Cholangiocarcinoma: what are the most valuable therapeutic targets – cancer-associated fibroblasts, immune cells, or beyond T cells?

Abstract

Introduction:

CCAs are dense and desmoplastic tumors with an abundant tumor microenviroment (TME). The evolving TME is characterized by reciprocal interactions between cancer cells and their environment and is essential in facilitating tumor progression. The TME has nonimmune and immune components. Nonimmune cell types include cancer-associated fibroblasts (CAFs) and endothelial cells accompanying tumor angiogenesis. Immune cell types include elements of the innate and adaptive immune response, and can have pro-tumor or antitumor roles. The TME can shape treatment response and resistance. Therefore, elements of the TME are attractive therapeutic targets. TME targeting therapies have been evaluated in preclinical and clinical studies but only a small subset of patients has a meaningful response.

Areas covered:

We discuss the TME components and potential TME targeting strategies. Literature search was performed on PubMed and ClinicalTrials.gov until October 2021.

Expert opinion:

Elucidating the CCA TME is essential for developing effective treatment strategies. Preclinical models that recapitulate the disease (such as organoids) are important tools in uncovering the intricate cross talk in the CCA TME. Characterization of patient-derived specimens using multi-omic and single-omic technologies can dissect the cellular interplay in the CCA TME, which can guide development of effective treatment strategies and identify biomarkers for patient stratification.

1. Introduction

Cholangiocarcinomas (CCAs) are heterogeneous epithelial tumors with features of biliary tract differentiation (e.g. cells that resemble cholangiocytes and form glands). CCAs are classified into three subtypes based on their anatomic location within the biliary tree: intrahepatic (iCCA), perihilar (pCCA), and distal CCA (dCCA) [1]. CCAs are the most common biliary malignancy and the second most common hepatic malignancy. CCA risk factors include underlying inflammatory conditions such as primary sclerosing cholangitis and choledocholithiasis, as well as environmental factors [2]. CCAs have a low 5-year survival (7–20%), in part related to diagnosis typically occuring at an advanced disease stage [1]. Advanced disease at presentation precludes potentially curative surgical treatment options [1]. For patients with advanced disease, the standard of care remains systemic chemotherapy with gemcitabine and cisplatin [3]. However, the overall survival with this combination is only 11.7 months [3].

CCAs are dense and desmoplastic tumors with an abundant tumor microenviroment (TME). The TME is complex and diverse across a variety of cancers. The constantly evolving TME is characterized by reciprocal interactions between cancer cells with their environment and is essential in facilitating tumor progression. The TME has nonimmune and immune components. Nonimmune cell types include cancer-associated fibroblasts (CAFs) and endothelial cells accompanying tumor angiogenesis. Immune cell types include elements of the innate and adaptive immune response, and can have pro-tumor or antitumor roles (Figure 1). The CCA TME is rich and has abundant immunosuppressive elements such as CAFs and myeloid cells. The TME can shape treatment response and resistance. Therefore, elements of the TME are an attractive therapeutic target [4]. In this review, we will discuss the TME components and potential TME targeting therapeutic strategies. Relevant literature was searched by combining the keywords ‘cholangiocarcinoma’ or ‘biliary tract cancer’ or ‘hepatobiliary malignancy’ with the following keywords: ‘tumor microenviroment,’ ‘cancer associated fibroblast,’ ‘ vascular endothelial cells,’ ‘macrophages,”myeloid-derived suppressor cells,’ ‘T cells,’ ‘dendritic cells,’ ‘NK cells,’ ‘NKT cells,’ ‘MAIT,’ ‘B lymphocytes’ and ‘immunotherapy’ until October 2021.

Figure 1. Tumor microenvironment of cholangiocarcinoma.

The TME of CCA has nonimmune and immune components. The immune components include both antitumor and pro-tumor immune cells. Antitumor adaptive immune cells include CD4⁺ T cells, CD8⁺ T cells, and B cells. Antitumor innate immune cells include natural killer cells and dendritic cells. Pro-tumor immune cells include tumor-associated macrophages, myeloid-derived suppressor cells, regulatory T cells, and MAIT cells. Nonimmune cells in the CCA TME include cancer-associated fibroblasts and tumor-related vascular endothelial cells. CCA, cholangiocarcinoma; MAIT, mucosal-associated invariant T cells; TME, tumor microenvironment.

