Cellular immunotherapy for high-grade glioma

Abstract

The outcome for patients with the most common primary brain tumor, glioblastoma multiforme (GBM), remains poor. Several immunotherapeutic approaches are actively being pursued including antibodies and cell-based therapies. While the blood–brain barrier protects brain tumor cells from therapeutic antibodies, immune cells have the ability to traverse the blood–brain barrier and migrate into GBM tumors to exert their therapeutic function. Results of Phase I clinical studies with vaccines to induce GBM-specific T cells are encouraging and Phase II clinical trials are in progress. Nonvaccine-based cell therapy for GBM has been actively explored over the last four decades. Here we will review past clinical experience with adoptive cell therapies for GBM and summarize current strategies on how to improve these approaches.

Primary brain tumors continue to pose significant clinical challenges. Tumors derived from astrocytes are the most common and among them glioblastoma multiforme (GBM) is the most aggressive. Despite advances in surgical procedures, radiation therapy and chemotherapy, the outcome for GBM is only slowly improving and 5-year overall survival rates remain low [1,2]. Thus, new targeted therapies are needed to improve current treatment strategies. Among them, immunotherapy is an attractive approach since it does not rely on the cytotoxic pathways of conventional therapies [3–5]. Cellular immunotherapies for GBM, which are nonvaccine based, have been explored for the last four decades (Table 1) [6–25]. In this article, we will review these approaches and summarize potential strategies on how to improve them. Vaccine-based approaches have been recently reviewed elsewhere [3,5,26,27].

Table 1.

Cell therapies for glioblastoma multiforme

Cell product	Clinical glioblastoma multiforme study (number of published studies)	Ref.
Autologous
Leukocytes (unmanipulated)	Yes (4)	[6–9]
NK cells
LAK cells	Yes (7)	[10–16]
NK cells (generated with APCs)	No
T cells
γδ T cells	No
Mitogen-activated T cells	Yes (3)	[17–19]
Antigen-specific T cells
TIL	Yes (1)	[20]
In vivo activated, ex vivo expanded	Yes (3)	[21–23]
Ex vivo activated and expanded with tumor cells	Yes (2)	[24,25]
Ex vivo activated and expanded with APCs	No
Genetically modified T cells
αβ TCR	No
Chimeric antigen receptor	Yes (1)	[136]
Modifications that enhance T-cell function	No
Allogeneic
T cells	Yes (1)	[88]

APC: Antigen-presenting cell; LAK: Lymphokine-activated killer; TIL: Tumor-infiltrating lymphocyte; TCR: T-cell receptor.

Glioblastoma multiforme & the immune system

There is convincing evidence that GBMs express antigens that are recognized by the patient’s immune system and that the cellular immune response, which is designed to kill virus-infected cells, can also recognize and kill GBMs [3–5]. Nonspecific killer cells, such as NK cells and lymphokine-activated killer (LAK) cells, recognize changes on the cell surface, such as carbohydrate abnormalities or low expression of MHC class I. T cells recognize ‘foreign’ peptides derived from cytosolic proteins presented on the cell surface by MHC molecules [28–30].

T cells recognize peptides derived from tumor-associated antigens (TAAs) expressed by GBMs. TAAs are immunogenic because they:

• Are normally only expressed during fetal development or at immunoprivileged sites;

• Are expressed at higher than normal levels;

• Contain a novel peptide sequence created by gene mutation or rearrangement.

Over the past decades, numerous TAAs have been identified and antigens relevant for GBMs are listed in Table 2 [31–56]. While the expression of TAAs indicate that glioma cells have the potential to be recognized by the immune system, the immune system fails to prevent the development of GBMs. This is probably due to the tumor location as well as the elaborate immune evasion strategies developed by GBMs.

Table 2.

