Immunotherapy against gliomas- is the breakthrough near?

Abstract

Immunotherapeutic approaches have been and continue to be aggressively investigated in the treatment of infiltrating gliomas. While the results of late phase clinical studies have been disappointing in this disease space thus far, the success of immunotherapies in other malignancies as well as the incremental gains in our understanding of immune-tumor interactions in gliomas has fueled a strong continued interest of their evaluation in these tumors. We discuss a range of immunotherapeutic approaches including, but not limited to, vaccines, checkpoint inhibitors, oncolytic viruses, and gene therapies. Potential biomarkers under investigation to help elucidate which patients may respond or not respond to immunotherapeutic regimens are reviewed. Directions for future investigations are also noted.

1. INTRODUCTION

The interplay between glioma and the immune system is complex due to several factors which contribute to the significantly immunosuppressive tumor microenvironment. The ability of the immune system to rid both the central nervous system (CNS) space and the extra-CNS space of malignancy, under specific circumstances, has led to the generation of tremendous interest in utilizing the immune system to control and potentially eradicate glioma. While clinical expectations should be tempered by the reality that immunotherapy has yet to demonstrate an objective benefit among all phase III clinical trials to-date, hope remains for this treatment approach to provide a survival benefit to patients with glioma in the future.

2. IMPACT OF THE STANDARD OF CARE ON IMMUNOTHERAPEUTIC APPROACHES

2.1.

The current standard of care (SOC) for infiltrating glioma often consists of surgical resection, radiation therapy (RT), and chemotherapy.[1,2] Patients with glioblastoma (GBM) may also be treated with tumor-treating electrical fields (TTF) delivered by arrays directly applied to the scalp.[1–3] It is not currently understood whether SOC treatments affect immunotherapeutic efficacy in a favorable or unfavorable manner.(Table 1) Therefore, a need exists to further develop a better understanding of how they can and should be applied in the age of immunotherapeutic investigation.

Table 1.

The Standard of Care Treatment of Glioblastoma and It’s Potential Impact on Immunotherapies.

Standard of Care	Potential Detrimental Factors	Potential Beneficial Factors
Surgery	• Limited antigenic tumor epitopes after extensive resection	• Surgically mediated cytoreduction leads to a smaller tumor burden for the immune system to address • May provide direct immune stimulation • May serve as a means of delivering local immunotherapies
Radiation	• May increase the likelihood of lymphopenia associated with temozolomide • May select for more aggressive tumor cell clones • May lead to increased tumor PD-L1 expression, IDO expression, and immunosuppressive macrophage infiltration	• May increase mutations within the tumor leading to the expression of more antigenic tumor epitopes • Radiation mediated cytoreduction leads to a smaller tumor burden for the immune system to address • May stimulate effector T cell infiltration into tumor
Temozolomide	• Contribute to leukopenia and particularly lymphopenia • May increase mutations within the tumor increasing its malignancy • May select for more aggressive tumor cell clones	• May cause an imbalanced reduction in T cell subtypes leading to a greater proportion of effector T cells
Tumor Treating Fields	• May increase tumoral PD-L1 expression	• May increase immune cell infiltration into tumors

