Vanderbilt University Researchers Combine Genetic Data and EHR Records to Identify Undiagnosed Disease in Patients
Multi-university research group discovers that heart arrhythmia genes may be more common than previously thought
For years, big data has been heralded as the key to unlocking the promise of personalized medicine. Now, researchers at Vanderbilt University are bringing that goal a step closer to reality by combining genetic testing data with data stored in electronic health record (EHR) systems to reveal undiagnosed disease in individual patients.
Should their research result in new ways to identify and diagnose disease, doctors and clinical laboratories would do confirmatory testing and then initiate a precision medicine plan.
Vanderbilt University Medical Center (VUMC) led a multi-university team of researchers that used data from the eMERGE (Electronic Medical Records and Genomics) network in two separate studies. eMERGE is a consortium of medical centers funded by the National Human Genome Research Institute (NHGRI) for the advancement of EHR data in genomics research.
The first study, published in the journal Circulation, titled, “Arrhythmia Variant Associations and Reclassifications in the eMERGE-III Sequencing Study,” looked at 10 arrhythmia-associated genes in individuals who had no prior indication for cardiac genetic testing.
The second study, published in Jama Oncology, titled, “Association of Pathogenic Variants in Hereditary Cancer Genes with Multiple Diseases,” explored the spectrum of diseases associated with hereditary cancer genes.
Dan Roden, MD, Senior Vice President for Personalized Medicine at VUMC and Senior Author of the Circulation study, said in a VUMC news release that the findings support the growing use of genetic information in clinical care.
“The questions we asked were: How many people who had no previous indication for cardiac genetic testing had pathogenic or likely pathogenic variants, and how many of those people actually had a phenotype in the electronic health records?” he explained.
Arrhythmia More Common than Previously Thought
The VUMC researchers drew data for their reports from the eMERGE Phase III study, which investigated the feasibility of population genomic screening by sequencing 109 genes across the spectrum of Mendelian diseases—genetic diseases that are caused by a mutation in a single gene—in more than 20,000 individuals. The scientists returned variant results to the participants and used EHR and follow-up clinical data to ascertain patient phenotypes, according to a Northwestern University Feinberg School of Medicine news release.
The research team looked specifically at the 120 consortium participants that had disease-associated pathogenic or likely pathogenic (P/LP) variants in the arrhythmia-associated genes. An analysis of the EHR data showed that 0.6% of the studied population had a variant that increases risk for potentially life-threatening arrhythmia, and that there was overrepresentation of arrhythmia phenotypes among patients, the VUMC news release noted.
The research team returned results to 54 participants and, with clinical follow-up, made 19 diagnoses (primarily long-QT syndrome) of inherited arrhythmia syndromes. Twelve of those 19 diagnoses were made only after variant results were returned, the study’s authors wrote.
Carlos G. Vanoye, PhD, Research Associate Professor of Pharmacology at Northwestern University (NU), said the study suggests arrhythmia genes may be more common than previously thought.
“A person can carry a disease-causing gene variant but exhibit no obvious signs or symptoms of the disease,” he said in the NU news release. “Because the genes we studied are associated with sudden death, which may have no warning signs, discovery of a potentially life-threatening arrhythmia gene variant can prompt additional clinical work-up to determine risks and guide preventive therapies.”
“The take-home message is that 3% of people will carry a pathogenic or likely pathogenic variant in a disease-causing gene and many others will carry variants of uncertain significance,” said Dan Roden, MD (above), Senior Vice President for Personalized Medicine at VUMC and Senior Author of the Circulation study in the VUMC news release. “We can use genetic testing, electronic health record phenotypes, and in vitro technologies to evaluate and find people who have unrecognized genetic disease and save lives by making earlier diagnoses.” Clinical laboratories will play a key role in making those early diagnoses and in managing personalized medical treatment plans. (Photo copyright: Vanderbilt University.)
Variants of Uncertain Significance
According to the NU news release, the scientists determined the functional consequences of the variants of uncertain significance they found and used that data to refine the assessment of pathogenicity. In the end, they reclassified 11 of the variants: three that were likely benign and eight that were likely pathogenic.
In the JAMA Oncology study, the VUMC scientists and other researchers conducted a phenome-wide association study to find EHR phenotypes associated with variants in 23 hereditary cancer genes. According to the VUMC news release, they identified 19 new associations:
- Seven for different cancers, and
- 12 for conditions including ovarian cysts, acute pancreatitis, and chronic gastritis, in addition to replicating established gene-cancer associations.
The VUMC study findings could improve disease diagnosis and management for cancer patients and help identify high-risk individuals, the researchers noted in their published report.
Other Scientists Urge Caution
In an editorial published in Circulation, titled, “First Steps of Population Genomic Medicine in the Arrhythmia World: Pros and Cons,” the professors noted that using genomic information in the case of potentially lethal inherited arrhythmia syndromes could be “lifesaving,” but questioned the benefits of reporting such secondary findings when patients are undergoing genome sequencing for other indications such as cancer.
“The likelihood that these ‘genetic diagnoses’ are translated into clinical diagnoses have not been completely evaluated,” they wrote. “In addition to the challenge of accurately identifying disease-causing genetic variants, defining the penetrance of such variants is critical to this process, i.e., what proportion of individuals in the general population with apparently pathogenic variants will develop the associated phenotype? If penetrance is low for particular gene/phenotype combinations, the costs associated with clinical screening and the psychological effects on individuals informed that they have potentially life-threatening variants may outweigh the benefits of the few new clinical diagnoses.”
These latest studies provide further evidence of the value of big data in healthcare and offer another lesson to clinical laboratories and pathologist about the future role data streaming from clinical laboratories and pathology assays may have in the growth of personalized medicine.
—Andrea Downing Peck