TUMOR IMMUNOLOGY

Co-Chairs: Darell D. Bigner, M.D., Ph.D., and Richard M. Ransohoff, M.D.

Participants: David Ashley E. Antonio Chiocca Glenn Dranoff Carol Krutchko Andreas Kurtz Lois Lampson

Linda M. Liau Sherie Morrison David Parkinson Sam D. Rabkin Tom Rozman John H. Sampson

Diane Traynor Carol Wikstrand Han Ying Richard Youle Jeanne Young John S. Yu

STATEMENT OF THE PROBLEM

There is a current lack of basic information concerning the functioning of the immune system within the central nervous system (CNS). It is imperative to narrow this sizeable gap in understanding of the interaction of these two systems in order to gain insight into the pathogenesis of primary brain tumors and the immunotherapeutic approaches that are most likely to produce successful outcomes in their treatment.

For broad purposes of discussion in this report, tumor immunotherapy may be considered to be both cell mediated (T-lymphocyte cell) and antibody mediated. Both types of immunotherapy are characterized by great specificity and low toxicity, as long as they are carefully administered. Current T-cell strategies involve recruiting tumor-reactive T-cells after exposure of autologous professional antigen-presenting cells to tumor antigens, and reinfusing them. Tumor-specific antibodies or their smaller fragments, such as single-fragment chains, are generated by hybrid technology from conventional or double-knockout/double-transgenic mice or from phage display systems. Antibodies may kill tumor cells in "unarmed" fashion or by directing drugs, radionuclides, or toxins to tumor cells.

CHALLENGES AND QUESTIONS

The areas in which further information would be most beneficial include the following:

• Identifying the antigen-presenting cell(s) (APC) of the CNS: For example, is there a CNS APC, or must we use APCs such as peripheral immune system dendritic cells?

• Determining the extent and pattern of the lymphoid drainage pathways of the CNS: Evidence for lymphoid drainage pathways in the CNS has been found in animals, but little is known of their existence or characteristics in humans.

• Elucidating the mechanisms or pathways by which marrow-derived immunocompetent cells traffic to the CNS: Chemokines, which are important in this process, are produced by resident neural cells, including glioma cells, and are essential for leukocyte recruitment to the normal and inflamed CNS. Chemokines also provide growth regulatory signals for glia during development and neoplasia and thus may play a complex role in the response of the CNS host organ to tumor.

• Identifying the important immune effector mechanisms in the CNS, both cell mediated and antibody mediated, and complement

• Identifying the antigens expressed on CNS tumors that also have encephalitogenic potential: Powerful immune responses underlie the relatively rare paraneoplastic syndromes, and immunotherapy targets potentially exist inside the CNS. However, unless directed appropriately, antigenic therapy bears the risk of extensive CNS destruction.

• Further understanding of the mechanisms by which CNS tissues and CNS tumors exert both local and systemic immunosuppressive effects: Although brain tumor-related immunosuppression is not severe enough to cause opportunistic infections, these effects may interfere with cell- and antibody-mediated tumor immune mechanisms.

BARRIERS

The development of predictive large and small animal models of primary CNS tumors is of great importance in the application of immunotherapy to primary brain tumors. Animal models have yielded disconcertingly little cross-species applicability in the treatment of brain tumors in the past; for example, the pharmaceuticals that have been found to be effective in mice with experimental allergic encephalomyelitis, a murine model of multiple sclerosis, have not proven effective in human trials. There are currently no predictive animal models for immunotherapy. However, the development of genetically modified mice provides promise of the ability to extrapolate the findings of mouse studies more effectively to human applications.

In mice, such models could be used to address critical problems, including the effector mechanisms by which immune responses to tumors can eliminate neoplastic cells. The creation of animal models of the paraneoplastic, immunologically mediated disorders of the CNS would also be advantageous for the examination of immunologic effector mechanisms.

Another barrier to optimal progress in tumor immunology is the lack of methodological consistency. Different immunotherapy groups use differing immunization strategies, thus diminishing the ability of clinicians to reach significant conclusions by comparing their results across trials. For example, at least four institutions are currently involved in dendritic cell trials, but each group uses a different method of cell preparation. Another area in which establishing a consistent methodology would be beneficial is the key investigative analyses of the immune responses of patients involved in clinical trials. Finally, although generation of CTL responses is not generally considered a gold standard for therapeutic success, it is an important technical outcome measure and should be determined in similar fashion by all investigators. Such is currently not the case.

There is a strongly felt need among immunotherapists that some outcome measures short of survival would be advantageous in evaluating investigative therapies rapidly and efficiently. Possible outcome measures could involve well-defined immunological endpoints and determination of toxicity or demonstration that the desired immune response has indeed been elicited.

