Targeted inhibition of the FGF19-FGFR4 pathway in hepatocellular carcinoma; translational safety considerations

Authors

  • Howard R. Mellor

    Corresponding author
    1. Drug Safety & Metabolism, Innovative Medicines, AstraZeneca R&D, Macclesfield, UK
    • Correspondence

      Howard R. Mellor, Drug Safety Evaluation, Vertex Pharmaceuticals (Europe) Limited,

      86-88 Jubilee Avenue, Milton Park, Abingdon, Oxfordshire OX14 4RW, UK

      Tel: +44 (0)1235 438328

      Fax: +44 (0)1235 835880

      e-mail: howard_mellor@vrtx.com

    Search for more papers by this author

Abstract

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death and new therapies are urgently required to treat this disease. Recent data suggest that the FGF19-FGFR4 axis may be a key driver in certain forms of HCC, making the pathway an interesting, emerging molecular target for potential therapeutic intervention. A complication is that, outside of malignant disease, FGFR4 plays an important physiological role in the regulation of hepatic bile acid (BA) synthesis. FGF19 signalling via FGFR4 suppresses de novo BA production in the liver, tightly maintaining hepatic and systemic levels of these detergent-like molecules at a physiological threshold and preventing pathological complications of raised BA levels, such as cholestatic liver injury and bile acid diarrhoea. In some cases of HCC, the malignant disease causes bile duct obstruction, preventing BA secretion from the liver and resulting in cholestasis. Here, the role of FGFR4 signalling in both HCC and BA homoeostasis is discussed. The potential effects of therapeutic FGF19-FGFR4 inhibition on human hepatobiliary/gastrointestinal physiology are considered along with the potential safety implications of FGF19-FGFR4 blockade in patients with HCC.

Soluble Fibroblast Growth Factors (FGF) and their transmembrane receptors are involved in a broad range of cellular functions including proliferation, migration, differentiation and survival [1]. FGF signalling has an important role in development, for example in controlling mesoderm patterning in the embryo [2]. The mammalian FGF family comprises of 23 members, including 18 ligands that signal through four FGF receptors (FGFR1-4). Four of the FGF (FGF11, FGF12, FGF13 and FGF14) do not function as ligands but rather as homologous factors [1]. In addition, humans do not possess an FGF15 gene; human FGF19 is the orthologue of mouse FGF15 [3]. A fifth receptor (FGFR5) can bind FGF but does not have a tyrosine kinase domain and so may be a negative regulator of FGF signalling [4].

The FGFR family have an overall structure similar to other receptor tyrosine kinases, existing as single pass transmembrane proteins with a ligand-binding extracellular domain, a transmembrane domain and an intracellular tyrosine kinase domain [5] (Fig. 1). The extracellular domain is composed of three units or (Ig) domains (Ig I, II and III). The Ig II and III domains form the ligand binding pocket and have distinct regions for binding heparin sulphate proteoglycans (HSPG) and FGF, with the heparin sulphate interacting with both ligand and receptor (Fig. 1). HSPG are required to both protect FGF from degradation and to support the formation of the FGF-FGFR complex. Dimerization of FGFR is ligand-binding-dependent and is required for signalling, with FGF molecules binding to FGFR dimers in a 2:2 ratio (2 FGF molecules per dimer). FGFR molecules only form homodimers, as to date there are no reports of heterodimerization or interactions between different FGFR family members [1, 5].

Figure 1.

FGFR topology and ligand binding. FGFR possess a ligand-binding extracellular domain, a transmembrane domain and an intracellular split tyrosine kinase domain. The extracellular domain is composed of three units or (Ig) domains (Ig I, II and III). The Ig II and III domains form the ligand binding pocket and have distinct regions for binding HSPG and FGF, with the heparin sulphate interacting with both ligand and receptor. Binding of an FGF molecule to an FGFR monomer leads to receptor dimerization, which is required for signalling. The specificity of ligand binding is generated by alternative splicing of the Ig III domain.

Ligand-dependent FGFR dimerization leads to a conformational change in the receptor that activates the tyrosine kinase domain, resulting in intermolecular transphosphorylation of tyrosine residues in the kinase domain and intracellular tail. These phosphorylated tyrosine residues act as docking sites for adaptor proteins, which can also be phosphorylated by FGFR, leading to multiple signal transduction pathway activation [1, 5]. The specificity of ligand binding is generated by alternative splicing of the Ig III domain. The first part of this domain is encoded by an invariant exon IIIa; this is then alternatively spliced to either exon IIIb or IIIc, followed by splicing to the TM domain. FGFR4 does not undergo alternative splicing, always containing only the IIIb isoform [6].

