Overview: Mechanisms of Immune Evasion

The precise mechanisms that enable antigenic neoplasms to escape host immunity are incompletely understood. The tumor cells appear to escape recognition by the immune system by a variety of redundant immune suppressive mechanisms, many of which have been appropriated from normal immune physiologic mechanisms of down-modulating the immune responses. For example, neoplastic cells express weakly immunogenic tumor-associated antigens (TAAs), which are typically overexpressed self antigens, and immunogenic tumor-specific antigens (TSAs), which are only expressed on the malignancy. Tumor cells can evade immune responses by losing expression of these antigens or MHC molecules. Many gliomas have low expression of human leukocyte antigens (HLAs) and induce muted MHC class II responses. The inhibitory effects of these suppressor mechanisms are further mediated by cytokines. Successful methods to induce immunity to TAAs could lead to tumor cell destruction and prolong the survival of cancer patients.

The secretion of immunosuppressive factors by the tumors themselves has recently been identified as a further target for immunotherapy. In particular, interleukin-10 (IL-10) and transforming growth factor-β (TGF-β) are two examples of such immunosuppressive cytokines secreted by tumor cells. Furthermore, there is increased expression of FasL and galectin by glioma that triggers T-cell depletion and T-cell apoptosis.¹³

Recently, a subset of regulatory T cells (T regs) have been identified in the local tumor environment and in the draining lymph nodes of systemic tumors, and they are elevated in the CD4 compartment of the systemic blood circulation. These cells coexpress CD4 or CD25, or both, and display potent immunosuppressive T-cell activity.¹⁴ In vaccines used for tumors outside the CNS, monoclonal antibody inhibition of these T regs has been associated with enhanced antitumor immunogenicity (details provided later). The inhibitory effects of these suppressor cells may also be mediated by cytokines. Interestingly, IL-2–mediated immune regulation involves not only the stimulation of effector cells but also the regulation of T regs.¹⁵

Thus, another source of immunosuppression, other than factors secreted by the tumor, is the host’s own immune system. Finally, many patients are treated with corticosteroids in the perioperative period, which furthers inhibits the host’s immune responses.

Regulatory T Cells

T regs inhibit T-cell activation and effector function by downregulating IL-2 production, inhibiting interferon-γ (IFN- γ) production, and increasing T_H2 cytokine production,¹⁶ thereby inhibiting the expression of MHC class II molecules, CD40, CD80, and CD86 and suppressing antigen presentation by monocytes and macrophages.^17,18 The presence of T reg–mediated suppression has been documented in a variety of human cancers.¹⁹ Several studies on immunogenic murine tumors have shown that adoptively transferred T regs are responsible for inhibition of tumor-reactive effector T cells and that the elimination of T regs enhances antitumor immunity.¹⁶ Within glioblastoma multiforme (GBM) patients, although reduced in absolute number, T regs comprise a disproportionately large fraction of the patient’s peripheral CD4 T-cell compartment. This increase corresponds with a decrease in T-cell effector functions. Furthermore, in vitro depletion of peripheral blood T reg cells results in successful reversal of effector T-cell function, including increased proliferation and a switch to a T_H1 (IL-2–positive, tumor necrosis factor-α [TNF-α]–positive, IFN-γ–positive) cytokine profile. Within murine tumor challenge models of established gliomas, in vivo depletion of CD4⁺ and CD25⁺ cells resulted in significant long-term survival.^20,21

Within the glioma microenvironment, the host-generated effector T cells can be critically suppressed or overwhelmed by the T reg cells. Tissues from glioma patients obtained after surgical resection have been dissociated and stained for the CD8⁺ and CD4⁺ subsets. The tumor-infiltrating CD8⁺ T cells were phenotypically CD8⁺ and CD25⁻, indicating that these cells were not activated or proliferating. CD4⁺ T cells were more numerous than CD8⁺ T cells within glioma tissue and the majority of CD4⁺ T cells were T regs, as evident by positive intracellular staining of FoxP3, a transcription factor that is elevated only in T regs and not in activated effector CD4⁺ T cells.²² These data indicate that T regs not only can inhibit initial systemic immune activation but also can prevent the effector responses in the tumor microenvironment. Of note, lower grade gliomas have only rare T regs, and oligodendrogliomas none at all. Even within the GBM patients’ specimens, the presence of T reg infiltration is highly variable and not an independent predictor of survival, emphasizing the redundancy of immunosuppressive pathways. Human T regs have been shown to accumulate in tumors and in ascites induced by the tumor and macrophages by the chemokine CCL22, which mediates trafficking of T regs to the tumor.¹⁹ Similarly, gliomas elaborate the chemokine CCL2, which induced the intratumoral migration of T regs.²³

Regulatory T-Cell Inhibition by Chemotherapeutic Agents

Overcoming T reg immunosuppression is an ongoing immune therapeutic approach that can be achieved through a variety of approaches, including the use of denileukin diftitox (Ontak),²⁴ cyclophosamide,²⁵ or signal transducer and activator of transcription type 3 (STAT-3) blockade agents,²⁶ or by inhibiting the T reg trafficking (i.e., inhibition of CCL2) with temozolomide.²⁷ Inactivation of T regs with an anti-CD25 antibody in murine glioma models has been shown to enhance vaccination-induced antitumor immune responses and to result in the eradication of intracerebral astrocytomas without induction of autoimmunity.²⁸ Similar results were also obtained in this murine model system with systemic cytotoxic T-lymphocyte–associated protein type 4 (CTLA-4) blockade. The CTLA-4 blockade reversed the CD4⁺ T-cell deficit, similarly seen in malignant glioma patients, and normalized the ratio of T regs in tumor-bearing mice.²⁹ Although eliminating the suppression of endogenous antitumor immune responses through the elimination of regulatory T cells may enhance tumor immune clearance, there is a potential risk for inducing autoimmunity; however, that was not found in the murine model. It is likely that the induction of T_H17 responses, and not necessarily the inhibition of T regs, is the primary mechanism of inducing CNS autoimmunity.^30,31 In patients with metastatic renal carcinoma, Ontak specifically depleted the T regs without inducing toxicity and significantly improved the stimulation of tumor-specific T-cell responses when those patients were stimulated with RNA-transfected DCs.³² Without vaccination, Ontak has been shown to be efficacious as a single agent in relapsed and refractory B-cell non-Hodgkin’s lymphoma³³ and in cutaneous T-cell lymphoma.³⁴ However, Ontak was not particularly effective in clearing the T reg cells in melanoma patients.³⁵ No published studies to date have evaluated this approach in malignant glioma patients. Another set of potential T reg modulation includes agents that block or disrupt Notch signaling. Notch signaling has also been identified as playing a role in regulating the responses of T cells and can affect the differentiation of CD4⁺ T cells into T regs.³⁶ However, it is unclear which members of the Notch family need to be blocked, and inhibitors are currently under development.

Lymphodepletion to augment immunologic responses has been described in both murine model systems^37,38 and human cancer patients.³⁹ These enhanced antitumor responses after lymphodepletion may be secondary to the removal of competition at the surface of APCs,⁴⁰ enhanced availability of cytokines that augment T-cell activity (e.g., IL-7 and IL-15),⁴¹ or depletion of immune inhibitory T regs.⁴² Cyclophosphamide (CTX), an alkylating agent with a therapeutic effect against tumors at high doses, has preferential effects on inhibiting T regs at lower doses. CTX can abolish the function of CD4⁺, CD25⁺, FoxP3⁺ T cells and enhance cytotoxic T-cell responses.⁴³ Pretreatment with CTX before antitumor vaccination results in the activation of CD8 tumor-specific T cells.⁴⁴ When CTX is administered at subtumoricidal doses, there is improved immune response and tumor eradication of large established murine tumors (sarcoma) when treated in combination with IL-12. This combination enhances CD4, CD8 T cells and macrophage infiltration within the tumor and skews the immune responses to the T_H1 phenotype.⁴⁵ Similar findings have been made in the murine system treating established melanoma with IL-2, IL-1, and CTX.⁴⁶ In addition to its induction of immunity to new antigens, CTX can also overcome tolerance. For example, the administration of CTX to mice bearing an established plasmacytoma resulted in the cure of 92% of mice, and further studies demonstrated that the cured mice rejected a subsequent tumor challenge.⁴⁷ The mechanism appeared to involve both the generation of CD8⁺ cytotoxic T cells and the upregulation of B7-1 (costimulatory molecule) on tumor cells.⁴⁸

Multiple clinical trials have demonstrated enhanced immune responses and improved clinical efficacy when CTX was administered before immunotherapy.^49,50 Administration of CTX to patients with advanced melanoma and colorectal cancer resulted in enhanced delayed-type hypersensitivity (DTH) reactions and the development of antibodies.⁵¹ When CTX was administered to augment autologous melanoma vaccine in patients with metastatic melanoma, the inclusion of CTX resulted in enhanced DTH responses.⁵² Similar potentiation of immune responses was seen in metastatic breast cancer and renal cell carcinoma. CTX has also been used in a phase I clinical trial of glioma patients and demonstrated some evidence of clinical efficacy. Plautz and colleagues⁵³ treated newly diagnosed glioma patients with a single dose of CTX before the administration of adoptively transferred T cells harvested from the patients’ lymph nodes. This study was conducted before our current understanding of the influence of CTX on T regs, and thus the contribution of the CTX to the clinical efficacy is unknown.

