Emerging Molecular Therapeutic Targets for Cholangiocarcinoma

FGFR Gene Fusions and Inhibitors

FGF signaling regulates a multitude of biological processes including cell proliferation, differentiation, survival, wound repair, angiogenesis, and migration (13). The integral role of the FGF-FGFR axis in essential cellular processes fosters the oncogenic potential of aberrant FGF signaling. Deregulation of FGF signaling with consequent carcinogenesis has been implicated in various malignancies including CCA. FGFR2 gene fusions have been detected in iCCA in several recent studies (14–17). In three different CCA cohorts, FGFR2 gene fusions were identified in 11–14% of iCCAs with report of several different FGFR2 gene fusions including FGFR2-BICC1, FGFR2-AHCYL1, FGFR2-TACC3, and FGFR2-KIAA 1598 (15, 16, 18). A higher frequency of FGFR gene fusions was noted by Sia et al. in a cohort of 107 iCCA patients (45%, 17/107) (19). FGFR2 gene fusions are rare in pCCA/dCCA. Mechanistic studies indicate that the FGFR binding partners mediate oligomerization and resultant activation of the respective FGFR kinase in tumors harboring FGFR translocations (17). Moreover, FGFR2 fusion proteins activate FGFR signaling with mitogen-activated protein kinase (MAPK) activation and confer anchorage-independent growth (18). Uncovering of these gene fusions is significant as these are often driver mutations and potentially targetable. For instance, targeting of the BCR-ABL gene fusion with Imatinib in chronic myelogenous leukemia has been one of the earliest instances of effective precision medicine (20).

Intrahepatic cholangiocarcinomas harboring FGFR2 gene fusions appear to have distinct clinical and pathologic features (15). Morphologically, cases harboring FGFR2 gene fusions appear to have a prominent intraductal growth pattern, anastomosing tubular structures, and desmoplasia (15). Immunohistochemically, iCCAs with FGFR2 gene fusions have strong cytokeratin (CK)-7 positivity but only weak and patchy expression of CK-19 (15). In addition to describing the pathologic features of tumors with FGFR2 gene fusions, Graham et al. also reported a survival advantage (median cancer specific survival 123 months versus 37 months) indicating that the presence of FGFR2 gene fusions may have prognostic significance (15). However, in an Asian cohort the presence of FGFR2 gene fusions in iCCA did not appear to have an impact on overall survival, clinical stage or tumor differentiation (18).

MET-HGF

MET tyrosine kinase is a plasma membrane protein which is activated when its extracellular domain binds to hepatocyte growth factor (HGF) or scatter factor (30). HGF-MET signaling has a critical role in essential cellular behaviors including proliferation, resistance to apoptosis, increased cell motility, and angiogenesis (30). Tumors can harness these processes to promote tumor growth and invasion. MET overexpression occurs in iCCA and correlates with the degree of tumor differentiation (31). An integrated molecular analysis identified two distinct biological classes in iCCA, an inflammation class (38% of iCCAs) and a proliferation class (62% of iCCAs) (32). The proliferation class featured activation of oncogenic signaling pathways such as MET, epidermal growth factor receptor (EGFR), and MAPK (32). Cross-talk between MET-HGF and ERRB family of receptors occurs and may account for resistance to MET and ERBB2 inhibitors (30). MET amplification has been detected in iCCA, albeit the incidence of these mutations is relatively low (16, 33).

A number of MET kinase inhibitors are being evaluated in clinical trials for various malignancies (34). Phase I results of a study of tivantinib, an oral MET inhibitor, in combination with gemcitabine in patients with solid tumors including CCA demonstrated partial response in 46% of patients and stable disease in 27% of patients (35). A phase II study of cabozantinib in patients with advanced CCA after progression on first or second line chemotherapy is currently ongoing (NCT01954745).

