NEW YORK – Researchers have tied numerous genetic loci to various aspects of cardiac biology captured by electrocardiograms.
ECGs are used to diagnose a range of cardiac disorders, including conduction disturbances and arrhythmias. Genetic studies have previously focused on variants associated with particular aspects of the ECG, such as the QT interval or QRS duration.
In a new analysis, the University of Groningen's Pim van der Harst and his colleagues instead used ECG data collected by the UK Biobank to search for genetic variants that affect the complete cardiac cycle. As they reported Friday in the journal Cell Systems, the researchers uncovered more than 300 loci associated with ECG features, including a number of variants associated with risk of dilated cardiomyopathy.
"Studying the complete ECG in an integrative manner provided a unique opportunity to obtain more meaningful insights into biology compared with studying individual ECG fragments one at a time," van der Harst and his colleagues wrote in their paper.
They obtained the raw ECG signal data for more than 77,000 UK Biobank participants for phenotyping analysis. For each person in the dataset, they averaged the ECG recordings of multiple heartbeats to one that represented the heart at rest. Each of these averaged ECG recordings contained 500 data points reflecting the different voltages the ECG detected.
To confirm that this approach captured known ECG traits, they conducted an association analysis between the ECG morphology captured by those 500 data points and the polygenic risk scores associated with classical ECG traits. They found, for instance, that the polygenic risk score for the PR interval was also associated with a shift in the P wave, and that the polygenic risk score for QRS duration was associated with Q and S wave durations.
They also noted, however, a range of ECG morphologies beyond those classical patterns, suggesting underlying biological differences.
Through a genome-wide association study, the researchers identified 414 variants in 331 two-megabase regions associated with ECG morphology. Of these genetic association signals, 179 were novel, and the strongest signals were near the TBX3, SCN5A, and KCNQ1 genes.
Clustering analysis of the ECG morphology profiles uncovered five groupings, four of which broadly aligned with differences in Q, P, R-S, and T wave morphology. The fifth cluster was more diverse.
The Q-wave cluster, the researchers noted, overlapped with dilated cardiomyopathy risk. In particular, one variant near BAG3 that was associated with the Q-R upslope at -18 ms of the ECG was previously identified in a GWAS of dilated cardiomyopathy. That GWAS similarly identified a variant near HASP/CLCNKA that is also associated with a similar ECG pattern as the BAG3 variant, indicating there may be a common feature between this portion of the ECG trace and dilated cardiomyopathy.
This suggested to the researchers that ECGs might be able to spot people at risk of dilated cardiomyopathy, a severe disease with a five-year survival rate following diagnosis of 50 percent.
To test this hypothesis, they conducted a Mendelian randomization analysis using the UK Biobank dataset and found the -18ms data point was associated with dilated cardiomyopathy, a finding they further confirmed in the MAGNet study cohort. They additionally uncovered variants near PRKCA, TMEM43, and OBSCN linked to dilated cardiomyopathy.
"Our results demonstrate how an integrated approach to analyze high-dimensional data can further our understanding of the ECG, adding to the earlier undertaken snapshot analyses of individual ECG components," the researchers wrote.
They noted, however, that their analysis is limited by its reliance on common variants, a study population that was primarily European, and an ECG that used a limited number of sensors.