Immunotherapy of Diffuse Gliomas: Biological Background, Current Status and Future Developments

Abstract

Despite aggressive multimodal treatment approaches, the prognosis for patients with diffuse gliomas remains disappointing. Glioma cells often extensively infiltrate in the surrounding brain parenchyma, a phenomenon that helps them to escape surgical removal, radiation exposure and chemotherapy. Moreover, conventional therapy is often associated with considerable local and systemic side effects. Therefore, the development of novel therapeutic approaches is essential to improve the outcome of these patients. Immunotherapy offers the opportunity to specifically target residual radio—and chemoresistant tumor cells without damaging healthy neighboring brain tissue. Significant progress has been made in recent years both in understanding the mechanisms of immune regulation in the central nervous system (CNS) as well as tumor‐induced and host‐mediated immunosuppression elicited by gliomas. In this review, after discussing the special requirements needed for the initiation and control of immune responses in the CNS, we focus on immunological phenomena observed in glioma patients, discuss different immunological approaches to attack glioma‐associated target structures and touch on further strategies to improve the efficacy of immunotherapy of gliomas.

INTRODUCTION

Gliomas are the most common primary brain tumors in adults. They are categorized according to the World Health Organization (WHO) classification, most recently revised in 2007 (162). Most gliomas, the “diffuse gliomas,” are characterized by extensive infiltration of brain parenchyma. The diffuse gliomas can be subtyped as astrocytic, oligodendroglial and oligoastrocytic tumors, with a malignancy grade (low grade = WHO grade II; malignant = WHO grade III; highly malignant = WHO grade IV) assigned to each tumor (162). The most aggressive, that is WHO grade IV diffuse glioma, is glioblastoma (GBM). Although GBMs are generally considered purely astrocytic in nature, they in fact form a heterogeneous group of tumors, representing 15%–20% of all intracranial neoplasms and approximately 50% of the malignant gliomas. As most studies on the immunobiology and immunotherapy of gliomas were carried out on diffuse gliomas, especially GBMs, the present review focuses on these tumors as well. Currently the standard of care for patients with GBM after maximal tumor resection is concomitant chemoradiotherapy followed by adjuvant treatment with temozolomide (TMZ). The median overall survival for patients with GBM is 14.6 months and the overall survival rate is 27.2% at 2 years. After 3 years, nearly all patients have progressed, with mortality well over 90% after 5 years 249, 250. Factors associated with an increased risk of death are increased age, lower Karnofsky performance score (KPS), corticosteroid use, shorter time from original diagnosis to recurrence, tumor outside the frontal lobe, unmethylated O6‐methylguanine methyltransferase (MGMT)‐promotor status and the presence of tumor‐initiating glioma stem cells 40, 100, 196, 299. Overall, the prognosis for patients with GBM remains poor. It is thus essential to consider novel treatments in the hope of attacking residual radio‐ and chemoresistant tumor cells and to improve the survival of glioma patients. With increasing knowledge of central nervous system (CNS) immunity, immunotherapeutic approaches are promising options for the treatment of diffuse gliomas.

CENTRAL ROLE OF DENDRITIC CELLS (DCs) IN THE ACTIVATION OF THE IMMUNE SYSTEM

In general, the immune system is based on two distinct types of responses, the innate and the adaptive immune response. The innate immune system functions as a first line of defense against invading microorganisms. It discriminates between the different microbes by recognizing a set of conserved molecular structures, so called pathogen‐associated molecular patterns (PAMPs), shared by large groups of microorganisms. Cells that orchestrate the innate immune response are DCs, macrophages, monocytes, granulocytes and natural killer (NK) cells. The specific, adaptive immune system forms the second line of defense, which depends on the activity of effector T and B cells. It uniquely distinguishes distinct microorganisms and confers immunological memory. The specifity is mediated by membrane receptors bearing a great diversity of antigen binding sites.

DC subsets

DCs are the professional antigen‐presenting cells (APCs) of the immune system that instruct and control primary immune responses. Two major subtypes of DCs have been defined, myeloid and plasmacytoid DCs. Both DC subsets differ in morphology, tissue distribution, expression of markers and function (270). Plasmacytoid DCs are circulating cells with a plasmacytoid morphology that are capable of producing large amounts of type I interferons upon activation by microbial stimuli (242). In addition, they can differentiate into DCs that are capable of activating naive T cells against allo‐antigens (91) and exogenous antigens (22). Myeloid DCs can be further divided into migratory DCs, which serve as sentinels of the immune system in peripheral tissues and present foreign antigens to T cells after migration to the draining lymph nodes, and lymphoid‐tissue‐resident DCs that capture local (foreign and self‐) antigens and present them to local T cells (238).

Migratory DCs

At an immature stage, myeloid DCs take up tissue antigens, both soluble and particulate antigens, very efficiently. After challenge with microbial or inflammatory (“danger”‐associated) stimuli, DCs undergo a process of activation, also known as maturation. Several environmental stimuli, such as signals from pattern‐recognition receptors (PPR), inflammatory cytokines and members of the TNF superfamily (eg, CD40L, RANKL) have been shown to mediate DC activation (246). During DC maturation, captured antigens are processed and bound to both classical major histocompatibility complex (MHC) class I and class II molecules that are translocated to the cell surface, accompanied by an increased expression of costimulatory molecules and the secretion of cytokines that influence T cell proliferation and differentiation. In response to the appropriate stimuli, DCs also upregulate CC chemokine receptor‐7 (CCR7), a receptor that drives DC migration to the lymphoid vessels and T cell areas of secondary lymphoid organs (13). There, DCs are mature and are well‐equipped to attract and activate naive CD4+ and CD8+ T cells (3) In addition, DCs are also able to directly activate innate lymphocytes such as NK cells or NKT cells via non‐classical (CD1 family) MHC molecules presenting glycolipid structures 68, 72, thus providing a link between the adaptive and innate immune system (Figure 1).

Figure 1.

Central role of dendritic cells (DCs) in the initiation of an immune response. DCs are present in virtually all tissues and organs including the brain and continuously monitor their environment for the presence of danger, for example, invading microorganisms and tissue damage. Immature DCs are very efficient in antigen uptake, mediated by high endocytotic activity and expression of an array of cell surface receptors. Upon recognition of danger signals, DCs undergo maturation and migrate via lymph vessel (LV) to the draining lymphoid organs. Here, mature DCs interact with and activate naive lymphocytes specific for their cognate antigen and initiate a primary immune response. After expansion, activated T effector cells (Teffs) and natural killer/natural killer T cells (NK/NKT cells) exit the lymphoid tissue and home back via the blood stream to the site of antigen deposit to eliminate the threat in the endangered tissue (eg, brain). Under steady state conditions immature, non‐activated DCs contribute to immune homeostasis by deleting autoreactive T cells and inducing T‐cell tolerance, or leading to the generation of regulatory T cells (Tregs).

PPRs expressed by DCs

Four main PPRs expressed by DCs can be distinguished. These include Toll‐like receptors (TLRs), C‐type lectins, NOD‐like receptors and cytoplasmic RNA helicases 149, 268.