Tumor associated antigens expressed in glioblastoma multiforme^†

Brain tumor antigens	T-cell therapy glioblastoma multiforme studies		Ref.
	Animal	Clinical
Mutations/novel epitopes in oncogenic proteins
EGFRvIII	Yes	–	[31,32]
Cancer testis antigens
MAGE, GAGE, SSX family	–	–	[33–35]
Overexpressed antigens
IL-13Rα2	Yes	Yes	[36,37]
HER2, gp100	Yes	–	[38–41]
Aim-2, Art-1, EphA2, EphB6, Ezh2, Fos11, GALT-3, GnT-V, HNRPL, hTert, B-cyclin KIF1C, KIF3C, MRP-3, NKG2D ligands, PTH-rP, Sart family, SOX5, SOX6, survivin, tyrosinase, Trp-1&2 Mart-1, Whsc2, YKL-40	–	–	[38,39,42–52]
Viral antigens
CMV	–	–	[53–56]

^†Determined either by RT-PCR, western blot or immunohistochemistry.

CMV: Cytomegalovirus; EGFR: EGF receptor; GAGE: G melanoma antigen; GALT: Galactosyltransferase; HNRPL: Heterogeneous nuclear ribonucleoprotein L; IL-13Rα2: IL-13 receptor subunit α-2; MAGE: Melanoma-associated antigen; MRP: Multidrug-resistance protein; NKG2D: NK group 2, member D; RT-PCR: Reverse transcriptase PCR; SOX: SRY-related HMG-box; SSX: Synovial sarcoma, X breakpoint; YKL: Tyr–Lys–Leu.

While the CNS is not strictly ‘immunologically privileged’, it is ‘immunologically quiescent’ with a low frequency of professional antigen-presenting cells (APCs) and lack of an organized lymphatic system [57–59]. In addition, GBMs are protected by the blood–brain barrier (BBB) [60]. However, the mere fact that GBMs have developed intricate immune evasion strategies strongly argues that the BBB does not present an insurmountable barrier to the cellular arm of the immune system.

Glioblastoma multiforme evade the immune responses by a variety of mechanisms and recent studies have highlighted the prominent role of glioma stem cells in creating the immunosuppressive microenvironment in gliomas [61–64]. Nevertheless, glioma stem cells can be recognized and killed by antigen-specific T cells [40,65]. While several intra-cellular signaling pathways are involved, persistent activation of the signal transducer and activator of transcription (STAT)3 most likely plays a major role, and the advent of STAT3 inhibitors offer the potential to pharmacologically reverse the ‘immunosuppressive glioma haven’ [41,62,66].

Glioblastoma multiforme:

• Release inhibitory cytokines;

• Interfere with the antigen-presentation pathway or mutate the antigen;

• Downregulate cell adhesion or costimulatory molecules, resulting in failure to activate specific immune responses either directly or indirectly;

• Induce inhibitory cells such as myeloid-derived suppressor cells, immunosuppressive microglia or regulatory T cells [67–71].

At present, the relative contribution of each of the aforementioned immune evasion mechanism is unknown and further studies are needed. Besides GBMs, many malignancies including Hodgkin’s and non-Hodgkin lymphoma have developed intricate immune evasion strategies [72,73]. T-cell therapies for these malignancies indicate that this hostile tumor microenvironment can be overcome, best exemplified by long-term remissions achieved with infused T cells, specific for the Epstein–Barr virus (EBV)-associated antigen, LMP2, which is expressed in EBV-positive lymphomas [74]. In addition, as discussed in the section ‘Enhancing effector T-cell function’ of this article, genetic engineering of T cells enables the generation of T cells that are resistant to the immunosuppressive glioma microenvironment.

Adoptive transfer of immune cells is one strategy to enhance anti-GBM immune responses. In preclinical or clinical studies, the use of leukocytes, NK cells, γδ T cells, activated T cells, tumor-infiltrating lymphocytes (TILs), antigen-specific T cells, or genetically modified T cells has been evaluated and we will review each of these cellular immunotherapy approaches in the following sections.

Leukocytes

Several investigators demonstrated in the 1970s that lymphocytic infiltrates could be detected in a subset of patients with malignant glioma, suggesting that the patients’ immune systems are able to recognize the tumors and mount a response, albeit an ineffective one [75,76]. Evidence suggested that the degree of lymphocytic infiltrate, particularly in the perivascular regions of tumors, was positively correlated with survival; however, this correlation was contentious [77,78]. Nevertheless, these findings provided rationale for early trials of adoptive cell therapy in which patients with recurrent GBM were given autologous leukocyte infusions intracranially [6–9]. Although these studies demonstrated the safety of infusing autologous leukocytes into the tumor resection cavity, the efficacy of using leukocytes was limited.

