Malignant primary brain tumors now cause more deaths each year than some of the most notoriously prevalent malignancies, including melanoma.
Ultimately, the infiltrative nature of the tumor, the impracticality of optimal resection, and the comparative intolerance of the normal brain for cytotoxic therapies create the need to pursue more specific treatment modalities.
The immune system may provide some direction in this pursuit; as such, desired specificity is the hallmark of its normal surveillance function.
The suggestion that the immune system might target tumors is admittedly not new.
Attempts to derive tumoricidal cancer vaccines date back to at least 1909 with Coley's efforts to treat inoperable sarcoma with bacterial toxins.
It was not until many years later that pioneering work by Zinkernagel, Van Pel, and Boon established that the immune system, via CD8+ cytotoxic T lymphocytes (CTL), can kill neoplastic cells following recognition of tumor antigens bound to tumor cell major histocompatibility complex (MHC) class I molecules.
However, classical ideas regarding antigen presentation and CTL priming presumed the in vivo presentation of MHC I-restricted tumor antigens to CD8+ T lymphocytes to be exclusively a function of the tumor cell itself.
The discovery of the dendritic cell (DC) and the resultant understanding of professional antigen presentation have revised the classical thinking and ushered in a new approach toward cancer immunotherapy.
Steinman and Cohn championed the DC in 1973 as a novel stellate cell present in the spleens of mice.
Shortly thereafter, the concept of a professional antigen-presenting cell (APC) evolved.
Studies demonstrated that host MHC-restricted CTL could be primed in vivo to generate cytotoxic responses against antigen introduced on transplanted MHC-disparate cells.
Since the transplanted cells were not syngeneic, they would be incapable of stimulating directly a CTL response to the antigen presented in context of the incompatible MHC.
It was thus concluded there existed an intermediate cell type that negotiated both the uptake of the antigen (released, for example, from dead or dying MHC-disparate cells) and its re-presentation on host (i.e. non-disparate) MHC in order to manufacture and evoke an antigen-specific T-cell response.
The same studies also highlighted the ability of these bystander cells to present MHC class-I restricted minor histocompatibility antigens in vivo.
The existence of an intermediate or bystander cell capable of stimulating cytotoxicity would indeed require the ability to take up antigen exogenously and present it on MHC I, despite the classical view that this mode of presentation was reserved for endogenously produced antigen.
The capacity for these specialized APC to ingest exogenous antigens and present them on MHC I molecules, (as well as on the expected MHC II), became known as cross-presentation.
Although it was understood that other populations, such as macrophages and B-lymphocytes, possessed an antigen presenting function (consistent with their ability to present via MHC class II), DC, following their discovery, were soon established as the fundamental APCs.
By the early 1990s, a firm role for the DC system in immunogenicity was recognized, this encompassing more specifically the functions of peripheral antigen capture and presentation; migration to the T-dependent areas of lymphoid organs; and presentation to and activation of naive T-cells.
Indeed, it is now known that immature DC develop from hematopoietic progenitors and situate themselves at sites of interface with the environment, such as the skin (i.e. epidermal Langerhans cells) and the mucosa.
Here, in the absence of any provocative inflammatory stimuli, they continuously sample both self and non-self antigens, including those obtained from live cells and other DC.
These antigens they load somewhat inefficiently onto both MHC I and MHC II molecules for presentation.
However, in the context of an inflammatory or 'danger' signal (i.e. microbial products such as lipopolysaccharide (LPS), viral dsRNA, and CpG DNA; mechanical stress; or inflammatory cytokines such as IL-1, IL-2, and TNF-α), the immature DC is obliged to undergo a maturation program.
This entails down-regulation of its antigen uptake capabilities; up-regulation of adhesion and co-stimulatory molecules and chemokine receptors; and increased synthesis and surface transport of long-lived MHC molecules, particularly class II.
The final product is a mature DC that is selectively recruited to the paracortical regions of lymph nodes, where it encounters antigen-specific, naive CD4+ and CD8+ T-cells.
Further activation may be elicited by CD4+ T-cell help and enhances the DC's capacity to mediate cellular immunity and prime CTL.
Ideally, by connecting the processes of antigen-uptake, MHC loading, and APC migration to the presence of a danger signal, DC may ultimately present preferentially those antigens encountered in the context of maturation signals and, hence, may stimulate primarily those T-cells specific for pathogenic antigens.
In 1994, investigators provided the first demonstration that tumor antigens were exclusively presented by host bone marrow-derived cells, understood to be DC.
Several observations continued to suggest that DC played a role in physiologic antitumor immunity, and pathology reports indicated a positive correlation between DC tumor infiltration and prognosis.
However, studies outlined defects in professional APC within tumor-bearing hosts, and several morphologic and functional changes in DC were described in patients with cancer.
Hypotheses evolved that in vivo hindrances to tumor antigen presentation by DC might explain failures in the physiologic antitumor response.
This was supported by the absence of in vivo evidence for capture and processing of tumor cell antigens by DC and by reports that tumors elaborated factors, such as vascular endothelial growth factor (VEGF) and IL-10 that inhibited the functional maturation of DC.
