The Tumor Microenvironment of Bladder Cancer

Abstract

Bladder cancer has been well known as immunotherapy-responsive disease as intravesical therapy with BCG has been the standard of care for non-muscle invasive disease for several decades. In addition, immune checkpoint inhibitors have dramatically changed the treatment of metastatic bladder cancer. However, only a small fraction of patients with bladder cancer can benefit from these therapies. As immunotherapies act on the tumor microenvironment, understanding it is essential to expand the efficacy of modern treatments. The bladder cancer microenvironment consists of various components including tumor cells, immune cells, and other stromal cells, affecting each other via immune checkpoint molecules, cytokines, and chemokines. The development of an antitumor immune response depends on tumor antigen recognition by antigen presenting cells and priming and recruitment of effector T cells. Accumulated evidence shows that these processes are impacted by multiple types of immune cells in the tumor microenvironment including regulatory T cells, tumor-associated macrophages, and myeloid derived suppressor cells. In addition, recent advances in genomic profiling have shed light on the relationship between molecular subtypes and the tumor microenvironment. Finally, emerging evidence has shown that multiple factors can impact the tumor microenvironment in bladder cancer, including tumor-oncogenic signaling, patient genetics, and the commensal microbiome.

17.1. Introduction

Bladder cancer is the most common malignancy of the urinary system, and estimated to cause 165,084 annual deaths worldwide, including 17,240 in the United States [1, 2]. Urothelial carcinoma is the most predominant histological type and accounts for 90% of bladder cancers in the United States and Europe. Recent comprehensive gene expression analyses have identified molecular classifications that further characterize urothelial carcinoma more with more precision. These classifications may have important implications for prognosis and treatment response [3–7]. Bladder cancer can generally be staged into three categories in accordance with progression, namely non-muscle invasive bladder cancer (NMIBC), muscle invasive bladder cancer (MIBC), and metastatic bladder cancer in relation to treatment. Historically, NMIBC has been well known to be an immunotherapy-responsive cancer type owing to the success of intravesical therapy with Bacillus Calmette-Guerin (BCG). This treatment following transurethral resection of bladder tumors (TURBT) has been the standard therapy for several decades [8–11]. Despite incomplete understanding of the mechanism of its therapeutic effect, available evidence suggests that the cellular inflammatory response includes immune cells such as CD4⁺ and CD8⁺ lymphocytes, natural killer cells, granulocytes, and macrophage as well as cytokines such as interleukin-2 (IL-2), IL-8, IL-12, interferon-gamma (IFNg), tumor necrosis factor (TNF)-alpha, and tumor necrosis factor apoptosis inducing ligand (TRAIL) [12–15].

More recently, monoclonal antibodies targeting the PD-1/L1 immune checkpoint pathway were found to be effective for advanced bladder cancer. Since 2016, five immune checkpoint inhibitors targeting this pathway received regulatory approval. However, response rates of these drugs remain less than 25% in the first line setting, and those in the second line setting are only 15–20% without biomarker selection [16–22]. Thus, the majority of patients with bladder cancer still fail to respond to immunotherapy. Even in patients with PD-L1 expression by immunohistochemistry, which is currently the most used biomarker for predicting treatment response, the response rates are still less than 50% [21, 22]. Importantly, only a limited proportion of patients have tumors with PD-L1 expression, so the benefit of immune checkpoint inhibitors in bladder cancer has been limited so far.

Immune checkpoint blockade, including those targeting the PD-1/L1 immune checkpoint pathway, is based on the inhibition of the tumor-mediated suppression of antitumor immune response and the promotion of cytotoxic effects of T cells on tumor cells. The presence of an existing antitumor T cell response is necessary for the activity of immunotherapy [23]. Although cytotoxic T cells play a major role in antitumor immune response, their activation and ability to attack tumor cells is regulated by the balance between the stimulatory and inhibitory signals among various types of immune cells and tumor cells in the tumor microenvironment (TME) [24, 25]. Therefore, gaining a deeper understanding of the TME is crucial for advancing the efficacy of immunotherapy. In this review, we summarize the current knowledge about TME focusing on bladder cancer in relation to its component immune cells, including both preclinical clinical data. In addition, we describe recently emerging evidence linking to molecular pathways to the TME.