So far, the best‐characterized PPRs are the TLRs 128, 154. The TLR family in humans consists of 11 transmembrane members that are located on the cell surface (TLR1, 2, 4, 5, 6, 10 and 11) or within the endosomal compartments (TLR3, 7, 8 and 9) (Figure 2). Many specific ligands for TLRs have now been identified. TLR2 recognizes bacterial lipoproteins, peptidoglycan and lipoteichoic acid from Gram‐positive bacteria, and zymosam from fungi, and can associate with TLR1 and TLR6 for functional responses to mycobacterial lipopeptides. TLR3 recognizes viral double‐stranded (ds) RNA and synthetic dsRNAs, such as polyinosinic–polycytidylic acid (Poly I:C). TLR4 binds lipopolysaccharides (LPS) from Gram‐negative bacteria and viral envelope proteins, whereas TLR5 recognizes flagellin. TLR7 and TLR8 recognize viral single stranded RNA and synthetic molecules like imidazoquinoline or its derivates. TLR9 recognizes unmethylated CpG motifs within bacterial and viral DNA. The specific ligand of TLR10 is currently unknown. The recently discovered TLR11 recognizes uropathogenic bacteria 117, 254. TLR expression by DCs is dependent on the cellular subset and differentiation state. Myeloid DCs express virtually all TLRs with the exception of TLR9, which is selectively expressed by plasmacytoid DCs. In contrast to humans, TLR9 is expressed both by plasmacytoid and myeloid DCs in mice (212). Importantly, DC activation and cytokine production are strongly dependent on the type and combination of the TLRs that are triggered 121, 183. Consequently, the nature of specific T‐cell responses generated also depends on the different stimuli that a DC encounters locally (244).

Figure 2.

Toll‐like receptor (TLR)‐mediated recognition of pathogen‐associated molecular patterns (PAMPs) and signaling pathways. TLR1, 2, 4, 5 and 6 are expressed on the surface of dendritic cells, and are involved in detecting lipids, carbohydrates and protein ligands from extracellular pathogens. In contrast, TLR3, 7, 8 and 9 are mainly expressed in intracellular vesicles and function within the acidified endosomes to detect nucleic acids derived from viruses and bacteria. TLR2 recognizes bacterial lipoproteins, peptidoglycan and lipoteichoic acid from Gram‐positive bacteria, and zymosam from fungi, and can associate with TLR1 and TLR6 for functional responses to mycobacterial lipopeptides. TLR3 recognizes viral double‐stranded (ds) RNA and synthetic dsRNAs, such as polyinosinic‐polycytidylic acid (Poly I:C). TLR4 binds lipopolysaccharides (LPS) from Gram‐negative bacteria and viral envelope proteins, whereas TLR5 recognizes flagellin. TLR7 and TLR8 recognize viral single‐stranded (ss) RNA and synthetic molecules like imidazoquinoline or its derivates. TLR9 recognizes unmethylated CpG motifs within bacterial and viral DNA. TLR3 uses Toll/Interleukin‐1 receptor (TIR) domain‐containing adapter‐inducing interferon (TRIF), whereas all other TLRs use MyD88 as an adapter molecule. Both TRIF and MyD88‐dependent pathways lead to the activation of MAPK kinases such as p38 and the activation of nuclear factor‐κB (NF κB) and interferon (IFN)‐regulatory factors (IRFs) to induce type I IFN and inflammatory cytokines, respectively.

T‐cell polarization

Inside the lymph nodes, antigen‐loaded activated DCs interact with naive T cells and induce their differentiation into T effector cells. It is a unique feature of DCs to efficiently present captured antigens on MHC class‐II molecules to CD4+ T cells. Naive CD4+ T cells can be differentiated into either T‐helper type 1 (Th1), T‐helper type 2 (Th2) or TH17 cells. Th1 cells develop in the presence of interleukin 12 (IL‐12) and secrete IFN‐γ, whereas Th2 cells develop in the presence of IL‐4 and secrete IL‐4, IL‐5 and IL‐13 191, 256. For the development of TH17 cells from naive precursors a combination of IL‐6 and TGF‐β is required (23). Moreover, IL‐1 and IL‐21 seem to play important roles in promoting the induction of TH17 cells, whereas IL‐23 is vital for the expansion and survival of this population 24, 143, 148. In contrast to other APCs, DCs are also capable of presenting captured antigens on MHC class‐I molecules to CD8+ T cells, a process termed “cross‐presentation”5, 132. In vivo priming of CD8+ cytotoxic T lymphocytes (CTLs) generally requires the participation of CD4+ T‐helper cells, CD40‐CD40L interaction, innate lymphocytes or stimulation via certain TLRs 121, 160, 233. Besides effector T cells, DCs are also efficient at activating naturally occurring CD4+ regulatory T cells (Tregs) 14, 130, 295. In addition, DCs are able to induce several subsets of Tregs capable of controlling effector T cells, including induced Tregs, type 1 regulatory T cells (Tr1 cells) and Th3 cells. The regulatory T‐cell‐polarizing factors are IL‐10 and transforming growth factor‐β (TGF‐β) 44, 65, 273 (Figure 3).

Figure 3.

Dendritic cell (DC)‐mediated T‐cell polarization. After activation of Toll‐like receptors (TLRs) by pathogens (in green) DCs increase cell‐surface expression of MHC‐peptide complexes, upregulate costimulatory molecules (eg, CD86) and secrete immunomodulatory cytokines that direct T‐cell polarization. Ligation of distinct TLRs trigger differential cytokine production in DCs. Production of interleukin‐4 (IL‐4) promotes the development of T helper type 2 (Th2) cells, whereas the production of IL‐12 favors the development of Th1 cells. Th2 cells secrete IL‐4, IL‐5, IL‐10 and IL‐13, whereas Th1 cells secrete IL‐2 and IFN‐γ. IL‐4 and IL‐12 enhance the generation of their own Th‐subset and simultaneously inhibit the generation of the opposing subset. A combination of IL‐1 or IL‐6 and TGF‐s contributes to the development of Th17 cells from naive precursors, whereas TGF‐s alone promotes the development of regulatory T cells (Tregs). Both IL‐4 and IFN‐γ inhibit Th17 development. IL‐6 inhibits the development of Tregs. Black arrows indicate promoting activity, whereas red lines indicate inhibitory activity.

Naturally occurring CD4+ regulatory T cells

Naturally occurring CD4+ Tregs represent 5%–10% of the overall CD4+ T cell population and are generated in the thymus (223). They constitutively express the IL‐2 receptor α‐chain (CD25), cytotoxic lymphocyte‐associated antigen‐4 (CTLA‐4) and different members of the tumor necrosis factor (TNF) receptor superfamily such as the glucocorticoid‐induced TNF related protein (GITR) or Ox40 237, 265. Other candidate markers of Tregs have been described and currently include CD62L, CD103, CD122 and CD223 (178). So far the most specific marker for naturally occurring CD4+ Tregs is FoxP3, a member of the forkhead family of DNA‐binding transcription factors, which has been shown to be expressed specifically in mouse CD4+ Tregs and acts as a master switch in the regulation of their development and function 71, 105. Per se, Tregs are hyporesponsive to T‐cell receptor (TCR) stimulation in vitro, but high amounts of IL‐2 or stimuli bypassing the TCR can overcome their anergic state (224). After polyclonal or antigen‐specific TCR stimulation, Tregs potently suppress the proliferation and cytokine production of effector CD4+ and CD8+ T cells by inhibiting IL‐2 gene transcription. Once activated in an antigen‐specific manner, Tregs also suppress effector T cells, which are specific for other antigens. Furthermore, Tregs have been shown to inhibit the proliferation and antibody production of B cells and to profoundly suppress NK‐cell function 33, 77. Tregs show their suppressive effects by a complex and overlapping set of mechanisms (225) (Figure 4).

Figure 4.

