Immunotherapy for glioblastoma: current progress, challenges, and future outlook

Immune checkpoints are a class of regulatory molecules expressed on the surface of certain immune cells, particularly T cells, that are crucial for maintaining immune system homeostasis and preventing autoimmunity [14–16]. Under physiological conditions, these molecules are essential for self-tolerance and protecting tissues from immune-related damage [17–20]. However, in the context of cancer, they can be exploited by tumor cells to evade immune surveillance and suppress antitumor immunity [15, 16, 21, 22]. In the past few decades, several immune checkpoints (Fig. 1A) have been identified, including programmed cell death protein 1 (PD-1) and its ligand PD-L1, cytotoxic T-lymphocyte associated protein 4 (CTLA-4), lymphocyte activation gene 3 (LAG-3), T-cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT), T-cell immunoglobulin and mucin domain 3 (TIM-3), V-domain Ig suppressor of T-cell activation (VISTA), and indoleamine 2,3-dioxygenase 1 (IDO1) [14, 23–26]. Blocking these immune checkpoints with monoclonal antibodies, known as ICIs, has shown significant success in treating multiple advanced cancers, such as melanoma, non-small cell lung cancer, colorectal cancer, gastric cancer, and hepatocellular carcinoma [10, 27]. However, a large number of clinical trials have shown minimal clinical benefit for GBM patients [28].

Three phase III trials have been designed to test the antitumor efficacy of the anti-PD-1 antibody nivolumab in GBM patients [29–31]. In the CheckMate 498 study, 560 newly diagnosed GBM patients with an unmethylated O⁶-methylguanine-DNA methyltransferase (MGMT) promoter received RT in combination with nivolumab or TMZ. However, the study did not meet its primary endpoint of OS, with mOS values of 13.4 and 14.9 months for the nivolumab and TMZ groups, respectively [29]. Similarly, the CheckMate 548 study involving 716 newly diagnosed GBM patients with a methylated MGMT promoter also failed to improve the mOS (28.9 vs. 32.1 months) or median progression-free survival (mPFS, 10.6 vs. 10.3 months) upon the addition of nivolumab concurrent with the SOC [30]. In another setting, the CheckMate 143 study compared the antitumor efficacy of nivolumab combined with the antiangiogenic drug bevacizumab in recurrent GBM patients. This study enrolled 369 GBM patients at their first recurrence following RT and TMZ therapy and revealed that nivolumab treatment did not improve patients’ mOS (9.8 vs. 10.0 months), mPFS (1.5 vs. 3.5 months), or objective response rate (ORR, 7.8 vs. 23.1%) [31]. Similarly, another anti-PD-1 antibody, pembrolizumab, has shown limited clinical benefit both as monotherapy and in combination with bevacizumab for recurrent GBM patients in phase I and II trials [32–34]. Notably, neoadjuvant anti-PD-1 immunotherapy has demonstrated encouraging outcomes in selected recurrent GBM patients in window-of-opportunity trials [35, 36]. For example, a notable study with 35 recurrent GBM patients demonstrated that neoadjuvant pembrolizumab therapy combined with adjuvant therapy prolonged patient survival compared with patients who received only postsurgical pembrolizumab treatment (mOS, 13.7 vs. 7.5 months; mPFS, 3.3 vs. 2.4 months) [36]. Mechanistic studies have demonstrated that this neoadjuvant pembrolizumab therapy increases T-cell infiltration and interferon-γ (IFN-γ)-related gene expression [36] and enhances chemokine expression and T-cell receptor (TCR) clonal diversity [37]. In addition to these investigations using anti-PD-1 antibodies, the efficacy and safety of PD-L1 antibodies (e.g., durvalumab, atezolizumab, and avelumab) have also been evaluated in GBM patients, with results showing potential antitumor efficacy and favorable tolerability [38–41].

For the anti-CTLA-4 antibody ipilimumab, a phase II clinical trial has been performed, and the results showed that ipilimumab combined with TMZ treatment did not lead to clinical benefit over TMZ treatment alone (mOS, 22.7 vs. 26.4 months; mPFS, 10.9 vs. 12.5 months) [42]. Another ongoing randomized open-label phase II/III trial (NCT04396860↗) aims to test the antitumor effect of combining ipilimumab, nivolumab, and RT in newly diagnosed and unmethylated MGMT GBM patients. For recurrent GBM patients, the phase I CheckMate 143 trial revealed that intravenous administration of ipilimumab and nivolumab has no clinical benefit but increases immune toxicity compared with nivolumab treatment alone (mOS, 9.2 vs. 10.4 months; mPFS, 1.5 vs. 1.9 months) [43]. However, one phase I clinical trial in recurrent GBM patients involving the intracerebral administration of ipilimumab and nivolumab showed a promising survival benefit (mOS, 8.9 months; mPFS, 2.7 months), encouraging further investigations [44].