NK cells

The observation that stimulating peripheral blood lymphocytes for 3–4 days with IL-2 was sufficient for obtaining cells with significant antitumor activity prompted a number of groups to evaluate LAK cells as cellular immunotherapy [79]. LAK cells are a mixture of lymphokine-activated CD3⁺ T lymphocytes and CD3⁻/CD56⁺/CD16⁺ NK cells and exhibit ex vivo cytolytic activity against a broad range of solid tumors. More than 100 GBM or high-grade glioma patients have been treated with LAK cells by local injection [10–16]. Most of these trials were conducted in the 1980s and early 1990s in the pre-temozolomide era. While some of the clinical studies have shown promising results in prolonging disease-free survival, a randomized Phase II clinical study was never conducted. This has limited the enthusiasm to pursue these cells as immunotherapy for GBM, especially since for other malignancies the use of LAK cells in combination with IL-2 was not superior to the use of IL-2 alone [80]. However, owing to recent advances in the field of NK cell biology, there is renewed interest in NK cell-based immunotherapy for cancer [28].

Several strategies are being pursued to enhance the antitumor activity of NK cells. First, the use of artificial APCs expressing membrane bound IL-15 and 4-1BB ligand has allowed, for the first time, the generation of a highly cytotoxic NK-cell population with enhanced antitumor activity against malignancies [81]. Second, genetic modification of NK cells with chimeric antigen receptors (CARs), as described in the section ‘Antigen-specific T cells’ of this article, has shown promise in preclinical studies to enhance the effector function of NK cells [82,83]. For example, NK cells expressing CARs specific for CD19 have demonstrated enhanced anti-leukemia activity in preclinical models, and a Phase I clinical study with NK cells expressing CD19-specific CARs is in progress [83]. This approach could be readily adapted to GBMs since CARs specific for GBM-associated tumor antigens such as IL-13 receptor subunit α-2 (IL-13Rα2) and HER2 are available [32,37,40].

Other strategies to improve the efficacy of NK cell-based therapy are based on the observation that NK cells express activating receptors such as NKG2D as well as inhibitory receptors called killer-cell immunoglobulin-like receptors (KIR). Thus, NK cell activation by tumor cells depends on the balance of activating and inhibitory ligands on their cell surface. Several investigators have shown that epigenetic modifiers such as histone deacetylase inhibitors enhance the expression of activating NK cell ligands on tumor cells, resulting in enhanced NK cell-mediated killing [84]. Since inhibitory ligands are encoded by HLA-C molecules, another strategy to overcome the presence of inhibitory ligands is the use of haploidentical NK cells, which lack the corresponding KIR [85,86]. Indeed, the infusion of haploidentical NK cells is safe and has resulted in promising antitumor effects [87]. Since allogeneic T cells have been injected locally into GBMs with an encouraging safety profile, exploring the use of allogeneic, KIR-mismatched NK cells might also be feasible [88].

γδ T cells

γδ T cells are a subset of T lymphocytes, which express T-cell receptors (TCRs) that consist of one γ-chain and one δ-chain. Unlike conventional αβ T cells that recognize only specific peptide antigens presented in the context of a MHC molecule, γδ T cells recognize a broader range of antigens in a MHC-independent fashion. These antigens include MHC-like stress-induced self-antigens such as the NKG2D ligands, glycolipids presented by CD1c and phosphoantigens produced as a byproduct of bacterial metabolic pathways [89].

