Advanced T and Natural Killer Cell Therapy for Glioblastoma

Lymphokine-activated killer (LAK) cells

Adoptive T cell therapy has developed from LAK cell therapy, which transfers a mixture of IL-2-activated T cells and NK cells obtained by culturing patients’ peripheral blood mononuclear cells (PBMCs) in the presence of IL-2 to patients with malignant glioma [67]. The main cytotoxic property of LAK cells is mediated by CD3^-CD56⁺ NK cells rather than CD3⁺CD56^- T cells [159]. However, their limited cytolytic activity due to a lack of tumor specificity and their IL-2-related toxicity, such as brain edema and aseptic meningitis, have prevented widespread use of this strategy [9,70].

NK cells, CD3^-CD56⁺ lymphocytes, play a pivotal role in the innate immune response. They contribute to the antitumor immune response by modulating T cell activation through the regulation of DC maturation, as well as by directly eliminating tumor cells [215]. They frequently infiltrate glioblastomas, but their lytic activity is decreased by the abundant immunosuppressive mechanisms of tumor cells and their microenvironment. Tumor recognition and the elimination of tumor cells by NK cells can be markedly enhanced through the expression of genetically engineered CARs. NK cell therapy will be discussed later.

TILs

TILs are effector T cells that are thought to have tumor specificity because they are already present in the tumor. Ex vivo expanded TILs are apt to proliferate in vivo and show functional activity and trafficking to the tumor [186]. It has been very difficult, however, to expand TILs from tumor tissues in most cancers, including glioblastomas, except melanomas [8]. Although some clinical studies in patients with recurrent [162] or newly diagnosed malignant glioma [163] used autologous TILs expanded ex vivo from cells in the draining inguinal lymph nodes after inoculation with irradiated autologous tumor cells have demonstrated a partial radiographic response, no survival benefit for the patients has been found. These results indicate that gliomas undoubtedly present immunosuppressive obstacles to TIL therapy, and a drastic improvement is needed in TIL expansion and the maintenance of TIL function in the immunosuppressive microenvironment of gliomas.

Adoptive CD8⁺ T cells

Selecting the target antigens is an important step in the efficient induction of antitumor immunity in effector T cells. Human CMV has been verified as a contributing factor of glioma progression [32] and suggested as a therapeutic target [81]. CMV pp65 antigens are recognized by a high fraction of T cells [127]. Because they are expressed in most glioblastomas (>90%) but not in normal brain tissue [33], they have been in the spotlight as target antigens for glioblastoma immunotherapy. In the first clinical study of adoptive immunotherapy using CMV-specific T cells in patients with recurrent glioblastoma, the treatment was shown to be safe with minimal toxicity to patients. The median overall survival (OS) of 19 patients was >57 weeks, with a median progression free survival (PFS) of >35 weeks [187]. Although CMV-specific immunotherapy showed disease stabilization and prolonged PFS in some patients, no correlation between antigen-specific T cell frequency and clinical outcomes was detected in that study. The tumor infiltrating CMV-specific T cells of some patients who showed tumor progression after T cell infusion displayed poor cytotoxic capacity and increased expression of inhibitory receptors such as PD-1, CTLA-4 and TIM-3, compared with T cells from peripheral blood. Moreover, regulatory T cells (Tregs) were detected at levels almost 5-fold higher in glioblastomas than in peripheral blood. Those disappointing results were ascribed to the infusion of CMV-specific CD8⁺ T cells without CD4⁺ helper T cells, the possible expansion of both Tregs and T cells in the presence of IL-2 loading, and including patients who were treated with T cells without prior lymphodepleting chemotherapy to make room for the infused T cells [223]. A phase I/II clinical trial using CMV pp65-specific T cells to remedy those limitations in patients with recurrent and newly diagnosed glioblastoma after lymphodepleting dose-dense temozolomide (TMZ) treatment, however, also found that the infused cells had insufficient cytotoxic function to control glioblastoma [223]. CMV-specific T cells might kill autologous tumor cells effectively in only a subset of CMV seropositive patients, which suggests that heterogeneity in CMV antigen expression could attenuate the effector function of T cells. Furthermore, the finding that CMV-specific T cells within the glioblastoma microenvironment were immunologically dysfunctional indicates that the tumor microenvironment (TME) of glioblastomas might be exceptionally immunosuppressive. Therefore, additional modulation will be needed to obtain effective antitumor immune responses to adoptive T cell therapy that targets CMV pp65.