FGF19 has unique specificity for FGFR4, in that the receptor mediates almost all FGF19 activities, with multiple signals at both the N- and C-terminus of FGF19 contributing to FGFR4 activation [7, 8]. In addition, cellular co-expression of a 130 kDa type 1 transmembrane protein called β-Klotho is required for FGF19 binding to FGFR4 and for the subsequent intracellular signalling, leading to modulation of downstream gene expression [9]. The requirement for β-Klotho restricts the signalling activity of FGF19 to tissues expressing both FGFR4 and β-Klotho. The human and mouse tissue expression profiles of β-Klotho and FGFR4 are similar, with the liver being the only tissue where both proteins are highly co-expressed in both species [9].

Recent data, reviewed in this article, suggest that the FGF19-FGFR4 signalling axis may be a key driver in certain forms of hepatocellular carcinoma (HCC) [10-13], raising interest in therapeutically inhibiting the pathway in this disease setting. However, it is important to first consider the potential tolerability and safety implications of inhibition of this pathway in man and particularly in HCC patients. FGF15/19 and FGFR4 play an important homeostatic physiological role in the regulation of bile acid (BA) homoeostasis. FGF15/19 signalling via FGFR4 suppresses de novo BA production in the liver, tightly maintaining hepatic and systemic levels of these detergent-like molecules at a physiological threshold [8, 14, 15]. This prevents pathological complications of raised BA levels, such as cholestatic liver injury [16] and BA diarrhoea [17]. Further, in some cases of HCC, the malignant disease causes bile duct obstruction, preventing BA secretion from the liver and resulting in cholestasis [18]. Below, the role of FGFR4 signalling in the context of both HCC and BA homoeostasis is discussed. The potential effects of therapeutic FGF19-FGFR4 inhibition, particularly on human hepatobilliary/gastrointestinal physiology, are considered along with the potential safety implications of FGF19-FGFR4 blockade in patients with HCC.

The FGF19-FGFR4 pathway in hepatocellular carcinoma

Hepatocellular carcinoma is the sixth most common form of cancer in terms of incidence, accounting for ~630,000 new cases per year globally [19]. Greater than 90% of cases of HCC arise as a consequence of underlying liver disease (e.g. viral hepatitis B or C, alcoholism), often resulting in impaired liver function [20]. Cirrhosis is the major risk factor for tumour development, present in more than 80% of patients prior to them developing liver cancer. Patients with HCC frequently present with tumours at advanced and incurable stages [19]. In advanced HCC, where patients may be ineligible for surgical resection or other forms of treatment with curative potential, treatment options are limited.

Growing evidence indicates that the FGF19-FGFR4 pathway is potentially a key driver in HCC. Multiple FGFR family members are expressed in the adult liver, however FGFR4 is the most highly expressed and is the only FGFR expressed in mature hepatocytes [21]. It has been directly demonstrated that it is through FGFR4 that FGF19 stimulates hepatocyte proliferation in vivo [7, 8]. Recombinant FGF19 inhibits apoptosis in HCC cell lines in addition to increasing proliferation and invasion capabilities, while siRNA knockdown of FGF19 or FGFR4 evokes the inverse effect on HCC cellular proliferation and apoptosis [10]. FGF19 clearly has mitogenic potential, as transgenic mice overexpressing FGF19 in skeletal muscle (FGF19-mice) have been observed to develop HCC [11]. Crucially, crossing these animals with FGFR4−/− mice inhibited the development of hepatic neoplasia. Furthermore, treatment of FGF19-mice weekly from 4 weeks of age with anti-FGF19 or anti-FGFR4 antibodies inhibited HCC formation when assessed at 6 months [12, 13].

In HCC, FGFR4 positively modulates secretion of alpha-foetoprotein [22], which is a widely used serological biomarker for the clinical disease, higher levels being associated with advanced disease progression and aggressiveness in about 70% of cases [23]. FGF19 is significantly overexpressed in HCC relative to adjacent normal human liver tissue, and HCC patients with high FGF19 expression have lower 5-year survival rates than those with low FGF19 expression [10]. FGFR4 expression is moderately, but not significantly, increased in HCC relative to normal liver in broad patient populations. However, a significant elevation in FGFR4 expression (>two-fold) is observed in a subset (~30%) of patients, indicating that FGF19-FGFR4 could be targeted in HCC using a personalized approach [13].

