Defining the Future of Neurofibromatosis Research

NINDS Workshop: Defining the Future of Neurofibromatosis Research

May 4-5, 2000

Neuroscience Center, Rockville MD

• Introduction
  
• Background
  
• Workshop Summary

• Session 1: Basic Research - Advances and Priorities
• Session 2: Therapeutics
• Session 3: Strategy: Research Priorities and Funding Mechanisms

---

Introduction

This workshop was initiated by NINDS, with assistance from five other NIH Institutes (NCI,NHLBI,NEI,NIDCD,and NICHD) and the Office of Rare Diseases. It was organized by Drs. Robert Finkelstein and Philip Sheridan of NINDS, and included representatives from the Department of Defense, the VA, the National Neurofibromatosis Foundation, Neurofibromatosis Inc, and other patient advocacy groups. It consisted of three sessions, the goals of which were to define scientific priorities in neurofibromatosis research (Session 1), assess areas in which therapeutic progress is essential (Session 2), and discuss ways in which the effort to combat NF1 and NF2 can be intensified (Session 3).

---

Background

The neurofibromatoses are genetic diseases that cause tumors to grow along nerves and produce other abnormalities such as skin changes and bone deformities. They occur in both sexes and in all races and ethnic groups, and are classified into two types, neurofibromatosis type 1 (NF1) and neurofibromatosis type 2 (NF2). NF1 and NF2 are inherited in a dominant fashion. However, about half of all cases of NF1 and of NF2 are sporadic. In these cases, the parents are unaffected and the disease arises spontaneously through a change (mutation) in the patient's genes.

NF1 affects about 1 in 3,000 people at birth. The presence of light brown skin spots (café-au-lait spots) during childhood is one of the symptoms of NF1. Patients also usually exhibit neurofibromas, tumors that grow along nerves under the skin, which can be cosmetically disfiguring. About one-third of patients have plexiform neurofibromas, tumors that involve multiple nerves. Other tumors, such as of the spinal cord and optic nerve can also develop. About half of NF1 patients have learning disabilities. Treatment for NF1 is aimed at controlling symptoms. The extent of severity of NF1 varies a great deal from case to case. In mild cases, patients can live normal and productive lives, but in severe cases, the disease is debilitating.

The gene responsible for NF1 encodes a large protein called neurofibromin. A region of neurofibromin exhibits sequence similarity to the Ras-GAP family of proteins. Neurofibromin and Ras-GAP can suppress the action of another protein, Ras, which can promote tumor growth. When neurofibromin does not function properly, as in NF1, the Ras signaling pathway is aberrantly activated.

NF2 affects about 1 in 30,000 people at birth. NF2 patients develop tumors along the eighth cranial nerve. Hearing loss results and patients can have problems with balance. About half of NF2 patients have intracranial meningiomas, tumors growing on the covering of the brain, and half of these patients have resulting symptoms. Intracranial meningiomas are difficult to treat. About 40% of NF2 patients have visual difficulties. NF2 is usually a severe disease that substantially affects the life and lifespan of patients. Apart from the treatments already noted, treatment is aimed at controlling symptoms.

The gene responsible for NF2 encodes a protein called merlin or schwannonim. Merlin appears to regulate growth in Schwann cells and in meningeal cells, but not much else is known about how it works.

---

Workshop Summary

---

Session 1: Basic Research - Advances and Priorities

Mouse Models of NF1

Luis Parada

Dr. Parada and colleagues produced mice that were heterozygous (had one normal gene and one defective gene) for each of two genes, the NF1 gene and the p53 gene, which encodes a tumor suppressor. These mice developed tumors and died more rapidly than mice heterozygous only for the p53 gene, which have a high tumor frequency to begin with. Also, the mice that were heterozygous in both genes had tumors that were similar to those seen in NF1 patients. Many (but not all) neurofibromas in NF1 patients have mutations in the p53 gene and in other genes that work with the p53 gene. The results suggest that different tumor suppressors cooperate in the development of NF1.

How Merlin Might Work

David Gutmann

The merlin protein can assume two different comformations. When the two end-regions of merlin interact with each other, the merlin molecule becomes "closed". When one of the end-regions of merlin folds up on itself, the merlin molecule remains "open".

One model suggests that when merlin is closed, it is active, but when it is remains open, it is inactive. For example, Dr. Gutmann and colleagues separated the ends of merlin and genetically engineered mice to have separate genes encoding these two ends. Then they stimulated these genes to get merlin to function. Stimulating only one of the genes was insufficient. For merlin to work, each of the genes had to stimulated to encode their respective ends. These and other experiments suggest that merlin acts as a molecular switch. When it is closed, merlin binds to proteins, resulting in suppression of growth. When merlin is open, it binds to a different set of proteins, resulting in lack of growth suppression.