Nevertheless, when combining the different forms of treatment, all aspects of treatment must be taken into account. Different dosing schedules and methods of vaccination can abrogate the desired therapeutic response. The use of low-dose doxorubicin after vaccination with a granulocyte-macrophage colony-stimulating factor (GM-CSF)–secreting cancer cell vaccine in a CT26 colon carcinoma murine model showed a significant increase in CD8⁺ T-cell response compared with the use of vaccine alone. Yet, when low-dose CTX was administered after or concurrently with a GM-CSF–secreting cancer cell vaccine, the desired immune response was inhibited, and the result was equivalent to that seen in mice treated only with chemotherapy.⁵⁴ Thus, with CTX administered before vaccination, the reduction of suppressive immunity increased the immunotherapy-mediated tumor clearance. Similar results were also obtained with paclitaxel.⁵⁵

Another agent that is capable of suppressing T regs is temozolomide (TMZ), the standard of care for GBM. TMZ is an alkylating agent that causes cell death by inducing G₂/M-phase cell cycle arrest and likely autophagy without apoptosis.⁵⁶ TMZ can inhibit the proliferation of lymphocytes, but it also depletes T regs^27,57 and the trafficking of T regs into the glioma microenvironment.²³ This mechanism may prove beneficial in combination with vaccine therapy and was recently investigated in a phase II clinical trial (ACT II) in combination with PEP-3-KLH vaccination.

In murine models of established intracerebral tumors, the combination of IL-2 immunotherapy and intratumoral administration of bischloroethylnitrosourea (BCNU) polymers enhanced survival.⁵⁸ The investigators suggested that the cytotoxic agents might increase the number of tumor peptide antigens released or abrogate tumor-derived T-cell suppressor factors, or both. When mice with established tumors (lymphoma) were treated with BCNU, they rejected secondary tumor challenges, indicating the induction of immunologic memory. Furthermore, mice treated with BCNU demonstrated high cytotoxic activity against tumor cells and low T-cell suppressor activity. Thus, although BCNU was directly tumoricidal, it is also possible that the BCNU inhibited the T regs, resulting in antitumor immunity.⁵⁹ The combination of BCNU with the newer efficacious immunotherapies has not been explored in animal models or clinical trials.

Immuno suppressive Cytokines

Human glioblastoma cells secrete a variety of factors, such as prostaglandin E (PGE), IL-10, and TGF-β2 that suppress multiple immune functions. For example, TGF-β2 inhibits cytotoxic responses of T cells against tumor targets, PGE₂ regulates the generation of immune responses, IL-10 downregulates MHC class II expression on monocytes and suppresses T-cell proliferation,^60–62 and vascular endothelial growth factor (VEGF) inhibits the maturation of DCs and subsequently downregulates MHC class II molecules. Suppression of TGF-β with an antisense TGF-β has been shown to be efficacious in 9L and C6 glioma intracerebral models.^53,63 Other attempts at TGF-β inhibition in murine models have included SD-208, a TGF-βRI kinase inhibitor, which slightly increased median survival,⁶⁴ and Tranilast, a TGF-β1 antagonist, which resulted in a 50% reduction in intracranial tumor volume.⁶⁵ A phase I/II clinical trial using an antisense oligodeoxynucleotide (AP 12009) targeted to suppress TGF-β2 in high-grade glioma patients demonstrated an increase in median survival after recurrence that exceeded the current data for chemotherapy.⁶⁶ The success of this type of approach is dependent on the dependency of the glioma on this particular mechanism of immune suppression and is compounded by the fact that gliomas usually express a variety of immunosuppressive cytokines; the blockade of any one cytokine may not be expected to significantly impact the overall immunosuppressive milieu.

The taxanes, paclitaxel and docetaxel, bind to ß-tubulin, stabilizing microtubules, and resulting in G₂/M-phase mitotic arrest. Paclitaxel can diminish the cytotoxicity of NK and T cells by interfering with IL-2–mediated activation. However, the best-known immunomodulatory effect of paclitaxel is its ability to trigger macrophages to secrete a variety of proinflammatory factors, such as IL-1ß, GM-CSF, tumor necrosis factor-α (TNF-α), and IL-12. This lipopolysaccharide-like activity enables paclitaxel, in the presence of a priming signal, to induce the tumoricidal activity of macrophages.⁶⁷ To date, there have been no reports of combining paclitaxel treatment with cancer immunotherapy.

Clearly, further studies investigating the cytokine profile before chemotherapy are warranted to enhance our understanding of the endogenous immune response. For example, in patients with breast cancer that were evaluated before treatment with chemotherapy, there was a significant decrease in immune cells such as NK and LAK cells and immunostimulatory cytokines such as IL-2, GM-CSF, and interferon compared with healthy individuals. After the administration of docetaxel and paclitaxel, there was a significant increase in the number of immune cells and a decrease in PGE₂ and TNF.⁶⁸ Doxorubicin has also been shown to have immunomodulating effects, including increasing the activity of monocytes and increasing the secretion of IL-1 and IL-2.⁶⁹

Immune Inhibitory Molecules

To initiate an immune response, T cells must receive two signals: one through the T-cell receptor (TCR), which recognizes antigen presented within the context of MHC; and the second through the costimulatory receptor CD28, which recognizes the costimulatory molecules CD80 and CD86 expressed on the surface of the APC. In response to antigen-specific stimulation, activated T cells produce IL-2 and proliferate. B7-H1 is a negative costimulatory molecule that inhibits CD4⁺ and CD8⁺ T-cell activation, can induce T-cell apoptosis,^70,71 and is upregulated within the tumor microenvironment⁷² and specifically in gliomas.⁷⁰ Blockade of B7-H1 with a monoclonal antibody can activate T cells, which are then able to inhibit tumor.⁷³ This represents a novel mechanism by which glioma cells evade immune recognition and destruction. Blockade of B7-H1 with antibodies in combination with adoptive T-cell transfer has been shown to enhance cancer clearance⁷³ and enhance myeloid DC-mediated immunotherapy⁷²; however, whether sufficient quantities of these types of antibodies can be achieved in the CNS remains to be seen. T-cell anergy can also be induced if the inhibitory molecule CTLA-4 is present on the gliomas.⁷⁴ CTLA-4 blockade has had some success in promoting antitumor T-cell responses in both preclinical and clinical settings,^29,75 but it is not clear whether this approach would be successful against primary brain tumors or CNS metastasis.

HLA-G is a nonclassic MHC molecule with limited tissue distribution that has an immune regulatory function. Soluble HLA-G (sHLA-G) has been identified in the plasma of patients with melanoma, glioma, breast cancer, or ovarian cancer. sHLA-G can inhibit the functions of T and NK cells at high concentration⁷⁶; thus, at levels within the tumor microenvironment, it would be anticipated that there is reduced immunologic responsiveness. Gene transfer of HLA-G1 or HLA-G5 into HLA-G–negative glioma cells (U87-MG) rendered these cells highly resistant to direct cytolytic clearance, inhibited T-cell proliferation, and prevented efficient priming of cytotoxic T cells. The inhibitory effects of HLA-G were directed against CD8 and CD4 T cells, but appeared to be NK-cell independent.⁷⁷ Only a few HLA-G–positive cells within the tumor microenvironment are necessary to exert significant immune inhibitory effects⁷⁷; thus, the inhibition of HLA-G blockade would need to be comprehensive. However, the frequency of expression of HLA-G and B7-H1 has not been assessed across glioma grades or types on tissue arrays; thus, the universality of this type of immune evasion is not currently known.

