Targeting FGFR inhibition in cholangiocarcinoma

Cholangiocarcinomas (CCAs) are rare but aggressive tumours of the bile ducts, which are often diagnosed at an advanced stage and have poor outcomes on systemic therapy. Somatic alterations with therapeutic implications have been identified in almost half of CCAs, in particular in intrahepatic CCA (iCCA), the subtype arising from bile ducts within the liver. Among patients with CCA, fibroblast growth factor receptor 2 ( FGFR2 ) fusions or rearrangements occur almost exclusively in iCCA, where they are estimated to be found in up to 10–15% of patients. Clinical trials for selective FGFR kinase inhibitors have shown consistent activity of these agents in previously treated patients with iCCA harbouring FGFR alterations. Current FGFR kinase inhibitors show differences in their structure, mechanisms of target engagement, and specificities for FGFR1, 2, 3 and 4 and other related kinases. These agents offer the potential to improve outcomes in FGFR-driven CCA, and the impact of variations in the molecular profiles of the FGFR inhibitors on efficacy, safety, acquired resistance mechanisms, and patients’ health-related quality of life remains to be fully characterized. The most common adverse event associated with FGFR inhibitors is hyperphosphatemia, an on-target off-tumour effect of FGFR1 inhibition, and strategies to manage this include dose adjustment, chelators, and the use of a low phosphate diet. As FGFR inhibitors and other targeted agents enter the clinic for use in FGFR-driven CCA, molecular testing for actionable mutations and monitoring for the emergence of acquired resistance will be essential.


Introduction
Cholangiocarcinomas (CCAs) are heterogeneous epithelial tumours arising from the biliary tree with features of cholangiocyte differentiation [1]. The anatomical subtypes of cholangiocarcinoma include intrahepatic cholangiocarcinoma (iCCA), which arises in the bile ducts within the liver, and extrahepatic cholangiocarcinoma (eCCA), which involves the ducts outside of the liver including the left and right hepatic ducts and the common bile duct.
The prognosis of both types of CCA is poor, but is particularly poor in iCCA, where only 30-40% of patients present with surgically resectable disease [2] and unfortunately, the majority of cases recur even in apparently resectable disease. The 5-year overall survival for patients with iCCA is <10% [3], so treatments to improve survival are urgently needed. As iCCA symptoms may be non-specific, such as vague abdominal discomfort, nausea, fatigue, and weight loss, delayed diagnosis is particularly common [4] In recent years, precision oncology has emerged as an promising approach for CCA.
One of the most promising range of targets are the fibroblast growth factor receptor 2 (FGFR2) fusions, gene alterations present in 10-15% of iCCAs, but in almost no eCCAs.
Multiple efforts to drug this target led to the first US Food and Drug Administration (FDA) approval in CCA. Pemigatinib, an oral selective FGFR inhibitor with potent activity against FGFR1-3, gained approval for treatment of patients with previously treated, locally advanced or metastatic CCA harbouring an FGFR2 fusion or rearrangement [5]. This review focuses on the molecular biology driving biliary tract malignancies, the clinical development of FGFR inhibitors in FGFR-altered CCA, and future considerations as this promising new precision medicine-based option moves into the clinic.