2. Tumor Nonimmune Microenvironment of CCA

2.1. Therapeuting Targeting of CAFs

CAFs are activated fibroblasts that express alpha-SMA, and foster a pro-tumor microenvironment [5]. CAFs exert pro-tumorigenic effects either directly on tumor cells or indirectly by recruitment of immunosuppressive immune cells via production of a variety of factors including growth factors, cytokines, and soluble factors [6–9]. Indeed, genetic deletion of CAFs suppresses CCA development and growth in murine models [10]. The platelet-derived growth factor (PDGF)/PDGF-receptor (PDGFR) axis facilitates cross talk between CAFs and CCA cells. Myofibroblast-derived PDGF-BB activates the hedgehog pathway and protects CCA cells from tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-mediated cytotoxicity [11]. Moreover, CCA cell–secreted PDGF-D enhances myofibroblast recruitment via PDGFRβ, and promotes early tumor metastasis [12,13]. Both PDGF-B and D isoforms increased the sensitivity of myofibroblasts to apoptosis [14]. The BH3 mimetics, navitoclax and ABT-199, leveraged this apoptotic priming to induce CAF apoptosis with consequent reduction in CCA tumor burden [14] (Figure 2). Placental growth factor (PIGF), a member of the vascular endothelial growth factor (VEGF) family, is another growth factor implicated in CCA progression. PIGF, which is overexpressed in CCA and secreted by iCCA cells, shifts hepatic stellate cells (HSCs) to a myofibroblast-like phenotype by activating the AKT/NF-KB pathway [6]. Meanwhile, CAF-secreted PIGF enhances CCA cell invasion, an effect that is abolished by PIGF blockade. In a murine model of CCA, PIGF blockade resulted in a reduction of the tumor stiffness and augmented vascular perfusion, effects that contributed to increased sensitivity to chemotherapy and prolonged murine survival [6]. Tumor stiffness, which is largely modulated by the extracellular matrix (ECM), can function as a barrier to T cell infiltration. Inhibition of lysyl oxidase (LOX), which crosslinks collagen, reduced ECM content and augmented intratumoral T cell mobilization in a human CCA subcutaneous model. Inhibition of LOX augmented the efficacy of anti-PD-1, providing rationale for approaches that combine immunotherapy with ECM targeted therapy [15].

Figure 2. Therapeutic targeting of cancer-associated fibroblasts in CCA.

Strategies for targeting CAFs include BH3 mimetics, PIGF blockade, and GIONF. PDGF-B and D-sensitive CAFs to apoptosis. Addition of BH3 mimetics such as ABT-199 leads to CAF apoptosis. GIONF also exert their antitumor effect via selective targeting of CAFs. PIGF blockade reduces tumor matrix and enhances sensitivity to chemotherapy. CAF, cancer-associated fibroblasts, GIONF; iron oxide nanoflowers decorated with gold nanoparticles; PIGF; placental growth factor; PDGFR, platelet-derived growth factor receptor.

Inhibition of hedgehog signaling has been investigated as an antitumor approach that selectively depletes myofibroblasts. HSCs can stimulate CCA cell proliferation, migration, and invasion via hedgehog signaling activation. Accordingly, the hedgehog antagonist cyclopamine enhanced susceptibility of CCA tumors to necrosis with attenuation of tumor growth in a mouse model [16]. One way in which CAFs can stimulate migration and invasiveness of CCA cells is by secretion of interleukin- (IL)-6. Resveratrol, a nutraceutical, blocks the secretion of IL-6 from CAFs [17].

Mechanical strategies have also been employed to deplete CAFs from the TME in an effort to reduce tumor burden. Multifunctional iron oxide nanoflowers decorated with gold nanoparticles (GIONF) modulate the TME by inducing light-activated nanohypothermia. GIONF are preferentially taken up by CAFs in CCA, which results in selective targeting of CAFs via GIONF-mediated photothermal therapy [18]. Mice bearing tumors derived from the human CCA cell line EGI-1 were treated with GIONF with consequent tumor regression [18]. In aggregate, these studies indicate that the cross talk between CAFs and CCA cells promotes CCA progression. Consequently, depletion of CAFs has therapeutic potential for the treatment of CCA.

2.2. Tumor-Related Vascular Endothelial Cells

Microvessel density (MVD) is an indicator of tumor-associated angiogenesis. In CCA, high MVD is characterized by intratumoral staining of CD31⁺, an endothelial specific antibody, and is positively correlated with advanced tumor stage and higher rate of recurrence in patients [19,20]. However, the data regarding the prognostic role of MVD in CCA are conflicting as a low MVD has also been associated with poor patient outcomes, low CD8⁺ T cell infiltration and higher infiltration of regulatory T cells (Tregs) [21].

