in lipid and inflammatory markers (IgM, CRP, haptoglobin, fibrinogen) and a decrease in 7α-hydroxy-4-cholesten-3-one, or C4, an intermediate of bile acid synthesis. ELA at both dosages was well tolerated and did not cause induction or exacerbation of pruritus. In general, all these effects suggest that ELA is a promising drug candidate for PBC.

A 12-week double-blind, randomized, placebo-controlled phase 2 trial investigated the effect of seladelpar, a selective PPAR-δ agonist [48]. A total of 70 PBC patients with inadequate response or intolerance to UDCA were randomly divided into 3 experimental groups treated with 50 or 200 mg/day of seladelpar or placebo. Since 3 patients treated with seladelpar developed a grade 3 increase in ALT, even if fully reversible and asymptomatic, the study was terminated early. Despite these results, the safety and tolerability of seladelpar have been tested in a 52-week, phase 2, open-label uncontrolled dose-finding study [49,50]. This trial enrolled 120 patients who were treated for 12 weeks: 53 patients were treated with seladelpar 5 mg/day, 55 with seladelpar 10 mg/day, and 11 were assigned to the 2 mg/day group (United Kingdom sites after interim analysis), after which the dose could be increased to 10 mg/day. One year of observations indicated that seladelpar appeared to be safe and well-tolerated, while not inducing pruritus. A total of 4 patients discontinued seladelpar due to ADRs, 2 of which have been correlated to the drug treatment (grade 1 heartburn and grade 2 transaminase elevation). The composite endpoint (ALP < 1.67 x ULN, −15% reduction in ALP, total bilirubin < ULN) was met in 64% and 67% of seladelpar-treated patients. ALP normalization rates were 9%, 13%, and 33% in the 2 mg-, 5 mg-, and 10 mg-treated groups, respectively. After one year of treatment with seladelpar, 101 patients included in this trial self-reported using the pruritus VAS, the 5D-itch scale, and the PBC-40 questionnaire (evaluating itch and fatigue domains) [51]. Seladelpar led to consistent improvement in both pruritus and fatigue, along with a reduction in serum bile acid profiles. A phase 3, international, randomized, placebo-controlled study (ENHANCE) further assessed the safety and efficacy of seladelpar in PBC patients not responding to first-line treatment [52]. Enrolled participants were randomized into three groups of 80 patients: seladelpar 10 mg/day, seladelpar 5 mg/day, or placebo. After a first analysis after 26 weeks, patients were treated for an additional 26 weeks with either 5 mg or 10 mg of seladelpar. The primary endpoint was a composite response at month 3,

which included an ALP of less than 1.67 times the ULN, a ≥15% ALP reduction, and total bilirubin at or below the ULN. After one year of treatment, this study demonstrated a mean ALP decrease of 40% in the 5/10-mg group and of 45% in the 10-mg group. In addition, in the 5-mg group uptitrated to 10 mg, 53% of the patients reached the composite endpoint, as well as 69% of patients in the 10 mg group. However, this trial was terminated early due to an unexpected histological finding of non-alcoholic steatohepatitis, even though the causality assessment with seladelpar treatment was not demonstrated. These results suggest that seladelpar is a drug candidate for the second-line therapy of PBC, although further evidence about its tolerability should be obtained.

#### *3.5. Agents Targeting the FGF19 Pathway*

Fibroblast growth factor 19 (FGF19) is a hormone encoded by the *FGF19* gene, directly reducing the gene expression of CYP7A1, a key enzyme catalyzing the first rate-limiting step of bile acid synthesis [53]. Since the suppression of hepatocyte bile acid synthesis is a rational mechanism for the improvement of bile acid homeostasis and the management of cholestasis, some attempts to find novel agents acting via the FGF19 axis have been made.

An engineered analogue of FGF19, NGM282 (aldafermin), was tested in a multicentric, randomized, double-blind phase 2 trial [54]. A total of 45 PBC patients with inadequate UDCA responses were randomly assigned to three groups: one received subcutaneous daily administration of NGM282 at a 0.3 mg dose (n = 14), another received 3 mg (n = 16), and the latter received the placebo (n = 15). NGM282 treatment significantly reduced ALP (primary endpoint) at both doses compared to placebo at the end of the treatment. Moreover, 50% of the patients receiving 0.3 mg and 46% of those receiving 3 mg were shown to have an ALP reduction higher than 15% from baseline compared to 7% in the placebo group. Most ADRs were gastrointestinal disorders of grade 1 and 2. Overall, the tolerability profile of NGM282 was acceptable. However, further studies are encouraged to ascertain whether the biochemical response is durable and related to a real improvement of robust clinical outcomes, rather than to a decrease in decompensation or death.

#### *3.6. Agents Targeting the NADPH Oxidase (NOX) Enzymes*

Besides their physiological functions, NADPH oxidases (NOXs), enzymes devoted to the production of reactive oxygen species [55], play a role in multiple pathological processes characterized by excessive oxidative stress. The NOX inhibitor GKT831 (setanaxib) was investigated in a phase 2 trial including 111 patients with PBC divided into 3 arms, one receiving 400 mg of GKT831 once daily (n = 38), another twice daily (n = 36), and the latter receiving placebo (n = 37) [56]. The primary endpoint was the change in GGT vs. baseline, and the secondary endpoints were the modification in ALP, liver stiffness evaluated by means of FibroScan, and overall quality of life after 24 weeks. GKT831 led to a reduction in cholestatic markers. Particularly, a greater GGT reduction was observed in patients with higher baseline values, thus suggesting that this NOX inhibitor may be useful in patients with more advanced disease. Moreover, GKT831 was shown to be well-tolerated, with no reported treatment discontinuation or interruption due to pruritus or fatigue. Due to the positive results obtained in this trial, a phase 3 trial in PBC patients is planned.

