A perspective on molecular therapy in cholangiocarcinoma: present status and future directions

SUMMARY

Cholangiocarcinoma (CCA) is an orphan cancer with limited understanding of its genetic and genomic pathogenesis. Typically, it is highly treatment-refractory and patient outcome is dismal. Currently, there are no approved therapeutics for CCA and surgical resection remains the only option with curative intent. Clinical trials are currently being performed in a mixed cohort of biliary tract cancers that includes intrahepatic CCA, extrahepatic/perihilar CCA, distal extrahepatic CCA, gallbladder carcinoma and, in rare cases, even pancreatic cancers. Today, clinical trials fail primarily because they are underpowered mixed cohorts and designed without intent to enrich for markers to optimize success for targeted therapy. This review aims to emphasize current clinical attempts for targeted therapy of CCA, as well as highlight promising new candidate pathways revealed by translational genomics.

Practice Points.

• Patient classification and characterization may be needed regardless of successful implementation of individualized treatment. This may guide secondary or tertiary treatment options in the case of drug resistance or disease recurrence.

• Detailed genetic, genomic and/or epigenomic data from large cohorts in combination with genomic analyses of individual patients will facilitate both faster clinical decision-making and the appropriate choice of therapies among the available drug options and pathways to target.

• Implementation of either conventional monotherapy or a single pathway-specific targeted therapy is unlikely to be successful, particularly in drug-resistant heterogeneous cancers such as cholangiocarcinoma (CCA).

• Coadministration of either unique targeted therapy based on molecular characteristics of the patient's tumor, with cytotoxic standard chemotherapy or combination of several targeted therapies inhibiting multiple dysregulated pathways, is likely to achieve a synergistic effect and be successful. However, the concern with simultaneously targeting multiple pathways is the increased risk of drug toxicity and adverse effects for the patient.

• CCA is characterized by diverse anatomical locations – that is, intrahepatic and extrahepatic bile ducts – as well as clinical and molecular heterogeneity that complicates assessment of treatment efficacy. This is further confounded when patients with biliary tract cancer are considered altogether as a group.

• Recent interest in understanding the molecular complexity of CCA has been evident; however, access to well-preserved and annotated cohorts for detailed genetic and genomic assessments is a limiting factor. Mainly due to the anatomical complexity of this malignancy, heterogeneity of the tumor samples/cohorts studied may be considered as a confounding factor that should be addressed in each study according to the 2013 International Liver Cancer Association guidelines.

Cholangiocarcinoma (CCA) is a rare and devastating malignancy with a dismal outcome. Whereas the incidence of most malignancies has declined in the past two decades, the incidence of CCA has increased worldwide [1], and the mortality rate of CCA has risen by more than 9% in the USA [2]. Intrahepatic CCA (iCCA) is classified as a peripheral-type tumor of the interlobular bile ducts, whereas tumors designated perihilar (pCCA)/hilar are generally considered extrahepatic, and originate in the main hepatic ducts or at the bifurcation of the common hepatic duct. iCCA is often diagnosed at a late stage in the disease progression; partly due to the anatomical location, diverse growth pattern and underlying pathobiological heterogeneity. In addition, at diagnosis, chemotherapy and radiation therapy are generally ineffective. Thus, the clinical management of this disease offers limited therapeutic options – that is, in the absence of unresectable advanced and metastatic disease, surgical resection remains the only curative treatment for patients with CCA [3]. Standard therapy administered to these patients is therefore typically palliative. In addition, intrahepatic recurrence is frequent following attempted curative resection and serves as a confounding variable [4].

The etiology of CCA remains partly undetermined [5]. A background of chronic liver inflammation, for example, primary biliary cirrhosis or primary sclerosing cholangitis are linked to increased risk of developing CCA. Other risk factors include bile duct injury (cholestasis), hepatitis B and C viral infection, alcohol consumption, diabetes or more regional specific hazards such as parasitic liver infestation.

Multitarget tyrosine kinase inhibitor (TKI) sorafenib, which is used as first-line therapy for advanced hepatocellular carcinoma (HCC), has had limited success in trials of CCA patients [6]. The lack of therapeutic efficacy in the clinical management of CCA is, in part, the result of inadequate molecular and pathobiological understanding of this disease. Stratification of class-specific risk groups or focus on individual patients may be essential for clinical success in treating CCA patients in the future.

Conventional chemotherapy: current standard of care

Standard of care is usually administered to CCA patients with palliative intend. Only few Phase III randomized controlled trials have been conducted in CCA, and typically these studies are in mixed biliary tract cancer (BTC) cohorts. Surgical resection remains the only clinical option with curative intent in which an estimated 30% 5-year survival rate may be achieved. Patients with unresectable iCCA have a 0–5% 5-year survival rate [7].