In another clinical trial (NCT00693095) that tested CMV-targeting T cells with TMZ in 17 patients with newly diagnosed glioblastoma, an additional vaccination with CMV pp65 RNA-loaded DCs enhanced the frequency of polyfunctional CMV pp65-specific T cells after adoptive T cell therapy, correlating with prolonged OS [172]. Isolating CMV-specific T cells from glioblastoma patients with deficient polyfunctionality and then stimulating them with antigenic peptides in the presence of the γC cytokine ex vivo might reverse their inability to generate multiple cytokines and improve their ability to mount an effective antitumor response in vivo [35]. Repeated endogenous antigenic stimulation to adoptively transferred CMV pp65-specific T cells via DC vaccination seems to restore T cell polyfunctionality. The adoptive T cell therapy + DC vaccine platform can target tumor ribonucleic acid (RNA) instead of CMV pp65. Clinical trials to evaluate the safety and antitumor immune response of a tumor RNA-loaded DC vaccine and subsequent tumor RNA-specific T cell therapy in pediatric patients with newly diagnosed high-grade gliomas (NCT03334305) and patients with diffuse intrinsic pontine glioma (DIPG) (NCT03396575) are underway.

Genetically modified T cells

Genetic modification of T cells has been developed to enhance the antitumor efficacy of adoptive T cell therapy. Two approaches have commonly been used for this strategy : (a) transfer of complementary deoxyribonucleic acids in the α and β chains of the TCR cloned from high affinity TAA-specific T cells, and (b) insertion of CARs that recognize tumor cells through a single-chain variable fragment (scFv) isolated from TAA-specific antibodies [31,115].

CAR T cells

CAR T cell therapy refers to the adoptive transfer of T cells genetically modified to express CARs, recombinant molecules typically composed of an extracellular domain of a tumor antigen recognition molecule that contains the scFv of a monoclonal antibody, intracellular domains with a TCR signaling domain and an additional costimulatory domain that lead to T cell activation, and a transmembrane domain as a spacer [51,98]. The intracellular domain has been optimized in successive generations of CAR T cells to enhance its signaling capacity (Fig. 1). First-generation CARs that used only the CD3ζ chain as the intracellular activation domain demonstrated limited persistence and efficacy because they lacked costimulatory signals [16]. Second- and third-generation CARs were developed by combining the CD3ζ of the first-generation CARs with one (second-generation) or more (third-generation) costimulatory domains, such as CD28 and OX40 or 4-1BB [73,88]. The addition of cytokine signaling domains such as IL-15Rα or Janus kinase-signal transducers and activators of transcription into the intracellular domain of third-generation CARs produced fourth-generation CARs and enhanced the antitumor activity of third-generation CAR T cells [84,144]. Modified CAR T cells can recognize and kill target cells that express TSAs without the need for MHC presentation and costimulatory signals. In spite of the successful engineering of more potent and immunogenic CAR T cells, on-target off-tumor effects, poor tumor infiltration, and a highly immunosuppressive TME remain serious barriers to the clinical efficacy of CAR T cells for solid tumors, including glioblastomas [169].

Fig. 1.

Advances in chimeric antigen receptor (CAR) generation. CARs are composed of an extracellular domain of a tumor antigen recognition molecule that contains the single-chain variable fragment (scFv) of a monoclonal antibody, intracellular domains with a T cell receptor signaling domain and an additional costimulatory domain that lead to T cell activation, and a transmembrane domain as a spacer. The intracelluar domain has been optimized in successive generations of CAR T cells to enhance its signaling capacity. First-generation CARs utilized CD3ζ chain only as an intracellular activation domain. Second- and third-generation CARs were developed by combining CD3ζ with one (second-generation) or more (third-generation) costimulatory domains such as CD28 and OX40 or 4-1BB. Fourth-generation CARs incorporated a cytokine signaling domain such as interleukin-15 receptor alpha (IL-15Rα) or Janus kinase-signal transducers and activators of transcription (JAK-STAT) into the intracellular domain of third-generation CARs.

It is also important to select appropriate antigens to target in CAR T cell therapy. An ideal target antigen is homogenously expressed on all tumor cells and completely absent on normal cells. To prevent normal cell attack by CAR T cells, which have greater cytotoxicity than non-engineered T cells, tumor antigens should be undetectable or have minimal expression on normal tissues not enough to mediate the elimination of normal cells. It is very important, therefore, for phase I clinical trials in patients with glioblastomas that mainly express TAAs to confirm the safety and feasibility of this treatment. Antigens selected for CAR T cells in glioblastomas to date include IL-13Rα2, EGFRvIII, HER2, and EphA2. To overcome the antigenic variability of glioblastomas and increase the efficacy of CAR T cells, bi- and tri-specific CAR T cells that target multiple TAAs have been developed and shown increased antitumor efficacy [12,71]. Ongoing clinical trials of CAR T cell therapy for glioblastoma are summarized in Table 1.

Table 1.