With the reality of in vivo hindrances to antigen presentation gaining acceptance, and with the capacities of DC for priming CTL understood, the expansion and loading of DC with tumor antigens ex vivo were advocated as first steps in a potentially powerful vaccination strategy for inducing antitumor immunity.
In one of the first clinical trials, B-cell lymphoma patients were vaccinated with DC loaded with tumor-specific, idiotypic antigens.
B-cell malignancies express immunoglobulin receptors that are tumor-specific, insomuch as they can be distinguished from those of normal B-cells by virtue of specific idiotypic determinants.
As the malignancies are monoclonal, all the cells of a given tumor express identical immunoglobulin receptors, making the task of identifying a ubiquitous, tumor-specific antigen considerably less daunting.
Therefore, following the development of methods for the isolation of human DC from peripheral blood, investigators initiated a pilot study to investigate the ability of DC pulsed with tumor-specific idiotype protein to provoke antitumor immunity.
All patients developed measurable antitumor cellular immune responses; anti-idiotype antibody responses were notably absent.
One patient experienced a complete tumor regression; a second had a partial regression, and a third patient with equivocal pre-vaccine findings (but molecular detection of disease) appeared to resolve all evidence of disease.
No significant side effects were reported, and the feasibility of DC-based antitumor immunization in humans was demonstrated.
DC have since served clinically as a platform for immunizations against other cancers, including melanoma, prostate cancer, renal cell carcinoma, non-small cell lung carcinoma, and colon cancer.
Due to the altered immune status of many late stage cancer patients, the validation of DC function in an immunized host can be difficult.
Therefore, a clinical trial in normal volunteers receiving antigen alone or unpulsed DC failed to develop specific immunity.
On the other hand, priming of CD4+ T-cells to keyhole limpet hemocyanin (KLH) and boosting of tetanus immunity were observed in patients immunized respectively with KLH- and tetanus toxoid (TT)-pulsed DC.
Definitive in vivo proof was thus obtained that DC immunizations can prime in humans an antigen-specific, cell-mediated immunity.
As most conventional vaccines elicit primarily humoral immunity, the demonstrated ability of DC-based vaccines to negotiate a combined CD4+ and CD8+ T-cell immunity has fairly broad immunologic significance.
Numerous preclinical studies have indicated that immunizing either mice or rats with DC pulsed with tumor cell antigens can prime a CTL response that is tumor-specific and that engenders protective immunity against CNS tumors in the treated animals.
In one notable study, DC was pulsed with tumor homogenate from a syngeneic astrocytoma cell line that arose spontaneously in the mice used.
This tumor model was unique, insofar as the targeted tumor was not chemically induced and was more likely to mirror the antigenic properties of human tumors that arose spontaneously.
Furthermore, the cell line secreted a biologically active form of TGF-β, modeling the immunosuppressive abilities of human gliomas.
Following pulsing, DC were administered to mice in four weekly intraperitoneal (i.p.) injections; animals were later challenged with intracranial (i.c.) injection of tumor cells.
Immunization produced a >160% increase in survival in treated mice, and 50% survived long-term.
Immunologic memory was demonstrated by survival of mice rechallenged with tumor; both cell-mediated and humoral immunity were induced.
The early preclinical results offered promise that DC would be a more effective platform than other immunotherapeutic strategies for brain tumors, and their use has now proceeded to clinical trials.
In the initial clinical study, systemically ascertainable cytotoxicity successfully developed in four out of seven testable patients who received DC pulsed with MHC-I peptides eluted from the surface of autologous glioma cells.
Furthermore, two out of four that underwent re-operation demonstrated robust CD8+ and memory (CD45RO+) T-cell infiltrates in areas of tumor.
Based on the small sample size, no reliable data on survival could be generated, but the treatment proved safe.
One of the potential side effects of immunotherapy is the generation of immunity against self.
Such autoimmunity may even be tolerable in some cases, i.e. in the generation of immunity against normal prostate during immunotherapy of prostate cancer.
This is decidedly not so in the brain, where autoimmunity can assume a particularly lethal form: allergic encephalitis.
The induction of lethal experimental allergic encephalitis (EAE) has been described in primates and guinea pigs after vaccination with human glioblastoma tissue.
This has raised concerns that vaccination with DC pulsed with unfractionated tumor-derived antigens may similarly elicit autoimmunity.
These concerns have gathered espousal for pulsing DC with tumor-specific antigens or tumor RNA, the latter of which could allow the use of subtractive hybridization to reduce the number of shared antigens between tumor and normal CNS.
It has become clear that many antigens shared by tumor and other tissues can be recognized by CTL that are allegedly tumor-specific.
Therefore, an effective antitumor immune response may very well tip the scales toward immunity against self.
Indeed, in melanoma models, an association between high frequencies of Melan-A-specific CTL and the development of autoimmune vitiligo has been reported.
Moreover, clinical studies have demonstrated that induction of autoimmune depigmentation is a marker for successful control of melanoma by the immune system, and accordingly, is linked to a more favorable prognosis.
The fear exists then that autoimmune phenomena, such as EAE, might be obligatory side effects of any truly effective anti-CNS tumor vaccine.