Studies on Neurofibroma in Mice

Nancy Ratner

One of the hallmarks of NF1 is the appearance of neurofibromas. However, neurofibromas are not normally observed in mice. Dr. Ratner and colleagues tried to induce these tumors in mice that were heterozygous for the NF1 gene.

Because wound healing and tumor formation have similarities, one possibility is that trauma triggers the development of neurofibromas. Accordingly, Dr. Ratner and colleagues wounded the skin of the mice to injure the nerves. The wounds healed abnormally. For example, the mice had skin spots resembling those typically noted in NF1 patients. A few mice had tumors resembling neurofibromas. The results showed that the mouse NF1 protein controls the response to injury, and that NF1 gene mutations in mice cause abnormalities similar to those seen in human neurofibromas.

In a related work effort to describe these abnormalities, Dr. Ratner and colleagues found that human neurofibroma cells have a characteristic that normal Schwann cells do not. This additional characteristic is the presence of a particular growth factor receptor, that enables the neurofibroma cells to respond to certain factors. Normal Schwann cells do not have this receptor and do not respond to these factors.

Function of the NF1 Gene

André Bernards

Some adolescent and young-adult NF1 patients have no neurofibromas, but others have 100 to 500. Researchers at Cambridge University in England found that the more closely related the patients were to each other, the more similar the disease characteristics. The researchers concluded that the characteristics of NF1 disease is greatly affected by genes besides the NF1 gene itself. However, the results did not exclude the possibility that differences in environment were responsible for the differences in disease characteristics.

To study the function of the NF1 gene and to find out the identity of other genes that affect its function, Dr. Bernards and his coworkers used the fruit fly Drosophila melanogaster. First, they isolated the fly's NF1 gene and showed that its DNA sequence is similar to that of the human NF1 gene. Next, they bred flies lacking the NF1 gene. These flies were about 20% smaller than wild-type (normal) flies. They also appeared to have a defect in the transmission of nerve impulses to muscle. Other studies showed that increasing the activity of a protein called PKA could rescue these flies from having the smaller size.

In collaboration with another research group, Dr. Bernards's group is trying to find out if the human NF1 gene can substitute for the Drosophila NF1 gene. Preliminary results showed that introducing the human NF1 gene into flies that are missing the NF1 gene can rescue the flies from the growth defect.

Continuing this work, Dr. Bernards and coworkers wanted to find out if the Ras-GAP region of the NF1 gene is needed to rescue the NF1 mutant flies from the growth defect. To do the experiment, they took the Drosophila NF1 gene and made it defective in the Ras-GAP region, so this mutant gene could not encode active Ras-GAP--like protein. Then they introduced this mutant gene into the NF1 mutant flies. The flies were not rescued. The researchers hope that studies like these will enable understanding of the genes that affect how the NF1 gene works, and, ultimately, why NF1 disease is less severe in some patients but more in others.

Perhaps the NF1 mutant fly is smaller than wild-type because the mutants has fewer cells or smaller cells. It turned out the eyes from the mutant flies have fewer cells, but the same size as wild-type eyes. In contrast, the mutant wing has as many cells as the wild-type wing, but smaller.

In other experiments, Dr. Bernards and coworkers managed to grow a patch of mutant wing cells in a wing of a fly that had one wild-type NF1 gene and one mutant NF1 gene. The mutant wing cells grew to normal size. This result showed that the size of the wing cells is not totally determined by the cells themselves. Instead, the cell size depended in part upon the environment in which they are growing.

NF1 and Pathways of Learning in Drosophila

Yi Zhong

Dr. Zhong and his coworkers are trying to understand why learning disabilities occur in NF1 patients. To do this, they are studying the role of the NF1 protein in learning and memory in Drosophila.

Learning and memory require many biochemical steps. A fly mutant called rutabaga has deficiencies in learning and short-term memory. The rutabaga gene encodes a protein that is important for the function of an important series of chemical events. This pathway is separate from the Ras-GAP pathway that NF1 protein is understood to affect. It turns out that the fly NF1 protein is essential for the chemical pathway that is affected by the rutabaga gene, and for proper learning and memory in Drosophila. Therefore, NF1 protein can act on other pathways besides the Ras-GAP pathway. The results suggest that drugs could be used to make up for the defects in the chemical pathways, thereby easing the learning and memory disabilities in NF1 patients.