#### *3.7. Agents with Immunomodulatory Properties*

In recent years, many studies have pointed out that immunomodulators, such for example anti-IL antibodies, Janus kinase (JAK) 1/2 inhibitors and sphingosine-1-phosphate receptor agonists, may have a potential efficacy in the treatment of PBC, since the dysregulation of innate immune system plays a fundamental role in its pathogenesis. To date, some agents with immunomodulatory properties are in early-stage preclinical and/or clinical development for PBC treatment.

Budesonide, a synthetic corticosteroid displaying a high first-pass metabolism, has been evaluated in a placebo-controlled, double-blind trial in 62 non-responder patients to UDCA [57]. Participants were randomly assigned 2:10 to receive budesonide (9 mg/day) or placebo once daily for 36 months while maintaining UDCA treatment. The primary endpoint was the improvement of liver histology with respect to inflammation and no progression of fibrosis. Comparing patients with paired liver biopsies (n = 43) the histologic endpoint was not met; moreover, serious adverse events occurred in 10 patients receiving budesonide and 7 receiving placebo. Improvements in biochemical markers of disease activity were obtained with budesonide.

Recently budesonide has been recommended for patients diagnosed with AIH/PBC overlap syndrome [58]. This treatment can improve liver function tests and is relatively safe, although the risk of portal vein thrombosis remains a concern [59].

The efficacy of rituximab, an anti-CD20 chimeric monoclonal antibody, was evaluated in two open-label studies enrolling PBC patients with incomplete UDCA response. The results of both studies suggested a limited efficacy of rituximab in PBC patients, even though an impressive reduction in ALP levels was observed [60,61] in a limited number of patients. Moreover, the treatment with rituximab was demonstrated to be ineffective in reducing fatigue in a phase 2 trial in PBC patients [62].

Since PBC hepatic histology shows a lymphocytic infiltration in portal tracts and segmental inflammatory destruction of intrahepatic bile ducts, some studies have investigated the potential effects of antibodies directed against chemokine (C-X-C motif) ligand 10 (CXCL10) in PBC patients. CXCL10 is a chemokine secreted in response to interferon-γ-stimulation by several cell types, e.g., monocytes, endothelial cells, fibroblasts, cholangiocytes, and hepatocytes, and is implicated in the hepatic recruitment of inflammatory T cells. This effect is elicited through its binding to chemokine (C-X-C motif) receptor 3 (CXCR3), expressed on effector T cells [63]. Moreover, both CXCL10 and CXRC3 have been demonstrated to be upregulated in the serum and livers of PBC patients [64]. In particular, CXCR3+ cells have been found in the hepatic tissue of PBC patients [65]. Interestingly, in situ hybridization of PBC liver samples demonstrated the presence of the CXCL10 messenger in hepatocytes surrounded by infiltrating monocytes. The anti-CXCL10 monoclonal antibody NI-0801 was evaluated in a phase 2 study enrolling 29 UDCA-nonresponder patients with PBC [66]. Each patient received an intravenous infusion of NI-0801 (10 mg/Kg, 6 doses) every 2 weeks. A 3-month follow-up was performed after the last infusion. No serious ADRs have been reported after treatment, and the most common ADRs were headaches (52%), pruritus (34%), fatigue (24%), and diarrhea (21%). However, the trial was terminated due to no significant therapeutic benefits obtained, despite the good pharmacological response observed in the blood, since the high rate of CXCL10 production makes it difficult to reach drug levels leading to an effective sustained neutralization of this chemokine [66].

Ustekimumab, a monoclonal antibody specifically binding the two interleukins IL-12 and IL-23, has been investigated in a multicentric, open-label study including PBC patients with an inadequate response to UDCA. Unfortunately, the results of this study failed to demonstrate the efficacy of this antibody in achieving a decrease, even moderate, in ALP levels [67].

Another open-label trial investigating abatacept, a fusion protein formed by the extracellular domain of the CTL4 and Fc region of the immunoglobulin IgG1, has demonstrated the inefficacy of this protein in achieving the required clinical outcomes [68].

Baricitinib is a JAK inhibitor, selective for the two subtypes JAK1 and JAK2, already approved in the US and Europe for the treatment of other autoimmune diseases, e.g., rheumatoid arthritis, and alopecia areata. JAK is a family of intracellular tyrosine kinases transducing cytokine-mediated signals. A randomized, double-blinded placebo-controlled trial in patients with PBC and inadequate response or intolerance to UDCA was performed [69]. Endpoints included change in ALP, itch numeric rating score, and fatigue scoring at 12 weeks post-baseline. Only two patients were enrolled and completed the trial (one received baricitinib and the other placebo). The patient treated with baricitinib demonstrated a 30% decrease in ALP and a 7-point improvement in itch scoring, but a 2-point increase in fatigue scoring.

FFP-104, an anti-CD40 monoclonal antibody, is a novel agent proposed for the treatment of PBC, since CD40 promotes the efficient T cell activation caused by the paracrine communications of antigen presenting cells, fibroblasts, and other non-lymphoid cells. As a consequence, its blockade could be exploited to counteract PBC autoimmune activation. A phase 2 trial including PBC patients is currently ongoing to determine the initial safety, tolerability, and pharmacodynamics of this antibody in PBC patients (NCT02193360). Interestingly, in a murine model of autoimmune cholangitis, administration of the anti-CD40 ligand reduced liver inflammation and lowered the levels of AMA, but these reductions were not sustained [70].

Mesenchymal stem cells (MSCs) transplantation has been studied as alternative to liver transplantation for patients with end-stage PBC [71]. MSCs are able to modulate and repair the injured tissue by affecting immune response by different mechanisms, such as cell-to-cell interactions and the production of useful paracrine factors [72]. The first clinical trial evaluating MSCs for PBC treatment was conducted in China (NCT01662973). This pilot study enrolled a small number of patients (n = 7) with an incomplete response to assess the safety and efficacy of umbilical cord-derived mesenchymal stem cells (UC-MSCs) [73]. All patients received 3 infusions of UC-MSCs every 4 weeks. After 48 weeks of follow-up, the treatment was well tolerated, and no relevant ADRs occurred. UC-MSCs significantly decreased serum levels of ALP and GGT. After these encouraging results, a second study was performed by the same research group testing MSCs derived from allogenic donors of the patients' family members [74]. Their efficacy was evaluated using a 1-year of follow-up. Although transaminases, GGT, and IgM were significantly improved, the histological analyses evaluating the presence and severity of fibrosis were stabilized by the treatment. Overall, further studies seem to be necessary to discriminate the real efficacy of the use of MSC therapy in PBC. A new study is currently ongoing (NCT03668145).