The role of the FGF15/19-FGFR4 pathway in the regulation of hepatic bile acid synthesis

One of the major physiological roles of FGFR4 is in the regulation of hepatic bile acid (BA) synthesis. BAs are physiological detergents, which are synthesised by the liver, stored in the gallbladder and facilitate the intestinal absorption of dietary components including vitamins, lipids and nutrients. The primary step in BA synthesis is performed by the enzyme Cyp7A1, which converts cholesterol into the BA precursor 7α-hydroxycholesterol.

The majority (~95%) of BA that are secreted into the intestine are reabsorbed efficiently by the terminal ileum and undergo enterohepatic recirculation and resecretion by the liver [24]. To ensure that systemic BA are maintained at an appropriate level, hepatocyte-expressed FGFR4 and the circulating ligand FGF15/19 form part of a negative feedback system controlling de novo hepatic BA synthesis (summarized in Fig. 2A). Intestinally absorbed BA bind to the nuclear hormone receptor FXR in the villus epithelium of the ileum leading to induction of FGF15/19 expression in enterocytes followed by FGF15/19 release into the circulation (Fig. 2A) [15]. FGF15/19 then acts as an enterohepatic hormone to activate FGFR4 in the liver, which suppresses Cyp7A1 expression resulting in reduced hepatocyte BA synthesis. Binding of BA to FXR in hepatocytes also leads to reduced BA production via induced expression of the atypical nuclear receptor Small Heterodimeric Partner, which cooperates with FGFR4 to suppress Cyp7A1 expression [15]. It is of note that in mouse, FGF15 is exclusively expressed in the small intestine [15, 25]. However, human FGF19 is expressed in the small intestine [26], gallbladder [27, 28] and hepatic expression has been observed in patients with extrahepatic cholestasis [29].

Figure 2.

Perturbation of BA homoeostasis through FGF19-FGFR4 blockade. (A) Normal bile acid control – Under normal circumstances, hepatocyte-expressed FGFR4 and the circulating ligand FGF19 form part of a negative feedback system controlling de novo hepatic BA synthesis. Intestinally absorbed BA bind to the nuclear hormone receptor FXR in the villus epithelium of the ileum leading to induction of FGF19 expression. FGF19 then acts as an enterohepatic hormone to activate FGFR4 in the liver, which suppresses Cyp7A1 expression resulting in reduced hepatocyte BA synthesis. Binding of BA to FXR in hepatocytes can also lead to reduced BA production via induced expression of FGF19. However, hepatic FGF19 production under normal physiological conditions is minimal. (B) Primary BAM syndrome – Patients with primary BAM have significantly reduced serum levels of FGF19 for unknown reasons. A reciprocal increase in the BA precursor C4 is found in serum from the patients, consistent with reduced hepatic suppression of Cyp7A1 by FGFR4. The elevated BA production and secretion into the gut exceeds ileal BA reabsorption capacity, leading to chronic BA diarrhoea. (C) FGF19/FGFR4 blockade – Genetic or therapeutic blockade of FGF19-FGFR4 results in increased Cyp7A1 expression leading to elevated hepatic BA production, secretion of BA into the ileum and increased enterohepatic BA recirculation. FGF19 blockade in primates is poorly tolerated, leading to diarrhoea and liver toxicity. (D) FGF19/FGFR4 blockade in cholestatic HCC – In patients with cholestatic HCC or HCC with partial bile duct obstruction treated with an FGF19-FGFR4 inhibitor, a potential scenario of increased BA synthesis together with blockade or interruption of hepatic BA efflux (as a consequence of bile duct blockage) exists and therefore there would be an increased risk of exacerbation of cholestatic liver injury.