Vascular endothelial growth factor receptor (VEGFR) promotes angiogenesis and invasiveness. High mobility group box 1 (HMGB1) released from cancer cells upregulates VEGFR-2 expression in CCA, resulting in tumor angiogenesis [22]. Apatinib, a tyrosine kinase inhibitor targeting VEGFR-2, suppresses CCA proliferation, migration, and angiogenesis [23]. These antitumor effects of apatinib are a consequence of VEGFR2/STAT3/HIF-1α axis inhibition. VEGF/VEGFR2 signaling also exerts anti-apoptotic effects in CCA via activation of the PI3K-Akt-mTOR signaling pathway. Apatinib inhibits this effect and promotes CCA cell apoptosis in vitro [24]. Given it has shown promsing effecacy in preclinical studies, aptinib has been evaluated in early phase human clinical trials. In a prospective, open-label phase II trial, patients with advanced CCA who received apatinib (n = 26) had an objective response rate (ORR) of 11.5% and disease control rate (DCR) of 50% [25]. Apatinib was evaluated in another prospective, open-label phase II trial of patients with advanced CCA who had progressed on gemcitabine-based systemic therapy [26]. In 24 patients, apatinib had an ORR of 20.8% and DCR of 62.5%. Although there have been a number of preclinical studies that have examined the protumor effects of CCA angiogenesis, the results of human clinical trials investigating anti-angiogenic drugs have been disappointing. A comprehensive and in-depth understanding of the complex biology of tumor vascularization is needed to design optimal therapies targeting angiogenesis in CCA.

3. The Tumor Immune Microenvironment of CCA

3.1. Pro-tumor, Immunosuppressive Immune Cells

3.1.1. Tumor-associated Macrophages

Tumor-associated macrophages (TAMs) play an essential role in tumor development and progression through extensive cross talk with other cell TME cell types. In tumor biology, macrophages can have both pro-tumor and antitumor activity. However, the preponderance of data suggests a protumor effect [27]. A higher proportion of TAMs is generally associated with poor patient outcomes in CCA. For instance, increased infiltration of CD163⁺ macrophages is associated with poor progression-free survival (PFS) and extrahepatic metastasis in iCCA [28,29]. However, CD163⁺ macrophage infiltration was not associated with poor prognosis in pCCA/dCCA [30]. Patients with increased infiltration of CD86^lowCD206^high TAMs have a significantly worse prognosis with higher risk of recurrence following surgical resection [31].

The tumor-promoting role of TAMs has been extensively studied. Kupffer cells, the resident macrophages in the liver, promote cholangiocellular proliferation and oncogenic transformation in a JNK-dependent manner [32]. Although resident macrophages foster tumor development and growth in CCA, the preponderance of TAMs in CCA are recruited [33]. The TAM pool is largerly replenished with recruited macrophages induced by chemokine (C-C motif) ligand 2 (CCL2) and colony stimulator factor 1 (CSF-1) [34]. Hence, targeting macrophage recruitment may have therapeutic potential. However, depletion of recruited macrophages from the TME by targeting CSF1-R or CCR-2 failed as an antitumor strategy in murine CCA due to compensatory emergence of myeloid-derived suppressor cells (MDSCs) [33] (Figure 3).

Figure 3. Therapeutic targeting of immune cells in CCA.

Strategies for targeting pro-tumor immune cells and antitumor immune cells are depicted. Inhibition of pro-tumor immune cells ((TAMs, MDSCs, Tregs) can reduce their abundance and lead to enhanced T cell cytotoxicity. Therapeutic strategies for targeting TAMs include anti-CSF1R and anti-CD47. LXR agonism or CXCR2 blockade can reduce MDSC abundance, while anti-CD25 attenuates Treg accumulation. Strategies to augment antitumor immune cells include ICI such as anti-PD-1 and anti-CTLA-4. CSF1R, colony-stimulating factor 1 receptor; LXR, liver X receptor; MDSC, myeloid-derived suppressor cells; TAM, tumor-associated macrophage; Treg, regulatory T cell.

An alternate strategy to therapeutically target the antitumor effect of TAMs is to promote macrophage phagocytosis (Figure 3). Cluster of differentiation 47 (CD47), also known as the ‘don’t eat me’ signal, is an innate immune checkpoint that is overexpressed on cancer cells [35]. CD47 binds to signal regulatory protein alpha (SIRPα), a myeloid inhibitory immunoreceptor that is expressed on macrophages. The activation of the CD47/SIRPα axis inhibits the destruction of cancer cells by phagocytes such as macrophages and neutrophils [35]. CD47 has a higher expression in CCA compared to hepatocellular carcinoma (HCC). In a model of liver metastasis, blockade of the CD47-SIRPα interaction by anti-CD47 lead to attenuation of tumor initiation [36]. In a phase II trial of patients with aggressive and indolent lymphoma, the combination of Hu5F9-G4, a CD47 targeting antibody, and rituximab had an ORR of 50%, with 36% of patients having a complete response [37]. These promising results need to be validated in additional human studies. Macrophage receptor with collagenous structure (MARCO) is another macrophage immune checkpoint [38]. MARCO, a scavenger receptor, is associated with a subtype of immunosuppressive TAMs with high expression in M2-like TAMs [39]. Anti-MARCO antibody augments immune checkpoint inhibition (ICI) in preclinical models of melanoma and colon cancer [35]. Targeting MARCO can also facilitate tumor killing by natural killer (NK) cells [40]. Despite these reports of an immunosuppressive, pro-tumor role of MARCO, studies in HCC have demonstrated that MARCO expression is decreased in HCC and high intratumoral MARCO expression is associated with a favorable prognosis including significantly improved overall survival [41,42]. Thus, the role of MARCO in CCA remains to be defined.

