Targeting FGF19-FGFR4 pathway for cholestatic, metabolic and cancerous diseases

Abstract

Human FGF19 (FGF15 in rodents) plays a central role in controlling bile acid (BA) synthesis through a negative feedback mechanism. This process involves a postprandial crosstalk between the BA-activated ileal farnesoid X receptor (FXR) and the hepatic betaKlotho (KLB) coreceptor complexed with FGFR4 kinase. Additionally, FGF19 regulates glucose, lipid, and energy metabolism by coordinating responses from functional KLB and FGFR1-3 receptor complexes on the periphery. Pharmacologically, native FGF19 or its analogs decrease elevated BA levels, fat content, and collateral tissue damage. This makes them effective in treating both cholestatic diseases like primary biliary or sclerosing cholangitis (PBC or PSC) and metabolic abnormalities such as nonalcoholic steatohepatitis (NASH). However, chronic administration of FGF19 drives oncogenesis in mice by activating the FGFR4-dependent mitogenic or hepatic regenerative pathway that could be a concern in humans. Agents that block FGF19 or FGFR4 signaling have shown great potency in preventing FGF19-responsive HCC development in animal models. Recent phase 1/2 clinical trials have demonstrated promising results for several FGF19-based agents in selectively treating patients with PBC, PSC, NASH, or HCC. This review aims to provide an update on the clinical development of both analogs and antagonists targeting the FGF19-FGFR4 signaling pathway for patients with cholestatic, metabolic, and cancer diseases. We will also analyze potential safety and mechanistic concerns that should guide future research and advanced trials.

Introduction to the pathophysiology of FGF19-FGFR4-KLB signaling

Human fibroblast growth factor 19 (FGF19, or FGF15 in rodents), along with FGF21 and FGF23, constitutes the endocrine FGF (eFGF) subfamily [1-5]. These eFGFs are characterized by their ability to support physiological interorgan crosstalk to regulate homeostasis of bile acids (BAs), glucose, lipids, energy, and phosphate/Ca²⁺ metabolism, thereby maintaining metabolic fluctuations within healthy levels. Although sharing structural homology and a general three-dimensional folding structure with canonic FGFs, FGF19, 21, and 23 diverge from them at the C-termini. Such divergence allows eFGFs to bind the transmembrane co-receptors Klotho (KL or Klotho alpha) or Klotho beta (KLB) that are complexed with FGFRs [6-8]. KL/KLB also provides target-tissue specificity for eFGFs, as both KL and KLB are predominantly expressed in selected endocrine and metabolic tissues, compared to the wide distribution of FGFR isotypes in the body. Furthermore, eFGFs can act in both paracrine/autocrine and endocrine modes due to their low affinity for heparan sulfate proteoglycan moieties, allowing them to escape extracellular matrix. Unlike canonic FGFs, eFGFs do not appear to have overt activity for promoting cell proliferation in vivo. Lastly, eFGF analogs, receptor agonists, or antagonistic agents have shown robust pharmacologic effects in rodents, non-human primates and man on treating various metabolic diseases such as cholestasis, NASH, hyperlipidemia, and X-linked hypophosphatemia, as well as obesity, diabetes, and insulin resistance [5, 9-14].

Physiologically, FGF19/15 is a crucial postprandial enterokine, a pivot to the gut-liver-gallbladder axis, ensuring daily maintenance of BA synthesis and enterohepatic flux homeostasis [1] (Fig. 1). Following a meal, BAs in the ileum’s chyme activate the nuclear FXR in enterocytes, inducing a spike in the expression and secretion of enteric FGF19/15 during the early fed state. Through enterohepatic circulation, FGF19/15 then binds to and activates hepatic FGFR4 in complex with KLB during the late fed-state [15, 16]. This is believed to trigger an ERK1/2-associated intracellular signal pathway, leading to feedback inhibition of cholesterol 7a-hydroxylase (CYP7A1) expression, the rate-limiting enzyme for de novo hepatic BA synthesis [1, 17]. This enteric BAs-FXR to hepatic FGFR4-KLB crosstalk pathway mediated by FGF15 and FGF19 completes an essential endocrine circuitry critical for governing BA homeostasis (Fig. 1). In the liver and intestine, FGF19 may also regulate other postprandial responses, including glycogen and protein synthesis independent of insulin, and lipid absorption [18, 19]. Furthermore, FGF19 acts on other peripheral organs such as adipose tissue and potentially the CNS to regulate glucose, lipid, and energy homeostasis via the FGFR1-3 and KLB complexes (Fig. 1) [19-23]. It is worth noting on the differences in species specificity, receptor selectivity and functions among human FGF19, mouse FGF15, and FGF21 [24]. Studies have found that both FGF19 and FGF15 exert potent BA-lowering, cholesterol-rising activity in hepatocytes while FGF21 does not. On the other hand, FGF15 lacks glycemic effects typical of FGF19 and FGF21 in adipocytes, owing to its inability to activate the FGFR1-KLB complex. Further, FGF15 binds to only mouse KLB while FGF19 and FGF21 recognize KLB of both species. FGF21 is much less potent in binding to FGFR4-KLB than either FGF19 or FGF15, and thus, lacks tumorigenic activity typical of FGF19, and likely, FGF15 as well [25]. These differences contribute to the differences in their physiology, pharmacology, and clinical potential as we will discuss hereafter.

Fig. 1. Experimentally observed physiological roles and therapeutic effects of the FGF19-FGFR4-KLB signal pathway.

In normal physiology, the daily efflux of bile acids from the liver and gallbladder in response to food ingestion to the intestine allows the activation of ileal epithelial FXR, which promotes the expression and secretion of FGF19 as an enterokine. The major role of the secreted FGF19 is thought to transcriptionally suppress hepatic CYP7A1, encoding for the key rate-limiting enzyme for bile acid synthesis, by activating the hepatocyte-residing FGFR4-KLB complex and pERK1/2 or JNK mediated pathway, leading to negative control of bile acid content while inducing gallbladder relaxation and refilling. FGF19 is also found to inhibit lipogenesis and gluconeogenesis while promoting protein and glycogen synthesis in the liver. In the intestine, FGF19 may help to reduce the intake of lipids or liposoluble vitamins. As FGF21, a homolog in the same endocrine FGF subfamily, FGF19 has been reported to act on the white and brown adipose tissues and central nervous system to affect lipolysis, fatty acid oxidation and thermogenesis. Pharmacologically, FGF19 inhibits the development of fatty liver, NASH, obesity, diabetes, and cholestasis in experimental mouse models. CL, cholesterol; HPA, hypothalamus-pituitary-adrenal axis.

BAs are amphipathic steroid molecules primarily destined to facilitate the digestion and absorption of dietary lipids, steroids, and lipophilic vitamins while also lowering cholesterol [26]. However, if BAs accumulate to over-physiological concentrations in the liver, biliary ducts, intestine, blood and other tissues, they can become highly toxic. This can lead to cholestasis, inflammation, cell proliferation, necrosis, and fibrosis - all risk factors for spontaneous hepatocellular carcinoma (HCC) development [27, 28]. Thus, tight regulation of BA synthesis and flux is essential for cholesterol, metabolic, and cellular homeostasis [29]. In this context, the FGF19-mediated enteric FXR to hepatic FGFR4-KLB axis is paramount to BA-associated metabolic homeostasis and organismal heath (Fig. 2) [30]. Disruptions in BA sensing, biosynthesis, transport, reabsorption, and signaling in association with FXR, FGF19, FGFR4, and KLB can alter BA homeostasis. This contributes to an array of cholestatic and metabolic diseases, from extra- and intra-hepatic cholestasis, fibrosis and cirrhosis to HCC in the hepatobiliary system as mentioned above, and from BA malabsorption-associated diarrhea, irritable bowel syndrome (IBS), and inflammatory bowel disease (IBD) to colorectal cancer (CRC) in the intestine [31]. In addition, evidence has linked FGF19 signaling system to other common metabolic diseases, notably, the non-alcoholic fatty liver disease (NAFLD), nonalcoholic steatohepatitis (NASH), and type 2 diabetes (T2D) [32]. The FGF19 analogs Aldafermin (M70 or NGM282) and M52 are currently in clinical trials or in preclinical development for treating some of these diseases (Table 1) [33, 34].