On-going clinical trials of CAR T cell therapy for glioblastoma

Targeting multiple antigens

The antigenic and molecular profiles of glioblastoma are strikingly heterogeneous in terms of pathology and genetic changes, even within a single tumor [156]. So, glioblastoma cells without the targeted antigen can escape CAR T cell recognition and elimination. In addition, preclinical and clinical studies have shown that targeting a single antigen can result in antigen loss variants during subsequent tumor recurrence [99,150]. Strategies to prevent such escape include efforts to expand the list of available TSAs, such as mutation-derived neoantigens, and to engineer CAR T cells to achieve multispecificity. To date, two preclinical studies from the same research group have used CAR T cells to target multiple antigens in glioblastomas. One study used bispecific CARs composed of signaling domains (CD28 + CD3ζ) and a tandem CAR (TanCAR) exodomain that fused a HER2-binding scFv to an IL-13Rα-binding IL-13 mutein. The TanCAR T cells displayed enhanced antitumor efficacy and improved animal survival compared with the effects of single CAR T cells encountering HER2 or IL-13Rα2 [71]. The other study designed trivalent CAR T cells that targeted HER2, IL-13Rα2, and EphA2. The trivalent CAR T cells exhibited improved cytotoxicity and cytokine release compared with monospecific and bispecific CAR T cells in vitro. They were also able to control tumor growth at low T cell doses in autologous glioblastoma patient-derived xenografts [12]. Further clinical information is required to determine whether CAR T cells targeting multiple antigens can be efficient in the immunosuppressive TME of human glioblastomas.

Neoantigens

Neoantigens are real TSAs because they are a series of peptides present in tumor cells but not in normal cells. Therefore, neoantigens derived from tumor-specific mutations can generate potent immune responses with central tolerance and without toxicity to normal tissue [242]. The most common method for identifying personalized neoantigens compares DNA sequences in tumor tissues with those in normal tissue [108]. An efficient sequencing tool currently in wide use is whole exome sequencing technology [209].

Since the first personalized neoantigen-pulsed DC vaccine began to be tested in a phase I clinical trial for melanoma in 2015 [22], various clinical trials of neoantigen-loaded DC vaccines for solid tumors have been conducted [44,151,152,180,184] and shown therapeutic value against cancer. In a phase I/Ib study of personalized neoantigen DC vaccines for eight patients with newly diagnosed O⁶-methylguanine-DNA methyltransferase (MGMT)-unmethylated glioblastoma, neoantigen-specific T cells from the peripheral blood were shown to migrate into an intracranial tumor without dose-limiting toxicity. However, all patients showed tumor recurrence and ultimately died of progressive disease, with a median OS and PFS of 16.8 months and 7.6 months, respectively [89]. Thus, even the T cell response induced by a DC vaccine targeting neoantigens might be insufficient to produce clinically effective antitumor activity in the immunosuppressive TME of glioblastoma. Perhaps, reducing the number of steps between effector T cells and naïve T cells in the body would improve the antitumor activity of effector T cells. For example, in mouse tumor models, neoantigen-pulsed DC vaccines that used fewer steps to produce effector T cells endogenously were superior to neoantigen-adjuvant vaccines with one more step of antigen priming in both activating immune responses and inhibiting tumor growth [252]. Therefore, ex vivo expanded neoantigen-specific effector T cells that do not require an endogenous process to stimulate T cells might improve the antitumor immune response.

Cytokine overexpression

Incorporating a signal of immune stimulatory cytokines into third-generation CAR T cells produces fourth-generation CAR T cells, which have improved cell expansion and persistence [25]. Transgenic cytokine expression could also stimulate CAR T cells to lyse antigen-negative cancer cells not otherwise recognized by CAR T cells in the tumor. The cytokine tested in a glioblastoma animal model was IL-15, which plays an important role in T cell expansion and survival, especially in absence of the antigen [92,99]. IL-13Rα2 CAR T cells modified to express transgenic IL-15 (IL-13Rα2 CAR T/IL-15) showed greater proliferation and longer persistence, and produced more cytokines than IL-13Rα2 CAR T cells in vitro. Those results also produced a survival benefit in vivo, but late recurrence of tumors with downregulated IL-13Rα2 expression and antigen loss was observed [99]. Although genetically modifying CAR T cells to express transgenic cytokines might be a powerful method to improve their antitumor activity, multiple antigen targeting might also be required to prevent the occurrence of antigen loss variants during tumor recurrence.