17.2. Main Text

17.2.1. T Cell-Inflamed and Non-T Cell-Inflamed Tumor Microenvironment

T cell-inflamed tumors are those with a pre-existing antitumor immune response, characterized by cytotoxic T cells and other important immune cells, such as antigen-presenting cells. The presence of negative immune regulators is also typically also be found in T cell-inflamed tumors (Fig. 17.1). These immune microenvironment phenotypes can be identified by gene expression profiling, immunohistochemistry, or other emerging technologies. Divergent phenotypes are characterized by the presence or absence of a gene expression profile indicating presence of an existing antitumor T cell response. These signatures include T cell markers and chemokines necessary for recruiting effector T cells to TME, such as CD8A, CCL4, CXCL9, CXCL10, ICOS, and GZMK [23, 26]. The T cell-inflamed phenotype shows spontaneous T cell priming and recruitment effector T cells in to TME even prior to any immunotherapy treatment. In contrast, non-T cell-inflamed phenotype lacks such T cell response at baseline. Clinically, these phenotypes have been correlated with prognosis and response to immunotherapies such as checkpoint inhibitors [26–28]. We previously reported that T cell-inflamed and non-T cell-inflamed phenotypes can be also identified in bladder cancers based on gene expressions of a cluster of 725 genes derived from a T cell-inflamed signature using RNA-Seq gene expression data of the Cancer Genome Atlas (TCGA). These phenotypes were highly correlated with CD8⁺ T cell infiltration in immunohistochemistry as well as mRNA expression of negative regulatory molecules such as PD-L1, IDO, TIM3, and LAG3 [29]. More in depth immune phenotyping of the TME in bladder cancer is needed, and current research efforts have been focused on this goal.

Fig. 17.1.

T cell-inflamed versus non-T cell-inflamed tumor microenvironments in bladder cancer. T cell-inflamed tumors show infiltration of T cells, antigen presenting cells (APCs), and other immune cells. T cell-inflamed tumors are also characterized by the presence of negative regulators such as PD-L1 expression and regulatory T cell (Treg) infiltration (Treg). Non-T cell inflamed tumors lack infiltrating T cells and immune cells and immune exclusion in the tumor microenvironment may be a result of multiple factors including tumor-intrinsic oncogenic signaling

17.2.2. Cytotoxic T Cells

As cytotoxic CD8⁺ T cells have an ability to directly attack cancer cells, they have been studied in the bladder cancer TME for more than a decade. Prior reports have concluded that greater infiltration of CD8⁺ T cells was related to favorable outcomes, such as overall survival, disease-free survival, and disease-specific survival both in patients with MIBC and NMIBC [30–32]. The backbone of immunotherapy for advanced bladder cancer is anti-PD-1/PD-L1 antibodies, which enhance cytotoxic activity of CD8⁺ cytotoxic T cells via blocking the suppressive signal. Thus, expression of PD-1, the receptor for PD-L1 and PD-L2, on T cells is also of interest in the TME. Expression of PD-1 is a marker of T cell exhaustion in the context of chronic antigen stimulation. Infiltration of CD8⁺ T cells and PD-1⁺ T cells has been highly correlated in the stromal area, suggesting that an exhausted immune state occurs in TME and is a target of immune checkpoint blockade [33, 34]. According to recently published meta-analyses, PD-L1 expression in bladder cancer is related to advanced tumor stage and poor overall survival. However, conflicting results were demonstrated in some of the included studies and there were several confounding factors including patient characteristics, varying treatments, and definition of positivity. Thus, these associations should be interpreted with a caution [35, 36].

