2

Existing Precision Medicine Initiatives

UNITED STATES: ALL OF US

Petra Kaufmann, director of the Office of Rare Diseases Research at the National Center for Advancing Translational Sciences, told workshop

participants about the All of Us research program that is under way at the National Institutes of Health. The program is an effort to move precision medicine forward by collecting and analyzing a wide array of data from at least 1 million Americans. Kaufmann first laid out the rationale behind the program, observing that the current system of research and clinical care has disadvantages for patients, providers, and researchers. Clinical treatments are usually geared toward the average patient, and finding the right treatment for any individual patient may be the result of years of trial and error. Providers lack the time to thoroughly analyze each patient’s biology, medical history, and symptoms, and there are too few data underpinning the available clinical guidelines. Kaufmann described the challenging process for a provider trying to do the detective work to diagnose a patient: Internet research, emails to people in the field, and “pulling together puzzle pieces from PubMed” to diagnose one patient. She noted that this process is not sustainable or scalable in the current healthcare environment, but that precision medicine initiatives like All of Us may address these issues. Challenges for researchers, she said, include difficulties in collecting large and diverse sample sizes, building and maintaining systems for storing and analyzing the data, and sharing and collaborating with other researchers.

The mission of All of Us is to accelerate health research and medical breakthroughs, enabling individualized prevention, treatment, and care for all Americans (https://allofus.nih.gov), said Kaufmann. This goal will be met by nurturing relationships with 1 million participant partners, developing a large and rich biomedical dataset, and catalyzing a robust ecosystem of researchers and funders to use and support the system. The program will collect clinical, environmental, lifestyle, and genetic data, with participants consenting to data collection on an ongoing basis throughout their lifetime. Data will be collected using a wide variety of methods, including biospecimens, wearable technologies, electronic health records, and surveys. The program seeks to recruit a highly diverse group of participants and will particularly over-recruit populations who are traditionally underrepresented in biomedical research. Participants will have access to information about themselves, and the program will seek to earn the trust of participants through engagement and full transparency. Kaufmann noted the pervasive concern about data privacy and said that, while of course it is critically important to protect privacy, she does not want the pendulum to swing too far toward not collecting or sharing data that could lead toward treatments and cures for disease. The program will seek participation from a diverse group of researchers, from citizen scientists to premier university labs. By making the data accessible to all, All of Us hopes to apply “more brainpower per problem” and be a catalyst for innovative research. Kaufmann said that All of Us is seeking to transform the traditional method of research—in which data are brought to researchers through a complex maze of issues including

data sharing protocols, huge infrastructure, and cumbersome access—to a new approach that brings researchers to the data quickly and efficiently.

Currently, the program is in the closed beta phase. In this phase, Kaufmann said, the program will enroll 10,000-15,000 participants and will expand across the country slowly. All aspects of the program—including the website, consent forms, online tools, and staff workflow—will be tested and refined during this time, and participants will be asked to give feedback on their experiences. The national launch for the program is set for spring 2018.

UNITED KINGDOM: 100,000 GENOMES

Damian Smedley, director of genomic interpretation at Genomics England, explained the United Kingdom’s 100,000 Genomes project. The project involves sequencing 100,000 genomes, including genomic, phenotypic, and other clinical data from 70,000 patients with rare diseases or cancer. The project is not just a genomic research project, said Smedley, but rather is aimed at transforming the healthcare system. While the main goal is to benefit patients in the UK healthcare system, a second goal is to kick-start the development of a UK genomics industry by creating the capacity, capability, and legacy of personalized medicine in the United Kingdom. The project utilizes 13 National Health System (NHS) genomic medicine centers around the country, responsible for recruiting patients, obtaining informed consent, and collecting samples. Sequencing is performed, and electronic health record data on the patients are collected and stored at a specially commissioned data center that is government-approved, secure, and behind an NHS firewall. Researchers can access and analyze de-identified data through a system akin to a reading library, which is designed to protect patient data while also making translations from research to the healthcare system quicker and more efficient. Academic researchers who want access to the data become members of a Genomics England Clinical Interpretation Partnership.