2.2. Potential Detrimental factors

Numerous aspects of SOC treatment may impede the effectiveness of immunotherapy. Alkylating chemotherapies such as temozolomide (TMZ) are typically utilized in the treatment of infiltrating gliomas with concomitant radiation and in the adjuvant setting.[1,2] These chemotherapies are myelosuppressive and are associated with leukopenia and in particular, lymphopenia.[1] The myelosuppression is exacerbated by the pre-existing low peripheral lymphocyte counts noted in 25–40% of glioma patients prior to the initiation of treatment.[4,5] Approximately 18% of treatment-naïve steroid-naïve GBM patients exhibit lymphopenia. The incidence increases greater than 2-fold when steroid-exposed patients are included. The enhancement of lymphopenia can be pronounced with 15% of GBM patients presenting with ≤200 cells/μL, which is roughly equivalent to an AIDS-defining lymphocyte count in HIV⁺ patients.[5] This process is thought to be driven, at least in part, by the central nervous system tumor. While the full mechanism governing this effect is under investigation, it has been shown that the loss of sphingosine 1 phosphate 1 (S1P1) on the surface of naïve T cells leads to their sequestration in the bone marrow in models and patients with malignant glioma.[5] The concurrent use of radiation and TMZ appears to be associated with a higher incidence of lymphopenia as compared to adjuvant TMZ alone. It’s also possible that this could undermine the immunostimulatory benefit of RT in promoting the conversion of tumors with immunologically “cold” microenvironments into a “hot” phenotype. Risk is highest in patients who are older and female, suggesting an underlying biologic driver for establishing the effect.[6,7] The size of the radiation field, a potentially modifiable risk factor, also appears related to the risk of severe acute lymphopenia.[6] Treatment induced lymphopenia can persist long-term with absolute lymphocyte counts remaining low for around one year [8], which can have knock-on effects that negatively impact the efficacy of subsequent therapeutic approach considerations. Treatment related lymphopenia has also been associated with a negative impact on survival, with a notable effect in the elderly population.[9,10] In a subset of patients with unmethylated MGMT promoter GBM who are less likely to benefit from the treatment with TMZ[11], omission of TMZ from the SOC can be considered and may even enhance the efficacy of immunotherapeutically-geared clinical trials.

TMZ-induced lymphopenia may reflect an unbalanced decrease in effector T cells as compared with immunosuppressive Tregs. This has been most extensively studied in the context of GBM expressing the tumor-specific epidermal growth factor (EGFR) variant 3 (EGFRvIII) mutation. However, this may also be related to the overall dose of chemotherapy. In patients with EGFRvIII glioblastoma treated with concomitant TMZ, granulocyte-macrophage colony-stimulating factor (GM-CSF), and rindopepimut, an EGFRvIII peptide-based vaccine, increased Tregs was noted in patients receiving a dose-intense (100 mg/m2 for 21/28 days) TMZ regimen but not in the standard dosing regimen. However, the dose-intense population also experienced higher EGFRvIII-specific antibody titers as well as more robust delayed type hypersensitivity reactions. It’s hypothesized that these findings may relate to the specific timing of vaccination in relation to the fluctuations of homeostatic cytokines secreted in response to chemotherapy-induced lymphopenia.[12]

After gross total resection of enhancing tumor or extensive subtotal resection, there is a theoretical concern that there are a limited amount of remaining GBM cells expressing antigenic epitopes that are required to stimulate optimal immune activity. In addition, both primed and more quiescent tumor infiltrating lymphocytes (TILs) are removed during GBM resection. This hypothesis is supported by analyses of trials during treatment with immunotherapy for patients with glioma. These include post hoc analyses of the phase III trial of the EGFRvIII targeting vaccine rindopepimut for newly diagnosed GBM whereby patients with a greater volume of residual tumor trended toward possessing better outcomes in the investigational arm [13]. Similarly, 2 recent studies reported better clinical and immunological responses for GBM patients treated with single agent anti-programmed death 1 (PD-1) blockade in the neo-adjuvant setting as compared to post-resection resection adjuvant setting.[14, 15]

SOC may increase the mutational burden in GBM over time potentially promoting a more malignant phenotype. Low grade astrocytoma treated with cytotoxic therapies such as TMZ increase the number of mutations when re-biopsied or re-resected at time of progression. This was more frequent in patients with methyl guanine methyltransferase (MGMT) promoter methylation.[16] In patients with treatment-dependent hypermutation, mutation of the mismatch repair (MMR) pathway gene family was likely.[17] This could facilitate development of additional mutations with subsequent lines of DNA damaging treatments.

SOC may select for the most aggressive tumor cell clones to dominate the remaining tumor after less resistant clones are selected against. Stem-like tumor cells would likely be an exception to this rule due to their immunologically silent nature. Stem-like tumor cells have decreased expression of antigen presenting molecules including MHC-I, MHC-II, and NKG2D.[18,19] Therefore, while facile eradication of a substantial tumor cell population may be transiently beneficial, the small subset of resistant tumor cells with regenerative potential will likely circumvent curative efforts.