In addition to “classic” immune checkpoints, novel targets such as LAG-3, TIM-3, TIGIT, and IDO1 are actively under investigation for GBM treatment. Given that LAG-3 is an early marker of exhausted T cells, treatment with anti-LAG-3 drugs might offer therapeutic benefits in cancer patients [45]. In the context of GBM, two phase I clinical trials (NCT02658981↗ and NCT03493932↗) are underway to explore the antitumor effect of the LAG-3 inhibitor relatlimab (BMS-986016) alone or in combination with nivolumab in recurrent or newly diagnosed GBM patients. TIM-3 is a coinhibitory molecule expressed on immune cells, and TIM-3-targeted therapies are currently under investigation in various cancer types, including GBM [46]. For example, treatment with the TIM-3 inhibitor sabatolimab combined with the PD-1 inhibitor spartalizumab and stereotactic radiosurgery is underway for recurrent GBM patients (NCT03961971↗). TIGIT is a T-cell coinhibitory receptor that is expressed mainly by T cells and natural killer (NK) cells [47]. Preclinical studies have shown that inhibition of TIGIT improves the antitumor immune response [47]. A phase 0/I study (NCT04656535↗) is recruiting recurrent GBM patients to evaluate the safety and tolerability of the anti-TIGIT antibody domvanalimab (AB154) combined with the anti-PD-1 antibody zimberelimab (AB122). In addition, the antitumor efficiency of IDO1 inhibitors (e.g., BMS-986205 and indoximod) combined with RT and TMZ is also currently under investigation in different clinical trials (NCT04047706↗ and NCT02052648↗) for newly diagnosed GBM patients.

Adoptive T-cell therapy is a personalized cancer immunotherapy technique in which a patient’s T cells are isolated, expanded ex vivo and then reinfused back into the patient to target tumors [48]. Before reinfusion, patients undergo a lymphodepleting regimen with lymphocyte-directed chemotherapy to increase treatment effectiveness by creating a favorable environment for infused cells [48]. An increasing number of clinical trials have demonstrated a transient antitumor response with manageable side effects, providing hope for the development of effective adoptive T-cell therapies for GBM. Here, we discuss the current status of adoptive T-cell therapies, including tumor-infiltrating lymphocyte (TIL) therapy, cytomegalovirus (CMV)-specific T-cell therapy, and chimeric antigen receptor (CAR)-T-cell therapy, in GBM (Fig. 1B).

Cancer vaccines aim to activate the patient’s adaptive immune system against specific endogenous and exogenous tumor antigens and are being explored in GBM [69]. Endogenous antigens, including tumor-specific antigens (TSAs, also known as neoantigens) and tumor-associated antigens (TAAs), are intracellular proteins from tumor cells. Exogenous antigens originate from the infection of exogenous pathogens, such as CMV [70, 71]. Cancer vaccines can be created using “predefined” antigens (known antigens, which include shared antigens expressed in the majority of patient tumors or personalized antigens specific to individual patients) or “anonymous” antigens (unknown antigens, which can be colocalized with antigen-presenting cells (APCs) ex vivo or engineered for reinjection into patients to activate endogenous APCs in situ) [72, 73]. After administration, antigens are presented by APCs to naïve or memory T cells, and these primed T cells then migrate into tumor sites, initiating tumor regression and establishing long-lasting memory responses against tumor recurrence [72, 73]. Various vaccination strategies are under investigation for GBM treatment [69]. In this section, we discuss four different vaccine platforms, including peptide vaccines, dendritic cell (DC) vaccines, DNA/RNA vaccines, and viral vector vaccines, and their use for GBM treatment (Fig. 1C).

OV therapy represents another class of immunotherapies being intensively investigated for GBM [116] (Fig. 1D). OVs are replication-competent viruses that selectively infect and replicate in cancer cells, causing lysis of GBM cells followed by the release of viral pathogen-associated molecular patterns, damage-associated molecular patterns, TAAs, and proinflammatory cytokines into the TME, which, in turn, induce antitumor immune responses and transform the TME from “cold” to “hot” [117]. Specifically, OVs can facilitate the activation and migration of APCs to activate cytotoxic CD8⁺ T cells, enhancing the overall immune response against GBM.