Molecular biology of the FGFR gene
The FGF-FGFR signaling pathway Fibroblast growth factors (FGFs) and their associated fibroblast growth factor receptors (FGFRs) have been studied extensively, with a focus to exploit the therapeutic potential of FGF-FGFR signaling being made over the past 10 years [6]. The FGF pathway consists of 22 human FGFs and four highly conserved transmembrane receptors with intracellular tyrosine kinase domains, FGFR 1-4 [7]. The FGFRs are expressed on multiple cell types [8]. FGF-FGFR signaling is triggered by the ligand-dependent receptor dimerization following binding of FGF at the cell surface. This leads to intracellular phosphorylation of receptor kinase domains, a cascade of intracellular signaling, and gene transcription that activates a number of intracellular survival and proliferative pathways ( Figure 1A) [9]. The specificity of the FGF-FGFR interaction is influenced by the differing ligand binding capacities of the receptor paralogues, by alternative splicing of FGFR, and by tissue-specific expression of ligands and receptors, coupled with cell surface or secreted proteins that facilitate the FGF-FGFR interactions and increase ligand specificity [10]. Alterations in FGFR genes, including activating mutations, chromosomal translocations, gene fusions, and gene amplifications, can result in ligand-independent signaling which, in turn, leads to constitutive receptor activation ( Figure 1B).
FGF-FGFR signaling has been shown to have oncogenic roles in many cancers. Key downstream signaling pathways altered by FGF-FGFR activation are the Ras-Raf-MEK-ERK pathway, the PI3-AKT-mTOR pathway, and JAK-STAT pathway ( Figure 1A) [11]. In an analysis of 4,853 solid tumours, FGFR aberrations were found in 7.1% of all cancers, with the majority (66%) being gene amplifications, followed by mutations (26%), and rearrangements (8%) [8]. Among the CCA tumours in that study (N=115), 7% harboured FGFR aberrations. These aberrations were mostly in the gene encoding for FGFR2 (6.1%), with a small proportion in the FGFR1 gene, and none identified in the FGFR3 or FGFR4 genes.
Genomic profiling of CCA Biliary tract cancers (BTCs) arise from epithelial cells lining the bile duct and can occur at distinct anatomical locations: intrahepatic, extrahepatic and in the gallbladder [12]. Analyses of the genomic and transcriptomic landscape of the anatomical subtypes of BTC show that molecular profiles vary between iCCA, eCCA, and gallbladder cancer (GBC), with multiple small cohorts of patients having mutually exclusive or co-existent aberrations [12][13][14][15]. Given the heterogenous nature of BTCs, it is unsurprising that multiple genetic factors are implicated in CCA development, including chromosomal aberrations, and genetic and epigenetic alterations in tumour suppressor genes and oncogenes.

Discovery of targetable FGFR2 aberrations in iCCA
The earliest report of FGFR2 fusions in CCA was in 2013 by Wu and colleagues [19]. The two fusions identified occurred in patients with iCCA, and subsequent studies have shown that FGFR2 fusions occur nearly exclusively in iCCA compared to other BTCs and epithelial cancers, making them a useful diagnostic marker. Across multiple tumour genotyping studies in CCA, the frequency of FGFR2 fusions in iCCA is estimated to be approximately 10-15% [20,[25][26][27]. Geography and etiology may impact reported frequencies in FGFR2 fusions.
Kongpetch and colleagues evaluated 193 CCA tumours from Thailand, Romania, and Singapore, and reported that rates of FGFR2 fusions were 0.8%, 6.8%, and 15.7%, respectively [28]. These authors also reported that the rate of FGFR2 fusions in flukeassociated and non-fluke associated CCA were 0.8% and 11.6%, respectively (p=0.0006), suggesting that FGFR2 fusions might play a crucial role in the evolution of non-liver flukeassociated CCA, but less so in liver fluke-associated CCA. An integrated data analysis from whole-genome sequencing/targeted DNA sequencing with RNA-fusion sequencing showed mutations in FGFR1, FGFR2, FGFR3 and FGFR4 were present in 1.0%, 3.6%, 1.0% and 0.5% of CCAs, respectively. FGFR2 fusions and FGFR mutations were mutually exclusive in this study [28].
FGFR2 fusions generally encode a functional fusion protein with FGFR2 fused to a partner gene at the C-terminus that has strong dimerization or oligomerization capabilities [19,20]. The most common partner is BICC1, but various other fusion partners with FGFR2 have subsequently been identified in iCCA [13,[19][20][21]25,[27][28][29] (Table 1), most of which fuse at a consistent breakpoint within the FGFR2 gene on chromosome 10 [21]. In vitro and in vivo experiments show that the oncogenic ability of FGFR2 fusion proteins can be suppressed by treatment with FGFR kinase inhibitors [19,20,30], this has been mirrored clinically.
FGFR2 fusions in iCCA have been associated with a better prognosis [27,31] and younger age at diagnosis [27,28] in some studies. They are also mutually exclusive with KRAS and BRAF [20] and ERBB2/BRAF/NRAS alterations [28] in some studies. FGFR2 fusions have been found to be frequently co-altered with mutations in the chromatin-remodeling gene BAP1 [27], which acts as a tumour suppressor in iCCA [32]. The implications of these genetics on the therapeutic potential of combination therapy have yet to be realized.
iCCA epidemiology and current systemic treatment for iCCA CCA epidemiology and risk factors for developing CCA Globally, the incidence and mortality rates of CCA show substantial geographical variation, which may reflect exposure to different geographical risk factors and genetic determinants [33,34]. Multiple studies from Europe, the USA, Japan and Australia have reported rising rates of iCCA [33,35], which appear to have plateaued over the past 10 years. This increase may be due to advances in imaging, molecular diagnostics and pathology, enabling more accurate diagnosis of iCCA [34,35], however, in contrast, the incidence of both perihilar CCA and distal CCA appears to be stable or decreasing [33,35] suggesting the increase is real. However, a recent international analysis of population-based incidence rates of CCA, the Cancer Incidence in Five Continents Plus (CI5plus), showed that the incidence rates of both iCCA and eCCA increased in a majority of countries worldwide during the period 1993-2012, with iCCA incidence rates being higher than eCCA incidence rates in most countries between 2008 and 2012 [34].
The highest rates of CCA are in South East Asia (Northeast Thailand, Cambodia, and Laos), where the incidence is approximately 80/100,000 per year compared to 1-2/100,000 in the UK and USA, the former primarily associated with liver fluke infection [4,36]. Other risk factors include primary sclerosing cholangitis, hepatolithiasis, liver fluke infections, chronic viral hepatitis, metabolic syndrome, alcohol use, and congenital anomalies of the bile ducts, such as choledochal cysts [4,33]. Risk factors may overlap, for example, parasitic infection often induces hepatholithiasis [37]. In Western countries about 50% of cases are still diagnosed without any identifiable risk factor despite advances in the knowledge of CCA etiology [33].
Current systemic treatment for iCCA The standard of care for patients with unresectable or metastatic disease is combination chemotherapy with gemcitabine and cisplatin, based on the ABC-02 and BT22 trials showing an improved median overall survival (mOS) with this combination compared to gemcitabine alone [38,39]. In patients with unresectable, liver-confined disease, liverdirected therapy with external beam radiation, radioembolization, chemoembolization or ablation can be considered [40].
If the disease progresses, second-line treatment with FOLFOX is the preferred regimen based on the ABC-06 trial findings, which demonstrated a mOS of 6.2 months for modified FOLFOX plus active symptom control versus 5.3 months for active symptom control alone [41]. The response rate of 5% and disease control rate (DCR) of 33% for patients in that study underline the urgent need for improvements in therapy for refractory patients with iCCA. Although the overall survival for iCCA treated with standard chemotherapy seems to be better than that for other BTCs [42], overall, systemic chemotherapy has a low survival benefit for patients with unresectable iCCA as the majority of patients have a chemorefractory course [43].
A recently published multicenter, randomized, double-blind, placebo-controlled phase III trial demonstrated the efficacy of the IDH1 inhibitor, ivosidenib, in a majority intrahepatic CCA study population [44]. It is anticipated that this will be licensed for second line in iCCA patients with an IDH1 mutation.

