With genetics providing new targets and identifying new potentially druggable pathways, and with the development of novel tools and technologies, the field of neuropsychiatry is at an inflection point, said Stacie Weninger, president of FBRI. “We now have a better understanding of the real pathology behind disease, how to affect it, how we can measure it, and the new ways of interacting with it,” she said; the challenge is translating that knowledge into innovative treatments.
The workshop highlighted some of the emerging technologies to better define genetic variants, genes linked to these variants, and interactions among these genes and pathways, said Dimitri Krainc. These technologies will make it possible to identify genetically validated targets and demonstrate target engagement in patients, develop a deep mechanistic understanding of disease, produce high-quality and reproducible data, and better stratify diverse patient populations using genetics and biomarkers. The final goal, he said, is to develop targeted therapies for well-defined subgroups of patients.
To achieve this goal, data sharing and collaboration are critical, said Krainc. Moreover, said John Ngai, the lasting legacy of scientific enterprise for future generations also requires making the generated data findable, accessible, interoperable, and reusable (FAIR), which will require a close partnership between funders and the research community.
COLLABORATION AND DATA SHARING
As mentioned frequently throughout this workshop by several participants, collaboration and data sharing are essential to build very large and diverse datasets that can be analyzed with new innovative technologies, said Carole Ho, chief medical officer and head of development at Denali Therapeutics, Inc. Partnerships have sprung up across the neurology and psychiatry space, including academic partnerships, industry–academic partnerships, advocacy group partnerships, and public–private partnerships (PPPs) to achieve this.
Eline Appelmans, director of neuroscience research partnerships at the Foundation for the National Institutes of Health (FNIH), has been involved in developing and managing many such partnerships, including those focused on identifying and developing biomarker-based tools for clinical trials, which are needed to ensure better outcomes for patients. Uniting resources through PPPs enables funds to be leveraged in a precompetitive manner, she said.
Appelmans called for all sectors and stakeholders to continue to work collaboratively to ensure the best use of resources and build the trust of the community. She added that one of the advantages of partnerships such as the Accelerating Medicines Partnership (AMP) consortia is their ability to interact with other consortia. For example, AMP-Alzheimer’s disease (AD) scientists can draw on the expertise of scientists at other National Institutes of Health (NIH)-funded consortia, such as MODEL-AD (Model Organism Development & Evaluation for Late-Onset Alzheimer’s Disease)1 and TREAT-AD (TaRget Enablement to Accelerate Therapy Development for AD).2
FNIH is but one entity establishing PPPs. Ho cited a partnership between The Michael J. Fox Foundation and multiple industry and academic partners to understand the biology of LRRK2 mutations in Parkinson’s disease. Through this partnership and collaboration with academia, assays were developed to be taken into the clinic, said Ho. “Not having boundaries between industry, public advocacy groups, and others enabled us to come together very quickly to address some of the challenges in drug development,” said Ho. She said Denali has now completed Phase 1a and 1b studies on an LRRK2 inhibitor.
Henne Holstege of the Amsterdam Alzheimer Center also mentioned the need for large datasets to understand the impact of rare variants on AD. She and her colleagues are working on leading the analysis of 50,000 exomes as part of a large international collaboration of Alzheimer researchers. Yet she said due to the rarity of the variants, the current dataset only enables identification of low-hanging fruit, and that much larger datasets are necessary to confirm the promising signals they currently observe beyond the low-hanging fruit.
Many of these partnerships encourage data sharing. For example, Bradford Casey, senior associate director of The Michael J. Fox Foundation’s Research Programs division, said the Parkinson’s Progressions Marker Initiative (PPMI)3 has invested both in building biosample reposi-
___________________
1 To learn more about MODEL-AD, see https://www.model-ad.org (accessed November 29, 2021).
2 To learn more about TREAT-AD, see https://treatad.org (accessed November 29, 2021).
3 To learn more about PPMI, see https://www.ppmi-info.org (accessed December 1, 2021).
tories as well as in generating very specific datasets that go along with those samples, and making these resources available to the broader research community. Ernest Fraenkel mentioned the Pooled Resource Open-Access ALS Clinical Trials Database (PRO-ACT),4 which provides researchers with access to clinical trials data from 23 late-stage trials.