3.1.2. Myeloid-Derived Suppressor Cells

Myeloid-derived suppressor cells are a heterogenous population of immature myeloid cells that are classified into two broad subsets: granulocytic (G-MDSCs) or polymorphonuclear (PMN) MDSCs and monocytic MDSCs (M-MDSCs) [43]. In mice, G-MDSCs are CD11b⁺Gr1^hi (Ly6C^lowLy6G^high) whereas M-MDSCs are CD11b⁺Gr1^hi (Ly6C^highLy6G^low) [43]. MDSCs support tumor growth by suppressing function of cytotoxic cells such as CD8⁺ T cells and NK cells, while fostering tumor cell proliferation. High levels of circulating MDSCs confer resistance to ICI [44]. Accordingly, MDSCs are an attrative therapeutic target in the tumor immune microenvironment. Activation of liver X receptor (LXR), a ligand-activated transcription factor, reduces MDSC survival and abundance [45]. The LXR agonist, RGX-104, repressed MDSCs and augmented antitumor immunity in preclinical models of melanoma, glioblastoma, and lung cancer. Moreover, preliminary results demonstrated that RGX-104 had promising antitumor activity in a first-in human, dose escalation phase I trial [45]. This trial is currently ongoing in patients with advanced solid tumors (NCT02922764). Emerging data indicate that targeting MDSCs in CCA may be a viable therapeutic approach. In a murine model of CCA, TAM blockade resulted in a compensatory emergency of G-MDSCs [33]. A therapeutic approach targeting both G-MDSCs and TAMs reduced tumor burden and augmented efficacy of ICI using anti-programmed cell death protein 1 (PD-1). In this study, pharmacologic targeting of MDSCs employed both anti-Ly6G as well as the LXR agonist, GW3965 [33].

Primary sclerosing cholangitis (PSC) is a known risk factor for CCA, particularly in the Western world. In murine models of colitis as well as murine models that mimic a PSC-like state, an increase in CXCR-2⁺ PMN-MDSCs with immunosuppressive properties was observed [46]. Gut dysbiosis with a decrease in gut barrier function induced expression of the chemokine CXCL1 by hepatocytes with consequent accumulation of CXCR-2⁺ PMN-MDSCs. CXCR-2 blockade using SB225002 reduced PMN-MDSCs infiltration and murine CCA tumor burden [46]. These studies suggest that MDSCs are an essential immunosuppressive myeloid cell population and targeting MDSCs can augment ICI in CCA (Figure 3).

3.1.3. Regulatory T cells

Regulatory T cells (Treg) are another major immunosuppressive element in the tumor immune microenvironment, contributing to tumor immune escape and resistance to immunotherapy. An elevated intratumoral Treg denstity is associated with poor overall survival in several malignancies including CCA [47,48]. Thus, targeting Tregs has the potential to restore and enhance the antitumor immune response. In addition to their lineage marker Foxp3, Treg express several surface marker including CD25 and cytotoxic T lymphocyte-associated protein 4 (anti-CTLA-4). Tumor-infiltrating Tregs have high CD25 expression. CD25, also known as the IL-2 receptor alpha subunit (IL-2Rα), has been proposed as a target for cancer immunotherapy as it can be selectively targeted to deplete Tregs. In preclinical models of cancer, anti-CD25 antibody depleted Tregs in the tumor immune microenvironment and synergized with anti-PD-1 to reduce tumor burden [49]. However, in human clinical trials, anti-CD25 antibodies have failed likely due to off-target effects on IL-2 receptor signaling on effector CD8⁺ T cells [50,51]. An optimized human anti-CD25 clone (RG6292) demonstrated efficient Treg depletion without any overt off-target effects on effector T cells [50]. A multicenter, phase I clinical trial of human anti-CD25 monoclonal antibody is currently under investigation in solid tumors (NCT04158583) [50]. Although there is limited evidence to-date on therapeutic targeting of Tregs in CCA, the existing body of literature in tumor immunology suggests that Tregs depletion has therapeutic potential in CCA.

