Dendritic Cell Vaccines for Brain Tumors

Introduction

Dendritic cells (DC) have long been regarded as the most potent antigen presenting cells (APCs) within the immune system. Their ability to sample environmental antigens and stimulate T cell activity in a major histocompatibility complex (MHC)-restricted manner has attracted much attention given the poor antigen presenting ability and immunogenicity of tumor cells [1,2]. Although DCs constitute approximately 0.3% of all circulating blood leukocytes, they serve as the sentinels of the immune system and are found nearly ubiquitously throughout the body [3]. In their immature state, DCs are highly specialized antigen samplers capable of surveying their microenvironment through several mechanisms including engulfment, macropinocytosis, and receptor mediated endocytosis [3]. Upon encountering an antigen the DC processes it through MHC pathways and directs it to the cell surface to form an MHC-peptide complex (Figure 1). In line with traditional antigen presentation following uptake from the environment, many antigens are channeled through MHC-class II pathways with resultant MHC-peptide complexes being capable of stimulating CD4+ T cells. In addition, dendritic cells possess the unique ability to “cross-present” acquired antigens. In this process, DC endosomes release captured antigenic material into the cytosol where it is broken down by proteasomes [4]. The degraded peptides are then transported to the ER via a transporter-associated protein (TAP) and bound to MHC-class I molecules for presentation to CD8+ T cells [5,6]. These distinct mechanisms allow DCs to stimulate T cells in an MHC-class I and II manner, overcoming classical restrictions in antigen processing and presentation [7] and diversifying the resultant immune response.

Figure 1

Schematic of dendritic cell antigen processing and presentation via distinct MHC-I and MHC-II targeted pathways. Foreign antigens are sampled from the environment via dendritic cell phagocytosis or pinocytosis. Once vacuolized, antigen-containing vesicles ...

Dendritic cells are capable of handling a vast range of antigenic mediums. The sources of antigen that have been used in DC immunotherapy include exogenous MHC-restricted peptides, acid-eluted tumor peptides, tumor RNA and cDNA, viral vectors, apoptotic tumor cells, tumor cell lysate, and whole tumor cells. Many of these methods have been employed with varying degrees of success. However, a growing sentiment has emerged that argues for the use of a diverse range of antigens that cover both MHC classes rather than constructing specific MHC-matched peptides. The reasoning for this is multifold. First, stimulating T cells with a broad range of antigens reduces the likelihood of an escape phenomenon in which tumor cells lacking the specific antigens of interest avoid immune detection and continue to grow unhindered. Second, it is now well established that the stimulation of both CD4+ and CD8+ T cells is crucial in the activation and maintenance of anti-tumor immunity [7–10]. By allowing DCs to present and cross-present antigens on MHC-class II and I molecules, respectively, one avoids having to laboriously engineer peptides for each MHC class [9,11]. Finally, the methods employed to load the spectrum of antigens for a particular tumor obviate the need of characterizing each individual antigen used. Although the use of unfractionated tumor material containing unknown antigens has long raised the concern of inducing auto-immunity, particularly in the form of experimental allergic encephalomyelitis (EAE), no reports of this complication have been seen following DC vaccination in humans to date [3].

A dendritic cell vaccine is defined as DCs loaded with antigens, for example, those found on glioma, which are administered to patients in order to induce an antigen-specific T cell mediated anti-tumor response [12]. However, though immature dendritic cells are not functionally ideal for the loading of antigens, they are unable to activate lymphocytes until an inflammatory signal or pathogen induces their maturation [3,9,11]. Some groups argue that ex vivo maturation of DCs through CD40L or interferon (IFN)-γ [13] is thus necessary prior to vaccine administration to ensure proper antigen presentation and T cell activation [14–17]. Others maintain that maturation occurs naturally, and that no prior stimulus is required [18]. In the process of maturation, DCs lose their ability to uptake and process antigens. Moreover, they exchange their immature molecular signature for a mature (CD83+) phenotype, increasing expression of MHC-antigen complexes, lymphocyte costimulatory molecules (e.g. CD80/B7-1 & CD86/B7-2), TNF and TNF-receptor molecules (e.g. CD40), and many chemokines and chemokine receptors (e.g. IL-12, IL-15, IL-18) to aid in T-cell recruitment and DC navigation to lymphoid tissues (as reviewed by Steinman [11] and Soling [9]).