Mcl-1 and JAK/STAT Pathway

Myeloid cell leukemia sequence 1 (Mcl-1) is a potent antiapoptotic Bcl-2 protein. MCL1 amplification occurs in iCCA (16–21%) (16, 33). Mcl-1 mediates tumor necrosis factor-related apoptosis ligand (TRAIL) resistance in CCA by blocking the mitochondrial pathway of cell death (36). Interleukin (IL)-6, an inflammatory cytokine implicated in CCA biology, upregulates Mcl-1 via an AKT-dependent mechanism (37). IL-6 inhibition overcomes apoptosis resistance via downregulation of AKT and Mcl-1 (37). IL-6 upregulation of Mcl-1 is also mediated via Janus kinase (JAK) and signal transducer and activator of transcription (STAT) signaling cascade (38). STAT3 directly regulates Mcl-1 transcription suggesting that STAT3 inhibition may have therapeutic utility in CCA (38). Epigenetic silencing of suppressor of cytokine signaling 3 (SOCS-3) mediates sustained IL-6/STAT-3 signaling and enhanced Mcl-1 expression in CCA (39). Furthermore, biliary transduction of constitutively-activated AKT and yes-associated protein and systemic IL-33 administration induces tumor formation in mice via an IL-6 sensitive mechanism, underscoring the role of inflammatory cytokines in CCA oncogenesis (40). Mcl-1 amplification is a critical event in many cancers as cancer cells with Mcl-1 amplification are dependent on this potent anti-apoptotic protein for survival (41). Detection of Mcl-1 amplification utilizing FISH analysis is an essential component of a FISH probe set with high specificity for the detection of cholangiocarcinoma (42). Collectively, these findings provide several modes of therapeutic targeting in CCA. Novel, selective Mcl-1 inhibitors have been developed and preliminary preclinical studies have demonstrated evidence of their efficacy in pancreatic cancer (43, 44). Mechanistic studies have also demonstrated that S63845, a small molecule inhibitor of Mc-1 which binds with high affinity to the BH3-binding groove of Mcl-1, potently induces cell death in various solid cancer-derived cell lines (45). Preclinical studies have also demonstrated efficacy of the JAK2 inhibitor, AZD1480, in suppression of STAT-3 mediated oncogenesis in solid tumor cell lines (46).

IDH/IDH2 Mutations and Inhibitors

Isocitrate dehydrogenase 1 and 2 are enzymes that catalyze the oxidative decarboxylation of isocitrate to α-ketoglutarate. IDH mutations are heterozygous point mutations in catalytic amino acid residues and these alteration occur in Arginine 132 of IDH1 and Arginine 140 or Arginine 172 of IDH2 (47). IDH1/IDH2 mutations results in elevated levels of 2-hydroxyglutarate, an oncometabolite which induces widespread epigenetic changes. Consequently, IDH mutations have pleiotropic effect on differentiation, growth factor dependence, collagen processing and hypoxia signaling (47).

IDH1/IDH2 mutations, albeit a relatively common occurrence in gliomas, were previously thought to be rare in other solid organ tumors (48). However, several recent mutational profiling studies have demonstrated that IDH mutations are a relatively frequent genetic aberration in CCA (16, 33, 49). IDH mutations are more frequently observed in intrahepatic cholangiocarcinomas than pCCAs and dCCAs (23–28% versus 0–7%) (48, 50). Histopathologic analysis of 94 surgically resected primary cholangiocarcinomas demonstrated that IDH mutations are associated with clear cell change and poorly differentiated histology (48). In this study, patients with IDH1/2 gene mutations also appeared to have a better overall survival a year after surgical resection compared to patients without the presence of IDH mutations (48). A positive prognostic association with IDH mutations was also observed in a study of 326 iCCAs which demonstrated that IDH mutations were associated with longer overall survival and longer time to tumor recurrence after iCCA resection (51). However, subsequent studies have demonstrated that the prognostic significance of IDH mutations remains unclear. In a smaller cohort of 34 iCCAs, Jiao et al. demonstrated that patients with IDH mutations had a 3-year survival of 33% compared to 81% for those with wild-type IDH genes, although patients with IDH mutations were somewhat older and had a higher tumor stage (12). Additionally, the sample size of this study (n = 32) was much smaller compared to the prior study. A subsequent mutational profiling of 200 resected iCCA specimens demonstrated that although IDH-mutant tumors were more frequently multicentric in the liver, there was no impact on long-term prognosis (49). Although these 4 studies had conflicting results regarding prognosis, they focused mainly on early-stage or resectable iCCA, whereas targeted therapy with IDH inhibitors would be considered in patients with unresectable or advanced iCCA (52). Accordingly, in a subsequent study Goyal et al. assessed the correlation of IDH mutations with prognosis in 104 patients with unresectable or advanced iCCA (52). The presence of IDH mutations did not have a significant impact on the median overall survival (52).