Strong evidence for the role of FGFR4 in BA regulation comes from the phenotypic evaluation of FGFR4−/− mice. These animals have a small (~70% of wt), depleted gallbladder and a two- to three-fold increase in secreted BA relative to wt animals [14]. In addition, in FGFR4−/− mice, liver HMG-CoA reductase, the rate-limiting enzyme in cholesterol biosynthesis, is increased 7-fold, and Cyp7A1, which converts cholesterol into the BA precursor 7α-hydroxycholesterol, is increased 2.5-fold [14]. Therefore, it is clear that the negative control FGFR4 exerts on cholesterol metabolism to BA is abrogated by disruption of the FGFR4 gene. FGFR4−/− mice also express significantly higher levels of Cyp8B1, obligatory for cholic acid (CA)-derived but not chenodeoxycholic acid (CDCA)-derived BA synthesis. Correspondingly, basal levels of CA and its metabolites relative to CDCA are increased in the FGFR4−/− mice [8]. The absolute requirement for FGFR4, and not other FGFR members, in FGF19-mediated BA regulation has been further demonstrated in vivo using an engineered FGFR4-specific FGF19 variant, which can only activate FGFR4 but still downregulates Cyp7A1 in the liver [7].

Manipulation of FGF15/19 levels also results in severe disruption to BA homoeostasis. Mice deficient in FGF15 exhibit increased faecal BA secretion [15] and conversely, forced hepatic overexpression of FGF15 has the opposite effect [30]. Infusion of wt mice with human recombinant FGF19 leads to a significant reduction in CA-derived BA species [8]. FGF15/19 also acts on the gallbladder and stimulates filling with bile [31]. The negative effect of FGFR4 signalling on Cyp7A1 and BA synthesis appears to be entirely dependent on stimulation by FGF15/19. Although FGF21 can bind to FGFR4 (in the presence of β-Klotho), unlike FGF15/19, FGF21 cannot suppress Cyp7A1 expression in hepatocytes [32]. Further, β-Klotho−/− mice exhibit increased BA synthesis and secretion linked to elevated Cyp7A1 and Cyp7B1 levels and a failure to supress Cyp7A1 in response to dietry BA [33]. This finding is consistent with the loss of hepatic FGF15/19-FGFR4 signalling in the absence β-Klotho expression [9]. It is therefore clear that the FGF15/19-FGFR4 axis has an essential role in the post-prandial negative feedback regulation of BA homoeostasis.

A key finding in patients with extrahepatic cholestasis, i.e. that caused by blockade of bile flow because of an obstruction outside the liver, is marked elevations in hepatic FGF19 mRNA, increased plasma FGF19 levels and greatly reduced hepatic Cyp7A1 levels [29]. This adaptive response is entirely consistent with our understanding of the pathway and is clearly directed towards suppressing hepatic de novo BA production under circumstances of hepatic BA accumulation. The contribution of enterocyte-derived FGF19 to the elevated circulating FGF19 pool is considered to be minimal in these patients because of the blockade in BA enterohepatic recirculation. FGF19 elevation in these individuals is also linked to a range of other adaptations, including expression of hepatobiliary transporters, aimed at minimizing the impact of BA toxicity on the liver. Patients receiving a biliary stent to restore bile flow and BA levels have reduced hepatic FGF19 mRNA and circulating FGF19 levels comparable to control patients [29].

Metabolic roles of the FGF15/19-FGFR4 pathway

In addition to maintaining BA homoeostasis, FGF19 also has various metabolic effects. The transgenic FGF19-mice described earlier have a specific reduction in white adipose tissue, as a result of increased metabolic rate, and fail to become obese or diabetic on a high-fat diet [34]. These animals also have decreased liver and muscle triglyceride levels, lower serum glucose and insulin, and improved glucose tolerance and insulin sensitivity compared to wt mice [34]. A similar effect profile has been observed after intravenous administration of recombinant human FGF19 to high-fat-fed mice [35]. FGF19-treated mice lose body weight because of increased fatty acid (FA) oxidation in addition to having decreased serum insulin, leptin, cholesterol and triglycerides and improved glucose tolerance [35].

Interestingly, the hepatic expression levels of various genes regulating lipid metabolism are altered in FGF19 transgenic or FGF19-treated mice. This includes repression of steroyl CoA desaturase 1 (SCD1), involved in monounsaturated FA biosynthesis, and acetyl CoA carboxylase-2 (ACC2), a negative regulator of FA transport into the mitochondria in preparation for β-oxidation [34, 35]. Like FGF19-mice, ACC2−/− mice have lower body weight and are resistant to the adverse effects of a high-fat, high-carbohydrate diet [36]. Similarly, liver-specific loss of SCD1 expression protects mice from high-carbohydrate diet induced adiposity and hepatic steatosis [37]. Furthermore, FGFR4−/− mice fed a regular diet develop metabolic syndrome, displaying hyperlipidemia, glucose intolerance, insulin resistance and weight gain [38], adding support to the protective metabolic effects of FGF15/19-FGFR4 signalling. However, the contribution of hepatic FGFR4 to the overall metabolic effects of FGF19 remains unclear, as the direct action of FGF19 in adipose tissue, where FGFR4 expression is very low, also seems to play a role [39].