Herpes simplex virus (HSV) is the first engineered viral strain for GBM treatment in a murine model [118]. HSV-1, a neurotropic virus from the Herpesviridae family, is a double-stranded DNA virus. Several genetically engineered versions (e.g., HSV-1716, G207, and G47∆) of HSV-1 have been evaluated in GBM. Notably, both copies of the RL1 gene encoding the virus protein ICP34.5 were deleted in all HSV recombinants, resulting in increased tumor selectivity [119, 120]. HSV-1716 is the first-generation HSV, and its safety has been proven in several phase I clinical trials [121–123]. The second generation of the HSV construct, G207, featuring the deletion of both copies of the RL1 gene and an inactivating insertion of the UL39 gene encoding infected cell protein 6 (ICP6), can generate preferential replication in dividing cells [124, 125]. The results from several clinical trials demonstrated that G207 is safe for GBM patients [126–128]. G47∆ (Teserpaturev) represents the third generation of oncolytic HSV, which is constructed on the basis of G207 with an additional deletion of the α47 gene, thus increasing viral replication and triggering antitumor immune responses via the upregulation of MHC I molecules [129]. A phase I/II trial at the University of Tokyo confirmed the safety of G47∆ in recurrent GBM patients [129, 130], followed by a phase II trial revealing its antitumor efficacy following intratumoral injections in residual or recurrent GBM [131]. As a result, G47∆ has been granted conditional time-limited approval by Japan’s Pharmaceuticals and Medical Devices Agency for brain tumor treatment. Moreover, a recent study developed a new version of G47∆ expressing IL-2 (G47Δ-mIL2), which can significantly prolong the survival of different GBM mouse models associated with increased intratumoral CD8⁺ T cells, suggesting that G47Δ-mIL2 may represent a new direction of HSV treatment for GBM patients [132], but further clinical trials are needed.

Polioviruses are positive- and single-stranded RNA viruses, and PVSRIPO is a nonpathogenic poliovirus/rhinovirus chimeric virus that targets tumor cells via the poliovirus receptor CD155, which is overexpressed in GBM tumors [133]. A phase I clinical trial (NCT01491893↗) demonstrated that intratumoral infusion of PVSRIPO in recurrent GBM patients was tolerated and produced increased mOS (12.5 vs. 11.3 months) and 24-month (21% vs. 14%) and 36-month (21% vs. 4%) survival rates compared with those of historical controls [134]. Currently, several clinical trials, including a phase II trial (NCT02986178↗) evaluating PVSRIPO as a monotherapy and phase I/II (NCT03973879↗) and phase II (NCT04479241↗) trials investigating the combination of PVSRIPO with the anti-PD-L1 antibody atezolizumab or the anti-PD-1 antibody pembrolizumab, are underway in GBM patients.

Adenoviruses are nonenveloped, double-stranded DNA viruses with icosahedral structures that have been extensively studied for cancer treatment [135]. DNX-2401 (known as tasadenoturev or Delta-24-RGD) is an engineered oncolytic adenovirus designed to selectively target and replicate in cancer cells with aberrant retinoblastoma (Rb) pathways. Specifically, DNX-2401 contains a genetic modification in the E1A gene, which is a 24 bp deletion responsible for Rb binding, enabling this adenovirus to replicate selectively in cancer cells with defects in the Rb pathway and decreasing virus replication in normal cells. In addition, it involves the insertion of an Arg-Gly-Asp (RGD) peptide motif, which can improve its attachment to GBM cells by binding to αV integrins. Phase I clinical trials in recurrent GBM patients demonstrated that DNX-2401 was safe and had significant antitumor effects on long-term survival in some patients [136, 137]. Furthermore, DNX-2401 in combination with TMZ was also well tolerated [138], whereas the addition of IFN-γ to DNX-2401 resulted in poor tolerability without an additional clinical benefit [139]. CRAd-S-pk7 is another glioma-tropic oncolytic adenovirus that contains the tumor-specific survivin promoter (S) and a fiber protein polylysine modification (pk7), with potential antineoplastic activity [140, 141]. Interestingly, this oncolytic adenovirus can be delivered by neural stem cells (NSCs), which can enhance the glioblastoma stem cell (GSC)-targeting ability of CRAd-S-pk7 by combining the unique tumor tropism of NSCs [142]. The safety and therapeutic potential (mOS, 18.4 months; mPFS, 9.1 months) of NSC-CRAd-S-pk7 in newly diagnosed GBM has been proven by a phase I study (NCT03072134↗) [143], and further clinical trials are needed to confirm its antitumor efficacy in GBM.