3.2. Other Innate and Innate-Like Immune Cells

3.2.1. Dendritic Cells

Dendritic cells (DCs) are antigen-presenting cells that take up and present antigens to adaptive immune cells. DCs present antigens on MHC-II molecules to CD4⁺ T cells and then cross-present (present exogenous antigens on MHC class I molecules to initiate CD8⁺ T cell response) to CD8⁺ T cells [52]. DCs can be classified into two categories: classical or conventional DCs (cDCs) and plasmacytoid DCs (pDC). cDCs are further divided into two subsets: (i) cDC1, characterized by CD141 in humans and CD103 in mice and (ii) cDC2, characterized by CD1c in humans and CD11b in mice [53]. DCs play a major role in augmenting the antitumor response, and the absence of DCs or the presence of dysfunctional DCs confers a poor outcome. Thus, increasing DC density and/or restoring the function of DC can be viewed as potential therapeutic approaches for the treatment of malignancies including CCA. Recruitment of CD103⁺ DCs by chemokine CC motif ligand 4 (CCL4) amplifies the CD8⁺ T cell response and enhances the antitumor effect of immune checkpoint inhibition in murine models of melanoma and breast cancer [54]. The growth factor FMS-like tyrosine kinase 3 ligand (FLT3L) is involved in the maturation and differentiation of DCs [55] (Figure 3). Treatment of mice with recombinant FLT3L dramatically increased infiltration of DCs into tumor sites resulting in significant repression of tumor growth [56]. CD40, a TNF receptor superfamily member, is expressed on DCs and has a role in the activation of antigen-presenting cells. Higher expression of CD40 is associated with improved patient outcomes in CCA. Combined CD40 agonism and FLT3L modulated the tumor immune response with a marked influx of DCs, robust CD8⁺ T cells infiltration, as well as notable increase in NK, NK T cells (NKT) and γδT cells in pancreatic ductal adenocarcinoma tumors [57]. Similarly, in murine models of CCA combination anti-CD40 and anti-PD-1 therapy reshaped the antitumor immune response with increased infiltration and activation of CD4⁺ and CD8⁺ T cells and NK cells with resultant reduction in tumor burden [58]. Furthermore, this combination enhanced the efficacy of gemcitabine/cisplatin, first-line systemic therapy in CCA, with a greater survival benefit. Based on these encouraging results, a phase I clinical trial of the CD40 agonist CDX-1140 alone or in combination with either FLT3L (CDX-301) or anti-PD-1 (Pembrolizumab) in solid organ malignancies including CCA is currently ongoing (NCT03329950).

Vaccines directed at DCs are another strategy to promote antitumor immunity. DC vaccines are typically pulsed with tumor-associated antigen (TAA) in vitro and then injected in vivo. Several TAAs have been investigated in CCA [59–61]. Self-differentiated monocyte-derived DCs presenting cAMP-dependent protein kinase type I-alpha regulatory subunit (SD-DC-PR) activated T cells for CCA cell killing. SD-DC-PR activated autologous effector T cells had enhanced cytotoxic activity compared to T cells activated by conventionally derived DCs [61]. In small clinical studies in PDAC patients, DCs loaded with Wilms’ tumor 1 (WT1) and/or mucin1 (MUC1) peptide enhance effector T cell responses [62,63]. Further studies are needed to assess the safety and efficacy of DC viruses in CCA and to investigate DC vaccines in combination with other immunotherapies.

3.2.2. NK Cells, NKT Cells, and MAIT Cells

NK cells are cytotoxic innate immune cells that play an essential role in maintenance of homeostasis. NK cells participate in antitumor response by producing cytokines such as granzymes, perforin, and interferon-γ. However, NK cell function is typically impaired in the context of malignancy. Similar to cytotoxic T cells, NK cells express a variety of inhibitory and activating receptors. The primary activating receptors are natural killer group 2D (NKG2D) and natural cytotoxicity receptors, which include NKp30, NKp46, and NKp44 [64]. High expression of NKG2D ligands (NKG2D-L) on tumor cells is associated with improved disease-free survival and overall patient survival in human CCA, suggesting that regulating the NKG2D/NKG2D-L axis has the potential to curb CCA progression [65]. Approaches to increase NKG2DL expression include utilization of histone deacetylase (HDAC) inhibitors such as valproic acid, and prevention of NKG2D-L shedding from the cell membrane by the metallopeptidase ADAM10 or matrix metalloproteinase inhibitors [66–68].

Inhibitory NK receptors include killer cell immunoglobulin-like receptors (KIRs) [69]. Patients with CCA have multiple alterations at the KIR gene loci; these alterations may impact NK cell function in tumor surveillance [70]. Accordingly, immunotherapy targeting KIR on NK cells may hold therapeutic potential. However, a phase I study of the combination of nivolumab plus either ipilimumab or the KIR inhibitor lirilumab in relapsed/refractory lymphoid malignancies did not demonstrate any meaningful benefit of combinatorial therapy over single-agent nivolumab [71]. An early phase study of nivolumab in combination with lirilumab or ipilimumab in advanced/metastatic solid organ malignancies including CCA is currently ongoing (NCT03203876). Thus, further preclinical and clinical data are needed to determine whether there is therapeutic potential of targeting KIRs in CCA.

NKT cells are innate-like immune cells that express surface receptors for NK cells, part of the characteristic of innate immunity, but also express a T-cell receptor (TCR), characteristic of adaptive immunity [72]. NKT cells have potent antitumor activity including in liver cancer. Moreover, modulation of the gut microbiome can promote accumulation of NKT cells with a resultant antitumor effect in HCC as well as metastatic liver tumors [73].