Regulatory T cells (Tregs) and immune cell suppression. Tregs are characterized by a set of markers (in blue), suppress different types of immune cells (red lines) and show their suppressive effects by a number of different mechanisms (orange box). Tregs have cytolytic activity and can kill target cells directly in a granzyme/perforin‐dependent manner or induce cytokine deprivation‐mediated apoptosis. Two ectoenzymes, CD39 and CD73, expressed on activated Tregs, induce the extracellular release of adenosine nucleosides that suppress effector T‐cell functions. Tregs also promote the transfer of the potent inhibitory second messenger cyclic adenosine monophosphate (cAMP) to effector T cells. In addition, Tregs can condition DCs, via a CTLA‐4:CD80/CD86‐dependent mechanism, to express indoleamine 2,3‐dioxygenase (IDO). Expression of the lymphocyte activation gene 3 (LAG3) on Tregs can further modulate the maturation and immunostimulatory capacity of DCs. Finally, potent inhibitory cytokines, produced by Tregs, such as membrane‐bound or soluble TGF‐s (mTGF‐s; sTGF‐s), IL‐10 or IL‐35, contribute to their suppressive function.

THE CNS—AN IMMUNOLOGICALLY SPECIALIZED SITE

In the past, the CNS has been described as an immunologically privileged site based on the presence of the blood–brain barrier (BBB), graft acceptance, lack of conventional lymphatics, low MHC expression and low T cell trafficking 41, 50. However, with recent advances in understanding CNS immunity it is more accurate to regard the CNS as an immunologically specialized site that can be divided into three different compartments: the brain parenchyma; the ventricles containing choroid plexus and cerebrospinal fluid; and the meninges. The immune reactivity of the ventricles and meninges is similar to that of the periphery, whereas immune responses are delayed or even abrogated within the brain parenchyma 17, 248, 280.

Initiation of an immune response in the CNS

Accumulating data suggest that DCs play a central role in the initiation of immune responses in the CNS 135, 198. DCs can be found in low numbers in the meninges, perivascular spaces, the choroid plexus and in the CSF where they reside in an immature state 169, 172, 197.

Under non‐inflammatory conditions, DCs present antigens derived from tissue debris and apoptotic cells in the absence of costimulatory molecules. Such antigen presentation by non‐activated DCs induces T‐cell anergy, T‐cell tolerance or the induction of Tregs 159, 236, 247. Moreover, it has been reported that non‐stimulated, brain‐derived DCs preferentially target B‐cell areas instead of T‐cell areas, again skewing the immune system towards a humoral response (96). Similarly, drainage of brain‐derived antigens was shown not to be sufficient to induce a Th1‐type and CTL response by itself, but leads to tolerance induction or skewing of the immune response towards a B cell and Th2‐type response 94, 95, 283. However, under a variety of inflammatory conditions, DCs are quickly recruited to the site of the brain lesion and appear in both the brain parenchyma and the cerebrospinal fluid in high numbers 69, 157, 235. Importantly, it was recently demonstrated that these DCs do not differentiate from CNS‐resident precursors like microglia cells, but are of peripheral origin (175). At the site of the lesion, inflammatory mediators and danger signals promote maturation and rerouting of DCs to the secondary lymphoid organs where they prime antigen‐specific immune responses.

Although conventional lymphatics are missing in the CNS, antigen‐loaded DCs have access to secondary lymphoid organs through several pathways. The most prominent one is the drainage of antigen‐loaded DCs along the subarachnoid space surrounding the olfactory bulbs through discrete channels in the cribriform plate into the lymphatics of the nasal submucosa and, subsequently, to the cervical lymph nodes. Regional lymph nodes that are associated with other cranial nerves might also be involved in the CNS drainage. With access to the cerebrospinal fluid and the perivascular spaces, antigen‐loaded DCs can also migrate out through the arachnoid villi into the venous blood 103, 208, 282.

These data further support recent analyses demonstrating that brain‐derived antigens can reach the cervical lymph nodes very quickly, but similar to the skin system, a second wave of incoming brain‐derived APCs, most likely DCs, is necessary for optimal T‐cell activation 80, 119, 135. After clonal expansion, activated antigen‐specific T cells preferentially home back to the site of antigen deposit in the brain 34, 35.

T cell trafficking into the CNS

In general, the traffic of leukocytes into the CNS is a highly regulated process. Resting T cells fail to enter the CNS, whereas activated T cells are able to penetrate the BBB under inflammatory conditions 279, 281. For CD4+ T cells, the activation‐stage rather than the antigen‐specificity determines BBB crossing (102). CNS recruitment of CD8+ T cells is regulated differently than CD4+ T cells. Recent data show that antigen specificity is a factor that governs CD8+ T cell infiltration into the brain and that this process is dependent on luminal expression of MHC class I by cerebral endothelium, but independent of antigen presentation by perivascular APCs (75).

The current paradigm of blood leukocyte transendothelial migration involves a sequential, multistep adhesion cascade between leukocyte and endothelial cell adhesion molecules 19, 63. T‐cell priming in CNS‐draining lymph nodes is usually associated with a rapid upregulation of α4 and β1 integrin (35). These α4β1 integrins have been shown to interact with VCAM‐1 on cerebral vascular endothelium and this interaction has been reported to facilitate CNS entry of encephalitogenic or virus‐specific T cells 39, 264. An important function for ALCAM in the recruitment of CD4+ T cells and monocytes across the BBB endothelium has recently been demonstrated (43). After CNS entry, antigen‐specific reactivation of effector T cells by perivascular APCs is needed for T‐cell persistence and amplification of T‐cell‐mediated immune responses in the CNS. In general, those T cells that have encountered their target antigen within the brain are retained for longer periods within the CNS (168).

After contact with perivascular APCs, effector T cells undergo rapid reactivation with renewed induction of activation markers such as CD25 and CD134 (OX‐40) 49, 90, 136. Moreover, CD8+ T cells further differentiate in the brain, showing enhanced IFN‐γ and granzyme B expression and induction of αEβ7 integrin, mediating increased adhesion to the brain parenchyma (168). Finally, effector T cells migrate to the site of the brain lesion towards a chemotactic gradient. Chemokines such as monocyte chemoattractant protein (MCP)‐1 also induce the expression of matrix metalloproteinases and other ectoenzymes that enable penetration of effector T cells into the brain parenchyma 6, 11.

Perivascular macrophages and parenchymal microglial cells

Next to the DCs, two other major populations of phagocytic cells of myeloid origin function as APCs in the CNS: perivascular macrophages and parenchymal microglia cells. Based on their morphology and immunophenotype, perivascular macrophages appear to be very similar to blood‐derived macrophages (228). Moreover, it has been shown that brain perivascular cells contain a population of replaceable, rather than resident phagocytic cells (18).

Microglial cells are the primary immunocompetent cells in the brain. In addition to their phagocytic function, they may also participate in the regulation of innate and adaptive immune responses. They show a very low turnover rate and remain in an undifferentiated state with little expression of MHC class II under steady‐state conditions. Upon activation, microglial cells can quickly upregulate their antigen presenting capabilities in response to inflammatory or microbial stimuli both in vitro and in vivo; they can also develop into more DC‐like cells 70, 195, 227. However, even upon full activation, the expression levels of MHC and costimulatory molecules on DC‐like microglial cells are much lower than generally found on DCs. Furthermore, CNS‐resident microglial cells purified from the inflamed CNS were found to be largely incapable of activating either naive or effector T cells (CD4+ T helper cells), in contrast to peripherally derived myeloid DCs 171, 175. In addition, there are currently no data available on the ability of microglial cells to migrate out of the CNS in significant numbers and present antigens in the draining lymph nodes. Instead of T‐cell priming, both cell types might play a direct role in reshaping the function CD8+ effector T cells or a more indirect role in the conditioning of the DC precursors to differentiate into DCs 136, 168.