Moreover, adenoviruses can be modified to act as delivery vectors for exogenous genes, such as IL-12, the Fas-chimera transgene, and the HSV thymidine kinase (HSV-TK) gene. Ad-RTS-hIL-12 is an engineered adenoviral vector that delivers IL-12 controlled by the RTS promoter upon the oral administration of veledimex (VDX) [144, 145]. The tolerability and antitumor effect of Ad-RTS-hIL-12 in patients with resected recurrent GBM were evaluated in a phase I clinical trial (NCT02026271↗), with the results showing that peritumoral injection of Ad-RTS-hIL-12 was tolerated and resulted in an mOS of 12.7 months [146]. VB-111 is a nonreplicating adenovirus-engineered vector that can carry a Fas-chimera transgene, leading to Fas-mediated tumor apoptosis and vascular disruption [147]. In a phase I/II study (NCT01260506↗), combined treatment with VB-111 and bevacizumab significantly improved survival in recurrent GBM patients compared with treatment with VB-111 alone (mOS, 13.8 vs. 7.4 months; mPFS, 3.0 vs. 3.0 months) [148]. However, the subsequent phase III trial (GLOBE, NCT02511405↗) using combination therapy did not show a clinical benefit in recurrent GBM patients (mOS, 6.8 vs. 7.9 months; mPFS, 3.4 vs. 3.7 months) [149]. Another adenoviral vector that delivers tumoricidal gene is the HSV-TK gene, which converts prodrugs (e.g., valacyclovir, acyclovir, and ganciclovir) into toxic nucleotide analogs to kill tumor cells. The aglatimagene besadenovec (AdV-tk), a nonreplicating adenovirus expressing the HSV-TK gene, was found to be tolerated in phase I trials [150–152]. Two phase II trials demonstrated favorable PFS and OS associated with AdV-tk-based therapy in GBM patients [153, 154]. Moreover, a randomized and controlled phase III study also revealed that, compared with SOC treatment, AdV-tk gene treatment extended the survival of GBM patients (mOS, 12.9 vs. 8.6 months) [155]. However, another randomized phase III trial (ASPECT) demonstrated that sitimagene ceradenovec, another adenoviral vector used to deliver the HSV-TK gene, failed to improve survival in newly diagnosed GBM patients (mOS, 16.6 vs. 15.1 months; mPFS, 10.3 vs. 8.9 months), although it increased the time to death or reintervention [156]. Therefore, adenovirus-based delivery of HSV-TK needs to be further investigated for GBM treatment. However, a phase I study (NCT01811992↗) in newly diagnosed GBM patients revealed an mOS of 21.3 months when AdV-tk was combined with another adenovirus expressing FMS-like tyrosine kinase 3 ligand (Flt3L) [157].

In addition to adenovirus, the HSV-TK gene can be delivered by retroviruses. However, a phase III study revealed that retrovirus-based HSV-TK gene therapy did not generate significant clinical benefit for GBM patients compared with SOC treatment (mOS, 12.0 vs. 11.8 months; mPFS, 6.0 vs. 6.1 months) [158]. Vocimagene Amiretrorepvec (Toca 511) is another retrovirus-based treatment iteration that encodes a yeast cytosine deaminase gene and can convert the prodrug 5-fluorocytosine (5-FC) to toxic 5-fluorouracil (5-FU) [159]. Although the phase I trial resulted in encouraging efficacy data and a good safety profile [160, 161], the subsequent phase II/III trial failed to confirm the survival benefit of Toca 511/FC in GBM patients compared with SOC treatment (mOS, 11.1 vs. 12.2 months; mPFS, 6.0 vs. 6.1 months) [162]. Therefore, certain approaches (e.g., selecting the most effective virus type, optimizing the delivery method, refining genetic engineering, and combining other immunotherapy strategies) should be taken into consideration for enhancing the efficacy of virus-based therapies in GBM.

The brain has historically been considered an “immune privileged” organ, given that many studies have shown that heterotopic tissues are not rejected by animal brains, whereas the same tissues are eradicated by the host immune system when they are introduced into peripheral tissues [164, 165]. This is attributed to the lack of lymphatic drainage in the brain [166–170]. However, more studies have demonstrated that the brain engages with the immune system. Routes for antigenic egress from the brain to deep cervical lymph nodes were identified in the 1980s [167, 168, 170]. In 2015, the discovery of the presence of a dural lymphatic system within the CNS revealed a mechanism for draining CNS antigens from cerebrospinal fluid into cervical lymph nodes, facilitating immune surveillance [166, 169]. Moreover, the glial-lymphatic (glymphatic) pathway has been recently identified, providing insight into how fluids and solutes are cleared from the brain [171, 172]. Specifically, cerebrospinal fluid enters the brain along arterial perivascular spaces, engages in the interstitium through aquaporin 4 water channels, and then exits through venous perivascular spaces, draining into deep cervical and lumbar lymph nodes [172]. Therefore, the brain is an immunologically dynamic organ and should be described as “immunologically distinct” rather than “immune privileged” (Fig. 2).