Phenomenon of Antigen-Loss Variants in Gliomas

Although tumor specificity has the advantage of not inducing potentially fatal autoimmunity, the intrinsic antigenic heterogeneity will prevent clearance of all cancer cells and the development of antigen-loss variants, especially for those antigens whose expression is not required by tumor cells for the maintenance of a transformed phenotype. The propagation of such antigen-loss variants is secondary to epitope immunodominance, that is, the preferential immunodetection of one or a few epitopes among many expressed on a given target.⁷⁸ The parental tumor cells carrying the immune dominant epitope can be eliminated, and the formerly immune recessive epitopes then become dominant, allowing unrestricted tumor growth. It is well documented that tumor antigen expression is heterogeneous, even within the same tumor,⁷⁹ and antigen loss has been observed with a variety of immune therapeutic approaches in various cancers.^80,81

The epidermal growth factor receptor (EGFR) is often amplified and structurally rearranged in malignant gliomas, with the most common mutation being EGFRvIII. Treatment in a murine model with a vaccine consisting of a peptide encompassing the tumor-specific mutated segment of EGFRvIII (PEP-3) conjugated to keyhole limpet hemocyanin (KLH [PEP-3-KLH]) resulted in the loss of antigen expression. Specifically, EGFRvIII expression in mice that failed to respond to the PEP-3-KLH vaccination was lost in 80% of relapsing tumors, indicating that EGFRvIII-negative escape variants were a potential mechanism of treatment failure in active immunotherapy.¹¹ Although EGFRvIII appears to represent a nearly terminal branch of tumor progression for malignant brain tumors and is expressed clonally within this lineage, antigen-specific tumor targeting might confer a selective growth advantage on neoplastic cells not expressing this epitope. It has been shown previously that treatment failure with antigen-specific passive humoral immunotherapy is not a result of antigen-loss variants,⁸² suggesting that antigen-negative escape variants are an etiology of treatment failure in active immunotherapy. In two separate phase II clinical trials (ACTIVATE and ACT II) in newly diagnosed EGFRvIII-expressing GBM patients who received the PEP-3-KLH, on tumor progression, the EGFRvIII expression was lost.^57,84 To overcome this obstacle, cocktails of different antigens or whole-tumor lysates can be used as immunogens⁸⁵; however, with nonspecific antigen approaches, there is a risk for autoimmune responses.¹⁰ To circumvent this, future approaches may involve screening the patient’s tumor for a panel of antigens⁸⁶ and targeted immunotherapy selected based on these parameters—similar to selecting antibiotics for a given bacterial infection. Alternatively, upregulating the expression of these antigens with some forms of chemotherapy may restore the ability of the immune system to clear these cancer cells.⁸⁷

A variety of agents have been shown to be capable of inducing or upregulating antigen expression. For example, gemcitabine, a nucleoside analogue, can increase tumor antigen cross-presentation and T-cell infiltration of the tumor.⁸⁸ The demethylating agent DAC is capable of inducing the tumor-associated antigen MAGE-1 on melanoma cells with subsequent cytolytic T-cell clearance.⁸⁹ Irinotecan (CPT-11) has been shown to induce the LYGD/E48 antigen, which resulted in synergistic tumor regression with a cytotoxin-armed antibody in a colorectal xenograft murine model.⁹⁰ The combination of 5-fluorouracil and cisplatin (CDDP) can upregulate the antigen CEA and MHC class I in human colorectal cancer cell lines⁹¹ and was functionally relevant as evidenced by increased CEA-specific, MHC-restricted cytotoxic T-cell killing.⁸⁷ Thus, many different types of chemotherapy can enhance antigen expression; however, none of these agents has been tested to determine whether it can restore antigen expression after clonal elimination with immunotherapy.

An alternative therapeutic approach for the issue of antigen heterogeneity would be to selectively screen each patient’s tumor for a panel of potential antigens, but this would be somewhat labor intensive and costly at present. Another alternative approach would be to vaccinate cancer patients with a panel of tumor-associated or tumor-specific antigens. Tissue arrays of gliomas from a large population of patients could be used to determine the frequency of the most common tumor antigens. This strategy would be very similar to that used in manufacturing flu vaccines annually, in which the most common virulent strains are incorporated as immunologic targets. Finally, exploration has begun on immunologic approaches for inducing immune responses against immune escape variants based on defects in the transporter associated with antigen processing,⁹² but these are not yet sufficiently developed for clinical trial application.

STAT-3—A Key Switch Mediating Immuno suppression

The ability of tumor cells to proliferate uncontrollably, resist apoptosis, sustain angiogenesis, and evade immune surveillance is regulated, in part, by the signal transducer and activator of transcription type 3 (STAT-3), which acts as a cytoplasmic signaling protein and nuclear transcription factor. Under normal physiologic conditions, latent STAT-3 activation is dependant on ligand-receptor interaction, primarily under the control of growth factor receptor tyrosine kinases, cytokines, and G-protein receptors.⁹³ For example, EGFR and IL-6 activate Jak2, which then activates STAT-3 by phosphorylation of the tyrosine residue in the transactivation domain of STAT-3.⁹⁴ In cancer cells, these tyrosine kinases are among the most frequently activated oncogenic proteins. STAT-3 has been shown to be persistently activated in most human cancers, including gliomas.^95,96 The phosphorylation of STAT-3 has been attributed to EGFRvIII.⁹⁷ IL-6 is expressed in the CNS, especially by reactive astrocytes, under a wide variety of conditions, such as hypoxia,⁹⁸ traumatic and metabolic injury,⁹⁹ and inflammation,¹⁰⁰ which in turn also induces expression of p-STAT-3. The use of decoy antisense STAT-3 oligonucleotides and dominant-negative vectors has provided convincing evidence that STAT-3 is highly relevant to the growth and survival of many tumor types, including gliomas.^101–103

STAT-3 has also emerged as a key switch that drives the underlying immuno suppression¹⁰⁴ identified in cancer patients by preventing maturation of DCs and inhibiting the proliferation and activation of immune effector populations. STAT-3 has been shown to be a potent regulator of these anti-inflammatory responses by suppressing macrophage activation and limiting inflammatory responses.^105,106 It also has been shown to become constitutively active in diverse tumor-infiltrating immune cells.¹⁰⁷ For example, STAT-3 activity within NK cells and neutrophils directly reduces their cytotoxicity, whereas STAT-3 activity in DCs reduces the expression of MHC class II, CD80, CD86, and IL-12 in these cells, rendering them unable to stimulate T cells and generate antitumor immunity. Investigators have recently shown that by ablating STAT-3 in only the hematopoietic cells in mice resulted in marked enhancement of activated and functional T cells, NK cells, and DCs in tumor-bearing mice. This ablation of STAT-3 in only the hematopoietic cells also resulted in marked antitumor effects in vivo, indicating that STAT-3 expression in immune cells restrains antitumor immune eradication.¹⁰⁷ The tumor microenvironment induces STAT-3 activity in tumor-associated immune cells.^104,107,108

The blockade of STAT-3 with small molecular inhibitors in microglia obtained from the CNS of patients with GBM resulted in the upregulation of costimulatory molecules (e.g., CD80, CD86), induced systemic APCs to elaborate proinflammatory cytokines essential for T-cell effector responses, and induced activation and proliferation of T cells, indicating that STAT-3 blockade is a potent approach to modulating both the systemic and local tumor immune microenvironments. Of note, this potent immune activation occurred in immune cells obtained from patients with disease refractory to other conventional immune activators, such as toll-like receptor agonists. When the immune cells from immunosuppressed GBM patients were treated with a STAT-3 inhibitor, Western blot analysis demonstrated phosphorylation of ZAP-70 in T cells, thus bypassing the impairment of signal transduction typically identified in the immune systems of patients with cancer.¹⁰⁹ The mechanism of T-cell activation is likely secondary to the inhibition of T regs present in these immune preparations, thus allowing immune activation (i.e., releasing the “brake” on immune cell activation and proliferation). IL-2 has been shown to regulate FoxP3 expression in human CD4⁺, CD25⁺ T regs by inducing STAT-3 binding of the first intron of the FoxP3 gene.¹¹⁰ Suppressor of cytokine signaling type 3 has been shown to be an inhibitor of STAT-3 signaling¹¹¹ and transcriptional activity^112,113 but is deficient in T regs.¹¹⁴ STAT-3 inhibitors should have a multiplicity of immune modulatory activities, such as allowance of expression of costimulatory molecules in the tumor microenvironment, inhibition of immune suppressive cytokines, maturation of DCs, and inhibition of T regs, and represents a newly emerging class of drugs that will be used for immune therapeutic purposes.