---

Session 2: Therapeutics

Goals in Treating NF1

Bruce Korf

The goal in treating a young child with NF1, who has abnormal skin spots but does not yet have neurofibromas, is to try to prevent the development of neurofibromas and other pathologic features of NF1. It is better to start treating NF1 as soon as the diagnosis is made, rather than to wait for more serious manifestations of the disease.

NF1 is a chronic disease, and early in the disease, patients feel well and can function. Tumors in NF1 grow slowly, so it might not be realistic to think of tumor shrinkage as a measure of success in treating NF1. In this regard, it is important that treatment for NF1 not be worse than the disease. In designing treatment for NF1, perhaps it is better to look at the example of how diabetes is treated rather than at the example of how leukemia is treated. Clinical trials of NF1 treatments would be appropriate when reasonably plausible treatment and measurements of treatment outcome are available.

It is difficult to define outcomes in treating NF1. The disease is complex and its characteristics differ from patient to patient. Although some aspects of the disease make life miserable for the patient, they are not life-threatening.

The frequency and impact of complications can be summarized as follows: Optic glioma occurs in about 15% of NF1 cases, but fewer than half of these require treatment. However, more harm has been done by overtreating optic gliomas that did not need to be treated.

Points to consider are as follows:

• Will the experience that might be gained in treating malignant tumors apply to treating non-malignant tumors?
• It is important to follow all lesions in a treatment trial.
• Treatment of neurofibromas is likely to be preventive.
• Reliable outcome measures of functional impairment and of psychological adjustment are needed.

Approaches to Cancer Therapy

Jackson B. Gibbs

Significant advances in cancer therapy are in three categories: cytotoxics (cell poisons), antihormonals (hormone blockers), and adjuvants (drugs added to assist the main drug). These categories were defined with molecular targets in mind. The issue is now to define new molecular targets for cancer therapy. Approaches to treatment of NF should be based on specific targets characteristic to the disease.

As mentioned, proteins such as Ras work through specific biochemical pathways. These pathways have been closely studied in the pharmaceutical industry for the development of new drugs that block them. The reason is that cancers are associated with changes in these pathways, so these pathways offer targets for cancer chemotherapy. Various drugs are being developed to aim at various targets. Most of these drugs are being designed for use in combination with traditional anticancer drugs. So far, the side effects of these drugs appear to be reasonably tolerated.

Ras acts in many ways, offering many targets for chemotherapy. However, the body has multiple types of Ras proteins, which are encoded by different genes. Therefore, in developing drugs to treat NF1, it will be important to figure out which Ras pathway is affected.

Pharmaceutical companies are developing drugs to interfere with Ras action. For Ras to work, it must first have a small fat-like molecule attached to it. This feature suggests a therapeutic approach: Perhaps blocking the attachment of the fat-like molecule would be useful in treating Ras-associated cancers. Several pharmaceutical companies have been testing drugs that block the attachment of the fat-like molecule to Ras. So far, it does not appear that the blocking drugs will be useful by themselves. Clinical trials are underway to test these drugs in combination with others.

Optic Gliomas and Plexiform Neurofibromas in NF1

Roger J. Packer

NF1 patients can have optic nerve gliomas (tumors of a type of nerve cells called glial cells), which can take an unpredictable, erratic course. These tumors can shrink without treatment, which makes it difficult to evaluate if treatment is effective or not. Optic nerve gliomas in children with NF1 are mostly low-grade (less severe), and often the treatment is confined to observation. Surgery can be used to correct cosmetic disfigurement. Radiation therapy and chemotherapy are also used, such as when the tumor appears in the brain.

Various drugs are being tested for use for treating plexiform neurofibromas.

Gene Therapy for NF2 Brain Tumors

Xandra Breakfield

Gene therapy involves designing vectors (often viruses) that contain therapeutic genes and trying to deliver these vectors to specific targets in the body. Viruses are used because they efficiently deliver the therapeutic gene. The viruses have been stripped of viral genes that would otherwise make them dangerous. Gene therapy for cancer presents a problem in that it is difficult to deliver the therapeutic gene to the tumor. For example, some viruses can go only to cells that are multiplying. This is a problem for treating tumors that are growing slowly. Furthermore, the safety of gene therapy is not established, so gene therapy poses a risk to NF2 patients who are functional.