Systemic chemotherapy, as well as targeted therapies, have typically had limited success in CCA. A randomized Phase III trial in 90 patients with advanced BTC and pancreatic cancers compared 5-flurouricil and leucovorin to best supportive care [8]. This trial demonstrated improved quality of life and prolonged survival as a result of the treatment. Interestingly, more than 100 Phase II trials that included >2800 patients were conducted between 1985 and 2006; however, they only had an average cohort size of 25 subjects [9]. This meta-analysis demonstrated an overall response rate – that is, partial plus complete response – following systemic chemotherapy of 22.6%, time to progression of 4.1 months and median overall survival (OS) was 8.2 months. In a small study including 23 treatment-naive unresectable BTC patients, gemcitabine was given as a single agent [10]. In this nonrandomized Phase II trial, gemcitabine alone was generally well tolerated, showed limited adverse events and had clinical efficacy with 26.1% rate partial. In addition, gemcitabine in combination with either irinotecan [11] or capecitabine [12] showed limited adverse events, but overall modest improved efficacy. The recent ABC-02 Phase III randomized controlled trial included 410 patients and established a new guideline for the standard of care in advanced and metastatic BTC [13,14]. In this trial, the authors reported a significant benefit in time to progression and increase in the median OS by 11.7 months after combined chemotherapy of gemcitabine plus cisplatin (CisGem) versus gemcitabine alone [14]. This trial was validated in a smaller Japanese Phase II multicenter study (BT22) of CisGem in patients with advanced BTC [15], reporting similar efficacy as observed in the ABC-02 trial. Although most clinical trials include patients with BTC and are typically underpowered, nonrandomized, single-center studies, these recent studies are encouraging as a baseline to design combination trials with molecular-targeted therapies in unique well-characterized patient groups.

The potential of neoadjuvant chemotherapy to achieve operable margins in patient downstaging would also represent a promising step to improve management of the disease. As for adjuvant therapy, currently there is limited evidence of any clinical efficacy. However, in a Phase III randomized clinical trial with mitomycin C plus 5-flurouricil, clinical benefit was reported, but only in the patient subgroup diagnosed with gallbladder cancer [16]. However, we are currently awaiting the ongoing UK NCRI BILCAP trial that will report its findings in 2014 on postoperative capecitabine versus observation alone.

Despite some encouraging progress, there are clearly still many limitations and unmet needs. Discovery of new systemic agents with clinical efficacy and actionable targets are urgently required. To achieve this goal in a rare neoplasms, such as CCA, with diverse clinicopathology, multicenter studies are necessary to acquire the required cohorts. Future trials may include CisGem or derivatives in the control arm of the study as current standard of care. Detailed molecular evaluation is recommended and, as such, it is suggested that future clinical trials plan for procurement of tissue samples for molecular characterization to: evaluate the biomarker and/or drugable target network; establish a drug response or resistance signature; and identify genetic events pre- and post-treatment, for example, difference in the mutational profile caused following the treatment either directly drug-related or due to accrual of new or underlying somatic mutations, which would indicate resistance to the treatment.

Rationale for targeting candidate oncogenic signaling pathways

To date, some of the most interesting approaches include strategies targeting the EGF receptor (EGFR) family (EGFR and HER2), the RAS/RAF/MEK/ERK signaling network and angiogenesis (VEGF) (Figure 1). Besides these pathways, well-described molecular alterations include the MET proto-oncogene and inflammatory stimuli from the tumor microenvironment – that is, the tumor stromal compartment, such as IL-6/STAT3, as well as COX2. However, integration of the new genomic data [17–20], for example, genetic, epigenomic and/or transcriptomic [21] patterns that may improve patient stratification may emerge and help guide clinical decisions and provide new opportunities to treat CCA, as is the case for other malignancies.

Figure 1.  Canonical pathways and classical targeted therapy.

EGFR: EGF receptor; IL-6R: IL-6 receptor; PDGFR: PDGF receptor; VEGFR: VEGF receptor.

▪ EGFR family

The EGFR family of receptor tyrosine kinases (RTKs) is comprised of ERBB1–4. However, ERBB1 (EGFR) and ERBB2 (HER2/Neu) are the two RTKs most frequently dysregulated in the molecular pathogenesis of CCA [22,23]. In addition, this is an attractive pathway for therapeutic targeting since TKIs are effective antitumor agents in other cancers.