Control of immune responses in the CNS

The time of onset, the intensity and duration of immune responses within the CNS have to be tightly controlled; otherwise permanent tissue damage can occur. Immunosuppressive Tregs have been shown to be essential for the control of immune pathology. It has been reported that, similar to effector T cells, Tregs traffic to the inflamed site within the CNS, become reactivated, and further expand in situ. Interestingly, it has been shown that Tregs are not functional at the disease peak of experimental autoimmune encephalomyelitis (EAE). However, in the recovery phase, Tregs regain suppressor activity and potently curb the autoimmune responses locally 144, 190. Moreover, it was recently demonstrated that CNS DCs show a dual role during the disease course of an EAE: these cells promote CNS inflammation at the onset of the disease, whereas by the peak of the disease, they lose their optimal T‐cell stimulatory capacity and support Treg‐mediated immunosuppression to limit the extent of CNS inflammation (58).

INTERACTION OF MALIGNANT GLIOMA AND THE IMMUNE SYSTEM

Immune cell infiltrates

The immune recognition of glial tumors arising within the brain is usually prevented at the earliest stages of neoplastic transformation where the tumor cells are totally contained within the normal brain parenchyma. As such, primary brain tumors present a unique challenge for CNS immunosurveillance. However, at later stages of tumor growth, when massive tissue damage with destruction of the BBB and tumor cell necrosis with antigen drainage to the periphery occurs, brain tumors become accessible to the peripheral immune system. In this case, albeit at varying degrees, immune cell infiltrates can regularly be found within malignant gliomas. The vast proportion of immune cells, mainly macrophages, microglial cells, DCs and T cells, is found at perivascular regions, but can also be observed throughout the tumor bed and within necrotic tissue regions 110, 199, 216, 272. Attraction of these immune cells is directed by a number of molecules, which are mainly produced by oxygen‐deprived tumor cells surrounding necrosis. Hypoxia inducible factor‐1α (HIF‐1α) has been identified as a key factor in gliomas promoting the release of chemoattractants, such as CXCL5, CXCL8/IL‐8 and CXCL12/SDF‐1 57, 253. Other chemoattractants, such as CCL2/MCP‐1, CCL3/MIP1‐α, CCL4/MIP1‐β, CCL22/MDC also contribute to the recruitment of immune cells to gliomas 56, 118, 252. However, early studies have already determined that the presence of immune cell infiltrates usually does not correlate with the clinical course of glioma patients or may even be a negative prognostic factor 217, 220, 221.

Immunosuppression elicited by malignant gliomas

Compelling evidence exists that immune activation within malignant gliomas is suppressed by the local tumor microenvironment 61, 274, 275 (Figure 5). Malignant gliomas are characterized by the presence of a large number of immunosuppressive soluble factors, such as vascular endothelial growth factor (VEGF), IL‐10, TGF‐β, whereas immunostimulatory molecules, such as IL‐12, IL‐18, INF‐γ, are lacking. In addition, immunosuppressive enzymes such as indoleamin 2,3‐dioxygenase (IDO), cyclooxygenase‐2 (COX‐2), arginase and nitric‐oxide synthase‐2 (NOS‐2) are overexpressed in glioma cells 28, 32, 42, 48, 83, 185. Moreover, many immunoinhibitory molecules are abundantly expressed on the surface of glioma cells. Non‐classical MHC ligands such as HLA‐G and HLA‐E 286, 290, co‐inhibitory molecules such as PD‐L1 (B7‐H1) 288, 289, minor brain gangliosides such as GM2 and GM3 (88), and galectin‐1 and galectin‐3, mammalian lectins with specificity to beta‐galactosides have been detected 74, 218. Furthermore, inhibitory cytokine signaling molecules such as signal transducers and activators of transcription 3 (Stat3) are known to be constitutively activated in several human glioma cell lines, promoting tumor cell growth and survival 2, 142, 207.

Figure 5.

Immune escape mechanisms of diffuse gliomas. Once antigen‐specific effector T cells (Teffs) reach the central nervous system (CNS) parenchyma through the intact or disrupted blood–brain barrier (BBB) (1), they may face regulation by CNS‐residents cells such as astrocytes and microglial cells that are able to inhibit T‐cell proliferation and to induce T‐cell apoptosis by the secretion of immunosuppressive molecules and the expression of death receptors ligands such as FasL (2). Constitutive immunosuppressive factors in the brain parenchyma such as TGF‐s may also impede T‐cell differentiation or T‐cell effector functions (3). Within the tumor, Teffs are confronted with different subpopulations of immune suppressor cells that are attracted to the tumor site, such as regulatory T cells (Tregs), myeloid‐derived suppressor cells (MDSCs) and mesenchymal stem cells (MSCs). Moreover, tumor‐infiltrating Teffs are exposed to high concentrations of tumor‐derived soluble immunosuppressive factors as well as immunoinhibitory surface molecules expressed on the tumor cells (4). Finally, efficient immune recognition of glial brain tumor cells may be compromised by low MHC molecule expression and inefficient processing and presentation of tumor‐associated antigens (5). Adapted from Walker et al (2002) (274).

As a consequence, differentiation, maturation and function of tumor‐infiltrating DCs and other APCs, as well as the generation and activation of immune effector cells are all profoundly suppressed in the tumor microenvironment. In addition, the cytolytic activity of macrophages, NK cells and cytotoxic T cells is deeply disturbed 81, 111, 179, 239, 241, and other players in tumor‐induced immunosuppression are recruited to the tumor, preventing tumor‐specific immunity. Among those are myeloid‐derived suppressor cells, representing a phenotypically heterogeneous cell population that includes immature myelomonocytic cells, terminally differentiated monocytes and granulocytes 82, 203, 263, but also mesenchymal stem cells with immunosuppressive functions 26, 260. Recent studies clearly show that CD4+ Tregs infiltrate and fbaccumulate within gliomas 62, 125 and profoundly suppress antiglioma immune responses. Also, patients with malignant glioma show an elevated proportion of Tregs in peripheral blood in the setting of an overall diminished CD4+ T‐cell pool (66). This predominance of Tregs was shown to be responsible for the multiple immunological defects that have been previously described in glioma patients, including abnormal delayed hypersensitivity responses, depressed mitogen responsiveness of T and B cells, decreased antibody responses, and impaired T‐cell cytotoxicity (60).

GLIOMA‐ASSOCIATED TARGET STRUCTURES FOR IMMUNOTHERAPY

The ultimate goal of glioma immunotherapy is to induce prominent antitumor immune response leading to the complete eradication of malignant cells without eliciting serious negative side effects. An essential step in the development of immunotherapeutic strategies against malignant gliomas is the identification of suitable target structures. An ideal target for immunotherapy is an antigen that is specifically and stably expressed by the tumor, absent from normal tissues, and crucial for the survival of the cancer cell. Cancer/testis (CT) antigens have the potential to match such requirements. CT antigens are encoded by genes that are normally expressed only in the human germ line, but are aberrantly expressed in various tumor types. So far, more than 40 CT antigen family members have been identified, most of them in melanomas. Of note, these antigens are frequently found in only a relatively small proportion of tumor cells. Interestingly, cancer germline genes (CGGs) are frequently co‐expressed, and tumors that express them tend to express several CT antigens 93, 124.

CT antigens

The knowledge of tumor‐specific target structures for glioma immunotherapy is not far advanced. The fact that glial cells and melanocytes ontogenetically have a common neuroectodermal origin may explain the finding that melanoma‐associated CGGs such as the melanoma antigen‐encoding (MAGE) genes MAGE‐1 and MAGE‐3, MAGE‐E1 or GAGE‐1 151, 229, 230 can also be detected in malignant gliomas. Other CGGs, such as the expression of homo sapiens testis (HOM‐TES)‐14 [also known as stromal cell‐derived protein (SCP)‐1], synovial sarcoma X breakpoint (SSX)‐1, SSX‐2, SSX‐4, HOM‐TES‐85 and Sry‐related high‐mobility group (HMG) box‐containing gene (SOX)‐6 have also been described in malignant gliomas, although their antigenic determinants are not yet known 222, 261. Most studies have analyzed antigen expression at the mRNA level by reverse transcription polymerase chain reaction (RT–PCR). Protein expression or immunogenicity of CT antigens have only rarely been investigated in malignant gliomas (158). Efficient immune recognition of CT antigens on glioma cells may be compromised by low or absent expression of classical MHC molecules and inefficient processing and presentation of such antigens 64, 173.