The finding that IDH mutations are a relatively frequent occurrence in several human malignancies logically lead to speculation whether inhibition of mutant IDH activity may have therapeutic benefits. Small molecule inhibitors of mutant IDH were assessed in two proof of concept preclinical studies (53, 54). AGI-5198, a selective IDH1 inhibitor, blocked the ability of this enzyme to produce the oncometabolite 2-HG with resultant impaired growth of IDH1-mutant cells (53). Likewise, AGI-6780, a selective mutant IDH2 inhibitor, induced differentiation in hematopoietic cell lines (54). AG-120 is a first-in-class, orally bioavailable inhibitor of mutant IDH1. Phase 1 results of AG-120 in solid organ malignancies (NCT02073994) including iCCA presented at the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics in 2015 demonstrated that AG-120: i) is well-tolerated in the solid tumor patient population, ii) had encouraging evidence of clinical activity, iii) reduced intra-tumoral 2-HG levels. On the basis of these preliminary results, an advanced phase multi-center, randomized, double-blind, placebo-controlled study of AG-120 in previously treated subjects with unresectable or metastatic cholangiocarcinoma with an IDH1 mutation is planned (NCT02989857). AG-221, an orally available, selective inhibitor of mutant IDH2, has received fast-track designation from the Food and Drug Administration (FDA) and is currently being evaluated in multiple clinical trials including a Phase 1/2, multi-center study in subjects with advanced solid tumors including iCCA who harbor an IDH2 mutation (NCT02273739). Saha et al. recently demonstrated that iCCA cells with mutant IDH have a unique dependency on SRC kinase (55). Consequently, the SRC inhibitor, dasatinib, had a significant anti-tumor effect in mutant IDH xenografts, signifying a potential new therapeutic strategy against mutant IDH iCCA (55)

Protein tyrosine phosphatases (PTPs)

Members of the protein tyrosine phosphatase (PTP) are enzymes which remove phosphate from tyrosine residues in proteins (56). Genetic aberrations of PTPs can result in disequilibrium of protein tyrosine-kinase phosphatase activity with consequent oncogenic potential (56). A prevalence screen of 124 pairs of iCCAs and nontumor samples demonstrated that 41.4% of iCCAs had gain of function mutations in PTPN3 (57). Presence of these mutations conferred markedly enhanced pro-tumor activity and significantly increased postoperative tumor recurrence (57). This study has important therapeutic implications as it is well-known that gain of function mutations are potential therapeutic targets in various malignancies. Consequently, the high frequency of these mutations and their significant functional relevance makes these mutations an important potential therapeutic target.

Chromatin-remodeling Genes and HDAC Inhibitors

The role of dysregulated chromatin remodeling in iCCA was highlighted through exome sequencing of 32 iCCAs which detected genetic aberration of at least one chromatin-remodeling gene in 47% of cases (15 of 32) (12). Inactivating mutations were frequently identified in BAP1, ARID1A and PBRM1. BAP1 encodes a nuclear deubiquitinase involved in chromatin remodeling, whereas ARID1A and PBRM1 encode subunits of the SWI/SNF chromatin remodeling complexes (12). Subjects with mutations in any one of these genes tended to have shorter survival times and worse survival compared to subjects with wild-type genes, albeit these observations were not statistically significant (12). Frequent alterations of ARID1A (36% of iCCAs) were noted in next-generation sequencing of 28 iCCAs (16). ARID1A mutations appear to be more frequent in iCCA than pCCA/dCCA as indicated by a recent mutational profiling which identified ARID1A aberrations in 20% of iCCAs compared to 5% of pCCA/dCCAs (33). BAP1 and PBRM1 alterations were identified in 9% and 11% of iCCAs, respectively (33). BAP1 mutations were also detected in 10% of pCCA/dCCA cases, and in these patients the presence of BAP1 mutation was significantly associated with reduced progression-free survival (median 3 months vs. 8.8 months) and reduced overall survival (median 8.9 vs. 19.9 months) (33). Overall, in all cases harboring BAP1 mutations in this study, progression after first-line chemotherapy was observed within 4 months (33).