Aberrant EGFR and HER2 expression is found in many cancers, for example, breast, gastric and ovarian. In CCA, overexpression of EGFR and HER2 was reported in 10–32% and 20–30% of tumors, respectively [22,23]. BTC with increased EGFR and HER2 expression displayed copy number (CN) gains in 77 and 79% of cases, respectively [24]. However, two recent large genomic studies of CCA, showed overexpression of EGFR and HER2 signaling in patient subgroups associated with poor outcome and KRAS mutations, despite a lack of CN amplification [17,18,25]. Although HER2 and EGFR are rarely mutated in CCA, genetic variants in EGFR (0–13.6%) have been reported [17,18,26]. In the case of ERBB3 and ERBB4, information is limited, although these receptors were shown by immunohistochemistry to be moderate-to-strongly positive in 40 and 11% of cases, respectively [27]. Following laser-capture microdissection in a subset of CCAs [17], the authors found a significant HER2–ERBB3 gene network associated with the epithelial tumor compartment, suggesting cooperation between the two receptors. Recently, ERBB3 was shown in association with HER2 to correlate with survival in a subset of pCCA patients [28].

Preclinical evidence supports the oncogenic role of the ERBB receptor family, particularly HER2, in the development and treatment of CCA. Kiguchi et al. have developed a transgenic mouse model where expression of the rat ErbB2, under the control of the bovine keratin-5 promoter (BK5.erbB2), is targeted to the basal layer of multiple epithelial tissues, which include the bile ducts, giving rise to CCA in 30% of cases as a consequence of the elevated HER2 expression [29]. Other evidence for the involvement of HER2 originates from chemical-induced animal models using furan [30], thioacetamide [31], p53-deficient/CCL₄ [32], as well as the orthotopic transplantation BDEneu model (in which adult rat BDE1 cholangiocytes are transfected with a mutated constitutively active HER2 gene, resulting in the cells becoming tumorigenic) [33]. Elevated expression levels of EGFR and COX2, as well as activation of ERK and AKT, generally characterize these models, which are similar to human CCA with poor prognosis.

To assess RTKs as potential drug targets [23], cell culture and orthotopic animal models using anti-EGFR therapies, such as erlotinib and cetuximab, have demonstrated the ability of these drugs to inhibit proliferation of CCA cell lines [34]. Furthermore, simultaneous targeting of EGFR and HER2 by lapatinib has demonstrated antiproliferative effects, as well as the ability to reduce tumor growth [17,35]. A multi-TKI (vandetanib), which targets EGFR, VEGF receptor (VEGFR) 2 and RET was demonstrated in vivo to cause significant tumor growth inhibition and prolong the time to metastasis [36]. Currently, several clinical Phase I/II trials are ongoing, evaluating vandetanib alone or in combination with standard chemotherapy.

The clinical experience with molecular-targeted therapies against EGFR in CCA has been inadequate compared with other therapeutic areas, suggesting that future trials may require improved patient stratification and likely combination strategies, targeting several oncogenic pathways simultaneously. Interestingly, COX2 activation may be involved in the regulation of RTKs and the MAPK signaling network. As such, it was demonstrated that bile acid from patients with congenital choledochal cysts has a growth promoting effect on CCA cell lines and that this proliferation can be inhibited by the COX2 inhibitor, celecoxib [37]. Other preclinical models include the EGFR inhibitor iressa as a radio sensitizer [38]. Interestingly, introduction of radiation therapy in CCA was shown to activate AKT, which in turn was suppressed by administration of iressa.

▪ MAPK signaling cascade

The RAS/RAF/MEK/ERK tyrosine kinases pathway is frequently dysregulated in cancers, including CCA, supporting unrestricted growth and survival of cancer cells [39,40]. This dysregulation is driven by numerous growth promoting signals, including RTKs, PDGF and inflammatory stimuli.

Activating mutations in KRAS are one of the most frequently reported disease-causing genetic events in CCA, with an incidence ranging from 8 to 100% [17–20,41–43]. The authors previously reported that 24.6% of CCA patients have mutations in the KRAS gene [17]. However, when classifying patients according to tumor site, 53.3% of hilar-type (pCCA) versus 16.7% of peripheral-type CCAs (iCCA) were positive for KRAS mutations, suggesting the importance of the anatomical location for the tumor in selecting for KRAS mutations, and the subsequent choice of therapy.