Building these extremely large datasets will require multiple technologies and areas of expertise and international collaboration, said Weninger. In addition to collaboration, Weninger noted the importance of infrastructure that will ensure coordination to prevent different groups from working at cross-purposes. Moreover, said Weninger, enormously large genetic datasets will only be as good as the phenotyping associated with samples and the identification of subtypes within a given diagnosis. “Unless we can do a better job of characterizing, we are not going to be able to interpret the genetics or biomarkers, and all of these need to inform each other,” she said. John Ngai added that the community will also need to develop and agree on data and metadata standards, data integration tools, new and accessible algorithms and software, and an appropriate and inclusive governance structure to make sure all voices are heard and all data are being used, deposited, and organized for the benefit of everyone. Carrots work better than sticks, he said. The ultimate reward is to do better science by having access to data and ideas beyond those generated in one’s own group, Ngai added.
However, Weninger noted that complex international rules around privacy and data sharing present a major challenge. For example, while she applauded the NIH requirement for data sharing, many international participants are governed by the European Union’s General Data Protection Regulation (GDPR), which Weninger said “is seemingly in conflict with the NIH request to upload the data.” Relying on individual investigators to figure out how to navigate GDPR rules will not work, said Weninger. She urged the international scientific community to come together to develop non-conflicting rules that will enable data sharing.
SHIFTING THE TRAJECTORY OF THERAPEUTIC DEVELOPMENT FOR NEUROLOGICAL AND PSYCHIATRIC DISORDERS: STRATEGIES PROPOSED BY INDIVIDUAL WORKSHOP PARTICIPANTS
In the final session of the workshop, participants discussed specific next steps that could shift the trajectory of therapeutic development for neurological and psychiatric disorders.
___________________
4 To learn more about PRO-ACT, see https://ncri1.partners.org/ProACT/Document/DisplayLatest/9 (accessed December 1, 2021).
-
Operationalizing approaches for bringing together disparate areas of expertise and methodologies. Steven Hyman posed to workshop participants a call to operationalize how to bring all the disparate pieces together. One challenge he cited was the gulf between (1) neuroscientists’ reductionist experimental approach wherein as many variables as possible are controlled as a means of providing clear answers, and (2) the approach taken by geneticists and other population-based scientists who typically conduct more open-ended inductive work that yields statistical associations. Ngai agreed on the need to address this problem as a multidisciplinary group, adding that when bringing many different kinds of expertise together, it is important to find a balance between allowing the best science to bubble up from the bottom and providing a light top-down touch. For example, the BRAIN Initiative funds large, multisite, multi-investigator U19 grants that require a strong theoretical basis, but allow the team of investigators to self-assemble, giving them the space to realize their creativity.
Kafui Dzirasa, K. Ranga Rama Krishnan endowed associate professor in psychiatry and behavioral sciences, neurobiology, biomedical engineering, and neurosurgery at Duke University, added that engineers should also be included in these groups. Ngai mentioned that clinical partners are also needed. In support of the idea of bringing people together, Casey said he has been heartened by the emergence of platforms and consortia willing to not only share data, but also to build out the goals of a project that serves the entire community. Examples include GP2, the Global Parkinson’s Genetics Program, and the BRAIN Initiative Cell Census Network.5
Reflecting on the workshop discussions, Joshua Gordon, director of the National Institute of Mental Health, challenged workshop participants to develop a vision of what it would take to go from frog to mouse, to non-human primate, to human cells, to a drug for a neurological or psychiatric condition, including the collaborations and scientists needed for this approach to research.
- Recruiting and retaining a highly talented and diverse group of scientists to sustain progress. Complex problems require not only large amounts of data, but also diverse perspectives from scientists of different backgrounds and expertise and a diversity of model systems that reflect the ancestral heterogeneity of humans, said Ngai. Recruiting and retaining highly talented statisticians and bio-
___________________
5 To learn more about the BRAIN Initiative Cell Census Network, see https://braininitiative.nih.gov/brain-programs/cell-census-network-biccn (accessed November 30, 2021).
-
informaticians into academia is particularly challenging because of the great demand for these skills in the biotech industry and across many other sectors, said Ngai. This represents just one aspect of the larger challenge of recruiting and retaining scientists given the traditional academic models for career advancement, he added. Much of the work today is done by senior staff scientists, he said, and there is an important place for highly trained professional career scientists in mathematics, engineering, and wet-lab sciences to make huge contributions. New strategies are needed to address issues of compensation, job stability, workplace environment, and credit attribution for these important contributors, he said.