Bispecific T cell engagers (BiTEs)

BiTEs are a subclass of bispecific antibody composed of two different antibody fragments, one of which is specific for CD3ζ on T cells and the other of which recognizes a tumor antigen [233]. Therefore, BiTEs can act as an immunologic synapse to facilitate an optimal interaction between cytotoxic T cells and tumor cells without the need for co-stimulation or MHC recognition [48]. Their antitumor effects and safety have been shown in clinical trials for various hematologic malignancies [10,143] and solid tumors [94,200]. In glioblastoma mouse models, bispecific T cells expressing both EGFRvIII CAR and a BiTE against EGFR eliminated heterogeneous EGFRvIII-expressing tumors that monospecific EGFRvIII CAR T cells were unable to treat completely [28]. Fn14, a cell surface receptor of the TNF-related weak inducer of apoptosis, is upregulated in gliomas, and its overexpression can stimulate the migration and invasion of glioma cells, so it is associated with poor prognosis [206]. In a preclinical study testing the antitumor activity of three therapeutic approaches, an Fn14×CD3ζ BiTE antibody, Fn14-specific CAR T (Fn14 CAR T) cells, and Fn14 CAR T cells engineered to secrete IL-15 (Fn14 CAR T/IL-15) against glioblastomas, both the Fn14xCD3ζ BiTE antibody and Fn14 CAR T cells showed cytotoxic effects in vitro and in vivo, and IL-15 production augmented the antitumor effects of CAR T cells, resulting in longer remission and survival [106]. Although BiTEs or BiTE-secreting CAR T cells have shown promising antitumor activity in preclinical studies, it is necessary to further verify whether they will work well in the immunosuppressive TME of human glioblastomas. A phase I clinical study (NCT04903795) to evaluate the safety of EGFRvIII×CD3ζ BiTEs alone and in combination with a peripheral autologous T cell infusion in patients with EGFRvIII-mutated grade IV malignant glioma is ongoing.

Disrupting immunosuppressed molecules

The widespread adoption of gene-editing technologies has enabled the random insertion or deletion of specific transgenes to or from CAR T cells. Incapacitating immune inhibitory molecules such as PD-1 and transforming growth factor-β (TGF-β) on T cells could be one strategy for overcoming the immunosuppressive TME of glioblastomas.

Methods for blocking or reversing immune inhibitory molecules include combination therapy of CAR T cells and PD-1 blocking antibodies [190,194], CAR T cells that secrete PD-1 blocking antibodies [198], CAR T cells with PD-1 gene-knockout [29], and CAR T cells with a PD-1 chimeric switch receptor that reverses the inhibitory signal of PD-1 activation into a stimulatory signal [116]. Because Treg cells also express PD-1, systemic treatment with PD-1/PD-L1 blocking agents could enhance Treg cell function, leading to significant suppression of antitumor immune responses and subsequent hyper-progression of cancers [23,90]. The advantages of CAR T cells with an intrinsic PD-1 blockade produced by genetic engineering are that they provide more sustainable activity and more tumor-limiting PD-1 inhibition than CAR T cells combined with antibody treatment. Targeted disruption of PD-1 using the clustered regularly interspaced short palindromic repeats-CRISPR associated protein 9 (CRISPR-Cas9) system enhanced the antitumor activity of EGFRvIII CAR T cells in vitro and significantly prolonged the survival of mice bearing glioblastomas [29]. However, a sustained PD-1 blockade by genetic deletion can promote the accumulation of terminally differentiated, exhausted CD8⁺ T cells [148,224] and raise safety concerns about the occurrence of CRS through the supraphysiological activation of CAR T cells that would result from continuously blocking the physiologic function of PD-1, which is to inhibit excessive T cell activation. Clinical trials, therefore, should only be conducted after sufficient consideration of the problems that could arise from a persistent PD-1 blockade.

TGF-β is a powerful immunosuppressive factor that has been shown to promote T cell exclusion and dysfunction in most solid tumors, including glioblastomas [128,161]. TGF-β blockade can facilitate the efficacy of adoptive T cell therapy for glioblastomas. Recent advanced CAR engineering can convert immunosuppressive molecules into T cell stimulants for CAR T cell therapy. Actually, CARs responsive to a variety of soluble ligands, including TGF-β, can be constructed to effectively convert TGF-β from a potent immunosuppressive cytokine into a strong stimulant for T cells [24]. Such CAR T cells could inhibit endogenous TGF-β signaling in T cells. Because that approach is not directly involved in tumor cell lysis, it might require a combination of receptors to recognize surface antigens for direct tumor cell killing.

Enhancing T cell trafficking

Making CAR T cells accumulate in the TME for a long time is another method for increasing antitumor efficacy. CAR T cells can be engineered to express chemokine receptors, such as C-X-C motif chemokine receptor 1 (CXCR1) and CXCR2, and thereby enhance intratumoral T cell trafficking. CD70, a member of the TNF family, is a novel immunosuppressive ligand and glioma target [79]. CD70 CAR T cells modified to express IL-8 receptors, CXCR1, and CXCR2 had greater trafficking to the tumors via radiotherapy-induced IL-8 upregulation, which resulted in complete tumor regression and a long-lasting memory T cell response in preclinical models of malignant tumors, including glioblastoma [80]. A subsequent phase I study (NCT05353530) of IL-8 receptor-modified CD70 CAR T cell therapy in CD70 positive and MGMT-unmethylated adult glioblastoma is underway.