Preclinical animal models, such as the genetically engineered tissue-specific activation of the hotspot mutation KRAS ^G12D, display low penetrance and long latency for the development of invasive iCCA [44]. In combination with p53 inactivation the penetrance is greatly increased, suggesting these genes are important ‘drivers’ in the development of iCCA [44]. To date, there are no proven KRAS-targeted therapies. Furthermore, the therapeutic efficacy of EGFR-targeted therapies, for example, erlotinib, cetuximab and panitumumab, is restricted to patients with wild-type KRAS gene status. Similarly, patients with rare NRAS mutations showed limited respond to EGFR inhibition [45]. However, we have shown the potential of sensitizing CCA to TKI-based therapies using a murine xenograft model designed to mimic TKI-resistant human CCA with a mutated KRAS gene (KRASG12D). In this model, mice treated with the HER2 inhibitor (trastuzumab) in combination with celecoxib demonstrate a significant decrease in tumor growth and benefit on survival, whereas trastuzumab administered alone is ineffective, suggesting that celecoxib sensitize these tumors to HER2 inhibition [Andersen JB et al., Unpublished Data]. Interestingly, preclinical mouse models of orthotopic transplanted gallbladder carcinoma with KRAS mutations also showed a significant survival benefit following treatment with downstream MEK inhibitors (U0126 and PD98059) [46]. Targeting MEK is an attractive strategy, particularly since concerns of drug resistance similar to what is observed with EGFR inhibitors are limited in patients with KRAS^G12D CCAs. Several clinical trials evaluating drugs targeting MEK (selumetinib [AZD6244] [47], ARRAY-438162 and GSK1120212) are either ongoing or concluded.

▪ BRAF^V600E

BRAF is an upstream component of the MAPK pathway. The genetic variant BRAF^V600E was identified in 0–22% of CCA, dependent on the population studied [17,18,48,49]. Several BRAF-targeted therapies, for example, GSK2118436 and vemurafenib (PLX-4032), are in clinical development; however, vemurafenib is thus far the only US FDA-approved BRAF inhibitor for metastatic melanoma. The V600E mutation in the BRAF gene was recently highlighted as a promising target in appropriate iCCA patients – limited to Europe, with a potential of an estimated 3000 iCCA patients annually [18].

▪ Angiogenesis: VEGF

VEGF, a key factor in promoting tumor growth and metastasis via its important role in neo-angiogenesis, is overexpressed in >50% of CCA [50,51]. Moreover, high expression of VEGF is associated with poor prognosis. Experimental studies have shown that sorafenib, a multi-TKI primarily targeting VEGF, but also PDGF receptor (PDGFR) and RAF kinases, has potent activity in vitro and in vivo against iCCA [52]; however, clinical trials in BTC have failed [6,53]. Evaluation of other VEGF inhibitors includes, sunitinib (against VEGFR, PDGFR and KIT), vandetanib (against VEGFR, EGFR and RET), bevacizumab (against VEGF-a) and cediranib (against VEGFR) has either failed or is ongoing.

Targeted therapies in CCA have mainly in the past focused on inhibition of RTKs (EGFR and HER2), VEGFR and MAPK pathways. However, several promising ‘classical’ targets include the HGF receptor (MET), the PI3K/AKT pathway and mTOR.

▪ MET proto-oncogene signaling

Binding of the HGF to the MET receptor triggers an activation cascade of multiple downstream signal transduction pathways, for example, MAPK, PI3K/AKT and STAT [54]. Aberrant signaling and constitutive MET activation caused by either mutation or CN amplification was demonstrated in numerous malignancies [48]. Increased expression of MET is observed in 12–58% of CCA patients [24,55,56] and is associated with poor prognosis [55,57]. Both human data [57] and preclinical experimental models [55,58] suggest an oncogenic signaling loop between overexpression of MET and members of the ERBB-family, suggesting a likelihood of drug resistance to EGFR inhibitors in patients with aberrant MET expression.

Several MET-targeted inhibitors are in clinical development [59]. Data from a recent Phase II randomized clinical trial evaluating efficacy of tivantinib (ARQ-196) in advanced HCC as second line after sorafenib failure are very encouraging [60]. The trial demonstrated an approximate doubling of progression-free survival and OS compared with placebo in patients with MET overexpression, which includes approximately 25% of patients. A dose escalation Phase I trial for MET inhibitor LY2801653 is currently enrolling patients with advanced cancers in three different tumor types, including CCA (NCT01285037) [201]. This safety and later efficacy trial is, unfortunately, not expected to report before late 2015. MET inhibitors (tivantinib, cabozantinib, golvatinib, foretinib or MetMAb) have not been evaluated in CCA; however, future clinical trials evaluating MET-targeted therapies alone or in combination with standard of care, combined with EGFR/HER2 inhibitors or sorafenib in patients stratified for MET expression, would be warranted.