The peri-operative use of dexamethasone during the course of immunotherapy can have a negative impact on immune responsiveness to immunotherapy. The negative association between steroid utilization and outcomes has been described across aa range of preclinical and settings.[20,22] This is a modifiable intervention, as compared with some patients that are symptomatic due to mass effect that benefit from steroids, the administration of dexamethasone at time of anesthesia during the beginning of surgery is common practice, even for asymptomatic patients.

2.3. Potential Favorable factors

While circulating lymphocyte levels are frequently decreased, neutrophil counts are also suppressed in a subset of GBM patients. Somewhat unexpectedly, radio-chemotherapy-induced decreases in peripheral neutrophil counts has been associated with improved overall survival (OS) in patients with isocitrate dehydrogenase (IDH) wild type (wt) GBM.[23] which is inverse of the relationship between OS and absolute lymphocyte counts. This phenomenon is a reflection of the well-described negative prognostic increased neutrophil to decreased lymphocyte ratio, which is associated with select types of inflammatory responses.[24] These observations lack a clear mechanistic understanding of how this process would drive favorable outcomes.

RT and chemotherapy have both immunostimulatory and immunosuppressive effects. Clinical trial evaluation of immunotherapies in cancer patients have previously demonstrated that the combination of a few moderately sized RT doses may be synergistic with immune checkpoint inhibitors across a wide range of advanced cancers.[25] It is thought that the synergy may arise via complementary mechanisms. First, RT can lead to rapid cytoreduction while also serving as a means for non-surgical tumor debulking.[26] Also, RT promotes the increase of effector T cells into tumors. The RT-induced tumor cell death can also enhance the exposure of a larger and broader repertoire of potentially immunostimulatory antigens, which can be observed in preclinical models of glioma when co-neutralizing non-redundant immunosuppressive pathways.[27] In preclinical models, treatment with TMZ increases major histocompatibility complex 1 (MHC-I) expression on glioma stem cells through a nuclear factor of kappa-light chain-enhancer of activated B cells (NF-κB)-dependent mechanism.[28] In other malignancies, there is clear synergy when traditional cytotoxic chemotherapies are combined with immunotherapies.[29] This, as stated earlier in relation to EGFRvIII vaccine, may be due to a disproportionate toxicity to specific subsets of immune cells (including those which contribute to immunosuppression). The dosing of cytotoxic chemotherapy, routinely below a myeloablative concentration, and its impact on the tempo of recovery of various immune cell lineages requires further understanding. It is possible that dosing of cytotoxic chemotherapies when utilized in conjunction with immunotherapies may require recalibration from standard dosing schedules.

The effect of tumor treating fields (TTFields) on immune system-tumor interactions is not well understood and has not been systematically studied in the clinical setting. A preclinical study using mice and rabbits implanted with melanoma and VX-2 cells showed an increase in the intratumoral immune cell infiltration with TTFields. In addition, local delivery of TTFields decreased the risk of distant metastases and improved survival of these animals.[30] Another preclinical Lewis lung carcinoma murine model suggested an additive benefit when TTFields and anti-PD-1 mAb are combined. Combined modality treatment is associated with increased tumoral programmed death ligand 1 (PD-L1) expression and a higher number of tumor infiltrating immune cells.[31] It will be of interest to learn whether TTFields synergize with various immunotherapeutic approaches in gliomas.

