President's Corner - Accessible Patient Data
Published on
I applaud the efforts of scientists and physicians from around the globe to save and make available genetic and clinical information from individual patients (Mandl and Kohane 2015. PubMed ID: 25850061; Topol 2015. PubMed ID: 26065035; The Global Alliance for Genomics and Health, 2016. PubMed ID: 27284183). Data must be saved even after a patient leaves a health care system and even after the patient dies. Data on individual patients from different health care systems must be combined. And the data must be available to health care providers, testing labs, and researchers. This is indeed a huge undertaking, and progress in the beginning will likely be slow, but the benefits to patients are enormous.
Providers treating an individual patient must of course have access to the patient’s genomic sequence, even if that sequence was generated many years ago in a different state or country. Providers also need access to information from relatives of the patient. As human geneticists, we know the importance of family studies for variant interpretation. Detecting de novo variants, determining variant phase, and following the segregation of variants in affected and unaffected family members dramatically improves interpretation. As genome sequencing expands, more and more patients will have extensive genetic data, but of course, we also need at least a synopsis of clinical information. Relying on family lore for clinical information is a distant second best.
Testing labs need data from unrelated patients to accurately interpret variants. If we have clinical information from 10-20 unrelated individuals who are either heterozygous (dominant disorders) or homozygous (recessive disorders) for a specific variant of interest, then we can often determine whether the variant is a cause of disease. We can also make rough estimates of penetrance, which are essential for interpretation, prognosis and reproductive planning. Because carriers of rare variants (and specific combinations of more common variants) are present in populations at low frequencies, very large databases from millions of patients will be necessary.
Researchers, both from not-for-profit and for-profit organizations, will of course use the patient data in myriad ways (see for example Haggerty et al. 2017. PubMed ID: 28471438; Abul-Husn et al. 2017. PubMed ID: 28008010; Bastarache et al. 2018. PubMed ID: 29590070). We can no longer rely on just dedicated research studies for data. We must use patient data. There are not enough research funds to collect the information from the millions of patients that will be needed.
Funding of these databases is another obvious problem that will need to be solved. Sale of the de-identified data to pharmaceutical and other for-profit companies may be a way to provide at least a portion of the required financial support. Many testing labs are selling patient data to such companies today. More about that in the next President’s Corner.