Immune Modulation by Central Nervous System Antigen-Presenting Cells

T-cell activation requires signals through both the MHC and costimulatory molecules, and the expression of MHC alone results in T-cell anergy.¹¹⁵ Low levels of expression of CD80 provide an immune escape advantage to cancer cells, and CD28-mediated costimulatory signals are essential for differentiation of functional tumor-specific cytotoxic CD8⁺ effector T cells.^116,117 In tumor-bearing animals, repeated stimulation of the T cells is necessary, especially at the effector site, to generate tumor responses,¹¹⁸ indicating that if the APC failed to provide appropriate stimulation, the APC may contribute to immune suppression. Although microglia may be able to act as APCs during autoimmunity (multiple sclerosis) and infection (HIV),^119,120 glioma-associated microglia are functionally impaired, are not able to stimulate T-cell responses, and are refractory to activation stimulation (i.e., toll-like receptor agonists).⁷ Intratumoral DCs may be able to provide these immune-stimulating conditions, but these are rarely present in malignant gliomas.⁷ Additionally, CD11b⁺ inhibitory macrophages (iMACs) have been described, which induce apoptotic death in CD8⁺ T cells. This inhibitory phenotype could be changed into one reflective of highly activated DCs by culturing the iMACs with IL-4 and GM-CSF,¹²¹ suggesting that the tumor microenvironment can be manipulated for therapeutic purposes. However, the delivery of cytokines directly to the CNS is problematic. Costimulatory molecules on glioma-associated microglia and macrophages have been shown to be lost or significantly downregulated. In attempts to overcome the failure of APCs, and specifically the lack of sufficient costimulation, vaccines with enhanced costimulatory capacity have been tested and demonstrated antitumor effects that have maintained high avidity effector-memory CD8⁺ T cells. Other approaches for potentially upregulating the costimulator molecules on APCs include CpG DNA,¹²² chemotherapy with cytosine arabinoside,¹²³ transfection with poxviruses expressing the costimulatory molecules,¹²⁴ or transfection of the replication-defective fowlpox recombinant vector rF-TRICOM (TrIad of COstimulatory Molecules).¹²⁵

An emerging strategy for upregulating costimulatory molecules in the glioma microenvironment is with STAT-3 blockade agents. A small molecular inhibitor (WP1066) of STAT-3 has been shown to upregulate costimulatory molecules and proinflammatory cytokines on microglia obtained from the CNS of patients with GBM that were refractory to modulation by other conventional immune activators, such as toll-like receptor agonists. Furthermore, WP1066 can modulate enhanced T-cell functional activities in physiologic ranges that can be obtained within the CNS.¹⁰⁹ These types of agents are anticipated to be in clinical trials of human patients within the next several years.

Immune Modulation by Glioma-Infiltrating Lymphocytes and T_H2 Lymphocytes

After activation, CD4⁺ helper T cells are classified, based on their elaborated cytokine responses, as T_H1 (promoting cell-mediated immunity) or T_H2 (promoting humoral immunity), which predominantly secrete either interferon-γ (IFN-γ) or IL-4, respectively. A predominance of T_H2-type cytokine response has been correlated with enhanced tumor growth.^126,127 This T_H2 polarization has been demonstrated in gliomas¹²⁸ and is likely mediated by microglia-secreted macrophage-derived chemokine (MDC)¹²⁹ or neuropeptides, or both.¹³⁰ One therapeutic approach altering the T_H2 intratumoral skewing would be to limit the chemokines that direct migration of the immunosuppressive immune cells to the tumor microenvironment, such as with a STAT-3 inhibitor.²⁶ Alternatively, preferential T_H1 responses could be induced with cyclooxygenase-2 (COX-2) inhibition.¹³¹

Impaired Immune Responses

The proliferative response of stimulated T cells, along with the levels of the IL-2 receptor and tyrosine phosphorylation in response to IL-2, from malignant glioma patients are reduced compared with healthy donors.¹³² These results suggest a defect in the expression of functional IL-2 receptor and T cells isolated directly from human gliomas that are not activated as reflected by expression of the IL-2 receptor CD25.⁷ Soluble products from tumors may suppress T-cell proliferation through a mechanism that involves downregulation of Jak3 expression and inhibition of IL-2–dependent signaling pathways.¹³³ A potential mechanism for overcoming this is the constitutive activation of the T cells by transfection with CD3 or CD28, or both, for adoptive transfer therapeutic purposes. An antibody, r28M, directed against a melanoma-associated proteoglycan expressed on glioblastoma cells activates T cells through the CD28 molecule without additional stimulus from the TCR-CD3 complex and induces T-cell–mediated killing of glioblastoma cells in vitro and in vivo.¹³⁴

Doxorubicin, an anthracycline antibiotic, can enhance the cytotoxic T-cell compartment. Pretreatment of mice with doxorubicin produced enhanced cytotoxic T-cell responses and activation of macrophages, as reflected by the secretion of IL-1 and IL-2.⁵¹ There was a twofold increase in the number of macrophages that could support the generation of the cytotoxic T cells. Ultimately, the macrophages became tumoricidal, and the effect persisted for 14 to 18 days. Further studies demonstrated that the administration of doxorubicin 5 days after a GM-CSF–secreting vaccine injection potently induced cytotoxic T-cell activity.⁵⁴ Cancer patients treated with doxorubicin at a dose of 25 mg/m² developed an increase in the percentage of circulating CD8⁺ T cells having twofold higher cytotoxic activity.¹³⁵ Ideally, the chemotherapy selected for use in combination with immunotherapy should have some intrinsic tumoricidal activity. Although doxorubicin has little clinical efficacy against gliomas, this approach could potentially be used to boost the T-cell responses in glioma patients with further clinical development.

T-Cell Apoptosis

T-cell apoptosis has been observed frequently within malignant gliomas.¹³⁶ The induction of apoptosis involves the activation of caspases, leading to nuclear fragmentation and apoptosis, and is triggered by deprivation of survival stimuli such as costimulators⁷ or by binding of ligands to death-inducing membrane receptors.¹³⁷ Both of these mechanisms are operational within gliomas. FasL is predominantly expressed on T cells after activation by antigen and IL-2. When mature T cells are repeatedly stimulated by antigens, they coexpress Fas and FasL.¹³⁷ Gliomas can evade killing by tumor-infiltrating T cells by overexpressing FasL, rendering the tumor resistant to apoptosis¹³⁸ while delivering a death signal to Fas-expressing cells, which include the activated cancer-infiltrating T cells. However, high-grade gliomas also express the apoptotic receptors Fas and CD95, rendering them susceptible to antibody-stimulated Fas- and CD95-mediated apoptosis and CD8⁺ cytotoxic killing.¹³⁹ Alternatively, agents such as topotecan can be administered to patients to upregulate Fas on gliomas that renders the tumor more susceptible to cytotoxic immune cell clearance.^139–141

Topotecan penetrates the blood-brain barrier but has demonstrated only modest clinical efficacy in glioma patients, probably owing to failure to achieve in vivo levels sufficient to induce apoptosis of the tumor. Nevertheless, topotecan has been shown to upregulate Fas and CD95 in high-grade gliomas, rendering them more susceptible to immune clearance.^139,141 Thus, although levels achieved are insufficient for direct glioma clearance within the CNS, this agent potentially could be exploited as an immune modulator. Other chemotherapeutics, such as doxorubicin¹⁴² (Adriamycin) and its analogues (epirubicin and pirarubicin),^143,144 VP-16 (etoposide), cisplatin,¹⁴⁴ and the combination of cisplatin and 5-fluorouracil¹⁴⁵ can also increase cytotoxicity mediated by T cells and NK cells by upregulating Fas.