3. AVENUES UNDER INVESTIGATION

3.1. Vaccines

A range of therapeutic modalities have been and are currently being investigated for potential patient benefit in the treatment of glioma.(Table 2) The modality that has been most thoroughly explored is vaccine therapy.[32,33] Although vaccination against glioma is a derivative of the canonical meaning of the term which is normally associated with, ‘preparation used as a preventive inoculation to confer immunity against a specific disease’. In patients with malignant glioma, one cannot truly vaccinate, since the disease is already present and cannot therefore be prevented. However, it is possible to ‘stimulate’ an immune response against antigens expressed by malignant glioma using a variety of “personalized-medicine” and “off the shelf” vaccines. Of personalized medicine approaches, vaccines can be divided into autologous patient-derived treatments and regimens designed to specifically address a panoply of epitopes within the patients’ tumor. Amongst the patient specific vaccines, the two which progressed furthest in development were the heat-shock protein peptide complex 96 (HSPPC96) () and dendridic cell vaccine (DCvax) (). The HSPPC96 patient derived vaccine demonstrated promise in a very select population of patients evaluated in the non-randomized phase 2 trial.[34] Unfortunately, the randomized phase 2 trial for recurrent glioblastoma failed to pass the interim futility analysis.[35] Antigen presenting dendritic cells pulsed with autologous tumor lysate form the basis of another patient-specific vaccine approach. The DCvax randomized phase 3 trial completed accrual and the preliminary results have been published. The treatment appears to be well tolerated. Interpretation of the current results is complicated somewhat by the substantial amount of crossover (90%) from the control arm to the investigational treatment.[37,38] The mature results are pending. Another patient-specific approach being pursued by numerous groups centers on the ex vivo development of a vaccine based approach that predicts neoantigens present in a patient’s tumor.[38,39] This type of approach is undergoing early phase clinical evaluation (, , , ).

Table 2.

Summary of vaccine and immune checkpoint trials highlighted in manuscript.

NCT number	Phase	Patients	Treatment	OS	PFS	Reference
			Vaccine
	3	Newly Dx GBM	RT/TMZ/rindopepimut RT/TMZ	20.1 mo 20.0 mo	7.1 5.6	13
	1	DIPG	H3K27M	NA	NA	NA
	1	IDH1 mutated recurrent grade 2 glioma	PEPIDH1M/tetanus-diphtheria toxoid/TMZ	NA	NA	NA
	1	Newly Dx IDHR132H mutated grade 3 and 4 astrocytoma	IDH vaccine/topical imiquimod+/−TMZ+/−RT	NA	NA	NA
	2	Recurrent GBM	HSPPC96/bevacizumab bevacizumab	7.5 mo 10.7 mo	NA	35
	2	Recurrent GBM	HSPPC96	42.6 wks	19.1 wks	34
	3	Newly Dx GBM	DCVax	23.1 mo*	NA	36,37
	1	Newly Dx MGMT unmethylated GBM	RT/NeoVax/pembrolizumab	NA	NA	39
	2	Recurrent GBM	bevacizumab/rindopepimut bevacizumab	NA	NA	40
	1	Newly Dx GBM	APVAC1/APVAC2/polyICLC/GM-CSF/TMZ	NA	NA	41
	1	Newly Dx MGMT unmethylated GBM	RT/GNOS-PV01/INO-9012/nivolumab+/−ipilimumab	NA	NA	42
	1	GBM or DIPG	cyclophosphamide/bevacizumab/DC vaccine	NA	NA	NA
	1	DIPG	H3K27M	NA	NA	NA
	1	IDH1 mutated recurrent grade 2 glioma	PEPIDH1M/tetanus-diphtheria toxoid/TMZ	NA	NA	NA
	1	Newly Dx IDHR132H mutated grade 3 and 4 astrocytoma	IDH vaccine/topical imiquimod+/−TMZ+/−RT	NA	NA	NA
			Immune Checkpoint Inhibitor
	1	Recurrent GBM	Neoadjuvant pembrolizumab	417 days 228.5 days	99.5 days 72.5 days	14
	2	Recurrent GBM	Neoadjuvant nivolumab	7.3 mo	4.1 mo	15
	3	Recurrent GBM	nivolumab bevacizumab	9.8 mo 10.0 mo	1.5 mo 3.5 mo	44
	3	Newly Dx MGMT unmethylated GBM	nivolumab	NA	NA	45
	3	Newly Dx MGMT methylated GBM	nivolumab	NA	NA	46
	1	Recurrent GBM	atezolizumab	4.2	1.2	48