#### *3.8. Antiretroviral Therapy*

After the proposal of a Canadian research group of a viral involvement in the pathogenesis of PBC, a multicentric trial was designed to investigate the efficacy of antiretroviral therapy in PBC patients (NCT01614405) in a limited number of patients (n = 13), since most enrolled patients were intolerant to the lopinavir-ritonavir (LPRr) combination. Patients were randomized and received a combination of tenofovir-emtricitabine (TDF/FTC 300/200 mg), LPRr (800/200 mg), or placebo for 6 months [75]. A significant 25% reduction in ALP was observed after antiretroviral therapy (*p* < 0.05). However, an important limitation to the use of the antiviral combination was represented by the frequency of ADRs, which were much higher than those reported in HIV patients receiving the same therapy. A new trial investigating better-tolerated combination regimens is ongoing (NCT03954327). Another antiretroviral therapy with tenofovir/emtricitabine-based regimens in combination with lopinavir or raltegravir in recurrent PBC following liver transplantation improved hepatic biochemistry, but the antiretroviral therapy was associated with side-effects [76].

#### **4. Agents for the Treatment of Specific Symptoms of PBC**

#### *4.1. Agents Targeting Pruritus*

Pruritus represents a frequent and troublesome symptom, reported in 60–70% of patients [77,78]. Its pathogenesis is complex, and the results regarding therapy with UDCA showed that it was mostly ineffective in improving this symptom. Since the principal guideline-approved anti-pruritic agents, e.g., cholestyramine, rifampicin, naltrexone, and sertraline, are often ineffective to improve PBC-related pruritus, novel agents targeting this symptom have been developed and are under evaluation.

#### Ileal Bile Acid Transporter (IBAT) Inhibitors

The use of ileal bile acid transporter inhibitors has been suggested for the treatment of PBC-related pruritus due to their ability of decrease retained circulating BAs. IBATs is physiologically devoted to BA reabsorption from the ileum, thus maintaining their enterohepatic circulation. Since in many cholestatic liver diseases, ileal BA absorption is increased, several compounds capable of altering the ileal reabsorption of bile acids have been proposed and are discussed below.

Maralixibat, a selective, sodium-dependent, ileal apical, BA transport inhibitor was tested in a phase 2 trail in which its efficacy and safety were assessed in PBC patients with pruritus (CLARITY study [79]). Patients were divided into arms and treated for 13 weeks with either maralixibat (10 or 20 mg/day) or placebo, in addition to the standard UDCA therapy, when tolerated. The primary endpoint was defined as "adult itch reported outcome average sum score" from baseline to the end of the study. The main ADRs were gastrointestinal disorders, which were common in treated (78.6%) but also placebo (50%) patients. Despite an improvement of baseline risk scores, maralixibat caused no significant improvement of pruritus with respect to placebo.

The IBAT inhibitor GSK2330672, also called linerixibat, was evaluated in a phase 2 trial enrolling 21 patients [80] to assess the safety and tolerability of GSK2330672. The secondary endpoints were changes in patient-reported pruritus scores, assessed by means of different scales, namely a 0 to 10 numerical rating scale, PBC-40 itch domain score, 5-D itch scale, and changes in circulating bile acid levels (NCT01899703). Linerixibat was well tolerated, and diarrhea was the most experienced ADR. The percentage decrease in itch scores was −57% in the numerical rating scale, −31% in the PBC-40 itch domain, and −35% in the 5-D itch scale in linerixibat-treated patients, and these differences were statistically significant with respect to the placebo-treated group. A larger phase 2 study enrolling 147 patients is still ongoing to confirm these beneficial effects on PBS-related pruritus and to further assess the drug tolerability (NCT02966834).

The last proposed IBAT inhibitor, A4250 (odevixibat), was tested in an open-label phase 2 study that aimed to assess drug tolerability and efficacy in improving pruritus in 9 PBC patients (NCT02360852) [81]. Patients were treated with odevixibat at a dose of 0.75 mg (n = 4) or 1.5 mg (n = 5) for 4 weeks. A remarkable improvement in pruritus was observed in all 9 odevixibat-treated patients assessed by VAS, the 5-D itch scale, and the pruritus domain of the PBC-40 questionnaire [82]. Unfortunately, tolerability was low because of gastrointestinal symptoms. Odevixibat received its first approval in the EU in July 2021 for the treatment of progressive familial intrahepatic cholestasis (PFIC) in patients aged ≥ 6 months, followed by its approval in the US for pruritus in patients with PFIC aged ≥ 3 months [83].

#### *4.2. Agents Targeting Fatigue*

Another frequently reported PBC manifestation is fatigue, a complex syndrome characterized by feelings of discomfort, exhaustion, and lethargy that could significantly reduce the quality of life. The probability of improving fatigue after LT in advanced PBC is roughly 50% [84]. Currently, no pharmacological treatment is approved for PBC-related fatigue. The only prescribed suggestion is an exercise increase, even though this kind of prescription needs further evaluation. The results of a pilot study showed an improvement in muscle pH in PBC patients, and an amelioration of fatigue, social, and emotional symptoms in patients following an exercise program [85].

