Precision medicine informatics

This special issue on precision medicine informatics flowed from the AMIA 2015 Translational Bioinformatics Summit theme of “Accelerating Precision Medicine”¹ and President Obama’s 2015 State of the Union call “to give all of us access to the personalized information we need to keep ourselves and our families healthier.”² The goal is to focus on the inherent translational informatics challenges, concerns, and opportunities afforded by precision medicine to provide an accurate, personalized characterization of patient populations based on various characteristics including molecular (eg, genomic, proteomic), clinical (eg, comorbidities), environmental exposures, lifestyle, patient preferences, and other information.^4–13 Informatics is a necessary component in a comprehensive initiative to tackle precision medicine. Its roles include (1) managing big data,^15–17 (2) creating learning systems for knowledge generation,^18–21 (3) providing access for individual involvement,^22–25 and (4) ultimately supporting optimal delivery of precision treatments derived from translational research.^26–29 The papers submitted for this special issue cover these 4 areas and expand on the themes of informatics emerging in the field of precision medicine.

BIG DATA

Data become “big” when the scale of the data and analyses challenge our traditional systems in size, velocity, or complexity.⁸ The large scale of the Million Veterans Program is an example, with systems and analysis methods being developed for application in a repository that has established policies and procedures. The Million Veterans Program can serve as a model for involving thousands of individuals in precision medicine research.³⁰ Similarly, the Precision Medicine Initiative Cohort Program seeks to connect electronic health records with participant-provided data, molecular determinants, environment, and lifestyle patterns to deeply impact our knowledge of health and therapies.³¹ The ambitious goal of enrolling 1 million volunteers whose demographics reflect the diversity of the US population can only be accomplished with robust and scalable informatics. Tenenbaum et al.^3,32 present an informatics perspective and describe key informatics innovations required to advance precision medicine.

Big data can also be characterized by the velocity of data, such as in real-time data collection. We have come to expect nearly instant access and tracking of information that improves our quality of life.^33–35 With precision medicine, the informatics research community is responding to the challenge and is providing applications relevant to the health of individuals. The Substitutable Medical Applications, Reusable Technologies (SMART) application by Warner et al.³⁶ provides the ability to visualize genomic information for oncology treatment using mobile technology. This can facilitate individualized communication about cancer treatment options in the context of genomic information. Personalizing treatment using the vast volume of available molecular data requires tools that reduce cognitive load.^14,37 The tools described by Xu et al.³⁸ create a knowledge base of 2024 genomically informed clinical trials and treatments to support the delivery of personalized cancer therapy. Fathiamini et al.,³⁹ Hintzsche et al.,⁴⁰ and Cheng et al.⁴¹ describe resources that identify variants relevant to therapeutic treatments, variant calls, and natural language processing approaches to create structured information resources.

LEARNING SYSTEMS

Informatics has the potential to enable the collection, analysis, and reuse of data to facilitate learning knowledge representations for a wide range of diseases using phenotypes, genotypes, and predictive analytics.^42–51 Halpern et al.⁵² present an approach to derive computable phenotypes using machine learning techniques that reduce the need to manually create phenotypes, which can be expensive and time-consuming. The use of machine learning techniques to establish disease-mutation relationships is described by Singhal et al.⁵³ Rioth et al.⁵⁴ describe automatic parsing and categorizing of molecular profiles gathered in routine care into a database that can inform research and clinical practice. Hoffman et al.⁵⁵ describe guidelines for incorporating pharmacogenetic tests into clinical decision support systems. The integration of epidemiological evidence into knowledge representations that can inform care is explored by Torosyan et al.⁵⁶ The integration of information from a wide variety of sources combined with learning systems promises to speed discovery.

With patients and physicians making decisions based on genetic tests that identify actionable risks, a patient’s individual attitudes can influence how the information is accepted and used. Strategies and frameworks that address engagement and disparities will need to be developed.^57–62 Dye et al.⁶³ show that some minority populations are less likely to want genetic testing or participate in genetic research, which has the potential to exacerbate existing disparities. Precision medicine is focused on the individual patient. Given that molecular measurements are increasingly guiding care, studies will be needed to understand how to improve acceptance and participation by minorities. Adams and Petersen⁶⁴ discuss ethical, legal, and social issues that can arise when large numbers of individuals participate in genetic research. A framework for addressing ethical, legal, and social challenges can facilitate trust while protecting the individual and advancing research. Ni et al.⁶⁵ describe a learning framework to predict patient involvement in clinical trials with the goal of improving the effectiveness of recruitment.

PRECISION TREATMENT

The evaluation of methods to match genomics to therapeutics is an active area of research.^66–74 Eubank et al.⁷⁵ developed an informatics solution to manage genetic/genomic and clinical data to expedite clinical trials of targeted cancer therapies at Memorial Sloan Kettering Cancer Center. As of August 2015, the system contained data on 159 893 patients, with 64 473 being tracked across 134 research cohorts with 51 192 patients in a genotype-matched eligible pool. The system can serve as a model for cohort programs in cancer clinical trials. Patients will want to know the outcomes for other patients in similar situations, and Warner et al.⁷⁶ provide a system that tracks clinical outcomes over time. Access to such information is vital to informed decision makers who are involved in keeping themselves and their families healthy. The direct relationship between clinical trials and knowledge representations is demonstrated through a knowledge base for proper cancer drug selection and a The Cancer Genome Atlas (TCGA) analysis of triple-negative breast cancer, where 71.7% of 85 cancer patients had Food and Drug Administration (FDA)-approved “drug-able” genomic targets.⁷⁷

Informatics tools will be required for precision medicine to both catalyze research on huge populations and enable implementation of precision care now done for isolated cases in routine care. With the powerful response of the translational informatics community to the request for papers for this precision medicine special focus issue, we can chart the main themes of the course ahead. The initial response is the beginning of a transition in health care that is supported through informatics to propel participating individuals into the center of research and care. The needs of such an interactive model of care are broad and will need to be further defined as the initial systems are tested and measure how, when, and where they improve the health and outcomes of individuals and families.

FUNDING

The work of L.J.F. was supported in part by National Institutes of Health (NIH) grants 1R01GM108346-01 and U54-GM104941, Health Equity and Rural Outreach Innovation Center grant CIN 13-418, and funding from the Hollings Cancer Center’s support grant P30 CA138313 at the Medical University of South Carolina. The work of E.V.B. was supported by National Cancer Institute (NCI) U01 CA180964, the Sheikh Bin Zayed Al Nahyan Foundation, the Cancer Prevention Research Institute of Texas Precision Oncology Decision Support Core RP150535, National Center for Advancing Translational Sciences (NCATS) grant UL1 TR000371 (Center for Clinical and Translational Sciences), the Bosarge Foundation, and an MD Anderson Cancer Center support grant (NCI P30 CA016672). The work of J.C.D. was supported by R01 LM 010685, R01 GM 103859, and R01 GM 105688 from the NIH.