TREATMENT

Co-Chairs: Howard Fine, M.D., and Larry E. Kun, M.D.

Participants: Mitchel Berger Joseph C. Gloriosa Stuart A. Grossman Alison Hannah Peter T. C. Ho William G. Kaelin, Jr. David Kaplan Kathleen Lamborn Bertrand Liang

Dennis C. Shrieve Malcolm Smith Philip J. Tofilon Diane Traynor Michael Walker Roy S. Wu Mark Yarborough John S. Yu W. K. Alfred Yung

BACKGROUND: STATUS OF THE FIELD

Surgery

The tremendous evolution in surgical capabilities of recent decades has been driven by technology. The advantage of maximal resection has been documented in most primary central nervous system (CNS) tumor systems. At present, the state of the art in treatment for brain tumors is the incorporation of preoperative imaging, both metabolic and functional, with image-guided surgical techniques. For example, "navigation" technology allows imaging in a patient by using external landmarks before surgery, and during surgery, the surgeon is guided to the lesion and can appreciate its extent based on the preoperative image set. The newest technologies provide intraoperative imaging, allowing the neurosurgeon to realign the three-dimensional image set during surgery, further enhancing safe maximal resection. Surgery can also be important in local drug delivery, including the strategic placement of catheters to deliver small molecules. Careful and consistent mapping of tumor specimens allows coordination with imaging, drug delivery, and cellular/molecular correlates of disease characteristics and therapeutic response.

Radiation Oncology

Technology has also driven the field of radiation oncology. Radiation therapy has proven efficacy in many common tumor histiotypes. Trials in malignant gliomas in particular have provided opportunities to test altered fractionation and three-dimensional, image-guided delivery (including conformal photon irradiation, radiosurgery, and brachytherapy) to achieve substantial dose escalation. Unfortunately, however, there are few data encouraging additional explorations of dose escalation with radiation therapy alone for these inherently resistant tumors. Further studies of enhanced radiation dose are ongoing in other histiotypes (e.g., low-grade gliomas, and ependymomas).

Interactions of radiation with pharmacological and biological agents represents an important area for further development. Understanding ways to increase tumor response and potentially to limit normal tissue toxicities requires additional laboratory and clinical testing. Explorations of genetic radiosensitization offer exciting research opportunities in the malignant gliomas.

In pediatric tumors, trials of restricted volumes of radiation therapy and dose reductions defined by response to chemotherapy seek to improve the therapeutic ratio in tumor types with known radioresponsiveness (e.g., medulloblastoma, other embryonal tumors, low-grade gliomas, intracranial germ cell tumors). Exquisite localization of radiation volume in ependymomas is the basis for ongoing and planned trials even in very young children. Image definition of tumor extent is important in guiding restricted radiation therapy volumes; better assessments of tumor volumes and margins are needed.

Chemotherapy

Chemotherapy has had a significant impact on the treatment of selected CNS tumors, such as primary CNS lymphomas, anaplastic oligodendrogliomas, and pediatric embryonal tumors. Nevertheless, the exact drug regimen, timing, and duration of treatment remain areas of uncertainty. Despite the clear benefits for these selective tumor types, the role of standard chemotherapy is limited for the majority of primary CNS tumors, particularly tumors of the astrocytic lineage. Identification of the role of chemotherapy has been hampered by relatively small numbers of many brain tumors other than glioblastoma multiforme, making large randomized trials problematic, and by the apparent poor responsiveness of glioblastoma multiforme to cytotoxic chemotherapy.

STATEMENT OF THE PROBLEM

The main problems in brain tumor treatment research listed below are addressed in detail in the following section (see "Challenges and Questions").

• There are few active therapeutic approaches or agents for the treatment of brain tumors.

• There are no adequate or reliable preclinical screening systems.

• Few new agents are marked for CNS tumor development.

• There is little understanding of drug-radiation interactions in normal CNS tissue.

CHALLENGES AND QUESTIONS

Few Active Therapeutic Approaches or Agents for the Treatment of Brain Tumors

Generally, there have been few significant advances in the treatment of malignant gliomas over the last two decades. Difficulties in identifying effective approaches include heterogeneity of tumor types, the rarity of some tumors, and relative difficulties in accruing adult patients to clinical trials. All of these problems result in small patient populations for clinical study.

