Gene Therapy for the Treatment of Malignant Glioma

INTRODUCTION

In the United States, approximately 21,000 new cases of malignant gliomas are diagnosed per year [1]. Although patients with low-grade glioma generally have good prognosis [2], long-term survival is rarely achieved in patients with high-grade glioma (HGG). Patients with glioblastoma (GBM), the most aggressive form of HGG, have a median survival of 14 to 21 months [3,4]. Current standard of care involves aggressive surgical resection followed by concomitant and adjuvant temozolomide (TMZ) with radiotherapy [3] and potentially tumor-treating fields [3,4]. However, tumor recurrence invariably occurs, on which treatment options are limited. Several US Food and Drug Administration (FDA)–approved therapeutics, such as bevacizumab, carmustine wafers, tumor-treating fields, and lomustine, are able to prolong survival, but they are not curative [5].

HGGs are notoriously difficult to treat. The sensitive location and extensive infiltration of these tumors make complete resection impossible without accompanying neurologic deficits, whereas the blood-brain barrier (BBB) limits the use of chemotherapies effective in other forms of cancer [6]. Radiotherapy can provide initial therapeutic benefit in newly diagnosed patients with HGG, but radioresistance is acquired on tumor recurrence [7]. Immunotherapy approaches such as immune cell cycle inhibitors or chimeric antigen receptor T-cell therapy, which have shown great promise in other malignancies, have failed to show meaningful therapeutic response in patients with HGG, which is likely caused by systemic and local immunosuppression characteristic of HGG [8,9]. Furthermore, genetic heterogeneity within and between tumors limits the use of molecular targeted therapies [10]. Thus, there remains a need for improved therapeutic options for these patients.

Gene therapy, which involves the transfer of exogenous genetic material into target cells in order to produce a therapeutic effect, has been extensively studied for the treatment of hereditary diseases as well as cancer [11]. In the past 2 decades, large-scale clinical studies involving gene therapy for HGG have increased the understanding of how this therapeutic modality behaves in patients and has also established the safety profile of a variety of methods used to deliver genetic payloads.

This article discusses contemporary clinical studies involving gene therapy in both recurrent and newly diagnosed patients with HGG. The definition of gene therapy here is limited to treatment methods that strictly involve the introduction of exogenous nucleic acids meant to elicit a therapeutic response. Studies involving oncolytic viral therapies are not included because there have been several high-quality reviews written on this subject in the past several years [12-15]. The successes and lessons learned from clinical investigation of these experimental gene therapies are discussed.

GENE THERAPY FOR THE TREATMENT OF CANCER

As a general overview, gene therapy for the treatment of cancer involves the use of a transgene in order to elicit apoptosis, control tumor growth, or lead to the locoregional production of a product that could prove detrimental to the tumor cells. The mechanism by which the inserted gene of interest causes cellular death can be used to classify the different forms of gene therapy currently under investigation. Suicide gene therapy involves transducing cancer cells to express a nonmammalian enzyme that can convert inert prodrugs into cytotoxic agents [16]. Immunomodulatory gene therapy is intended to encourage an immune response against tumors by increasing tumor cell expression of inflammatory cytokines [17]. In addition, gene therapy has also been used to modulate the tumor microenvironment, such as blocking angiogenesis [18].

An important consideration in the application of gene therapy for the treatment of cancer is the method by which exogenous nucleic acids are delivered and integrated into the target cell, because therapeutic response is greatly determined by the transduction efficiency. Various vehicles have been developed to deliver genetic material into cancer cells, including liposomal carriers, viral vectors, and more recently nanoparticles [19]. Although many of these vectors have been tested in the preclinical setting, only a few have been tested in the clinic, and, of those, even fewer have shown the requisite transduction efficiency to generate therapeutic response.

Viruses have been the most studied gene therapy vector because of their natural ability to transduce host cells. Adenoviruses and retroviruses are the 2 different viral vectors that have undergone the most investigation in patients with HGG. There are important differences between these 2 vectors (Fig. 1). Adenoviruses can package double-stranded DNA, whereas retroviruses envelop single-stranded RNA. Both vectors can incorporate inserts up to 8 kb. Importantly, retroviral vectors are able to integrate their genetic payload within the host genome, allowing sustained gene expression, whereas adenoviral vectors can only provide transient gene expression [20]. A comprehensive list of recent clinical trials investigating the use of viral vector–mediated gene therapy can be found in Table 1.

FIG. 1.

Basic structure of viral vectors used in gene therapy for glioma. dsDNS, double-stranded DNA; ssRNA, single-stranded RNA.

TABLE 1.