Fig. 2. Potential therapeutic strategies for the pathological roles of the FGF19-FGFR4-KLB signal pathway.

In a normal physiological context, the postprandial bile acids (BA) in the food digestion in the ileum activates nuclear receptor FXR, leading to a spike in the transcription of FGF19/15 in the enterocytes, which is secreted to circulation as an endocrine signal. FGF19/15 then activates the hepatic FGFR4-KLB complex, leading to a transcriptional repression of CYP7A1, the rate-limiting step in bile acid synthesis. Therefore, the endocrine FGF19 to hepatic FGFR4-KLB pathway is critical for bile acid homeostasis and FGF19 analogues or agonists can be utilized to treat bile acids-associated cholestatic diseases as well as metabolic diseases. However, FGF19 can also activate FGFR4 independent of KLB albeit with a much lower affinity, which is thought to play a role in liver regeneration by promoting cell proliferation, but chronically, the development of hepatocellular carcinoma (HCC). Whether the KLB-dependent pathway contributes to hepatocellular homeostasis such as cell proliferation and growth and liver regeneration is unclear. Nevertheless, small molecular inhibitors or neutralizing antibodies of FGFR4 have been developed for treating HCC. Because FGFR4 as the obligatory transducer of FGF19 signal lies in the center of both therapeutic approaches, it raises concerns about the possibility that one approach would compromise the other as a potential side-effect, although a so-called mitogenically defective variant of FGF19 has been devised and tested for cholestasis patients, while BA sequestrants have been used to reduce the increased BA content in HCC patients treated with FGFR4 inhibitors.

Table 1.

Clinical trials of FGF19 analogues and FGFR4 antagonists in human patients with different diseases

FGF19 analogues	Aldafermin
Alias	M70 Aldafermin, NGM282
Molecular feature	FGF19 variant: Pro24-Ser28 deletion, Ala30Ser, Gly31Ser, His33Leu IC50: 6.8 pM, 1.7 pM (native FGF19)
Receptor target	FGFR4-KLB
Company	NGM Biopharmaceuticals
Disease indications	Cholestatic disease PBC	Cholestatic disease PSC	NASH	T2D	Constipation, BA associated-Diarrhea
Dose, duration	0.3 mg & 3 mg 4 weeks	1 mg & 3 mg 12 weeks	3 mg & 6 mg 12 weeks	2, 5 & 10 mg 4 weeks	1 mg & 6 mg 2 weeks
Clinical trial registration	NCT01776528 NCT02026401 NCT02135536	NCT02704364	NCT02443116 NCT03912532 NCT04210245	NCT01943045	NCT02649062 NCT05130047
7a-hydroxy-4-cholesten-3-one (C4)	Significant decrease
Fibrosis Pro-C3, Liver injury markers - ALT & AST	Significant decrease
Total cholesterol	Significant increase, managed with Statin co-administration

FGFR4 inhibitors
Small molecular inhibitors	Fisogatinib	H3B-6527	Roblitinib
Alias	BLU9931/BLU-554		FGF401
Targeting strategy	Covalent modification of cysteine-552 at the ATP-binding site of FGFR4
Company	Blueprint Medicines	H3 Biomedicine	Novartis
Disease indications	HCC patients positive for FGF19-FGFR4-KLB pathway
Clinical trial registration	NCT02508467	NCT02834780	NCT02325739
Overall Response rate	17%	16.7%	N/A
Bile acids	Significant increase, managed with bile acid sequestrants

Note: PBC, primary biliary cholangitis. PSC, primary sclerosing cholangitis. BA, bile acid. HCC, hepatocellular carcinoma

In addition to controlling BA, sugar and lipid homeostasis, chronically elevated levels of human FGF19 may also stimulate hepatocyte proliferation and contribute to the development of HCC (Fig. 2), intrahepatic cholangiocarcinoma (ICC), and gallbladder carcinoma (GBC) [35-37]. An early study demonstrated that the ectopic expression of FGF19 could promote the development of HCC in mice [35]. Interestingly, some HCC patients have been found to express of FGF19 ectopically in hepatocytes, which is not the typical site of its production. Approximately 10% of human HCC tumors show co-amplification of FGF19 and cyclin D1 on 11q13.3 chromosome [38]. Given these findings, antagonizing the FGF19-FGFR4 axis has been proposed as a precision-target approach to combat HCC. Several agents are currently in clinical trials or under preclinical development (Table 1). However, this strategy may unavoidably raise potential safety concerns regarding the dysregulation of the BA/FGF19/FGFR4 pathway and the development of BA-associated non-tumorous diseases. A key question is whether it is possible to antagonize the same associated signaling pathway for treating one disease while promoting it for another sickness without causing a potential side-effect for each other (Fig. 2). Moreover, understanding the oncogenic role of the FGF19-FGFR4 axis is complicated by findings suggesting that bile acids and associated FXR, SHP and YAP pathways may also contribute to the development of HCC and ICC. These complicating factors need to be carefully considered [27, 39-41].

In this review, we aim to summarize and discuss the outcomes of clinical trials involving novel therapies that target the FGF19-FGFR4 pathway. These therapies function as either agonists or antagonists in the context of both BA homeostasis and cancer. We will also delve into the potential mechanisms that underlie the therapeutic outcomes of drug candidates for both strategies, with a particular focus on hepatobiliary and intestinal pathophysiology. An analysis of potential safety concerns will be also included.

We will not be detailing the structure and function of the FGF and eFGF family in this review, as this information is readily available in several other excellent reviews [42-44]. Therefore, due to the intended scope of this review, some mechanistic studies in the field of FGF19/15 will not be discussed.