The first phase 2 randomized controlled trial of treatment of PBS-associated fatigue and daytime somnolence (NCT2376335, [62]) was performed in 57 PBC patients with moderate to severe fatigue. Patients were randomized to receive two doses of rituximab (1000 mg) or placebo. The primary outcome was assessed by measuring fatigue severity using a questionnaire at 3 months. The rationale of the use of rituximab was an improvement in the fatigue associated with a variety of other autoimmune diseases, e.g., Sjogren's syndrome, which has been also association with PBC. Rituximab, however, failed to show an improvement in fatigue in PBC patients.

Modafinil, an agent acting on the central nervous system and used for the treatment of daytime somnolence in narcoleptic patients, was tested in an open study enrolling 21 patients with PBC experiencing daytime somnolence and fatigue [86]. The starting dose

of modafinil was 100 mg/day, which was titrated according to patient's tolerability and response. Unfortunately, only 14 patients could tolerate the full 2-month treatment, although in those patients, an improvement of excessive daytime somnolence and associated fatigue was observed. The suggestion from these data was to improve the design of the study with a placebo-controlled trial, to confirm modafinil's efficacy against fatigue.

A preclinical study on an animal model of hepatic cholestasis induced by the ligation of the bile duct demonstrated that early OCA administration was able to improve cognitive impairment [87]. Otherwise, these preclinical observations need to be validated in further studies.

A systematic meta-analysis of 16 studies evaluating UDCA, liver transplantation, serotonin reuptake inhibitors, colchicine, methotrexate, cyclosporine, modafinil, and OCA found some improvement in fatigue with liver transplantation, but a lack of high-quality evidence supporting the efficacy of any other intervention in the treatment of PBC-related fatigue [88].

#### **5. Conclusions**

In recent years, new candidate drugs have undergone or completed phase 2 and 3 clinical trials on PBC patients who did not respond to the first line therapy with UDCA. OCA represents the most promising drug and is approved in many countries for this indication. Fibrates seem to effectively ameliorate biochemistry alteration and symptoms typical of PBC. Moreover, a variety of new agents, acting with different mechanisms of action, are under clinical evaluations for PBC treatment, e.g., PPAR agonists, NOX inhibitors, immunomodulators, and mesenchymal stem cells transplantation. Even though most of these approaches seem to have beneficial effects on biochemical endpoints, no data are currently available regarding robust endpoints, such as transplant-free survival; thus, their clinical use needs to be supported by the consistent improvement of these parameters. In general, data on the efficacy of the new therapeutic agents are still undergoing investigation in clinical trials and are too premature to provide practical information to physicians. The crucial point when designing clinical trials is the choice of a combination treatment with nuclear receptor ligands and other agents with different mechanism and therapeutic effects [89]. This huge armamentarium of new therapeutic options will likely lead to a novel treatment landscape for PBC in the near future, with novel therapies based on the combinations of multiple agents acting on different pathogenetic mechanisms [90]. Another crucial point is that the ideal therapy for PBC would achieve a complete biochemical remission, namely normalization of serum ALP and bilirubin, and would be well tolerated. Furthermore, the ideal therapy must be safe for patients with advanced or decompensated disease and should aim to reduce the need for liver transplantation [91].

**Author Contributions:** Conceptualization, A.F. and S.D.M.; writing—original draft preparation, A.F. and D.G.; writing—review and editing, S.D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** A.F. has received consultancy fees during the last two years from Intercept.

#### **References**


## *Review* **Obeticholic Acid for Primary Biliary Cholangitis**

**Annarosa Floreani 1,2,\*, Daniela Gabbia <sup>3</sup> and Sara De Martin <sup>3</sup>**


**Abstract:** Primary biliary cholangitis (PBC) is a rare autoimmune cholestatic liver disease that may progress to fibrosis and/or cirrhosis. Treatment options are currently limited. The first-line therapy for this disease is the drug ursodeoxycholic acid (UDCA), which has been proven to normalize serum markers of liver dysfunction, halt histologic disease progression, and lead to a prolongation of transplant-free survival. However, 30–40% of patients unfortunately do not respond to this first-line therapy. Obeticholic acid (OCA) is the only registered agent for second-line treatment in UDCAnon responders. In this review, we focus on the pharmacological features of OCA, describing its mechanism of action of and its tolerability and efficacy in PBC patients. We also highlight current perspectives on future therapies for this condition.

**Keywords:** primary biliary cholangitis; obeticholic acid; ursodeoxycholic acid; farnesoid X receptor

### **1. Introduction**

Primary biliary cholangitis (PBC) is a chronic disease characterized by the accumulation of bile acids in the liver, potentially progressing to cirrhosis, end-stage liver disease, hepatocellular carcinoma, and even death [1]. The existence of gender differences in PBC development has been widely reported. Indeed, PBC develops more frequently in females than males [1]. In the global population, a prevalence of 14.6 cases per 100,000 people has been observed, with a female:male ratio of 9:1, and 1.76 new cases diagnosed per 100,000 people each year [2]. Due to more careful routine testing and/or incompletely understood changes in environmental factors, the definition and outcome of PBC have been reconsidered over the last 30 years, from a severe symptomatic disease characterized by symptoms of portal hypertension to a milder disease with a long natural history [3]. As a consequence, many patients are asymptomatic, and most new diagnoses (up to 60%) are made after the discovery of increased serum biochemical markers of liver function during check-ups performed for unrelated purposes [4,5]. This autoimmune cholestatic disease is characterized by increased plasma levels of alkaline phosphatase (ALP) and the presence of a high titer of antimitochondrial antibodies (AMAs) in over 90% of patients, as well as a PBC-specific anti-nuclear antibody (ANA). The current EASL guidelines suggest that a diagnosis of PBC can be determined in adult patients in the presence of cholestasis and the absence of other systemic diseases, when the ALP value is elevated and AMAs are present with a titer >1:40 [6].