Recent Clinical Trials Investigating Gene Therapy for the Treatment of Glioma

First Author	Year	Phase	N	Histology	Newly Diagnosed vs Recurrent	Therapeutic Agent	Concurrent Treatment	Overall Survival
Rainov [25]	2000	III	248	GBM	Newly diagnosed	HSVtk + GCV	Surgical resection with RT	365 d (364 d)
Sandmair et al [27]	2000	I	14	GBM or AA	Newly diagnosed or recurrent	HSVtk + GCV VPC vs Ad as vehicle	Surgical resection,	15.0 mo (Ad), 7.4 mo (VPC), 8.3 mo (control)
Immonen et al [29]	2004	II	36	GBM or AA	Newly diagnosed or recurrent	HSVtk + GCV	Surgical resection with RT	62.4 wk (37.7 wk)
Westphal et al [30]	2013	III	250	GBM	Newly diagnosed	HSVtk + GCV	Surgical resection with RT (TMZ used at discretion of clinician)	497 d (vs 452 d)
Cloughesy et al [36,39]	2016	I/II	45	GBM or AA	Recurrent	Toca511 + TocaFC	Previously received TMZ with radiotherapy, repeat surgical resection	13.6 mo
Cloughesy [44]	2019	III	403	GBM or AA	Recurrent	Toca511 + TocaFC	Investigators’ choice of lomustine, TMZ, or bevacizumab	11.07 mo (12.22 mo)
Chiocca et al [50]	2018	I	31	GBM or AA	Recurrent	Ad-RTS-hIL-12 + veledimex	Resection	12.7 mo
NCT03636477	Ongoing	I	21	GBM	Recurrent	Ad-RTS-hIL-12 + veledimex	Resection and nivolumab	NA
NCT04006119	Ongoing	II	30	GBM	Recurrent	Ad-RTS-hIL-12 + veledimex	Resection and cemiplimab-rwlc	NA
Brenner [18]	2019	I/II	72	GBM	Recurrent	VB-111	Bevacizumab	414 d
Cloughesy et al [64]	2019	III	256	GBM	Recurrent	VB-111	Bevacizumab	6.9 mo (vs 7.9 mo)

Abbreviations: AA, anaplastic astrocytoma; Ad, adenovirus; GCV, ganciclovir; hIL-12, human IL-12; HSVtk, herpes simplex virus-1 thymidine kinase; IL-12, interleukin-12; RT, radiation therapy; RTS, RheoSwitch Therapeutic System; TMZ, temozolomide; Toca511, vocimagene amiretrorepvec; TocaFC, extended-release 5-fluorocytosine; VB-111, ofranergene obadenovec; VPC, retroviral vector packaging cell.

SUICIDE GENE THERAPY FOR GLIOMA

Suicide gene therapy involves the insertion of a suicide transgene into target cancer cells. The transgene codes for a nonmammalian enzyme that can convert normally inert prodrugs into cytotoxic agents (Fig. 2). Traditional chemotherapy is often limited by systemic side effects, especially in the hematopoietic compartment, because most chemotherapeutic agents act nonspecifically on dividing cells. Suicide gene therapy circumvents this problem, because, theoretically, only cells expressing the suicide transgene are affected by the systemically administered prodrug. However, early clinical trials investigating suicide gene therapy in patients with HGG revealed that nontransduced cells within the vicinity of transduced cells could also undergo apoptosis because of the transfer of the activated prodrug through cellular gap junctions, or through their release into the local microenvironment. This phenomenon has been termed the bystander killing effect [21,22].

FIG. 2.

Mechanism of suicide gene therapy. After transfection by viral vector, transfected cells begin producing viral protein, which is able to convert inert prodrug into activated cytotoxic agent. The activated drug can act on cells within the vicinity, termed the bystander effect.

Two suicide gene systems have been intensively investigated for the treatment of patients with HGG. The first system uses the herpes simplex virus thymidine kinase (HSVtk) enzyme with ganciclovir (GCV), a synthetic analogue of the guanosine. The second system uses cytosine deaminase (CD) with 5-fluorocytosine (5-FC). Although phase III studies involving both systems did not meet primary endpoints of increasing overall survival (OS), a portion of patients seem to show durable complete responses [23,24] Further analysis of both studies is warranted.

HERPES SIMPLEX VIRUS THYMIDINE KINASE/GANCICLOVIR SYSTEM

HSVtk phosphorylates GCV into monophosphate GCV (p-GCV). Once phosphorylated, GCV can be recognized by cellular kinases and is converted into its triphosphate form. Triphosphorylated GCV interferes with DNA synthesis in S-phase cells, thereby triggering apoptosis [21]. Many clinical trials investigating the HSVtk/GCV system were conducted before the addition of TMZ as a standard therapy. The earliest clinical trial investigating suicide gene therapy for the treatment of patients with HGG involved stereotactically implanting murine retroviral vector packaging cells (VPCs) containing the HSVtk gene into tumor resection cavity followed by GCV administration (5 mg/kg intravenously) twice a day for 14 days [25]. The main end points of this study were to evaluate safety, determine the degree of gene transfer to tumor cells, and look for evidence of unintentional gene transfer to organs outside of the brain. The treatment was generally well tolerated, with most adverse events being attributable to the surgical nature of the gene therapy procedure. Thirteen patients were evaluable for response determined through MRI contrast enhancement. Four patients showed a 50% decrease in volume of contrast enhancement after GCV treatment. In situ hybridization for messenger RNA (mRNA) revealed that transfection efficiency of tumor cells was low and limited to tumor cells bordering the needle tracks, and retroviral vector DNA was undetectable in patients’ peripheral blood samples.