Allogeneic CAR T cells

CAR T cell therapy based on autologous T cells has some limitations, including treatment delay due to production time, high cost, and the risk of manufacturing failure, and the functional availability of T cells is often reduced by the disease itself or previous therapies [254]. Allogeneic or universal “off-the-shelf” CAR T cell therapy using T cells obtained from healthy donors could be an alternative. Advances in gene-editing technologies, including zinc finger nuclease (ZFN) [168], transcription activator-like effector nuclease [164], and CRISPR/Cas9 [174], might be used to remove the two main barriers to allogeneic CAR T cells : graft-versus-host disease (GVHD) and allorejection. These exquisite gene-editing tools can generate TCR-deficient T cells to prevent GVHD [164,174,204] or T cells that eliminate MHC class I molecules by disrupting the β2-microglobulin locus to reduce allorejection [217]. In a recent phase I clinical trial, the safety and feasibility of allogeneic CAR T products were evaluated in patients with recurrent glioblastomas. Healthy donor-derived CAR T cells targeting IL-13Rα2 were generated and engineered using ZFNs to permanently disrupt the glucocorticoid receptor GRm13Z40-2, making them resistant to glucocorticoid treatment. Allogeneic Grm13Z40-2 T cells combined with an intracranial infusion of IL-2 and systemic dexamethasone maintained their effector function in the presence of dexamethasone, which was used to reduce the tumor-related brain edema as well as the rejection of therapeutic allogeneic T cells, and induced transient tumor reduction and/or tumor necrosis at the T cell infusion site in four of six treated patients without evidence of GVHD [19]. This first-in-human experience demonstrated the feasibility of using allogeneic CAR T products to treat glioblastomas.

Autologous NK cells

In early clinical trials, autologous ex vivo-expanded NK cell-rich effector cells derived from PBMCs were administered to patients with recurrent glioblastoma [77,113]. The autologous NK cell therapy was safe, but its antitumor effects were limited.

Allogeneic NK cells

Allogeneic HLA-mismatched NK cells have been in the spotlight as an alternative to autologous cells. Because allogeneic NK cells can bypass inhibitory signals and carry a low risk of GVHD [158,178], they are expected to have more potent antitumor efficacy than autologous cells. Moreover, glioblastoma stem-like cells seem to be highly susceptible to the cytotoxicity of allogeneic NK cells [6,69]. The sources of allogeneic cells include embryonic stem cells (ESCs), induced pluripotent stem cells (iPSCs) [49], umbilical cord blood [211], cell lines such as NK-92 [93], and the PBMCs of healthy donors [133]. Because of the low yield of NK cells available from allogeneic sources and the low transduction efficiency of CAR constructs, it could be reasonable to use NK cell lines. The NK-92 cell line is approved by the United States Food and Drug Administration (US FDA) for use in clinical trials [202]. NK-92 cells can expand without limit and are uniform, well-characterized, reproducible, and easily modifiable through genetic engineering [192,199]. They present impaired ADCC due to a lack of CD16, which can be re-expressed by CRISPR/Cas9 editing to increase antitumor cytotoxic activity [75]. Because the NK-92 cell line is derived from human NK cell lymphoma, it requires irradiation prior to infusion into patients due to safety concerns, such as chromosomal abnormalities and the risk of malignant transformation [175]. This radiation treatment does not affect the cytotoxicity of NK-92 cells, but it impairs their proliferation and reduces trafficking to the tumor, so it does limit their therapeutic efficacy [72]. Therefore, CAR NK-92 cells must be administered several times. Allogeneic NK cell transfer therapy has also been found to be safe. The antitumor efficacy of allogeneic NK cell therapy has been demonstrated in treating hematologic malignancies and to a lesser extent in studies on solid tumors [60]. The US FDA approved a clinical trial (NCT04489420) to evaluate the safety and feasibility of using human placental hematopoietic stem cell-derived NK cells (CYNK-001) in patients with recurrent glioblastoma in July 2020, but unfortunately that study was terminated in January 2022.