Amir Tamiz added that mid-career scientists interested in moving into translational research are well positioned because the toolbox has expanded so much. He added that the most prolific Principal Investigators are those who never stop collaborating and have a network of trusted experts to give them advice. They surround themselves with consultants and other partners, seek out industry help, and work with sophisticated advocacy groups that have resources and data banks available for interrogation.
-
Promoting international information sharing and addressing privacy concerns. Another challenge mentioned earlier by Weninger is the difficulty of integrating international research. This would be necessary, for example, to unite Alzheimer’s centers throughout the world with the genetic data and biomaterials that have been collected over time, as was suggested by Henne Holstege of the Amsterdam Alzheimer Center.
Hyman added that biorepositories are needed that are not strictly focused on a single disease and that bring together diverse sources of genetics and diverse ancestries, and collect follow-on materials. For example, he said, 30,000 genetic samples have been collected in East and South Africa to study schizophrenia, but to interpret common variants would also require brain tissue.
Hyman suggested that the international community come together to stipulate the importance of sharing data and strategize about how to do that within the confines of different national laws. Weninger suggested getting well-versed legal teams to look at these issues and issue recommendations about what is and is not possible, which scientists and institutions could reference when designing projects.
- Involving patients and advocates at all levels of the research enterprise. “When advocates and patients get involved, things happen,”
- said Tamiz, noting that they should be involved at many levels, including educational, regulatory, and other policy decisions; study design; and postapproval monitoring. Weninger agreed, adding that patients need better education about privacy, the true risk of sharing their data, and the advantages to society when they are willing to take on some risk. “These are important questions that we need to discuss as a whole—patients, scientists, governments, everybody,” she said. Casey also agreed, saying that his experience interacting directly with patients and family members indicates that they want their data to be used toward cures. “This is why they contribute, why they participate,” he said. He added that many people have not been well served by how their data has been used in the past. “We need to be ready to answer some hard questions and advocate for solutions that are going to work toward responsible use and sharing.”
-
Learning from failed trials. When everything possible is done to ensure good trial design, yet the trial fails to demonstrate efficacy, determining why the trial failed is essential, yet is often a step that is overlooked, said Krainc. Ho agreed, noting that when this happens and sponsors are left with no answers after a very large and expensive trial, they often have little appetite to try again. “The way I look at drug discovery is that it is okay to fail as long as you have learned something,” said Ho. If you have gone in with the right dose and the rationale is strong, she said, what you have probably learned is that at the stage of disease that you targeted, the disease is not modifiable.
One strategy that Ho believes could help alleviate this problem would be to build collaborative cross-sector studies to understand better the trajectory of the disease from the time a patient becomes symptomatic to the time when the disease is significantly affecting their lives. Hyman added that such natural history phenotyping and biomarker studies that require substantial patient involvement also require the expertise of physician-scientists, who are currently in short supply. He suggested that the need for these studies might motivate NIH to increase support for training physician-scientists. Ho added that strategies are also needed to reimburse clinicians for the time they spend collecting data and samples and providing genetic counseling to patients.
- Creating incentive structures that encourage generalization. Dzirasa said that most of the incentive structures for researchers encourage overfitting data on a certain pathway to a specific mouse strain or
- population rather than ensuring that the phenomenon generalizes. Declaring victory over progress made in genomics, he said, encourages investment in understanding and developing treatments that may be applicable to a small percentage of the world because of overfitting, he said.
- Integrating biomarkers much earlier into therapeutic development. To support translational research, Tamiz suggested making stratification a key criterion when creating a target product profile. He noted that stratification science requires both early discovery and validation, and urged funders, including NIH in its role as a catalytic force, to focus resources on validation of biomarkers that could be readily adopted by industry. He suggested that de-risking this aspect of therapeutic development could galvanize industry to commit resources to neuroscience and other areas.
- Investing more heavily and incentivizing multimodal biomarker development. Hyman said multimodal biomarkers, each independently significant and robust and with studies adequately powered and designed, are needed to advance therapeutics in psychiatry. Fraenkel agreed, suggesting that a challenge modeled on the DREAM Challenges may be needed to incentivize researchers in the field to come up with robust prediction technologies.6
In closing the workshop, Hyman acknowledged the many critical issues raised and solutions that have been suggested. The next step, he said, will be “turning all of these good ideas into realities, outcomes, and sure ways of implementing this vision of science.”
___________________
6 To learn more about DREAM Challenges, see https://sagebionetworks.org/researchprojects/dream-challenges-powered-by-sage-bionetworks (accessed December 1, 2021).
