The FGF/FGFR System in Breast Cancer: Oncogenic Features and Therapeutic Perspectives

Simple Summary The fibroblast growth factor/fibroblast growth factor receptor (FGF/FGFR) system represents an emerging therapeutic target in breast cancer. Here, we discussed previous studies dealing with FGFR molecular aberrations, the alterations in the FGF/FGFR signaling across the different subtypes of breast cancer, the functional interplay between the FGF/FGFR axis and important components of the breast microenvironment, the therapeutic usefulness of FGF/FGFR inhibitors for the treatment of breast cancer. Abstract One of the major challenges in the treatment of breast cancer is the heterogeneous nature of the disease. With multiple subtypes of breast cancer identified, there is an unmet clinical need for the development of therapies particularly for the less tractable subtypes. Several transduction mechanisms are involved in the progression of breast cancer, therefore making the assessment of the molecular landscape that characterizes each patient intricate. Over the last decade, numerous studies have focused on the development of tyrosine kinase inhibitors (TKIs) to target the main pathways dysregulated in breast cancer, however their effectiveness is often limited either by resistance to treatments or the appearance of adverse effects. In this context, the fibroblast growth factor/fibroblast growth factor receptor (FGF/FGFR) system represents an emerging transduction pathway and therapeutic target to be fully investigated among the diverse anti-cancer settings in breast cancer. Here, we have recapitulated previous studies dealing with FGFR molecular aberrations, such as the gene amplification, point mutations, and chromosomal translocations that occur in breast cancer. Furthermore, alterations in the FGF/FGFR signaling across the different subtypes of breast cancer have been described. Next, we discussed the functional interplay between the FGF/FGFR axis and important components of the breast tumor microenvironment. Lastly, we pointed out the therapeutic usefulness of FGF/FGFR inhibitors, as revealed by preclinical and clinical models of breast cancer.


Introduction
Breast cancer is the most frequently diagnosed malignancy and the second leading cause of cancer death in women worldwide [1]. The high incidence of this disease, the complex mechanisms involved in its progression, as well as the resistance to chemotherapy underline the need to better elucidate the molecular features fueling breast tumors. Over the past few years, substantial advances have been made in the discovery of new drugs for the treatment of this malignancy. The improved understanding of breast cancer heterogeneity has allowed the development of more effective and individualized therapeutic approaches [2]. Nevertheless, only a limited number of targeted therapies have been approved and are currently used in breast cancer patients. In this context, anti-human epidermal

FGFR Genetic Alterations in Breast Cancer
FGFRs' genetic alterations may contribute to tumor progression and poor outcomes in breast cancer patients. In this regard, an analysis performed in a cohort of 391 tumor patients has evidenced that FGF or FGFR aberrations are frequent in breast cancer (32.1%) and co-exist with other aberrations. For instance, univariate analysis identified in 15 genes certain anomalies associated with FGF/FGFR alterations [37]. A large next-generation sequencing analysis of approximately 5000 tumors showed that FGFR aberrations are harbored in about 7.1% of cancers [38]. Among the FGFR alterations, gene amplification has been found in an elevated percentage of cases (66%), however point mutations (26%) and chromosomal rearrangements (8%) have also been detected [38]. On the basis of the aforementioned observations and considering that many FGFR aberrations may lead to a gain-of-function, it has been hypothesized that FGFR inhibitors would be useful as anticancer agents [39].