Overexpressed self‐antigens

Further potential target structures with restricted protein expression include overexpressed self‐antigens, such as mutated versions of the epidermal growth factor receptor (EGFR), especially of variant III (EGFRvIII), the IL‐13 receptor α chain 2 (IL‐13Rα2) or the apoptosis inhibitor protein, survivin 109, 133, 287. Other tumor‐associated targets can be detected in malignant gliomas, such as the squamous cell carcinoma antigen recognized by T cells, SART1 and SART3 114, 180. Furthermore, transferrin receptors that are overexpressed on rapidly dividing cells, most notably on hematopoietic cells and various tumor cells, including GBM cells, are identified as target structures (210). Currently, much attention is drawn to viral antigens unique to human cytomegalovirus, which are expressed within the vast majority of high‐grade gliomas, but not within surrounding normal brain (176).

Differentiation antigens

Additional potential targets include differentiation antigens that are normally only seen at particular phases of cell differentiation, such as tyrosinase related protein (TRP)‐1 and TRP‐2 or gp100, but also antigen isolated from immunoselected melanoma (AIM)‐2, ADP ribosylation factor 4‐like protein ARF4L, and betaGlcNAc beta1, 3‐galactosyltransferase, polypeptide 3 (GALT3). However, as these antigens often demonstrate a broader tissue expression, their value as targets for antiglioma therapy might be limited 45, 158, 187, 230, 257.

Stromal antigens

Target structures within the tumor stroma have also been identified. One of the best examples is tenascin, a polymorphic glycoprotein of the extracellular matrix, which is expressed in about 80% of malignant gliomas, especially in GBM (20).

GLIOMA IMMUNOTHERAPY

For treatment strategies of gliomas, passive, adoptive and active immunotherapeutic approaches have been evaluated.

Passive immunotherapy

To target glioma‐specific structures, different monoclonal antibodies (mAbs) have been designed that are coupled to radionucleotides (radioimmunoconjugates) or exotoxins (immunotoxins) and are administrated locally. In some studies, a survival benefit in glioma patients has been documented. However, a disadvantage of this approach is that the penetration of these molecules into the tumor tissue is quite limited, because of high interstitial pressures in the tumor and surrounding tissue. High‐flow convention enhanced methods are needed to efficiently deliver these molecules to the more peripheral, diffusely infiltrative portions. Moreover, the effectiveness of the therapy seems to be restricted to those patients with a low tumor burden 29, 150, 277.

Adoptive immunotherapy

Alternatively, the adoptive transfer of ex vivo generated and activated effector cells into the resection cavity of patients with malignant gliomas has been extensively evaluated. Treatment approaches have differed in the types of cells administered, the route of administration, and the activation status of the cells. These approaches offer the advantage of circumventing deficits of antitumor effector cell populations in the host by providing optimal conditions for the culture and amplification of these cells in vitro, in the absence of tumor‐derived immunosuppressive factors. However, despite good tolerability, clinical efficacy has rarely been seen with the local administration of lymphokine‐activated killer cells (LAK‐cells) 27, 59, 126, mitogen‐activated killer cells (MAKs) 115, 129, NK cells, tumor‐infiltrating lymphocytes (TILs) and allogeneic or autologous, tumor‐specific cytototoxic T cells 139, 147, 200, 206, 291. Proinflammatory cytokines, such as IL‐2 have been coadministered locally to enhance antitumor responses, but has induced significant toxicity, particularly in patients with large tumor loads 16, 98.

Active immunotherapy

Active immunization of patients with malignant gliomas is regarded as a powerful strategy to induce potent antitumor response in vivo. As a vaccine, autologous inactivated tumor cells have been used, either gene‐modified or applied together with cytokine‐secreting fibroblasts 193, 245. Alternatively, tumor‐specific peptide vaccines targeting EGFRvIII have been administrated intradermally along with granulocyte macrophage colony‐stimulating factor (GM‐CSF) as an adjuvans (226). One of the most promising immunotherapeutic approaches to amplify tumor‐specific T‐cell responses is to use ex vivo generated, autologous tumor antigen‐loaded DCs as a vaccine (79). Several phase I/II studies have been carried out with this approach, and although these clinical trials differed in terms of DC generation, loading of tumor antigens and the route of application, robust intratumoral cytotoxic and memory T‐cell infiltration could be detected in patients who underwent resurgery after vaccination. DC vaccinations appeared to be mainly beneficial for patients with younger age, minimal residual tumor burden and expression of low levels of TGF‐β2 54, 156, 219, 292, 294, 296, 297. The principles and results of both peptide‐based vaccination and DC therapy against malignant gliomas are discussed in more detail in the companion articles of van Gool et al (267) and Choi et al (46).

ENHANCEMENT OF VACCINE IMMUNOGENICITY

Data from phase I/II studies support the view that strong glioma‐specific immune responses can be induced by a DC vaccine, but that therapy efficacy is still impeded by rapid tumor progression and a strong local immunosuppressive environment. To improve glioma immunotherapy, higher numbers and more potent glioma‐specific effector cells should be induced that home in to the CNS glioma site, including the peripheral, diffusely infiltrative areas (Figure 6).

Figure 6.

Principles of dendritic cell (DC) vaccination and further strategies to improve immunotherapy of diffuse gliomas. The most common method used to generate ex vivo DCs for vaccination is to culture autologous monocytes obtained from leukapheresis in the presence of granulocyte monocyte colony‐stimulating factor (GM‐CSF) and IL‐4. Following 5–7 days in culture, the monocytes differentiate into immature DCs. DC maturation is induced by culturing immature DCs for an additional 24–48 hours in the presence of a maturation cocktail, most commonly consisting of TNF‐α, IL‐1β, IL‐6 and PGE2, also known as monocyte‐conditioned medium. Tumor antigen loading with tumor lysate, tumor RNA or synthetic tumor‐specific peptides occurs at either the immature or mature DC stage. Finally, mature tumor antigen‐loaded DCs are injected back into patients. Critical issues to enhance the persistence of the vaccine‐induced immune responses for future immunotherapeutic approaches against diffuse gliomas are to improve the immunogenicity of a DC vaccine and to break the local immunosuppressive environment of diffuse gliomas. Toll‐like receptor (TLR) agonists, activation of costimulatory receptors, blockade of co‐inhibitory receptors and/or depletion or modulation of regulatory T cells (Treg) could all be used to intensify antiglioma immune responses induced by peripheral DC vaccinations. In addition, immunosuppressive molecules could be targeted directly within the tumor microenvironment to enhance treatment efficacy. Peri‐ or intratumoral administration of distinct TLR agonists might serve as an additional option to activate local immune cells and to potentiate antitumor immunity. Abbreviation: BBB = blood–brain barrier; Teff = T effector cells.

DC maturation

When using ex vivo generated DCs as a vaccine, the induction of effective immune responses is dependent upon proper DC maturation. The most widely used maturation cocktail in clinical DC trials consists of four reagents: TNF‐α; IL‐1β; IL‐6; and PGE2, also known as monocyte‐conditioned medium. A major disadvantage of this maturation cocktail is that the resulting DCs secrete little IL‐12, most likely as a result of the presence of PGE2 present during maturation 134, 213. Moreover, cytokine cocktail‐matured DCs were found to be more effective than immature DCs in expanding a population of immunosuppressive Tregs 55, 259.