What could be the pathophysiological consequences of FGF19-FGFR4 blockade in humans?

Mice lacking FGFR4 are born at the expected frequency and, with the exception of the gallbladder changes described above, show no gross abnormalities or pathological changes in tissues normally expressing FGFR4, when assessed up to 1 year of age [14, 40]. In contrast, most FGF15−/− mice die during late embryonic development or shortly after birth, but the few that survive appear normal [15]. However, a complication with translating the phenotypic observations in knockout (KO) rodent models to effects in humans is the possibility of functional redundancy between protein homologues in KO animals. In addition, developmental compensation, which could mask the true extent of a pathway blockade in a developed animal, may also be an issue. A further important consideration is that certain metabolic aspects differ between mice and humans. For example, diurnal BA synthesis in humans has two major peaks during the daytime, opposite from the circadian rhythm of cholesterol synthesis, but in rodents both cholesterol and BA synthesis peak at midnight [41]. A further example is that LXRα-mediated feed-forward induction of Cyp7A1 by dietary cholesterol occurs in mice but this mechanism does not exist in humans [42, 43]. It is also important to consider that mouse FGF15 and human FGF19 actually share only 51% sequence identity [3], so despite these orthologues exhibiting significant functional overlap, operational divergence may also exist.

While the data described above suggest the possibility of FGF19-FGFR4 blockade evoking a form of metabolic syndrome in humans, strong supportive data is lacking. However, with respect to potential disruption of BA homoeostasis, a key finding linking dysregulation of the FGF19-FGFR4 axis to human disease was made by Walters et al. [17]. Primary (idiopathic) BA malabsorption (BAM) is a chronic diarrhoeal syndrome resulting from excess colonic BA, yet until recently the pathogenesis of this disease was unclear. Primary BAM is unusual, as the majority of patients have no ileal resection or ileal disease and no apparent defect in ileal BA transport. Walters et al. showed that patients with primary BAM have significantly reduced serum levels of FGF19 compared to healthy controls. A reciprocal increase in the BA precursor 7α-hydroxy-4-cholesten-3-one (C4) was also detected in serum from the patients, consistent with reduced hepatic FGFR4-mediated suppression of Cyp7A1 (Fig. 2B). The conclusion is that primary BAM is caused by impaired FGF19 feedback inhibition of hepatic BA synthesis, which leads to elevated BA production and secretion, exceeding the capacity for ileal reabsorption, resulting in chronic BA diarrhoea [17]. The specific nature of the defect in the enterocyte that leads to impaired FGF19 secretion is unclear at present. There are as yet no reports describing any liver tissue pathology in patients with this form of primary BAM. However, it is reasonable to assume that persistent therapeutic blockade of the FGF19-FGFR4 pathway will almost certainly result in primary BAM-like effects in humans, including chronic diarrhoea.

There are close similarities between non-human primates and man regarding BA and biliary biology, hepatic physiology and the diseases associated with these systems [44]. In a recent study, Pai et al. described the preclinical toxicological evaluation of an anti-FGF19 therapeutic antibody in cyanomologus monkeys, which raised significant safety concerns regarding potential therapeutic blockade of this pathway in man [45]. A single dose administration of anti-FGF19 to the animals caused severe diarrhoea, low food consumption and decreased body weight. These toxicities ultimately resulted in the unscheduled euthanasia of all animals in the 10, 30 and 100 mg/kg dose groups in response to poor clinical condition, with only the lowest dose group (3 mg/kg) receiving the scheduled 5 doses. Substantial dose-related serological increases in aspartate aminotransferase, alanine transaminase, total BA and total bilirubin were observed across the 10, 30 and 100 mg/kg dose groups. Only modest elevations in these biomarkers were measured in the 3 mg/kg dose group [45]. Although histopathological changes were not seen in the livers of the majority of animals, two had mild to marked hepatocellular single-cell necrosis [45]. Interestingly, based on the clinical severity, anti-FGF19 appeared to be more toxic to females than males, the reason for which is currently unknown. A substantial increase in absolute faecal BA concentration was observed in the high-dose anti-FGF19 treated group relative to the control group. Anti-FGF19 had no apparent direct cytotoxic or deleterious effect on cultured primary monkey hepatocytes. However, anti-FGF19-treatment increased Cyp7A1 and BA efflux transporter expression in both primary monkey hepatocytes and in the livers of treated animals. The authors concluded that high doses of anti-FGF19 increase Cyp7A1-mediated BA synthesis leading to diarrhoea and increased enterohepatic BA recirculation, resulting in liver toxicity (Fig. 2C) [45]. It is possible that this liver toxicity, resulting from the increase in de novo and recirculated hepatic BA, represents a form of cholestatic-type liver injury, similar to that caused by small-molecule inhibitors of hepatic BA transporters [46].