A synergistic antitumor effect was observed with paclitaxel and FasL in two human malignant glioma cell lines, T98G and LN-229.¹⁴⁶ The synergy of paclitaxel and FasL was achieved independently of either the G₂/M-phase cell cycle arrest or p53 activation. At low concentrations of paclitaxel, Bcl-2 phosphorylation was induced in the glioma cells, which in turn interfered with the heterodimerization of Bcl-2 with Bax and the inhibition of Fas-induced apoptosis by Bcl-2. The investigators hypothesized that the synergistic antiproliferative effect of paclitaxel and FasL on malignant glioma cells stemmed from the upregulation of the Bcl-2/Bax rheostat in favor of Bax, thereby sensitizing the tumor cells to apoptosis. Interestingly, this mechanistic synergy was not observed with other agents such as teniposide.¹⁴⁷

The upregulation of Fas and CD95 could be combined with a tumor vaccine or activated lymphocytes to further potentiate induced immune responses. For example, chemotherapy could be administered first, to upregulate the expression of Fas on the glioma targets, followed by administration of the immunotherapy. In the case of active immunotherapy, the timing for topotecan treatment would depend on when an optimal effector response is obtained. For instance, if a DC vaccination induces a potent immunologic response 1 week after administration, then the topotecan would be administered 6 days after the vaccination. There are no previous reports of topotecan affecting the cytotoxic T-cell responses, but this would need to be addressed before the employment of this agent in a clinical trial. The window would need to be identified at which there is optimal upregulation of the immunologic target with minimal derangement of the cytotoxic effector response. Theoretically, the induced or administered cytotoxic T cells would be able to clear glioma cells more effectively at a lower effector-to-target ratio. Alternatively, this approach could provide a potent synergy by enhancing different types of immune responses in concert, such as enhancing cytotoxic responses with the delivery of antibody.

T-cell apoptosis is also induced by glioma-expressed CD70 and gangliosides. The glucosylceramide synthase inhibitor (PPPP) has been shown to reduce the ability of GBM cell lines to induce apoptosis in T cells.¹⁴⁸ Alternative strategies for blocking the T-cell death pathway include proteosome inhibitors, antisense oligonucleotide inhibitors, and other small molecule inhibitors. Although preclinical trials using some of these compounds have been successful in regressing established tumors, their therapeutic efficacy in cancer patients has yet to be established. Furthermore, apoptotic or necrotic tumor death is a desirable effect, and specific agents that block only T-cell apoptosis have yet to be devised.

Resistance to Innate Immune Clearance

The innate immune system includes NK cells, macrophages, granulocytes, and the complement system and is triggered by the recognition of microbial nonself molecules. Human tumor cells lose the expression of the ligands for NK activation, and this constitutes a tumor escape mechanism for NK cell–mediated responses. Accordingly, activating the innate immune system by forcing glioma cells to express danger signals such as NKG2D ligands is a potential strategy for immunotherapy for these tumors.¹⁴⁹ However, the use of adenoviral vectors to express a variety of markers has been problematic in human clinical trials. NK inhibitory receptors can also regulate NK cell cytotoxicity and T cells. The nonclassic MHC class I molecule HLA-E, expressed on gliomas,¹⁵⁰ can transmit either an activating signal through CD94/NKG2C or an inhibitory signal mediated by CD94/NKG2A. By small interfering RNA (siRNA)-mediated silencing of HLA-E or blocking of CD94/NKG2A, NK and CTL activity could be enhanced, and this represents a potential therapeutic strategy; however, the prevalence of this type of immune evasion within gliomas has not been clarified.

Cytokine-Mediated Strategies

Recent advances in our understanding of the biology of the immune system have led to the identification of numerous cytokines that modulate immune responses (Tables 97-1 and 97-2).^167–169 These agents mediate many of the immune responses involved in antitumor immunity. Several of these cytokines have been produced by recombinant DNA methodology and evaluated for their antitumor effects. In experimental clinical trials, the administration of cytokines and related immunomodulators has resulted in objective tumor responses in some patients with various types of neoplasms.^170–172

IL-2 is an important cytokine in the generation of antitumor immunity.¹⁷² IL-2 has no direct toxic effect on cancer cells but mediates its antitumor activity by the modulation of the host’s immune response to the neoplasm. In response to tumor antigens, the helper T-cell subset of lymphocytes secretes small quantities of IL-2. This IL-2 acts locally at the site of tumor antigen presentation to activate cytotoxic T cells and NK cells that mediate systemic tumor cell destruction. In preclinical animal studies, cytotoxic immune responses were induced in animals with an intracerebral glioma treated with allogeneic IL-2–secreting fibroblasts administered intracerebrally.¹⁵⁶ In addition, it has recently been shown that IL-2 regulatory effects involve T regs.¹⁷³ Intravenous, intralymphatic, or intralesional administration of IL-2 has resulted in clinically significant responses in several types of cancer.^170–172174 However, severe toxicities (hypotension and edema) limit the dose and efficacy of intravenous and intralymphatic IL-2 administration.^170,171,174 The toxicity of systemically administered cytokines is not surprising because these agents mediate local cellular interactions, and they are normally secreted in quantities too small to have systemic effects. To circumvent the toxicity of systemic IL-2 administration, several investigators have examined intralesional injection of IL-2.^{152,163,175–177} This approach eliminates the toxicity associated with systemic IL-2 administration. However, multiple intralesional injections are required to optimize therapeutic efficacy.^175,176 These injections are impractical for many patients without potential significant morbidity, particularly when tumor sites are not accessible for direct injection.

Another cytokine that has been investigated for immunotherapeutic use in brain tumors is IFN-γ. IFN-γ induces the expression of MHC class I determinants, augments the sensitivity of tumor cells to cytotoxic T-lymphocyte–mediated lysis, and may make weakly antigenic neoplastic cells more immunogenic. Transfection of the IFN-γ gene into cells has led to an inhibition of tumor growth.^156,159

Similarly, studies using IL-4–secreting cells demonstrated increased survival and antitumor cytotoxic responses, suggesting that an immune response can take place against a tumor of the central nervous system in situ.^158,178,179 In other preclinical animal studies, a specific and significant peripheral systemic immunocytotoxic response (by cromium-51 release assay) was induced in animals with an intracerebral glioma treated with allogeneic IL-2–secreting fibroblasts administered intracerebrally.¹⁵⁶ Thus, the secretion of IL-2, or other cytokines, by the cellular immunogen, or an immunogenic derivative of the cells, may have altered the blood-brain barrier, enabling the immunogen to reach the spleen and lymph nodes in the periphery.^180,181

Cytokine gene transfer has resulted in significant antitumor immune responses in several animal tumor models.^{158,159,182,183} In these studies, the transfer of cytokine genes into tumor cells has reduced or abrogated the tumorigenicity of the cells after implantation into syngeneic hosts. The transfer of genes for IL-2,^153,182,183 IFN-γ,^156,159 and IL-4^157,158 significantly reduced or eliminated the growth of several different histologic types of murine tumors. Other cytokines capable of producing similar results include GM-CSF¹⁶⁰ and IL-12.^161,184 In the studies employing IL-2 gene transfer, the treated animals also developed systemic antitumor immunity and were protected against subsequent tumor challenges with the unmodified parental tumor.^182,183,185 Similar inhibition of tumor growth and protective immunity were also demonstrated when immunizations were performed with a mixture of unmodified parental tumor cells and genetically modified tumor cells engineered to express the IL-2 gene. No toxicity associated with expression of the cytokine transgenes was reported in these animal tumor studies.^{159,182,183,186} An alternative strategy is to genetically modify tumor cells to express an antisense gene to TGF-β, which is a cytokine highly expressed in glioma cells that acts to inhibit the function of cytotoxic T cells.⁶³

Although preclinical studies with cytokine gene therapy appear promising,^{157,160,162,187–189} clinical trials for brain tumors have been limited. Some of these trials have involved immunization with tumor cells modified with the IL-2 gene,¹⁹⁰ IL-4 gene,¹⁷⁹ or TGF-β2 antisense gene.¹⁶⁶ The toxic effects of cytokines in the CNS may limit the quantity that can be administered.^191,192 Neurologic effects and cerebral edema have been seen in animals injected intracranially with syngeneic cytokine-secreting cells.¹⁷⁷ In contrast, however, studies using allogeneic cytokine-secreting cells demonstrated an increase in survival and no long-term or short-term toxicity.^163,186

Most of the systemic toxicities of cytokine therapy should be avoided by the introduction of the gene for cytokine directly into the tumor mass, resulting in primarily local concentrations of the cytokine. This form of treatment is particularly attractive in the treatment of primary gliomas because these tumors usually only recur locally and are rarely metastatic. Another strategy to overcome the toxicity associated with systemic administration of cytokines is the use of genetically modified cells as a delivery vehicle for the cytokine of interest.

