CAD’s Genetic Basis Strongly Overlaps With Other Vascular Conditions and Risk Factors

Coronary artery disease shares many of the same genetic origins as peripheral artery disease, hypertension, hypercholesterolemia, and type 2 diabetes, new research confirms. The analysis, which applied a polygenic risk score to UK Biobank participants, also showed that other seemingly related diseases are in fact secondary to CAD.

“The main message is the extent of genetic overlap we saw through our study among cardiovascular diseases,” said Panos Deloukas, PhD (Queen Mary University of London, England), joint senior author of the study along with Heribert Schunkert, MD (Deutsches Herzzentrum München, Germany).

But in the absence of CAD, the risk score wasn’t predictive for patients with heart failure or atrial fibrillation, indicating no genetic overlap. Here, “the polygenic risk score for CAD that we tested shows that you can consider stratifying patients into basically two groups: those that have CAD as a comorbidity versus those that have heart failure without CAD,” Deloukas explained to TCTMD.

Guillaume Paré, MD (McMaster University, Hamilton, Canada), commenting on the findings for TCTMD, said the study “certainly has an original slant, in that to the best of my knowledge this is the first one that systematically tested polygenic scores with both risk factors and secondary outcomes as well as unrelated diseases.”

What’s most interesting, he said, isn’t actually new but is a message worth repeating: “These scores are quite predictive to a level that we haven’t really seen in the genetics world before just a few years ago, and that’s extremely encouraging.” These latest results, published in the Journal of the American College of Cardiology, further support the validity of polygenic risk scores, according to Paré. “The reason why I’m so excited is we finally have a tool that is . . . on par with some of the classical risk factors.”

The question is how these scores might now apply in a clinical setting, Deloukas and Paré agreed.

UK Biobank Database

Led by Ioanna Ntalla, PhD (Queen Mary University of London), the researchers developed their polygenic risk score by examining 425,196 people whose genotypes are contained in the UK Biobank, focusing on 300 CAD-associated variants that were linked to 22 traits. These traits included risk factors, diseases secondary to CAD, and comorbid conditions.

Deloukas emphasized that they took a conservative approach in creating their score, basing it on a relatively small number of markers that prior studies have found to show true associations with the disease. Some scores, he said, try to create a “signature” that is more inclusive and based on up to 2 million variants that have some level of association with the phenotype.

The risk of CAD, as predicted by the score employed here, overlapped with the risks of hypercholesterolemia, type 2 diabetes, and hypertension, indicating “that the score contained variants predisposing to these conditions,” Ntalla et al explain. The risk score also predicted CAD in patients who were free of traditional risk factors.

Additionally, links were found between the polygenic risk score and peripheral artery disease, abdominal aortic aneurysm, and stroke; these remained significant in sensitivity analyses, also suggesting shared genetic roots. For heart failure, A-fib, and premature death, however, the associations with the score disappeared in sensitivity analyses. Interestingly, there was a slight but significant inverse association between the polygenic risk score and migraine, which was “very robust,” Deloukas pointed out, adding that the relevant genes have to do with cell wall biology and blood pressure. As noted in the paper, the observed “overlap between CAD and migraine was based on a range of genetic loci with opposing effects to the two diseases.”

Between individuals whose risk scores were in the lowest versus highest quantiles, there were clear gradations in risk.

Odds Ratio Based on CAD Polygenic Risk Score

Most of the above findings make sense, Paré commented, but the migraine relationship is surprising and intriguing. “If you look at the genetics of longevity, it appears that variances that increase the risk of cardiovascular disease [on the whole] decrease longevity. Often we think of genetics as this fine equilibrium between things that might be deleterious to us and things that might be beneficial, and it’s tempting to think that we might develop these [traits] that predispose us to disease because it might be beneficial in another way. When it comes to cardiovascular disease, there just doesn’t seem to be any benefit.”

Thinking creatively, Paré suggested it’s possible that humans “might have sacrificed the risk of late-onset cardiovascular disease to protect us against migraine. [The trade-off] might have, for hunter-gatherers, been more useful.” He emphasized that this idea is highly speculative.

I think we are likely to see this technology enter preventive medicine in a clinical sense. Guillaume Paré

The current study mainly takes an epidemiology perspective, but “polygenic risk scores also might have a place in a clinical setting for prediction,” he said, pointing out that the American Heart Association listed these scores’ potential as one of the top 10 research advances of 2018. Based on his own research, Paré said that scores are most clinically useful at the extreme ends of the spectrum. Nor is this approach restricted to the cardiovascular realm, he added. “Parallel work has been done in . . . breast and prostate cancer, and they came to similar conclusions. So I think we are likely to see this technology enter preventive medicine in a clinical sense.”

For Deloukas, while “there is now in our field a very strong push and interest in demonstrating clinical utility,” it’s important to be pragmatic and understand that current polygenic risk scores reflect only the population from which they were derived. “Their transferability, if you like, to other ethnic groups is a bit of a gradient and you lose a considerable amount of power. And therefore if you wanted to apply it to a population like the UK population you will face this problem, because of course the composition . . . is not just, for example, of European descent.”

This has not yet been done because this vast level of genetic data isn’t yet available from other areas of the world, Deloukas said. This information would allow the creation of more comprehensive scores.

Even when used in a closely matched population, it remains to be seen whether altering care based on the results of polygenic risk scores would bear fruit years down the line, he pointed out. “Their value is you capture people at risk at the very early stage and you can propose measures to affect lifestyle and therefore reduce the risk of the disease. In some instances you can also intervene with medication. A lot of that is now being tested.”