▪ PI3K/PTEN/AKT/mTOR network

The PI3K/AKT/mTOR pathway is activated downstream of RTKs. Altered regulation of this pathway results in proliferation, changes in cell adhesion and motility, as well as angiogenesis and, ultimately, tumorigenesis. Subgroups of CCA patients with poor outcome are characterized by activation and elevated expression of RTKs, in particular, EGFR, IGF receptor, HER2 and MET, which result in phosphorylation of AKT observed in 22–65% of tumors [61,62]. Data obtained from transcriptomic profiling show that the AKT/mTOR pathway is activated in a subset of CCAs [17,63]. This is supported by immunohistochemical analyses of the downstream targets of mTOR, RPS6, eIF4-E and the 4E-BP1 [17,63,64].

The gene PIK3CA is a subunit (P110α) of PI3K, which is an activator of the downstream AKT/mTOR pathway. Mutations in PIK3CA were identified in many cancers and shown to result in constitutive activation of AKT signaling. Recent studies in CCA suggest that mutations in PIK3CA (0–32%) are relatively rare, with the exception of in the Chinese population [20,49,64,65]. Reiner et al. identified hotspot PIK3CA mutations in one out of 11 iCCAs [64]. Similarly, Deshpande et al. found no mutations (none out of 44) in iCCA and pCCA [65], whereas 13% of the examined gallbladder carcinomas were positive for PIK3CA mutations. By contrast, a study in the Chinese population reported PIK3CA mutations in 32% (11 out of 32) of CCA patients [49].

A genetically engineered mouse model of liver-specific PTEN and SMAD4 disruption was found to develop aggressive iCCA [66]. Given the strong activation of downstream mTOR and MAPK pathways in this animal model, as well as in CCA patients with poor outcome, targeting either of the nodes in the PI3K/AKT/mTOR pathway may present an alternative therapeutic strategy in the clinical management of CCA. Preclinical evaluation of inhibition of AKT (MK2206) and mTOR (RAD001/everolimus) alone or in combination demonstrated a synergistic effect of both targeted therapies in CCA, significantly reducing the tumor growth [67]. In addition, we have shown that MK2206 alone is sufficient to inhibit tumor growth in a TKI-resistant xenograft model of KRAS ^G12D CCA [Andersen JB et al., Unpublished Data]. The efficacy of MK2206 in unresectable, advanced and metastatic refractory biliary cancer is currently ongoing and results from this Phase II trial are anticipated in early 2014 (NCT01425879) [202].

Although a recent review suggested that more than 26 PI3K inhibitors are in development and used in >150 clinical trials, we found no current clinical trials using PIK3CA-targeted therapy in CCA [68]. The only clinical trial, which planned to enroll patients with CCA, was a Phase II study scheduled to evaluate BKM120 in PI3KCA-mutated cancers. However, this trial was terminated owing to limited accrual (NCT01501604) [203]. Although multiple trials evaluating mTOR-targeted therapies in HCC are ongoing [69], implementation of these inhibitors in the treatment of CCA is currently limited to two ongoing single-arm Phase II studies (RADiChol and CRAD001T) using everolimusin first-line treatment against advanced CCA (NCT00973713 [204] and NCT01525719 [205]).

Molecular-targeted therapies

The standard of care for patients with advanced inoperable CCA is palliative chemotherapy. Since the efficacy of systemic drugs is limited to marginal prolonged survival compared with best supportive care, recent years have led to intensified investigations into molecular-targeted therapies. The implementation of these therapies in the treatment of CCA is difficult, largely because the majority of clinical trials are designed for BTC and, as such, the CCA patient group is typically too underpowered to allow for subgroup analysis and there is an absence of a comprehensive genomic and genetic understanding of the pathogenesis of CCA.

▪ Tyrosine kinase inhibitors

In past years, clinical trials have mainly been focused on TKIs as single agents in first-line treatment or in combination with standard chemotherapy. Results from multiple Phase II clinical trials that have assessed targeted agents – for example, erlotinib (against EGFR) [70], lapatinib (against EGFR and HER2) [71], sorafenib (multi-TKI against VEGFR2/3, PDGFR and RAF kinases) [6], sunitinib (against VEGFR, PDGFR and KIT) [72] and selumetinib (against MEK) [47], have unfortunately not warranted further investigation. It is important to mention that these were all relatively small monotherapy-driven trials performed in mixed BTC cohorts. Furthermore, clinical trials designed to evaluate efficacy of targeted therapy should be encouraged to include molecular analysis of the particular pathway inhibited by the drug (i.e., marker-driven study). In this regard, the study design of the selumetinib trial [47] required tissue sampling with analysis of BRAF and KRAS mutational status, as well as quantification of the activation of ERK and AKT. The level of ERK phosphorylation was determined to significantly correlate with drug response.