γδ T cells have been shown, in a number of preclinical studies, to have potent cytolytic activity against GBM cells [90]. In early studies, it was shown that γδ T cells could be effectively isolated and expanded from the blood of GBM patients by removing the CD4⁺, CD8⁺ and CD16⁺ fractions from peripheral blood mono-nuclear cells (PBMCs) and culturing the negative fraction with OKT3 and IL-2 [91]. These γδ T cells were able to lyse autologous GBM in cytotoxicity assays, and this activity was enhanced by the addition of IL-12 and IL-15 [92,93]. More recently, it has been shown that although the absolute count of γδ T cells decreases and their proliferative capacity is diminished in GBM patients, these γδ T cells can still be activated and expanded ex vivo and are cytotoxic against primary GBM tumors, while sparing normal astrocytes [94]. Finally, γδ T cells had antitumor activity in GBM xenograft models [95]. To date, no clinical experience with the adoptive transfer of γδ T cells is available. One of the major limitations in the past has been the inability to generate sufficient numbers of γδ T cells that retain their broad antitumor activity without becoming exhausted or anergic from overstimulation. However, recent studies indicate that these limitations can be overcome [95]. Phase I/II trials have demonstrated the feasibility of stimulating and expanding γδ T cells in vivo through the use of aminobisphosphonates in patients with non-Hodgkin’s lymphoma, multiple myeloma, and metastatic prostate cancer [96,97]. The efficiency of in vivo activation and expansion of γδ T cells correlated in general with antitumor activity. In vivo activation of γδ T cells has not been evaluated for high-grade glioma.

Activated T cells

Several clinical studies have been conducted with mitogen-activated T cells (mitogen-activated killer cells or autologous stimulated lymphocytes) [17–19]. As for other malignancies, these cells only had marginal antitumor activity in GBM patients. With the advent of better ex vivo stimulation techniques to overcome tumor-induced T-cell anergy using anti-CD3- and anti-CD28-coated beads, there is renewed interest in this approach. While initial studies have shown that the adoptive transfer of these cells is safe and reconstitutes cellular immune responses post-hematopoietic stem cell transplantation, antitumor responses have been limited [98–101]. One strategy to enhance the anti-tumor activity of adoptively transferred CD3/CD28-activated T cells is to vaccinate patients post-T-cell infusion with tumor antigens. In this regard, results of a single Phase I/II clinical study for patients with myeloma are encouraging [102]. At present, no clinical studies have been conducted with GBM patients.

Tumor-infiltrating lymphocytes

Tumor-infiltrating lymphocytes represent a source of effector T cells that presumably have already been selected for their ability to recognize and respond to the specific tumor antigens that are present in the tumor. In contrast to LAK cells, TILs recognize tumor cells in a MHC-restricted fashion [103]. While TILs may not possess sufficient antitumor activity in the highly immunosuppressive microenvironment established by tumors, activation and expansion of TILs ex vivo can overcome the tumor-induced immunosuppressive effects and allow for the generation of sufficient numbers of TILs for adoptive immunotherapy. Most clinical experience with TILs is available for melanoma. Although TILs by themselves had only limited antitumor activity, clinical studies have shown that high-dose chemotherapy and radiation, in combination with TIL transfer and IL-2, results in significant antimelanoma effects [104,105]. However, for malignancies other than melanoma, it has been very difficult to expand TILs from tumor tissues [106]. For GBM, only one study has been published. TILs were obtained from the resected tumors and expanded ex vivo in media containing IL-2. The patients received two injections of the cells 2 weeks apart with doses between 3 × 10⁸ and 1 × 10⁹ along with intratumoral IL-2 injected three-times a week [20]. Of the six patients, one had a complete response, two had partial responses and three died of progressive disease at long-term follow-up. In summary, while TIL therapy continues to be explored for melanoma, enthusiasm to pursue TILs for GBM, as for other malignancies, is limited.

Antigen-specific T cells

The ex vivo generation of clinical grade polyclonal tumor antigen-specific T cells for the adoptive immunotherapy of malignancies has proven difficult unless viral-associated tumor antigens are targeted. For example, the infusion of EBV-specific T cells in patients with EBV-positive malignancies has resulted in promising antitumor responses, not only for lymphomas but also for EBV-positive nasopharyngeal carcinoma [74,107–109]. In this regard, the recent demonstration of cytomegalovirus (CMV) antigens in GBM opens the opportunity to evaluate the safety and efficacy of CMV-specific T cells for patients with GBM since robust ex vivo expansion protocols are available for the generation of CMV-specific T cells [53,110,111]. Indeed, Phase I clinical studies with CMV-specific T cells for GBM patients have been initiated or are being developed.