Mucosal-associated invariant T (MAIT) cells express an invariant TCRα chain. MAIT cells comprise up to 45% of the lymphocytes in the liver [74]. In a variety of cancers, a high NK cell signature is associated improved progression-free survival. However, a high MAIT cell infiltration, as assessed by a MAIT cell gene signature in publically available TCGA datasets of several malignancies including CCA, is associated with poor outcomes in patients with a high NK cell infiltration. These results imply that accumulation of MAIT cells may attenuate NK cell-mediated antitumor immunity in CCA [75].

3.3. Antitumor Adaptive Immune Cells

3.3.1. CD8⁺ T lymphocytes

CD8⁺ T cells are cytotoxic adaptive immune cells that play a central role in the antitumor response. CCAs have an infiltration of dysfunctional CD8⁺ T cells characterized by expression of exhaustion markers such as PD-1, CTLA-4, lymphocyte activation gene 3 (LAG-3), T-cell immunoglobulin and mucin domain-3 (TIM-3), and T cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT) [76,77]. Strategies aimed at restoring the antitumor, adaptive immune response include targeting inhibitory checkpoints on the T cell receptor, and adoptive transfer of engineered T cells. However, ICI monotherapy has had limited efficacy in human CCA. In the KEYNOTE-158 trial, pembrolizumab, an anti-PD-1 antibody, had an ORR of 5.8% in patients with advanced CCA (6/104; 95% CI 2.1%–12.6%). Tumor PD-L1 expression did not impact the response rate [78]. In the KEYNOTE-028 trial, ORR was 13% (3/23; 95% CI 2.8%–33.6%) [78]. In a phase 2 multicenter study of 54 patients with advanced CCA that had progressed on at least one line of treatment, nivolumab, an anti-PD-1 antibody, had an ORR of 11% based on an independent review [79]. In a prospective, multicenter, non-randomized phase I trial of patients with advanced rare cancers including CCA and gallbladder carcinoma, the combination of nivolumab and ipilimumab, an-anti-CTLA-4 antibody, had an ORR of 23%. The responders had either gallbladder carcinoma or iCCA; no responses were observed in the patients with pCCA and/or dCCA (n = 10). These results imply that each subtype may have a unique tumor immune microenvironment, and response to immunotherapy may differ based on the anatomic subtype. There are several ongoing clinical trials assessing dual immune checkpoint blockade in CCA (NCT03704480, NCT03473574, NCT03046862). Co-glucocorticoid-induced tumor necrosis factor receptor (GITR), a co-stimulatory molecule, is also overexpressed on tumor-infiltrating lymphocytes (TILs) in CCA, implying that agonistic targeting of GITR may have therapeutic implications in CCA [80]. However, preliminary results of clinical trials investigating GITR agonists have demonstrated limited antitumor response in advanced solid tumors [81,82]. These results indicate that monotherapy has limited efficacy in CCA, and combinatorial approaches targeting the various components of the CCA TME are needed.

Data on adoptive T cell therapy in CCA are limited. The combination of adoptive transfer of T cells (activated CD3⁺ T cells) plus a dendritic cell vaccine was assessed in the adjuvant setting [83]. The median PFS and OS were 18.3 months and 31.9 months, respectively, in patients receiving adjuvant therapy compared to those who underwent surgical resection alone. A phase II clinical trial assessing autologous TILs in advanced biliary tract cancer is currently ongoing (NCT03801083).

3.3.2. CD4⁺ T Cells

CD4⁺ comprise a subset of TILs, and they have an antitumor role via secretion of cytokines. Correlative studies suggest that CD4⁺ infiltration is associated with significant longer overall survival in patients with CCA [84]. Whole-exome sequencing in a patient with metastatic CCA, demonstrated CD4⁺ T helper 1 cells that recognized a specific mutation expressed by the tumor. Adoptive transfer of TILs containing approximately 25% of the mutation-specific T(H)1 cells decreased tumor burden and prolonged stabilization of disease [85]. Nonetheless, the existing knowledge regarding the role of CD4⁺ T cells in CCA is limited, and further work is necessary to elucidate their function in the CCA TME.

4. ‘B’eyond T cells: B lymphocytes

In contrast to T lymphocytes, the role of B lymphocytes in CCA development and progression is far less clear. B cells have an integral role in humoral immunity via the production of immunoglobulins. Upon antigen recognition, naïve B cells become activated B cells and subsequently differentiate into plasma cells, which are capable of antibody production [86]. There are several subsets of B lymphocytes, including pro-tumor or regulatory B cells that can dampen the antitumor immune response and promote tumorigenesis. In pancreatic ductal adenocarcinoma, IL-35 producing B cells stimulated tumor cell proliferation [87]. This B cell subset was recruited to the tumor by the chemoattractant CXCL-13, and CXCL13 blockade resulted in a decline in this population with reduction of tumor growth [87]. Although these findings imply that B cell–based immunotherapeutic approaches may hold promise, the role of B lymphocytes in CCA has yet to be elucidated.

