Fresh research suggests genetic testing should be recommended for early-onset atrial fibrillation
A new cohort study from Vanderbilt University Medical Center explores the prevalence of disease-associated variants in genes in some typical cardiology panels for patients with early-onset atrial fibrillation.
Advances in genomics research have seen recent breakthroughs in clinical diagnosis, and as the technologies improve so does the accuracy and pricing of whole-genome sequencing (WGS). Initiatives like The Trans-Omics for Precision Medicine (TOPMed) program collect whole-genome sequencing and other -omics data to help study a range of health issues and diseases in order to advance precision medicine.
This cohort study, led by Dr. Benjamin Shoemaker, used Franklin’s AI-powered genomics engine to analyze WGS data sequenced through the TOPMed program. The aim was to discover whether genetic testing should be recommended for patients with early-onset atrial fibrillation, since prescribing cardio panels is not currently standard practice for people diagnosed with this heart-rhythm abnormality.
Atrial Fibrillation (AF) affects more than 33 million people worldwide
AF is the most common type of arrhythmia and can be the initial manifestation of more serious underlying inherited heart diseases, such as cardiomyopathy or arrhythmia syndromes.
In an effort to explore the relevance of genetic testing in cases of early-onset AF, researchers at Vanderbilt University conducted a cohort study with nearly 1300 patients diagnosed with this cardiologic abnormality before the age of 66.
Automated AI-based classification to accelerate genetic research
The study analyzed whole-genome sequencing data from participants in 145 genes found in typical commercial arrhythmia and cardiomyopathy gene panels.
The research team used Franklin for variant interpretation and prioritization, harnessing the platform’s community-powered, automated, artificial intelligence-based ACMG classification.
As time is a major constraint in scientific studies, the Franklin platform allowed the research group to process thousands of VCF files of the patients’ genomes more efficiently due to its automated variant classification.
Franklin’s AI classification helped substantially narrow down the list of relevant variant candidates from millions to only 427 variants
Based on well-known open data sources like ClinVar, ClinGen, Uniprot, and gnomAD, Franklin’s AI engine builds disease association and deleteriousness prediction models on gene and variant levels. The platform’s proprietary algorithm classified each variant according to the ACMG guidelines as Benign (B), Likely Benign (LB), Likely Pathogenic (LP), Pathogenic (P), and Variant of Unknown Significance (VUS).
As shown in the figure below, the automated variant classification engine helped dismiss the benign and likely benign variants present on the patients’ genomes, leaving researchers with only 427 variants (181 likely pathogenic or pathogenic + 246 VUS leaning to pathogenic) requiring manual curation.
In addition, it’s worth noting that over 93% of the variants retained Franklin’s automated classification after being manually reviewed by a panel of independent, blinded reviewers.
Disease-associated rare variants were found in patients with diagnosed AF
Franklin’s automated AI-based variant classification was a key player in helping Dr. Shoemaker and his team achieve this goal, significantly reducing their workload and allowing the team to focus on manually reviewing the critical variants’ pathogenicity in depth.
This led to the Vanderbilt scientists finding that disease-associated rare variants in cardiomyopathy and arrhythmia genes appeared in 10.1% of all the patients, with an increase to 16.8% in those younger than 30 years old when diagnosed with AF.
Additionally, they observed that those variants were more likely to be on genes associated with inherited cardiomyopathies.
Community-powered knowledge on variants can make a difference in the clinic
This cohort study showed the presence of LP/P variants in heart disease-associated genes, which means that genetic testing could promise a significant impact on an AF patient’s future prognosis.
The study not only generates new evidence regarding the prevalence of disease-associated variants in patients with early-onset AF, but also shows how community data and Franklin’s AI can be crucial in providing new clinically-relevant insights, especially for cohort studies.
In line with Genoox’s commitment to encouraging collaboration within the genetics field, data from all the variants identified in this analysis were uploaded to the Franklin Community, and are now available for the benefit of geneticists worldwide.
View the full study here: https://jamanetwork.com/journals/jamacardiology/fullarticle/2783693
About Genoox and Franklin
Genoox, the company behind Franklin’s AI engine and the community, believes that the availability of shared scientific knowledge will guide geneticists to better-informed decisions for their patients.
It’s one of the reasons that we’re so eager to collaborate with research initiatives — not only to support our own research but to help the community make AI prediction technology a viable diagnostic tool.
Learn more about the Franklin platform and how it can help accelerate your research.