Polygenic Risk Scores: A Cardiovascular Conundrum

Cardiovascular disease dominates the global health landscape, causing nearly one-third of all deaths worldwide and costing over 350 million years of life.¹ In the United States alone, the American Heart Association estimates that almost 130 million people are living with cardiovascular disease—and that these diseases cost well over $400 billion each year.² With mortality from these conditions on the rise,³ screening tools are needed that accurately identify people at elevated risk of cardiovascular disease so that care providers can engage in early, effective prevention and treatment.

Polygenic risk scores for cardiovascular disease

Cardiomyopathy

Although more than one in every 500 people may be affected by a cardiomyopathy, these diseases often go undiagnosed.⁴ Recently, researchers have investigated the value of polygenic risk scores (PRS) for the two most common types: dilated cardiomyopathy and hypertrophic cardiomyopathy.

Using 13 dilated cardiomyopathy risk alleles, a group of researchers developed a polygenic risk score⁵—the DCM-PRS—to determine which members of a healthy population were more predisposed to developing the disease. They found significant correlations between the DCM-PRS and decreased left ventricular ejection fraction (a measure of left-sided heart dysfunction), increased left ventricle dimensions, and development of dilated cardiomyopathy, but no correlation with biomarkers of heart failure.

Another group sought to evaluate the performance of a new PRS in risk stratification of patients with hypertrophic cardiomyopathy.⁶ The PRS_IVS, which was based on genes with a potential effect on intraventricular septum diameter, was significantly elevated in patients with both monogenic and causal variant-negative hypertrophic cardiomyopathy—but showed no association with intraventricular septum diameter in any of the groups, meaning that although the PRS_IVS may have diagnostic value, it shows no evidence of utility for risk stratification.

Coronary artery disease

Coronary artery disease (CAD) is the largest single cause of death and disability globally; it is responsible for nearly one in every six deaths overall¹ and one in three deaths of individuals over 35.⁷ As a result, these conditions are not only in the public eye, but also strong candidates for genetic screening options.

One group sought to determine whether a CAD PRS could more accurately identify candidates for statin treatment than conventional clinical approaches.⁸ To do so, they applied an existing 2.3 million-variant CAD PRS to 294,371 participants clinically classified as low-risk, reclassifying 14 percent of patients as “borderline risk” (≥5 percent 10-year risk of disease) and 5 percent as “intermediate risk” (≥7.5 percent)—the equivalent of reclassifying approximately 14 million people if applied to the entire US population. Patients in the borderline risk category and above may warrant consideration of treatment with a moderate-intensity statin.

But could a CAD PRS determine a higher-risk patient’s likelihood of a major adverse cardiovascular event?⁹ To find out, researchers applied a PRS based on 180 literature-derived single nucleotide polymorphisms (SNPs) to patients in the Sanford Heart Screening Program with an existing coronary artery calcium (CAC) score. Patients were classified as low (<15^th percentile), moderate (15–85^th percentile), or high (>85^th percentile) PRS. Of the study’s 1,380 patients, 6.6 percent experienced a major adverse event, with high PRS patients more likely to experience such an event than those with moderate or low scores. However, directly comparing CAD PRS to CAC scores in the same patient group indicated that the PRS was not superior to CAC in predicting major adverse cardiovascular events.¹⁰

As clinical science moves increasingly toward data-driven and computer-based approaches to prediction, prevention, and diagnosis, PRS may have a growing role to play in noninvasive patient assessment. That’s why one research group used a machine learning algorithm to combine 228 SNPs with patients’ clinical and demographic characteristics.¹¹ The goal: to determine whether the addition of SNPs enhanced the algorithm’s predictive accuracy. After identifying the 15 most relevant SNPs (eight predicting CAD risk and seven predicting disease severity), the model incorporating the CAD PRS demonstrated superior performance in both risk assessment and severity prediction—suggesting a promising future for both PRS and artificial intelligence (AI) in cardiovascular disease management.

Stroke

For those at risk of stroke or embolism, a normal day can quickly become an emergency situation. Understanding an individual’s risk of these issues can not only inform preventive measures, but also increase the likelihood of a rapid response, accurate diagnosis, and appropriate treatment when needed.