17.2.3. Dendritic Cells

Dendritic cells (DCs) are an important group of antigen presenting cells and are central regulators of the adaptive immune response in T cell-mediated anti-cancer immunity. Among DCs, a subset called conventional DCs (cDCs), especially cDC1s adept at inducing cellular immunity against cancer. They endocytose cancer cells or their debris and transport tumor-associated antigens (TAAs) to the draining lymph node, where cross-presentation of TAAs and priming of CD8⁺ T cells occurs [37, 38]. Mechanistic studies have identified several regulators of DC recruitment to the TME. CCL5 and XCL1 produced by natural killer cells have been found to recruit cDC1s to TME, which is impaired by tumor-derived pros-taglandin E2 [39]. Tumor-intrinsic signaling has also been linked to lack of priming of T cells by DCs. For example, in a melanoma mouse model, β-catenin signaling was found to inhibit CCL4 production and insufficient DCs recruitment and production of CXCL9 and CXCL10, which are required for effector T cell recruitment [40, 41]. The inverse correlation of β-catenin signaling and a T cell inflamed TME was subsequently demonstrated in human cancers including bladder cancer.

In human cancers expression of Batf3, a lineage marker of DCs necessary for T cell priming, was correlated with expression of CXCR3-binding chemokines (CXCL9, CXCL10, CXCL11) and CD8A expression [40]. In several types of cancers, it has been reported that patients with abundant infiltration of DCs in tumors by IHC or gene expression is correlated with better survival outcomes compared to patients with less DC infiltration [39, 42–44]. In bladder cancer, DCs infiltration in tumors and DCs in urine have both been associated with efficacy of BCG therapy [45, 46]. We recently reported that CD8⁺ cells and Batf3⁺ DCs tended to spatially cluster rather than to distribute randomly, which suggests that those two types of immune cells cooperate in antitumor immunity (Fig. 17.2) [47].

Fig. 17.2.

Multichannel immunofluorescence for CD8 and Batf3 markers on immune infiltrating cells from human bladder tissues showing clustering in a T cell-inflamed tumor (a and b), whereas a non-T cell-inflamed tumor (c and d) shows rare T cells without Batf3+ cells (60× magnification)

17.2.4. Regulatory T Cells

Regulatory T (Treg) cells are a subset of CD4⁺ T cells characterized by CD25 expression and the master transcription factor forkhead box protein P3 (FOXP3). These cells show immune-suppressive effects via various mechanism including cytotoxic T lymphocyte antigen 4 (CTLA-4)-mediated suppression of antigen-presenting cells, consumption of IL-2, and production of immune inhibitory cytokines and molecules [48–50]. Although the molecular mechanism responsible for Treg expansion in TME is heterogeneous and remains uncertain, sphingosine 1 phosphate receptor 1 (S1P1) is overexpressed by bladder cancer cell and promotes the expression of TGF-β and IL-10 in vitro. [51] This process results in Treg expansion in bladder cancer.

Treg cells heavily infiltrate cancer tissue, and are absent or far less abundant in the peripheral circulation [50, 52]. This point highlights the importance of analyzing the TME. Similar to other cancers, human bladder cancer has abundant infiltration of Treg cells [53–55]. Although abundant Treg cell infiltration has been reported to be associated with a poor prognosis, there have been several reports indicating the opposite association. For example, Treg infiltration is a favorable prognostic marker in colorectal cancer and Hodgkin’s lymphoma [56–58]. In bladder cancer, while abundant infiltration of Treg cells was reported to be associated with a poor recurrence-free survival (RFS) or PFS after endoscopic resection [59–61], the opposite relationship between Treg cell infiltration and survival outcome was also reported [62]. This paradoxical relationship between abundant Treg cell infiltration and favorable survival was explained by the suppression of matrix metalloproteinase 2 (MMP2) in TME, a key pro-invasive factor induced by tumor-promoting inflammation by Treg cells specifically at the invasive margin of the tumors [55]. In addition, several studies reported that high Treg cells/total T cells and high Treg cells/cytotoxic T cells were associated with poor survival outcomes. Also, the ratio of high cytotoxic T cells/Treg cells was associated with response to neoadjuvant chemotherapy [63–65]. Therefore, the number of Treg cells is important, but in addition, the location and balance between Treg cells and other T cells are relevant. A more detailed classification of Treg cells including those with potential effector function might explain the discordant prognostic relevance in bladder cancer [66].