CAR NK cells

Because NK cells have potent antitumor cytotoxicity and a favorable safety profile, they are the most frequently explored candidate for generating CARs. NK cells do not produce IL-1 or IL-6, the main cytokines involved in CRS [192], and they display a low risk of GVHD [158,178]. If CARs are genetically engineered on NK cells, they will have greater antitumor activity than CAR T cells and minimal toxicity. The development of CAR NK cells for glioblastomas has followed a path similar to that of CAR T cells. CAR NK cells used CD3ζ as the first signal domain and then costimulatory domains such as CD28 and CD137 (4-1BB) were added. These conventional costimulatory domains, which are not found in NK cells [72], can be changed into NK-specific signaling domains such as NKG2D, CD244 (2B4), DAP10, or DAP12 to promote NK cell activation and cytotoxicity [62,130]. Compared with CAR T cells, CAR NK cells carry significantly fewer safety concerns, such as CRS and GVHD [158]. Moreover, CAR NK cells maintain their activating receptors, including NKp30, NKp44, NKG2D, and DNAM-1, which could reduce tumor recurrence caused by the loss of CAR targeting antigens [192]. The CAR targets of NK cells, including HER2, EGFR, EGFRvIII, and NKG2D, are very similar to those of T cells discussed above.

The potent anti-glioblastoma activity of CAR NK cells targeting EGFR [61], EGFRvIII [61,141,142], both EGFR and EGFRvIII [61,68], and ErbB2 (HER2) [4,249,250] has been shown in various preclinical studies. The route of delivery is also important in CAR NK cell therapy. Intravenously injected CAR NK-92 cells did not cross the BBB without ultrasound disruption in murine models, resulting in no therapeutic efficacy for intracranial tumors [4]. Even though the BBB environment in animals might be different from that in human glioblastoma patients whose BBB can be broken by the tumor, intratumoral delivery has been the preferred route of administration. In preclinical glioblastoma mouse models, repeated intratumoral injections of ErbB2 (HER2)-specific CAR NK-92 cells (NK-92/5.28.z cells) induced endogenous antitumor immunity and persistent protection against the tumor, with cures of initial syngeneic glioblastomas and the rejection of rechallenged tumor cells at distant sites [249]. A subsequent phase I clinical trial, CAR2BRAIN (NCT03383978), to investigate a clonal intracranial ErbB2-specific NK-92/5.28.z CAR NK product combined with intravenous ezabenlimab in patients with recurrent HER2-positive glioblastoma is ongoing.

Protection from an immunosuppressive microenvironment

NK cell therapy for glioblastoma can encounter several obstacles, including the inhibition of NK cell infiltration into tumor sites, downregulation of target ligands or maintenance of cognate MHC class I molecule expression on the tumor cells, and the release of inhibitory cytokines and secretory factors such as TGF-β in the TME [183]. TGF-β impairs NK cell cytotoxicity and proliferation by inhibiting IFN-γ [117], and it also inhibits activating receptors such as NKKG2D [104] and ADCC [207]. So blocking the TGF-β signaling pathway could be a strategy to increase NK cell function [238]. Genetically engineering cord blood-derived NK cells to express dominant negative TGF-β receptor II, a mutant receptor lacking the kinase domain of TGF-β, has shown enhanced antitumor activity in preclinical studies of glioblastoma [246] and medulloblastoma [165]. Another approach is administering TGF-β inhibitors with the NK cells. In a xenograft glioblastoma mouse model, treatment with allogeneic NK cells in combination with inhibitors of integrin or TGF-β signaling or treatment with allogeneic NK cells whose TGF-β receptor 2 gene was edited to abrogate glioma stem cell-induced NK cell dysfunction produced significant tumor control and prolonged survival of the animals [189]. Those findings suggest that the integrin and TGF-β axis could be a potential therapeutic target of NK cell therapy in glioblastomas, as well as an important NK cell immune escape mechanism. A phase I clinical trial (NCT04991870) evaluating the feasibility and toxicity of engineered allogeneic cord blood NK cells with TGF-βR2 and NR3C1 deletion in recurrent glioblastoma is in progress. Mothers against decapentaplegic homolog 3 (SMAD3) could be an another target for NK cell therapy. SMAD3 can induce TGF-β mediated NK cell suppression, so the suppression of SMAD3 could enhance NK cell activity [207]. Genetically engineered SMAD3-silenced NK-92 cells promoted IFN-γ production in NK-92 cells and inhibited tumor progression in xenograft mouse models of hepatoma and melanoma [222]. In addition, prostaglandin E2 (PG E2) secreted by cancer cells can promote cancer progression by inhibiting NK functions [153]. Blocking PG E2 has enhanced NK cell activity in preclinical models of metastatic breast cancer [123] and gastric cancer [110].

An approach targeting immune checkpoints can be applied to NK cells as well as T cells to improve their potential antitumor immunity. These immune checkpoints include NK cell-specific receptors such as KIR and NKG2A, and NK cell-expressed TIM-3, T cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine-based inhibition motif domains, CD96, and LAG-3 [5,101]. Although the role of those checkpoints in regulating NK cell function remains unclear and the mechanisms involved in their enhancement of NK cell function are controversial [91], further studies evaluating the therapeutic benefits of targeting them will provide new treatment options and improve NK cell function, producing better clinical outcomes.