Targeting FGFR in iCCA
History of FGFR-targeted therapies in CCA Several candidate drugs targeting this pathway are under development, including nonselective and selective FGFR tyrosine kinase inhibitors (TKIs), anti-FGF/FGFR monoclonal antibodies, and FGF traps [45]. Although the non-selective TKIs pazopanib and ponatinib showed anecdotal anti-tumour activity in patients with iCCA harbouring an FGFR2 fusion [21], other preclinical and clinical trials have highlighted the pitfalls of using non-selective FGFR TKIs, including issues with off-target side effects [45]. The use of selective FGFR kinase inhibitors has therefore been a rational approach to address these issues. Several FGFR inhibitors have been evaluated in early phase clinical trials in patients with refractory iCCA harbouring FGFR2 gene rearrangements, either in trials specifically enrolling patients with iCCA or in trials evaluating a variety of advanced solid tumours harbouring FGFR2 gene rearrangements or other alterations ( Table 2 and Table 3). Derazantinib differs in that it is not a selective FGFR inhibitor, but rather a multi-kinase inhibitor with potent pan-FGFR activity [46].
All of the compounds discussed in the following section and shown in Table 2 and Table 3 bind reversibly to FGFR, with the exception of futibatinib which covalently binds to a highly conserved P-loop cysteine residue in the ATP pocket of FGFR (C492 in the FGFR2-IIIb isoform) [47]. The earliest reported data of selective FGFR inhibition in patients with CCA was with the oral agent infigratinib [48], while pemigatinib is the first FGFR-targeted agent to gain regulatory approval from the US FDA for use in previously treated patients with iCCA with FGFR2 fusions or rearrangements [5,49]. Note that in the following section, the FGFR-targeted agents of interest are presented and discussed in alphabetical order.