FGFRs Gene Amplification
The amplification of FGFR1 represents the most frequent genomic alteration in breast cancer, whereas the amplification of the FGFR2-4 genes is less common [40]. In particular, the FGFR1 locus (8q12) has been found to be amplified in nearly 15% of hormone receptor (HR)-positive breast cancer and in approximately 5% of the aggressive triple-negative breast cancer (TNBC) [40]. The genomic analysis of The Cancer Genome Atlas (TCGA) and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) databases has confirmed that the amplification of FGFR1 is the highest among the FGFR family members, as it occurs in nearly 14% of breast cancer patients [9]. Furthermore, those patients with either a high expression of FGFR1 or copy number gain do exhibit reduced overall survival rates compared to the remaining cohort of patients [9]. It is worth mentioning that the amplification of FGFR1 has been correlated with aberrant ligand-dependent and ligand-independent signaling, which may lead to resistance to endocrine treatments in breast cancer [41,42]. In particular, breast cancer patients with the FGFR1 amplification frequently harbor activating alterations in the PIK3CA gene [38]. Likewise, FGFR1 has been found to be co-amplified with the CCND1 gene towards poor outcomes in estrogen-receptor (ER)+ breast cancer [23,43]. A recent study has also shown that genetic aberrations of FGFR, TP53, FLT1, and HER2+ may forecast the occurrence of brain metastases in breast tumor patients [44]. Among the FGFR aberrations, FGFR1 abnormalities were the most frequent and significantly higher in patients with brain metastases with respect to the remaining cohort of patients [44]. In addition, patients with ER+ metastatic breast cancer displayed the amplification of FGFR1 in circulating tumor DNA (ctDNA), hence mirroring the findings obtained in genomic studies of ER+ breast cancer tissues [45]. Together, these results indicate that the amplification of FGFR1 may represent an oncogenic driver contributing to breast cancer progression.
Cancers 2020, 12, 3029 4 of 20 cytoplasm, where they undergo lysosomal degradation or the recycling process, which are partially controlled by CbL-mediated monoubiquitylation or the LIS1/NDE1 complex [25,32]. FGFR signaling may be negatively regulated by MAPK phosphatase 3, Sprouty proteins, and SeF (similar expression to FGF, also known as IL17RD) family members [33][34][35] (Figure 1). As the down-regulation of the activated receptors is an important process in order to prevent altered signaling, a defective FGFR ubiquitination system and/or an error in the mitigation pathway could induce aberrant cell growth and malignant transformation [36]. The FGFR comprises three extracellular immunoglobulin-like subdomains (IgI, IgII, and IgIII), a transmembrane α-helix, and an intracellular bilobed cytoplasmic tyrosine kinase domain. IgI and IgII are separated by an acidic box. FGFs, which are secreted by tumor cells and/or the stromal compartment, bind to membrane-anchored FGFR monomers, then induce FGFR dimerization and the phosphorylation of the tyrosine kinase domains. The binding of FGFs to FGFRs is stabilized by heparan sulphate proteoglycans (HSPGs). Subsequently, the complex leads to the docking of adapter proteins and the activation of downstream pathways. For instance, ligand-stimulated FGFRs phosphorylate the FGFRassociated cytosolic docking protein FRS2. Once phosphorylated, FRS2 recruits Son of Sevenless adapter protein (SOS) and growth factor receptor-bound protein (GRB2) to activate rat sarcoma (RAS) and the downstream Rapidly Accelerated Fibrosarcoma (RAF) and MAPK/ERK Kinase (MEK) pathway. A different complex involves GRB2-associated binding protein 1 (GAB1), which recruits phosphoinositide 3-kinase (PI3K) that activates the Protein Kinase B (AKT) pathway. Phospholipase Cγ (PLCγ) hydrolyzes phosphatidylinositol-4,5-bisphosphate (PIP2) to phosphatidylinositol (3,4,5)triphosphate (PIP3) and diacylglycerol (DAG). PIP3 releases calcium, whereas DAG activates protein kinase C, which helps to strengthen the activation of the MAPK pathway by phosphorylating RAF in a RAS-independent manner. Other cascades may also be activated by FGFRs, such as the signal transducer and activator of transcription (STAT)-dependent signaling. The aforementioned effectors regulate, in turn, gene expression as well as cell proliferation, migration, invasion, and angiogenesis. FGF/FGFR signaling may be down-regulated by receptor internalization or through negative Figure 1. Schematic representation of aberrant FGF/FGFR signaling in breast cancer. The FGFR comprises three extracellular immunoglobulin-like subdomains (IgI, IgII, and IgIII), a transmembrane α-helix, and an intracellular bilobed cytoplasmic tyrosine kinase domain. IgI and IgII are separated by an acidic box. FGFs, which are secreted by tumor cells and/or the stromal compartment, bind to membrane-anchored FGFR monomers, then induce FGFR dimerization and the phosphorylation of the tyrosine kinase domains. The binding of FGFs to FGFRs is stabilized by heparan sulphate proteoglycans (HSPGs). Subsequently, the complex leads to the docking of adapter proteins and the activation of downstream pathways. For instance, ligand-stimulated FGFRs phosphorylate the FGFR-associated cytosolic docking protein FRS2. Once phosphorylated, FRS2 recruits Son of Sevenless adapter protein (SOS) and growth factor receptor-bound protein (GRB2) to activate rat sarcoma (RAS) and the downstream Rapidly Accelerated Fibrosarcoma (RAF) and MAPK/ERK Kinase (MEK) pathway. A different complex involves GRB2-associated binding protein 1 (GAB1), which recruits phosphoinositide 3-kinase (PI3K) that activates the Protein Kinase B (AKT) pathway. Phospholipase Cγ (PLCγ) hydrolyzes phosphatidylinositol-4,5-bisphosphate (PIP2) to phosphatidylinositol (3,4,5)-triphosphate (PIP3) and diacylglycerol (DAG). PIP3 releases calcium, whereas DAG activates protein kinase C, which helps to strengthen the activation of the MAPK pathway by phosphorylating RAF in a RAS-independent manner. Other cascades may also be activated by FGFRs, such as the signal transducer and activator of transcription (STAT)-dependent signaling. The aforementioned effectors regulate, in turn, gene expression as well as cell proliferation, migration, invasion, and angiogenesis. FGF/FGFR signaling may be down-regulated by receptor internalization or through negative regulators such as FGFR-like 1 (FGFRL1), with a similar expression to FGF (SEF), Sprouty (SPRY), Casitas B-lineage Lymphoma (CbL), and MAPK phosphatase 3 (MKP3).
The amplification of the FGFR2 locus (10q26) is a rare event in breast cancer, as it occurs in less than 1% of all cases [38]. However, the amplification and overexpression of FGFR2 have been detected in approximately 4% of TNBCs [46]. In FGFR2-amplified TNBC cells, a constitutive receptor activation highly sensitive to FGFR inhibition has been observed, suggesting that FGFR2 might be a potential therapeutic target in this occurrence [46][47][48]. In addition, high levels of FGFR2 have been found to be correlated with low overall and disease-free survival rates [49]. However, a further study performed on 148 TNBC samples reported that the FGFR2 amplification did not affect the survival of patients [50]. In this context, it should be mentioned that FGFR2 amplification may occur in normal breast tissue [20,[51][52][53], making therefore uncertain the consideration of FGFR2 as a potential therapeutic target. Regarding the FGFR3 locus (4p16), its amplification has been detected Cancers 2020, 12, 3029 5 of 20 in less than 1% of patients with breast cancer. In particular, analyses performed using the TCGA and METABRIC datasets have displayed the amplification of FGFR3 in five (0.5%) and nine (0.5%) cases, respectively [38,54]. A small frequency (2.3%) has been observed for the amplification of FGFR4 (5q35 locus) [38]. Yet, increased FGFR4 mRNA levels have been found in up to 30% of breast cancer patients; thus, new investigations are warranted in order to better assess such discrepancy occurring between the amplification of FGFR4 and its mRNA levels [55]. In this context, it should be noted that a retrospective study has assessed high FGFR4 levels independently associated with the therapeutic response and the survival of ER+ breast cancer patients treated with tamoxifen [56].