Because the treatment was well tolerated and showed signs of clinical efficacy in a portion of the patients, this treatment using VPCs was evaluated in a controlled, multicenter, open-label, phase III trial [25]. As noted earlier, this study was conducted before the TMZ era. Two-hundred and forty-eight newly diagnosed patients with GBM were randomized 1:1 to receive standard surgical resection with high-dose radiotherapy (56–60 Gy) with or without HSVtk/GCV gene therapy. Median survival was indistinguishable between gene therapy and standard therapy groups (365 vs 364 days). The investigators cite low rate of HSVtk transduction as the main source of therapeutic failure. A preclinical study with rat glioma models found that 10% of tumor cells must be transduced in order to reduce tumor size significantly [26]. However, evaluation of 7 patients’ tumor tissues found that transduction efficiency was always less than 0.002% [24,25].

Another study investigating HSVtk with GCV in humans directly compared the efficacy of retroviral VPCs with a replication-deficient adenoviral vector (AdV) containing HSVtk [27]. A cohort of patients received AdV containing lacZ gene (whose product is β-galactosidase) served as a control group. All patients treated with VPCs showed tumor progression at the 3-month follow-up, whereas 3 of 7 patients treated with the HSVtk AdV remained stable. OS was also improved in the AdV-treated patients compared with VPC and control patients (15.0 months vs 8.3 months, log rank test P<.012). Based on these results, the use of retroviral VPCs was discarded and a nonreplicating adenoviral vector containing the HSVtk gene was developed by Ark Therapeutics, named Cerepro (sitimagene ceradenovec) [28].

A phase II study in newly diagnosed or recurrent GBM or anaplastic astrocytoma (AA) comparing sitimagene ceradenovec with GCV with standard surgical resection with radiotherapy provided evidence of therapeutic efficacy in the gene therapy group [29]. Thirty-six patients were randomized into the 2 study groups. Median OS was 65% longer for patients receiving gene therapy compared with control treatment (62.4 weeks vs 37.7 weeks, log rank test P = .0095). Recurrent versus primary brain tumors were evenly distributed between control and experimental gene therapy groups; however, the gene therapy group had a higher ratio of patients with AA/GBM compared with the control therapy group (1:3.5 vs 1:18). Despite this, post hoc subgroup analysis showed that the survival benefit seen with gene therapy held even when excluding all patients with AA. At the time of the study, the contemporary molecular classification of gliomas was not in use. The median OS for patients with GBM receiving gene therapy was 55.4 weeks versus 37.0 weeks for those receiving standard therapy (log rank test, P = .0214). Vector DNA was detected in plasma via polymerase chain reaction in 2 patients 3 days following gene therapy, but not at 4 or 7 days after therapy. The therapy was well tolerated in most patients, except for 2 patients who developed cerebral edema during GCV administration, which resolved following surgical intervention. The results of the phase II study supported the initiation of a randomized, open-label, multicenter phase 3 trial in newly diagnosed patients with GBM.

The phase III trial investigating sitimagene ceradenovec with GCV enrolled 240 patients; 124 patients were assigned to the experimental group and 126 to the standard care group [30]. Patients received perilesional injection of Cerepro (1 × 10¹² viral particles [vp]) followed by the same GCV regimen administered in the phase II study in addition to standard care. The results of the study were as follows: median time to death or reintervention was longer in the experimental group (308vs 268 days; hazard ratio [HR], 1.53; 95% confidence interval [CI], 1.13–2.07; P = .0057); however, in terms of OS, there was no difference. Note that in this study an HR greater than 1 signified a benefit for sitimagene ceradenovec. The most common study-specific adverse events were fever, increased frequency of seizures, transient focal neurologic deficits, and hyponatremia. However, none of these events were life threatening or fatal.

Standard care was heterogeneous because the study was conducted during the time TMZ use, an active agent in this disease, began to become widespread. TMZ was allowed at the discretion of the investigator in both the standard-of-care and experimental group. Median time to death or reintervention was still improved with the use of gene therapy irrespective of TMZ use. Because O6-methylguanine-DNA methyl-transferase (MGMT) promoter methylation emerged as a prognostic factor in treatment with TMZ, MGMT promoter methylation analysis was also conducted. The investigators found that patients with nonmethylated MGMT promoters also showed improved time to death or reintervention in the gene therapy group (HR, 1.72; P = .008).

Interestingly, the HSVtk/GCV system has been shown to transduce and trigger apoptosis in endothelial cells (ECs) within the tumor microenvironment, in turn blocking angiogenesis [24]. An explanation as to why investigators observed an increase in time to reintervention in the gene therapy group could be that sitimagene ceradenovec inhibited angiogenesis, thereby increasing time to increased radiographic contrast enhancement.