A combination therapy comprising both erlotinib and bavacizumab showed marginal benefit over previous trials with 12% rate partial, time to progression was 4.4 months and median OS was 9.9 months [73]. As a secondary end point, the trial was designed to integrate molecular evaluation of EGFR and KRAS mutational status, as well as VEGF serum levels. Several clinical trials that are designed to evaluate VEGFR inhibitor (vandetinib) alone or in combination with conventional chemotherapy are ongoing (NCT00753675 and NCT00551096) [206,207]. Similarly, two trials designed to study the efficacy of cediranib in advanced BTC are currently in progress (NCT01229111 and NCT00939848) [208,209]. The UK-based trial ABC-003 is a randomized Phase II study with 136 enrolled patients, which will compare cediranib in combination with CisGem to CisGem alone. This study is active but not recruiting, and expected to report in 2014.

In a continuation of monotherapy-based trials using sorafenib or erlotinib, a combination of both drugs in first-line treatment also failed to meet the primary end point [74]. Interestingly, the study concluded that improved patient selection may be required. Analysis of molecular markers prior to enrollment as part of inclusion/exclusion criteria may be beneficial for the outcome of targeted therapies in future clinical trials.

In recent years, combination trials including both chemotherapy – for example, gemcitabine and platinum-based drugs with targeted therapies, for example, bevacizumab [75], cetuximab (EGFR monoclonal antibody) [76] and erlotinib [77], have given rise to more encouraging results. The erlotinib trial is currently the only randomized Phase III trial, with 268 patients being administered GEMOX plus erlotinib versus standard chemotherapy alone [77]. This trial demonstrated a significant improved objective response (40 vs 25 patients in either arm, p < 0.005); marginal benefit in progression-free survival was 4.2 versus 5.4 months and a comparable median OS of 9.5 months was achieved. The two single-arm Phase II trials both demonstrated better objective response rates (40 vs 63%) and median OS (12.7 vs 15.2 months) [75,76]. Particularly, GEMOX plus cetuximab showed encouraging results as three patients experienced complete response and a total of nine patients downstaged after therapy and subsequently underwent surgical resection with curative intent.

To improve clinical efficacy of targeted therapies, strengthened patient selection may be required either based on the tumor biology – that is, mixed BTC cohorts and/or evaluation of molecular targets as part of the inclusion criteria. A small Phase II trial recently reported significant efficacy of GEMOX, capecitabine plus panitumumab (EGFR monoclonal antibody) in first-line treatment of BTC with a KRAS wild-type gene status as an inclusion criteria [78]. However, the evaluation of the study is complicated by the addition of capecitabine in the single-arm trial compared with previous trials. Since the study met its efficacy criteria with a primary end point of 6 months progression-free survival at 74%, it may warrant a larger randomized trial. A randomized Phase II trial (PiCCA) evaluating the efficacy of panitumumab and CisGem compared with CisGem alone is currently only recruiting patients with KRAS wild-type tumor status and no prior systemic treatment; however, results from this trial are not scheduled until mid-2014. Beside this study, only few trials are currently ongoing [22].

Emerging molecular targets

In recent years, promising new candidate targets have been uncovered by detailed genomic characterization and patient stratification, which may result in ‘innovative therapeutics’ and strategies to improve clinical outcome in CCA (Figure 2). We recently showed that a stromal-specific gene signature following microdissection of CCAs is associated with poor prognosis and that predominant gene networks correlate with hepatic stellate cells activation, chemokine and cytokine expression [17]. In another stromal gene signature of iCCA, osteopontin was identified as an independent predictor of survival [79]. CCA is a very desmoplastic cancer type [80], and the microenvironment is a key component of the tumor development and progression in many cancers [81]. The CCA microenvironment has received attention in recent years and various studies suggest that this area could be of therapeutic interest in the future [10,11,29,79,82–84].

Figure 2.  Emerging-targeted therapy.

APC: Antigen-presenting cell; FGFR: FGF receptor; IGFR: IGR receptor; TCA: Tricarboxylic acid cycle.

▪ Polo-like kinase

This is a family of serine/threonine kinases involved in key regulatory processes, such as cell cycle progression (G2/M transition) and cytokinesis. Interestingly, Polo-like kinase (PLK)1 is overexpressed in several cancers and correlates with poor outcome [85]. In a large genomic subgroup analysis, the authors found PLK1 differentially expressed in CCA from patients with overall poor prognosis [17]. Preclinical evaluation found that the selective PLK1 inhibitor (BI2536) reduced proliferation of 14 BTC cell lines by inducing G2/M arrest and apoptosis, whereas the phosphorylation status of AKT and ERK was unchanged by the drug [86]. Recently, a role of PLK2 linked PLK and Hedgehog (Hh) signaling in TRAIL-induced cell death, by demonstrating the ability of Hh to regulate PLK2 expression, stabilize Mcl-1 and thus, convene resistance to TRAIL [83]. Fingas et al. concluded that targeting either Hh or PLK signaling may restore CCA cell susceptibility to TRAIL-induced apoptosis [83].

