Urothelial bladder cancer is among the 10 most common tumor entity in industrialized countries. While 75% are non-invasive tumors with a good patients’ prognosis, the remaining 25% are invasive and still associated with poor outcomes (1). The standard of care for muscle-invasive bladder cancer (MIBC) is currently radical cystectomy with perioperative platinum-based chemotherapy and adjuvant nivolumab treatment in selected cases (2). Molecular analyses identified three major molecular lineages within muscle-invasive urothelial bladder cancer (MIBC): basal, luminal, and neuronal-like/neuroendocrine. These subtypes have considerable prognostic and predictive implications (3–8) and correlate with urothelial subtype histology (9). Moreover, the basal and luminal subtypes of MIBC exhibit distinct associations with inflammation levels, infiltration through T lymphocytes, and secondary lymphoid structures, which are valuable immunological indicators predicting survival outcomes and therapy responses of patients (10–13).

Human leukocyte antigen G (HLA-G) is known for its immune regulatory properties in the placenta to prevent fetal rejection in pregnant women (14–16) and (aberrantly) expressed in many tumor entities and at immune-privileged sites to induce immune tolerance (17–21). HLA-G binds to the inhibitory ligands ILT2, ILT4, and KIR2DL4 and suppresses effector mechanisms, in particular in CD8+ T cell and NK cell mediated cytotoxicity, thereby contributing to the immune escape of tumors (18, 22). The tumor-specific expression of HLA-G and its biological characteristics, particularly its involvement in tumor immune evasion, render it as an interesting biomarker and a potential therapeutic target for novel immunotherapies. However, to date, the expression of HLA-G in MIBC remains unexplored. Thus, the global expression patterns and association of HLA-G with determinants of the spatial immune microenvironment, molecular subtypes and pathological features was analyzed in the well characterized consecutive population-based CCC-EMN-ER MIBC patient cohort (n=241).

MIBC continues to exhibit a poor prognosis with a 5-year survival of 40%-60% (38). The immune system plays an important role in the anti-tumour response and immune-targeted therapies, such as immune checkpoint blockade have already been tested and successfully implemented for MIBC and metastatic UC treatment (39). However, the development of future immunotherapies requires a deeper understanding of the tumor immune microenvironment (TIME) and immune evasion mechanisms of MIBC. Prior research has demonstrated an increase in HLA-G expression across diverse tumor entities, which is associated with a poor patients’ prognosis (21, 40). Systematic reports on the prevalence and biological relevance of HLA-G in MIBC however are missing until today. Thus, we conducted the present study to better understand if and which role HLA-G play in the clinical and biological framework of MIBC with a specific focus on elucidating its role in MIBC immune evasion, and thus studying HLA-G as a potential therapeutic target for restoring the anti-tumoral immunity.

Based on our findings, HLA-G appears to be a promising biomarker associated with the biological characteristics, the composition of the tumor immune microenvironment (TIME), and clinical outcomes in MIBC. This is supported by the observation that patients with high HLA-G expression in tumors experience reduced overall and disease-specific survival. Additionally, the ability to quantify soluble HLA-G in the blood, its predictive value when co-expressed with other immune checkpoints, and its consistent presence across various tumor regions further highlight its utility as a potential biomarker and therapeutic target. However, our study is limited by assessing total HLA-G expression while a functional heterogeneity of HLA-G isoforms exists, and the retrospective and monocentric nature of the study.

Our data demonstrated a strong correlation between HLA-G expression on tumor cells and both overall survival and disease specific survival rates in MIBC patients, highlighting the clinical and biological significance of HLA-G in cancer progression and prognosis. In contrast, HLA-G expression on TILs was not linked to a decrease in survival rates, suggesting that it is HLA-G expression on tumour cells, rather than on TILs, that orchestrates biological behavior of MIBC and thus HLA-G’s impact on clinical outcomes. Notably, the impact of HLA-G tumor cell expression on overall and disease specific survival remained independent of clinicopathological parameters in the multivariable regression analysis. This is validated by recent studies which revealed a similar clinical impact of HLA-G tumor cell expression and an association with shorter overall survival in other tumor entities (40). For example, Yie et al. demonstrated an upregulation of HLA-G in esophageal squamous carcinomas that represents an independent adverse prognostic factor (41), and similar results were observed for non-small cell lung cancer (NSCLC) (42).