While there is no clinical study published with the adoptive transfer of polyclonal, tumor-antigen-specific T cells, the infusion of antigen-specific T-cell clones has been reported. These studies indicate that T-cell clones can induce tumor regressions; however, antigen-loss variants rapidly emerge, highlighting the risk of tumor escape mutants with the use of T-cell clones [112]. A case report indicates that this problem can be potentially overcome if the initial T-cell-mediated tumor cell destruction results in the activation of endogenous immune responses against other tumor antigens (epitope spreading) [113].

Several reports indicate that is feasible to generate polyclonal, GBM-specific T cells by stimulating patients’ T cells with autologous GBMs in the presence of IL-2 [24,25]. In general, this approach is not very effective because tumor cells are poor APCs: they do not express costimulatory molecules and even inhibit T-cell responses by the secretion of inhibitory cytokines such as TGF-β. Despite these concerns, investigators have used this approach to generate GBM-specific T cells. The antitumor activity of these T cells was restricted to autologous tumors, suggesting that antigen recognition was HLA dependent; however, the recognized antigens have not yet been identified. Adoptive transfer of these cells resulted in transient antitumor responses, warranting further exploration of this approach [24,25].

Since initial priming and/or activation of polyclonal, tumor-antigen specific T cells is difficult ex vivo, one strategy to overcome these limitations is to prime/activate tumor-specific T cells in vivo and expand these cells ex vivo before reinfusion. This approach has been evaluated in three clinical trials. Activated T cells were either isolated from draining lymph nodes of the vaccine site or from PBMCs. Plautz et al. employed the first strategy in two Phase I studies for patients with recurrent or newly diagnosed high-grade gliomas [21,22]. For both studies, resected tumors were cultured short term and irradiated prior to intradermal injection with GM-CSF into the upper thigh. T cells were isolated from the draining inguinal lymph nodes a week after vaccination and expanded ex vivo using a combination of CD3 monoclonal antibodies, staphylococcal enterotoxin A and low-dose IL-2. T cells were administered intravenously, given previous experience with animal models demonstrating that T cells could be found infiltrating intracranial tumors following systemic injection. Of the ten patients with recurrent disease, two patients showed radiographic regression of at least 6 months, one patient had stable disease of over 17 months, and the rest had progressive disease. Of the 12 patients in the second study, four patients showed partial regression of residual tumor by MRI. The systemic infusion of these T cells was associated with only minor short-term side effects including fever, chills, nausea and myalgias. In the third study, Peres et al. vaccinated three patients with autologous tumor cells [23]. Postvaccination, patients’ PBMCs were collected, expanded ex vivo and reinfused into patients following a course of high-dose chemotherapy and peripheral blood stem-cell rescue. Two out of three patients had a clinical and radiographic response. Clearly, both of these approaches are promising; however, larger studies are needed to better evaluate the efficacy of this strategy.

Genetically modified T cells

Gene transfer allows the rapid generation of antigen-specific T cells for adoptive immunotherapy or T cells with enhanced effector function (Table 3) [114–135]. This approach can circumvent tolerance to the self-antigens expressed by tumor cells. Successful gene transfer strategies to generate antigen-specific T cells include the forced expression of α/β TCRs or antigen-specific CARs.

Table 3.

Genetic modifications to improve cell therapy

Goal	Genetic modification	Ref.
Tumor specificity	αβ TCR	[11–119]
	Chimeric antigen receptors	[120–125]
Cell survival and persistence	Cytokines or cytokine receptors	[126–128]
	Costimulatory molecules	[129]
	Antiapoptosis genes	[130]
Resist tumor-mediated inhibition	Dominant negative TGF-β receptor	[131,132]
	shRNA to knock down FAS	[133]
Integrating cell therapy with conventional therapies	Zinc-finger nuclease for glucocorticoid receptor	[134]
	Expression of chemoresistance genes	[135]

TCR: T-cell receptor.