What could be the consequences of FGF19-FGFR4 blockade in patients with HCC?

It is important to note that endogenous BA are carcinogens [47] and that dysregulation of BA homoeostasis is known to contribute to the development of HCC. Children with a deficiency in the hepatic bile salt export pump (BSEP, ABC11) develop severe cholestasis and HCC in early life [16] and certain BSEP polymorphisms are associated with an enhanced HCC susceptibility [48]. Furthermore, HCC patients have significantly higher fasting total serum BA than patients with cirrhosis, hepatitis or healthy controls [49]. Conjugated BA species (e.g. glycocholic acid, glycochenodeoxycholic acid and taurocholic acid) are higher in HCC patients compared to healthy controls, in contrast to unconjugated BA species, which are lower [50, 51]. Interestingly, serum conjugated BA levels vary depending on the extent of HCC progression, with the highest levels being associated with stage I disease. This initial BA peak may relate to acute hepato-functional dysregulation in early disease [50]. Increased levels of 3-oxo-∆4 BA species have also been consistently observed in HCC, which is interesting as these are generally considered to be fetal BA not usually detected in the plasma of adults [52, 53]. Although the mechanism(s) underlying BA dysregulation in HCC are unclear, one contributing factor may be reduced BA transporter expression in HCC tissues [54]. This of course returns us to the question of how FGFR19-FGFR4 blockade would impact on the pathophysiology of the mixed normal/malignant liver in the HCC setting.

It is clearly difficult to predict the impact of dysregulated BA production on the diseased livers of HCC patients, particularly in the context of the potential anti-tumour benefits of FGF19-FGFR4 blockade. However, an HCC setting where the risk appears significant is where the malignant disease involves or obstructs the bile duct. This type of liver cancer has also been referred to as ‘icteric’ or ‘cholestatic’ HCC [18] and has been reported to account for up to 12% of HCC cases [55]. The initial presentation of cholestatic HCC is usually clinical jaundice (obstructive jaundice secondary to HCC) [56], but it can be clinically challenging to diagnose pre-operatively. Firstly, the radiological characteristics of cholestatic HCC are not well known [57] and secondly, jaundice frequently occurs in advanced HCC cases anyway, in the absence of bile duct obstruction, as a consequence of liver failure [58]. Cholestatic HCC may involve the bile duct in four main ways: tumour compression, tumour thrombosis, haemobilia and diffuse tumour infiltration [55]. It has been reported that the incidence of bile duct thrombi is increasing in HCC and this could be linked to the use trancatheter arterial chemoembolization therapy [59]. The prognosis for individuals with obstructive jaundice secondary to HCC is poor, but not worse than for HCC patients presenting without jaundice, and still significantly better than when the presenting jaundice results from hepatic failure [56, 57].

As described above, anti-FGF19 antibody-mediated BA elevations have caused single-cell hepatocyte necrosis in primates in the absence of pre-existing liver disease [45]. Although there is clear preclinical evidence in support of the beneficial effects of FGF19-FGFR4 blockade in HCC, one could also envisage a potential exacerbation of liver complications if HCC patients were treated with an inhibitor of this pathway. Further, in patients with cholestatic HCC or HCC with partial bile duct obstruction, there would be a scenario of increased BA synthesis together with blockade or interruption of hepatic BA efflux (as a consequence of bile duct blockage) and therefore an increased risk of exacerbation of cholestatic liver injury (Fig. 2D). This hypothetical risk increases the importance of identifying the subset of HCC patients with underlying cholestatic HCC and potentially excluding them from this therapeutic approach. Alternatively, non-surgical [55] or surgical [58] biliary drainage strategies could be employed, where appropriate, in an attempt to reverse the bile duct blockage prior to systemic therapy.