FGFRs Point Mutations
FGFR somatic mutations, which represent 26% of FGFR aberrations, are frequently observed within the receptor ligand binding and transmembrane domains [57]. Although rare in breast cancer, certain FGFR activating mutations are able to trigger an aberrant signaling in a ligand-independent manner towards an oncogenic activity exerted in this tumor [38]. The analysis of the Catalog Of Somatic Mutations In Cancer (COSMIC) database has identified a limited number of mutations of FGFRs in breast cancer samples. In particular, FGFR1 mutations have been assessed in 2 (S125L and K566R) out of 1031 samples, an FGFR2 mutation (R203C) has been shown in 1 out of 637 samples, and an FGFR4 mutation (V550E) has been identified in 1 out of 550 samples [7]. Certain FGFRs mutations with unknown functional significance have also been identified in patients with breast cancer [58]. In addition to the abovementioned data, several studies have hypothesized that germline Single Nucleotide Polymorphisms (SNPs) in the FGFR loci may play a role in breast cancerogenesis. In this vein, the Genome-Wide-Association-Studies (GWAS) has ascertained that some SNPs (rs2981582, rs1219648, rs2420946, rs2981579) within the FGFR2 intron 2 are associated with an increased breast cancer risk in postmenopausal women [59][60][61]. An SNP in the FGFR4 transmembrane region (rs351855), which involves the substitution of a glycine (G) into an arginine (R), has been associated with a poor prognosis in solid tumors, including breast cancer [62]. Accordingly, two recent studies performed in 1492 and 747 patients have suggested that the presence of the FGFR4-R388 allele confers an increase in breast cancer risk and correlates with lymph node metastasis and poor survival [63]. This FGFR4-associated SNP is important for breast cancer susceptibility, as it causes a receptor conformational change that allows functional interaction with the signal transducer and activator of transcription 3 (STAT3) [64]. Conversely, an FGFR1-associated SNP (rs17182023) has been linked to a low FGFR1 protein expression and a reduced breast cancer risk [65]. Certain genetic variants, such as FGFR4 rs1966265 and FGFR2 rs2981578, have also been involved in the outcome of breast cancer patients treated with chemotherapeutic agents [66]. Overall, the aforementioned findings are not strong enough to suggest that these SNPs may be considered as a therapeutic target in breast cancer patients.