An excellent method to enhance the immunogenicity of DCs is to include immune response modifiers, such as TLR agonists into the maturation cocktail for optimal DC activation. Several groups reported that DCs matured with more extensive maturation cocktails containing TLR‐agonists such as Poly I:C (TLR3) or R848 (TLR7/8) and cytokines such as TNF‐α, IL‐1β and INF‐γ show superior immunogenicity to standard cocktail matured DCs; this strategy enhanced T‐cell and NK‐cell responses, while decreasing the chance to induce Tregs. Addition of TLR‐agonists to the maturation cocktail also enabled the use of sufficiently low concentrations of PGE2 that the capacity of DCs to produce IL‐12 was not disturbed. Supplementation with PGE2 was essential to improve DC handling by reducing cell adherence and enhancing migratory capacity via upregulation of CCR7. Moreover, TLR triggered DCs retained a stable phenotype and IL‐12 producing capacity, even when exposed to TGF‐β, Th2 or Treg promoting environments 30, 73, 84, 166, 302. At present, a phase I/II‐study (NCT00766753) using TLR‐polarized DCs (type‐1 αDCs) loaded with glioma‐associated tumor antigens is being carried out in patients with recurrent malignant gliomas.

The importance of IL‐12 for an effective immunotherapy against gliomas has been demonstrated in several animal and human studies. Coadministration of IL‐12 with a DC vaccine was reported to enhance antitumor responses resulting in prolonged survival 116, 137, 293. In addition, IL‐12 was shown to directly inhibit Treg differentiation from naive T cells (278). Collectively, these data imply that mature DCs with retained capacity to produce IL‐12 are the most promising candidates for glioma immunotherapy and point towards an important role for TLR‐agonists in inducing potent anti‐glioma responses.

In addition, TLR‐agonists can be used as immune adjuvans. Pretreatment of the injection site with proinflammatory compounds or TLR‐agonists such as imiquimode (TLR7) have been shown to significantly enhance both the persistence and trafficking of DCs into the draining lymph nodes after vaccination, resulting in increased priming of tumor‐specific T cells 167, 181, 204.

Discovery of major glioma rejection antigens

To create a powerful DC vaccine, it is not only important to use proper maturation stimuli, but also to efficiently load DCs with relevant tumor antigens. A major challenge is to discover new, glioma‐specific tumor antigens that can be used for targeted immunotherapy without immunological side effects. Particular attention should be paid to detection of major glioma rejection antigens that are recognized by tumor‐infiltrating lymphocytes during endogenous antitumor responses, especially those expressed on tumor‐initiating glioma stem cells. Moreover, immunodominant CD4 and CD8 epitopes of these antigens should be defined to be able to generate tumor peptides that are presented both by MHC class I and class II molecules. Such information could be further used to create potent antitumor vaccines consisting of synthetic long peptides in combination with TLR‐agonists to elicit strong effector T‐cell responses 25, 243. In addition, large numbers of antigen‐specific effector T cells could be generated ex vivo that are adoptively transferred into lymphopenic patients followed by peptide‐pulsed DC vaccination. This approach has already resulted in effective antitumor responses against preestablished intracranial tumors in a preclinical model (205).

Irrespective of the detection of new glioma‐specific tumor antigens, future work should focus on the potential of tumor‐initiating glioma stem cells to provide tumor antigens for vaccination. The possibility of generating cancer stem cell lines from primary GBM patients has been demonstrated by several groups 21, 92, 155. These cancer stem cell lines could be used to specifically target glioma stem cells, for example, by replacing whole tumor lysate or total tumor RNA with lysate or RNA from glioma stem cells for loading DCs.

Further improvement could be obtained by immunizing not only against tumor cells, but also against the tumor stroma, which is important for tumor cell growth and survival. Despite the presence of tumor‐specific T cells, tumor cells can escape immune destruction as a result of inappropriate expression of MHC molecules or the presence of antigen‐loss variants. However, stromal cell subpopulations that can capture and present tumor‐derived antigens appear to show less immune escape mechanisms, and therefore might be a better target for immune effector cells (298).

Intensification of T‐cell costimulation

A powerful option to enhance the immunogenicity of a vaccine is to increase T‐cell costimulation. This can be achieved by coadministering agonistic antibodies specific for costimulatory molecules belonging to the TNF receptor superfamily, such as GITR, OX40 (CD134), 4‐1BB (CD137) and RANK 47, 89, 170, 237, 265. In murine glioma models some of these antibodies have exerted a promising effect 138, 140.

Alternatively, T‐cell costimulation can be increased by coadministering antagonistic antibodies specific for co‐inhibitory molecules expressed on T cells. For instance, many inhibitory pathways within the B7–CD28 family have been described that can attenuate T‐cell responses and promote T‐cell tolerance. CTLA‐4 and the programmed‐death (PD)‐1 receptor are two of the key negative regulators of T cell responses. Anti‐CTLA‐4 antibodies have been studied as potential therapeutics for glioma. We and others have shown that systemic CTLA‐4 blockade prolongs survival in treated mice, without eliciting signs of autoimmunity. Anti‐CTLA‐4 treatment reestablished normal CD4 counts and abrogated increased Treg fractions in the CD4 compartment observed in tumor‐bearing mice. Moreover, the CD4 T‐cell proliferative capacity could be restored and glioma‐specific antitumor immunity was enhanced 67, 85.

Recent data suggest that CTLA‐4 blockade can result in the effective treatment of tumors arising in the CNS. It was demonstrated that the systemic administration of a human mAb directed against CTLA‐4 (Ipilimumab, MDX‐010, Bristol‐Myers Squibb, New York, NY, USA and Medarex Inc. Princeton, NJ, USA) induced a significant clinical benefit in a female patient with metastatic melanoma to the CNS. Pathological examination of the brain metastasis following Ipilimumab treatment revealed abundant infiltrates of CD8+ lymphocytes and extensive tumor necrosis (104).

Counteracting Treg‐mediated immunosuppression

The immunogenicity of a vaccine can be further enhanced by removing suppression of Tregs. Especially in preclinical murine models, different strategies have been successfully used to eliminate Tregs such as CD25‐specific mAbs, immunotoxins or different chemotherapeutic agents.

We recently showed in a murine glioma model that depletion of Tregs by systemic administration of CD25‐specific mAbs can elicit an otherwise suppressed immune response against glioma‐specific tumor antigens, leading to the rejection of the tumors. Additional experiments revealed that combining Treg depletion with administration of CTLA‐4 mAbs further boosted glioma‐specific CD4 and CD8 effector T cells, as well as anti‐glioma IgG2a antibody titers in this model without any signs of autoimmunity (85). The potency of Tregs to suppress an effective anti‐glioma immune response was further demonstrated by the fact that mice were not protected from tumor outgrowth when vaccinated with tumor‐lysate pulsed DCs in a “simultaneous setting” where tumor cells and DCs are given at the same day. In contrast, depletion of Tregs before vaccination considerably increased the ability of vaccinated mice to survive the tumor challenge. Moreover, DC vaccination with anti‐CD25 treatment allowed the development of long‐lasting tumor protective immunity 87, 164. These results confirm that counteracting Treg‐mediated immunosuppression is an essential component in the success of anti‐glioma vaccines. However, a disadvantage of CD25 depletion is that treatment of mice with anti‐CD25 mAbs is only beneficial within a limited time window and its efficacy is dependent on tumor burden. Treg depletion after immunotherapy inhibited clonal expansion of tumor antigen specific T cells and reduced the efficacy of the vaccine (51).