Debio 1347
Debio 1347 is an ATP-competitive, oral TKI with high selectivity for FGFR1-3 [50]. Nail toxicity: Nail toxicities also occur on FGFR inhibitors, especially with increased duration on treatment; most are grade 1 and 2, and grade 3 nail toxicity rarely occurs.

Future directions in targeting FGFR in iCCA
Resistance to FGFR kinase inhibitors in iCCA treatment Primary and acquired resistance limits the efficacy of FGFR inhibitors, similar to other TKIs in oncogene-driven cancers [72].
With respect to primary resistance, Silverman and colleagues describe a tendency towards a shorter progression-free survival amongst FGFR fusion patients with co-occurring tumour suppressor gene alterations including BAP1, CDKN2A/B, PBRM1 and TP53, although the numbers of patients do not allow any significant conclusions [73]. Assembly of datasets as we have greater clinical experience will be critical in describing the optimal genomic environment to predict benefit from treatment.
With respect to acquired resistance, Goyal  mutants, but remained relatively active against V565F compared to infigratinib and futibatinib. These results highlight the critical role of serial biopsy and ctDNA analysis to identify resistance mechanisms; this can guide selection of the next FGFR inhibitor for patients currently in the clinic, and also guide the development of the next generation of FGFR inhibitors beyond futibatinib. This study also showed that such a guided approach is feasible and effective in prolonging benefit for patients from FGFR inhibition in these FGFRdependent tumours [30].
Beyond the development of more effective FGFR inhibitors, combination strategies may also improve outcomes for patients with FGFR resistance in the setting of upregulation of alternative pathways in FGFR. Krook and colleagues showed via proteomic analysis of Incorporation of molecular testing within iCCA algorithm The approval of FGFR kinase inhibitors and the emergence of first line trials with these agents require the wider and potentially earlier ordering of molecular testing in iCCA. As discussed previously, the number of potentially actionable targets in iCCA is growing (e.g. FGFR fusions, IDH1/2 mutations, NTRK fusions), so the National Comprehensive Cancer Network (NCCN) recommends consideration of molecular testing for patients with unresectable and metastatic CCA [40].

The European Society for Medical Oncology (ESMO) Precision Medicine Working
Group has recommended that tumour multigene next generation sequencing (NGS) could be used to assess level I actionable alterations in advanced CCAs based on the ESMO Scale for Actionability of molecular Targets (ESCAT) criteria. Larger panels can be used only on the basis of specific agreements with payers taking into account the overall cost of the strategy (drug included), and if they report accurate ranking of alterations. RNA-based NGS can be used [77].
In the UK, genomic testing in the National Health Service (NHS) will be incorporated into the Genomic Laboratory Hubs of which there will be seven in the country. As targeted therapies, specifically FGFR inhibitors for iCCA, requiring genomic description become approved for standard of care, these centres will undertake standard of care profiling.

Combining FGFR inhibition with other therapy approaches
Although the clinical use of FGFR kinase inhibitors as monotherapy is still in an early stage, future trial results may support combination strategies using FGFRis with standard of care drug therapy options in solid tumours (see Supplemental information online, Table S1

Patient and Provider Education
Given the demonstrated promise of FGFR inhibitors in clinical trials, patients and their caregivers are also closely following developments in this area. Figure 2 illustrates the multiple issues that physicians and patients must address when considering targeted therapy.
Various barriers remain, for instance the availability of material, the accuracy and funding of the test, the availability and funding of the therapy, and finally, the toxicity and efficacy of the treatment. Despite these obstacles, the potential advantages of oral therapies are evident.
In addition to being an option a non-fusion patient would not receive, there would appear to be clear advantages of the FGFR inhibitors over chemotherapy with respect to toxicity, efficacy and quality of life, although data have yet to be generated. The increased complexity consequent on testing for FGFR alterations and treating with FGFR inhibitors does have resource implications that need to be addressed and acknowledged. Dr. Crolley declares no conflicts of interest that pertain to this work.

Financial support
Medical writing and editorial assistance were provided by Patrick Foley, PhD, of NexGen Healthcare (London, UK) and funded by Taiho Oncology, Inc.

Authors' contributions
Drafting of the manuscript, revision of the manuscript, and approval of the final version of the manuscript: JB, LG, SK and VEC.