FGFRs Gene Fusions
It is well known that fusion genes derived from the rearrangement of independent genes may trigger cancer development [67]. In this regard, it has been reported that several FGFR translocations may generate fusion proteins by combining the N-terminus part of a transcription factor and the TK domain of FGFR1-3. These chimeric proteins dimerize constitutively, allowing the activation of the TK domain [68,69]. For instance, a chromosomal rearrangement between the endoplasmic reticulum lipid raft-associated 2 (ERLIN2) and FGFR1 generates a chimeric protein, which displays a constitutive activation of the FGFR1 kinase domain, similar to that observed with the fusion of the AF4/FMR2 Family Member 3 (AFF3) with FGFR2 [68]. In breast tumor specimens and in SUM185PE TNBC cells, an oncogenic rearrangement similar to that reported in glioblastoma and bladder cancer has been found [54]. This chimeric gene is generated by the fusion of the coiled-coil domain of transforming acidic coiled-coil 3 (TACC3) with FGFR3. The presence of the coiled-coil domain of TACC3 enhances the dimerization of the fusion protein and activates FGFR3 tyrosine kinase, which in turn stimulates tumor Cancers 2020, 12, 3029 6 of 20 formation [54]. Further partners such as TACC1, TACC2, BAIAP2L1, BICC1, NPM1, PPAPDC1A, SLC45A3, and AHCYL1 should also be mentioned among the FGFR1-3 fusion genes leading to relevant biological actions [38,68].

Luminal A/B Breast Cancer Subtypes
Over the past two decades, several reports have suggested the existence of a functional crosstalk between FGFRs and the HR signaling pathway towards cancer progression and the failure of endocrine therapy [77]. For instance, FGF overexpression in ER+ breast cancer cells has been shown to confer estrogen-independent cell proliferation and metastatic spread [78]. FGFR1 amplification and overexpression have been frequently observed in ER+ breast tumors exhibiting an increased proliferation rate and reduced distant metastasis-free survival [42]. In this regard, it has been demonstrated that ER+ breast cancer cell lines harboring the amplification of FGFR1 rely on active FGFR1 signaling for anchorage-independent cell growth and resistance to endocrine therapy [42]. In addition, estrogen deprivation led to an increased expression of FGFR1 and FGF ligands in ER+/FGFR1-amplified primary tumors and breast cancer cells. In estrogen-free conditions, FGFR1 was associated with ER in tumor cell nuclei and regulated the transcription of ER-dependent genes. Likewise, the growth of ER+/FGFR1-amplified cells and PDXs was potently inhibited by the combination of drugs targeting both receptors [79]. Next, an adverse prognostic impact of FGFR1 expression has been assessed in the luminal A subtype of breast cancer [80]. In line with these results, FGFR1 amplification and overexpression have been shown to contribute to resistance to the CDK4/6 inhibitors used in combination with endocrine therapy in either in vitro or in vivo patient-derived xenograft models [23]. Recently, it was demonstrated that, in ER+ breast cancer cells, the activation of the FGFR1/3-STAT3 signaling pathway leads to resistance to the PI3K/mTOR inhibitor named NVP-BEZ235 [81]. Tomlinson and co-workers have also shown that the expression of FGFR3 is higher in a subset of tamoxifen-resistant ER+ breast cancers with respect to tamoxifen-sensitive ER+ breast tumors [82]. Likewise, the stimulation of FGFR3 has been found to trigger resistance to tamoxifen via the activation of the PLCγ signaling cascade [82]. It has been therefore suggested that FGFR3 may be considered as a promising target in the treatment of breast cancer patients exhibiting resistance to endocrine therapy [82]. It should be mentioned that, in breast cancer cells, the activation of the FGF7/FGFR2 axis may induce the proteasomal degradation of ER, preventing the inhibitory effects elicited by tamoxifen [83]. It is noteworthy that, in the analysis of tissue samples of patients with invasive ductal breast carcinoma, an inverse correlation between the expression of FGFR2 and ER was assessed, further supporting the aforementioned in vitro data [83]. Altogether, these findings suggest a potential association between FGFR overexpression and/or activation and acquired resistance to endocrine therapy in hormone-responsive breast cancers.

