Therapeutic Targeting of Fibroblast Growth Factor Receptors in Gastric Cancer

Chemotherapy has become the global standard treatment for patients with metastatic or unresectable gastric cancer (GC), although outcomes remain unfavorable. Many molecular-targeted therapies inhibiting signaling pathways of various tyrosine kinase receptors have been developed, and monoclonal antibodies targeting human epidermal growth factor receptor 2 (HER2) have become standard therapy for HER2-positive GC. An inhibitor of vascular endothelial growth factor receptor 2 or MET has also produced promising results in patients with GC. Fibroblast growth factor receptors (FGFR) play key roles in tumor growth via activated signaling pathways in GC. Genomic amplification of FGFR2 leads to the aberrant activation found in GC tumors and is related to survival in patients with GC. This review discusses the clinical relevance of FGFR in GC and examines FGFR as a potential therapeutic target in patients with GC. Preclinical studies in animal models suggest that multitargeted tyrosine kinase inhibitors (TKIs), including FGFR inhibitor, suppress tumor cell proliferation and delay tumor progression. Several TKIs are now being evaluated in clinical trials as treatment for metastatic or unresectable GC harboring FGFR2 amplification.


Introduction
Gastric cancer (GC) is the second leading cause of cancerrelated mortality, with 738,000 deaths per year [1]. Median overall survival was only 10 to 13 months in patients with metastatic or unresectable GC who received combined chemotherapy with cytotoxic agents [2][3][4].
Aberrant or oncogenic activation of receptor tyrosine kinase (RTK) is involved in carcinogenesis or tumor progression. Inhibition of signaling pathways of RTK is most intensively pursued as an anticancer target. Trastuzumab, a monoclonal antibody against human epidermal growth factor receptor 2 (HER2/ERBB2), was the first RTK-targeting agent approved for the indication of unresectable or metastatic GC worldwide [5]. However, several agents targeting epidermal growth factor receptor (EGFR) provided no additional benefits in clinical trials [6][7][8]. Bevacizumab, a monoclonal antibody targeting vascular endothelial growth factor-(VEGF-) A, which activates VEGF receptor-(VEGFR-) 1 and VEGFR-2, provided significant benefits in terms of progression-free survival (PFS), but not overall survival (OS) [9]. Ramucirumab is a monoclonal antibody targeting the extracellular domain of VEGFR-2. Ramucirumab as secondline chemotherapy prolonged overall survival [10,11] and was recently approved for the indication of unresectable or metastatic GC. Rilotumumab is a monoclonal antibody designed to inhibit binding of HGF to c-MET. Its additive effect was clinically significant in GC with high c-MET expression [12].
Fibroblast growth factor receptors (FGFRs) are one of the RTK families that belong to the immunoglobulin (Ig) superfamily [13]. Binding of fibroblast growth factors (FGFs) with high-affinity to FGFR results in kinase activation of downstream signaling pathways. The FGFR family consists of 5 receptors, named FGFR1 to FGFR5. The extracellular regions of FGFRs comprise 3 extracellular Ig-like domains (I-III), a single transmembrane domain, and the cytoplasmic tyrosine kinase domains TK1 and TK2. However, FGFR5 lacks an intracellular tyrosine kinase domain. The extracellular Ig-II and Ig-III domains are the FGF ligand-binding sites. Alternative splicing of Ig-III occurs in FGFRs 1-3, creating IIIb and IIIc variants of the receptors with diverse ligandbinding specificities that are expressed in a tissue-specific manner [14][15][16]. Binding of FGFs to FGFRs induces receptor dimerization, conformational changes within the FGFR structure, and phosphorylation of tyrosines in the intracellular part of the receptor, including the kinase domain and the C-terminus [17]. Subsequent downstream signaling is activated in two main pathways via the intracellular receptor substrates FGFR substrate 2 (FRS2) and phospholipase Cg, leading ultimately to upregulation of the Ras-dependent mitogenactivated protein kinase (MAPK)/extracellular signal-regulated kinase (ERK) and Ras-independent phosphoinositide 3kinase (PI3K)/Akt signaling pathways [18]. The other signaling pathway, dependent on signal transducer and activator of transcription (STAT), is activated by FGFRs [14].