Leukemia and NF1

Kevin M. Shannon

Juvenile myelomonocytic leukemia is a rare, usually fatal disease of early childhood. About 10% to 15% of children who have this disease also have NF1. This means that children who have NF1 are predisposed to having juvenile myelomonocytic leukemia.

Loss of the NF1 gene frequently occurs in human leukemia. This supports the idea the NF1 gene acts to suppress tumors. Various investigators developed a way to prepare mice that lack the NF1 gene in their bone marrow cells. These mice develop a disease that resembles juvenile myelomonocytic leukemia in humans. Additional studies revealed the role of a naturally occurring growth-stimulating compound in the development of the juvenile myelomonocytic leukemia-like disease in mice. The involvement of NF1 and of a growth-stimulating compound in the development of juvenile myelomonocytic leukemia suggests therapeutic approaches to this disease. Mice afflicted with leukemia could be used to test compounds that are being considered for treating the disease in humans.

---

Session 3: Strategy: Research Priorities and Funding Mechanisms

Technology Transfer

Allan Rubenstein

The National Neurofibromatosis Foundation (NNFF) is developing a division of technology transfer. Two full-time staff, a director of technology transfer and a clinical trial coordinator, will be hired. The director of technology transfer will work to implement the transfer of potential advances in diagnosis and treatment into the clinic. For example, the director of technology transfer will encourage pharmaceutical and biotechnology companies to become involved in phase I and phase II trials. The intention is to provide the infrastructure for drug testing; the companies need only provide drugs. The focus will be on plexiform neurofibromas in NF1 and on schwannomas in NF2.

The National Neurofibromatosis Foundation has also started working on a project to develop an imaging/archiving system for NF patients. Ideally this will make it possible for patients to have a their records kept in a standardized way that makes them more readily available for review by other physicians.

Basic Research Priorities

Tyler Jacks

Basic research involving NF1 has many important goals. These include target cell identification (identifying which cells are affected in NF1 tumors); understanding cell autonomy and non-autonomy in tumor formation; understanding the relevance of the Ras pathway and of other pathways in relation to the NF1 gene; understanding what happens to cells that have one normal NF1 gene and one mutated NF1 gene; defining the growth factors that affect NF1 disease; finding other genes that might affect the function of the NF1 gene; and understanding the molecular basis of tumor initiation (why neurofibroma growth starts and stops).

Basic research involving NF2 should aim at understanding the normal function of merlin; understanding target cell identification; understanding the function of isoforms (different forms) of merlin; and understanding the effect that removing the NF2 gene in fruit flies has on growth.

Although individual laboratories will be funded to work on projects intended to address these research priorities, interactive funding is probably also needed. For example, developing NF mouse models that fully reproduce the disease will be complicated, long-term, and expensive. Other needs for work on mice are access to proprietary materials and preparation of materials that are designed for research in the mouse. For example, drugs that are designed for use in humans might need to be modified for testing in mice.

Other needs include understanding how to do preclinical testing of drugs in mice; understanding how to screen many compounds (high-throughput screening) in NF1 cells; and understanding how to use mice to try to learn to prevent the disease. These projects are also expensive and not likely to be done through RO1 proposals.

It is unclear what pharmaceutical companies will be willing to do, such as develop a system to screen compounds for treating NF. Academic investigators are less well trained and less inclined to do this work.

Clinical Priorities

Bruce Korf

Current clinical definitions of NF are based on previous definitions that should be reconsidered. A lot remains to be learned about the clinical characteristics of NF. In designing clinical trials, treatment goals, such as change in tumor size and growth, need to be defined. NF is a genetic disease with an uncertain outcome, raising issues of research ethics that need to be considered. The lay community could contribute to the formation of a series of ethical considerations.

Core facilities, such as a tissue bank, are needed. The database has been in existence for longer than a decade, and should be reexamined for its usefulness. A network facilitating communication between patients and clinicians and among clinicians is needed.

The opportunity for beginning clinical trials is there, but the research community is probably not yet prepared. Issues include the kinds of trials that should be done, and who should be doing them. Geneticists and neurologists are the practitioners who have the most clinical experience treating NF1 patients. Oncologists are the practitioners who are most familiar with using cytotoxic drugs. However, oncologists are accustomed to treating diseases that have a much different course than NF1.

One possible goal is to establish specialized research centers that define clinical characteristics involve cooperation with research laboratories.

Concerns have been raised about revealing the nature of the genetic mutation to the family. Revealing this information has two immediate and obvious benefits. In clinically uncertain cases, identifying the mutation can verify a diagnosis. Many families find it difficult to handle the uncertainty about the diagnosis. The second benefit is the possibility of offering genetic counseling with the choices of prenatal diagnosis or preimplantation diagnosis.