Although HSVtk/GCV did not produce a robust increase in OS, a few key lessons emerged from the sitimagene ceradenovec phase III study. First, progression-free survival can be confounded by pseudoprogression or pseudoresponse, especially in the context of new experimental therapies. Second, although OS is a robust marker for therapeutic efficacy, it may mask meaningful treatment effects because treatment on recurrence is not controlled. The investigators also discuss the role the immune system may play in suicide gene therapy, because they observed that efficacy of HSVtk/GCV increased in patients with a robust baseline immune systems.

CYTOSINE DEAMINASE WITH 5-FLUOROCYTOSINE

The CD with 5-fluorocytosine (5-FC) suicide gene system is another form of gene therapy systematically evaluated in patients with HGG. 5-FC is converted by CD into cytotoxic agent 5-fluorouracil (5-FU), which is then converted into several active metabolites that act to disrupt RNA synthesis and the nucleotide synthesis through inhibition of thymidylate synthase [31]. Yeast CD was found to be nearly 15-fold more effective at converting 5-FC to 5-FU [32]. Thus, Ostertag and colleagues [33] developed a yeast CD, optimized to be thermally stable, for cancer therapy.

In contrast with the sitimagene ceradenovec approach, a nonlytic replicating retroviral vector (RVV) to deliver the yeast CD transgene was developed [33]. RVVs are an attractive vector for neuro-oncological gene therapy because nuclear envelope breakdown during mitosis is required for stable genome integration [34]. Because most nervous tissue is postmitotic, RVVs selectively integrate with malignant dividing cells in the brain. Furthermore, because RVVs are slightly immunogenic, they infect glioma cells more efficiently because of the blunted innate and adaptive immune responses in these tumors [35]. This system was named Toca 511 (vocimagene amiretrorepvec), and the investigators showed complete brain tumor eradication with Toca 511 activation of 5-FC in 2 syngenic GBM mouse models.

The therapeutic efficacy prompted the initiation of a phase 1 clinical trial [36]. The investigators enrolled 45 patients with recurrent HGG in an open-label, dose-ascending, multicenter study. All patients had previously received radiotherapy with TMZ. Safety, efficacy, and molecular profiling were evaluated. Patients underwent surgical resection for their recurrent tumors and received Toca 511 injected directly into the resection cavity wall, followed by orally administered cycles of Toca FC, an investigational extended-release formulation of 5-FC.

Forty-three out of 45 patients were evaluable for efficacy; of these, 35 were patients with GBM. Twenty-seven out of these 35 patients were at their first or second recurrence. Toca 511 with Toca FC was generally well tolerated with no treatment-related deaths. In patients with GBM with first or second recurrence, treatment with Toca 511 and Toca FC resulted in fewer treatment-emergent grade 3 or greater adverse events compared with an external matched control of patients with GBM in first or second recurrence receiving systemically delivered lomustine (3.7% vs 36.9%). Toca FC showed a dose-dependent increase in plasma concentration, with serum concentration of Toca FC reaching 100 μg/mL in the 220-mg/kg patient cohort. Therapeutic efficacy of Toca 511/FC treatment also trended toward dose dependence. Cohorts of patients receiving higher doses of Toca 511/FC had improved median OS compared with patients receiving lower doses, although this difference was not statistically significant (14.4 vs 11.8, log rank test P = .13). Compared with the external lomustine control group, Toca 511 and Toca FC showed improved median OS in patients with GBM at first or second recurrence (13.6 vs 7.1 months, log rank test P = .003). Median OS from date of diagnosis was also improved for these patients compared with an external control of patients with GBM in first or second recurrence receiving systemically delivered lomustine (29.2 vs 21.3 months; log rank test P value, 0.025; HR, 0.55; 95% CI, 0.32–0.93). Although lomustine is a frequently used second-line therapy, patients receiving off-study systemic chemotherapy for recurrent disease may not reflect the patient population, which qualifies for a surgically based clinical trial.

In an attempt to characterize a transcriptional profile of patients that might better benefit from Toca 511/Toca FC therapy, mRNAs from tumor biopsies before Toca 511 treatment were sequenced. Five out of 8 patients with GBM living greater than 12 months displayed increased levels of mRNAs encoding proteins involved in neuronal function, bearing great similarity to the neuronal subtype originally classified by Verhaak and colleagues [37]. The previously reported neuronal subtype has been called into question as a true subtype instead of normal neuronal tissue contamination [38]. Because the investigators do not report whether the sequenced tumor samples were confirmed via histopathology to contain mostly tumor cells, it is impossible to rule out whether or not the signature they detected was caused by normal neuronal tissue contamination, possibly suggestive of a lesser overall tumor cell burden. Further molecular analysis is warranted to confirm transcriptional profiles related to greater therapeutic response.