Clinical Analysis of Expression or Genomic Alteration of FGFR in GC
The results of immunohistochemical analyses of FGFRs are summarized in Table 1. We previously showed that protein overexpression of FGFR1, FGFR2, and FGFR4 is significantly associated with tumor depth, lymph-node metastasis, tumor stage, and poorer survival in GC, while FGFR3 is not [19]. Others have shown that overexpression of K-sam, a FGFR2 homologue, is significantly related to pathologically undifferentiated or diffuse-type GC [20,21]. Nagatsuma et al. reported that FGFR2 overexpression is significantly associated with tumor depth, lymph-node metastasis, and tumor stage in a larger analysis [22]. Moreover, patients with FGFR2 overexpression had a significantly higher incidence of peritoneal or lymph-node recurrence and a significantly shorter survival than those without FGFR2 overexpression. Ye et al. showed that FGFR4 is not associated with any clinicopathological factors or with survival, although patients with far advanced GC and FGFR4 overexpression had significantly worse survival [23]. The mRNA expression of FGFR1, FGFR2, or FGFR4 was upregulated in GC as compared with that in normal tissues, although FGFR3 mRNA was barely detectable in normal as well as cancer tissue [24]. Studies of FGFR genomic alterations are summarized in Table 2. FGFR2 amplification is a well-known phenomenon in GC. The frequency of FGFR2 amplification on comparative genomic hybridization had been reported to be 7% (2 of 30) in GC in one study and 16% (3 of 19) in diffuse-type GC in another [25,26]. In a study using Southern blot analysis, the frequency of FGFR2 amplification was 5% (3 of 57) [27]. Betts et al. reported that FGFR2 amplification was detected on fluorescence in situ hybridization (FISH) analysis in 1.8% (3 of 171) of GCs, and survival was very poor in three patients who had tumors with FGFR2-amplification [28]. In a study by Jung et al., FGFR2 amplification was detected on FISH in 4.5% (14 of 313) of GCs and was significantly associated with the depth of tumor invasion, lymph-node metastasis, distant metastasis, tumor stage, and poorer survival [29]. In that study, FGFR2 amplification was not detected in papillary or well-differentiated subtypes of GC. Das et al. reported that FGFR2 amplification was found in 7.3% (10 of 137) of patients, while FGFR2 deletion was detected in 5.8% (8 of 137), and patients with FGFR2-amplified GC had worse survival than those with FGFR2-deleted GC [30]. Interestingly, they showed that not only FGFR2 amplification but also deletion was more common in undifferentiated type than in differentiated type. In an international multicenter study using FISH, the presence of FGFR2 amplification did not differ appreciably among three countries: 7.4% (30 of 408) in the UK, 4.6% (9 of 197) in China, and 4.2% (15 of 356) in  Korea [31]. In each country, patients with FGFR2-amplified GCs had worse survival than those with nonamplified GCs. In addition, 24.1% of FGFR2-amplified GCs displayed intratumoral heterogeneity within multiple samples extracted from the same tumors on tissue microarray analysis. In the FISH studies mentioned above, FGFR2 gene amplification was determined on the basis of the presence of tight signals of FGFR2 clusters or a ratio of FGFR2 signals to chromosome enumeration probe-10 signals of 2.0 or higher. Matsumoto et al. reported that FGFR2 amplification on copy number assay (more than 5 copies) was detected in 4.1% (11 of 267) of GCs, whereas amplification of other FGFRs was not detected [32]. Patients with FGFR2-amplified tumors had slightly but not significantly shorter survival than those with nonamplified tumors. In a comprehensive survey assessing genomic alterations of GCs by high-resolution single nucleotide polymorphism arrays, FGFR2 amplification was detected in 9.3% (18 of 193) of GCs, and coamplification of FGFR2 with EGFR, ERBB2, KRAS, or MET was rarely detected [33]. In that study, the overall survival of patients with FGFR2 amplification did not differ from that of patients with nonamplification, although the survival of patients with high mRNA expression of FGFR2 was significantly worse than that of patients with low mRNA expression of FGFR2 in the extended population. Wang et al. detected FGFR2 amplification in 3.0% (3 of 100) of GCs on single nucleotide polymorphism (SNP) genotyping arrays [34]. In addition, two mutations of FGFR2 were identified in GC: a missense in exon IIIa and a splice site mutation in exon IIIc [35].
Among other FGFR genes, Ye et al. investigated the SNP of FGFR4 (Gly388 to Arg388) in GC samples and showed that 45% (46 of 103) of patients were heterozygous and 13% (13 of 103) homozygous for Arg388 allele [36]. Patients with tumors in which FGFR4 Arg388 allele was found had significantly shorter survival, and the presence of FGFR4 Gly388Arg allele was an independent prognostic factor. FGFR1 somatic mutation on whole-exome sequencing was detected in 1.1% (1 of 87) of diffuse type GCs and in 3.9% (2 of 51) of intestinal type GCs [37].