HER2-Enriched Subtype
An integrated molecular analysis of breast carcinoma has evidenced that HER2+ breast tumors harbor high expression levels of many receptor tyrosine kinases, including FGFR4 [84]. Experimental Cancers 2020, 12, 3029 7 of 20 studies using lapatinib-resistant HER2+ breast cancer cells have found the amplification and overexpression of the FGFR2 levels [75]. Furthermore, mouse mammary tumor models have revealed that combination therapies targeting FGF and ErbB receptors, including HER2 (ErbB2), may lead to synergistic anticancer effects [76]. In this context, amplified FGFR1 signaling has been included among the mechanisms involved in acquired resistance to lapatinib+trastuzumab co-targeted therapies in HER2-enriched breast cancers [6]. Overall, these results suggest the role prompted by aberrant FGFR signaling in HER2-driven tumor development, as well as the effectiveness of combination treatments targeting both pathways.

TNBC Subtype
Numerous studies have focused on the identification and characterization of the aberrant signaling pathways involved in the aggressive features of TNBCs [85,86]. In this vein, the amplification of the FGFR1 and FGFR2 gene loci has been described as recurrent in this breast cancer subtype [38,42]. For instance, a bioinformatic analysis based on publicly available datasets has unveiled that a subgroup of TNBC patients harbors the amplification of FGFR2 [46]. The same study has also shown that TNBC cells exhibiting FGFR2 overexpression display a high susceptibility to the apoptotic cell death prompted by the selective FGFR inhibitor PD173074 [46]. In accordance with these data, PD173074 induced either cell cycle arrest or apoptosis in a panel of breast cancer cell lines, including TNBC cells [74]. In addition, an analysis of the TCGA dataset performed on TNBC samples highlighted the association between the mRNA expression of FGFR1, FGFR2, and FGF2 and the low methylation status of the respective promoter sequences [50].

The Functional Interplay between FGF/FGFR Signaling and Breast Tumor Stroma
Breast cancer evolution is driven not only by tumor-cell intrinsic factors but rather relies on an intricate connection between cancer cells and the surrounding stroma. In this scenario, it should be noted that the tumor stroma accounts for various cell types, such as fibroblasts, endothelial and immune cells, and adipocytes embedded in a highly dynamic glycoprotein-rich extracellular matrix [87]. Together, this complex system represents a major hurdle to better understand the multiple mechanisms hidden behind the therapeutic failure [88,89]. Therefore, to better appreciate the role elicited by FGF/FGFR signaling in breast cancer progression, unraveling its contribution to the functional interplay occurring among the key players within the tumor microenvironment may represent a primary challenge.

Cancer-Associated Fibroblasts
Cancer-associated fibroblasts (CAFs), which are abundantly present in the tumor stroma, may facilitate cancer cell proliferation, migration, invasion, and angiogenesis [90][91][92]. In this respect, it should be pointed out that CAFs and their secretome (i.e., chemokines, cytokines, and growth factors) exert an important role in the acquisition of certain malignant features [93]. Of note, several studies have recognized that CAFs are a source of FGF ligands, which may regulate multiple processes acting in an autocrine and/or paracrine manner [90,94]. In this regard, the CAFs derived from hormone-independent breast cancers displayed higher levels of FGF2 than CAFs derived from hormone-dependent breast tumors [95]. It is noteworthy that the FGF2 secreted by CAFs triggered PR-mediated responses in tumor cells towards the stimulation of cell proliferation [95]. Next, CAFs may alter the beneficial effects of endocrine therapy through the involvement of the FGF/FGFR system. For instance, the overexpression of FGF4 has been shown to promote the transition from estrogen-dependent to estrogen-independent tumor growth and to trigger a metastatic phenotype in xenograft mouse models [96]. Besides this, the release of FGF7 by CAFs and its interaction with the cognate receptor FGFR2 have been shown to induce ER phosphorylation, ubiquitination, and subsequent proteasomal degradation, therefore counteracting the endocrine treatment in breast cancer cells [83]. Besides this, the FGF/FGFR axis has been involved in a feedforward stimulatory loop coupling CAFs to breast cancer cells [97,98], whereas the PDGF released by breast tumor cells stimulated the production of FGFs by CAFs towards the activation of proliferative and pro-metastatic pathways in breast cancer cells [97]. Likewise, the Hedgehog ligand produced by TNBC cells prompted the release of FGF5 by CAFs, leading to the acquisition of a chemo-resistant and cancer stem cell phenotype [99]. Recently, estrogens have been shown to mediate FGF2/FGFR1 paracrine activation through the G protein estrogen receptor (GPER), which coupled CAFs to breast cancer cells towards migratory and invasive tumor cell responses [9]. Together, these findings may extend previous studies showing that diverse FGF ligands are also secreted by normal fibroblasts [100].