The toxicity of a cellular-based cytokine gene therapy for tumors is likely to depend in part on the survival of the genetically modified cells in the CNS. In studies investigating the survival of an allogeneic IL-2–secreting vaccine in the CNS by two different means, polymerase chain reaction and bioassay,¹⁸⁶ it was shown that allogeneic cells survive in the CNS less than 28 days. The cells, like other allografts, were rejected. The cells were well tolerated, and the animals did not demonstrate any significant neurologic or systemic toxicity. This suggests that cytokine-secreting allogeneic cells may serve as a safe and useful vehicle for the safe delivery of cytokines into brain tumors and supports the possibility and safety of using a monthly retreatment schedule in a clinical protocol.

In general, for the cytokine therapy clinical trials for gliomas conducted to date, the number of patients enrolled in these types of trials has been limited, and no large study showing significant efficacy has been reported.

Cellular Immunotherapy

Previous immunotherapy strategies initially used classic immunologic cell types, including activated lymphocytes and LAK cells. More recently, a variety of cells have been investigated for their usefulness in tumor oncology, including tumor cells themselves (syngeneic or allogeneic), DCs, and fibroblasts (syngeneic or allogeneic). Although syngeneic tumor cells have the advantage that they express most of the appropriate antigens needed for targeted therapy, many types of tumors are difficult to establish in culture. It is also conceivable that a subpopulation of the primary tumor, selected for its capacity to grow in vitro, may not reflect the tumor cell population as a whole, especially because tumors such as gliomas are known to be heterogeneous. In addition, cytokine gene therapies requiring the transduction of autologous tumor cells may not be practical for many cancer patients. Modification of neoplastic cells taken directly from tumor-bearing patients may be difficult. In particular, a primary tumor cell line, required for retroviral modification, has to be established. An alternative cell type that can be used for therapeutic immunizations is the DC, which is a specialized APC. Preclinical studies have indicated that immunizing either mice or rats with DC pulsed using tumor cell antigens can stimulate a cytotoxic T-cell response that is tumor specific and that engenders protective immunity against CNS tumor in the treated animals.^193,194

There are a number of advantages to using genetically modified allogeneic cells, such as an allogeneic fibroblast cytokine-secreting cell line as an “off-the-shelf” cellular vaccine. Fibroblasts obtained from established allogeneic fibroblast cell lines may be readily cultured in vitro and genetically modified to express and secrete cytokines.^{6,154,161,162,165,190} The cells can be subsequently injected directly into the tumor bed. The use of allogeneic rather than syngeneic cells was initially based on evidence that allogeneic MHC determinants augment the immunogenic properties of the tumor vaccine.^155,165 Application of genetically modified fibroblasts in therapeutic vaccines facilitates titration of single or multiple cytokine doses independent of tumor cell doses. Like other allografts, the allogeneic cytokine-secreting cells are rejected. Furthermore, the number of cells can be expanded as desired for multiple rounds of therapy. In addition, the slow continuous release of cytokines and the eventual rejection of the allograft may be a useful advantage in the treatment of brain tumors in which long-term secretion of high concentrations of certain cytokines may be associated with increased morbidity. Thus, an allogeneic cytokine-secreting vaccine is readily available, easily expanded, possibly less toxic, and more immunogenic. These considerations provide the rationale for examining the use of allogeneic fibroblasts genetically modified to secrete cytokines in further studies as a means of enhancing antitumor immune responses in treatment of malignant intracerebral tumors.^{154,156,161,162,165,190}

Two of the more promising cellular immune therapeutic approaches are DC immunotherapy and adoptive T-cell transfers. Earlier cellular-based studies included the use of alloreactive cytotoxic lymphocytes¹⁹⁵ or LAK cells¹⁹⁶ combined with IL-2. More recently, therapeutic immunizations have been with DCs. Several clinical trials have been completed with promising results at the University of California Los multiple centers. The DCs are loaded with acid-eluted peptides, tumor homogenates, or the tumor-specific antigen EGFRvIII after being stimulated and matured with GM-CSF and IL-4. The patients are then systemically vaccinated with the DC preparations. In one of the phase I clinical trials,¹⁹⁷ nine newly diagnosed high-grade glioma patients received three separate vaccinations spaced 2 weeks apart. A robust infiltration of T cells was detected in tumor specimens, and median survival was 15.1 months (compared with 8.6 months for a control population). A subsequent report¹⁹⁸ involving eight GBM patients produced a median survival time of 33.3 months, compared with a median survival of 7.5 months for a comparable group of patients receiving other treatment protocols. Currently, a phase II multi-institutional clinical trial is under way that is using DCs pulsed with tumor homogenate in GBM patients sponsored by Northwest Biotherapeutics. To generate this vaccine, the patients must undergo surgery to obtain the tumor to generate homogenate and have sufficient tumor volume.

In a phase I clinical trial (VICTORI) conducted at Duke University Medical Center employing an antigen-specific approach to EGFRvIII with DCs, 15 newly diagnosed glioma patients were enrolled (World Health Organization [WHO]; WHO grade IV, n = 12). After surgical resection, each patient underwent leukapheresis to obtain peripheral blood mononuclear cells (PBMCs) to generate the autologous DCs and for baseline immunologic monitoring. The first three patients received 3 × 10⁷ mature DCs per vaccination intradermally beginning 2 weeks after completion of radiation therapy, and the remaining patients received one third of their total generated DCs per injection (up to 1.1 × 10⁸ cells). Patients underwent magnetic resonance imaging (MRI) every 2 months for evidence of toxicity or tumor progression. Enrolled patients demonstrated immunologic responses without any evidence of adverse events aside from grade I and grade II local reactions at the vaccine injection site. Humoral and cellular immune responses that were specific for EGFRvIII were detected ex vivo after vaccination. Two patients were allowed on the trial with MRI enhancement at the time of vaccination, and both these patients had near-complete resolution of radiographic enhancement after the vaccine. Median time to progression (TTP) in patients with GBM (n = 12) was 6.8 months, and median survival was 18.7 months weeks.¹⁹⁹

In perhaps the largest study to date, De Vleeschouwer and associates vaccinated 56 recurrent GBM patients with autologous mature DCs loaded with tumor lysates derived from autologous, resected GBM. They found a median progression-free survival of 3 months and overall survival of 9.6 months. After accounting for other prognostic variables, they found that total resection of the recurrent tumor and an escalated vaccination schedule were predictive of progression-free survival. The only serious adverse event was cerebral edema in a patient with residual disease.

The research group at Cedars-Sinai Medical Center in Los Angeles reported the results of a clinical trial using chemotherapy after patients received the vaccine protocol. Mean survival for patients receiving only the vaccine was 18 months, with a 2-year survival rate of 8%, whereas those receiving both the vaccine and chemotherapy had a mean survival time of 26 months with a 2-year survival rate of 42%.²⁰⁰ The authors of the study hypothesized that the improved outcome was due to the vaccine having primed the apoptotic machinery of the cancer cells so that chemotherapy was then able to trigger the apoptotic pathway. However, an alternative explanation exists—the immunotherapy may be selecting the more aggressive, infiltrative tumor cells, leaving behind more chemosensitive cells. Regardless, this study indicates that a synergy may exist between immunotherapy and chemotherapy that needs to be further explored.