The BBB, which consists of capillary endothelial cells, the basement membrane, the perivascular space, and glia limitans, is one of the inherent obstacles that impedes the entry of immune cells and immunotherapeutic drugs into the brain [173]. The BBB serves as a structural barrier that restricts the diffusion of large and hydrophilic molecules into the CNS. It also protects the brain from most blood-borne pathogens and exogenous compounds (i.e., drugs and neurotoxins) that might damage the CNS. Although the BBB is relatively impenetrable to circulating immune cells and antibodies in a quiescent state, peripheral immune cells can cross the BBB and execute vigorous inflammatory responses when danger signals are detected, providing the foundation for immunotherapy against brain tumors [174, 175]. In GBM, the BBB is partially disrupted with increased permeability due to increased angiogenesis and the release of cytokines and chemical mediators [176]. Although BBB disruption in GBM primarily occurs within the tumor core, which may enhance therapeutic delivery, the intact BBB at tumor edges can still impede the distribution of immune-related drugs [177]. To address this problem, an increasing number of studies have aimed to improve drug penetration ability by increasing drug liposolubility with liposomes or by remodeling the BBB properties via the regulation of efflux pumps and tight junctions. In addition, local administration of immunotherapeutic drugs (e.g., intrathecal drug administration, convection-enhanced delivery, and the use of implantable pharmaceutical formulations) might be another effective method for enhancing drug delivery [178]. However, randomized clinical trials are needed to evaluate the safety and effectiveness of these approaches.

Another unique immune characteristic of the brain is its resident microglia, which originate from yolk sac myeloid progenitors during development [179]. Microglia may undergo phenotypic changes in response to inflammatory stimuli [179]. Although the role of resident microglia in adaptive immunity is not fully understood, they perform a variety of critical functions, including phagocytosis, cytotoxicity, and immune regulation [179]. In GBM, resident microglia can be polarized from proinflammatory to anti-inflammatory phenotypes, which exert multiple protumorigenic activities (e.g., promoting GBM cell proliferation and invasion and stimulating angiogenesis) and inducing immunotherapy resistance. These immunosuppressive microglia inhibit effector T-cell infiltration, proliferation, and immune reactivity via distinct mechanisms, thus contributing to tumor immune evasion [180].

GBM is considered an immunologically ‘cold’ tumor largely due to the low number of TILs and other immune effector cell types, which are related to poor responses to immunotherapies, such as ICIs [181]. Additionally, GBM can also induce systemic immunosuppression by sequestering naïve T cells within the bone marrow, primarily through the downregulation of sphingosine-1-phosphate receptor 1, which is crucial for T-cell egress from the thymus and secondary lymphoid organs into the bloodstream [182].

Within the GBM TME, many immunosuppressive cells, such as tumor-associated macrophages and microglia (TAMs), myeloid-derived suppressor cells (MDSCs), and Tregs, together contribute to immunosuppression and GBM cell immune evasion [179, 183, 184] (Fig. 2). Among them, TAMs, which include yolk sac-derived microglia and bone marrow-derived macrophages (BMDMs), constitute the predominant cell population in the GBM TME, representing as many as 50% of the total number of cells in the entire tumor mass [179]. Recent studies demonstrated that TAMs can be educated by tumor cells, which, in turn, promote tumor progression, inhibit antitumor immunity, and induce immunotherapy resistance [34]. Tregs are a subset of T cells that contribute to tumor progression and immunotherapy resistance [185]. Specifically, cancer and microenvironmental cell-secreted IDO1, IL-10, C-C motif chemokine ligand 2 (CCL2) and transforming growth factor beta (TGF-β) promote the expansion of immunosuppressive Tregs within the GBM TME [186]. As a result, these Tregs express immune checkpoint molecules (e.g., PD-1 and CTLA-4) to suppress the effector functions of T cells [184]. MDSCs are a highly heterogeneous population of myeloid cells that contribute to tumor immunosuppression [187]. GBM cell-derived macrophage migration inhibitory factor recruits MDSCs from the bone marrow into the TME [188], and tumor-infiltrating MDSCs can suppress cytotoxic T-cell activity by expressing arginase to reduce TCR expression and induce oxidative stress to secrete reactive oxygen species [189]. Additionally, MDSCs can promote T-cell exhaustion by expressing PD-L1 [190]. Together, these findings highlight that the presence of an immunosuppressive TME represents one of the key challenges for immunotherapy in GBM.