OS, overall survival; PFS, progression free survival; Dx, diagnosed; GBM, glioblastoma; RT, radiotherapy; TMZ, temozolomide; GM-CSF, granulocyte-macrophage colony-stimulating factor; DIPG, diffuse intrinsic pontine glioma; IDH1, isocitrate dehydrogenase 1;APVAC1, personalized vaccine; APVAC2, personalized vaccine; PEPIDH1M, IDH1 targeting vaccine; HSPPC96, heat shock peptide protein complex 96; DCvax, dendritic cell vaccine; NeoVax, personalized vaccine; GNOS-PV01, personalized vaccine; INO-9012, liposomal IL-12; mo, months; wks, weeks; GPVAC, glioma actively personalized vaccine consortium; NA, not available; MGMT, methyl guanine methyl transferase; DC vaccine, dendritic cell vaccine.

^*for the overall study population (combining the investigational arm and the standard of care arm)

In contrast to the patient-tailored vaccines, the more easily scalable “off the shelf” vaccines represent another approach. The “off the shelf” vaccine that has been most thoroughly evaluated is rindopepimut, the EGFRvIII-targeting vaccine. This study did not demonstrate improved survival with the addition of the vaccine to SOC in patients with newly diagnosed glioblastoma. Interestingly and as noted earlier, there was a trend toward improved survival in patients with substantial residual disease post-operatively [13]. In a smaller randomized phase II trial () in recurrent GBM improved survival was noted when used in conjunction with bevacizumab[40]. These findings require further validation. A variety of other vaccine-based approaches are currently undergoing study. These include the targeting of other tumor specific or tumor enriched epitopes, targeting multiple epitopes, selecting patient-specific epitopes, and combining vaccine-based approaches with other immunotherapeutic and non-immunotherapeutic approaches. Recent clinical vaccine studies of particular note include the Glioma Actively Personalized Vaccine Consortium (GPVAC) phase 1 trial combining a premanufactured library of vaccines followed by personalized vaccines targeting neo-epitopes.[41] This type of approach limits potential delays in initiating treatment. Another approach in the pediatric setting utilizes a vaccine that stimulates immunity against the H3K27M mutation frequently present in children with midline gliomas (). This trial combines the vaccine therapy with the immunostimulant, poly-ICLC. Finally, vaccine strategies (, [42]) targeting another mutation, IDH1, found in a subset of more treatment responsive gliomas, are undergoing investigation.[43] Clinical results have not been published thus far.

3.2. Immune checkpoint inhibitors

Immune checkpoint inhibitors (ICI) are under intensive investigation in the setting of high-grade glioma. These agents are most frequently antibodies that bind to and block immunosuppressive checkpoints that promote resolution of the immune response. The interest in and excitement for these agents is driven, at least in part, by the durable responses observed in other non-CNS cancers. Among the multiple ICIs investigated to-date, programmed death 1 (PD1) antibodies (Ab) have been most extensively evaluated. In the setting of recurrent glioblastoma, the utilization of nivolumab in a non-biomarker selected population was not associated with any improvement in survival when studied in a phase 3 clinical trial ().[44]. There were, however, 8% of confirmed radiographic responses in the nivolumab arm. More importantly, the median duration of responses was > 10 months, suggesting a durable benefit in a subset of patients. In the newly diagnosed MGMT promoter unmethylated GBM population, RT and nivolumab did not meet the primary endpoint of OS in the phase 3 CHECKMATE 498 trial ().[45] Full study results are eagerly awaited, as are the results from the companion phase 3 CHECKMATE 548 trial () in patients with newly diagnosed MGMT promoter methylated GBM. Preliminary results from CHECKMATE 548 note no significant improvement in progression free survival. OS results are still maturing.[46]

Despite these notable disappointments, it is possible that ICI may still have an important role in the future treatment of high-grade glioma. When initiated 14 days pre-operatively and continued post-operatively in patients with recurrent GBM, the anti-PD1 mAb, pembrolizumab, demonstrated improved OS (HR=0.39, p=0.04) in a small (n=35) randomized trial. Interestingly, there was some improvement in progression free survival (PFS) (HR=0.43, P=0.03) as well.[14] This is somewhat surprising as there has been notable difficulty in fully understanding the radiographic imaging results of immunotherapeutic approaches in patients with primary brain cancer. A high incidence of inflammatory pseudoprogression is suspected but has not been confirmed. These concerns have led to development of criteria to help guide scan interpretation.[47] Even with the development of radiographic criteria an understanding exists in the neuro-oncology field that progression free survival is an artificial construct that does not well represent what is occurring at the cellular level, below the resolution of contemporary standard imaging modalities.