Upon localization to lymph organs rich in naïve T cells, mature DCs present their processed antigens in a MHC-restricted manner. Through various interactions they are able to mobilize many different arms of the immune system, including CD8+ cytotoxic T cells (CTLs), CD4+ helper T cells, natural killer (NK) and NK-like T cells [11]. Each of these cell types plays an essential role in the anti-tumor response (Figure 2). T cells expressing CD8 co-receptors recognize and lyse tumor cells in an MHC-class I restricted fashion, and have received much of the credit as the primary effector cell in immunotherapy. CD4+ T cells have traditionally been known for their part in the expansion and maintenance of CD8+ CTLs, secretion of stimulatory cytokines, and the induction of lasting immunity. Their critical role in immunotherapy has become increasingly appreciated over the past few years, as studies have demonstrated that their absence may result in deficient DC maturation as well as CTL tolerance [7,8]. Finally, NK and NK-like T cells have a unique niche in the leukocyte armament, being able to recognize and kill tumor cells that do not express surface markers such as MHC-class I. Although the exact mechanism of recognition and elimination of tumor cells in the absence of MHC-restriction is yet to be elucidated, they serve as an important complement in the killing of tumor cells that may possess diminished surface marker presentation and avoid CTL detection [3,11].

Figure 2

Diagram depicting the multiple dendritic cell-lymphocyte interactions that take place in the immune cascade following antigen processing and presentation by DCs. Dendritic cell activation of NK T cells through self ligands (not shown) and IL-12 results ...

Animal Models

Pre-clinical animal models explored many of the methodologies and safety concerns regarding dendritic cell vaccines. In the late 1990’s, we were among the first to describe the effectiveness of dendritic cell vaccines in the anti-tumor immunity of gliomas using a rat model [19]. Over the past decade, many groups have published similar studies with varying permutations in the choice of antigen, timing of vaccinations, and measures of therapeutic efficacy. The dissimilarities between study designs make it difficult to compare methodologies and their associated outcomes. However, their value lies within their ability to demonstrate the effectiveness of multiple different DC vaccine techniques in inducing antigen-specific cytotoxicity in vitro and in vivo. These studies have shown improved survival outcomes, as well as the safety of the strategy.

Many of the initial animal studies were key in evaluating the effectiveness of DC vaccination techniques for tumors located in the “immunologically privileged” central nervous system (CNS). Antigen sources used included synthetic peptides [20,21], acid-eluted tumor peptides [19], tumor lysate [21–26], DC-tumor fusion cells [27], and antigen containing vectors such as cDNA/RNA carrying viruses [28,29] and tumor extract carrying liposomes [30]. Many of these studies adopted strategies for antigen loading of DCs from previous experiments in peripheral neoplasms, and initial treatment schedules were similarly based on regimens that showed promise in non-CNS immunotherapy[19]. Nevertheless, the timing of DC administration variedly widely, with vaccinations being given before tumor inoculation [22,24,26–28], simultaneously with [21,23,25], or some time following tumor implantation [19,24,27,28,30]. The number of vaccinations ranged from two to five times across the various studies.

Overall, most groups concluded that vaccination with antigen-pulsed dendritic cells was able to produce a significant anti-tumor immune response. This was evidenced by increased overall survival in rats and mice, greater degrees of T-cell infiltration (primarily CD8+) on histological analysis of tumors, and more robust anti-tumor cytotoxicity assays when splenocytes were incubated with mouse glioma in vitro in immune responders. Although the studies conceded that pre-tumor vaccination resulted in greater survival in animal models, there are conflicting reports regarding the efficacy of DC vaccines when administered simultaneously or following tumor implantation [21–25]. Groups reporting no improvement in survival with DC vaccination after an established tumor suggest that it may be due to the immune system failing to generate an appropriate response quickly enough to counteract a rapidly growing tumor within a confined cranium[21,24]. However, studies in which T-cell mediated tumor killing was achieved showed that the animals that did respond to DC vaccination obtained lasting anti-tumor memory, with significantly improved survival following tumor rechallenge compared to unvaccinated controls [21,22,26].

These animal studies played an important role in alleviating some of the concerns regarding DC immunotherapy for CNS tumors. Experimental allergic encephalomyelitis (EAE) in particular was perhaps one of the most feared side effects, as previous studies have shown this lethal form of autoimmunity to occur following the injection of glioblastoma tissue into animals [31]. However, such signs of autoimmunity were not reported in the more recent studies conducted to date [21,22]. In addition to the information they provided regarding the applicability of different antigen sources and treatment schedules, the pre-clinical reports corroborated the idea that DC immunotherapy could be effectively used for intracranial neoplasms, and thus set the stage for further clinical studies.