Small-molecule compounds fitting into the ATP-binding pockets of RTKs have been developed as anticancer drugs. PD173074 is a reversible inhibitor of FGFR and VEGFR. PD173074 blocks FGF2-induced angiogenesis [42] and also blocks mitogenesis of tumor cells via G1-arrest mediated by downregulation of cyclin D1 and cyclin D2 [43]. Treatment with PD173074 selectively and potently inhibited growth of FGFR2-amplified GC cell lines (KATOIII, SNU-16, and OCUM-2M), leading to a strong decrease in tumor cells in S phase accompanied by an increase in tumor cells in the sub-G1 population [38]. In addition, prominent induction of poly(ADP-ribose) polymerase, a marker of caspase activation associated with apoptosis, was observed after treatment. EGFR family kinases might have been downstream targets of amplified FGFR2 in that study, because the increased expression of phosphorylated HER receptors was dependent on FGFR2. PD173074 was more effective in FGFR2-amplified GC cell lines (SNU-16, TU-KATOIII, HSC-43, and HSC-39) than in nonamplified cell lines (OCUM1, IM95, 58Aa1, and 44As3) on growth inhibition assays [32].
Cediranib (AZD2171) is also a broad-range tyrosine kinase inhibitor (TKI) and inhibits FGFR, VEGFR, PDGFR, and KIT, as well as VEGF-induced proliferation of human endothelial cells [45]. Cediranib completely inhibited the phosphorylation of FGFR2 and downstream targets, including FRS2, Akt, and MAPK, in GC cell lines (KATO-III and OCUM2M) that strongly expressed FGFR2-IIIb mRNA, and then significantly and dose-dependently inhibited tumor growth in KATO-III and OCUM2M tumor xenografts [46]. AZD4547 is a highly selective and potent ATP-competitive TKI of FGFR1-3 and inhibited recombinant FGFR kinase activity in vitro and suppressed FGFR signaling and growth in tumor cell lines with deregulated FGFR expression [47]. After treatment of GC cell-lines (SNU-16 and KATO III) with AZD4547, expression levels of phosphorylated FGFR2 and its downstream signaling molecules, such as phospholipase C-gamma, FRS2, ERK, and S6, were all reduced [48]. Furthermore, treatment with AZD4547 also dose-dependently increased the sub-G1 population of GC cells. AZD4547 inhibited tumor regression in FGFR2-amplified xenografts (SNU-16) but not in nonamplified models (AZ521 and MGC803) in that study. In addition, antitumor efficacy was enhanced in vivo by combined chemotherapy with AZD4547 plus chemotherapeutic agents as compared with monotherapy.
Ponatinib (AP24534) was designed with a carbon-carbon triple bond to accommodate the T315I mutation in the ABL kinase domain [49]. Ponatinib potently inhibits the kinase activity of FGFR1-4 and had higher inhibitory activity in GC cells with FGFR2 amplification than did other FGFR inhibitors and inhibited the growth of SNU-16 xenograft tumors [50]. In addition, ponatinib potently inhibited cell proliferation and signaling in several cell lines of other cancers with FGFR mutation. S49076, a potent inhibitor of FGFR1-3, MET, and AXL, inhibited the autophosphorylation of those receptors and the phosphorylation of FRS2 [51]. S49076 inhibited viability in SNU-16 cell lines and tumor growth in SNU-16 xenografts. Combined treatment with S49076 and bevacizumab, a VEGF inhibitor, enhanced the antitumor effect in other cancer xenografts.
Small interfering RNA (siRNA), the intermediate product of the pathway of RNA interference, plays a key role in RNA silencing treatment. Silencing of FGFR expression by treatment with siRNA led to inhibition of proliferation and promotion of apoptosis accompanied by a reduction in VEGFR expression and a rise in caspase-3, an apoptosis-related protein, in an in vitro study [52]. In experimental in vivo studies using GC cells (MGC80-3), siRNA also suppressed the expression of FGFR and enhanced tumor shrinkage [52].
MicroRNAs (miR) negatively regulate protein expression by binding to protein-coding mRNAs and inhibiting translation. The 3 UTR of FGFR1 mRNA contains two putative binding sites of miR-133b [53]. Therefore, miR-133b reduced the protein expression of FGFR1. Furthermore, upregulation of FGFR1 expression was found to negatively correlate with miR-133b expression in several GC lines and GC tissues.