Summary and future perspectives

The data discussed here raise the possibility that blockade of FGF19-FGFR4 pathway may not be tolerable in man. The evidence derived from individuals with primary BAM suggest that, albeit debilitating, in an otherwise healthy individual, a degree of disregulation of the FGF19-FGFR4 pathway can be manageable. Patients affected by primary BAM may be treated with BA sequestrants to reduce colonic BA levels, thereby relieving symptoms of diarrhoea. However, from a targeted therapy perspective, a potential complication is that BA sequestrant comedications may adversely impact the gastrointestinal absorption and systemic exposure of orally administered, FGF19-FGFR4 targeting, small-molecule lipophilic compounds. Despite this, the systemic exposure of compounds or biologics administered via the intravenous route should be unimpacted by oral BA sequestrants. It is also possible that statins could be employed to reduce hepatic cholesterol synthesis, thereby reducing the substrate availability for subsequent hepatic BA production via Cyp7A1 [60] and possibly circumventing the adverse pathology observed in primates upon FGF19 treatment [45].

Despite the potential tolerability of primary BAM-like phenotype, it is presently unclear firstly, to what extent the FGF19-FGFR4 axis is disrupted in this disease, secondly, the degree of further blockade required to derive therapeutic benefit in HCC and thirdly, the adverse hepatic/GI toxicity this level of inhibition may cause in humans. The lack of an overt pathological phenotype in mature FGFR4−/− [40] and FGF15−/− mice [15] has perhaps led to the conclusion that inhibition of FGF19-FGFR4 may also have minimal adverse effect on human liver physiology [22]. However, the data from primates treated with anti-FGF19 indicate substantial target-mediated and dose-dependent tolerability issues, apparently intrinsically related to the degree of pathway inhibition. Further, this study demonstrated that, not only does prolonged inhibition of FGF19-FGFR4 lead to inappetence, diarrhoea and body weight decline, it is also associated with liver toxicity [45]. Therefore, it seems unlikely to be possible to inhibit the FGF19-FGFR4 axis in the liver of HCC patients sufficiently to positively impact the malignancy without also triggering the dose-limiting toxicities observed in primates treated with anti-FGF19.

Advanced cancers are life-threatening diseases where the risk-benefit needs to be considered not only on a disease basis, but on an individual case-by-case basis. Perhaps a possible way forward for FGF19-FGFR4 inhibiting therapies in HCC is to exclude from treatment any patients at risk of therapy-associated complications. Functional magnetic resonance imaging (MRI) approaches have been applied to confirm bile duct obstruction/thrombi in HCC patients and it has been suggested that surgical approaches could then be used to remove these blockages in operable HCC cases [58].

Conceptually, it would be interesting to explore preclinical orthotopic FGF19-FGFR4-dependent HCC animal models, combined with therapeutic blockers of FGF19/FGFR4 signalling. It is clearly crucial that the approach blocks FGF19-FGFR4 signalling in both the tumour and normal host tissues, so that the interplay, pathophysiology and therapeutic index can all be assessed. This would permit examination of the impact of FGFR4 blockade on livers with pre-existing malignant disease and perhaps provide a disease-relevant assessment of the potential safety impact on patients with HCC. These approaches would also allow the assessment of dosing schedules aimed at developing/maximizing a therapeutic window. An anti-FGFR4 antibody has been recently tested in a mouse model of HCC where the antibody cross-reacted with mouse and led to hepatic Cyp7A1 induction, with no report of deleterious effects [13]. However, it is clear from the data reviewed herein, that mouse lacks biological relevance to man with respect to this toxicological mechanism. Therefore, defining the translational relevance of BA-biology-related observations made in different preclinical species to the clinical (oncology) setting and gaining a greater understanding of the subtle differences in BA regulation by FGF19-FGFR4 between species would need to be a focus of future work in this area.

In conclusion, the strategy of inhibiting the FGF19-FGFR4 pathway in HCC carries intrinsic safety risks together with significant translational challenges. Therefore, experimentally ‘de-risking’ this approach, ahead of moving into the clinical drug development phase, remains immensely challenging. While potential ways forward exist, it may be pertinent to pursue other therapeutic targets in HCC that offer an improved weighting of therapeutic benefit vs. risk.

Ancillary