Ursodeoxycholic acid (UDCA) represents the gold standard for PBC therapy, and it is generally administered as a daily oral treatment (recommended dose: 13–15 mg/kg) [6]. UDCA therapy improves liver transplantation (LT)-free survival in PBC patients, including those with early and advanced disease, and also in patients who did not meet the accepted criteria for UDCA response [7]. Even though the improvement of biochemical parameters after UDCA treatment is modest, patients experience a long-term benefit in terms of improved survival. Regardless, non-responders represent 30–40% of all UDCA-treated patients, and globally have a higher risk of PBC progression and a greater need for transplant

**Citation:** Floreani, A.; Gabbia, D.; De Martin, S. Obeticholic Acid for Primary Biliary Cholangitis. *Biomedicines* **2022**, *10*, 2464. https://doi.org/10.3390/ biomedicines10102464

Academic Editors: Giovanni Squadrito and Francesco Vasuri

Received: 23 July 2022 Accepted: 28 September 2022 Published: 2 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

than responder patients, as well as a higher mortality [8]. A young age at diagnosis and male sex have been associated with a reduced chance of biochemical response to UDCA therapy in a large cohort study from the UK-PBC study group [9]. Accordingly, another large, multicenter long-term follow-up study (*n* = 4355) found that young PBC patients (aged <45) had significantly lower response rates to UDCA than their older counterparts (aged >65) [10]. However, the biological mechanisms underpinning this clinical observation in non-responders to UDCA are far from completely understood. for transplant than responder patients, as well as a higher mortality [8]. A young age at diagnosis and male sex have been associated with a reduced chance of biochemical response to UDCA therapy in a large cohort study from the UK-PBC study group [9]. Accordingly, another large, multicenter long-term follow-up study (*n* = 4355) found that young PBC patients (aged <45) had significantly lower response rates to UDCA than their older counterparts (aged >65) [10]. However, the biological mechanisms underpinning this clinical observation in non-responders to UDCA are far from completely understood. Therefore, the proposal of a second-line therapy devoted to UDCA non-responders

treated patients, and globally have a higher risk of PBC progression and a greater need

Therefore, the proposal of a second-line therapy devoted to UDCA non-responders provides the rationale to overcome the observed limitations of drug efficacy. To date, obeticholic acid (OCA) represents the only second-line treatment recommended for nonresponder PBC patients, which are intolerant to UDCA therapy or in whom a 12 monthtreatment haven't produced benefit. As demonstrated by clinical trials, including the phase III POISE study described in detail below, OCA is effective in improving the serum and histological endpoints of PBC patients in monotherapy. In this review, we focus on the mechanism of action of OCA and its tolerability and efficacy in PBC, and offer a perspective on the future treatment of this condition. provides the rationale to overcome the observed limitations of drug efficacy. To date, obeticholic acid (OCA) represents the only second-line treatment recommended for non-responder PBC patients, which are intolerant to UDCA therapy or in whom a 12 monthtreatment haven't produced benefit. As demonstrated by clinical trials, including the phase III POISE study described in detail below, OCA is effective in improving the serum and histological endpoints of PBC patients in monotherapy. In this review, we focus on the mechanism of action of OCA and its tolerability and efficacy in PBC, and offer a perspective on the future treatment of this condition.

#### **2. Pharmacological Actions of OCA 2. Pharmacological Actions of OCA**

*Biomedicines* **2022**, *10*, x FOR PEER REVIEW 2 of 11

OCA, a synthetic derivative of the bile acid (BA) chenodeoxycholic acid, is an agonist of the farnesoid X receptor (FXR) [11], a key nuclear receptor mainly expressed in the liver and gut, which orchestrates complex signaling pathways related to the homeostasis of bile acids (BAs) (Figure 1). In vitro pharmacological studies have demonstrated that OCA is an FXR agonist with a potency 100 times higher than endogenous BAs [12]. BA synthesis occurs in the liver starting from hepatic cholesterol. After their synthesis, BAs are secreted into the gut to help digestion and consequently the absorption of nutrients, in particular lipids and liposoluble vitamins, by virtue of their emulsifying ability [13]. After their secretion, about 95% of BAs are reabsorbed from the terminal ileum, thus entering into the enterohepatic circulation. As FXR agonists, BAs themselves participate in the finely tuned regulation of their own synthesis and secretion through the modulation of FXR activation. In PBC-related cholestasis, the enterohepatic circulation of BAs is impaired, leading to hepatic inflammation and damage. OCA, a synthetic derivative of the bile acid (BA) chenodeoxycholic acid, is an agonist of the farnesoid X receptor (FXR) [11], a key nuclear receptor mainly expressed in the liver and gut, which orchestrates complex signaling pathways related to the homeostasis of bile acids (BAs) (Figure 1). In vitro pharmacological studies have demonstrated that OCA is an FXR agonist with a potency 100 times higher than endogenous BAs [12]. BA synthesis occurs in the liver starting from hepatic cholesterol. After their synthesis, BAs are secreted into the gut to help digestion and consequently the absorption of nutrients, in particular lipids and liposoluble vitamins, by virtue of their emulsifying ability [13]. After their secretion, about 95% of BAs are reabsorbed from the terminal ileum, thus entering into the enterohepatic circulation. As FXR agonists, BAs themselves participate in the finely tuned regulation of their own synthesis and secretion through the modulation of FXR activation. In PBC-related cholestasis, the enterohepatic circulation of BAs is impaired, leading to hepatic inflammation and damage.

**Figure 1.** Molecular mechanism of hepatic OCA pharmacodynamics. OCA activates FXR, thereby triggering cellular pathways leading to a reduction in the synthesis and hepatic uptake of BAs, and an increase in their efflux from the liver. Furthermore, OCA acts on LSEC and KC, exerting anti-inflammatory and antifibrotic effects by reducing the production of proinflammatory **Figure 1.** Molecular mechanism of hepatic OCA pharmacodynamics. OCA activates FXR, thereby triggering cellular pathways leading to a reduction in the synthesis and hepatic uptake of BAs, and an increase in their efflux from the liver. Furthermore, OCA acts on LSEC and KC, exerting anti-inflammatory and antifibrotic effects by reducing the production of proinflammatory cytokines

and HSC activation, respectively. Abbreviations: farnesoid X receptor (FXR), retinoid X receptor (RXR), bile acid (BA), Kupffer cell (KC), liver sinusoidal endothelial cell (LSEC), hepatic stellate cell (HSC), small heterodimer partner (SHP), liver receptor homolog 1 (LRH-1), fibroblast growth factor-19 (FGF-19), sodium taurocholate co-transporting polypeptide (NTCP), bile salt export pump (BSEP), multidrug resistance protein-3 (MDR3), organic solute transporters (OST), transforming growthfactor β (TGFβ), connective tissue growth factor (CTGF), platelet-derived growth factor β-receptor (PDGFR-β), monocyte chemo-attractant protein-1 (MCP1), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), inhibitor of kB (IκB).