Subsequent follow-up analyses of the original trial results included an additional 11 patients and 2 years of safety and efficacy monitoring [39]. Six of the 53 total patients receiving Toca 511/FC showed complete responses of enhancing tumor determined through radiographic imaging. An additional 10 patients showed stable disease according to Response Assessment in Neuro-Oncology (RANO) criteria [40], bringing the total clinical benefit rate up to 30.2% (16 out of 53). The complete responses produced by Toca 511/FC treatment support an immunologic mechanism in addition to the direct cytotoxic mechanism.

Two preclinical studies conducted following the completion of the initial Toca 511 trial convincingly show that part of this therapy’s efficacy is caused by an immunologic response against the tumor [41,42]. By profiling the immune cell population of heterotrophic glioma xenografts, Mitchel and colleagues [41] found that immunosuppressive cells such as tumor-associated macrophages and myeloid-derived suppressor cells were decreased 6 and 9 days following the start of Toca 511/FC treatment. At the same time, CD8+cytotoxic T-cell levels were increased within the tumors. Taken together, these observations suggest that Toca 511/FC treatment is able to shift the normally immunosuppressive tumor microenvironment to a more inflammatory state [41]. This shift occurs while avoiding the more global immunosuppression noted with systemically administered cytotoxic chemotherapy.

In an accompanying study, Hiraoka and colleagues [42] tested the Toca 511/FC treatment in 2 different intracranial glioma xenograft models. In the immuno-deficient model, Toca 511/FC was able to control tumor growth through repeated and consistent administration of the therapy. Alternatively, in the syngeneic mouse glioma model, 1 cycle of Toca 511/FC administration was enough to produce long-term survivors. On rechallenge with naive glioma cells, antitumor immunity was enough to prevent tumor formation.

Further molecular analyses of patient blood and tumor samples found that Toca 511 is predominantly detected in tumor samples, and only transiently in blood [43]. Furthermore, because RVVs integrate into the host cell’s genome, integration that would disrupt tumor suppressor genes or promote oncogenes is a valid concern. Sequencing of patient samples for Toca 511 integration sites found no evidence of inappropriate integration or clonal expansion [43].

Based on these encouraging preclinical and clinical studies, a phase III study (NCT02414165) was planned investigating Toca 511/FC treatment in patients with recurrent HGG [44]. The inclusion criteria for this study included patients with GBM or AA at first or second recurrence, with recurring tumors less than or equal to 5 cm. Four-hundred and two patients were enrolled in the study and randomized 1:1 to experimental Toca511/FC treatment or standard of care using lomustine, bevacizumab, or TMZ. Randomization occurred intraoperatively after the surgeon completed a maximal resection but before potential injection of the investigational agent. Primary end point of this study was OS, and secondary end points included durable complete response rate, durable clinical response rate, duration of durable response, and OS at 12 months.

However, the results of this much-anticipated trial were negative. Toca 511/FC failed to improve OS compared with standard of care. Furthermore, the durable complete responses noted in the previous phase I trial were not seen in the phase III trial. However, subgroup analysis showed that patients with second recurrence did have a reduced risk of death when treated with Toca 511 and FC. Most of the patients with second recurrence in the treatment population were isocitrate dehydrogenase (IDH) mutant AA [44]. The therapeutic benefit seen in this subgroup of patients may be caused by greater potential to generate antitumor immune response, as determined by immune phenotyping of blood and tumor samples collected at baseline. Flow cytometry analysis of peripheral blood samples collected during this trial suggest that immune modulation does occur in a subset of patients receiving Toca 511/FC. Combinatorial treatment with immune checkpoint inhibitors or other forms of immunotherapy with Toca 511/FC treatment have been proposed.

Despite not meeting primary and secondary end points, the results of the randomized controlled phase III Toca 511/FC trial showed an acceptable safety profile and have shown the feasibility of intraoperative randomization in a large multicenter study. A subset of patients with recurrent AA with IDH mutant status may benefit from this therapy. Because of the pressing need for improved upfront therapies in newly diagnosed patients with GBM, an ambitious phase II/III open-label study comparing adjuvant Toca 511/FC treatment with current standard of care or standard of care alone is currently in development in the cooperative group setting (NCT04105374) [44].

GENE THERAPY FOR LOCAL PRODUCTION OF INFLAMMATORY CYTOKINES

Although immunotherapeutic approaches such as immune checkpoint inhibitors and chimeric antigen receptor T cells have yet to produce meaningful survival benefit in patients with glioma, the successes of these approaches in producing durable complete responses in other cancers have emphasized the delicate interplay between tumor recognition and evasion from the immune system [45]. The use of inflammatory cytokines such as interferon gamma (IFN-γ ) or interleukin-12 (IL-12) to overcome the immunosuppressive microenvironment characteristic of glioma offers another immunotherapeutic approach.

