Putting a finer point on it – resolving QT loci using multi-ethnic studies

While we are familiar with the major currents underlying cardiac repolarization, there are still many aspects this complex process that remain elusive. Given that abnormalities in repolarization can be associated with a range of disease states, familial arrhythmic syndromes, adverse drug effects and potentially deadly arrhythmias, there remains an ongoing need to define the mechanistic pathways underlying repolarization. One approach has been to use population based genetic studies to identify loci associated with variation in the QT interval. The first genome wide association study (GWAS) for the QT interval which was published a decade ago, highlights both the promise and the limitations of this technique.¹ While this initial effort led to the identification of a new gene, NOS1AP, for cardiac repolarization, translating genetic associations into a mechanistic understanding of a trait or disease can be challenging. In the case of the NOS1AP locus, it has taken years to develop a fuller appreciation of how this gene regulates repolarization and to identify the functional variation at the locus.^{2, 3}

These challenges are only magnified as the number of loci for the QT interval has continued to expand in recent years.⁴ For example, at many GWAS loci there may be multiple potential genes at a given locus so knowing which is the causative gene can be difficult. At some loci there may be an obvious potential gene, but at many loci, it is often unclear how any of the genes in the region may relate to repolarization. Further, since most GWAS signals are located in non-coding regions and are presumed to modulate the expression of a gene or genes in the region, it can be difficult to identify the functional genetic variants - particularly when the genetic loci can stretch over tens, or hundreds, of thousands of bases.

In order to narrow down a genetic locus to a more approachable size, it is helpful to look at the same trait across multiple races and ethnicities. Given the genetic divergence between different races, loci that are fairly broad in individuals of European ancestry may be dramatically narrowed when compared to individuals of other races and ethnicities. By fine mapping, or densely assessing the genetic variation in a given region, you have the opportunity both to narrow the region of interest and to determine if there is one or multiple distinct association signals in a region. These smaller, more tractable regions could then be combined with known epigenetic data to identify functional genetic variants that modulate enhancer activity, and in turn, gene expression.

In the current issue, Drs. Avery, North and colleagues have performed such fine mapping analyses for a number of QT interval loci in African-American and Hispanic/Latino populations.⁵ They used the Metabochip array, a genotyping array that contains both markers for known cardiovascular and metabolic GWAS loci, and a series of variants that allow for dense genotyping or fine mapping at 16 of the known QT loci. Using data from over 12,000 African-American and nearly 15,000 Hispanic/Latino individuals, they were able to refine the association signal for the QT interval at 3 loci. They were also able to identify unique associations in trans-ethnic or Hispanic/Latino populations at 6 other QT loci.

The strengths of the current manuscript are the large sample sizes, the multi-ethnic nature of the study and the dense coverage of some of the QT loci. As acknowledged by the authors, the challenge in the current work is that the chip used in this project is now somewhat dated and only contains a subset of the known QT loci.

Ultimately, to resolve these genetic loci for the QT interval as well as other cardiac traits and conditions, it will be necessary to perform large scale genome sequencing in individuals of multiple races and ethnicities. Ongoing projects such as the NHLBI TOPMed program, which is currently performing whole genome sequencing on over 70,000 individuals should go a long way towards addressing some of these challenges. However, given the cost of genome sequencing and the complexity of analyzing such large datasets, we can expect that it will be a number of years before genomic data will be available scale similar to the current work. Future efforts will also be required to streamline the identification of functional genetic variants in non-coding regions. Potential approaches such as massively parallel reporter assays or CRISPR tiling of intergenic regions to identify transcriptional regulators both show early promise, but further refinement of these methods will be required.^6–8