▪ Hh & Notch signaling pathways

The Hh signaling pathway is important in the regulation of proliferation, survival, self-renewal and development of embryo and adult tissues. The pathway is implicated in many cancer types, including biliary tract cancers [87].

The function of cancer-associated fibroblasts or myofibroblasts (MFs) has recently demanded more attention since a cross-talk between MFs and the tumor epithelium itself may be tumor promoting, and even correlate with CCA survival [88,89]. In the liver, MFs are derived from – for example, activated hepatic stellate cells, which transition is stimulated by Hh signaling [90]. The MF-to-CCA cross-talk involves several signaling pathways, PLK [83], PDGF [82,91], Hh [92] and Notch [84,92]. SHH expression was shown to regulate both PDGF-BB signaling [93] and Jagged-1 [92]. Recently, in a model of bile duct ligation, Xie et al. demonstrated that impaired Hh signaling inhibited the Notch pathway, and that Hh and Notch cooperate to control cell fate in adult liver repair [84]. Overexpression of both Hh and Notch has been implicated in chemoresistance and, as such, blocking Hh signaling may enhance therapeutic efficacy [88]. Preclinical models have shown that Notch signaling is critical in cholangiocarcinogenesis [94,95] and that γ-secretase inhibitors blocking Notch signaling may reduce tumor growth [94] and also attenuate liver fibrosis by affecting the gene profile in MFs [96]. In addition, macrophages are involved in the promotion of liver inflammation and profibrotic signals. Interestingly, activated hepatic stellate cells were recently demonstrated to cause the differentiation of macrophages [97]. The interplay between inflammation, profibrotic signals and tumorigenesis mediated in part through MFs and tumor-associated macrophages may present promising treatment options in the future.

▪ IL-6/STAT3 signaling cascade

Binding of IL-6 to its receptor (gp130) triggers the heterodimerization with the Janus kinases (JAK1, JAK2 or TYK2), thus facilitating a signaling cascade and activation of STAT3 – that is, the JAK/STAT pathway – and/or the MAPK pathway.

In experimental models, IL-6 was found to function in an autocrine manner, leading to increased expression of the antiapoptotic protein Mcl-1, TRAIL resistance through AKT and growth [98–100]. Continued activation of IL-6/STAT3 signaling was linked to epigenetic silencing of the key inhibitor of STAT3, SOCS3 [101], whereas restoring the SOCS3 expression using a demethylating drug sensitized CCA to TRAIL-induced apoptosis [101]. Inhibition of IL-6, by incubating cells with a neutralizing antibody, abrogated proliferation in an ERK and p38 MAPK-dependent pathway [100,101]. As part of the inflammatory response, COX2 expression is often increased in CCA with poor prognosis, and correlates with both EGFR and HER2. COX2-derived prostaglandin (PGE₂) induces IL-6 levels, suggesting a link between the COX2 and IL-6/STAT3 signaling pathways [102]. Targeting COX2 downstream of the RTKs, MAPK, IL-6/STAT3 and PI3K/AKT/mTOR pathways is very compelling and celecoxib has successfully been used to inhibit CCA cell proliferation and reduce tumor growth in preclinical models [103–105].

The expression of the proinflammatory cytokine IL-6 is often increased in CCA as well as in serum and bile from CCA patients [99]. Serum level of IL-6 correlates with tumor size, and following photodynamic therapy the level of IL-6 is decreased, suggesting that IL-6 may be used both as a metric of tumor size and response. Recently, the JAK/STAT pathway was estimated to affect more than 70% of the inflammation subclass in iCCA [18]. This subclass corresponds to the patient subgroup with good prognosis and is characterized by enrichment in cytokine pathway signatures, overexpression of IL-6 and activation of STAT3 [17,18,25,106]. Sia et al. proposed targeting the JAK/STAT pathway, taking advantage of novel STAT3 and JAK1–JAK2 inhibitors [18]. JAK inhibitors such as AZD1480 have been shown to potently and concomitantly block JAK/STAT signaling [107], as well as the activation of JAK2, STAT3, MAPK and Fgfr3 by phosphorylation [108].