IL-12 is a heterodimeric protein that links the innate and adaptive immune systems [46]. It is endogenously produced by antigen-presenting cells and acts on natural killer and T cells by differentiating naive CD4+ T cells to the T-helper 1 phenotype and activating naive T cells to activated CD8+ cytotoxic T cells. Systemic administration of IL-12 was able to control tumor growth in a variety of preclinical models [47]; however, phase I/II clinical trials administering recombinant IL-12 in humans were halted because of severe toxicity [48]. The results of these studies highlighted the therapeutic value of IL-12 but also the need for careful and controlled administration.

To overcome the issue of adverse systemic reactions but maximize IL-12 production within the tumor, Ziopharm is developing the RheoSwitch Therapeutic System (RTS) for the treatment of cancer (Fig. 3). Using an adenoviral vector, the RTS system is delivered into target cells. Under this system, transcription of the IL-12 gene only occurs in the presence of an activator ligand, veledimex (VDX). In a series of preclinical studies [49], VDX was shown to cross the BBB at high enough levels to activate RTS-mediated transcription of IL-12 in mouse and monkey models. Furthermore, IL-12 transcription through RTS activation was able to eradicate GL261 brain tumors in C57/BL6 mice. Following these preclinical studies, a phase I trial was recently completed in patients with recurrent HGG.

FIG. 3.

RTS for inducible transcription of human IL-12. The RTS gene program includes 2 receptor protein fusions: VP16-RXR (coactivation partner [CAP]) and Gal4-EcR (ligand-inducible transcription factor [LTF]). In the absence of the activator ligand (veledimex), the LTF binds to the inducible promoter and does not form a stable interaction with the CAP. In the presence of veledimex, a conformational change in the LTF leads to a stable, high-affinity interaction with the CAP allowing recruitment of basal transcription proteins and transcription of the target gene, IL-12. (Courtesy of Ziopharm Oncology, Inc., Boston, MA.)

In this multicenter dose-escalation phase I trial [50], the safety and biological effects of adenoviral-delivered RTS for human IL-12 (ad-RTS-hIL-12) with VDX was evaluated in 31 patients with recurrent high-grade glioma eligible for surgical resection. Following tumor resection, ad-RTS-hIL-12 was injected into the resection cavity walls at a fixed dose of 2 × 10¹¹ vp. In order to assay blood-tumor barrier penetration of VDX in humans, VDX was given preoperatively at levels of 10, 20, 30, and 40 mg. Plasma and tumor VDX concentrations increased in a dose-dependent manner, with tumor VDX levels being about 40% that of plasma VDX levels. VDX at 30 and 40 mg was poorly tolerated, so 20 mg was determined to be the optimal dose.

IL-12 and IFN-γ serum levels were assayed before, during, and after VDX dosing, and both cytokines showed a sharp spike in serum levels during VDX dosing, consistent with transient transcription of IL-12. Five patients with progressive radiographic contrast enhancement underwent reresection 1 to 5 months following initial ad-RTS-hIL-12 administration. Of these 5 patients, 3 had enough tumor tissue to compare immune cell populations within the tumor before and after treatment. Posttreatment tumors had increases in tumor-infiltrating lymphocytes (TILs), CD8+ T cells, programmed cell death protein 1 (PD-1)–positive immune cells, and programmed death-ligand 1 (PD-L1)–expressing cells, consistent with a cytotoxic inflammatory response and initiation of a negative feedback mechanism (PD-L1). Furthermore, intratumoral IFN-γ levels increased from 4 ± 3 pg/g to 226 ± 138 pg/g following ad-RTS-hIL-12/VDX treatment.

Median OS in the cohort of patients receiving 20 mg of VDX was 12.7 months, compared with 7.3 months for the 10-mg, 30-mg, and 40-mg patient cohorts. Cox regression analysis was used to determine potential prognostic variables for OS. This analysis found that cumulative dexamethasone exposure to be less than 20 mg, and being male to be positively associated with OS. Furthermore, increases in cytotoxic T cell to regulatory T-cell ratio from baseline in peripheral blood samples positively correlated with OS.

In conclusion, this study showed proof of concept that the ad-RTS-hIL-12 system can tightly regulate IL-12 production with the use of VDX. Furthermore, this study established 20 mg to be the optimal dose of VDX that resulted in efficacy with minimum toxicity and showed preliminary evidence of therapeutic benefit with this treatment. There are ongoing clinical investigations adding PD-1 antibody to overcome the negative feedback loop in both a phase 1 substudy (NCT03636477) and a nonrandomized phase 2 trial (NCT04006119). Within the context of the phase 2 trial, the clearance of all tumor clones with DNA mismatch repair deficiency has been demonstrated in a patient treated with the combinatorial regimen [51].