There are a myriad of biological reasons for the ineffectiveness of most current chemotherapeutic agents. These include inconsistent drug delivery secondary to issues related to the blood-brain and blood-tumor barriers, tumor hypoxia, intrinsic drug resistance, and acquired drug resistance through the variable exposure of tumor cells to different concentrations of delivered drug as a result of problems with drug delivery.

The "new biology" has led to the identification of a number of new signaling pathways that appear to be important for gliomagenesis, and with their identification has come the creation of a number of new and exciting molecular inhibitors of those pathways that appear to represent exciting therapeutic opportunities. There is significant concern, however, that the current drug development process, from preclinical screening to clinical trial design, patient accrual, and endpoint assessment, may be suboptimal for studying newer cytostatic agents for brain tumors. As a result of this concern, there is reluctance from private industry (the source of the majority of the newer anti-tumor agents) to invest resources into exploring the utility of these agents in patients with brain tumors.

No Adequate or Reliable Preclinical Screening Systems

Preclinical screening of potential agents is currently time consuming and inefficient. Spontaneously occurring brain tumor animal models are not available at present, and there is a growing belief that currently utilized xenograft models do not accurately reproduce the biology of human tumors and therefore are generally nonpredictive for identifying active clinical agents. Therefore, the routine use of these models for screening agents for clinical development not only may have allowed inactive agents to enter clinical trials in the past, but may have inadvertently excluded potentially active drugs from ever having been evaluated clinically. This problem has become increasingly more important as the number of rationally designed agents with cytostatic rather than cytotoxic mechanisms of action are being developed (e.g. anti-angiogenic, differentiating, anti-invasion agents). Such agents can be adequately tested only in vivo, making the lack of reliable tumor models a major obstacle for preclinical development. Clearly reliable in vitro and/or tissue surrogate markers and/or assays of biological activity would be very helpful for the preclinical and clinical development of these agents (see "Few New Agents for CNS Tumor Development"). Coupled with the establishment of biological endpoints correlated with survival, improvements in preclinical screening may lead to innovative clinical trial designs, and ultimately the entire field of therapeutic development for brain tumors could move forward quickly and effectively.

Few New Agents Marked for CNS Tumor Development

Novel therapeutic development for CNS tumors is modest, and progress has been meager. Despite the exciting advances made in the identification of potentially active, selective, rationally designed anti-cancer agents, there has been significant reluctance on the part of the pharmaceutical industry to move these agents into the brain tumor population. The reason for this reluctance relates to the inherent characteristics of brain tumors and of patients with these tumors, which make them problematic as the focus of industry research. Examples include difficulties quantifying toxicities in patients with brain tumors, the altered pharmacology of many agents secondary to induction or inhibition of the hepatic P450 cytochrome system from concurrent administration of anti-epileptic agents, the heterogeneity of the patient population (particularly as it relates to heterogeneity of tumors), modest numbers of patients, and slow patient accrual, although this last issue has been addressed in part by the brain tumor consortia.

Low Rate of Patient Accrual to Clinical Trials

Patient accrual into clinical trials in general is lacking: less than 10% of adult patients with brain tumors enter clinical trials. Ways should be investigated to systematically increase adult patient accrual to trials. Even though there are at present few novel approaches that warrant large-scale Phase III trials, the infrastructure for speedy accrual needs to be in place in order to expedite testing when new agents appear promising in pilot studies. In addition, determining groups of molecularly homogeneous tumors would allow for more, and more effective, clinical trials.

Improvements in "response" criteria are needed. To advance the field of novel therapeutic agents for brain tumors, it will be necessary to identify and validate meaningful biological endpoints for evaluating novel therapies. There continues to be significant discussion within the neurooncology and neuroradiology community as to the appropriate criteria for measuring radiographic tumor "response." This discussion arises from the realization that abnormalities detected by magnetic resonance imaging and computed tomography and associated with tumors may be the result of pathophysiological processes other than the tumor mass itself. These abnormalities include treatment-related effects (e.g., radiation necrosis), cerebral edema, inflammation, and postsurgical changes. Routine criteria for measuring perpendicular diameters may be limited in reliability and accuracy by the fact that brain tumors often grow as irregular, asymmetrical processes in three dimensions, and by variability in head positioning in sequential imaging studies. The problem is further compounded by the realization that many of the novel agents entering clinical trials will not necessarily have cytotoxic mechanisms of action, so it might be difficult to assess therapeutic activity within the first several months of therapy, even with highly reliable measurement techniques. This difficulty in turn may complicate early (Phase I and II) trials, in which too few patients may be treated for a sufficient period to allow an accurate assessment of effects on progression-free and overall survival, the most accurate and important measures of biological activity for a cytostatic agent.