Similar to other nuclear receptors [14,15], upon activation, FXR binds to the retinoid X receptor (RXR). The binding of the FXR–RXR heterodimer to DNA responsive elements results in the induction of the small heterodimer partner (SHP) gene, finally causing the transcriptional repression of rate-limiting enzymes in BA synthesis, such as cytochrome P450 (CYP)7A1 and liver receptor homolog 1 (LRH-1) [16]. LRH-1 is a transcription factor with a key role in the regulation of BA and cholesterol homeostasis, and also in coordinating a panel of other hepatic metabolic processes [17]. In addition, FXR stimulates the synthesis of fibroblast growth factor-19 (FGF-19), which in turn participates in the inhibition of CYP7A1 and CYP8B1 expression through the fibroblast growth factor receptor-4 (FGFR4) pathway in hepatocytes [18]. As a result, the above-described FXR/SHP and FXR/FGF19/FGFR4 pathways are major negative regulators of BA synthesis. Furthermore, FXR inhibits the sodium taurocholate co-transporting polypeptide (NTCP) via SHP, thereby repressing hepatic BA uptake [19]. FXR activation also increases the efflux of BAs from the liver to the canalicular lumen by targeting the transporter bile salt export pump (BSEP) and multidrug resistance protein-3 (MDR3), triggering another mechanism responsible for the anticholestatic effects of FXR agonists [20]. FXR activation also leads to an increase in the expression of the organic solute transporters OSTα and β, which also enhance BA efflux from the liver to the portal vein [21]. Besides its pivotal activity as a BA-responsive transcription regulator of BA synthesis and metabolism, as described in detail above, it has been demonstrated that FXR-mediated signaling plays a role in hepatic fibrogenesis, although controversial results have been obtained regarding this function. Hence, it has been observed that FXR knock-out mice develop hepatic inflammation, fibrosis, and liver tumors over time [22] and, accordingly, it has been demonstrated that OCA-induced FXR activation reduced liver fibrosis in two different experimental in vivo models of liver fibrosis [23]. Other authors have suggested that FXR in liver fibrosis models can be either detrimental or irrelevant, depending on the type of damage [24]. Notably, no direct effects of FXR agonists could be observed on the activation of cultured hepatic stellate cells (HSCs) [25,26], which are the main cell types triggering the fibrogenesis process [27].

OCA exerted both anti-inflammatory and ant-fibrotic effects by targeting the activation of both liver sinusoidal endothelial cells (LSECs) and Kupffer cells [26]. In particular, OCA reduces the production of inflammatory cytokines and chemokines (transforming growth-factor β, connective tissue growth factor, platelet-derived growth factor β-receptor, monocyte chemo-attractant protein-1) by these two types of sinusoidal cells, which in turn activate HSCs [28]. Hence, the mechanism of the anti-inflammatory effect relies on the inhibition of the NF-κB signaling pathway via the up-regulation of its inhibitor IκBα. In summary, OCA acts by a complex mechanism, comprising several actions: (a) the regulation of bile acid transport; (b) the reduction in inflammation; (c) the modulation of cellular pathways triggering fibrogenesis [29]. Due to the induction of a signaling pathway which modulates the activity of fibroblast growth factor-19 (FGF-19), OCA exerts greater hepatoprotection than UDCA. OCA also induces the expression and secretion of gut-derived hormones, e.g., FGF-19 [30]. This hormone is absorbed and secreted by enterocytes into the portal blood, thereby reaching the liver through the portal venous system. In the liver, FGF-19 is involved in the anticholestatic mechanisms described above.

#### **3. Pre-Registration Studies**

OCA has been evaluated in monotherapy in a phase II study in which PBC patients were enrolled with the aim of assessing its benefit in the absence of UDCA treatment [31]. After randomization, patients were treated with a placebo (23 patients), or two doses of OCA (10 mg in 20 patients and 50 mg in 16 patients) for 3 months, and followed up by a 6-year open-label extension. The ALP reduction, measured as the percentage difference from the baseline, was evaluated as the primary endpoint of this study. The treatment with both dosages induced a significant ALP reduction compared to the placebo. Accordingly, other plasma parameters were reduced in OCA-treated patients, e.g., conjugated bilirubin, GGT, AST, and immunoglobulins. In this study, the most common adverse effect reported after OCA treatment was pruritus, having been experienced by 15% of the 10 mg-treated patients and 38% of the 50 mg-treated patients.