Endothelial Cells
Angiogenesis is an essential process in tumor development and metastatic spread, as it constantly supplies tumor mass with oxygen and nutrients [101]. In this context, several reports have provided evidence that the endothelial cells of the surrounding tumor stroma express FGFRs that mediate the stimulatory action of FGFs [102,103]. Indeed, FGF/FGFR signaling has been demonstrated to act as an angiogenic driver, given that certain FGFs may be able to stimulate the formation of new blood vessels [28]. Additionally, the abnormal activation of the FGF/FGFR system has been reported to promote resistance to anti-VEGF therapy [104]. In this regard, preclinical data have shown an increased FGF2 expression in breast tumors treated with antiangiogenic agents, whereas a suitable antitumor activity was obtained, inhibiting both the VEGFR and FGFR-mediated transduction pathways [105].

Tumor-Infiltrating Immune Cells
Macrophages, neutrophils, myeloid-derived suppressor cells (MDSCs), and T and B lymphocytes mainly contribute to the immune tumor infiltrate, which plays an integral role in the evolution of breast cancer [106]. For instance, MDSCs that are highly up-regulated in pathological conditions such as the chronic infections and cancer may exhibit a strong immunosuppressive activity [107]. In this context, the infiltration of MDSCs during mammary tumorigenesis has been found to be decreased upon the FGFR inhibitor BGJ398 in a transgenic mouse model [108]. Likewise, the FGFR inhibition by AZD4547 reduced breast tumor-infiltrated MDSCs as well as their levels in the peripheral blood and the spleen [109]. The use of AZD4547 also prompted the recruitment of CD4(+) and CD8(+) T-cells in the tumor stroma and the spleen [109]. Noteworthy, FGFR inhibitors induced the immune modulations at least in part targeting the cytokines produced by CAFs, though the mechanisms involved remain to be fully understood [20]. To date, the action of the FGF/FGFR axis has been investigated in tumor-associated macrophages due to their peculiar action exerted in breast cancer [110]. For instance, it has been shown that the functional interplay linking the activation of FGFR1 in epithelial cells to the chemokine expression in macrophages (i.e., CXCL1 and CXCL5) may lead to mammary tumor formation and progression [111]. Moreover, FGFR1 activation in breast tumor cells promoted the release of inflammatory chemokine CX3CL1, which in turn stimulated the migration of macrophages during the initial stages of mammary tumor formation [112]. Next, the recruitment of macrophages and the angiogenic process mediated by FGFR1 were halted by the inhibition of CX3CR1 [112].

Cancer-Associated Adipocytes
Within the breast tumor microenvironment, relevant biological responses are triggered through the release of diverse mediators by cancer-associated adipocytes. For instance, the crosstalk between adipocytes and cancer cells may prompt functional changes in both cell types, thus supporting tumor progression [113]. In the diet-induced obesity xenograft model (DIOX), the expansion of the adipose tissue stimulated cancer progression through the production of FGF1 by the adipocytes and the crosstalk between FGFR and ER in the nearby breast cancer cells [114]. Of note, the weight gain and the adipocyte diameter positively correlated with (i) the levels of FGF1, (ii) FGFR1 activation in tumor cells, and (iii) estrogen-independent tumor growth [114]. It has also been reported that obese patients harbor an increased systemic concentration of FGF2 and exhibit resistance to anti-VEGF treatments [115]. Moreover, in a breast tumor mouse model obesity was associated with an increased release of FGF2 by adipocytes, whereas either the antidiabetic drug metformin or the inhibition of FGFR decreased the vessel density and restored the tumor sensitivity to anti-VEGF agents [115].
Altogether, these results indicate that the FGF/FGFR signaling may play a role in the functional interplay between cancer cells and the surrounding stroma towards breast cancer development, as recapitulated in Figure 1.