Clinical trials of FGF19 analogs in cholestatic and metabolic diseases

1. FGF19 analogs for treating cholestatic disease

Cholestatic disease, a condition marked by hindered bile flow and BA retention in the hepatobiliary and intestinal system, is a severe health issue [45]. The excessive accumulation of BAs can stimulate the growth of hepatic and biliary duct cells, inflammation, fibrogenesis, and necrotic damage, which predispose individuals to HCC development, often for which the liver transplant is the only medicinal option considered. The obstruction of bile flow can be due to extrahepatic cholestasis caused by tumors, cysts, or stones in the common bile duct or intrahepatic cholestasis resulting from conditions such as biliary cirrhosis, biliary atresia, cholangitis, viral hepatitis, alcoholic hepatitis, NASH, sepsis, drug toxicity, or even in pregnancy. The rise of BAs to high, intractable levels in the hepatic and bile duct system can be of other causes such as specific mutations in the regulatory genes involved in the gut-liver signaling axis that control de novo synthesis, modification, efflux, transport, and reabsorption of BAs [45]. For example, inheritable mutations in ATP8b1 (a canalicular phosphatidylserine flippase), ABCB11 (bile salt export pump), and ABCB4 (phosphatidylcholine floppase multidrug-resistant protein MDR2/3) cause Progressive Familial Intrahepatic Cholestasis (PFIC) types 1, 2 and 3, respectively [46-48]. To date, ursodeoxycholic acid (UDCA), a tertiary, hydrophilic BA produced in the liver, was the first approved drug for treating a selected group of cholestatic Primary Biliary Cholangitis (PBC) patients. It is a choleretic agent and works by increasing the hydrophilicity of the BA pool, stimulating bile flow, and competing with prevalent BAs for intestinal transporters and for FXR activation, among others [49, 50]. In 2016, Obeticholic Acid (OCA), a semi-synthetic BA analog and a potent, selective FXR agonist, was approved for PBC patients with advanced cirrhosis who have not responded adequately to UDCA [51]. However, these anti-cholelithic or choleretic treatments themselves can cause serious liver injury leading to liver decompensation or failure. Therefore, for patients with late-stage hepatocellular cholestasis, liver transplantation remains the only available therapy.

The accumulation of BAs in the hepatobiliary system is a central event to the pathogenesis and pathognomonic manifestation of cholestasis, leading to hepatobiliary injury, inflammation, fibrosis, cirrhosis, cancer, and liver failure. Therefore, reducing BAs levels could be a crucial strategy to prevent liver damage or slow disease progression. The endocrine FGF19-mediated gut-liver signaling axis, which involves enteric FXR engagement and hepatic FGFR4-KLB machinery, is vital as ‘cholestat’ for maintaining a normal negative feedback control mechanism on BA content. These pathways are key to tightly regulating BA homeostasis and preventing BA overshoot, as demonstrated in all Fgf15, Fgfr4, Klb or Fxr deficient mouse models [1, 17, 52, 53]. Furthermore, circulating FGF19 levels are increased in patients with PBC and Primary Sclerosing Cholangitis (PSC), along with hepatic BA levels. This suggests that the BA-FGF19 interplay may represent an adaptive mechanism in cholestatic and cirrhotic conditions [54-56]. Along the same line, serum and hepatic levels of FGF19 correlate with worsened liver biochemistry, BAs, and quality of life. Serum FGF19 is elevated in UDCA non-responders and its concentrations correlates with the severity of liver disease, potentially serving as a predictive biomarker of chronic cholestatic liver injury [57]. Since OCA treatment increases ileal FGF19 for more than 70-fold [58], utilizing the FGF19-FGFR4 agonist to reduce intrahepatic and extrahepatic BAs levels may therefore represent an ideal strategy for treating cholestatic liver disease.

The initial attempt to develop FGF19-based therapy involved the creation of a new variant of FGF19, known as Aldafermin. This variant has a deletion in the N-terminal portion of human FGF19 at Pro24-Ser28 and mutations of Ala30Ser, Gly31Ser, and His33Leu [59]. In the bile duct ligation (BDL) mice, Aldafermin showed a similar potency to native FGF19 in reducing total BA pool size by 46% and 51%, but a lower IC₅₀ (half-maximal inhibitory concentration) value than native FGF19 that was 6.8 pM and 1.7 pM, respectively. Interestingly, Aldafermin and another similar FGF19 analog, M52, were suggested not to possess tumorigenicity of native FGF19, while the reason for that remains unknown. However, this notion was recently challenged in https://www.biorxiv.org/content/10.1101/2023.09.15.557921v1. Nevertheless, Aldafermin and M52 exhibited significant effectiveness in protecting the liver from inflammation, fibrosis, and necrosis-associated injury in several mouse models of metabolic and cholestatic diseases, including db/db, ob/ob, HFFCD, Mdr2/3^−/−, Fxr^−/−, and rasH2 mice, as well as obstructive BDL and alpha-naphthylisothiocyanate treatment mouse models [33, 59-62]. These promising preclinical data on Aldafermin paved the way for the subsequent clinical trials (Table 1).

In the first clinical trial (NCT01776528), Aldafermin was administered daily to healthy volunteers for seven and fourteen consecutive days. This resulted in a rapid and potent reduction of serum 7a-hydroxy-4-cholesten-3-one (C4), a surrogate marker for CYP7A1 activity, in both fasted and fed states in all subjects [59]. Aldafermin was generally well tolerated with no serious adverse effects. In February 2014, a phase 2 intervention trial (NCT02026401) was conducted with 45 PBC patients who had an inadequate response to UDCA. The patients were given 0.3, 3 mg of Aldafermin and placebo for 28 days [63]. This study was followed by a phase 2b with extended Aldafermin treatment in PBC patients who have completed the prior clinical experiment (NCT02135536). Aldafermin led to a dose-dependent decrease in serum C4, indicating an excellent target engagement. At day 28, significant improvements in serum levels of ALP, ALT, AST, GGT, IgG, and IgM, which are markers of cholestasis, hepatocellular injury, immunity, and inflammation, were observed. PBC patients at baseline had elevated total cholesterol (TC), LDL-C, HDL-C, and hyperlipidemia. Aldafermin treatment at 3 mg dose induced a dose-dependent decrease in TC and LDL-C but not HDL-C, TGs, bilirubin, albumin, and body weight. Unfortunately, liver biopsies were not conducted in this study so the measures of improvements in fibrosis and cirrhosis were not assessed.

Another phase 2 trial (NCT02704364) was conducted in March 2016 with Aldafermin (1 or 3 mg) in 58 individuals with PSC [13] for 12 weeks. Similar to the PBC patients, significant reductions in serum C4 and BA, particularly the secondary BA levels, were observed with Aldafermin use. Significant decreases in liver enzymes (ALT and AST) and fibrosis biomarkers (ELF score and Pro-C3 levels) were also detected, indicative of a robust activity of Aldafermin against liver injury. However, there was no reduction in the pre-specified primary endpoint cholestasis marker ALP, as well as TC and LDL-C levels. Despite the fact that pathognomonic lesions in PSC and PBC suggest an association between liver fibrosis stage and transplant-free survival, liver biopsies again were not routinely performed. Further trials with Aldafermin in patients with PSC or PBC with longer term administration and imaging endpoints should be conducted.

In another report (NCT02704364 trial) involving 62 ALP-elevated PSC patients compared to the NCT02443116 study with 176 biopsy-confirmed NASH patients, Aldafermin treatment significantly suppressed the levels of more hydrophobic secondary BA species that have greater detergent activity and cytotoxicity [64]. Specifically, Aldafermin predominantly reduced the levels of glycine-conjugated BAs, and these changes correlated with changes in the fibrogenesis marker Pro-C3 in the pooled NASH and PSC individuals, suggesting that Aldafermin may be broadly applicable across multiple metabolic and cholestatic diseases. The comparison also suggests that BAs may contribute to providing a pro-fibrogenic microenvironment in the liver in these disease conditions. Unfortunately, the impact of Aldafermin on reducing fat content and lipogenesis and promoting energy expenditure beyond lowering BA levels was not analyzed. It would be also valuable to compare the effect of FXR agonists on the induction of FGF19, a major mediator of the effects of the BA-FXR signaling axis in both PBC and PSC patients. Along these lines, in a recent trial in PBC patients, Tropifexor, a potent non-BA FXR agonist, significantly attenuated markers of bile duct injury with a decrease of C4 and an increase of FGF19 [65]. Such changes were also seen in healthy volunteers and patients with NASH and primary bile acid diarrhea (BAD) treated with Tropifexor [66, 67], although other studies indicate the effect of FXR agonist Px-102 might be independent of FGF19 in health volunteers [68].