One of the primary reasons for and benefits of the project is being able to diagnose rare diseases. To do so, internal curation teams curate gene panels for every recruited disease category to look for variants using two software tools, Exomiser and Genomizer. Finally, all of the data and analysis are merged together for clinicians to review and share with patients. Smedley gave two examples of successful diagnoses stemming from the 100,000 Genomes project. One patient, a 4-year-old girl named Jessica, presented with epilepsy and developmental delay, and all of the standard genetic tests had not revealed the cause. After a genome-wide scan, researchers found more than 6 million variants in her genome, around 600,000 of which were rare. About 3,000 of these were predicted to cause

change in a protein, and only 67 of those were different from her parents’ genes. One of those was a gene that was listed in PanelApp, a crowdsourcing tool developed by Genomics England. A de novo mutation in SLC2A1 was identified as the cause of her Glut 1 deficiency syndrome, and she is now being successfully treated with a ketogenic diet. Another successful diagnosis was for a young girl named Georgia, who had developmental delays and multiple medical problems. Despite extensive genetic testing, no cause had been identified. 100,000 Genomes researchers found a de novo truncation in KDM5B, a newly recognized disease gene. While there is currently no treatment for her disease, the diagnostic odyssey has ended and gives her parents reassurance about the low risk of recurrence if they have another child, said Smedley.

The treatment cycle for diagnosing and treating a patient with a rare disease requires a complex chain of operations, most of which have not been designed or optimized for the purposes of genomic medicine, said Smedley. In particular, in order to integrate clinical and genetic data, the clinical data need to be captured and recorded in a standardized way across the system. Genomics England has created a clinical data capture system for rare disease diagnostics that includes defined Human Phenotype Ontology (HPO) terms for each disease category and a standardized system for imaging and lab test results. Based on the patient’s phenotypes, clinicians can decide on the relevant gene panels.

Genomics England is wrapping up a pilot test for rare diseases, and Smedley shared some of the results stemming from almost 5,000 enrolled patients with 170 different conditions. HPO phenotyping allowed for 12,966 positive and 43,088 negative annotations to be collected. The diagnostic rate is currently around 20-25%, but Smedley expects it to rise as further analysis is performed, because the diagnostic rates improve as more data are collected. The pilot project also demonstrated the benefit of software like Exosimer. Smedley said that, for the first 1,000 patients who received a likely diagnosis based on internal clinical review, Exosimer identified the right variant in 59% of the cases and identified the variant as a top five candidate in 88% of the cases. Exosimer has also found additional diagnoses for patients. For example, the gene panel for cataracts did not include a deletion in the SORD gene, due to limited evidence. However, after Exosimer flagged this mutation as a top five candidate for a patient with congenital cataracts, researchers found a mouse model with both the mutation and cataracts. This demonstrates the utility of the model organisms, said Smedley, as most useful in this realm of research candidates in the undiagnosed cases.

Animal models, said Smedley, will have two main purposes in the 100,000 Genomes research: to validate the variants, and to test potential treatments. He noted that they are likely to end up with 10,000 to 15,000

rare disease cases that do not have a diagnosis, and that model organisms will be critical for new disease gene discovery. He is particularly intrigued by the potential for using CRISPR/Cas-9 to quickly and efficiently develop precision animal models for functional validation and treatment testing. Smedley noted that the Genome Editing Mice for Medicine (GEMM) program, available for researchers to request mouse models, puts out a call every 9 months, and Genomics England has submitted several applications. He noted that similar programs are available in other countries as well, and that the International Mouse Phenotyping Consortium, with its huge number of mouse models, also takes requests for new ones. Genomics England is also participating in the Matchmaker Exchange program, which seeks to connect patients, diseases, and model organisms around the world to aid in diagnosis and mechanistic discovery, and also performs computational matching of rare disease patients across clinical and public sources.

Smedley concluded that the future of genomic medicine is dependent on deep and accurate phenotyping of clinical and model organisms and the use of precision animal models to validate genetic findings and test potential treatments. The 100,000 Genomes project will drive genomic medicine into the NHS by building capacity and capability for precision medicine, with the hope of achieving diagnoses, therapies, and opportunities for patients.

FRANCE: THE FRENCH PLAN FOR GENOMIC MEDICINE

Catherine Nguyen, director of ITMO Génétique, Génomique et Bioinformatique, told workshop participants about France’s recent foray into genomic medicine. Nguyen started by noting that, when a patient presents to a provider, there are a number of questions that are not immediately answerable, such as whether the patient’s therapeutic regime would be effective at this stage of the disease, what kind of rare disease this patient may have, what treatment is best suited for this patient, can the evolution of diabetes in this individual be predicted. Genomic medicine, said Nguyen, will hopefully provide new tools to start answering these types of questions.

Providing the historical context regarding the French plan, Nguyen explained that in 2015 the French prime minister asked Aviesan (the French National Alliance for Life Sciences and Health) to examine the current landscape of genomic medicine and its role in medicine, and to propose a plan to integrate genomics into France’s healthcare system and industrial sector. The prime minister specifically asked for a proposal of a long-term model that integrates the issues of insurance coverage along with the establishment of an industrial sector to support such an initiative. More than 160 people collaborated to answer the prime minister’s call and to produce a report entitled Genomic Medicine France 2025.