Therapeutic Targeting of the FGF/FGFR Axis in Breast Cancer
On the basis of the aforementioned findings, diverse therapeutic approaches have been recently developed in order to target the FGF/FGFR transduction pathway [116][117][118]. In addition, different clinical trials have been performed. Nevertheless, FGFR-targeted therapies have not been approved up to now for the treatment of breast cancer [119]. To date, the current pharmacological strategies involving FGFRs may be grouped into three main categories according to the agents used, as reported in Table 1: (i) small-molecule multi-targeting Tyrosine Kinase Inhibitors (TKIs) that block the tyrosine kinase activity; (ii) small-molecule TKIs that selectively target the kinase domain of FGFRs; (iii) monoclonal antibodies (mAbs) that block FGFRs as well as entrapping their ligands.

Multi-Targeting TKI
Non-selective TKIs include diverse compounds targeting the tyrosine kinase domains of FGFR, VEGFR, and PDGFR, as these receptors are related phylogenetically and display a high homology. Some TKIs showed promising results regarding their safety and tolerability in early clinical trials [117]. For instance, nanomolar concentrations of Dovitinib (TKI258) inhibited a panel of RTKs, including VEGFR1-3, FGFR1-3, and PDGFRβ, showing an anti-cancer activity in preclinical models of breast tumor [76,120]. Dovitinib was then evaluated in association with fulvestrant in postmenopausal patients with HER2-enriched or ER+ breast cancer. Despite the promising clinical activity, this trial was ended early due to the slow enrollment of the FGF-amplified subgroup (NCT01528345) [121]. The non-selective inhibitor named Lucitanib (E3810), which blocks the activity of VEGFR1-3 and FGFR1-2, has been tested in a phase II clinical study involving metastatic breast cancer patients with or without FGFR1 amplification. However, the results of this clinical trial are not yet available (NCT02202746). The small-molecule Lenvatinib (E7080), which targets FGFR1-4, VGFR1, PDGFR, RET, and KIT, suppressed lymph node and lung metastases in a breast cancer xenograft model [122]. Currently, a phase II clinical trial regarding the use of Lenvatinib for the treatment of early-stage ER+ breast cancer patients is ongoing (NCT03168074). Next, the potent dual inhibitor-namely, 3D185, which targets FGFR1-3 and colony stimulating factor 1 receptor (CSF1R)-showed promising anti-tumor effects in FGFR-dependent and macrophage-dominant cancer models [123]. Interestingly, this study emphasized the usefulness of targeting simultaneously tumor cells and components of the cancer microenvironment in order to achieve synergistic beneficial effects [123].

Selective FGFR-TKI
Selective FGFR agents have been developed to appreciate the on-target FGFR inhibition in patients harboring FGF/FGFR abnormalities. In this regard, the pan-FGFR inhibitor Infigratinib (BGJ398) was first evaluated to determine its maximum tolerated dose and the pharmacokinetic and pharmacodynamics properties (NCT01004224) [124]. Thereafter, the combination of BGJ398 with the PI3Kα inhibitor BYL719 was investigated in patients whose advanced or metastatic tumors express PIK3CA mutations with or without FGFR1-3 alterations (NCT01928459) [125]. However, no clear evidence of synergistic activity was observed in the cohorts of patients harboring dual PIK3CA and FGFR1-3 aberrations [125]. The FGFR1-3 inhibitor named AZD4547 was tested in a phase II clinical trial in patients with advanced tumors, including HER2-negative breast cancer, non-small cell lung cancer, and gastroesophageal carcinomas harboring FGFR1/2 amplification. In this vein, treatment with AZD4547 has shown a higher antitumor activity in FGFR2-amplified gastroesophageal cancers with respect to FGFR1-amplified breast tumors (NCT01795768) [126]. The therapeutic response to AZD4547 used in combination with anastrozole or letrozole was evaluated in a phase II clinical study in ER+ breast cancer patients, however the results are not yet known (NCT01791985). Further clinical studies are ongoing in order to determine the effectiveness of other selective FGFR inhibitors. For instance, the small molecule Debio-1347 is currently used in combination with fulvestrant in a phase II clinical study aiming to assess the response in ER+ metastatic breast cancer with FGFR amplification (NCT03344536). Next, a phase II clinical trial has been set in order to appreciate the efficacy of TAS-120 used alone or in association with fulvestrant in patients with advanced or metastatic breast tumor harboring FGFR amplification (NCT04024436). In addition, the pan-FGFR inhibitor Erdafitinib, used in combination with fulvestrant and palbociclib, is currently under evaluation in a phase I clinical study to assess the anti-tumor activity in patients with ER+/HER2-/FGFR-amplified metastatic breast cancer (NCT03238196).