17.2.5. Tumor-Associated Macrophages

Tumor-associated macrophages (TAMs) are found in tumors that can suppress antitumor immunity and promote tumor growth and progression by producing cytokines and chemokines, promoting genetic instability, nurturing niche for cancer stem cells, and supporting metastasis [67]. TAMs are polarized to M1 or M2 according to signals from the TME and show different activities in terms of receptor expression, effector function, and cytokine and chemokine production. M1 macrophages are characterized by the production of pro-inflammatory and immune-stimulatory molecules, such as IL-12 and tumor necrosis factor alpha (TNF-α). M2 macrophages are characterized by the production of immune-suppressive molecules, such as IL-10, TGF-β, and IL-1ra [68]. TAMs are thought to be recruited from the blood compartment through chemotactic signals released from tumors, such as CCL2, CCL5, and CSF-1. In bladder cancer, CCL2 was also shown to recruit TAMs, which promoted lymphatic metastasis via vascular endothelial growth factor production [69].

In the TME, M2 immune-suppressive macrophages are predominant and M1 pro-inflammatory macrophages are relatively rare [67]. TAMs can be identified by expression of the surface marker, CD68. M2 macrophages can be identified using surface markers such as CD163 and CD204. Abundant infiltration of TAMs has been associated with high tumor grades and poor survival outcomes in various types of cancer [70–75]. Interestingly, there are also some reports that describe discordant association between abundant infiltration of TAMs and survival outcomes according to treatments, suggesting that TAM infiltration is also related to efficacy of specific agents [76, 77]. In bladder cancer, abundant infiltration of TAMs, especially M2 phenotype, or a high M2/pan-macrophages ratio was linked to aggressiveness of the disease such as size, grade, nodal status, microvessels, and stage, as well as poor survival outcomes [78–80]. These findings are consistent with the above-mentioned preclinical observations. Additionally, a high TAM/TIL ratio has been associated with a poor disease-free survival after cystectomy in patients with MIBC [81]. In patients treated with BCG after TURBT, abundant infiltration of TAMs or M2 macrophages was related to response to BCG and survival [45, 82, 83].

17.2.6. Myeloid-Derived Suppressor Cells

Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of myeloid cells that strongly inhibit antitumor activities of other immune cells and also directly interact with tumor cells promoting tumor progression. MDSCs are not present in normal tissues, but appear in pathologic conditions such as cancer, chronic inflammation, infection, autoimmune diseases, trauma, and graft versus host disease. MDSCs are generally categorized into two phenotypes. Monocytic MDSCs (M-MDSCs) are phenotypically and morphologically similar to monocytes. Granulocytic or polymorphonuclear MDSCs (G-MDSCs or PMN-MDSCs) are phenotypically and morphologically similar to neutrophils. In humans, G-MDSCs and M-MDSCs are characterized by the markers CD11b⁺CD14⁻CD15⁺ or CD11b⁺CD14⁻CD66b⁺ and CD11b⁺CD14⁺HLA-DR^−/loCD15⁻, respectively [84, 85].