The first approval of OCA was obtained following the results of a phase III trial that enrolled 216 patients [32], and demonstrated that about 59% of UDCA-non-responders benefitted from a one-year treatment with a combination of OCA and UDCA. These patients reached the clinical endpoint, set as an ALP level of less than 1.67 times the upper limit of the normal range, with a reduction of at least 15% from the baseline). Thereafter, the study underwent an open-label extension phase in which 193 enrolled patients were switched to OCA treatment [33]. The results of the following 3-year interim analysis showed that OCA therapy was well tolerated and could be demonstrated to maintain its performance over time. Additionally, a post-hoc analysis revealed that OCA induced a significant bilirubin reduction (both total and direct) that was particularly evident in those patients with a high baseline value of direct bilirubin [34]. This analysis thus confirmed the beneficial effects of OCA therapy in high-risk patients. Furthermore, the histological analysis of liver biopsies at baseline and after a 3-year treatment with OCA in a subgroup of patients (*n* = 17) revealed the improvement or stabilization of a panel of histologic disease features, e.g., ductular injury, fibrosis, and collagen morphometry [35]. This analysis, despite the limited number of assessed liver biopsies, further demonstrated that OCA is effective in UDCA-non-responders. The most reported adverse effects related to OCA treatment were pruritus and fatigue, which were experienced by 77% and 33% of patients, respectively [34]. As regards pruritus, only 8% of the OCA-treated patients interrupted the treatment during the open-label extension phase and, in general, patients reported a mild-to-moderate pruritus, and those experiencing severe pruritus were treated with specific medication after a clinical consult. In general, the results of this clinical trial demonstrate that 3 years of OCA treatment were efficient in ameliorating or stabilizing multiple histological features of PBC in most patients with an inadequate UDCA response, and supported the approval of OCA from the FDA in 2016.

Another sub-analysis of the above-reported trial observed that OCA treatment induced a significant reduction in the AST to platelet ratio (APRI). This effect was observed after a 1-year treatment and in the open-label extension phase in the groups treated with 10 and 50 mg OCA with respect to the placebo [36]. Liver stiffness (LS) was evaluated in 39 patients randomized and dosed with the placebo, 35 patients dosed with OCA 5–10 mg, and 32 patients dosed with OCA 10 mg. LS at baseline was 12.7 ± 10.7, 10.7 ± 8.6, and 11.4 ± 8.2 kPa, respectively. During the double-blind and open-label phases, a decrease, while not significant, was only observed in the OCA 10 mg group, while both the OCA 5–10 mg and placebo groups displayed mean increases in liver stiffness [36]. In other words, a trend towards a reduction in LS was observed only in the arm treated with the highest dose of OCA. In another scenario, namely non-alcoholic steatohepatitis, patients enrolled in the phase III REGENERATE study with OCA showed a significant reduction in LS after 18 months in the OCA 25 mg group vs. the placebo [37]. Thus, the assessment of the antifibrotic activity of OCA in a clinical setting has several limitations, mainly considering that changes in LS occur during a median interval of 2 years.

The main pre-registration studies evaluating the efficacy and safety of OCA are reported in Table 1.


**Table 1.** Summary of the main pre-registration studies described in the text.

#### **4. Real-World Data on OCA**

Currently, OCA is available as tablets containing 5 and 10 mg under the brand name Ocaliva. Typically, therapy for PBC patients is started with the administration of an initial dose of 5 mg once daily, which can be titrated to a maximum of 10 mg daily [40]. The general recommendation for patients with advanced cirrhosis (Child–Pugh B or C) is to start with a dose of 5 mg once weekly, which is then increased to a maximum of 10 mg twice weekly if the drug is well-tolerated.

The most significant ADRs caused by OCA therapy which have been reported in clinical trials are pruritus, fatigue, nausea, and headache. To a minor extent, hypersensitivity reactions and depression have also been observed [40]. As far as pruritus is concerned, it appears to be less severe if the patients are initially treated with a low dose, which can then be gradually increased. As a consequence of the alteration of lipid metabolism, which is due to other molecular signaling pathways triggered by FXR activation, an increase in total serum lipid levels and a small decrease in high-density lipoprotein (HDL) have also been reported in PBC patients treated with OCA, but to date these effects have not been correlated to a long-term increased cardiovascular risk [30].

Real-world data are crucial for understanding treatment effectiveness and safety in everyday clinical practice where: (i) patients' characteristics are more heterogeneous with respect to sub-phenotypes, e.g., cirrhosis and overlap syndrome between PBC and AIH; (ii) the treatment schedule may be less rigid and more "personalized" by each treating physician. A number of post-registration clinical trials are ongoing and recruiting patients (Table 2).


**Table 2.** Ongoing clinical trials recruiting patients for post-registration efficacy assessment.

Three real-world cohorts have been published thus far (Table 3), all reporting results for 12 months of OCA treatment [41–43]. Altogether, 375 patients treated with OCA were included in these three studies. The main characteristics of the three cohorts are respectively described in Table 3. The inclusion criteria were: hepatologist's discretion for the Canadian cohort, lack of response to Paris II criteria [44] for the Iberian cohort and ALP >1.5 times the normal according to the Italian Medicines Agency (AIFA) for the Italian cohort. The percentages of patients with cirrhosis were 6.3, 10, and 15%. The percentages of response at 12 months according to the POISE criteria were respectively 18, 29.5, and 51.9%. Due to the retrospective design of these studies, a comparable evaluation of the response to OCA is impossible. However, it has to be pointed out that in the Italian cohort, with one third of cirrhotic patients, the response rate was lower due to the higher drop-out and higher

levels of bilirubin at baseline in cirrhotic patients. Within the Canadian cohort, 11 patients (17%) had a permanent discontinuation of treatment (2 of them with Child–Pugh A and B respectively) for suspected hepatotoxicity. The first case was a 67-year-old female who discontinued OCA due to an increase in ALP. The second patient was a 54-year-old female who developed severe cholestatic cirrhosis, who was transplanted for severe complications. Within the Iberian cohort, a total of 14 patients (11.67%) discontinued the treatment due to severe adverse events or decompensation of cirrhosis. Within the Italian cohort, 33 patients (17%) discontinued OCA for pruritus or other side-effects. In the same cohort, factors associated with a lack of response at 12 months were: previous treatment with fibrates, high levels of ALP at baseline, and high levels of bilirubin at baseline [43].


**Table 3.** Real-world data in three cohorts of patients with PBC.

A further analysis was performed in 100 cirrhotic patients of the Italian cohort [45]. The response to treatment was obtained in 41% of cases, according to the POISE criteria, confirming OCA efficacy at this stage as well. In this case, the use of the normal range criteria means that the endpoint was reached by only 11.5% of the cirrhotic patients. Regarding the reported severe adverse effects, 22% of patients discontinued OCA therapy: 5 patients due to jaundice and/or ascitic decompensation, 4 due to upper digestive bleeding, and 1 subject died after the substitution of a transjugular intrahepatic portosystemic shunt.