In humans, several anti‐CD25 antibodies such as daclizumab have been developed and tested for their ability to deplete human Tregs. However, recent studies demonstrated that these antibodies failed to eliminate Tregs or did not affect their suppressive function 145, 271. Currently, two phase I/II randomized studies (NCT00626483, NCT00626015) are being carried out in patients with newly diagnosed GBM to determine if daclizumab inhibits the functional and numeric recovery of Tregs after therapeutic TMZ‐induced lymphopenia.

Besides CD25 mAb, a recombinant IL‐2 diphtheria toxin conjugate, DAB389IL‐2 (also known as denileukin diftitox or ONTAK) and a CD25‐specific immunotoxin (LMB‐2), which is formed of the single‐chain Fv fragment of the anti‐CD25 mAb with a recombinant form of the pseudomonas exotoxin, have been investigated for the elimination of Tregs. A few studies have shown a reduction in the number of circulating and tumor‐infiltrating Tregs after a single dose of ONTAK, with a more prominent effect after successive ONTAK treatments or administrations of LMB‐2. Moreover, ONTAK significantly improved the stimulation of tumor‐specific T cell responses after vaccination with tumor‐antigen loaded DCs 53, 165, 202. However, despite inducing a reduction in Treg cell numbers in vivo, ONTAK or LMB‐2 treatment have only resulted in limited objective clinical responses so far 146, 202.

Alternate chemotherapeutics are currently being investigated for their potency to selectively deplete Tregs 10, 76, 78. In a rat glioma model, oral administration of low‐dose metronomic TMZ has recently been shown to reduce circulating Tregs, associated with a suppression of their inhibitory functions on effector T cells in a rat glioma model. In contrast, high‐dose TMZ failed to significantly modulate Treg numbers or function (15). Similarly, a significant reduction of CD4+CD25+ T cells could be demonstrated with metronomic administration of TMZ in melanoma patients (251). Therefore, a low dose metronomic TMZ regime in association with a cancer vaccine could be an attractive strategy for the treatment of malignant glioma and might be superior in efficacy to conventional TMZ chemotherapy alone.

Another interesting approach to reduce the number of tumor‐infiltrating Tregs is to selectively block their trafficking to the tumor site. Treg subsets have been reported to differ in homing receptor expression and chemokine responsiveness 108, 189, 240. Recently, it was demonstrated that Tregs from GBM patients express significantly higher levels of CCR4 than those from healthy controls and they respond to both the recombinant human chemokines CCL2/MCP‐1 and CCL22/MDC in vitro. Moreover, chemotherapeutic agents such as TMZ and carmustine [3‐bis (2‐chloroethyl)‐1‐nitrosourea] were shown to reduce the production of CCL2 by glioma cells (131). Other inflammatory chemokine receptors involved in the attraction of Tregs are CCR2, CCR5 and CCR8 (113). Gliomas can synthesize and release CXCL12/SDF‐1 in a TGF‐β dependent manner (252). Interestingly, we observed CXCR4 expression on glioma‐infiltrating Tregs and its upregulation during tumor growth. Implementation of CXCR4 inhibitors may thus improve immunotherapeutic strategies against malignant gliomas (211).

BREAKING THE LOCAL IMMUNOSUPPRESSIVE ENVIRONMENT OF GLIOMAS

In order to enhance the persistence of the vaccine‐induced immune responses and to achieve optimal clinical benefit, apart from optimizing antiglioma vaccines, additional treatments need to be developed to break the local immunosuppressive environment of gliomas.

Targeting TGF‐β

A powerful approach to overcome glioma‐induced tolerance mechanisms involves targeting immunosuppressive molecules directly within the tumor microenvironment. In malignant gliomas, the three existing isoforms of TGF‐β in mammals, termed TGF‐β1, 2 and 3, are differentially expressed. High‐grade gliomas mostly overexpress the TGF‐β2 isoform, which is processed and secreted in large amounts in its active form (141). In recent years, different strategies have been developed for inhibiting TGF‐β. The most advanced in the clinical application is the use of a phosphorothioate‐modified antisense oligonucleotide (trabedersen, AP 12009, ANTISENSE Pharma GmbH, Regensburg, Germany), which is complementary to the mRNA encoding TGF‐β2 isoform 122, 123, 232. Several phase I/II‐studies have been carried out applying TGFβ2‐antisense oligonucleotides intratumorally through continuous high‐flow microperfusion [(convection‐enhanced delivery (CED)] in patients with high‐grade gliomas; these studies confirmed the safety of this approach, as well as long‐term clinical benefit in several individuals (97). A multinational phase III study (NCT00761280) comparing TGFβ2‐antisense‐oligonucleotides with standard chemotherapy (TMZ or BCNU) in adult patients with confirmed recurrent or refractory anaplastic astrocytoma (WHO grade III) is currently recruiting patients.

An alternative to the treatment with antisense oligonucleotides is to prevent downstream signaling by interfering with TGF‐β receptor kinase activities. Such kinase inhibitors have already been successfully tested in preclinical models, resulting in increased survival of treated animals, associated with pronounced tumor infiltration by NK‐cells, CD8+ T‐cells and macrophages 255, 262.

Strategies combining active immunotherapy with local TGF‐β blockade could thus further enhance therapy efficacy. In a preclinical study using a rat glioma model, the combination of intracranial TGF‐β2 antisense oligonucleotide administration and vaccination with irradiated glioma cells not only inhibited TGF‐β protein production in vivo, but also significantly prolonged survival times when compared with either vaccine alone or no therapy (161).

Targeting Stat3 and immunosuppressive enzymes

There is accumulating evidence from preclinical studies that targeting immunosuppressive molecules other than TGF‐β improves antitumor immunity against gliomas as well.

Considering the suppressive role of signal transducers and activators of transcription 3 (Stat3) in antitumor immunity, selective inhibitors of Stat3 have been evaluated in murine glioma models and were shown to activate intratumoral macrophages and microglia, induce apoptosis in glioma cells, and inhibit tumor growth 112, 120, 300.

With regard to immunosuppressive enzymes, specific IDO inhibitors such as 1‐methyl‐l‐trypophan have been studied to break immune resistance of malignant gliomas through T‐cell inactivation caused by tryptophan depletion and metabolite accumulation 106, 177. Moreover, COX‐2 inhibitors have been shown to significantly reduce PGE2 levels in Cox2‐overexpressing gliomas, thereby abrogating the induction of IL‐10 and TGF‐β‐secreting Tr1 cells (4). Selective COX‐2 inhibitors also show antiproliferative and apoptosis‐inducing effects independent of their COX‐2 inhibitory activity, resulting in the reduction or inhibition of glioma growth in vivo 182, 234. Furthermore, antitumor responses could also be elicited by blocking arginase expression using COX‐2 inhibitors 214, 215. To restore T‐effector‐cell functions, selective iNOS inhibitors such as SD‐3651 might be attractive for the treatment of gliomas (186).

Local treatment of gliomas with TLR agonists

Local administration of immune response modifiers, such as TLR‐agonists are powerful tools to generate immune responses against antigens derived from the tumor in vivo. Pioneering experiments from Carpentier et al have shown that direct injections of synthetic phosphothioate‐stabilized CpG‐oligonucleotides (CpG‐ODNs) in neuroblastomas induced complete tumor rejection in the majority of mice and triggered long‐term immunity (37). Further studies have confirmed the antitumor effects of CpG‐ODN in different intracranial models of syngeneic glioma 38, 174. Recently, we could show in a murine glioma model that the therapeutic efficacy of CpG‐ODN is predominately mediated through a TLR9‐dependent immune activation leading to intratumoral accumulation of IFN‐γ producing CD4 and CD8 T cells, a marked increase in the ratio of CD4 effector T cells to Tregs, and long‐term protective immunity. Moreover, local administration of CpG‐ODNs (TLR9 agonist) was most efficient in inducing antitumor immunity as compared with other TLR‐agonists (86).