Findings suggestive of negative feedback loops related to the treatment with ICI, such as increased intratumoral PD-L1 expression will need to be carefully considered as future studies are designed. These include transcriptional increases in interferon (IFN)-γ related genes, as well as an increase in RNA sequencing data demonstrating IFN-γ and T cell pathway induction. There was no clear impact of baseline steroid dosing on these findings. [14] Yet, there is theoretical concern regarding the use of steroids. The negative CHECKMATE-143 study included 40% of patients receiving anti-PD1 mAb blockade receiving concomitant dexamethasone, possibly having an effect on the 8% response rate reported.[44]

Other ICI strategies are in earlier phase clinical trials for treatment of patients with glioma. PD-L1, the ligand for PD-1, also holds substantial interest for treating this class of tumor. In a phase 1 trial in recurrent GBM, the single agent use of the anti-PD-L1 mAb, atezolizumab, was associated with a subset of patients with partial responses and prolonged survival. IDH mutated glioblastoma, with their better prognosis, were not surprisingly over-represented amongst the long-term survivors. There was, however, a long term survivor with an inactivating mutation in the DNA-repair enzyme, POLE. This patient’s tumor was also found to have a high tumor mutational burden (TMB).[48] Similar findings have been noted in a case report of another patient with a germline POLE mutation and high TMB.[49] Germline mutations of DNA-repair genes would presumably confer a higher likelihood of an elevated TMB throughout the tumor, as opposed to somatic mutations in these same genes which may lead to increased mutations only in the subclones of tumor which carry the DNA repair deficit.

3.3. Oncolytic viral therapy

Oncolytic viruses are engineered to infect and kill tumor cells. While initial efforts investigating oncolytic viral therapies in glioma have focused on their direct cytotoxic effects, it has also become clear that they may be serving to help initiate and augment an immune response against the tumor.[50,51] The performance of two complimentary functions in a single therapeutic modality makes them particularly attractive. A number of different viruses are being investigated for the treatment of glioma. These include modified adenovirus, herpes virus, measles virus, polio-rhinovirus chimera (PVSRIPO), parvovirus, reovirus, and zika virus. Initial studies have broadly demonstrated the relative safety of such approaches. Some long term sustained responses have been observed to-date. Phase I trial evaluation of oncolytic poliovirus found some responses when the alkylating chemotherapy, CCNU, was initiated after presumed progression on study.[52] These responses are particularly remarkable as the anti-tumoral activity of CCNU in progressive GBM is relatively low. PVSRIPO is being evaluated as a monotherapy () and in combination with atezolizumab () in the setting of two phase 2 trials. Other groups are also evaluating the combination of treatment with an oncoloytic virus and ICI. The oncolytic adenovirus, DNX-2401 (delta-24-RGD), has demonstrated safety as well as a pronounced and sustained response in a subset of patients with recurrent high-grade glioma.[53] This agent is now being investigated in conjunction with the anti-PD-1 mAb antibody, pembrolizumab (). A novel approach to decrease degradation of the number of viral particles and to increase their delivery is to encase them in neural stem cells. These cells are able to traverse long distances and traffic to areas of injury, such as tumor, in the brain. This approach is currently under investigation in a phase 1 trial ()[54].