5. Conclusion

There have been significant advances in our understanding of the TME of CCA. A variety of TME targeting therapies have been evaluated in preclinical and clinical studies. However, to-date only a small subset of patients have had meaningful response to these therapies. ICI monotherapy in CCA has been disappointing, and the mechanism of resistance have not been defined. This highlights the intricacies of the CCA TME and underscores the significance of the extensive cross talk that is present. A more in-depth understanding of the CCA TME will guide the development of combinatorial approaches targeting multiple facets of the CCA TME.

6. Expert Opinion

CCA is a highly lethal malignancy with limited treatment options. The majority of patients present at an advanced disease stage and are not eligible for potentially curative surgical treatment options. The standard of care for advanced or metastatic CCA is systemic chemotherapy with gemcitabine and cisplatin. However, the overall survival with this combination is less than a year. Thus, there is a critical need for the development of effective systemic therapies that can prolong patient survival. Immuno-oncology has transformed treatment of a variety of cancers over the past decade. However, the efficacy of ICI monotherapy in CCA has been disappointing. This is likely due to an abundance of immunosuppressive elements in the CCA TME that mediate tumor evasion. Thus, it is essential to delineate the CCA TME and the cross talk between cancer cells, immune cells, and stromal components. This in turn can guide the development of effective therapies that can target various components of the TME, and perhaps augment the efficacy of ICI. An enhanced understanding of the CCA TME requires robust models, utilization of multi-omics to dissect the cellular source of TME signals, and identification of key targets.

6.1. Models for Examining TME Cross Talk in CCA

Human cell lines are derived from patient tumor tissues and broadly used in extensive in-vitro studies. However, in vitro studies examining the TME that employ cell lines are limited given that they are unable to mimic the interactive microenvironment of CCA with its extensive cross talk. Mouse models of CCA can overcome this limitation as they have an intact TME. However, patient-derived xenograft (PDX) mouse models are of limited utility in investigation of the immune microenvironment as they are established in an immunodeficient host. Syngeneic mouse models are more suitable for this purpose. In human CCA, genetic aberrations such as mutations, amplifications, and gene fusions are frequently detected [88]. In an effort to recapitulate the interaction between the tumor and TME in human CCA, tumor organoids have been established. CCA organoids are generated either from dissociated single cells or minced tissue of resected tumor derived from primary or metastatic tumors. Organoids can be cultured and expanded in a 3D matrix basement system. The histological organization of the native tumor is preserved in human CCA organoids. Moreover, the genomic landscape, gene expression profile, and tumorigenic potential of the original tumor is retained in the organoids. Thus, organoids can recapitulate primary human tumors and are an effective model to examine therapeutics [89]. Single cell-derived organoids can be manipulated to mimic genetical variation. Fibroblast growth factor receptor 2 (FGFR2) gene fusions occur in approximately 12–15% of human CCAs, primarily iCCAs [90,91]. FGFR2 fusion-driven CCAs were modeled in mouse Tp53^−/− liver organoids. These organoids were transplanted into immunodeficient mice and subsequent characterization demonstrated that these had phenotypic, histologic, and transcriptomic similarity to human CCA [92]. Combinatorial therapies were assessed in this model. However, similar to PDXs, organoids transplanted into immunodeficient mice lack an intact immune system, making it challenging to investigate the immune response ot the tumor. Multicellular organoids are comprised of malignant epithelial cells, immune cells, and stromal cells [93]. Such organoids can recapitulate the TME and facilitate investigation of immunotherapeutic combinations. However, CCA multicellular organoids have yet to be successfully generated.

6.2. Integrating Multi-omics to Dissect the Cellular Interplay in the CCA TME

CCAs are heterogeneous, dense desmoplastic tumors with a variety of cell types. Bulk omics such as RNA and DNA sequencing can detect broad signals in the tumor. However, bulk omic technologies cannot discern the cellular source of these signals. Identifying the cellular source is essential in the effort to develop effective therapies that target the TME, and single-cell technologies can help address this challenge. Such technologies include single-cell RNA sequencing, mass cytometry, imaging mass cytometry, and single-cell ProtEomics by mass spectrometry (SCoPE-MS). The latter examines the proteome, enabling analysis from transcriptional to translational regulation across single cells [94]. CITE-seq simultaneously profiles transcriptomics and proteomics in the same cell [95]. Single-cell omics can identify unique subsets of different cell types and examine signatures associated with specific cell types and has the potential to decode the heterogeneity of the CCA tumor ecosystem. Employing these technologies has the potential to elucidate cell–cell cross talk in the TME and identify novel biomarkers, which in turn can help stratify patients.