A phase I trial using convection‐enhanced intratumoral delivery of CPG‐ODN (CpG‐28) has already been completed in the setting of recurrent GBM and showed minor responses in two patients without showing serious adverse events (36). A randomized phase‐II trial with CpG‐ODN in GBM is still ongoing (NCT00190424). However, a major drawback of this approach is that TLR9 expression in humans is far more restricted and predominately confined to plasmacytoid DCs and B cells. Therefore, besides CpG‐ODN, it will also be interesting to explore local delivery of functionally similar agents, such as TLR7/8‐agonists to break local immunosuppression in glioma patients. Survival benefit was observed in therapeutic murine glioma models after intratumoral injection of synthetic TLR7/8‐agonists, such as R848 (resiquimod) or protamine‐protected mRNA 86, 231. Alternatively, recently identified phosphorothioate‐modified immunostimulatory RNA oligonucleotides might be attractive in this setting, as they proved to be potent inducers of INF‐α and TH1 cytokines via TLR 7/8 (secreted by DCs) and triggered an antigen‐specific cytotoxic T‐cell and IgG2a response (31).

In the future, it might be worthwhile to apply certain combinations of TLR agonists to further enhance the treatment efficacy. Synergy between different TLR agonists is now well documented 183, 276. Recent findings further imply that TLR agonists not only induce beneficial activating, but also potentially suppressive effector responses. Therefore, selective inhibition of immunosuppressive molecules induced during TLR immunotherapy might further enhance the efficacy of TLR‐agonists as tumor immunotherapeutics. Inhibition of IL‐10 with an anti‐IL‐10 receptor antibody combined with CpG‐ODN activation was shown to reverse the tolerogenic state of tumor‐infiltrating DCs and to augment their therapeutic efficacy in a mouse tumor model (269). Moreover, vaccinations with TLR‐matured DCs pulsed with tumor antigen combined with peri‐ or intratumoral injection of TLR‐agonists can potentiate antitumor immunity, as it was shown to eradicate large murine tumors resistant to chemotherapy (99).

It will be also interesting to evaluate stimulation of other innate immune receptors for glioma therapy. Recently, bifunctional RNA molecules that were designed to target key tumor survival factors and to activate cytosolic retinoic acid‐inducible gene‐I (RIG‐I)‐like helicases promoted strong antitumor effects in a mouse model of metastatic melanoma to the lung (201).

To move forward, however, additional studies are needed to define the protective mechanisms of action of distinct innate immune receptor agonists and their potential toxicity, because immune response modifiers that activate a wide range of innate and adaptive immune cells can have severe adverse effects, especially in the CNS.

Local glioma treatment with DCs

An alternative strategy to modify the tumor microenvironment is to inject DCs directly into the tumor bed. Intratumoral injections of DCs either transduced with INF‐α DNA or TLR‐polarized DC have been shown to enhance anti‐CNS immunity by promoting cross‐presentation of tumor‐associated antigens in draining lymph nodes and by enhancing CNS tumor homing of antigen‐specific type 1 CTLs through the induction of CXCL10 73, 258. Furthermore, intratumoral delivery of DCs enhances the antitumor efficacy of peripheral vaccines 73, 152, 194. Preliminary clinical results also confirm that intratumoral DC application in addition to intradermal DC injections are more beneficial in the treatment of glioma patients than intradermal DC injections alone (294).

COMBINATION OF IMMUNOTHERAPY AND RADIOCHEMOTHERAPY

Recent studies indicate that the clinical responsiveness of patients with malignant gliomas to chemotherapy is increased after DC vaccinations. Vaccinated patients receiving subsequent chemotherapy showed significantly longer times to tumor recurrence/progression and longer overall survival after chemotherapy than those who were treated only with chemotherapy. It has been postulated that therapeutic vaccinations increase the sensitivity to chemotherapy by eliminating chemoresistant tumor cells 284, 285.

There are a number of reasons to combine immunotherapeutic strategies with conventional therapeutic methods such as radiotherapy and chemotherapy 153, 266. Recent data show that tumor cell death triggered by chemotherapy or radiotherapy initiates an immunoadjuvant pathway that contributes to the success of cytotoxic treatments (301). Numerous endogenous danger signals transferred by dying tumor cells to innate immune effectors may account for the immunogenicity of tumor cell death (8). Both anthracyclines and alkylating chemotherapeutics such as TMZ have been shown to induce an immunogenic cell death by triggering innate receptors such as TLRs. As a consequence, the ability of DCs to present tumor antigens from dying tumor cells to T and B cells is enhanced 7, 52, 192. Moreover, activation of innate immune receptors leads to the upregulation of MHC molecules on tumor cells, thus increasing their sensitivity to T‐cell‐mediated killing 86, 184. In addition, radiotherapy and chemotherapy remove local suppressor cells, such as tumor‐specific Tregs or might affect their function, thus permitting a more effective T‐cell stimulation 188, 251. Furthermore, chemotherapy has been shown to induce lymphopenia, thereby allowing thymic‐independent antigen‐driven T‐cell regeneration within the context of T‐cell homeostasis 127, 163. The concept of tumor‐specific immunization at the time of immune reconstitution after chemotherapy has been successfully tested in different animal models and phase I/II clinical studies, demonstrating that the availability of tumor antigens during homeostatic T‐cell proliferation leads to effective antitumor immunity and enhanced memory T‐cell responses 9, 12, 107, 209.

Combination therapies integrating vaccination strategies into the standard chemoradiotherapy protocol are currently under evaluation. There is preliminary evidence of successful combination of immunotherapy and TMZ (101). However, administration of anti‐immunosuppressive therapy together with radiochemotherapy must be carefully timed so as to optimize the immune response. It is of critical importance to establish reliable, reproducible and quantitative assays to evaluate vaccine‐induced immune responses 1, 259. Ideally, pre‐ and postvaccination samples should be taken throughout the course of vaccination concomitantly with the evaluation of clinical parameters. Vaccine‐induced cellular and humoral immune responses should be monitored in different compartments, especially at the tumor site, in the circulation, and at the vaccination site [delayed‐type hypersensitivity (DTH) test biopsies]. Overall, the combination of radiotherapy, chemotherapy and immunotherapy has the potential for strong antitumoral activity against malignant glioma, when applied using a well‐designed strategy.

CONCLUSION

Although our understanding of glioma immunobiology is rapidly growing, the knowledge about the complex pathological interactions within the local tumor microenvironment is still limited. Further elucidation of the spatiotemporal organization of the different players in tumor‐induced immunosuppression is of utmost importance for improving intervention strategies that boost potent antitumor responses. Current vaccination protocols are still of limited efficacy for suppressing glioma growth. To achieve optimal clinical benefit, additional treatments need to be developed and combined with a vaccine to break the local immunosuppressive environment of gliomas without inducing autoimmunity. This will enhance the persistence of the vaccine‐induced immune responses and will increase the chance to eliminate diffusely infiltrating tumor cells embedded within the normal brain parenchyma. Moreover, understanding the details of immune evasion in individual patients is critical, because it will ultimately enable the development of tailored therapies that specifically overcome these mechanisms of resistance. Keeping in mind that many of these immune evasion strategies used by gliomas are also part of the regular defense mechanism of the brain to prevent autoimmunity, such therapies have to be monitored carefully to detect and limit immune responses that are harmful to normal brain tissue. With further unraveling of glioma immunobiology, immunotherapeutic strategies have the opportunity to become a standard component in the multimodal treatment of malignant gliomas.