A sub-analysis from the Italian and Iberian cohorts found that patients with PBC/AIH overlap syndrome had a similar response after OCA treatment [42,43].

Two further real-world studies were presented at an AASLD virtual meeting in 2020. The first study, derived from the GLOBAL PBC group, enrolled 290 patients in 11 centers located between Europe, North America, and Israel [46]. Among them, 215 patients met the POISE criteria for eligibility, 60 patients possessed available biochemical data for a period of 12 months, and 35% of patients reached the pre-defined POISE primary endpoint after 1 year of treatment. The second study was conducted on 319 patients that received OCA therapy between May 2016 and September 2019, and were considered eligible for OCA according to laboratory databases and American administrative claims [47]. According to the Toronto criteria, the proportion of patients achieving a biochemical response to the treatment was 48% after 1 year, 58% after 2 years, and 55% after 3 years which marked the end of the follow-up period [48]. More recently, a large nationwide experience of secondline therapy in PBC has been reported [49]. The study was conducted from August 2017 to June 2021 across 14 centers in the UK. A total of 457 PBC patients with an inadequate response to UDCA were recruited. Overall, 259 patients received OCA and 80 received fibrates (fibric acid derivatives) and completed 12 months of therapy, yielding a dropout rate of 25.7% and 25.9%, respectively. Treatment efficacy was quantified by the proportion of patients attaining a biochemical response according to propensity score matching. The 12-month biochemical response rates were 70.6% with OCA and 80% under fibric acid treatment, without reaching any statistical significance.

With the objective of evaluating the time to first occurrence of liver transplant or death, OCA-treated patients in the POISE trial and open-label extension were compared with non-OCA-treated external controls [50]. Propensity scores were generated for external control patients meeting POISE eligibility criteria from 1381 patients in the Global PBC

registry study and 2135 in the UK PBC registry. Over the 6-year follow-up, patients treated with OCA had a significantly greater transplant-free survival than comparable external control patients.

#### **5. Combined Therapy with OCA and Fibrates**

Fibrates, well-known agents with anti-lipidemic properties, were proposed as a secondline treatment because their beneficial effects on inflammation, cholestasis, and fibrosis are documented, resulting from their activity as peroxisome proliferator-activated receptor (PPAR) agonists. Fibrates have different affinities to the three main PPAR isoforms, PPARα, PPARβ/δ, and PPARγ, and consequently can activate different signaling pathways. As an example, fenofibrate, a PPARα agonist, upon binding to its receptor, increases the expression of multidrug resistance protein 3 (MDR3) [51]. Furthermore, it increases biliary phosphatidylcholine secretion, thus ameliorating a recognized biomarker of cholestasis. Bezafibrate acts as a dual agonist of PPARα and PPARγ and is also a pregnane X receptor (PXR) agonist [52]. The BEZURSO trial is a Phase III study, employing bezafibrate in combination with UDCA, and was the first placebo-controlled trial evaluating the use of fibrates as a second-line treatment for PBC. In this study, the second-line combination therapy of bezafibrate and UDCA was effective in obtaining a complete biochemical response with a rate significantly higher than that observed in patients treated with a placebo and UDCA [53]. This regression was associated with a concurrent improvement of both symptoms and surrogate markers of liver fibrosis. The most frequently reported ADRs of fibrates include increased levels of creatinine and transaminases and heartburn. As a consequence of its main mechanism of action involving a reduction in BA synthesis, clofibrate treatment can lead to the formation of gallstones and hypercholesterolemia [54], two events which have not been observed during treatment with fenofibrate or bezafibrate.

A triple therapy with UDCA, OCA, and fibrates was studied in a multicenter retrospective cohort of patients with PBC [55]. Fifty-eight patients were treated with a combination of UDCA (13–15 mg/day), OCA (5–10 mg/day), and fibrates (fenofibrate 200 mg/day or bezafibrate 400 mg/day). This combination achieved a significant reduction in ALP level compared to dual therapy (odds ratio for ALP normalization of 5.5). The primary outcome (change in ALP) and the effect on pruritus are summarized in Table 4.


**Table 4.** Outcomes of triple therapy (UDCA + fibrates + OCA) [55].

### **6. Conclusions**

In May 2021, the Food and Drug Administration issued a new warning restricting the use of OCA in patients with advanced cirrhosis (https://www.fda.gov/drugs/drug-safetyand-availability/due-risk-serious-liver-injury-fda-restricts-use-ocaliva-obeticholic-acid-pr imary-biliary-cholangitis, accessed on 1 September 2022). Advanced cirrhosis was defined on the basis of current or prior evidence of liver decompensation (e.g., encephalopathy, coagulopathy) or portal hypertension (e.g., ascites, gastroesophageal varices, or persistent thrombocytopenia). A practical guidance statement was published thereafter by the AASLD [56]. In this statement, the AASLD reported the contraindication on cirrhosis announced by the FDA, namely decompensated cirrhosis, and further recommended the careful monitoring of any patient with cirrhosis, even if not advanced, receiving OCA. In eligible patients, the recommended starting dose of OCA is 5 mg, which can be titrated to 10 mg after 6 months if OCA is well-tolerated. It is also recommended by the AASLD to monitor liver function before and after the initiation of OCA therapy.

In conclusion, due to its complex and fascinating mechanism, OCA represents a complete intervention for the therapeutic management of those PBC patients who cannot be treated satisfactorily with UDCA for efficacy or safety reasons. However, more real-world data are needed to gain a full understanding of its pharmacological and toxicological features.

**Author Contributions:** Conceptualization, A.F.; writing—original draft preparation, A.F., D.G. and S.D.M.; writing—review and editing, A.F. and S.D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