α/β T-cell receptors

α/β TCR genes have been cloned for several HLA-restricted epitopes encoded by TAAs [115–119]. Genetic modification of T cells with α/β TCRs requires high expression and correct pairing of two different receptor molecules from a single vector, which has proved problematic for transgenic α/β TCRs. However, in the last 5 years, there has been significant progress in overcoming both of these limitations and three Phase I clinical studies with α/β TCR T cells for patients with refractory, metastatic melanoma have been completed [118,119]. A total of 34 patients were infused with T cells expressing a low-affinity MART1-specific α/β TCR T cells and objective clinical responses, including two complete responses, were observed. A total of 36 patients received T cells expressing high-affinity α/β TCR specific for either MART1 or gp100. Infusion of high-affinity α/β TCR T cells was associated with antitumor activity in nine patients including one complete response and eight partial responses. However, T cells also recognized normal tissues, which expressed low levels of the targeted TAAs. At present, no clinical study with α/β TCR T cells has been conducted in GBM patients.

Chimeric antigen receptors

Tumor-specific T cells can be generated by genetically modifying human T cells with tumor-specific CARs. CARs consist of an extracellular binding domain, a transmembrane domain and a cytoplasmic signaling domain [120–123]. The extracellular binding domain is most commonly derived from a single-chain variable fragment of a monoclonal antibody; however, ligands specific for antigens expressed on the cell surface of tumor cells have also been used. The cytoplasmic signaling domains are derived from the TCR ζ-chain and costimulatory molecules such as CD28, CD134 and CD137. T cells with CARs have numerous advantages over immunotherapies based on monoclonal antibodies or T cells alone. They can be directed toward any antigen that is expressed on the cell surface. Because CARs provide T-cell activation in a non-MHC-restricted manner, they are immune to some of the major mechanisms by which tumors avoid MHC-restricted T-cell recognition, such as downregulation of HLA class I molecules and defects in antigen processing. CAR-expressing T cells are more likely to eradicate tumor cells than antibodies alone, since they can migrate through microvascular walls, extravasate and penetrate the core of solid tumors to exert their cytolytic activity, sequentially kill a multiplicity of target cells and recruit additional components of the immune system, thus amplifying the antitumor or antiviral immune response.

Chimeric antigen receptors have been generated for the glioma-specific antigens, including IL-13Rα2, HER2 and EGFRvIII [32,37,40]. T cells expressing these GBM-specific CARs had potent antitumor activity in preclinical animal models. In addition, the study with T cells expressing HER2-specific CARs showed that these cells had potent antitumor activity against HER2-positive, CD133-positive glioma stem cells, which are chemo- and radiotherapy resistant [40]. A Phase I clinical study with T cells expressing IL-13Rα2-specific CAR is ongoing and a study with T cells expressing HER2-specific CARs is in the development phase [136].

For other diseases, initial clinical studies with T cells expressing a CAR with only a TCR ζ-chain displayed limited clinical benefits [137,138]. The most pertinent issue being that CAR T cells failed to expand and rapidly lost their function in vivo. Several approaches have been pursued to overcome the limitation of CAR T cells. Incorporation of additional signaling domains from the costimulatory molecules CD28, CD134 and CD137 into CARs, as well as the coexpression of cytokines or their receptors, have resulted in enhanced effector function of CAR T cells [82,139,140]. A more pragmatic approach to overcome the signaling defect of CARs with a single ζ-signaling domain might be to express the receptors in antigen-specific T cells, which can be activated and expanded through their endogenous TCR. This concept of bispecific T cells was validated in an animal model with alloreactive T cells expressing CARs. Clinical relevant examples of this strategy include the expression of CARs in EBV-, influenza- or varicella zoster virus-specific T cells [123,141–143]. In a Phase I clinical study, EBV-specific T cells, expressing a GD2-ζ CAR, persisted significantly longer than autologous GD2-ζ T cells. In addition, the infusion of GD2-specific T cells resulted in tumor necrosis or regression (including a complete remission) in four out of eight patients with refractory/relapsed neuroblastoma [123]. Since a subset of GBMs is positive for CMV antigens, CMV-specific T cells might be an ideal ‘T-cell platform’ for GBM-specific CARs [144].