▪ The Hippo pathway

The Hippo signaling pathway is a novel tumor-suppressor pathway involved in development and stem cell biology [109]. Dysregulation of this pathway is implicated in tumorigenesis [110], and deletion of Mst1/2 was recently shown to cause both HCC and CCA [111]. Inactivation of Mst1/2 results in activation of the oncogene YAP1 and resistance to FAS-induced apoptosis. Interestingly, acute inactivation of Mst1/2 results in activation of mTORC1 and increased mTORC1-dependent phosphorylation of p70 S6 Thr389/412 phosphorylation. It was also suggested that overexpression of YAP1 may mediate activation of Notch and Wnt signaling through β-catenin transcriptional activity and induce the expression of downstream Notch targets. This suggests a cross-talk between these pathways, which needs to be investigated further.

Recently, Sia et al. examined the landscape of chromosomal aberration in iCCA [18]. The majority of regional CN alterations showed no apparent accumulation when analyzed compared with the prognostic gene classification. However, one interesting focal deletion containing the Hippo pathway gene, SAV1, which is enriched in the proliferation subclass, was also significantly associated with reduced SAV1 gene expression. Lately this pathway has attracted attention in numerous genetic studies, detailing its involvement in primary liver cancers [112]. However, validation of this may be required before it warrants a large translational study into the Hippo network in iCCA. To date, no Hippo–YAP-targeted therapy/inhibitor is available; however, recent evidence suggests a potential role for mTOR inhibitors in Hippo-dependent cancers [113].

Conclusion

CCA is an orphan malignancy with no current approved therapy. Currently, standard chemotherapy provides improved survival and the promise of potential downstaging, which would make surgery with curative intention possible. In the past few years, our understanding of the underlying biology of this heterogeneous cancer type has improved and provided target-enriched patient classification. While the molecular understanding of this malignancy is improving, clinical trials are still designed in mixed BTC cohorts. Besides the need to consider the anatomical locations of the tumor, the ethnic background of the patient may further complicate the clinical decision-making by differences in key genetic variants – for example, KRAS, PIK3CA, TP53, IDH1 and IDH2 – suggesting the need to involve translational genomics-driven evaluation as a part of the clinical management.

We have suggested several emerging molecular pathways with compelling treatment options, which are encouraging, but require further investigation. These include targeting ‘oncogenic drivers’, for example, mutations in the IDH1 and IDH2 genes [114,115]. Alternatively to targeted therapy, since IDH genes influence epigenetic control, this may suggest a potential benefit of DNA methyl transferase inhibitors. In a preclinical translational genomics/epigenomics study, we recently demonstrated the effect of zebularine on HCC and CCA [116]. Other promising pathways, in addition to the classical growth factor signaling such as FGF receptor and IGF receptor, include Wnt/β-catenin, Notch, Hippo, Hh and PLK. These pathways may involve targeting the microenvironment – that is, the stromal compartment of the tumor, including inhibition of signals from MFs and tumor-associated macrophages. These interlinked networks are complex and require further investigation. However, these observations emphasize the benefit and need to treat the underlying inflammation, targeting, for example, NF-κB (curcumin or bortezomib) or COX2 (celecoxib). Dysregulated expression of Wnt/β-catenin and genetic alterations – for example, CTNNB1, AXIN1 and APC – are rare events in CCA, compared with HCC [117,118]. Currently, the only exome-sequencing project in CCA is in a small group of patients whose disease is related to liver flukes [19]. We are confident that additional sequencing projects will significantly enrich our understanding of ‘drivers’ and concurrent genetic variants in this malignancy, and that this may advance patient classification, enlighten future clinical trial designs and improve outcome of this disease.

Future perspective

Traditional clinical trials, and consequently treatment schemes, against CCA with conventional chemotherapy have, at best, resulted in modest successes. Characteristically, the disease is ‘clinically silent’, presenting at a late stage where surgical resection is not an option and with limited or no therapeutic options. The clinical challenges for CCA, besides its treatment-refractory nature, gross anatomical and tumor heterogeneity and late-stage diagnosis, resides in essentially unknown disease etiology; limited understanding of the genetic, genomic and epigenomic abnormalities in the pathogenesis of the disease; and lack of standardized and well-powered clinical trials. To achieve successful outcome, future clinical trial designs should require multicenter, homogeneous and well-powered cohorts; biopsies taken pre- and post-therapy to extract maximum molecular information; consider possible metastatic sites, as these tissues may provide a ‘window’ into potential tumor clones, which may determine/alter treatment decisions; and include translational genomics specialists as a part of the multidisciplinary clinical decision team. These approaches should resolve if future clinical success will result from treatment of class-specific patient subgroups with innovative multidrug approaches combining, for example, conventional cytotoxic chemotherapy with well-designed precision therapy targeting several molecular pathways simultaneously to achieve drug synergy or, if indeed, the future will bring well-designed treatment approaches tailored for the individual patient.