Evidence for the significance of MDSC has been accumulating in several cancers including bladder cancer. In vitro studies have revealed that G-MDSCs inhibit CD4⁺ T cell proliferation via CD15 and induce Treg cells [86]. Bladder cancer cells recruit MDSCs via CXCL2/MIF-CXCR2 signaling and also activate group 2 innate lymphoid cells (ILC2), which recruit and activate MDSCs further via IL-13 production [87, 88]. In relation to chemotherapy, cisplatin directly inhibited proliferation and induced the apoptosis using a human bladder cancer cell line, T24, in vitro [89]. Using a C3H/He mouse bladder cancer model, cisplatin treatment was also shown to inhibit the progression of the tumor and decrease the proportion of G-MDSCs phenotype in the tumor [89].

In human bladder cancers, the numbers of circulating G-MDSCs or total MDSCs are higher in bladder cancer patients than non-cancer healthy controls. In human bladder cancer tissue, G-MDSCs are the predominant phenotype and more G-MDSCs exist in cancer tissue than the adjacent normal tissue [86, 90, 91]. In addition, G-MDSCs are reported to be increased in cancer tissues compared to the adjacent non-cancer tissues, and a negative correlation was observed between the number of G-MDSCs and CD8⁺ T cells in tumors, which can be explained the expression of immune-suppressive molecules by MDSCs including Arg1, iNOS, ROS, PD-L1, and pSTAT3 [88, 89]. Increased MDSCs in peripheral blood or in bladder cancer tissue are associated with size, pathological grade, and stage of the tumor [88, 90, 91]. Patients with a low T cell/MDSC ratio in urine samples demonstrated a lower recurrence-free survival after BCG in patients with NMIBC. Patients with a high number of circulating MDSCs demonstrated a low pathological CR rate after neoadjuvant chemotherapy compared to patients with a low number of circulating MDSCs [87, 92].

17.2.7. Pericytes

Angiogenesis has been rigorously studied as a part of TME, since the aberrant vasculature is one of the main hallmarks of tumors, and suppresses oxygen distribution and immune cells delivery [93]. In angiogenesis, activation of tumor-associated endothelial cells was shown to promote tumor growth and progression in multiple human cancer cell lines, and a high inflammatory gene signature of the tumor-associated endothelial cells was related to a poor survival in gene expression analysis [94]. Meanwhile, pericytes, another component of small vessels, play important roles in angiogenesis by stabilizing capillaries and limiting vascular permeability [95]. Among two main subsets of pericytes in mice (type-1 [nestin⁻NG2⁺] and type-2 [nestin⁺NG2⁺]), only type-2 pericytes were recruited during its angiogenesis in tumor plantation mouse models. In preclinical models, anti-angiogenetic agents showed increased pericyte coverage and enhances tumor perfusion and improved homogeneous oxygen distribution. Furthermore, it increased immune cells delivery and converted the TME from immunosuppressive state to immunostimulatory state [96–98]. This phenomenon was also positively related to clinical outcomes in several clinical trials [98].

Following the success of combination of immune checkpoint inhibitors and axitinib as the first line therapy for advanced renal cell carcinoma [99, 100], the combination of an immune checkpoint inhibitor and an anti-angiogenetic agent is being tested for multiple cancer types including bladder cancer [97].

17.2.8. The Relationship Between the TME, Tumor-Cell Intrinsic Molecular Pathways, and the Responses to Immune Checkpoint Inhibitors

Although PD-L1 expression in TME by IHC was initially thought to be a promising predictive biomarker for response to anti-PD-1/L1 therapy, further research indicated that its accuracy remained limited [101]. Further research has focused on identifying other characteristics of the tumor microenvironment that might predict response. A biomarker analysis was performed on IMvigor210, a phase II trial of the anti-PD-L1 antibody atezolizumab in patients with metastatic urothelial cancer. The authors found that responses were associated with a CD8⁺ T-effector cell signature based on RNA sequencing and tumor mutation burden. Transforming growth factor β (TGF-β) signaling in fibroblasts predicted lack of response in patients with CD8⁺ T cell excluded tumors that had CD8⁺ T cell in peritumoral stroma [102]. More evidence about the relationship between the bladder cancer TME and responses to immune checkpoint inhibitor is expected, given the rapid accumulation data from bladder cancer patients receiving PD-1/PD-L1 treatment.