In the short term, Aldafermin appeared to be reasonably well-tolerated in both PBC and PSC patients. However, higher rates of mild to moderate gastrointestinal side-effects, including diarrhea, loose and frequent stools, headaches, and injection site reactions, were all observed in Aldafermin treatment arms in both trials. Some Aldafermin-treated PSC patients also reported an increased appetite. Interestingly, the increased incidence of diarrhea, despite the decreased BA levels in the colon, contradicts the notion that high BA levels cause this outcome.

2. FGF19 analogs to treat NASH

NASH represents the irreversible progression of NAFLD to more aggressive conditions such as cirrhosis, ascites, variceal hemorrhage, hepatic encephalopathy, HCC, and liver failure, all of which may necessitate liver transplantation. NASH currently affects more than 80 million people worldwide, and the prevalence of NASH and NASH-related end-stage liver disease is projected to rise continuously over the next decade, posing a significant challenge for future public health management [69-71]. Moreover, most NASH patients also have coexisting conditions such as obesity, T2D, insulin resistance, hypertension, or dyslipidemia [72] significantly complicating disease treatment. As of now, there are no approved pharmacotherapies for this condition.

Emerging evidence suggests that an aberrant increase in BAs is a significant risk factor for chronic liver diseases such as NASH, in addition to the usual illnesses associated with cholestasis [73]. Individuals with NASH exhibit elevated CYP7A1 in hepatocytes and BA levels in blood, feces, and urine as compared to healthy subjects [74-77]. Given the physiological nature of bile acids as “mild detergents”, high BA levels could potentially induce harmful effects, such as the activation of hepatic stellate cells, mitochondrial dysfunction, and endoplasmic reticulum stress, which could ultimately lead to cell death, compensatory ectopic tissue overgrowth, liver injury, immune infiltration, inflammation, and fibrosis [78]. Several clinical studies have reported that patients with NASH have reduced circulating FGF19 concentrations in correlation with elevated BA levels [79, 80]. Conversely, these NASH patients also exhibited increased hepatic KLB and FGFR4 expression, suggesting a potential disruption in the the FGF19-KLB-FGFR4 signal pathway [77].

Preclinical studies using cell lines and animal models have demonstrated that Aldafermin or native FGF19/15 can improve NASH pathogenesis by potentially influencing two major pathways. First, FGF19 inhibits the de novo BA synthesis pathway, reducing hepatic, biliary, intestinal, and circulating BAs through the enteric FXR to the hepatic FGFR4-KLB signal axis. Second, FGF19 regulates key metabolic pathways to enhance insulin sensitivity, fatty acid oxidation and energy expenditure, and rectify peripheral mitochondrial dysfunction [20, 21, 62, 81, 82]. These two mechanisms activated by FGF19 can lead to reduced lipotoxicity, hepatic inflammation, injury, and fibrosis and a restoration of liver function. Overexpression or animal dosing with FGF19 reversed diabetes, fatty liver, obesity, and hyperlipidemia in ob/ob, db/db, and DIO mice [19, 83, 84]. Similarly, overexpression of FGFR4 prevents hyperlipidemia, insulin resistance, and fatty liver in rodents [85]. These preclinical results led to several clinical experiments with Aldafermin in NASH patients that produced favorable metabolic outcomes reminiscent of the clinical pharmacology of FGF21 analogs in animal models (Table 1).

In the first phase 2 trial (NCT02443116) in 82 patients with biopsy-confirmed NASH over a period of 12 weeks, both 3 and 6 mg doses of Aldafermin resulted in rapid and sustained improvements in liver fat content (LFC) as measured by non-invasive MRI-PDFF and fibrosis (ELF) score (Hyaluronic acid, PIIINP and TIMP-1), while also decreasing serum ALT, AST, and Pro-C3 [86]. It is worth noting that treatment efficacy was more pronounced in patients with higher baseline LFC and ALT. These changes were strongly correlated with a significant decrease in C4 and a significant increase in LDL-C, indicative of a potent target engagement. The increase in LDL-C has been previously observed in studies involving OCA to reduce BA [87], which was assumed to be likely associated with increased enteric FGF19. Given that patients treated with Aldafermin developed atherogenic dyslipidemia, which is associated with an increased risk of atherosclerosis [88, 89], the increases in LDL-C in these patients were successfully mitigated with concomitant use of statin, a HMG-CoA reductase inhibitor. This led to reductions in plasma triglycerides, TC, and LDL-C, and an increase in HDL-C [90].

In the next report aimed to assess liver histological changes using liver biopsies from 43 NASH patients, Aldafermin treatment for 12 weeks produced marked antifibrotic, antisteatotic, and anti-inflammatory activities [91]. 50% of NASH patients at 1 mg to 68% at 3 mg Aldafermin dose achieved a significant reduction in NAS (steatosis, hepatocellular ballooning, and lobular inflammation) of 2 or more points without worsening of fibrosis. 42% of the patients treated with 3 mg Aldafermin had a decrease in fibrosis score ELF of at least one stage with significant reversal of trichrome stain intensity for fibrotic collagen in liver biopsies and no worsening of NASH. Aldafermin was generally well tolerated in NASH patients, with most treatment-related adverse effects being transitory diarrhea, frequent bowel movements, and injection site reaction. In an extended trial on 78 biopsy-proven NASH patients with fibrosis stage of 2 or 3, Aldafermin treatment for 24 weeks produced significant LFC improvement, fibrosis regression, and NASH resolution. At a post-hoc analysis, 22% of Aldafermin-treated patients achieved the combined histologic end point of both fibrosis improvement and NASH resolution compared with none by placebo. No significant changes in body weight, body mass index, glucose, insulin, HMAO-IR (homeostatic model assessment of insulin resistance), hemoglobin A1c, blood pressure, and heart rate were seen with Aldafermin treatment. Comparing to this efficacy outcome, only 7% of patients achieved this combined end point by OCA treatment compared to 4% by placebo in the recent REGENERATE trial. To note, the peak blood exposure concentrations of Aldafermin reached 10 ng/mL, while OCA achieved a peak elevation of endogenous FGF19 to only 0.3–2 ng/ml via FXR engagement [67, 92]. Whether such a difference accounts for the potency of Aldafermin vs OCA in treating NASH remains to be seen in further studies.

A phase 2b study was recently reported (ALPINE 2/3, NCT03912532) in 145 patients with NASH at stage 2 or 3 of fibrosis. 24-week treatment resulted in significant improvements in steatosis, inflammation, liver injury, fibrosis, and serum BAs, which were associated with NASH resolution and improvement in disease activity including reductions in NAS score and individual components [34]. However, Aldafermin treatment did not reach a pre-specified significance in fibrosis improvement of at least one stage with no worsening of NASH. Again, except for increases in LDL-C levels which could be effectively managed by statin co-administration, no detrimental effects were detected on blood pressure, heart rate, hematological parameters, or kidney function, as well as levels of glucose, insulin, and HbA1c. Currently, Aldafermin is further being evaluated in an ongoing ALPINE 4 trial in patients with NASH and compensated cirrhosis (NCT04210245).