The report, said Nguyen, set a plan for genomic medicine over the next

10 years, with two major goals. First, sequencing should be integrated into routine clinical practice. Specifically, the goal is to implement a general healthcare pathway for all French patients with cancer, rare diseases, or common diseases that includes access to genomic medicine for all those concerned (patients and eventually family members depending on diagnosis) by 2025. Second, work must be done to develop a national genomic sector. Specifically, the goal is to place France among the leading countries in the field of genomic medicine in the next 10 years, to export its expertise developed in this area, and to establish a medical and industrial sector of genomic medicine.

Nguyen further elaborated that, on the first goal of integrating sequencing into routine healthcare, the report outlined a basic model for integration that included putting clinical and genetic data into a national database, and a framework for exchange between the diagnostic laboratory and clinical decision makers and patients. The report identified several needs that each stakeholder would have in such an integrated system. For example, the patient would want a diagnosis based on the integration of genomic data with clinical data, and adequate care by the healthcare system. The practitioner would need validated indications, digital files with genomics and clinical data for each patient, a system for identifying variants for a particular pathology, and tools for exploring data and developing therapeutic strategies. Based on the needs of stakeholders, there were three broad objectives identified: (1) implement the tools for a genomic healthcare pathway, (2) ensure operational implementation and growth, and (3) implement monitoring and management tools. Fourteen actions were identified that would help support these objectives. For example, in order to implement the tools for a genomic healthcare pathway, there were three actions that needed to be performed:

1. Create a network of sequencing platforms
2. Create a Central Analyzer of Data to process and use the volume of data generated
3. Allow the integration and use of patient data in the healthcare pathway

In order to test these objectives and actions, several pilot studies were initiated on diabetes, rare disease, and variant analysis of the general population. As part of this process, centers for reference, innovation, and expertise (CReflX) are being created. These centers will test innovations, evaluate the quality of data, and educate stakeholders.

Animal models, Nguyen said, will be important to the advancement of precision medicine in several ways: for increasing knowledge on gene function, for understanding variations in the genome and environmental

conditions, for enhancing knowledge of molecular mechanisms, and for validating therapeutic hypotheses. Importantly, there are drawbacks and limits to the use of animal models, including societal concerns about animal research generally, issues with reproducibility of animal studies, and determining the relevance of animal models to human health.

JAPAN: INITIATIVE ON RARE AND UNDIAGNOSED DISEASES

Kenjiro Kosaki, director of the Center for Medical Genetics at Keio University School of Medicine, told workshop participants about Japan’s Initiative on Rare and Undiagnosed Diseases (IRUD). IRUD is designed to help patients who have remained undiagnosed because their disease is so rare that doctors do not know it, or because their disease is a completely new and unknown condition. IRUD accepts undiagnosed patients who meet at least one of the enrollment criteria: congenital onset, familial occurrence, and/or multi-organ involvement. The IRUD process is multi-step, collaborative, and iterative, utilizing regional participating hospitals, clinical centers, and analysis centers. Eligible patients are referred up through the system, phenotyping and genetic sequencing are performed, and information and analysis are filtered back down to clinicians and patients for genetic counseling and potential treatment.

To analyze and evaluate the patients’ genetic information, IRUD uses a program that includes information about mouse phenotypes and associated human diseases. Normal Japanese variants are filtered out, using information from whole genome sequencing on 3,500 controls as well as a database of centenarians and super centenarians (i.e., over 110 years old). Kosaki noted that any variants that are observed in these long-living individuals are unlikely to be associated with one of the rare diseases.

IRUD started in July 2015, and in the 2 years since its inception, has already identified and published about 10 new diseases, said Kosaki. Twenty-five hundred patients have been accepted to the program, and the diagnostic success rate is around 30%. Kosaki gave two examples of new diseases that have been identified. In the first case, an 18-year-old undiagnosed patient with overgrowth, scoliosis, fragile skin, and intellectual disability was matched to three other patients from Japan and the United States with similar phenotypic features. Genetic analysis revealed that all patients had de novo mutations in the PDGFRB gene, though the mutations were not identical (Minatogawa et al., 2017). In addition to the three signaling pathways associated with overgrowth syndrome—IGF1, TGFb, and AKT1 receptors—the IRUD research demonstrated that PGDFRB can also cause overgrowth. A Belgian research group read the IRUD research findings and subsequently conducted in vitro studies that showed that the PDGFRB mutation could be suppressed by imatinib, a kinase inhibitor

(Arts et al., 2016). At the same time, another patient in Minnesota with a different mutation in PDGFRB was identified. Based on further in vitro work by the Belgian researchers, this patient was treated with imatinib, resulting in “objective improvement of debilitating hand and foot contractures and significant improvement in quality of life” (Pond et al., 2018). Kosaki said that this experience shows that making a diagnosis can lead to potential treatment, and that it took only 3 years between the discovery of the condition to the first human trial.