GENE THERAPY FOR TUMOR MICROENVIRONMENT MODULATION

Inducing angiogenesis is a hallmark of cancer [52]. As some (but not all) tumors grow in size, they require the generation of new blood vessels in order to provide sufficient oxygen and nutrients while eliminating metabolic waste and carbon dioxide. In glioma, vascularization increases with tumor grade because of the increased production of proangiogenic factors such as vascular endothelial growth factor (VEGF) [53,54]. However, it has been shown that glioma can progress with extensive tumor cell migration without the need for angiogenesis [55]. In turn, this leads to the questioning of this target as a means for improving OS in this disease. Antiangiogenic therapies seek to starve proliferating tumor cells of oxygen and nutrients by preventing the formation of new blood vessels. Bevacizumab is a humanized monoclonal antibody that targets VEGF, and it was shown to improve progression-free survival in patients with recurrent GBM (rGBM) [56]. However, randomized phase III trials (NCT00884741, NCT00943826) have not shown efficacy in improving OS using this agent [57,58].

Ofranergene obadenovec (VB-111) offers an alternative mechanism to disrupt tumor angiogenesis (Fig. 4). VB-111 is a gene therapy designed to elicit apoptosis not in the tumor cell itself but instead in the epithelial cells that form tumor neovasculature. VB-111 uses a nonreplicative adenoviral vector to carry a transgene for a chimeric protein, Fas-c, which contains the extracellular region of tumor necrosis factor receptor 1 (TNFR-1) and the transmembrane and intracellular region of Fas [59]. When tumor necrosis factor alpha binds to Fas-c, cells undergo Fas pathway–mediated apoptosis. Fas-c expression is controlled by a specially modified murine preproendothelin-1 (PPE1) promoter. Transgene expression under the control of the PPE1 promoter is specifically limited to proliferating ECs [60,61]. Unlike other gene therapies requiring direct intratumoral injection, targeting cells on the adluminal side of the BBB allows systemic administration. Preclinical studies using rodent glioma models established efficacy of a single dose of VB-111 administered intravenously. Median survival of nude mice bearing U87 human glioma xenografts treated with VB-111 improved by a modest 13 days compared with control mice. Furthermore, tumors treated with VB-111 had decreased levels of microvessel proliferation compared with control mice as measured through CD31 immunohistochemistry staining, confirming the antiangiogenic properties of the agent [62].

FIG. 4.

Antiangiogenic mechanism of VB-111. Transfected epithelial cells produce chimeric protein Fas-c, which activates Fas-mediated apoptosis when bound to VEGF. Unable to produce new vasculature, tumor cells are unable to eliminate metabolic waste or receive necessary nutrients.

A phase I dose-escalation study evaluating the safety and tolerability of VB-111 was conducted in 33 patients with advanced solid tumors [63]. Patients received a single intravenous infusion of VB-111 with doses from 1 × 10¹⁰ to 1 × 10¹³ vp. VB-111 was exceptionally well tolerated, with the most common adverse events being grade I to II fever, chills, and fatigue, with these events being more common in the higher-dose cohorts. There was only 1 occurrence of a grade III adverse event of pyrexia at the highest dose of 1 × 10¹³ vp, which resolved with administration of acetaminophen. Although the patients enrolled in the study had tumors of different origin, which makes survival comparison difficult, there was radiographic evidence of treatment efficacy in 1 patient with radioiodine-resistant metastatic papillary thyroid cancer.

Given the favorable safety profile of VB-111, a phase I/II study was conducted in patients with rGBM [18]. This study enrolled 72 patients with rGBM into 4 different treatment groups: a subtherapeutic dose-escalation group (SubT); a limited exposure (LE) group that received VB-111 (1 × 10¹³ vp) every 56 days until tumor progression; a primed combination group that received VB-111 (1 × 10¹³ vp) every 56 days as a monotherapy until tumor progression, on which patients also began receiving bevacizumab every 14 days in addition to VB-111; and an unprimed combination group that received VB-111 (1 × 10¹³ vp) every 28 days and bevacizumab every 14 days.

The safety profile of VB-111 was similar to the previously conducted phase I dose-escalation study. Approximately half of patients receiving VB-111 at 1 × 10¹³ vp had fever or flulike symptoms that responded to the use of antipyretics, which is consistent with viral vector infection. Median OS was significantly extended in the primed combination group (414 days) compared with the LE group (223 days; HR, 0.48; 95% CI, 0.23–0.998; P = .043) and unprimed combination group (141.5 days; HR, 0.24; 95% CI, 0.09–0.66; P = .0056). The 12-month OS rate for primed combination group was significantly improved compared with a historical control group of 694 patients with rGBM receiving bevacizumab monotherapy (57% vs 24%; P = .03). Although patients enrolled in this study were not randomized, the data suggest that differences in survival between the LE and primed combination groups are not caused by baseline prognostic factors because patients in these 2 groups had similar time to progression (61 vs 60 days).

The subsequent GLOBE randomized controlled phase III study (NCT02511405) compared VB-111 with bevacizumab combination therapy and bevacizumab monotherapy in patients with rGBM [64]. Based on the superiority of the combination regimen in the earlier trial, patients with rGBM (n = 256) were randomized 1:1 to receive VB-111 (1 × 10¹³) every 8 weeks in combination with bevacizumab every 2 weeks (10 mg/kg) or bevacizumab alone. Patients who had previously received antiangiogenic therapy were excluded from this trial. The GLOBE study failed to meet both primary and secondary end points, which were OS and overall response rate (ORR) as measured by RANO criteria respectively. Median OS in patients receiving combination therapy was 6.8 months versus 7.9 months in patients receiving bevacizumab monotherapy (HR, 1.20; 95% CI, 0.91–1.59; log rank test P = .19) with ORR being 27.3% versus 21.9% (P = .26).