Notably, in our MIBC cohort HLA-G expression on tumor cells was highly expressed in sarcomatoid/rhabdoid tumors that show an increased EMT (43). EMT is a crucial biological process where epithelial cells differentiate into mesenchymal cells leading to increased cell mobility and invasion capacity thus facilitating metastatic seeding to other organs. Matching with these observations, EMT has been associated with poor patient outcomes, and in the case of sarcomatoid NSCLC with a remarkable upregulation of immune inhibitory molecules and immune evasion programs (44).

Hence, analogous mechanisms could be implicated in the upregulation of HLA-G in sarcomatoid MIBC. Indeed we can show that sarcomatoid MIBCs, which exhibit the highest levels of HLA-G tumor cell expression, belong predominantly to a recently described spatially organized immune phenotype (“Evasion phenotype”) whose TIME is dominated by immunosuppressive T and myeloid cells as well as a marked expression of immune inhibitory immune checkpoint molecules like PD-L1 and PD-1 on tumor cells or immune cells (11). The spatially organized immune phenotypes described by Pfannstiel et al. are also closely related to consensus molecular subtypes of MIBC, where sarcomatoid MIBCs usually belong to the basal squamous molecular subtype (Ba/Sq) (11). While MIBC with absent or low T cell infiltration mostly belong to luminal molecular subtypes, which associate with conventional urothelial histology or urothelial subtype histology, strongly T-cell infiltrated tumors predominantly belong to the basal squamous molecular subtype (Ba/Sq) and present with sarcomatoid or squamous histology (11). Matching with these observations, we also observed higher HLA-G expression levels in MIBC with Ba/Sq molecular subtypes, which is a molecular subtype that has been associated with decreased patient survival rates (6, 45). While many MIBC with Ba/Sq molecular subtypes exhibit immune infiltrates consisting of CD8+ cytotoxic T lymphocytes (CTL), NK cells and macrophages (6), especially those with high HLA-G tumor cell expression belonged to the immune evasive “Evasion phenotype” that is characterized by a T-cell exhausted and myeloid dominated TIME.

Thus, upregulation of HLA-G tumor cell expression might be a counter-regulatory mechanism that 1) limits the TIME infiltration of CD8 and NK cells, 2) exhausts present effector T cells and NK cells via recruitment of immunosuppressive myeloid cells and regulatory T-cells, and 3) increases the immunosuppressive capacity of myeloid cells (46, 47). Similar observations have been made by Dong et al., where higher HLA-G expression levels was mainly found in clinically aggressive non-luminal (basal) breast cancers, and associated with depletion of activated TILs (48). Mechanistically, this immunosuppressive function of HLA-G has been partially shown to operate by inhibiting the cellular anti-tumor immune response and tumor cell killing through its interaction with receptors ILT2, ILT4, and KIR2DL4 expressed on cytotoxic T cells and NK cells (47, 49–51). In line with these observations, Mendel et al., for example, demonstrated that inhibition of ILT2 (preventing HLA-G from acting via ILT2) restores antitumor immune response leading to decreased tumor growth, less metastatic spread, and a prolonged survival in a humanized mouse model (52). In addition, HLA-G has also indirect potent immune regulatory capabilities, e.g., via induction of CTLA-4 and PD-1 expression on ILT-2 positive T cells regardless of prior T cell activation (53–55). This is in line with our observations, that immune cells in the TIME of HLA-G high MIBC – especially of evasion phenotype MIBCs- exhibited an upregulation of crucial immune checkpoint proteins like TIGIT, PD-L1, LAG3, and CTLA-4 on tumor cells (PD-L1) and immune cells (all proteins). In parallel to this direct inhibition of effector T cells and NK cells, HLA-G has been linked with an active recruitment of Treg cells and induction of CD4+ T cell differentiation into Treg cells via CD46 different regulatory T cells for ex (56–58). These findings are in line with our observation of a positive correlation of regulatory T-cell infiltration and HLA-G tumor cell expression especially in MIBC with evasion phenotypes (31). Additionally, an inverse correlation between the GZMB/CD8 ratio and HLA-G was found, which is indicative of a diminished cytotoxic potential of CD8+ cytotoxic T cells (59). Dumont et al. associated this immunosuppressive effect on ILT2+ cytotoxic T cells to HLA-G (60), and other groups demonstrated, that this effect can be reverted effectively by inhibition of HLA-G which leads to restoration of IFNG secretion and cytotoxic degranulation (60, 61). Another frequently observed immune evasion mechanism of MIBC is the downregulation of MHC class I (MHCI) presentation on tumor cell membranes in order to evade recognition by cytotoxic T cells; however, MHC-deficient tumor cells can still be recognized and eliminated by NK cells (62–64). Interestingly, MIBC with evasion phenotype exhibit a low expression of MHCI accompanied by a low infiltration with CD56+ NK cells, but high levels of HLA-G expression. Thus, HLA-G upregulation might represent a mechanism to evade the MHC independent tumor cell killing through NK cells (65). This immune evasion mechanism has also been functionally proven in ovarian cancer, where HLA-G directly inhibited NK cell mediated cancer cell lysis, which could be disinhibited by blocking of HLA-G (66). Taken together, previous reports and our findings thus support the hypothesis that HLA-G is a crucial immune checkpoint protein in (subsets of) MIBC, that represents an attractive therapeutic target to restore sustained cellular anti-tumor immune response.