Several small molecule inhibitors targeting chromatin remodeling have been approved by the FDA (58). These include the histone deacetylase (HDAC) inhibitors vorinostat and romidepsin and DNA methyltransferase (DNMT) inhibitors azacitidine and decitabine (58). As the role of chromatin remodeling in CCA carcinogenesis is being illuminated, these inhibitors may have increased therapeutic utility in CCA.

EGFR/HER2

Aberrant expression and signaling of the epidermal growth factor receptor (ERBB) family of receptor tyrosine kinases has been reported in cholangiocarcinoma. Activation of this pathway leads to downstream oncogenic pathway activation including the MAPK pathway. The ERBB family has four distinct receptors including ERBB1 (EGFR), ERBB2 (Her-2/neu), ERBB3 and ERBB4 (59). In a cohort of surgically resected CCA cases, HER2 amplification was identified using FISH analysis in a single iCCA case and two pCCA/dCCA cases (15). A recent mutation profiling of 75 CCA specimens identified ERBB2 genetic aberrations in 20% of pCCAs/dCCAs and in only 1.8% of iCCAs (33). The pCCA/dCCA ERBB2 alterations included one amplification and four mutations, the latter being a novel finding, as mutations had previously not been reported in CCA. These mutations included one in the kinase domain (V777L) and three in the extracellular domain (S310F) (33). The V777L mutation is an activating mutation with some demonstrated resistance to lapatinib, a reversible, dual inhibitor of EGFR and HER2 (60). However, the V777L mutation was sensitive to nertatinib, an irreversible, dual inhibitor of EGFR and HER2 (60). The presence of kinase domain mutations may explain the limited success of erlotinib, a reversible EGFR inhibitor, in human CCA clinical trials (61, 62). Irreversible EGFR inhibitors such as neratinib, afatinib, and dacomatinib may have greater utility in this setting (33).

Protein kinase cyclic AMP (cAMP)-activated catalytic subunit alpha PRKACA) and beta (PRKACB)

Protein kinase A (PKA) is a cAMP-dependent protein kinase. Both PRKACA and PRKACB are catalytic submits of PKA and are members of the serine/threonine protein kinase family (63). Novel gene fusions of the PKA signaling components and mitochondrial ATP synthase, α subunit (ATP1B) were recently detected in pCCA/dCCA (63). Nakumura et al. detected ATP1B-PRKACA and ATP1B-PRKACB fusions in pCCA/dCCA cases (63). These fusions stimulated significantly enhanced expression of the PRKACA and PRKACB genes with consequent downstream MAPK signaling activation (63). Interestingly, fusion of the same exon of the PRKACA gene to DNAJB1 has been demonstrated in fibrolamellar hepatocellular carcinoma (64). Genetic aberrations in the PKA regulatory subunits, PRKAR1A (nonsense mutation) and PRKAR1B (PRKAR1B- C7orf50 gene fusion), were also detected in two different cases of pCCA/dCCA. PRKAR1A expression is enhanced in CCA, and PRKAR1A knockdown induces growth inhibition and apoptosis of CCA cells with resultant decrease in MAPK, phosphatidylinositol 3-kinase (PI3K)/Akt, JAT/STAT, and Wnt/β-catenin pathway signaling (65). Isoquinoline H89, a small molecule inhibitor of PKA, significantly inhibited CCA cell proliferation, albeit this may be an off-target effect as H89 is known to have a number of PKA-independent effects (65).