For this and other reasons outlined below, it would be ideal if, in early-phase studies, novel endpoints could be used as measures of biological activity of the tested agent. Such endpoints optimally would be related to the mechanisms of action of the agent or to modulation of the putative target (e.g., radiographic demonstration of diminished blood flow after administration of an antiangiogenic agent). Ideally, such endpoints will have correlates to preclinical screens (e.g., the same endpoint used to screen and select a biological agent for clinical testing can be demonstrated to be modulated in the treated patient). Such endpoints could include (but are not limited to) in vitro assays with patient material (blood, urine, cerebrospinal fluid, tumor tissue) and imaging methodologies such as magnetic resonance spectroscopy or positron emission tomography with appropriate probes. Early clinical trials could be designed to allow for agents that achieve a specific threshold effect in the predetermined endpoint evaluation to move forward into further clinical development, ultimately leading to a definitive Phase III trial.

The development of large, historical, clinical trials is important in order to evaluate the use of endpoints such as time to tumor progression and survival in early-phase trials. Such evaluations must be based on meaningful historical data used as controls against which activity and efficacy of the agent can be preliminarily inferred. Clinical endpoints for evaluating new agents must include quality of life measures in all Phase III and novel Phase II trials. The importance of quality of life endpoints is readily apparent in pediatric tumors but are also relevant for malignant gliomas in adults.

The development of such endpoints and objective response criteria will not only aid in the development of novel agents but also allow for the accrual of knowledge about current therapies. Furthermore, they will allow earlier identification of response in individual patients. Determining which patients are helped by current therapies and why will aid more rational and focused development of new therapeutics.

Novel clinical trial designs should be based on the demonstration that a new agent is able to reach and affect its intended target. The rarity of most CNS tumor types and the lack of meaningful short-term endpoints correlated with survival lead to difficulties in designing timely and effective brain tumor clinical trials. In order to focus more quickly on the highest-priority agents, especially for less common or slowly growing brain tumors, it would be helpful to be able to select them for further study based on demonstration of their ability to actually reach and alter their putative molecular targets in patients.

An additional issue relates to the delays often encountered in developing combinations of new agents. Rational combinations may be expected to be no less important for newer drug classes directed at defined molecular targets than they are for conventional cytotoxic chemotherapies. However, the current regulatory system requires that drugs can be licensed only if they are sufficiently useful individually, even when combinations may reasonably be expected to be considerably more efficacious. To substantially speed the development cycle for effective combinations would require a systematic plan for combination testing that begins at the preclinical and Phase I and II levels of the process, and when carried out with appropriate rigor, acceptance of such a strategy by the Food and Drug Administration.

The two adult and one pediatric brain tumor consortia offer excellent means to expedite testing of new agents, including in novel trial designs and in combinations. Further interactions with industry should be encouraged to provide pharmaceutical and biotechnology companies with access to tools (e.g., patient data, markers, validated endpoints) for evaluating potential new drugs. Study designs for new pharmacological agents should encourage pharmacokinetic and scientific endpoints, including collection of tumor and, where appropriate, adjacent normal neural tissue for pharmacological and genetic studies after drug delivery. Involvement of neurosurgeons and basic scientists in the design and analysis of clinical trial may facilitate this objective.

Little Understanding of Drug-Radiation Interactions in Normal CNS Tissue

An important challenge remains the development of interventions that might enhance radiosensitivity while diminishing neurotoxicity. Spatially defined radiation interactions with pharmacological and biological agents represent an exciting area for potential enhancement of the therapeutic ratio in brain tumor treatment. Further understanding of the interactions of chemotherapeutic agents and irradiation in normal CNS tissues should be sought.

New avenues to exploit radiation effects require intensive laboratory development prior to expedited clinical trials in malignant gliomas. Such approaches include the delivery of tumorsensitizing or -neuroprotective molecules via gene therapy coupled with focal radiation delivery. Exploration of radiation-induced promoters represents an additional focal biological effect that may be exploited in this tumor system. (See also the report of the Radiation Biology breakout session.)

RESEARCH PRIORITIES

Priority 1: Facilitate novel therapeutic development and increase knowledge about the mechanisms of current therapies.

• Enhance preclinical screening:

-- Recognize that currently existing brain tumor animal models are unreliable predictors of clinical drug activity.