6.3. Developing New Combination Therapeutic Strategies

ICI monotherapy has had disappointing efficacy in CCA with response rate ranging from 6% to 11%. Phase 2 trial results of combination immune checkpoint inhibition in CCA have demonstrated only modest benefit with the addition of a second agent. In the KEYNOTE-158 trial, PD-L1 expression was not a robust biomarker of response [78]. Although microsatellite instability is a reliable biomarker of response to immunotherapy across a variety of malignancies, only a small subset of CCA patients have microsatellite instability. Hence, there is a need for robust biomarkers of response to TME targeting therapy in CCA [96]. Identifying subsets of patients who will respond to systemic therapy including immunotherapy in CCA remains a significant challenge. To address this challenge, an enhanced comprehension of the CCA TME is needed to develop combination strategies that target critical TME components. The TME of iCCA has been classified into four subtypes based on gene expression signatures of the TME: immune desert (I1), immune active or immunogenic (I2), myeloid rich (I3) and mesenchymal (I4) [97]. Correlating the subtypes with patient outcomes demonstrated that patients with an immune active subtype had the longest survival. The authors also concluded that patients with the immune active (I2) subtype may benefit from ICI therapy.

The subpar efficacy of ICI in CCA is likely due to an ‘immune cold’ TME with extensive cross talk between tumor cells and immunosuppressive elements of the TME. Combination strategies that target different aspects of CCA including the TME are needed. For instance, in a murine model of colorectal cancer with a highly angiogenic desmoplastic TME with exclusion of T cells, a therapeutic combination employing an anti-angiogenic agent (anti-VEGFA) and immunotherapy with a CD40 agonist resulted in a robust antitumor immune response, fibrosis remission, and tumor regression [98]. In a murine model of CCA characterized by an abundance of immunosuppressive cell types, dual targeting of TAMs and MDSCs augmented immune checkpoint blockade with anti-PD-1 with consequent prolongation of murine survival [33]. These studies exemplify the complex TME of desmoplastic cancers and make the case for combinatorial therapies targeting different aspects of the tumor and its TME. Ongoing clinical trials are assessing combination of ICI with chemotherapy (NCT04003636), tyrosine kinase inhibitors such as apatinib (NCT03092895), and radiation therapy (NCT03482102). The combination of ICI with TME targeting therapies such as anti-CSF1R (NCT03768531) and granulocyte-macrophage colony-stimulating factor (NCT02703714) is also being evaluated in human clinical trials. We eagerly await the results of these trials. Hence, well-designed and multifaceted combination strategies will likely incorporate integral components of the CCA TME.

In addition to the TME complexity, systemic factors need to be considered as well when designing combinatorial therapeutics. It has become apparent of late that factors beyond the tumor such as gut dysbiosis can alter the TME and affect response to systemic treatment. For instance, bacterial flora in the gut can impact the immune microenvironment by release of factors such as lipopolysaccharide. Hence, the CCA TME has tremendous complexity with extensive cross talk and multiple factors that can impact the tumor. In the future, enhanced characterization of tumors at baseline and post-treatment employing patient-derived specimens and multi-omic as well as single-omic technologies will be critical to deconvolute the cellular composition, functional relevance, as well as spatial distribution of the various components of the CCA TME. This information can help identify predictors of resistance to treatment as well as potential biomarkers of response, which can guide a biomarker-based stratification of patients.

Article Highlights.

• CCA is a highly desmoplastic malignancy characterized by myofibroblasts termed cancer-associated fibroblasts (CAF), a rich extracellular matrix, cells of the innate and adaptive immune system, and vascular endothelial cells all involved in a myriad of cellular cross-talk pathways facilitating cancer progression and permitting escape from effective immune surveillance.

• Checkpoint inhibitors designed to neutralize PD-L1 and PD-1 have minimal activity in human CCA, suggesting that CD8 T-cell exhaustion alone is not responsible for immune suppression.

• In experimental preclinical models, targeting CAF, which are derived from hepatic stellate cells, is tumor suppressive indicating a pro-tumor role for these cells in this cancer.

• Tumor-associated macrophages are abundant in CCA and can be either pro-tumor or antitumor activity and are derived from resident macrophages and bone marrow-derived macrophages. Interestingly, deleting pro-tumor macrophages in murine models of CCA fails as an antitumor strategy due to the emergence of myeloid-derived suppressor cells.

• CD40 agonism, an activator of antigen presenting cells such as dendritic cells, in combination with anti-PD-1 therapy enhances infiltration of CD4- and CD8-positive T cells and NK cells reducing tumor burden in murine models of CCA. An observation suggesting augmenting antigen presenting cell function has a therapeutic role in CCA.

• Given the complexities of the tumor immune microenvironment in CCA, combinatorial therapy will likely be needed for immunotherapy of CCA.