ELF3

ELF3 encodes an E26 transformation-specific transcription factor and hence is implicated in the regulation of several genes with essential roles in multiple cellular processes (67). By interacting with their promoter regions ELF3 increases transcription of TGFBR2 and EGF1 activation, two known tumor suppressor gene (68). TGFBR2 also has an integral role in TGF-β signaling. Two recent genomic analyses detected inactivating mutations of ELF3 in ampullary carcinoma/dCCA. In a cohort of 98 periampullary carcinomas and 44 dCCAs, Gingras et al. identified inactivating frameshift or nonsense mutations in 10.6% of periampullary tumors (68). Similarly, ELF3 mutations had been reported in 9.5% of extrahepatic CCAs (pCCA/dCCA) in a molecular characterization of biliary tract cancers (63). In a cohort of 172 ampullary carcinomas, driver mutations in ELF3 were also identified in 13.3% of cases (69). ELF3 mutations were present at high allele frequencies, indicating that ELF3 mutation may represent a founder or driver mutation in ampullary carcinoma (69). Functional studies were subsequently carried out and demonstrated that ELF3 knockdown promotes motility and invasion in epithelial cells (69).

HER2 (ERBB2)/ERBB3

Molecular alterations in the ERBB family are well-described in pCCA (please see above) and have recently been reported in dCCA/ampullary carcinoma. ERBB3 mutations were identified in 14% of pancreatobiliary-type ampullary carcinomas and ERBB2 mutations were identified in 11.6% of ampullary carcinomas (69). ERBB2 mutations overlapped with ELF3 patients.

KRAS

Activating mutations of the proto-oncogene KRAS are one of the most frequently encountered genetic mutations in CCA. The rate of KRAS mutations appear to be higher in pCCA/dCCA (40%) compared to iCCA (9%–24%) (16, 33, 49). The presence of these mutations has prognostic utility as patients harboring KRAS mutations have worse progression-free and overall survival (33, 70). Moreover, the KRAS mutant tumors are more likely to have adjacent organ involvement and R1 margin status (49). KRAS activation leads to upregulation of downstream effector pathways including PI3k-AKT-mammalian target of rapamycin (mTOR) and Raf/MEK/ERK. Presently, there are no effective direct inhibitors of KRAS and therefore the therapeutic approach for KRAS mutant tumors is inhibition of the downstream pathways. Selumetinib, a selective MEK 1/2 inhibitor, had demonstrated some efficacy in a phase II trial of advanced biliary tract cancers (26). Subsequently, phase I results of a study with selumetinib in combination with gemcitabine and cisplatin in advanced biliary tract cancer demonstrated a median progression-free survival of 6.4 months and acceptable adverse event profile (71). Other early phase trials of MEK inhibitors with conventional chemotherapy or tyrosine kinase inhibitors in biliary tract cancer are ongoing (NCT02042443, NCT01438554).

PI3K-AKT-mTOR Pathway

The PI3K pathway is activated by stimulation of several different receptor tyrosine kinases (e.g. c-MET, FGFR) with consequent activation of AKT (72). This in turn phosphorylates the mTOR complex which is integral in regulation of several cellular processes including proliferation, survival, tumor invasion, metabolism and angiogenesis (72, 73). Consequently, the activation of the PI3K-AKT-mTOR pathway mediates oncogenesis in a multitude of malignancies including CCA (74–76). Exome sequencing has identified mutations in multiple family members of the PI3K pathway including inactivating mutations of PIK3R1 and activating mutations of PIK3CA, PIK3C2A, PIK3C2G (12, 63). Activating mutations of PIK3CA have been reported in 4–8% of iCCAs and 4% of pCCA/dCCA cases (16, 49, 63).

Presently, there is a multitude of clinical trials investigation various inhibitors of the PI3K-AKT-mTOR pathway in different solid organ malignancies (77). Selective inhibition of mTOR can lead to feedback activation of AKT (78). However, preclinical data suggests that AKT inhibition can potentiate the anti-tumor efficacy of mTOR inhibition (78). The presence of functional redundancy within this pathway and extensive cross-talk with other pathways such as the RAF-MEK-ERK pathway makes combination chemotherapeutic approaches an attractive option. Preclinical data has demonstrated that dual targeting of the PI3K/AKT/mTOR and RAF-MEK-ERK pathways has synergistic effects in CCA (79). However, phase I trial of the PI3K inhibitor BKM120 in combination with mFOLFOX6 (a common chemotherapeutic regimen in gastrointestinal malignancies) revealed increased toxicity with the combination compared to both as single agents (80). Conversely, a phase I trial of the mTOR inhibitor everolimus in combination with gemcitabine and cisplatin, the current practice standard for metastatic CCA, in patients with biliary tract cancer demonstrated stable disease in 6/10 patients and an acceptable safety profile (81).