Clinical Trials

In 2000, we published a case report on the first brain tumor patient to be treated with DC-based immunotherapy [32]. A patient with histologically confirmed GBM received three biweekly intradermal injections of DCs pulsed with acid-eluted, allogeneic MHC-I matched GBM peptides. Although we were able to appreciate an immune response as evidenced by an increased infiltration of CD3+ T cells in post-vaccination tumor, there was no objective clinical response from the treatment. The patient’s poor Karnofsky performance score (KPS) in addition to the possible lack of antigen homology between the allogeneic GBM and the patient’s tumor may have contributed to the lack of clinical response or prolonged survival.

In a Phase I dose-escalation clinical trial, we treated 12 GBM patients using DCs pulsed with autologous acid-eluted MHC tumor peptides in a dose-escalation study [33]. Patients were separated into three cohorts, each receiving 1, 5, or 10×106 DCs per injection. Subjects tolerated the procedure well with no signs of autoimmunity. There were only minimal Grade I toxicities related to the study vaccine, which were distributed similarly across all three dose groups. Although this study was not powered to measure efficacy, patients undergoing DC-based immunotherapy appeared to have an increased median time to progression (15.5 months) and overall survival (23.4 months) compared to historical controls. Of note, tumor burden and disease progression at the time of vaccination was a critical determinant of systemic CTL activity, tumor infiltration by T cells, as well as overall survival. All patients that generated a systemic CTL response showed no MRI evidence of progressive disease at the time of vaccination. Conversely, no patient with actively progressive disease developed statistically significant cytotoxicity. Moreover, only patients possessing minimal tumor burden at the time of vaccination were found to have tumor infiltrating lymphocytes (TILs) upon post-vaccination tissue examination. These findings suggest that active tumor progression or bulky residual burden can debilitate the initiation and propagation of an anti-tumor response. Interestingly, expression of the inhibitory cytokine TGF-β2 was found to be inversely proportional to the number of TILs found in tumor tissue following vaccination (IL-10 was not), implicating TGF-β2 as a possible mediator of immune evasion following vaccination. This study argues for the need for maximal resection and/or minimal residual disease to improve the efficacy of DC-mediated immunotherapy for glioma. Importantly, this clinical trial established the feasibility, safety, and immunological potential of DC vaccines for brain tumor patients.

Yu et al. reported another Phase I clinical trial using DC pulsed with autologous, acid-eluted peptides for glioma patients [34]. In this study, nine patients with newly diagnosed malignant glioma received 3 biweekly subcutaenous injections of DCs loaded with acid-eluted tumor peptide. Systemic antitumor cytotoxicty was detected in 4 of the 7 patients assessed; intratumoral CD8+ CTL and CD45RO+ memory T-cell infiltration was found in 2 of the 4 patients who underwent a second resection due to tumor progression. Patients receiving DC vaccination were found to have an increased median survival (455 days) compared to that of those in the control group (257 days).

Kikuchi et al. used DC-glioma fusion cells (FCs) to vaccinate glioma patients in their Phase I clinical trial [35]. Eight patients with malignant gliomas received FCs intradermally every 3 weeks, with the total number of injections ranging from 1 to 9. An increased percentage of NK cells was found on FACS analysis in the peripheral blood of patients. In addition, an increase in IFN-γ release in PBMC-tumor co-incubation with both autologous as well as allogeneic glioma was seen following DC vaccination. Two patients experienced a minor response and no serious side effects were observed. These findings suggested that non-specific anti-tumor cytotoxicity may play a role in the DC-based immunotherapy of glioma.

Kobayashi et al. vaccinated five patients with autologous glioma RNA-pulsed DCs [36]. They were able to demonstrate the presence of a strong CD8+ CTL response against autologous glioma accompanied by a weaker NK cell-mediated cytotoxicity in their patients. This finding was significant in 3 of the 5 patients treated. Notably, in the two patients with minimal immune responses, a constitutively increased expression of the inhibitory cytokine IL-10 and decreased expression of IFN-γ by CD8+ T cells was found in vitro.

Yamanaka et al. compared different routes of DC injection in their Phase I/II clinical trial of 10 patients [37,38]. Dendritic cells were pulsed with autologous tumor lysate and administered to patients intradermally (n = 5) or both intradermally and intratumorally via an Ommaya (n = 5) reservoir every 3 weeks for a total number of injections ranging from 1 to 10. Immunologically, they observed an increased percentage of NK cells and increased T-cell mediated anti-tumor activity. In addition, there was an increased intratumoral infiltration of CD4+ and CD8+ T cells in the two patients who underwent reoperation following vaccination. Radiographically, the two minor responses seen were in patients included in the combined intradermal/intratumoral administration group, suggesting that the additional intratumorally injected DCs may stimulate a more efficient anti-tumor immune response.