The Department of Defense NF Program

James F. Gusella

The Department of Defense sponsors an NF research program, which is administered through the U.S. Army Medical Research and Materiel Command. Funds for the program are mandated yearly by Congress and must be spent or at least dedicated by a certain date. The needs and directions are set by an integration panel. Examples of award mechanisms include investigator-initiated research, natural history studies, postdoctoral traineeships, IDEA grants (Innovative Development and Exploratory Awards), new investigator awards, and clinical trials. The integration panel decides on the proportion of funds to be allocated to the different types of award pathways, and separate peer-review committees decide which applications will be funded. The Department of Defense program is arranged differently from NIH-sponsored extramural research programs.

The Department of Defense NF program is a useful way to develop infrastructure consortia for clinical trials. The program is flexible so that even though proposals for clinical trials might not get the highest scores from peer-review committees, they can still be funded if the integration panel so chooses.

The patient's disease and genetic characteristics should be the reference standard in NF research. Close cooperation among patients, clinical researchers, and basic scientists is needed, and these groups need to be supported by a stable infrastructure. The Department of Defense NF program funding is limited to 2 or 3 years, which provides questionable stability. Funding that provides greater stability is necessary.

Research involving inherited diseases is based on two approaches. One is based on the phenotype--the observable characteristics of the disease. The other is based on the genotype--the genes responsible for the phenotype. Before discovery of genes responsible for an inherited disease, the disease will have been described, and, using this description, assumptions made about how the disease occurred in the first place. These assumptions become the basis of tests about how to intervene. Thus this approach--the phenotypic approach--works backward from observations about a disease to inferences about how the disease process occurs.

In contrast, the genotypic approach works in a forward direction. With the discovery of a gene responsible for a disease, assumptions are made about how the gene works, and then interventions are designed that test these assumptions. Thus the genotypic approach starts with knowledge of the genetic variation and then works forward to try to understand the pathologic consequences of the variation and how to deal with it.

Discovery of the gene that causes a disease can reveal how the disease occurs. For example, discovery of the genes responsible for NF1 and for NF2 led scientists to begin to understand the relationships between neurofibromin and merlin and these diseases. These relationships were not predicted by investigators using the phenotypic approach. Thus the genotypic approach has more capacity to discover what is unknown than does the phenotypic approach. However, many steps occur between the presence of a disease gene in a patient and the presentation of the disease itself. Therefore, the goal in research in inherited disease should be to meet in the middle of the phenotypic and genotypic approaches, and in doing so attempt to work for the benefit of the patient.

Funding mechanisms that foster both approaches would be an important development for NF research. As it does for other diseases, the NIH should support centers that conduct NF research and provide care for NF patients. Such NF centers could support needed descriptive research, which is usually not funded by granting agencies looking for hypothesis-based proposals. Such centers typically attract motivated patients who provide tissue samples, and can attract basic scientists as well. A comprehensive program such as an NF research center would cost $1.5 to $2 million per year. Supporting a program for 5 years, as the NIH typically does, would provide the needed stability.

---

Discussion of Funding Priorities

Bruce Korf

Participants discussed the advantages and disadvantages of establishing NF centers. Some believed that it has to be decided whether or not establishing NF centers is the best use of available funds. Others were concerned about the overall quality of centers, and that despite large amounts of funding, centers concentrating on some other diseases have not accelerated the development of disease treatments. Still others believed that centers are useful because they provide important connections between individual investigators and patient populations.

Participants disagreed about the necessity of having basic and clinical researchers in the same location. A problem is that a centers with only clinical researchers do not get funded. A solution might be to have basic scientists and clinical researchers at different locations linked by a "virtual center".

Some participants believed that NINDS and other organizations should broadcast their interest in funding standard RO1 grant proposals. Perhaps different NIH components could cooperate to fund joint initiatives for NF research. Some participants suggested having more support for trainees such as postdoctoral fellows and residents.

Participants discussed having Specialized Programs of Research Excellence (SPOREs), which are used by the National Cancer Institute. The SPORE supports interdisciplinary teams of investigators who are dedicated to translational research. Translational research is defined as the movement of a laboratory discovery into a patient or a population research setting, or the movement of an observation in a patient or a population setting into a laboratory research environment.

Finally, the participants agreed that the results of this NINDS workshop should be presented to the larger NF research community. Bob Finkelstein agreed to lead a discussion of those results at the upcoming national NF meeting in Aspen, Colorado.