S100A4

S100A4, a cytoskeleton-associated calcium binding protein, has garnered recent attention for its role in metastasis of various malignancies including cholangiocarcinoma (82, 83). Depending on its subcellular localization, S100A4 contributes to the regulation of a multitude of cell biological processes including proliferation, survival, differentiation, and cytoskeletal rearrangement (82). S100A4 expression has prognostic implications as well, as it is has been identified in the subset of patients with worse outcomes in several malignancies including cholangiocarcinoma (83, 84). Fabris et al. identified enhanced nuclear expression of S100A4 in the CCA patients with a worse overall survival following surgical resection (83). Furthermore, nuclear expression of S100A4 was linked with increased CCA cell invasiveness and metastasization, suggesting that it may be a potential therapeutic target in CCA (83). A subsequent study demonstrated that low-dose paclitaxel down-regulated nuclear S100A4 expression. Consequently, a mouse xenograft model treated with low-dose paclitaxel had reduced lung metastasis, albeit no significant effect on local tumor growth (85). Overall, these findings support the notion that therapeutic targeting of S100A4 may have utility in hindering metastatic progression of cholangiocarcinoma.

Hedgehog Signaling Pathway

The Hedgehog (HH) signaling pathway is an evolutionary conserved, developmental pathway which is known to be deregulated in a number of malignancies including cholangiocarcinoma (86, 87). Preclinical studies have demonstrated that hedgehog signaling inhibition with cyclopamine hinders CCA cell proliferation, migration, and invasion and promotes tumor suppression in vivo (86, 88). Hh signaling-induced apoptosis resistance in CCA is mediated by polo-like kinase 2 (PLK2), a cell division regulating kinase (89). Profiling of human iCCA as well as pCCA/dCCA has demonstrated the presence of PLK immunoreactive cells (89). Accordingly, Hh signaling inhibition by cyclopamine reduces PLK expression (89). The role of the non-canonical Hh pathway in CCA has been examined as CCA cells often do not express cilia, a prerequisite for canonical Hh signaling (90). Indeed, the non-canonical Hh pathway promotes tumor progression in CCA. Moreover, a small molecule antagonist of the Hh plasma membrane protein Smoothened, vismodegib, resulted in tumor suppression and inhibition of metastasis in a preclinical CCA model (90). Another small molecule Hh antagonist, BMS-833923, had a more profound tumor suppressive effect in CCA xenografts when combined with gemcitabine, compared to either treatment alone (91).

Mesothelin

Mesothelin, a cell surface protein expressed in normal mesothelial cells, has aberrant tissue expression in a several malignancies including pancreatic adenocarcinoma and CCA (92, 93). Moreover, mesothelin expression appears to have prognostic implications in CCA. In a cohort of 61 surgically resected pCCA and dCCA specimens, mesothelin expression was associated with intrahepatic metastasis, peritoneal metastasis, and worse overall patient survival (94). A series of surgically resected iCCAs also demonstrated an association of mesothelin expression with poor overall survival following surgical resection (95). On the basis of these findings, mesothelin has become a potential target for antibody-based cancer therapy (96). Consequently, several therapeutic agents are under investigation in preclinical and clinical studies. Amatuximab, a chimeric monoclonal antimesothelin antibody, combined with cisplatin was well-tolerated and had a stable disease rate of 90% in patients with advanced pleural mesothelioma (97). However, a phase I study of amatuximab in 17 patients with advanced solid tumors including pancreatic adenocarcinomas demonstrated that only 3 patients had stable disease (98). Phase I results of SS1P, a recombinant anti-mesothelin immunotoxin, in 33 patients with mesothelin expressing malignancies including pancreatic adenocarcinoma, 19 had stable disease and 4 had minor responses (99). A phase II trial of SS1P and pentostatin plus cyclophosphamide in mesothelin expressing malignancies including pancreatic adenocarcinoma is currently ongoing (NCT01362790).