Moreover, recent reports suggest that environmental factors, including age and sex, are associated with GBM onset, severity, and the immune response. The onset of GBM generally occurs later in life, and aging remains a major variable that can impact outcomes [191]. In relation to immunotherapies, recent work has demonstrated changes in the immune microenvironment as a result of age, leading to older individuals being more refractory to ICIs [192]. In addition to age, sex is a major biological variable that impacts GBM growth and therapeutic resistance. From an epidemiological standpoint, males develop GBM more frequently [193] and have poorer outcomes [194]. A recent clinical report also revealed sex differences in MGMT methylation, where a survival advantage was observed only in females [195]. While these differences can be attributed to different signaling networks [196], recent work in preclinical models and human samples has shown that male T cells are more prone to exhaustion and more responsive to ICIs [197], whereas MDSC subset localization is sex biased, with males having an enrichment of monocytic MDSCs in the TME, whereas females having increased granulocytic MDSCs in their circulation [198]. Moreover, these differences were also leveraged for preclinical validation of sex-specific immunotherapies [198]. These observations further support earlier work in which sex differences in microglia also impacted GBM growth [199, 200]. Taken together, age and sex are emerging as key biological variables in GBM and have implications for the GBM immune response and immunotherapy strategies.

In addition to the immunosuppressive TME, the biological activities of cancer cells also contribute to immunosuppression and immunotherapy resistance in GBM (Fig. 2). First, the complexity and heterogeneity of GBM cells pose significant challenges for immunotherapies. GBM tumors exhibit both intertumoral and intratumoral heterogeneity, impacting their different responses to immunotherapies on the basis of molecular profiles, such as dysregulated p53, Rb, and phosphoinositide 3-kinase pathways [201]. In addition to interpatient heterogeneity, intratumoral variability further increases the difficulty of immunotherapy in GBM. For example, the individual tumor mass harbors a complex and dynamic architecture of tumor cells, exhibiting variability at the epigenetic, transcriptomic, protein, and metabolic levels [202, 203]. Moreover, treatment may induce GBM tumor phenotype switching, and relapsed GBM tumors might have different tumor subtypes and accessible targets than newly diagnosed tumors do [204]. Therefore, developing immunotherapy approaches that specifically target tumor cell subpopulations might be a new direction for GBM treatment.

Second, the highly invasive capacity of GBM cells might contribute to tumor recurrence and treatment resistance [205]. The different invasive potentials of GBM cells increase their intratumoral heterogeneity. Specifically, cells in the tumor core exhibit heightened proliferative capacity, whereas those at the tumor periphery are more prone to infiltration, enabling them to penetrate surrounding normal brain tissues [205]. After infiltrating brain tissues, GBM cells remodel the extracellular matrix, cytoskeleton, and metabolism [206]. Although GBM cells rarely metastasize to distant organs [207], this invasion and infiltration decrease the possibility of complete surgical resection and lead to a high chance of tumor recurrence even after multiple postoperative adjuvant therapies, including immunotherapy.

Third, GBM cells exhibit a low tumor mutation burden (TMB) and consequently have limited neoantigen presentation for effective T-cell recognition [208, 209]. This is primarily because the presentation of neoantigens depends on the abundance of mutations capable of generating neoepitopes [210]. Although some mutations can be immunogenic and presented by APCs, the majority of mutations do not become MHC-presented neoepitopes that can be recognized and targeted by T cells [211]. Moreover, tumor subclones without highly antigenic peptides might evade immune surveillance [211]. The low TMB-induced smaller antigen pool allows fewer immunogenic neoantigens to be exposed following immunotherapy, which increases the challenges of immunotherapy in GBM.

Fourth, GBM cells can express immunosuppressive molecules, such as PD-L1, IDO1, IL-10 and TGF-β, to cause the dysfunction and exhaustion of TILs [212]. For example, TGF-β secreted by GBM cells decreases the expression of the activated receptor natural killer group 2D on CD8⁺ T cells and NK cells, thereby reducing their cytotoxic effects on GBM cells [213]. Moreover, tumor cells also decrease their MHC expression levels to reduce the likelihood of tumor antigen presentation [214]. On the other hand, context-dependent signaling from GBM cells can affect myeloid cell biology, such as TAM migration, polarization, and activation (Fig. 3), generating context-dependent tumor-TAM symbiotic interactions to promote GBM progression and immunosuppression [212, 215]. For example, PTEN deletion/mutation in GBM cells increases the expression and secretion of lysyl oxidase to promote macrophage infiltration via activation of the β1 integrin-protein-tyrosine kinase 2 signaling axis in macrophages [216]. TP53 gain-of-function mutation leads to the upregulation of CCL2 and tumor necrosis factor α (TNF-α) via nuclear factor-kappa B (NF-κB) signaling, which, in turn, increases the infiltration of microglia and monocyte-derived immune cells [217]. In addition to genetic alterations, metabolic changes resulting from lactate dehydrogenase A (LDHA) overexpression in GBM cells increase CCL2 and CCL7 expression to attract macrophages into the TME [218]. Eventually, these GBM-associated TAMs are polarized toward an immunosuppressive phenotype, thus inhibiting T-cell-mediated antitumor immunity and inducing immunotherapy resistance [216–218].