It should be noted that FGF21 and GLP-1 analogs, both of which are well-known metabolic regulators of glucose, lipid, and energy metabolism, have also been evaluated for treating patients with NASH [93-95]. These treatments have shown promising clinical outcomes by therapeutically modulating metabolic dysfunction in NASH. However, these drug candidates have not demonstrated significant activity in reducing or affecting BA levels [96, 97]. On the other hand, Aldafermin primarily mitigates BA toxicity by directly inhibiting CYP7A1 through the FGFR4-KLB pathway in hepatocytes. It has also been observed that FGF19, which may escape the enterohepatic circulation, improves obese and diabetic phenotypes at the pharmacological levels in animal models. It was postulated that uncontrolled BA synthetic activity can lead to BA accumulation, toxicity, and local inflammation, which may contribute to liver injury in NASH [73]. Interestingly, Aldafermin does not affect fasting glucose, insulin, HOMA-IR, and HbA1c concentrations. This suggests that it may not correct glucose/insulin metabolic dysfunction as observed with GLP-1 and FGF21 analogs in man. However, all these treatments consistently exert strong inhibitory effects on LFC, inflammation and fibrosis. How much effect of Aldafermin on improving NASH is from its action on the peripheral tissues in comparison to FGF21 or GLP analogs warrants careful follow-up, which will help to elucidate the underlying mechanism. It should also be noted that, unlike Aldafermin, FGF21 analogs significantly reduce LDL-C while increasing HDL-C in patients with NASH. Interestingly, however, both types of these FGFs appear to cause diarrhea in patients. A neuronal effect rather than a metabolic consequence has been proposed to be at play for this side-effect [34, 94].

3. FGF19 analogs to treat T2D and obesity

Studies have shown that serum FGF19 levels are reduced in patients with T2D and are inversely correlated with the severity of diabetes progression, as determined by the C-peptide and HbA1c levels [98-101]. Pharmacologically, FGF19 has been compellingly demonstrated to have profound antidiabetic effects in various murine models, including obese DIO, monogenic ob/ob and db/db, polygenic TALLYHO and NONNZO, and beta-cell deficient mice [77, 83]. In addition to improvements in glucose metabolism, insulin sensitivity and lipid profiles, animal studies have also shown that overexpression of FGF19 or treatment with recombinant FGF19, similar to FGF21, enhances metabolic rates and energy expenditure while reducing adiposity [83, 84, 102]. However, a recent study showed that Aldafermin had only a modest glucose-lowering effect in DIO mice, which was possibly secondary to the observed weight reduction, and no improvement was detected in the severe hyperglycemic condition in db/db mice [77]. Bariatric surgery, including Roux-en-Y gastric bypass (RYGB) and vertical sleeve gastrectomy (VSG), has proven to be the most effective treatment for hyperglycemia and obesity, particularly in severely obese patients with T2D. Murine intestinal FGF15 was increased by VSG in mice. The gut-specific loss of FGF15 in HFD-fed mice undergoing VSG abrogated its effects on improving glucose tolerance and led to greater loss of muscle mass, body weight, and bone density than non-VSG mice. This was accompanied by a marked increase in hydrophobic BA level, indicating a role of gut FGF15 in protecting VSG mice from the deleterious effects of high BA levels [103]. It was also found that serum FGF19 increased significantly after RYGB in most patients under fasting or after a standard meal as early as 7 days and 21 days post-surgery yet unfortunately, the changes in GLP-1 and FGF21 levels were not parallelly evaluated [77]. However, another study showed that gastric bypass increased FGF19 while decreasing total BAs and FGF21 contents [99]. These animal and clinical results have renewed interest in Aldafermin for treating diabetic patients (Table 1).

Consequently, a clinical trial was conducted in 81 T2D patients who were inadequately controlled by metformin therapy (NCT01943045) [77]. The initial hypothesis was that FGF19 could be the long-seeking surgical factor mimetic for the glucose-lowering effects of RYGB. However, daily administration of Aldafermin (2, 5 and 10 mg vs placebo) for 28 days did not significantly alleviate hyperglycemia, as indicated by the lack of significant changes in plasma glucose and HbA1c levels. This contrasted with substantial reductions in C4, ALT, and AST, all indicative of potent target engagement. Nevertheless, improvements in insulin resistance based on HOMA-IR were observed in patient cohorts at both 5 and 10 mg doses of Aldafermin. Fasting insulin and HOMA-bcf (surrogate measures of insulin secretion), mean body weight, and fructosamine also decreased from baseline in the 10 mg Aldafermin treatment group.

Similarly, native FGF21, its analogs, and FGFR1-KLB co-agonists have demonstrated significant anti-diabetic effects in several diabetic mouse models. However, human clinical trials with LY2405319, PF05231023, Pegbelfermin (BMS-986036), or BFKB8488A on T2D patients failed to demonstrate statistically significant control of hyperglycemia [104-108]. Yet, later trials with Efruxifermin, Pegozafermin, or NGM313 on T2D or NASH patients showed improvements in markers of insulin sensitivity and glucose control, including fasting glucose, HbA1c, and HOMA-IR. These improvements were associated with reductions in LFC and body weight [93, 109, 110]. The exact mechanism for the variations in the efficacy of glycemic control between animals and human subjects and among FGF21-based agents in patients is currently unknown. It is postulated that factors like treatment duration, inadequate serum dose exposure, potency, and time action properties may play a role. It is also unclear whether the glycemic effects with FGF21 analogs are direct or secondary to the potent reductions in dyslipidemia and/or weight loss in humans.

4. FGF19 analogs to treat other metabolic and inflammatory diseases

BAs are known to increase mucosal permeability, stimulate water and electrolyte secretion, and accelerate colonic transit partly by triggering propulsive high-amplitude contraction. An excess of BAs can lead to bile acids-associated diarrhea (BAD), a condition that is often seen in patients with ileal resection, Crohn’s disease, idiopathic disorders such as functional diarrhea or diarrhea-predominant irritable bowel syndrome (IBS-D), or BA malabsorption in association with chronic pancreatitis or celiac disease. BAD is also connected to a higher body mass index, increased stool weight and stool fat, and faster colonic transit. The molecular causes of BAD are thought to include a decrease in the level of FGF19 and the genetic variations in the genes responsible for BA synthesis, transport, enterohepatic circulation, BA receptor TGR5 (which mediates BA effects on colonic secretion and motility), and regulators of BA homeostasis such as FXR, FGFR4, and KLB [111-113]. These variations could be associated with increased mucosal permeability and water secretion, inhibited apical Cl/OH exchange, increased mucus lubrication, and accelerated colonic motility. Currently, chronic BAD is treated with BA sequestrants such as cholestyramine, colestipol, and colesevelam. However, the use of FXR agonists such as OCA is emerging as a promising new approach to treat this condition.

The FGF19-FGFR4-KLB pathway has been shown to regulate BA synthesis and colonic transit in IBS-D [112, 114]. Up to 30% of patients with IBS-D have lower serum levels of FGF19 [111], which is associated with increased BA synthesis or fecal excretion [115]. FGF19/15 is known to reduce C4 and other components of BAs in the hepatic, biliary, enteric tissues and systemic circulation. It also promotes gallbladder distension and bile refilling for the next meal [116], which could potentially improve BAD. However, as alluded earlier, phase 2 trials with Aldafermin in T2D, PBC, PSC, and NASH patients have shown unexpected increases in abdominal cramping, diarrhea, loose stools, or stool frequency that were dose-dependent and present in all study subjects. These gastrointestinal symptoms contradict our current understanding of the roles of BAs reaching the colon in promoting diarrhea and of FGF19 activity in reducing BA synthesis and promoting gallbladder refilling. Therefore, the concentrations of BAs in colonic fluid or feces should be carefully measured in future trials.