3.4. CAR-T cells

Chimeric antigen receptor T cells (CAR-T cells), a type of genetically engineered T cell, have proven to be effective in treating select types of lymphoma and acute leukemia. The initial steps in their investigation for glioma was recently reviewed in detail.[55,56] CAR-T cells are engineered to engage cells expressing tumor-specific epitopes, subsequently leading to T cell-mediated tumor cell death. The exquisite specificity of the CAR-T cell tumor interactions is both weakness and strength. While it allows for a highly selective form of immune cell-target cell interaction, it also presents the same problems of epitope escape found in vaccine-based approaches. O’Rourke et al., reported the disappearance of EGFRvIII expression in GBM that recurred following adoptive transfer of EGFRVIII-targeted CAR-T cells.[57] Some investigators propose to circumvent this limitation with the development of multi-valent CAR-T cells.[58,59] For this type of approach to be truly effective a broad enough epitope targeting repertoire utilized early enough in the disease course prior to proliferation of resistant clones would be required. In addition the approach would necessitate the collateral eradication of tumor clones not harboring the predetermined epitopes engineered into the CAR-T cells.

A number of groups are actively investigating the role of CAR-T cells in high grade gliomas. The CAR-T cells currently in clinical trials have been engineered to target the EGFRvIII mutation (, , , , ), HER2 (FRP5 exodomain) (), and IL13Ralpha2 (E13Y epitope) (). Each approach is also utilizing distinct co-stimulatory domains (B-1BBz, CD28z, 4–1BBz) to activate the T cells. An extensive sustained radiographic response has been described in a case study of a patient with recurrent GBM after intracranial administration of IL13Rα2 targeting CAR-T cells.[60] This type of response was not recapitulated in the phase 1 trial of EGFRvIII targeting CAR-T cells in patients with EGFRvIII-expressing (from initial diagnosis, but not at the time of recurrence) GBM.[61] Loss of GBM EGFRvIII expression through disease progression [62] is one potential explanation for the lack of efficacy. Successful eradication of the EGFRvIII containing subclone and subsequent emergence of non-EGFRvIII subclones may be another more hopeful explanation.

3.5. Cytokine therapy

Cytokines are endogenous signals which influence cellular activity. Pro-inflammatory cytokines lead to stimulation of the immune system in numerous ways. Within the tumor microenvironment they may serve to convert an immunologically “cold” tumor to one which is “hot”. The utilization of pro-inflammatory cytokines has a longstanding history in the management of cancer. [63] This type of approach has been curative in a small subset of patients with non-CNS cancers, although the pronounced toxicity has limited the scalability of this treatment approach.[64,65] One strategy for circumventing the systemic toxicities of cytokine based therapies is the utilization of intratumoral administration of a viral vector for tumor cell vector production of the choice cytokine.[66] This allows for the controlled intratumoral/intracranial production of immunomodulatory cytokines, which was previously investigated with interferon-beta [67] and currently under investigation (, , ) with an IL-12 expression vector via the Ad-RTS-hIL12 modified adenovirus and oral activating ligand, veledimex.[22] As with other approaches, this is also under investigation as part of a combinatorial regimen in conjunction with systemic anti-PD-1 mAb treatment (, ). While early results appear promising [68], this approach awaits further validation.

3.6. Cytotoxic gene therapy

Late phase studies are also utilizing a viral vector (vocimagene amiretrorepvec), but for a different purpose. In this case it is for delivery of a gene encoding the pro-drug extended-release 5-fluorocytosine (Toca FC) which is converted locally in the CNS to the active cytotoxic 5 fluorouracil (5FU). Depletion of immunosuppressive myeloid cells and induction of anti-tumor immunity has been noted in the earlier phase studies.[69] This echoes earlier discussions regarding the effects of cytotoxic agents on the immune system, but limits their activity to the relevant compartment of the CNS. Disappointingly the preliminary results of the phase 3 trial evaluating this approach in recurrent glioblastoma were negative. Full results are awaited [70].