An additional barrier to obtaining the maximal benefit from clinical trials of immunotherapy has been the lack of an optimal data set from all patients entered in those trials. For example, stratification for the degree of pre-treatment immunological suppression is not always done. It has been suggested that future initiatives should include, for instance, a complete proteomic analysis of the components of acid-washed membrane preparations used to pulse dendritic cells, and ultimately molecular knowledge of each antigen employed.

A common theme of the preceding paragraphs, following but by no means eclipsing the discussion of animal models, is the need for consistency: in immunization strategies, in a consensus on the use of outcome models short of survival, and in obtaining the most complete and clinically-informative data set possible from patients in clinical trials.

RESEARCH AND SCIENTIFIC PRIORITIES

Among the areas of tumor immunology considered to be of greatest value and interest in the development of brain tumor immunotherapy, the following topics were selected by a consensus of the PRG participants as most deserving of research focus and funding:

Priority 1: Develop techniques of antigen identification, resulting in a readily accessible source of information on the genes and gene products that produce antigens.

A preliminary database already exists containing information characterizing genes that are differentially expressed in tumor cells as opposed to those typically found in non-neoplastic tissue.

• To accomplish antigen identification, researchers will require high-throughput screens to define the antigens recognizable by T-cells.

• Similarly, high-throughput screens are needed to permit the identification of cell surface antigens and antigens located in the extracellular matrix, which may be approachable by antibody-targeted therapy.

Priority 2: Characterize both CNS and systemic immune responses in patients with brain tumors.

Protective (tumor-destructive) responses are poorly understood. Deleterious reactions are frequently generated by interactions between neoplastic cells and the immune system response apparatus. These interactions are capable of generating tumor growth factors, angiogenic factors, and immunosuppressive components.

Tumor-associated immunosuppression is little understood, yet clinicians must learn to control this mechanism and develop means of dealing with it. Helpful investigatory areas would include characterization of the underlying mechanisms, implications for the natural history of the tumor, and consequences of the immunosuppressive reaction for the immunotherapy of the tumor.

Priority 3: Consider the problems and challenges posed by patient and tumor heterogeneity.

Individual patients are heterogeneous, for both genetic (e.g., human leukocyte antigen) and epigenetic reasons, in their abilities to respond to immunotherapy. Tumors are heterogeneous with regard to antigen expression. Heterogeneity needs to be understood in each tumor so that as many as three or four individual molecular antigenic species may be targeted at one time. Target cells also display heterogeneity with regard to intracellular components that confer susceptibility to enzyme- or Fas-mediated programmed cell death.

RESOURCES NEEDED

• Brain Tumor Clinical Consortia for expediting chemotherapy and radiation therapy are funded by the National Cancer Institute (NCI) and are productive. Similar clinical consortia for immunotherapy, especially for Phase I and II trials, are needed to expedite and standardize approaches. There are enough institutions (five or six) ready to form the critical mass necessary for such consortia. Human trials are more important than animal studies. For example, cytokines are species specific, and human cytokines can be evaluated only in primate systems.

• Dedicated support from the National Institute of Neurological Disorders and Stroke (NINDS) to understand basic mechanisms of immune responses in the CNS is important for demyelinating and viral diseases as well as for brain tumors.

• The voluminous data from genomic differential displays have revealed many expressed genes in brain tumors not present in normal CNS. Before such findings can be used for immunotherapy, high-throughput screening must be developed and used to identify linear peptide T-cell and conformational antibody-targeted antigens. Workshops may be necessary for implementation.

• The NCI Rapid Access to Intervention Technology (RAID) program for biologics should be expanded to include NINDS investigators.

• Brain Tumor Immunotherapy Center Grants, Program Project Grants, and SPORES should be solicited by NCI and NINDS.

• Centralized transgenic and scientific model shared resources are needed. An especially important animal need is the availability of double-deletion/double-transgenic mice with murine immunoglobulin genes replaced with human immunoglobulin genes. Such mice can be used for production of fully human monoclonal antibodies for repetitive administration to patients with brain tumors and other cancers.

• Clinical trial support mechanisms are needed to cover the cost of research activities, including immunological assays, scanning, and autopsies, not covered by third-party payers. The National Institute on Aging successfully worked with investigators and advocates in dementia to obtain funding for research autopsies. Advocates should encourage federal legislation to provide a payment mechanism for research autopsies.

• Accessible proteomic facilities are needed to provide molecular characterization of antigens now presented in unpurified, crude form, such as cell lysates.