Wheeler et al. published a report examining the correlation between thymic function, as manifest through CD8+ recent thymic emigrant production, age, and patient outcome in 17 GBM patients undergoing DC immunotherapy [39]. They found that thymic function, as reflected by its ability to produce CD8+ T cells, was directly proportional to good clinical outcomes in mice and human GBM patients and inversely proportional to age. Although patient age has long been a predictor of mortality and prognosis, their findings suggest that it is actually thymic function, which is inversely correlated with age, which may be the more telling factor. Thus, this non-specific immune parameter may later serve as an important prognosticator in glioma immunotherapy.

Caruso et al. conducted a Phase I study of 9 pediatric brain tumor patients undergoing immunotherapy via autologous tumor RNA pulsed DCs [40]. The cohort was comprised of a wide range of different tumor histologies (see Table 1). Although they detected a modest increase in anti-tumor antibodies in some patients, they did not appreciate any increase in T-cell mediated antitumor immunity. This may be explained by their findings that their patients had impaired immunocompetency prior to the start of the trial. Despite this, they reported clinical responses in three patients during the course of their study.

Table 1

Summary of Phase I & II clinical trials of DC vaccination for CNS tumors.

De Vleeschouwer et al. explored the possibility of assessing immunotherapeutic progress through the use of magnetic resonance imaging (MRI) and methionine positron emission tomography (MET-PET) [17]. By monitoring contrast enhancement changes in relation to metabolic uptake ratios they could postulate at which point an immune response had occurred. This group published the findings from their Phase I clinical trial of 12 recurrent malignant glioma patients that were vaccinated with tumor lysate-pulsed dendritic cells [16]. Interestingly, they were the first to induce DC maturation ex vivo for glioma immunotherapy based on recent evidence arguing that the injection of mature DCs may mediate a more potent anti-tumor response [41,42]. The extent of resection was stressed in this study as prolonged disease free survival was only achieved in two patients who underwent gross total resection (GTR) prior to vaccination. Moreover, one patient who received only partial tumor resection suffered Grade IV neurotoxicity (National Cancer Institute common toxicity criteria) secondary to vaccination-induced peri-tumoral edema. As such, they argue that maximal resection may help avoid such dangerous complications during CNS immunotherapy. Akin to our conclusions [33], this study further champions the need for maximal resection to improve the potential efficacy of vaccination strategies for malignant gliomas.

In a Phase I/II study of tumor lysate-loaded DC vaccination for malignant glioma, subcutaneous injections of DCs loaded with tumor lysate were administered biweekly for a total of three injections [43]. Elevated IFN-γ mRNA levels in PBMCs, positive cytotoxicity assays, increased peripheral CD8+ CTLs, and increased infiltration of CD45RO+ memory and CD8+ T cells in progressive tumor corroborated a positive immune response. Additionally, this study reported an increased median survival in patients receiving vaccinations (133 weeks) compared to historical controls (30 weeks), further substantiating the viability of DC immunotherapy for glioma.

After their initial Phase I clinical trial, Kikuchi et al. [44] continued their work with human patients through a Phase I/II series modeled after their animal studies involving DC-glioma fusion cell injection with peri-vaccination IL-12 [27]. Fifteen patients were vaccinated intradermally with FCs on a biweekly basis for a total of three injects per course, with IL-12 administration on days 2 and 5 following each injection. Interleukin-12 was given as it had been shown to enhance the antitumor effects of FCs in mouse models. Similarly, they found that treatment efficacy using FC/IL-12 vaccination in human patients was better than FCs alone. Although, they were able to demonstrate cellular anti-tumor immunity in only a few of their patients, they observed much improved clinical outcomes, including four partial responses and one mixed response as determined by imaging. Patients tolerated the treatment regimen well and there were no reported signs of autoimmunity despite the use of systemic IL-12.