4. SEARCHING FOR BIOMARKERS

While predictive biomarkers associated with response to immunotherapy have been well delineated in some malignancies, this is not the case with respect to glioma. Expression of PD-1 and PD-L1 within the tumor (tumor cells, immune cells) is prognostic of a poor outcome in both newly-diagnosed and recurrent IDHwt glioblastoma.[71,72], although neither PD-1 nor PD-L1 expression have been definitively shown to be predictive of response to immunotherapies in gliomas. A small case series in the pediatric glioma population is suggestive of higher PD-1 expression being associated with improved survival in patients treated with an anti-PD-1 mAb.[73] As noted earlier, increased TMB appears to be associated with a prolonged survival in patients treated with ICI,[48,49,74] however, in cohorts of patients with recurrent GBM, overall mutational burden was not associated with a response to immunotherapeutic treatment [75]. This may reflect that within a single tumor only some subclones of tumor cells are hypermutated while others harbor a more moderate TMB. The response noted for the rare patient with hypermutant GBM, it is likely that mutational burden in GBM is not predictive in general given as compared with the cancers with an overall high TMB and where this association has proven to be stronger. Our group recently reported that PTEN mutations, which have long been known to be associated with tumor microenvironment immunosuppression,[76] are associated with lack of benefit to anti-PD-1 blockade,[75] while conversely, enrichment of MAPK pathway aberrancies is associated with a sustained response.[75] These findings, however, require validation in a large, prospective trial evaluating PD-1 pathway blockade.

5. FUTURE DIRECTIONS

Combinatorial approaches with multiple immunotherapies, cytotoxic chemotherapies, anti-angiogenics, targeted therapies, RT, and TTFields are all undergoing investigation. These investigations are at various stages of development. Approaches with dual checkpoint blockade (anti-CTLA4 mAb + anti-PD-1 mAb), for example, have proven to be superior to ICI monotherapy in other malignancies such as melanoma. As agents are combined, there is always the potential for increased overall toxicity. Dual checkpoint blockade, as well as the combination of checkpoint inhibitors with vaccine, immunomodulators, oncolytic virus, and stimulatory cytokines is being utilized. It is unclear which, if any, of these approaches will prove beneficial in gliomas. The anti-angiogenic agent, bevacizumab, an Ab targeting vascular endothelial growth factor (VEGF), has been utilized in the management of recurrent GBM for over a decade.[77,78] Bevacizumab is able to decrease the need for the utilization of steroids for tumor related cerebral edema.[79] This may be of importance as steroid use is thought to be associated with decreased survival [80] and appears to abrogate the efficacy of immunotherapies.[81] In turn, the combination of immunotherapies with bevacizumab is rational. In some study designs a few low doses of bevacizumab can be administered as steroid sparing agents. This provides greater flexibility with respect to on-study treatments and better reflects contemporary clinical practice. The use of RT as a means of rapid cytoreduction and immune stimulation makes its combination with immunotherapies in glioma promising. This is further augmented by the favorable responses, including observation of abscopal effects, in other non-CNS malignancies when a similar approach is utilized. Finally, TTF are the most recently approved therapy for the treatment of newly diagnosed and recurrent gliomas. They are well tolerated in part because of a lack of overlapping toxicities with other tumor treatments. This makes them an attractive component of combinatorial regimens. Further understanding of the preclinical and clinical effects on immune activity will help provide insight regarding if they would likely provide additive or even synergistic benefit when used in conjunction with immunotherapies.

6. CONCLUSIONS

For immunotherapeutic approaches to prove successful in the setting of infiltrating glioma, it will require careful consideration of the extensive preclinical data and the emerging large-scale clinical data. This will be facilitated by robust discussion on why our progress has been disappointing thus far, and thoughtful consideration on how to best overcome our limitations. It is unlikely that haphazard combinatorial approaches will prove remarkably fruitful. The exploration of novel modifiers of immune response including immunosuppression due to advanced aging [82], the host-microbiome interaction,[83] psychoneuroimmunologic stress [84], and ambient temperature effects on immune activity,[85] may lend us greater insight. There is clear evidence of individual differences in response to immunotherapy noted across GBM, and identifying molecular features that account for this can lead to effective personalized use of immunotherapy.

KEY POINTS.

• The understanding of what drives the immunosuppression associated with infiltrating gliomas continues to evolve.

• While large trials of vaccines or immune checkpoint inhibitors did not meet their primary endpoints, they have lent insight into how immunotherapeutics may prove effective in treating these tumors.

• A range of exciting immunotherapeutic strategies, often employing combinatorial approaches are being actively investigated in infiltrating gliomas.