A phase 2 trial was launched in 2015 to study the pharmacodynamics of Aldafermin on colonic transit, BA homeostasis, and fecal fat in subjects with functional constipation and healthy individuals (NCT02649062) [117]. Aldafermin significantly accelerated gastric emptying and overall colonic transit at 24 and 48 hours. The 6 mg dose of Aldafermin also significantly increased the number of weekly bowel movements with softer stool and easier passage than at 1 mg dose or placebo. No changes were observed in stool weight and fecal fat excretion across all groups. As expected, fasting serum C4 was reduced by Aldafermin in a dose-dependent manner. Fecal BA excretion was decreased by Aldafermin without evidence of steatorrhea, suggesting that steatorrhea, secondary to BA deficiency, is not the cause of the increased bowel movement, colonic transit, or gastric emptying. The colonic transit-promoting effect of Aldafermin is consistent with its propensity to induce diarrhea and loose stools as revealed in several clinical trials in patients with PBC, PSC, T2D and NASH. Compared to the commonly used 5-HT4 (serotonin) receptor agonists in clinics, this FGF19 analog could therefore provide a novel alternative for use as prokinetics for accelerating gastric emptying and colonic transit in patients with gastrointestinal dysmotility such as gastroparesis and colonic inertia.

Aldafermin reduces BA excretion by decreasing BA synthesis, which should theoretically result in delayed colonic transit and inversely associate with gastric emptying [118]. However, Aldafermin actually speeds up both gastric and colonic transit without increasing fecal fat and colonic secretion, regardless of its effect on decreasing BA synthesis. This suggests a mechanism unrelated to BA kinetics or induction of steatorrhea but rather related to colonic motility. Moreover, the acceleration of colonic transit was more pronounced in patients with the KLB rs17618244 minor allele mutation. KLB and FGF15 have been shown to control neural tube development, and KL deficiency resulted in lower cholinergic signaling [119, 120], suggesting a potential effect of FGF19 on peripheral cholinergic neurons in the gastrointestinal tract. The mild adverse effects associated with Aldafermin include hyperphagia, flatulence, bloating, headache, injection site reaction, and interestingly, diarrhea. These results suggest that the clinical use of FGF19 analogs in treating metabolic and cholestatic diseases may necessitate the use of antidiarrheal agents. The same principle may also apply to the clinical use of FGF21 analogs or its receptor agonists that appear to cause diarrhea, which is also likely unrelated to BA effects.

Interestingly, a new trial has been initiated to compare the effects of Aldafermin on bowel functions and hepatic synthesis and fecal excretion of BAs in patients with diarrhea associated with BA malabsorption (NCT05130047). In addition, a phase 1 trial of Aldafermin is currently underway in participants with CKD (NCT04179630). The rationale and outcomes of both trials have not yet been reported.

Antagonists of FGF19-FGFR4 for hepatic and biliary cancers

1. Preclinical studies

While human FGF19 and its analogs have shown significant potential in treating PBC, PSC, and NASH, and in promoting bowel movement, they also carry significant therapeutic risks. These molecules function by activating the FGFR4-KLB receptor to regulate BA synthesis and/or the FGFR1/2/3-KLB complexes to promote metabolic homeostasis. However, chronic elevation of FGF19 can lead to the development of HCC, ICC, and potentially other cancers. Physiologically, FGF19/15 is produced by ileal enterocytes, gallbladder, and common bile ductal cells, but not hepatocytes [1, 121-124]. Yet, under certain disease conditions or challenges, FGF19 expression can be induced in the liver or primary hepatocytes [54, 57, 125, 126]. FGF19 can induce proliferation of hepatocytes [127] and activate cells of cholangiocytic, enteroendocrine, and enterocytic origins [121], through activating FGFR4 in the absence of KLB [128]. This aligns with the earliest report of various FGFR stimulation in a heparin dose-dependent manner, albeit at a much higher concentration than FGF1 [129] and is also consistent with the classic FGFs’ proliferation-promoting activity. However, the oncogenic capacity of FGF19 is peculiar given its pivot role in BA homeostasis. The main physiological function of FGF19 is to reduce the biosynthesis of BAs and their content, which is typically associated with reduced risk of tissue damage and cancerous lesions in hepatic, biliary, and intestinal system [27, 33, 39, 130]. The reason why evolution has endowed FGF19/15 with such unique dual activities is a topic that warrants further exploration.

The oncogenicity of FGF19 for HCC development was first discovered in a transgenic mouse model that overexpressed FGF19 in the skeletal muscle. These mice exhibited increased hepatocyte proliferation, hepatocellular dysplasia, and neoplasia, and developed HCC by 10–12 months of age. These pathologies could be prevented by deleting FGFR4 or by using FGFR4 or FGF19 neutralizing antibody [35, 131, 132]. FGF19 amplification (11q13 amplicon) and expression aberrations (e.g. gained ectopic expression in hepatocytes) were found in a small subset (6-12%) of HCC patients and in some HCC cell lines, where this factor activates the FGFR4-KLB axis in an autocrine fashion [38, 133-136], leading to cross-talk with STAT3/beta-catenin/ERK1/2 pathways that are typically associated with proliferative as well as metabolic effects [61, 137]. Ectopic expression of FGF19 was also found in non-parenchymal cells of the livers with cholestatic diseases that constitute a significant risk for HCC and ICC [54, 124]. Clonal growth and tumorigenicity of HCC cells harboring the 11q13.3 amplicon can be inhibited by RNAi-mediated knockdown of FGF19 or by an anti-FGF19 antibody. AAV-mediated chronic FGF19 delivery caused higher incidences of HCC development at 52 weeks in wildtype C57BL6/J and BDF mice specifically upon metabolic challenge [61]. FGF19 was implicated in the resistance to Sorafenib, a multi-kinase inhibitor, and was proposed to be an oncogenic driver [138] and a potential predictive marker of response to FGFR kinase inhibitors such as BGJ398 [133, 139]. FGF19 is secreted from gallbladder to stimulate the relaxation and postprandial bile refilling, and also from GBC cells to stimulate proliferation in an autocrine manner. In patients with GBC, FGF19 levels in bile are significantly elevated and both FGF19 and FGFR3 might be associated with poor prognosis [36].

However, some FGF19 analogs, such as FGF21/19^38-42, Aldafermin, and M52, seem to lack hepatocyte proliferation- or HCC-promoting activity as demonstrated in several mouse models including ob/ob, db/db, rasH2, Mdr2/3^−/−, Fxr^−/−, and HFFCD [33, 59, 61, 77, 127], permitting the use of such variants to treat chronic diseases. Aldafermin was even claimed to inhibit FGF19-FGFR4 pathway-dependent tumor growth in rodents even though the mechanism of this effect is unclear [61].