Off-the-shelf CAR NK cells

Off-the-shelf products in effector cell therapy are allogeneic immune cells that can be manufactured on a large scale and distributed to treat a broad range of cancer patients [193]. Although CAR NK-92 cells with potent antitumor activity do not require strict HLA matching, can expand easily, and do not present safety concerns such as GVHD and CRS, making them good candidates for off-the-shelf products, they require irradiation prior to infusion into patients, which suppresses their proliferation. Moreover, these cells do not express CD16, which mediates ADCC, so they might have decreased lytic activity. iPSCs might also be standardized as an off-the-shelf therapy. Human iPSCs can produce NK cells effectively [95], and they are easier to genetically modify than ESCs and hematopoietic stem cells [111]. Theoretically, any somatic cell can be reprogrammed into an iPSC, but in practice, easily accessible cells such as skin, urine, or blood are commonly used [85].

iPSC-derived NK cells expressing CARs that use NK-specific NKG2D instead of conventional CD28 as the costimulatory signal (NK-specific CAR iPSC NK cells) have demonstrated enhanced NK cell activation and longer survival than T cells or iPSC-NK cells expressing a conventional CAR in a mouse xenograft model of ovarian cancer [111]. NK-specific CAR iPSC NK cells could be an attractive option as a safe and renewable off-the-shelf CAR NK therapy, but a variety of clinical studies to evaluate their antitumor efficacy and safety in solid tumors with heterogeneous tumor populations and immunosuppressive TMEs, including glioblastoma, will be required. Ultimately, to be ideal off-the-shelf effector cells for the treatment of glioblastomas, NK-specific CAR iPSC NK cells will need to be produced in large quantities and modified to express cytokines that play a major role in stimulating NK cell expansion and cytotoxic functioning such as IL-15 [171,232], to relieve the immunosuppression mediated by TGF-β released from the TME [57], to express checkpoint inhibitors (as seen in CAR T cell therapy) [109], and to enable multi-specific targeting [220]. The advantages and the disadvantages of CAR NK cells close to off-the-shelf effector cells for glioblastoma therapy are summarized in Table 3.

Table 3.

Comparison of the CAR NK cells close to off-the-shelf cell therapy for glioblastoma

Combination with conventional chemo-/radiotherapy

Lymphoid cells independently perform homeostatic regulation of resting and memory cells, so a rapid proliferation of remaining or infused lymphocytes occurs to recover normal lymphocyte numbers after periods of lymphopenia [59]. Because these homeostasis-induced T cells respond to tumor antigens at a lower dosage than naïve cells [26], the administration of TSAs in the form of a vaccine or ex vivo expanded adoptive T cell transfer during this recovery time can induce a disproportionate enhancement of effector cell populations that increases antitumor efficacy [96,181]. The induction of lymphodepletion in patients before T cell-based immunotherapy can be achieved using total body irradiation or non-myeloablative chemotherapy [157]. Another therapeutic advantage of lymphodepletion prior to immunotherapy is the ability to eliminate immunosuppressive cells, including myeloid derived suppressor cells (MDSCs) and Tregs [7]. Lymphodepletion before immunotherapy has been applied to various types of adoptive T cell therapy [176].

Conventional adjuvant therapies for patients with glioblastomas, such as radiotherapy and chemotherapy, are independently immunosuppressive [52]. However, they can also induce favorable immune responses by changing the TME to increase the antitumor efficacy of T cell therapy. In addition to cancer cell death caused by DNA damage, which triggers the release of danger signals, radiation can cause phenotypic changes in tumor cells that enhance tumor cell recognition and elimination, including the upregulation of MHC class I, NKG2D ligands, co-stimulatory receptor CD80, death receptor Fas, and intercellular adhesion molecule 1 [41,56,173]; the induction of proinflammatory cytokines such as TNF-α, IL-1β, IFN-γ, CXCL9, CXCL10 and CXCL16, which attracts T cells into the tumor [41,56]; and possibly the disruption of the tumor vasculature and the BBB, which would also increase T cell trafficking [179]. These immune-favorable responses in the TME can produce an endogenous and systemic antitumor immune response. The abscopal effect, an antitumor immune response that occurs outside the radiation field, suggests the potential for radiation-induced systemic antitumor immunity [56]. However, the rare occurrence of the abscopal response indicates that this endogenous immune response is generally weak. Thus, local radiation therapy can be harnessed in combination with immunotherapy to induce a potent systemic antitumor immune response. Combining CAR T cell therapy with radiotherapy has been shown to improve the antitumor efficacy of each monotherapy in preclinical models with a subset of solid tumors and glioblastoma [42,226]. In two independent syngeneic mouse models of glioblastoma, a sublethal dose of local radiotherapy combined with NKG2D CAR T cell therapy exerted synergistic activity by promoting the migration of CAR T cells to the tumor site, which increased the effector functions and prolonged survival [226]. Chemotherapeutic agents similar to local irradiation can also enhance the antitumor immunity of adoptively transferred T cells or CAR T cells via the upregulation of tumor antigens [78], elimination of immunosuppressive cells [253], and extension of cell survival [214].