Finally, GSCs constitute a subpopulation of GBM cells that are characterized by stem cell-like capabilities, which drive tumor self-renewal and clonal tumor initiation, eventually promoting intratumoral heterogeneity and therapeutic resistance [219]. While GSCs account for only a small fraction (~10%) of the total GBM cells within tumor tissues, a large population of proliferative cancer cells are composed of GSCs [219]. GSCs can escape immune surveillance by recruiting immunosuppressive PD-L1⁺ macrophages [220]. Crosstalk between GSCs and TAMs is an important mechanism within the GBM TME that promotes the immune escape of GBM tumors [212, 215]. For example, a gain-of-function screen of epigenetic regulators identified circadian locomotor output cycles kaput (CLOCK, a gene amplified in approximately 5% of GBM cases) as a key hit in GSCs that not only promotes stemness but also triggers microglial infiltration and immunosuppressive polarization by transcriptionally upregulating olfactomedin-like 3 and legumain (LGMN). As a result, these microglia promote tumor growth and suppress antitumor immunity [221, 222]. The Kunitz-type protease inhibitor TFPI2 is another example that is amplified in approximately 4% of GBM cases and can regulate GSC–microglia interactions. Specifically, TFPI2 is highly expressed in GSCs, where it not only promotes GSC self-renewal and tumor growth but also triggers microglial infiltration and immunosuppressive polarization to induce immunosuppression [223]. Moreover, neurofibromin 1 (Nf1)-deficient GSCs isolated from tumors of the Nf1 genetically engineered mouse model can produce the chemokines C-X3-C motif ligand 1 and CCL5 to recruit microglia into the TME [224], which ultimately exhibit an immunosuppressive function.

The combination of immunotherapies with chemotherapy, particularly TMZ, is one of the most investigated strategies for GBM treatment. A phase II study in newly diagnosed GBM patients demonstrated that combining the anti-PD-1 antibody pembrolizumab with TMZ and tumor treatment resulted in an mPFS of 12.0 months compared with 5.8 months in case-matched controls [225]. However, the adjuvant anti-CTLA4 antibody ipilimumab combined with TMZ after surgery and chemoradiotherapy did not result in a survival benefit for newly diagnosed GBM patients (mOS, 22.7 vs. 26.4 months; mPFS, 10.9 vs. 12.5 months) [42]. IDO1-targeted ICI therapy is being investigated in combination with TMZ and RT in newly diagnosed GBM patients (NCT04047706↗ and NCT02052648↗). Moreover, other immunotherapeutic strategies, including the peptide vaccine SurVaxM [83, 84], the DC vaccine DCVax-L [100], and the oncolytic adenovirus DNX-2401 [138], in combination with TMZ are being evaluated in clinical trials, and some of them have already shown encouraging results in GBM patients.

RT can cause molecular damage (e.g., DNA breaks and base modifications) and trigger immunogenic cell death in GBM cells [226]. With respect to the immune response, RT can promote the release of tumor neoantigens and inflammatory cytokines, recruit effector T cells to the tumor site, and activate the cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway, laying the foundation for RT in combination with immunotherapies [226]. Recently, several clinical trials have been designed to explore the safety and efficacy of such combination therapy. For example, a phase I study demonstrated that combination treatment with hypofractionated RT (5 × 6 Gy), the anti-PD-1 antibody pembrolizumab, and the anti-VEGF antibody bevacizumab was well tolerated in GBM patients and induced promising survival results (mOS, 13.4 months; mPFS, 7.9 months) [227]. Further clinical studies are needed to explore the optimal dose and fractionation scheme of this combination strategy for GBM patients.

Although the development of a therapeutic approach that combines immunotherapy with SOC is promising, some concerns and associated solutions should be taken into consideration. First, systemic TMZ chemotherapy could cause a reduction in lymphocytes, and for this reason, local TMZ treatment might be a better administration approach when combined with ICIs [228]. Second, patients whose MGMT promoter is not methylated are resistant to TMZ treatment [229], and MGMT promoter methylation may be considered a biomarker for patient selection. Third, immunotherapy efficiency is affected by the dose and treatment timing, which should be determined before clinical trials. Together, the combination of immunotherapy with SOC shows great therapeutic potential, and further clinical trials are needed to validate the efficiency of such a combination treatment strategy in GBM patients.