An alternative cellular vaccine approach is the amplification of the T cells that are generated by the individual cancer patient in response to tumor cells. Glioblastoma tumor cells gathered during surgery were cultured in the presence of growth factors and then injected subcutaneously back into the patients. After development of an immune reaction, the lymph nodes draining the location of the injection were resected to obtain lymphocytes directed against the tumor cells and were stimulated with staphylococcus toxin and IL-2. This generated a large number of activated T cells, which were infused back into the patient. Of 10 patients, 2 had tumor regression, but 8 patients had progressive disease.⁵³ It is unknown whether these T cells are functional once they enter into the tumor microenvironment. Techniques to maintain T-cell activation despite immunosuppressive factors are under development. Another inherent problem with cellular immunotherapy is the ex vivo manipulation of the cells—these techniques can be labor intensive, require special facilities for generating a product that can be infused into human patients, and will probably only be available at specialty centers.

Antibody Therapy

Monoclonal antibodies can be used to induce apoptosis, to mediate immune responses, or as a delivery vehicle for chemotherapeutics, toxins, or radionucleotides. One of the most exploited targets has been tenascin, an extracellular matrix glycoprotein, ubiquitously expressed in malignant gliomas but not normal brain. The monoclonal antibody 81C6 binds an epitope within the alternatively spliced fibronectin type III region of tenascin. In a phase II clinical trial of newly diagnosed GBM patients, the iodine-131–labeled murine 81C6 was injected into the patient’s surgically created cavity. Median survival times were 19.9 months for newly diagnosed GBM patients and 12 months for recurrent GBM patients, which significantly surpassed the historical controls even when accounting for established prognostic factors, including age and Karnofsky performance score.^201,202 Patient-specific dosing in newly diagnosed GBM patients further pushed median survival to 23 months.

Targeting cell surface receptors with immunotoxins provides another strategy in the antibody-directed immunotherapeutic treatment of brain tumors. The targeted agents are composed of a toxin derived from bacteria, like pseudomonas endotoxin or diphtheria toxin, coupled to either an antibody specific for a membrane antigen (immunotoxin) or a ligand specific for cell surface receptors (cytotoxin).²⁰³ The specificity of immunotoxin therapy depends on the presence or overexpression of unique receptors or antigens on tumor cells and not normal cells. Specific ligands used for this type of therapy have included the transferrin receptor,²⁰⁴ the epidermal growth factor receptor,²⁰⁵ and two cytokine receptors, IL-4²⁰⁶ and IL-13.^207,208 To target IL-4 and IL-13 receptors, recombinant fusion cytotoxins composed of IL-4 or IL-13 and a mutated form of pseudomonas exotoxin (PE) have been engineered. It has been shown that these chimeric cytotoxins are highly toxic in vitro to glioma cell lines, whereas normal cells are spared their cytotoxic effects.

An important consideration that arises in treating malignant gliomas with these agents is that the blood-brain barrier does not allow delivery of large molecules to intracranial tumors. Thus, several routes of administration have been tried, including intrathecal, intravenous, intra-arterial, and intratumoral. However, the development of convection-enhanced intratumoral delivery of macromolecules into the brain²⁰⁹ may eventually overcome this challenge. In initial clinical trials using IL-13 PE administered to patients with malignant gliomas using a convection-enhanced delivery system, the patients tolerated the infusion of this drug, and there was some modest prolongation of survival.^209,210 Ultimately, optimization of delivery will be needed to bring this approach into the realm of standard care.

Another related treatment strategy has been used that takes advantage of the fact that EGFR is overexpressed in a significant number of malignant brain tumors, and recombinant toxins targeting this growth factor receptor have therefore been developed.²⁰⁶ Once again, convection-enhanced delivery techniques are needed to aid in drug delivery, although problems involving ineffective infusion are a recurrent theme in patients treated with these agents. Nevertheless, some durable radiographic responses have been reported, suggesting the potential efficacy of these drugs in treatment of patients with malignant brain tumors.

One other potential problem that must be considered is that these tumors represent a heterogeneous collection of cells and surface receptors, and thus combination therapy in which immunotoxins and cytotoxins targeting different receptors, in combination with other therapies, such as chemotherapy and radiation, will likely be necessary to achieve clinical efficacy.

Vaccination Strategies

The genetic alterations occurring within gliomas²¹⁰ can provide specific targets for immunotherapy, but none are universally ubiquitous or maintain expression throughout the disease course. The implications of tumor heterogeneity must be recognized in the design of rational therapeutic approaches and ultimately may require either an individualized approach or targeting with multiple agents. A number of different vaccination strategies are being evaluated^211,212 and include approaches to vaccination with tumor-associated antigens (TAAs), for example, those based on (1) defined antigens or antigenic peptides, (2) tumor cell lysates or lysate fractions, and (3) whole irradiated tumor cells or apoptotic tumor cell bodies. Vaccines prepared using TAAs or TAA-derived epitopes presented by APCs are loaded onto DCs and used in clinical trials for patients with gliomas.^{178,197,213,214} However, these TAAs have to first be identified and then purified, requiring a tremendous effort requiring an “antigen discovery” step. Often the quantity of purified antigen must be increased to enable multiple immunizations of the cancer patient. While new TAAs are being discovered, the question of which TAA to be used in the vaccine is uncertain and extensively debated. The choice of TAA is not a trivial decision. Not only are isolation and purification of TAAs or antigenic peptide highly labor intensive, but it also remains uncertain whether TAA- or peptide-based vaccines are superior to tumor cell vaccines. Selection of the immunizing antigen is generally based on its abundant expression in the tumor and lack of expression in normal tissues. Antigens that do meet this criterion may not always be immunogenic. Furthermore, the heterogeneity of antigen expression in the tumor cell population is likely to be a concern. Some tumor cells may not express the antigen chosen for therapy. For example, de Vries⁸⁰ found that expression of known tumor antigens such as gp100 and tyrosinase was variable in different melanoma lesions in the same patient. Not all the malignant cells in the patient’s neoplasm expressed these antigenic determinants. Because the tumor cell population is heterogeneous, tumor cells that fail to express the defined antigen chosen for therapy are likely to escape destruction by the activated immune system and are likely the source of recurrent tumor.

The use of tumor lysates, lysate fractions, or apoptotic tumor bodies for vaccination overcomes some of the tumor heterogeneity limitations. However, preparation of the vaccine requires the availability of autologous tumor, often in substantial quantities. Vaccines have been prepared by culturing patient-derived DCs and then pulsing or “feeding” the cells tumor lysates or apoptotic bodies. However, DC-based vaccines are laborious and costly to prepare. Their efficacy in generating antitumor immune responses capable of tumor rejection remains unproved. In all cases, the optimal adjuvant, protein or peptide concentration, ratios of DCs or apoptotic bodies and routes of delivery, and immunization schedules remain undefined. They may be critical for success. Immunization by injection of “naked” plasmid DNA or RNA encoding a tumor antigen is also currently under evaluation. However, as for immunization with defined antigens, there is a danger that the antigen specified by the polynucleotide chosen may not be the most appropriate for CTL or helper T-cell generation. Anichini and associates²¹⁵ found that CTLs in melanoma patients are not always directed toward known melanoma antigens, such as Melan-A/Mart-1, MAGE-3, gp100, or tyrosinase. It is likely that there are multiple other tumor antigens in addition to those previously identified that are expressed by different cells that comprise the malignant cell population. Only certain of these TAAs are able to induce tumor-specific immune responses. The identification of the most “clinically relevant” tumor antigen cannot always be accomplished a priori, without extensive preclinical studies. Even then, subsequent validation in the patient may not confirm that these are “tumor rejection” antigens.

Another strategy involves preparation of vaccines by transfer of tumor DNA or RNA into an immunizing cell. The major advantage of this approach is that TAAs do not have to be purified or produced in large quantities. DNA-based vaccines are able to elicit robust and long-lasting activation of the immune system, which results in tumor rejection.^163,164 In comparison with protein vaccines, DNA-based vaccines provide prolonged expression and direct presentation of tumor antigens. This offers an opportunity for the development of effector as well as memory immune responses to many different epitopes encoded by the tumor-derived DNA. From a practical point of view, these vaccines are easy and relatively inexpensive to prepare. Unlike the tumor lysate and homogenate strategies, these vaccines can be prepared from only a limited quantity of tumor DNA, which can be obtained from small surgical specimens or even biopsies. In the case of a recipient fibroblast, these cells can be selected to meet the requirement for rapid expansion in culture and MHC restriction. DNA-based vaccines offer a number of important advantages, which is greatly encouraging for their further development for cancer immunotherapy. The use of tumor vaccines as a “protective treatment” introduced after tumor resection may play an important role in delaying tumor recurrence in the clinical setting.