Enhancing effector T-cell function

Genetic modification not only allows for the generation of GBM-specific T cells, but also opens the opportunity to improve the effector function of T cells (Table 3) [126–135]. For example, preclinical models have shown that:

• Expression of a dominant negative TGF-β receptor renders T cells resistant to the inhibitory effects of TGF-β [131,132];

• Transgenic expression of cytokines improves T-cell persistence and function [126–128];

• Expression of chemokine receptors enhances T-cell migration [145,146];

• Zinc-finger nuclease-mediated disruption of the glucocorticoid receptor gene locus renders T cells resistant to the inhibitory effects of steroids [134].

Clearly, several of these strategies are directly applicable to cell therapies for GBMs and warrant further exploration.

Combinatorial cell therapy

There is increasing evidence that the efficacy of cell therapies can be enhanced when combined with other therapies. First, the use of lymphodepleting chemotherapy and/or radiation prior to cell transfer has resulted in enhanced in vivo expansion and antitumor effects in clinical studies [105]. Other strategies to enhance the in vivo expansion of T cells include lymphodepleting monoclonal antibodies as well as vaccines [147–149]. Second, investigators have shown in a preclinical melanoma model that VEGF antibodies enhance trafficking of adoptively transferred T cells to tumor sites resulting in enhanced antitumor effects [150]. This approach is readily applicable to GBMs, since the VEGF antibody Avastin was approved for the treatment of GBM in 2009. Third, agents that either increase the expression of tumor antigens or reverse the inhibitory microenvironment have the potential to enhance cell therapies for GBM patients in the clinic. For example:

• Combining the adoptive transfer of T cells with HDAC inhibitors, which increase MHC and tumor antigen expression in tumor cells, has resulted in enhanced antitumor effects in a preclinical melanoma model [151];

• Blocking STAT3 in combination with the adoptive transfer of T cells resulted in enhanced antitumor effects [41,152].

In summary, most combinatorial cell therapies have not been evaluated in clinical studies, however, the promising results obtained in pre-clinical models warrant further exploration of these approaches.

Conclusion & future perspective

Cell therapies for GBM have now been evaluated for more than four decades and most published clinical studies have used nonspecific cell products. While antitumor effects of adoptively transferred cells were observed in several studies, these findings were never confirmed in randomized clinical trials. This lack of definitive Phase II studies for cell therapies is not limited to GBM-directed therapies, but is a prevailing problem for the entire field of cell therapy. Currently, several ‘definitive’ clinical cell therapy studies are being developed for other malignancies, and GBM-directed cell therapies will hopefully follow suit. Improved methods to generate GBM-specific T cells, either using APCs or genetic approaches, and the ability to engineer T cells that are resistant to the inhibitory GBM microenvironment, offer exciting new approaches, which need to be tested in carefully designed clinical studies. Finally, combining cell therapies with other therapies holds the promise to enhance their efficacy. Paul Bucy, the former editor of Surgical Neurology, put out the following challenge in a commentary in 1973: “The field [brain tumor and immunity] is rapidly developing and we are in need of a better understanding of it” [153]. In the last 37 years, we have made significant progress in understanding the complex relationship between brain tumors and the immune system. The challenge now is how to translate these findings into immunotherapies that will benefit GBM patients. We believe that cell therapies will play an important role in this endeavor.

Executive summary.

• The outcome for patients with glioblastoma multiforme (GBM) remains poor.

• Immunotherapy with nonspecific cell products has been evaluated in a series of small clinical studies with some clinical benefit; however, these findings were never confirmed in a randomized clinical study.

• Initial clinical studies with the infusion of in vivo- or ex vivo-activated GBM-specific T cells has shown encouraging results; confirmatory studies are needed.

• Recent advances in the ex vivo generation of NK cells, γδ T cells and GBM-specific T cells make clinical studies of these cell products in GBM patients feasible.

• Genetic modification holds the promise to generate T cells that are not only GBM-specific but also have enhanced effector function.

• Combining cell therapies with other therapies has the potential to improve their antitumor activity.

• Phase II clinical studies are needed to define the role of cell therapies.