Combination with immunotherapy

Immunotherapy, such as immune checkpoint inhibitors and oncolytic viruses, is another candidate for combination with CAR T cell therapy. Immune checkpoint inhibitors restore the activity of effector cells that can recognize and attack cancer cells. Despite clinical success in various cancers, including melanoma, non-small cell lung cancer, and renal cell carcinoma [203], anti-PD-1 monotherapy has not shown a significant survival benefit in patients with glioblastoma [177]. Immune checkpoint blockades that target the PD-1/PD-L1 and CTLA-4 pathways have been found to increase the activity of CAR T cells in preclinical studies of glioblastomas [194,245]. Two clinical trials of CAR T cell therapy combined with immune checkpoint inhibitors are currently progressing. One of them is a phase I clinical trial (NCT04003649) testing the safety and feasibility of L-13Rα2 CAR T cells administered alone or together with nivolumab (anti-PD-1) and ipilimumab (anti-CTLA-4) to patients with recurrent/refractory glioblastoma, and the other is a clinical study (NCT03726515) to assess the safety and tolerability of EGFRvIII CAR T cells in combination with pembrolizumab (PD-1 inhibitor) in patients with newly diagnosed EGFRvIII+, MGMT-unmethylated glioblastoma. However, systemic immune checkpoint inhibitors can block the checkpoints of Treg cells, as well as those of CAR T cells, which can enhance Treg cell function, as mentioned above.

Oncolytic viral infection and subsequent immunogenic tumor cell death can turn tumor cells into large-scale producers of tumor-specific neoantigens, cytokines, and chemokines, which converts the TME from immunosuppression to immune stimulation and thus augments T cell effector function and trafficking [86]. Oncolytic viruses can also induce the production of type I IFNs (IFN-αβ) in the TME to promote T cell proliferation, effector function, and immune memory function [37]. In addition, IFN-β can modulate the TME by inhibiting Treg cell activation and proliferation and disrupting tumor microvessels [255]. This potential for TME modulation suggests that oncolytic viruses and CAR T cell therapy could have synergistic antitumor effectiveness [3]. Combined therapy of B7-H3 CAR T cells and IL-7-loaded oncolytic adenoviruses has been shown to enhance T cell proliferation and reduce T cell apoptosis in vitro, and it produced a synergistic survival benefit in glioblastoma xenograft mouse models [74]. In another study of glioblastoma, a combination of oncolytic adenoviruses armed with CXCL11 and B7-H3 CAR T cells produced increased infiltration of the effector cells and decreased proportions of immunosuppressive cells such as MDSCs and Tregs in the TME. In animal models, glioblastoma that was not inhibited by B7-H3 CAR T cells alone was inhibited when CXCL11-armed oncolytic viruses were added [219]. In addition, a herpes simplex 1-based oncolytic virus expressing IL-15/IL-15Rα combined with EGFR CAR NK cells demonstrated increased synergistic antitumor effects and a significant survival gain in glioblastoma-bearing mice [122].

Combination with targeted therapy

Small-molecule inhibitors block intracellular signal transduction pathways such as tyrosine kinases and mitogen activated protein kinases in tumor cells, which deregulates cell proliferation and differentiation. Small-molecule tyrosine kinase inhibitors (TKIs) have been shown to have great antitumor efficacy in treating hematologic malignancies [191] and a variety of solid tumors [197]. Monotherapy with small-molecule TKIs, however, has displayed limited treatment outcomes in clinical studies of patients with glioblastoma [205]. Combining TKI therapy with CAR T cells in murine models produced synergistic antitumor effects in other types of solid tumors [107,235]. The combination of LB-100, a small-molecule inhibitor of protein phosphatase 2 A (involved in cell-to-cell adhesion), and CAIX-specific CAR T cells has been found to have synergistic antitumor effects in glioblastoma animal models [36]. These results suggest that TKIs could induce synergistic effects in combination with CAR T cell therapy for glioblastoma.

In a particularly useful innovation, CAR T cells can be designed to switch off by administering a small molecule that chemically disrupts a heterodimer [63]. CAR T cells incorporate a protease and CAR degradation moiety (degron) that can be switched on in the absence of the protease inhibitor asunaprevir. The degron is cleaved from the CAR by the protease, and it is switched off in the presence of asunaprevir; thus in the absence of the inhibitor, the degron is cleaved from the CAR by protease, leading to the degradation of the CAR [83]. These CAR T cells with a switch off function provide a controllable way to improve the safety of CAR T cell therapy and reduce the risk of CRS.