Based on genomic analyses of bladder cancer, several molecular subtypes have been proposed, and a recently proposed consensus molecular classification included six subtypes for muscle invasive bladder cancer [4, 7, 103]. Among them, “basal/squamous” tumors, which is the commonest subtype accounting for 35% of bladder cancer, were enriched in cytotoxic lymphocytes markers, natural killer cells markers, and monocytic lineage markers. This subtype was also associated with strong expression of immune checkpoint markers and antigen-presenting machinery genes, suggesting that such tumors might be more responsive to immunotherapies. The “stroma-rich” subtype overexpressed T cell markers and B cell markers [104]. Tumors with “luminal papillary” or “luminal unstable” classification lacked immune infiltrating cell markers as did the “neuroendocrine-like” subtype. “Luminal nonspecified” tumors showed limited expression for T and B cell markers and were associated with a fibroblastic stromal molecular pattern.

Evidence has accumulated that alterations in certain tumor-cell intrinsic molecular pathways are associated with the non-T cell-inflamed TME and resulting immunotherapy resistance. Gain-of-function alterations in Wnt-β catenin signaling and MYC and loss-of-function alterations or deletions LKB1, PTEN, and p53 were associated with reduced T cell priming or infiltration [105]. Activation of tumor-intrinsic Wnt-β catenin signaling was shown to be enriched in non-T cell-inflamed tumors across cancer types including bladder cancer using TCGA data [106]. Using the data of TCGA Bladder Urothelial Carcinoma, PPAR-γ, and FGFR3 pathways were activated in non-T cell-inflamed tumors as well as Wnt-β catenin signaling [29]. Indeed, activated PPARγ/RXRα signaling suppressed the production of pro-inflammatory cytokines and chemokines, resulting in impaired CD8⁺ T cell infiltration leading to resistance to immunotherapies in preclinical models [107]. FGFR3 mutation was associated with low T cell infiltration compared to wild type bladder cancers. The responsiveness to immunotherapy was not linked with FGFR alterations in the biomarker analyses from IMVIGOR 210 and Checkmate 275, which tested atezolizumab and nivolumab, respectively, in metastatic bladder cancer patients. It was suggested that an inverse association between FGFR3 mutation and a stromal TGF-β signaling was suggested to be the reason for similar response rates between FGFR3 mutated tumors and wild-type tumors, despite the difference of T cell infiltration [108].

17.3. Future Directions

The tumor microenvironment in bladder cancer is a complex of factors promoting and inhibiting the antitumor immune response. Therefore, a multidimensional approach to its evaluation will be necessary to gain a deeper understanding of the biological underpinnings at play. In addition to FACS or CyTOF, recently developed multiplex immunohistochemistry technology enabled us to stain multiple markers on a single slide and to evaluate multiple phenotypes of immune cells [109]. Besides quantitative analysis of the numbers of multiple phenotypes of infiltrating immune cells, spatial analysis can be conducted using this technology [47, 73]. Cytokines and chemokines also play important roles in terms of activation or inactivation and recruitment of immune cells, the relationships between these molecules and immune cells should be investigated for comprehensive understanding of TME. Combination of in situ hybridization for RNAs and immunohistochemistry for proteins could shed light on their relationships [110]. Emerging data indicate that heritable genetics and the commensal microbiome are two additional factors that can influence the tumor microenvironment in bladder cancer [111, 112]. There have been some reports suggesting interactions between nervous system and the tumor microenvironment and cancer progression [113–117]. The roles of nerves impacting the TME in bladder cancer may be another important unexplored area of investigation. The incorporation of multiple interacting factors will necessitate the use of advanced statistical and computational approaches to characterize each unique tumor. These advances may enable us to better prevent, diagnose, prognosticate, and optimize treatments for bladder cancer patients in the future.