-- Develop a paradigm to move rationally developed new agents toward clinical trials primarily based on their ability to inhibit signaling pathways that are known to be important in the proliferation and/or survival of gliomas or other CNS tumors. Such a paradigm will require validation by experience, however, and perhaps could never displace entirely older designs, because our understanding of pathways and molecular targets is far from complete and will always be changing. Some agents will eventually be found to modulate pathways or inhibit targets in ways other than those originally assumed and intended.

-- Support the development of new animal models that more faithfully model human tumors as validated both by molecular characterization and through their pharmacological interactions and response to new and known cytotoxic agents.

• Provide added resources to facilitate moving new therapeutic agents developed in the academic setting from laboratory compounds to clinical-grade drugs for therapeutic trials.

• Increase patient accrual into clinical trials.

-- Increase patient awareness of clinical trials and the benefits of participation through advocacy groups.

-- Support a systematic effort to determine barriers to adult enrollment in brain tumor clinical trials, the results of which effort should be incorporated into educational and informational efforts to increase enrollment.

-- Facilitate the availability of innovative studies for less common CNS tumors or inclusion in ongoing trials of promising new therapies for more common CNS tumors in which patients can be analyzed as distinct subgroups.

• Identify meaningful biological endpoints for the evaluation of new therapeutic approaches.

-- Develop and validate molecular and genetic endpoints (surrogate markers).

-- Develop and validate imaging endpoint parameters (e.g., quantitative magnetic resonance imaging and magnetic resonance spectroscopy).

• Optimize study designs.

-- Identify genetic and epigenetic markers that more accurately group patients with biologically homogeneous tumors.

-- Formulate study designs by using specific biological endpoints that are relevant to the known mechanism of action of the agent(s) being tested.

-- Encourage study designs that incorporate tissue acquisition before and during treatment for assessment of drug delivery and measurement of biological endpoints.

-- Consider distinct designs appropriate to cytostatic and cytotoxic agents.

-- Encourage incorporation of functional and quality-of-life measures in adult and pediatric brain tumor studies.

-- Identify common data elements to facilitate the evaluation of new therapies across institutions and cooperative groups.

-- Establish a national data repository for clinical and genetic information to be available to investigators.

Priority 2: Stimulate research on improving the therapeutic index of new agents specifically relevant to the CNS.

• Develop improved methodologies for drug delivery (e.g., blood-brain barrier disruption, convection) for both primary intraparenchymal tumors and leptomeningeal tumors.

• Develop new methodologies for assessing neuropharmacokinetics.

• Develop tools for assessing toxic effects of drugs on the CNS (neurotoxicology).

• Support enhanced research into potential means of neuroprotection.

• Support research to improve the design and delivery of conditionally replicating oncolytic viral vectors and other gene therapy vectors used either alone or in conjunction with chemotherapy and/or radiotherapy.

Priority 3: Enhance research to improve the therapeutic ratio for radiation therapy for CNS tumors.

• Study means of enhancing radiosensitivity for malignant gliomas and other CNS tumors.

-- Develop gene transfer technologies providing radiosensitization or radioprotection.

-- Explore the role of radiation-inducible promoters as a means of enhancing temporal and spatially specific gene expression.

• Study outcomes in children treated with limited volume, high-technology radiation therapy, often at less than "conventional" dose levels.

RESOURCES NEEDED

• Resources to establish an Internet-based clinical/research database

• Incentives to encourage industry interactions with academic investigators to identify new endpoints and surrogate markers important to testing new CNS agents

• Increased access by academic scientists to centrally funded GMP capabilities in developing new CNS agents

• Funding for the development and validation of new animal models specific for testing new agents in CNS tumors

• Increased funding for statistical support in developing novel study designs

• Funding for prospective assessment of imaging endpoints of therapeutic response based on collaborations between clinical investigators in neurooncology and neuroimaging

• Development of central core facilities for in vivo animal evaluation of new therapeutic approaches, including pharmacological imaging and biological endpoints for pharmacological and biological agents, neurosurgery, and radiation therapy

• Support for studies of genetically related alterations in radiation sensitivity and toxicity

• Support for development of radiation-inducible promoters to enhance temporal and spatial specificity of gene expression

• Support for an infrastructure of coordinated (central or linked) tumor banks that would be available on a competitive basis to researchers in academia or industry in order to expedite development of new therapies. These banks would need to contain specimens collected and maintained with appropriate quality assurance and associated (with appropriate privacy safeguards) with validated clinical data.