Beside this direct and indirect HLA-G mediated inhibition of antitumoral effector cells, recruitment and activation of myeloid cells has been functionally attributed to HLA-G (67, 68). In line with these observations, MIBC belonging to the evasion phenotype also demonstrated a myeloid-rich TIME (increased CD68/CD8 ratio) where macrophages exhibit a predominant M2 phenotype (11, 13). In line, shifts in the balance between myeloid cells and cytotoxic T cells, i.e., towards a myeloid dominated TIME, have been shown to substantially decrease adaptive antitumoral immune responses (67). In addition, HLA-G has been shown to directly increase the suppressive activity of myeloid-derived suppressor cells (MDSCs) via ILT4 in the placenta, and that this effect can be reversed by ILT4 inhibition, indicating an indirect suppression of adaptive immune responses via shaping the activity of myeloid cells (68).

Beside the HLA-G high evasion phenotype, Pfannstiel et al., Erlmeier et al, and Zhou et al. identified another subgroup of highly T cell infiltrated MIBC with a predominant infiltration by activated T cells, T cells with preserved capacity for tumor cell killing and reduced presence or absence of immune inhibitory immune checkpoint proteins (highly inflamed cytotoxic environment) as well as a reduced myeloid cell component resulting in significantly improved outcome rates of patients (11, 13, 69). In line, MIBC with this cytotoxic immune phenotype showed lower HLA-G protein expression levels than those belonging to the immunosuppressed/exhausted “Evasion phenotype” in our cohort, thus underlining the mechanistic role of HLA-G shaping the functionality of tumor associated immune infiltrates to an exhausted state. The upregulation of PD-1 on T cells within the inflamed high phenotype may be elucidated by the phenomenon wherein lymphocyte activation is concomitant with the upregulation of co-inhibitory receptors, thereby averting the potential for untoward inflammation in normal tissue that might result in extensive tissue damage (70).

In summary, elevated HLA-G expression is strongly linked to direct and indirect inhibition of antitumoral effector immune cells and to remodeling of the TIME to an immunosuppressive state thus resulting in significantly decreased patient survival rates. Based on previous reports and other data, we thus believe that HLA-G functions as a crucial immunosuppressive and inhibitory immune checkpoint protein in MIBC, contributing to reduced T cell cytotoxicity, suppression of NK cells and maintenance of an immunosuppressive environment. Consequently, HLA-G emerges as a promising candidate both as a prognostic marker and – more important – as an interesting immunotherapeutic target, particularly in the context of tumors with an immune exhausted and evasive TIME characterized by a poor prognosis that require intensified treatment regimens.