This fear would seem to gain validity from studies that have optimized conditions for DC-priming of EAE.
One study demonstrated that presentation of the self-antigen myelin basic protein (MBP) by DC to transgenic CD4+ T-cells specific for an MBP epitope is sufficient for the generation of EAE in mice.
In other studies, DC specific for tumor antigens concomitantly expressed in peripheral nonlymphoid organs induced severe and even fatal autoimmune diseases, such as autoimmune diabetes, arteritis, myocarditis, and dilated cardiomyopathy.
Although autoimmunity has thus far been absent among initial trials in brain tumor patients, the importance of maintaining some sort of tumor specificity in vaccine design seems clear.
Immunization with DC is an attractive route to pursue in the search for more specific and more effective treatments for malignant brain tumors.
Currently, though, no consensus exists on the optimal DC subtype, generation, loading method, maturation, dose, or route of delivery.
An understanding of the roles that each of these factors plays in influencing clinical outcomes is essential.
To date, before being used in immunizations, DC have been most commonly pulsed with tumor extract or lysate; apoptotic bodies; RNA; acid-eluted peptides; synthetic peptides; and tumor-associated antigens or tumor-specific antigens.
Responses to antigens expressed broadly by a variety of cancers have also been studied.
The injection site used in immunization protocols is an important consideration.
In one study with mice, intravenously (i.v.) injected DC mainly accumulated in the spleen, whereas subcutaneously (s.c.) injected DC homed preferentially to the T-cell areas of the draining lymph nodes.
Moreover, the accumulation of the s.c. administered DC in the lymph nodes correlated with the induction of antitumor immunity.
In the case of humans, in vitro-generated, antigen-loaded, human DC injected i.v. localized to the lungs and then redistributed to the liver, spleen, and bone marrow, but were not detected in lymph nodes or tumors.
A small percentage of DC injected i.d., however, migrated rapidly to the regional lymphatics in some individuals.
No lymph node activity was detected after s.c. injection.
DC, to date, have not been detected in the normal adult brain parenchyma, and their absence has been considered a component of the immune privilege of the CNS.
The presumptive pool of local APC has been astrocytes and microglia/macrophages, which are decidedly less effective than DC in T-cell priming.
However, although seemingly absent from the normal CNS parenchyma, MHC class II+ dendriform cells appear to be located at potential sites of antigen entry, such as the meninges and choroid plexus.
Furthermore, in a variety of experimental models of brain inflammation, DC have been inveigled to appear intraparenchymatically, inviting the question of how such DC come to arrive in the CNS.
Hypotheses have been that they (1) develop from infiltrating blood monocytes in the context of inflammation; (2) get recruited from the meninges; (3) differentiate from perivascular macrophages; or (4) represent a population of brain DC that develop locally from intracerebral progenitors or resident microglia in the wake of inflammatory stimuli.
There is currently a line of evidence supporting this latter possibility, although the issue is undecided.
Similarly undetermined are the migration patterns and destiny of functional DC that arrive in the CNS.
In one study, DC injected into the brain migrated preferentially into the white matter tracts and elicited a disproportionate recruitment of CD8+ T-cells into the CNS.
However, while these injected DC tended to home to the T-dependent regions of cervical lymph nodes, some studies of physiologic antigen presentation in the brain have implied an additional in situ APC activity for DC appearing in the CNS, as these DC are often found clustered with T lymphocytes.
The questions of CNS DC origin and migration should be areas of further research and may be relevant to the issue of delivery for DC-based brain tumor vaccines.
For example, if DC derive from local progenitors within the brain and play some function in the recruitment of CD8+ T-cells to the CNS, then intratumoral or intra-CNS vaccination may be of benefit.
On the other hand, if brains DC do indeed migrate to the cervical lymph nodes and prime a local immune response possessing enhanced CNS trafficking, then i.d. or s.c. injections near the cervical nodes may be warranted.
There is still no evidence, though, that any of these alternative routes of delivery would offer advantages over peripheral vaccination.
If one assumes that T-cells primed in the periphery are capable of trafficking into the CNS and locating their target, then the site of T-cell priming may prove to be irrelevant.
Indeed, it has been demonstrated that subcutaneous vaccinations delivered peripherally are capable of eliciting CD8+ T cell mediated immunity against tumors harbored within the CNS.
The relative ease and safety of peripheral vaccinations should likely then not be supplanted until a final determination of the origin of brain DC and their possible route(s) of migration can be made.
Our evolving knowledge of antigen presentation, T-cell activation, and antitumor immune responses contribute a logical appeal to a DC-based vaccination platform.
Initial results with DC immunizations are encouraging, but information and consensus must be gained regarding methods of preparation and of antigen loading.
DC subsets need to be identified and characterized, and the pathways of antigen uptake and cross-presentation further studied.
The discovery of new tumor-associated antigens and tumor-specific antigens should aid in the tailoring of more encompassing yet tumor-specific immune responses, and future goals can include the development of means to directly expand DC in patients and load them with antigen in vivo.
Until such times, however, results obtained in preclinical models and clinical trials continue to provide DC-based immunization strategies with credibility, as well as to offer hope to patients battling with cancer.