CANCER GENETICS AND EPIDEMIOLOGY

Co-chairs: Webster K. Cavenee, Ph.D., and Ronald A. DePinho, M.D.

Participants: Francis Ali-Osman Rebecca Betensky Jaclyn A. Biegel Darrell D. Bigner J. Gregory Cairncross Richard A.Fishel

James F.Gusella Daphne A Haas-Kogen C. David James Bruce R. Korf Kenneth A. Krohn Greg Riggins

James Rutka Gail Segal Robert L. Strausberg Jeannine Walston Albert Wong

STATEMENT OF THE PROBLEM

Concerted efforts over the past few years have shown that brain tumors, like other major human neoplasms, result from the accumulation of genetic lesions during tumor progression. Despite the extensive catalogue of these somatic tumor-associated lesions, significant gaps exist in our understanding of how such lesions initiate the process, how they influence therapeutic response, and the nature of their biological function. Clearly, such information will be required in order to harness the knowledge of genetics for improved diagnostic and therapeutic modalities. This is of particular importance for this group of tumors, given their unique biological and clinical characteristics and heterogeneity.

CHALLENGES AND QUESTIONS

• Little is known about brain tumor predisposition genes in humans. This situation reflects the scarcity of specimens, poor record-taking of family medical histories, insufficient clinical and pathological information on the samples, the lethality and late onset of many of these diseases, the inaccessibility of early lesions, and pedigrees that often do not lend themselves to mendelian analysis. In rare instances, autosomal dominant patterns indicative of "hard" primary mutations have been reported (e.g., Turcot syndrome, neurofibromatosis 1 and 2, and Li-Fraumeni syndrome). Family patterns are more commonly consistent with the possibility of multigenic inheritance ("soft" but interacting mutations), modifiers that alter the penetrance or expressivity of the genes, or epigenetic gene inactivation.

• A special feature of many malignant brain tumors is their innate resistance to existing chemo- and radiotherapeutic approaches. Little is known about the genetic mechanisms responsible for this resistance. For example, the impact of somatic or germline allelic variation on these mechanisms remains to be determined.

• Limited information exists on how specific lesions behave in cells of different lineages thought to represent precursors of distinct brain tumor types. Such information could be of importance in the design of therapeutic protocols targeting such lesions.

• Little is known of the interactions between predisposing/somatic mutations and external or internal environmental perturbations, such as hormonal influences, in utero exposures, and workplace carcinogens. This issue might be particularly relevant in explaining tumor emergence in different age groups (pediatric versus adult) and their distinct clinical behaviors.

• The genetic basis of the unique biological features of brain tumors is largely not understood. This applies to integral and important features such as invasion, motility, angiogenesis, and necrosis, as well as tumor progression and maintenance.

• Existing genetic models and associated genomic infrastructure (particularly in mice) are inadequate to properly address the genetic and phenotypic aspects of the human diseases. In the absence of validated, refined models, rapid testing of candidate cancer genes and their therapeutic approaches is severely hampered.

• There is an inadequate compendium of gene expression profiles for precursor cells and their lineages and tumor derivatives. Little information exists on the nature of the physical and functional interactions of the gene products that are known to play a role in the development of brain neoplasia with other cellular components. Moreover, genome-wide genotypes have not been collected and so have not been tested for their correlation with tumor type or behavior.

• There exists a strong need for the development of genetic screens that will permit tumorigenesis. These screens need to be conducted on both the organism and cell levels. The latter will be depend on the development of in vitro systems that accurately reflect the in vivo process under investigation.

• There is no comprehensive tumor registry, tumor bank, and familial tissue bank. It is especially important that these be comprehensive and organized on a national level, given the rarity and heterogeneity of the most informative tumors and familial situations.

RESEARCH AND SCIENTIFIC PRIORITIES

Priority 1: Isolate genes causing predisposition to human brain tumors.

It is important to search for predisposition genes in families with brain tumors as the primary identifier of genes relevant to brain tumors. Families whose members are prone to a variety of other tumors may represent additional opportunities to isolate genes that are also relevant to brain tumor pathogenesis. Elucidation of the interaction of such genes with environmental agents may also play a significant role in understanding the etiology of brain tumors.

Priority 2: Identify the genes and genetic variations that underlie tumor resistance to chemotherapy and radiation therapy, as well as the allelic variations that influence responses to therapy in individual patients.

Priority 3: Understand genotypic influences on phenotypic behavior, tumor type, age at onset, anatomical position, cell of origin, and cellular biology.

Priority 4: Establish and refine genetically based model systems that can faithfully recapitulate the complexity, heterogeneity, and diversity of human brain tumors.

RESOURCES NEEDED

• There is a strong need for organized and coordinated brain tumor registries, including family histories and extensive clinical and pathological information. These registries should be coupled with tumor samples that are equally well characterized and with somatic noncancerous tissues from affected individuals and their family members. It would be particularly useful if such centralized resources maintained strong technical support to conduct routine genome-wide studies, including expression profiling, in situ hybridization of tissue arrays, and high-density genotyping and mutation analysis. Centralization of these technical efforts would provide for efficient and thorough utilization of these precious samples and enable investigators to obtain such information without the need to build or develop advanced capabilities themselves.

• It would be widely useful to establish a compendium of gene expression patterns and genome-wide genotypes of tumors of many different histologies to be used for correlative studies with regard to cell type, developmental stage, and their response to therapeutic agents. The Brain Tumor Genome Anatomy Project BT-GAP and the Cancer Genome Anatomy Project (CGAP) are making significant progress with regard to human tumors, but the genome infrastructure to analyze mice lags far behind. This is a serious problem in that it hampers the rapid isolation of genes based on interspecies sequence homologies.

• There is a strong need to support the design and development of novel and targeted genetic screens conducted on both the organism and cell culture levels. Particularly relevant is the need to fortify efforts for the development and refinement of mouse models of human brain tumors harboring commonly occurring genetic lesions. Such cancer-prone models will find great utility in the identification of pathways and their interactions, as well as for loci that modify the effects of these pathways. Significant emphasis should be placed on the development and design of mouse models that enable assessment of the role of genes in both tumor initiation and tumor maintenance.

• The interdisciplinary nature of neurooncology makes it essential to augment the opportunities for physicians to receive training in molecular oncology and developmental neurobiology. It is equally essential for basic scientist trainees to receive training that is medically relevant to molecular neurobiology. Similarly, there exists a need for programmatic funding mechanisms that can form a bridge between the activities of established investigators from different disciplines focused on common themes in neurooncology. The design of the peer review of such grants should consider the special needs and circumstances of such interdisciplinary efforts. Finally, the relative paucity of understanding of this dread disease underscores the desirability to develop rapid funding mechanisms emphasizing novel and, perhaps, preliminary ideas (cf. Department of Defense "concept grants").