The negative result of the GLOBE study, although disappointing, lent potential support for an antagonistic effect of coadministration of VB-111 and bevacizumab, especially in light of the promising results seen in the phase II primed combination therapy group. Bevacizumab works by blocking the proangiogenic signal VEGF, which limits EC proliferation. It was previously shown that, under PPE-1 promoter control, transgene expression is highest in proliferating EC populations compared with quiescent ECs [61]. It is possible that concomitant administration of bevacizumab with VB-111 may have interfered with VB-111–triggered vascular disruption. According to the investigators, a new randomized, placebo-controlled, phase II study of neoadjuvant and adjuvant VB-111 in patients with rGBM will soon open, applying the lessons learned from the negative GLOBE study.

CONCLUDING REMARKS

Understanding of glioma has deepened tremendously over the past several decades, but, despite the hopes of the scientific and medical community, these vast amounts of knowledge have translated into only incremental improvements in survival for this deadly disease. With the addition of TMZ to surgery and radiotherapy, median survival has improved from 12.1 to 14.6 months [3]. However, this agent only improves outcomes for patients with GBM with methylated MGMT promoters. Patients with unmethylated MGMT promoters seem to derive little benefit from TMZ therapy because median OS for patients with methylated MGMT promoters treated with TMZ is 18.2 months versus 12.2 months for patients with unmethylated MGMT promoter tumors [65,66]. The addition of tumor-treating fields has improved on this [4]; however, there remains an unmet need for initial upfront therapies offered to these patients, which comprise nearly 60% of all GBM cases [66]. The need for novel therapeutic strategies is even more apparent in patients with HGG with recurrent disease, because all currently approved agents are noncurative.

Gene therapy approaches have been evaluated in several large randomized controlled studies in patients with both newly diagnosed and recurrent HGG [25,30,44,64]. Although most of these therapies have not met primary end points to increase median OS in the intention-to-treat population, it is important to take note that a subset of patients within these trials do respond to experimental gene therapy [67]. Cox regression analyses of the phase III Toca 511 survival data indicate that patients with recurrent AA with IDH1 mutations may benefit from this therapy more than current therapeutic options. Furthermore, in the phase I/II Toca 511/FC trial, a subset of patients showed complete durable responses. Meanwhile, the favorable safety profile and successful cell transfection seen in patients with rGBM treated with VB-111 supports further investigation of intravenously administered viral vector gene therapies as potentially viable approaches.

Another important lesson that has emerged from the completion of the phase III studies of Toca 511/FC, sitimagene ceradenovec with GCV, and VB-111 highlights the importance and difficulty of determining appropriate end points to gauge therapeutic efficacy in the context of experimental therapies. Response is generally determined through radiographic imaging; specifically, through the absence or shrinkage of contrast-enhancing lesions. However, most of the gene therapy approaches discussed here often involve an immune component, which can also produce radiographic contrast enhancement because of inflammation in the tumor microenvironment [68]. Using progression-free survival as an end point may lead to premature termination of experimental gene therapy.

Although the therapeutic efficacy of gene therapy approaches such as Toca 511 and sitimagene ceradenovec for the overall population of patients with HGG remains in question, the completion of these studies has improved the ability to design clinical studies to evaluate new gene therapy agents such as the ad-RTS-hIL-12 and VB-11 system. Furthermore, the past several decades of preclinical and clinical investigation for the use of these agents have established the safety and tolerability of the use of viral vectors as delivery systems for gene therapy. Future directions investigating gene therapy approaches are now combining gene therapy with other immunotherapeutic approaches [69]. Perhaps these multimodal therapeutic strategies will succeed where single-agent therapies have not.

KEY POINTS.

• A subset of patients with glioma show complete durable response following treatment with gene therapy.

• Recent large-scale randomized controlled clinical trials have established the safety and tolerability of viral vector–based gene therapy, specifically the Toca 511, VB-111, and sitimagene ceradenovec systems.

• Preclinical and clinical evidence suggests many viral gene therapy approaches involve an immunologic component, and further studies combining gene therapy with immunotherapy are being investigated.

CLINICS CARE POINTS.

• A variety of gene therapy approaches have been investigated as means of treating malignant gliomas.

• Viral vectors may be the most practical means of delivering a gene to intracranial cells for therapeutic purposes.

• Gene therapy can serve as a way to utilize immunotherapy, antiangiogenic therapy, and cytotoxic chemotherapy in a controlled fashion for the management of malignant brain tumors.

• Additional investigation is required to validate benefit of these types of approaches.