Based on these cancer-promoting preclinical observations, the strategies directly targeting FGF19/FGFR4 pathway have been explored using either small-molecule inhibitors or neutralizing antibodies in mouse models of HCC as well as other cancers. H3B-6527 is a highly selective, covalent, and irreversible inhibitor targeting cysteine-552 in the middle-hinge region of the ATP-binding site of FGFR4 kinase domain that is getting developed by H3 Biomedicine. The proliferation of HCC cells in vitro and the growth of xenografts of Hep3B and patient-derived xenografts (PDX) in mice, all of which showed increased FGF19 expression, were reduced by H3B-6527 monotherapy or in combination with CDK4/6 inhibitor [140]. The BLU9931 and its successor BLU554 (Blueprint Medicines) are another highly selective and covalent inhibitor targeting FGFR4 kinase via the same mechanism [136]. BLU9931 shows a robust and dose-dependent inhibition of FGFR4 activity and cell proliferation in vitro and remarkable antitumor activity in mice bearing HCC xenografts and PDX that overexpress FGF19 due to either amplification (such as in Hep 3B, HUH-7, and JHH-7 cells), or mRNA transcriptional upregulation (such as in LIXC012 PDX), or even when FGFR4 signaling pathway is not perturbed. The FGF401 (Novartis) is also a highly potent, selective but reversible inhibitor for FGFR4 kinase [141]. FGF401 inhibited the FGF19-FGFR4 signaling and exerted remarkable antitumor activity in mice bearing HCC, PDX, NIH3T3 xenograft of rhabdomyosarcomas harboring a specific FGFR4 N535K mutation, or gastric cancer PDX that are all positive for FGF19, FGFR4, and KLB [142, 143]. Of note, other studies indicate that the tumorigenicity of FGFR4 could also be the consequence of an associated systemic, microenvironmental or cellular metabolic challenge [144, 145]. Loss of FGFR4 delayed TGFalpha-HER2 mediated breast cancer progression, which might be attributable to the altered metabolic effects of endocrine FGF15/21 signaling.

Of note, the FGF19 neutralizing antibodies were also shown to inhibit tumor growth in rodent models of HCC and CRC [38, 132]; however, safety study in cynomolgus monkeys with this molecule revealed dose-dependent liver toxicity, severe diarrhea, and low food consumption, which were due to the increases in Cyp7α1 expression, BA synthesis and alteration of the expression of bile transporters in the liver [146]. Similarly, single or repeated doses of FGF401 in dogs recapitulated the anti-FGF19 adverse effects in monkeys, including induction of Cyp7a1 in the liver, increased plasma and fecal BA content in association with decreased serum cholesterol, diarrhea, and secondarily elevation of serum ALT with these effects being mitigated by cholestyramine, a BA sequestrant [147]. Interestingly, Aldafermin and M52 themselves exhibited both anti-cancer and anti-cholestatic effects in mice. If such dual effects are repeated in independent studies, these FGF19 analogs could represent the class of ideal agents to treat cancer- and BA-associated diseases. To note, selective FGFR4 kinase inhibitors appeared to have little, if any, effect on inducing hyperphosphatemia or tissue calcification common upon the use of the pan-FGFR or MEK inhibitors that disrupt FGF23 signal [148, 149]. Thus, specific FGFR4 inhibitors as BLU554 (Fisogatinib), H3B6527, and FGF401 (Roblitinib) are currently tested in clinical HCC or ICC trials (Table 1).

2. Clinical trials with FGFR4 antagonists to treat cancers

Fisogatinib (BLU-554) was evaluated in 106 advanced HCC patients who had FGF19 expression (NCT02508467). Across 140 to 900 mg doses QD for 28 days, the overall response rate was 17% with median duration of response of 5.3 months but 0% for FGF19-negative patients [150]. These results validate FGFR4 as a targetable driver for FGF19-positive advanced HCC, supporting further clinical trial of Fisogatinib in FGF19-positive patients using mono or combination therapy. Levels of plasma cholesterol were reduced along with increased BA levels, reflecting activation of BA synthesis. 13 patients (12%) had increased ALT, AST, bilirubin, diarrhea, pyrexia, fatigue, and abdominal pain leading to discontinuation. Of note, levels of serum FGF19 were increased in a dose-dependent manner presumably to compensate for FGFR4 inhibition. Furthermore, acquired mutations of the gatekeeper (V550M/L) and hinge-1 (C552R) residues in FGFR4 were found in two patients of phase 2 trial at 600 mg QD by week 31-32, leading to Fisogatinib-resistant, progressive disease, further validating FGFR4 as a critical driver of oncogenicity [151].

H3B-6527 was granted an orphan drug designation by FDA in October 2017. A phase 1 trial (NCT02834780) was conducted in 90 advanced HCC, who progressed after at least one prior therapy. H3B-6527 was given orally at doses of 300-2000 mg QD or 500-700 mg BID for 21 days [152]. It demonstrated an encouraging clinical outcome in HCC patients with >2 prior lines of therapy, including overall survival of 10.6 month and progression-free survival of 4.1 month with an overall response rate of 16.7%. 1000 mg QD was recommended as the phase 2 dose.

Roblitinib (FGF401) alone and in combination with Spartalizumab, an inhibitory antibody against PD-1 receptor, was tested in phase 1 study with 74 patients and phase 2 with 86 patients having HCC or other solid tumors (NCT02325739). Four patients were reported with a partial effect in phase 2 (120 mg QD) and 1 complete and 3 partial responses (tumor remission) in phase 1 (50 to 150 mg) patients, all having HCC [153]. Such modest clinical activity was likely due to no pre-selection of the patients based on the FGFR4-FGF19 pathway aberrations, as FGF401 demonstrated a robust target engagement at all treatment doses based on elevated levels of C4, total BA, AST, ALT, and circulating FGF19 while reduced plasma cholesterol.

Of note, treatments with Roblitinib, Fisogatinib and H3B6527 were all associated with grade 1/2 gastrointestinal adverse effects, such as diarrhea, nausea, emesis, and fatigue. Interestingly, some of the adverse effects, diarrhea in particular, were also seen in trials with FGF19 agonist Aldafermin, which is mechanistically opposite to FGFR4 inhibition. INCB062079, a FGFR4 inhibitor developed by Incyte, went into a phase 1 trial in subjects with advanced HCC, ICC, and other malignancies in May 25, 2017 (NCT03144661), yet the outcome of this study has not been reported.

Overall, the current small molecular inhibitors targeting FGFR4 appear to present with good tolerability, and effectiveness against a subset of HCC positive for FGF19-FGFR4-KLB signaling, and only with mild BA elevation-associated adverse effects that are common in all modalities targeting either FGF19/FGFR4 or using FGF19 analogs. The selectivity of Fisogatinib, H3B6527 and Roblitinib for FGFR4 is excellent with IC₅₀ at 1 - 5 nmol/L compared to that for FGFR1-3 being in the triple digits or the μmol/L range. JNJ-42756493 (Erdafitinib, Balversa) from Janssen and LY2874455 from Lilly have an IC₅₀ of 1-6 nmol/L for all 4 FGFR isotypes, which may be effective for a broad range of cancers including the FGF19-FGFR4 positive HCC. Hatlen and colleagues demonstrated that LY2874455 overcame resistance to Fisogatinib in both in vitro and in vivo HCC models [151]. Erdafitinib, a first-in-class pan-FGFR inhibitor drug approved for bladder cancer with FGFR alterations [154], may also find its usage in overcoming HCC resistance to other FGFR4-targeting agents as well.

The positive clinical outcomes of FGFR4-targeting inhibitors should boost momentum to try other therapeutic modalities in targeting FGF19-FGFR4 signaling. Monoclonal antibodies to FGF19–FGFR4 pathway have been pursued but so far only in preclinical animal models. Neutralizing antibodies for FGF19 inhibited the binding of FGF19 to FGFR4 and the growth of colon tumor xenografts and HCC in FGF19 transgenic mice [38, 132]. Likely due to profound perturbations in BA metabolism, however, they did not advance to clinical trials [146]. LD1, U3-1784 and 3A11ScFvFc are high-affinity neutralizing antibodies against FGFR4 and have been tested in different HCC cells and xenograft models in mice and cynomolgus monkey [131, 155, 156]. However, no clinical experiments for any of these antibodies have been registered yet.