The second example Kosaki gave was a patient who suffered from intellectual disabilities, lymphedema of the legs, and increased platelet size with concomitant decreased platelet count. Genetic analysis showed a novo mutation in CDC42, which in mouse models was associated with large, albeit fewer, platelets, and brain defects. Kosaki presented this case at a local genetics meeting in Japan, and one of the attendees recognized the phenotype from one of his patients. This attendee’s patient had the same de novo mutation, while several more patients have been identified since then with the same or similar mutations and similar phenotypes. This new disease—a diagnosis of thrombocytopenia and intellectual disability, with a CDC42 mutation—has been classified as Takenouchi-Kosaki syndrome.

Kosaki noted that the process of identifying Takenouchi-Kosaki syndrome was dependent on a very astute audience member with a photographic memory, as well as the previous publication of CDC42 knockout mice studies. He warned, however, that we should not be relying on these types of fortuitous circumstances but should instead rely on more structured approaches, such as cross-species phenotype ontology and systematic case matching. The HPO is one tool that can help with this type of systematic approach, said Kosaki, as it provides a standardized vocabulary of phenotypes associated with human disease and allows researchers from around the world to compare cases and find matches. Kosaki reported that they have succeeded in using Google translation technology to translate Japanese medical items directly into HPO, which is a versatile method for not only Japan but other countries. Kosaki said that the IRUD project has begun collaborations beyond the borders of Japan and has analyzed samples from 37 patients from other countries. This type of global collaboration is essential to identifying rare and undiagnosed diseases and, hopefully, finding effective treatments for these patients.

WORLD ECONOMIC FORUM: CENTER FOR THE FOURTH INDUSTRIAL REVOLUTION

The World Economic Forum (WEF), said Genya Dana, project head for the Precision Medicine project, is a membership-based, international organization that serves as a convening platform for bringing together lead-

ers and stakeholders around the world who are committed to shaping the future of the world and making it a better place. In recent years, the WEF has moved beyond a convening function into taking a more active role, said Dana. As part of these efforts, WEF launched the Center for the Fourth Industrial Revolution in March 2017, focused on emerging technological areas. This project is aimed at developing and testing policy frameworks and protocols that maximize the benefits of emerging technologies while minimizing the risks to societies. The project has nine program areas: Artificial Intelligence and Machine Learning; Blockchain (Distributed Ledger Technology); Future of Drones and Tomorrow’s Airspace; Future of Urban and Autonomous Mobility; Internet of Things and Connected Devices; Digital Trade and Cross-Border Data Flows; Environment and the Fourth Industrial Revolution; Future of Production; and Precision Medicine.

The Fourth Industrial Revolution project collaborates with global partners, including companies that are committed to a leadership role in the revolution, as well as partners at all levels of government. These government partners will send a representative to the WEF Center in San Francisco for up to 18 months; these representatives will collaborate with each other and other partners on co-designing pilot projects and will help guide the development of governance norms, protocols, partnerships, and standards. Rwanda and Japan have signed on as government partners, and corporate partners include Kaiser Permanente, Microsoft, and the American Heart Association.

The Precision Medicine project, said Dana, has the objective of shaping the trajectory of precision medicine to promote societal benefits while minimizing risk, and will do so by collaboratively designing and testing governance approaches for precision medicine through pilot projects. Dana noted that precision medicine has long been an interest and a topic of discussion among WEF’s members, but that this specific project is new and in the earliest stages. The project will be divided into three phases. First, the landscape of precision medicine will be surveyed, key stakeholders will be identified, and the barriers that are preventing precision medicine from moving forward will be described. Next, the project will work with its government partners to determine what some of the priority areas are and how they could be translated into a pilot project. For example, if partners are concerned about data sharing, WEF would look at the barriers to data sharing, the policy challenges involved, and how current initiatives could be joined or utilized as part of a pilot project. Finally, pilot projects will be designed and launched in collaboration with government partners and other stakeholders. An assessment after 9 months will evaluate the projects and determine whether and how they could be expanded into other communities.

Dana enumerated the challenges identified thus far by the Precision Medicine project:

• Evidence of efficacy generation
• Data sharing and infrastructure
• Regulatory environments
• Integration into clinical practice
• Pricing and reimbursement pathways
• Patient and public engagement

Dana said that the Precision Medicine project is a global scoping exercise and a global endeavor; therefore she welcomes the opportunity to collaborate with all relevant stakeholders from around the world.