Ligand Trap and Antibodies
Although considerable efforts have been recently addressed to the development of drugs targeting the FGFR kinase activity, further approaches evaluating the action of monoclonal antibodies (mAbs) as therapeutic agents have been considered. For instance, mAb GP369 recognizes the FGFR2-IIIb isoform -nhibited tumor burden both in vitro and in vivo models [127]. The effectiveness of the mAbs FPA144 and MFGR1877S that block, respectively, FGFR2 and FGFR3 and the usefulness of the soluble decoy receptor named FP-1039 that sequesters FGFs and inhibits FGFR signaling have also been ascertained [128,129]. Whilst these mAbs have shown promising anti-tumor effects in advanced solid tumors, including breast cancer, their clinical potential has been only partially explored [20]. FP-1039, which has been evaluated in a phase I clinical trial including breast cancer, has shown promising results in terms of its tolerability and low toxicity with respect to FGFR-TKI (NCT00687505) [130].
Overall, anti-FGF/FGFR agents used alone and/or in combination with established treatments have displayed an intriguing potential to be considered in comprehensive therapeutic approaches in breast cancer.

Conclusions and Future Perspectives
The choice of a targeted therapy for each patient is currently made by analyzing the expression and/or the genomic status of a specific molecular target. Nevertheless, the heterogeneity of breast tumor cells frequently makes this decision challenging. In this context, it should be mentioned that numerous studies performed both in vitro and in vivo have evidenced that aberrant FGF/FGFR signaling may contribute to the onset and progression of breast tumor [117]. Furthermore, the overexpression and/or abnormal activation of FGFRs have been frequently associated with an acquired resistance to well-established anti-cancer therapies [20,39]. Moreover, diverse studies have documented that the FGFR signaling cascade may contribute to estrogen-independent cell growth in the endocrine-resistant breast cancer subset [41,42,[77][78][79].
Several reports have also highlighted the role of the FGF/FGFR system in the functional liaison between breast cancer cells and the surrounding tumor stroma [20,21]. Preclinical and clinical studies have indeed provided important hints on the feasibility of targeting FGFR in a novel anti-tumor strategy, at least in the subset of breast cancer patients harboring FGFR abnormalities [23,76,116,121,123]. On the other hand, the clinical trials described may evidence that considerable attention has been recently shifted to the usefulness of anti-FGFR drugs in combination with other chemotherapeutic agents [131]. Building up more knowledge of the FGF/FGFR pathway, these findings may represent the main goal in further studies aimed at addressing innovative anti-cancer strategies. In this vein, it would be critical to identify a sub-population of patients harboring peculiar molecular alterations associated with the dysregulation of the FGF/FGFR pathway towards novel co-targeting therapeutic strategies. Likewise, it remains crucial to better dissect the mechanisms involved in the action of FGF/FGFR signaling in breast cancer development in order to achieve therapeutic advances and overcome treatment hindrances. For instance, some FGFR alterations are potential targets for monotherapy, such as FGFR1-2 amplification or FGFR3 fusion genes. Other FGFR aberrations do not seem to be suitable biomarkers, hence remaining to be carefully evaluated in breast cancers patients.
Compelling evidence has suggested that intra-and inter-tumoral heterogeneity should be taken into careful consideration for an adequate selection of patients [2,132]. In this respect, liquid biopsy allowing the identification of circulating tumor cells (CTCs) in the peripheral blood and the sequencing of the tumor DNA extracted from CTCs would provide the genetic profile of the malignancy [133].
As several studies have unveiled comparable mutation frequencies between the genomic profile of breast cancer biopsies and ctDNA [134][135][136], the assessment of the FGFR aberrations in ctDNA might be a promising strategy to monitor the sensitivity or resistance to therapies, hence facilitating patient-tailored precision medicine [45,137,138].
Overall, we are only at the beginning of truthfully appreciating the role played by the FGF/FGFR transduction pathway in breast malignancy, as well as how it may be halted in a suitable manner. Thus, further studies are warranted in order to gain valuable insights into the treatments targeting the FGF/FGFR system in the diverse subpopulation of breast cancer patients.