Walker et al. investigated the interaction between chemotherapy and dendritic cell vaccines in their Phase I clinical trial [45]. Thirteen patients with malignant glioma were treated with 6 biweekly injections (and every 6 weeks thereafter) of DCs pulsed with autologous irradiated tumor cells. Immunologically, they were able to appreciate an antitumor response by the presence of increased cytotoxic and memory T cells on post-vaccination resected tumor. Of the 8 patients that received adjuvant chemotherapy in addition to immunotherapy, 5 were reported to show objective radiological response to treatment, including one patient who had a complete response. This mirrors findings by Wheeler and colleagues [46] who through retrospective analysis determined that patients who received chemotherapy following DC immunotherapy did better in terms of overall survival and time to recurrence than patients who received either one alone. Although it was previously believed that chemotherapy and immunotherapy were antagonistic forms of treatment [47], this and other studies have added to the accumulating evidence that these two therapies may in fact be synergistic in nature.

In a recent paper, De Vleeschouwer et al. published an update of their work on DC immunotherapy for brain tumors, including 56 patients with recurrent glioblastoma [14]. Patients received intradermal injections of mature DCs pulsed with autologous tumor lysate according to three vaccination schedules that varied in regards to frequency of injections and the presence or absence of tumor lysate boosts (Table 1). In addition, DTH was assessed in 21 patients from which enough tumor material could be removed for appropriate testing. The treatment regimens were well-tolerated with the exception of one patient who developed vaccination-induced Grade IV neurotoxicity as was mentioned in their previous study [16] and two patients who experienced Grade II transient hematotoxicity. Analysis of patient survival and time to progression revealed that gross total resection prior to vaccination was the only independent predictor of progression free survival. Younger age (<35) and GTR were predictive of better overall survival, however, only in univariable analyses. Although it did not reach statistical significance, the regimen that included frequent vaccinations with tumor lysate boosting seemed to have improved PFS. Interestingly, DTH reactivity was not shown to have any correlation with clinical outcome.

Recently, Wheeler et al. reported on their Phase II trial in which they treated 34 patients with new or recurrent glioblastoma [48]. Patients received a total of 4 subcutaneous injections of autologous tumor lysate-pulsed DCs on weeks 0, 2, 4 and 10. Primary outcomes of interest were time to progression and time to survival (TTS). Immunological responses were quantified through measuring the differential expression of IFN-γ mRNA in lysate pulsed DCs expanded from PBMCs collected before and after vaccination. Using normalized IFN-γ production values as previously reported [49], 17 of the 31 patients tested showed a positive vaccine response (≥1.5 fold expression) after 3 vaccinations (responders). The magnitude of increased IFN-γ expression correlated logarithmically with TTS, however, only in vaccine responders. This finding was striking in that it was the first immunological predictor of immunotherapy outcome to achieve statistical significance, likely due to the large number of vaccine responders in this trial. Clinically, vaccine responders had significantly longer TTS (642 ± 61 days) compared to nonresponders (430 ± 50 days). Moreover, disease free progression in vaccine responders was improved by 4.5 months, with responders and non-responders having TTPs of 308 ± 55 days and 167 ± 22 days, respectively. It should be noted that these trends were not significant in patients with recurrent glioblastoma, only in those with newly diagnosed tumors. Finally, it was found that patients in this study experienced a 186-day to 190-day increase in TTP when the course of DC injections was followed by adjuvant chemotherapy, compared to DC therapy alone. This treatment effect was observed indiscriminately between responders and nonresponders, with differences only appreciable when comparing patients with 5-fold IFN-γ increase and all others. These findings supported recent data suggesting that chemotherapy may possibly potentiate the clinical effects of DC-based immunotherapy [46,47].

Current Status of DC Vaccines for Brain Tumors

Conclusion

Over the past decade, dendritic cell-based immunotherapy for CNS tumors has progressed from pre-clinical rodent models and safety assessments to Phase I/II clinical trials in over 200 patients, which have produced measurable immunological responses and some prolonged survival rates. However, many questions regarding the methods and molecular mechanisms behind this new treatment option remain unanswered. Results from currently ongoing and future studies will help to elucidate which dendritic cell preparations, treatment protocols, and adjuvant therapeutic regimens will optimize the efficacy of DC vaccination. Additionally, it will be important to characterize the pathways underlying the immunosuppressive microenvironment of brain tumors that currently hinder anti-tumor responses. Combined with further advances in the manipulation of various lymphocyte subsets such as regulatory T cells and NK-like T cells, in addition to the usual armament of CD4+ & CD8+ T cells, understanding these immunologic intricacies will help maximize the cellular efficiency of immunotherapeutic techniques. As clinical studies continue to report results on DC-mediated immunotherapy, it will be critical to continue refining treatment methods and developing new ways to augment this promising form of glioma treatment.