Clinical Trials of FGFR-Targeted Treatment in GC
Clinical trials of FGFR inhibitors for GC are summarized in Table 3. Several phase II trials of FGFR inhibitors are ongoing in GC. Dovitinib was evaluated in a phase I study of 35 solid tumors including 2 GCs [54]. Enrolled patients were treated in four intermittent (25-100 mg/day) and three continuous (100-175 mg/day) dosing cohorts. Dose-limiting toxicities were grade 3 hypertension in one patient in the 100 mg continuous dosing cohort, grade 3 anorexia in a second patient at 175 mg, and grade 3 alkaline phosphatase elevation in a third patient at 175 mg. Unfortunately, neither patient with GC had stable disease for more than 4 months in this study. Nonetheless, three phase II studies of dovitinib are ongoing in GC. Dovitinib is being assessed as salvage monotherapy after failure of first-or second-line chemotherapy in patients with advanced or metastatic scirrhous GC in one study [55] and in patients with GC harboring FGFR2 amplification in another study [56]. Dovitinib was administered orally at 500 mg/day on days 1 to 5 of 7-day repeated cycles in both studies. In the third study, divided into phase I and phase II, dovitinib is  being assessed in combination with docetaxel as second-line chemotherapy in patients with GC [57]. A phase II study of AZD4547, an oral TKI of both FGFR and VEGFR, is also ongoing to assess the efficacy and safety of AZD4547 monotherapy versus paclitaxel in patients with locally advanced or metastatic GC associated with FGFR2 polysomy or amplification [58]. AZD4547 was administered orally at 160 mg/patients on days 1 to 14 of a 21-day cycle.

Results of Clinical Trials of FGFR-Targeted Treatment in Various Cancers
Phase III clinical trials in patients with other types of cancer are shown in Table 4. Cediranib (AZD2171) is an oral TKI of both FGFR and VEGFR. In one study of colorectal cancer, the addition of cediranib to standard first-line chemotherapy significantly prolonged PFS but not OS [59]. In the other study, the noninferiority of cediranib did not reach the predefined level of PFS as compared with bevacizumab [60]. No synergistic effect of cediranib was found in patients with nonsmall-cell lung carcinoma (NSCLC) [61]. Brivanib (BMS-582664) is an oral TKI of both FGFR and VEGFR, and the addition of brivanib increased toxicity and did not improve OS as compared with cetuximab alone in patients with colorectal cancer with wild-type KRAS [62]. In addition, no significant effect of brivanib was found in unresectable hepatocellular carcinoma [63,64]. Dovitinib (TKI258) is an oral multitargeted TKI, including FGFR, and was not superior to sorafenib in metastatic renal cell carcinoma [65]. Nintedanib (BIBF1120) is an oral TKI of FGFR, VEGFR, PDGFR, FLT-3, and lymphocyte-specific protein tyrosine kinase and significantly prolonged PFS in combination with docetaxel in patients with NSCLC [66].

Conclusions
Aberrant activation of FGFR signaling pathway, especially FGFR2 amplification, is related to disease progression or poor survival in GC; thus FGFR-targeted therapy is considered promising. Unfortunately, the superiority of multitargeted TKIs, including those with FGFR inhibitory activity, to standard chemotherapy has not been demonstrated in most phase III clinical trials in other malignancies. However, TKIs were evaluated as VEGFR inhibitors, but not FGFR inhibitors, in those studies. FGFR inhibitors were shown to have higher antitumor activity against FGFR2-amplified tumors than against nonamplified tumors in preclinical studies [32, 33, 38, 40-42, 46, 48, 50, 51]. Therefore, ongoing clinical trials of dovitinib or AZD4547 in patients with FGFR2-amplified GC are expected to show positive results. Scirrhous gastric cancer is known to be refractory to intensive treatment and to carry a poor prognosis; however, FGFR2 amplification is found in cell lines originating from scirrhous GC, such as KATO-III, SNU-16, and OCUM-2M. FGFR inhibitors may be a promising treatment for scirrhous GC and are now being evaluated in clinical trials. On the other hand, intratumoral heterogeneity of FGFR2 amplification has been found in GC samples [32]. Intratumoral heterogeneity of HER2 was also detected in GC, and the expression levels of primary lesions may not be consistent with those of metastatic sites. Intratumoral heterogeneity can be a critical issue for a single molecular-targeted treatment [67]. Amplification of other FGFRs has not been found in GC; however, overexpression of FGFR1 and FGFR4 or single nucleotide polymorphism of FGFR4 appears to be associated with tumor progression or survival [19,23,36]. Preclinical studies evaluating other FGFRs in GC remain scant. FGFR2 amplification was detected in only 1.8% to 7.3% of patients with GC, regardless of ethnic group; therefore, only a small subgroup of patients with GC can potentially benefit from FGFR2-targeted therapy alone. The development of FGFR inhibitors against tumors with overexpression not only of FGFR2 but also of FGFR1 or FGFR4 is likely to enhance potential treatment benefits in patients with GC.