Mechanistic considerations

FGF19-FGFR4-KLB signaling is responsible for keeping BA levels in the physiological range and maintaining metabolic health yet when disturbed it is promoting HCC. On such basis, clinical trials have since been conducted with both FGF19 agonists and antagonists targeting this pathway but for different types of patients having either metabolic or cholestatic diseases or cancer. The pharmacological outcomes from these short-term phase 1/2 trials are generally positive and indicative of well-defined target engagement. Some concerns, however, exist and remain to be explored in the future studies.

Firstly, both approaches, though opposite in nature, are designed to target the same pathway and hence, will likely act through the same mechanism (Fig. 1). The FGFR4 kinase is the ultimate determinant for the effects of this pathway as it directly transmits the FGF19 signal to intracellular mediators/adaptors. Since neither KLB nor KL are active kinases and there’s no evidence to suggest that KL/KLB has other direct signaling-relaying capacities beyond being structural scaffolds for eFGFs, it is predictable that agonists could potentially cause opposite effects of antagonists, and vice versa, deemed to be side-effects, especially over a prolonged duration. The main goal of FGF19 analogs is to decrease BA content and toxicity, along with associated hepatic, biliary, and intestinal cholestatic diseases such as PBC, PSC, BAD and others. These molecules also aim to reduce fat content, lipotoxicity, and associated metabolic diseases such as NASH. However, long-term exposure to FGF19 analogs could potentially promote the development of neoplastic lesions and HCC. Such concerns have led to the exploration of FGF21/19^38-42, FGF19dCTD, Aldafermin, and M52 molecules which are claimed to be non-oncogenic while fully metabolic [20, 33, 59, 127]. Attesting to these concerns, a most recent publication in bioRxiv showed that Aldafermin, like FGF19 and mouse FGF15, still reveals oncogenicity in mouse model of HCC (doi: https://doi.org/10.1101/2023.09.15.557921).

Secondly, evidence suggests that FGF19/15-FGFR4 pathway plays an important role in liver regeneration following partial hepatectomy (PH). FGF19/15 appears to directly stimulate the growth of hepatocytes and biliary epithelial cells. The relatively weak mitogenic activity of FGF19/15 may be crucial in promoting liver regeneration, and still safe, particularly in response to potential damage caused by anomalies in the diurnal rise and fall of BA tides [15, 157]. FXR activation is known to facilitate liver regeneration and repair after partial hepatectomy or injury, at least partially through the induction of enteric FGF15. Studies have shown that ectopic expression of FGF15 can compensate for defective liver regeneration and repair in mice with intestine-specific FXR deletion [158]. Following PH in mice, ileal and serum FGF15 levels were found to increase, and hepatic FGFR4 was activated. Administration of FGF19 accelerated the restoration of liver mass 48 and 72 h post-PH in wildtype mice, but not in FGFR4 knockdown mice [159]. The absence of hepatic FGFR4 was found to impair liver regeneration due to reduced hepatocyte proliferation or increased necrosis [160, 161]. Transgenic expression of activated FXR in the intestine of Mdr2^−/− mice was found to protect against liver damage by inducing FGF15 expression [123]. Mice lacking Fgf15 who underwent PH exhibited higher mortality due to toxic intrahepatic BA concentrations, while AAV-mediated FGF15 delivery improved survival rates after extensive liver resection [162]. Liver growth and proliferation of hepatocytes and cholangiocytes induced by cholic acid feeding were significantly reduced in Fgf15^−/− mice. These findings suggest that FGF15 is necessary for BA-induced liver growth and that FGF15-regulated BA homeostasis is vital for liver protection during regeneration. It is possible that this regenerative activity is exploited by cancers through chronic and constitutive elevation of FGF19/15-FGFR4 signaling (Fig. 1).

Thirdly, it is worth to mention the difference between murine FGF15 and human FGF19 pharmacology. FGF15 was first implicated in promoting CCl₄ and DEN-induced fibrosis and HCC development [163], but a later study showed that FGF15 lacked the ability to induce HCC even at supra-pharmacological levels in db/db, DIO, and Mdr2^−/− mice [25]. By marked contrast, long-term exposure to FGF19, but not FGF15, was found to induce HCC in db/db and DIO mice. In addition, FGF15 lacked the anti-diabetic effects in db/db mice and the ability to activate STAT3 pathway characteristic of human FGF19, although both FGF15 and FGF19 reduced BA levels and resolved hepatocellular and biliary injury [25]. The mechanism for such species-specific difference is currently unknown. However, the results raise concerns of relying on preclinical rodent models for safety assessment of clinical agents that have potential to raise FGF15 or activate hepatic FGFR4, such as FXR agonists for NASH and PBC patients, or even some FGF19 analogs. Of note, inhibitors for FGFR4 in patients also raise FGF19 in a dose dependent manner.

Fourthly, FGF19/15 are inversely associated with BA levels, presenting a paradox in their roles in tumorigenesis. Chronically elevated BAs are known to cause liver toxicity, BAD and hepatobiliary damage and to promote spontaneous HCC development, as showed by the anti-FGF19 antibody, FXR-SHP deficiency, and YAP activation in rodents and monkeys [39, 146, 164]. Chronic FGF19 treatment significantly lowered BA levels and inhibited some BA-associated diseases, but ironically at the expense of promoting HCC development. This argues against a direct association of BA with cancer development [62, 77]. On the other hand, Aldafermin and M52, the non-tumorigenic FGF19 variants, similarly reduced BA levels, inflammation, and fibrosis without causing HCC, suggesting an intricate balance among BA-FGF19, liver regeneration, and tumorigenesis to control the threshold levels [33, 62, 77]. A recent study showed that FGF15 could activate FGFR4 to recruit NF2-Raf-Mst1/2 Hippo pathway to restrain liver overgrowth and tumorigenesis while depressing bile acid synthesis, consistent with the idea of species or context specificity proposed in other reports [25, 165, 166]. These findings present a mechanistic paradox about the roles of the elevated BA and FGF19/15 in tumorigenesis, which needs to be clarified along with clinical trials.

Lastly, the question remains if the chronically depressed BA levels impose any pathological effects beyond the increased cholesterol levels in patients in a longer term, such as malabsorption of lipid, lipophilic steroids and vitamins to cause associated health issues.

Conclusion and future perspectives

In summary, the use of FGF19 analogs to promote FGFR4-KLB signaling seems to be an effective treatment for metabolic and cholestatic diseases. However, it carries significant risks, including the potential to promote neoplastic diseases and cardiovascular issues due to increased levels of plasma cholesterol and LDLc in humans. Current clinical trials have been short-time, leaving these concerns to be addressed in longer-term trials. On the other hand, FGFR4 inhibitors show promise for treating HCC in patients with defects in the FGF19-FGFR4 pathway. However, these inhibitors disrupt BA homeostasis by increasing new synthesis and total content. This effect is like an overdose of cholagogue, which can lead to BA toxicity, BAD, intra- and extra-hepatic cholestatic disease, or other inflammatory disease in the liver, gallbladder, bile duct, and gut such as IBS and IBD. While BA levels can be managed by sequestrants, the long-term effects are yet to be determined.