Although ICIs are widely used to treat a variety of cancers, such as melanoma and non-small cell lung cancer, the efficacy of ICIs in GBM patients is still disappointing [163, 230]. One of the key reasons for their ineffectiveness is the low immunogenicity of GBM with a relatively low number of TILs. Indeed, the efficacy of ICIs in GBM is highly dependent on the abundance and activation status of TILs, especially T cells [28]. Therefore, other immunotherapeutic approaches aimed at increasing T-cell infiltration and activation, such as CAR-T-cell therapy, cancer vaccines, and OV therapy, might be potential candidates for overcoming ICI resistance in GBM. Here, we discuss the therapeutic potential of combining ICIs with other types of immunotherapies for GBM treatment (Fig. 4).

Myeloid cells play crucial roles in regulating GBM progression and immunosuppression and inducing immunotherapy resistance [179, 243]. GBM is considered a “cold” tumor infiltrating high levels of immunosuppressive myeloid cells [212]. TAMs constitute the largest population of myeloid cells in the GBM TME, where their infiltration and immunosuppressive polarization are usually triggered by factors secreted by GBM cells [179]. As a result, these polarized TAMs contribute to several GBM tumor hallmarks, including immunosuppression [215]. Blockade of TAM infiltration and immunosuppression markedly suppresses tumor progression and activates antitumor immunity in GBM mouse models [218, 221–223]. These findings suggest that targeting TAMs is a promising antitumor strategy that has the potential to affect ICI therapy efficacy in GBM. Here, we discuss advances in therapeutic strategies that combine myeloid cell-targeted therapies with ICIs in GBM.

The cGAS-STING pathway serves as a sensor of cellular stress that can activate innate immunity and antigen presentation by myeloid cells and is recognized as the most promising therapeutic target for GBM treatment [244–246]. Indeed, STING agonists function as potential therapeutic drugs to influence the TME, including the infiltration of inflammatory macrophages and neutrophils and the upregulation of PD-L1 [247]. As a result, the combination of the clinical-grade STING agonist 8803 with an anti-PD-1 antibody markedly enhances the survival of GBM tumor-bearing mice [248]. Histone 3 lysine 27 demethylase (KDM6B) is an epigenetic enzyme that is highly expressed in immunosuppressive myeloid cells (e.g., immunosuppressive macrophages). Inhibition of myeloid cell KDM6B genetically (using myeloid cell Kdm6b-specific knockout mice) or pharmacologically (using the KDM6B inhibitor GSK-J4) sensitizes GBM-bearing mice to anti-PD-1 therapy [249]. LGMN is a protease that is highly enriched in TAMs. Inhibition of LGMN genetically and pharmacologically enhances T-cell-mediated antitumor immunity and synergizes with anti-PD-1 therapy in GBM tumor-bearing mice [250]. In addition to LGMN, sialic acid binding Ig like lectin 9 (SIGLEC9), which is expressed in a unique immunosuppressive macrophage subpopulation from GBM patients who do not respond to anti-PD-1 treatment, has been identified. Deletion of SIGLEC9 in GBM mouse models synergizes with anti-PD-1 or anti-PD-L1 therapy [251]. Similarly, MDSC-targeted therapies have the potential to improve the antitumor efficacy of ICIs. For example, C-X-C motif chemokine receptor 4 (CXCR4) is commonly overexpressed in tumor-associated MDSCs, and anti-CXCR4 therapy can synergize with anti-PD-1 therapy to increase the survival of GBM tumor-bearing mice [252]. Together, these findings suggest that strategies that target myeloid cells can increase the antitumor efficacy of ICIs in GBM mouse models.

Emerging evidence indicates that the symbiotic interaction between GBM cells and myeloid cells is critical for immunotherapy resistance [212, 253]. On the basis of the understanding and progress in the field, here, we summarize recent findings highlighting the therapeutic potential of targeting tumor–TAM symbiosis to increase the effectiveness of ICIs in GBM. For example, the inhibition of GSC–microglia symbiosis via the targeting of CLOCK and its downstream CD162 can activate antitumor immunity and synergize with anti-PD-1 therapy in GBM mouse models [221]. TFPI2 is another key mediator that can modulate GSC–microglia interactions. Blockade of this symbiosis by genetic depletion of TFPI2 in GSCs or pharmacologic inhibition of the TFPI2 receptor CD51 and its downstream STAT6 in microglia activates antitumor immunity and enhances the therapeutic efficiency of anti-PD-1 therapy in animal models of GBM tumors [223]. In addition to these newly identified strategies, we recently reviewed the accumulating approaches that can target tumor-TAM symbiosis to improve the effectiveness of ICIs in GBM [212, 253]. Although these findings highlight the importance and therapeutic potential of targeting myeloid cells and/or tumor-myeloid cell symbiosis to improve the effectiveness of ICIs in GBM, the current knowledge in this field is still at an early stage, and clinical trials are needed to validate this combination therapeutic strategy in GBM patients.