An ideal immunologic approach to cancer treatment is an off-the-shelf approach, similar to the commercially available vaccines. One such immunotherapeutic approach for newly diagnosed GBM is the vaccine CDX-110 (PEP-3-KLH), which is a peptide-based vaccination approach, again targeting the tumor-specific antigen EGFRvIII. Preclinical data demonstrated that when mice with established intracerebral tumors were treated with a single dose of PEP-3-KLH, there was a significant increase in long-term and median survival.¹¹ A phase II multicenter trial (ACTIVATE) was conducted at Duke University Medical Center and the University of Texas M. D. Anderson Cancer Center and enrolled 18 newly diagnosed GBM patients with tumor expression of EGFRvIII.⁸⁴ After completion of radiation therapy, patients received three vaccinations at 2-week intervals of PEP-3-KLH with GM-CSF. Subsequent vaccinations were continued monthly until tumor progression was evident. Adverse events consisted mostly of grade I and grade II vaccine site reactions. Both humoral and cellular immune responses to the EGFRvIII were enhanced after vaccination. Median PFS in this trial was 14.2 months and not significantly different between sites (P = .3445) and compared favorably to an institutional historical cohort matched for eligibility criteria and adjusted for age and Karnofsky performance score (n = 17), in which the median PFS was only 6.3 months. The median survival time for all study subjects was 26 months, which again compared well with the historical data, even after adjustments for age and Karnofsky performance score. Further evidence for efficacy was provided by the observation that among the patients with recurrent tumors in whom pathologic material could be obtained at recurrence, 82% lost EGFRvIII expression, suggesting a possible mechanism of failure involving antigen escape. The results are also notable for the fact that the expression of EGFRvIII on the GBM is a negative prognosticator in patients surviving more than 1 year. Typically, the median survival is 12.8 months for patients with EGFRvIII-expressing tumors, and no patient survives more than 18 months.²¹⁶ The ACTIVATE clinical trial was conducted before temozolomide became the standard of care,²¹⁷ so a second phase II clinical trial (ACT II) combined the CDX-110 vaccine with temozolomide.

One issue that arises for the peptide vaccine strategies is that the EGFRvIII is not ubiquitously expressed, and therefore many GBM patients are not candidates. In a phase I/II clinical trial at the University of California San Francisco, the vaccination strategy circumvented the identification of a precise tumor antigen by vaccinating with heat shock proteins, which act as immune chaperones for the tumor antigens.^217a The heat shock proteins are purified from the resected tumor by antigenics and then subsequently administered back to the patient. The patients who received the heat shock protein had induced immune response. Efficacy data are pending; however, this type of approach is limited logistically by the volume of tumor that needs to be resected to generate a vaccine.

Synergy of Chemotherapy with Immunotherapy

Conventional wisdom has been that chemotherapy and immunotherapy are two antagonistic forms of therapy. Model systems have demonstrated that the depletion of immune cell subsets can abrogate the efficacy of several types of immunotherapeutic approaches,^11,82 indicating that chemotherapy administered during the effector stages of immunotherapy may be deleterious to efficacy. However, this does not preclude using chemotherapeutic agents when appropriately timed with immunotherapy or at lower doses to minimize the aforementioned effects. Certain chemotherapy drugs do, in fact, have positive immunomodulatory properties that stimulate the production of proinflammatory cytokines (TNF, IL-12, and IL-1) and increase NK cell antitumor and LAK cell activity.²¹⁸ Furthermore, the depletion of certain effector cells, such as T regs, may be a highly desirable outcome of chemotherapy, yielding greater immunotherapeutic efficacy, or may promote a desirable cytokine profile for adequate tumor control. One particular area of investigation that has been underexplored is an analysis of the specific immunologic effects, or the lack thereof, of various chemotherapeutic agents. Of note, the immune modulatory effects of a particular chemotherapeutic agent cannot be extrapolated to all, and caution needs to be employed in using combination therapy. For example, the number of patients able to generate an immune response to a peptide was diminished and toxicity increased when they were vaccinated with IL-2 during chemotherapy.^219,220 Recent studies by Ahmadzadeh and coworkers²²¹ have shown that the administration of IL-2 in patients with metastatic melanoma and renal cell carcinoma leads to a significant increase in CD4⁺, CD25^hi, FoxP3⁺ T regs. Patients were infused with a high-dose bolus of IL-2 (720,000 IU/kg) intravenously every 8 hours to a maximum of 15 doses. About 20% of the patients demonstrated a clinical response, although the mechanism of action remains unclear. Moreover, there was a significant in vivo expansion of the T regs, and patients experiencing that fared much worse. Therefore, chemotherapies that induce strong IL-2 responses may not be desirable.

In recent animal studies, the combination of a cellular vaccine consisting of cytokine-secreting allogeneic fibroblasts and the drug thiazolidinedione (a nuclear hormone receptor involved in metabolism, signal transduction, apoptosis, and differentiation) led to increased survival and specific cytotoxic responses.²²⁵ The improved efficacy of combination drug with vaccine therapy suggests that these drugs may enhance immunotherapy by inhibiting immunosuppression (as described previously) or may actually augment the immunogenicity of the tumor itself, through either apoptosis or alteration of membranes or proteins, changing the antigenicity of the tumor. In the clinical trials that have attempted to address the mechanisms of immunologic failure, certain mechanisms predominate. In the case of adoptive T-cell transfers, many effector cells are administered, which can potentially overwhelm the T reg population. Clinical responses to this have been documented, especially in melanoma patients. Nevertheless, eventually the disease recurs, which has been attributed to the T regs regaining hemostasis and eventually predominating. Several agents have been identified that can reduce the relative T reg population. These agents include temozolomide, CTX, and denileukin diftitox. Thus, with clinical trials that are using adoptive T-cell transfer, it would be logical to determine when T regs become prominent. When a patient experiences an elevation of the T reg population, an agent that depletes the T regs can be administered, which may further enhance and prolong the clinical efficacy observed in such patients. A greater understanding of how these cells are involved in the basic biology of cancer patients is needed. In the specific circumstances of glioma patients, we do not know how the various phases of treatment, such as surgical resection, radiation, or recurrence, influence the T reg population, but we do know that the influx of T regs is not ubiquitous across all types of gliomas and is heterogeneous even within GBMs.²²³

One could consider that part of the observed clinical efficacy of temozolomide may be attributable to the trafficing to the glioma micro-environment rather than solely to its direct antiglioma effects. When a vaccination is administered during the nadir of the temozolomide level, there are likely enhanced effector responses.⁵⁷ Such effector responses may be secondary to a lag in the recovery of T regs, allowing a greater clonotypic expansion than would otherwise occur without the temozolomide. Given the fact that temozolomide has been established as a standard of care for glioblastoma patients, in future trials, it is difficult to justify failing to administer this to patients.

A phase II clinical trial was undertaken at University of Texas M. D. Anderson Cancer Center and Duke University Medical Center to assess the immunogenicity and efficacy of an EGFRvIII-specific peptide vaccine in patients with newly diagnosed EGFRvIII-expressing GBM in combination with simultaneous standard or continuous temozolomide.⁵⁷ After resection and radiation and concurrent temozolomide, consecutive cohorts received subsequent monthly cycles of 200 mg/m² or continuous 100 mg/m² temozolomide simultaneous with intradermal vaccinations with an EGFRvIII-specific peptide (PEP-3) conjugated to keyhole limpet hemocyanin (KLH) until tumor progression or death. Twenty-two patients were enrolled. There were no significant differences in vaccine immunogenicity, progression-free survival, or overall survival between temozolomide regimens. Although temozolomide induced grade II lymphopenia in 53.8% of patients, the coadministration of temozolomide with the EGFRvIII vaccine (CDX-110) resulted in strong sustained immune responses to EGFRvIII in 100% of evaluated patients. Median progression-free survival was 15.2 months.²²⁴ Median survival was 23.6 months. These data indicate that the concurrent administration of temozolomide with immunotherapy does not ablate immunologic or clinical efficacy and is the current design of the registration trial by Celldex.

Therapeutics