To determine how high genetic stroke risk and clinical characteristics are related, a research group applied a PRS¹² based on the 1.2 million-SNP GIGASTROKE dataset¹³ to 454 healthy adults with a mean age of 31. Despite their lack of overt cardiovascular disease, patients with higher PRS had higher systolic and diastolic blood pressure at rest, higher systolic blood pressure in response to exercise, higher mean arterial pressure, and lower hematocrit. These hemodynamic traits may indicate an elevated risk of stroke before other manifestations of cardiovascular disease.

Atrial fibrillation is commonly associated¹⁴ with stroke and systemic embolism, but can a PRS for atrial fibrillation also predict patients’ risk of these events? Researchers probed the question¹⁵ using samples from 41,337 individuals not known to have atrial fibrillation at baseline, 334 of whom went on to develop atrial fibrillation-related stroke or embolism. By adding PRS to the patients’ CHARGE-AF scores¹⁶ (which are based on demographic, behavioral, and clinical characteristics), the researchers observed a significant improvement in stroke or systemic embolism prediction in patients at high genetic risk of atrial fibrillation.

How effective are polygenic risk scores?

Although recent research into cardiovascular disease PRS is promising, data from University College London (UCL) have cast some doubt on these scores’ value in screening and risk stratification.¹⁷ The researchers examined 926 PRS in 310 diseases, analyzing a total of almost 4,000 performance metrics. Their findings? In CAD, a PRS in the 97.5^th percentile was associated with a one in eight likelihood of developing CAD within the next 10 years—little more than double the background odds of one in 19. A PRS in the 75^th percentile was associated with a one in 15 chance; the 25^th and 2.5^th percentiles had odds of one in 29 and one in 54, respectively. Professor Aroon Hingorani, UCL Chair of Genetic Epidemiology and the study’s lead author, said, “Strong claims have been made about the potential of polygenic risk scores in medicine, but our study shows that this is not justified. We found that, when held to the same standards as employed for other tests in medicine, polygenic risk scores performed poorly for prediction and screening across a range of common diseases.”¹⁸

Do these results signal the end for PRS despite other indicators of success? The matter remains under debate—but, as technology advances, it’s possible that a combination of PRS and other tools, such as AI-based analysis, could yield greater accuracy and predictive value in the future.

References:

1. 1. Institute for Health Metrics and Evaluation. GBD Compare. October 15, 2020. https://vizhub.healthdata.org/gbd-compare.

1. 1. Tsao CW et al. Heart disease and stroke statistics–2023 update: a report from the American Heart Association. Circulation. 2023;147(8):e93–e621. doi:10.1161/CIR.0000000000001123.

1. 1. Liu M et al. Cardiovascular health of middle-aged U.S. adults by income level, 1999 to March 2020: a serial cross-sectional study. Ann Intern Med. 2023; online ahead of print. doi:10.7326/M23-2109.

1. 1. Semsarian C et al. New perspectives on the prevalence of hypertrophic cardiomyopathy. J Am Coll Cardiol. 2015;65(12):1249–1254. doi:10.1016/j.jacc.2015.01.019.

1. 1. Paldino A et al. Utility of a polygenic risk score for dilated cardiomyopathy in the general population. Circulation. 2023;148:A15195. doi:10.1161/circ.148.suppl_1.15195.

1. 1. Nomura A et al. Usefulness of polygenic risk score for intraventricular septal diameter in patients with hypertrophic cardiomyopathy. 2023;148:A17207. doi:10.1161/circ.148.suppl_1.17207.

1. 1. Ralapanawa U, Sivakanesan R. Epidemiology and the magnitude of coronary artery disease and acute coronary syndrome: a narrative review. J Epidemiol Glob Health. 2021;11(2):169–177. doi:10.2991/jegh.k.201217.001.

1. 1. Bertot JH et al. Predictive utility of a CAD polygenic risk score in a low-risk primary prevention cohort. Circulation. 2023;148:A13054. doi:10.1161/circ.148.suppl_1.13054.

1. 1. Fohle E et al. High polygenic risk score is a predictor for major adverse cardiac event. Circulation. 2023;148:A13144. doi:10.1161/circ.148.suppl_1.13144.

1. 1. Maryniak A et al. CAD polygenic risk score, a novel screening tool for predicting likelihood of coronary artery disease, was not superior to coronary calcium score in predicting MACE. Circulation. 2023;148:A12423. doi:10.1161/circ.148.suppl_1.12423.

1. 1. Chatzopoulou F et al. Combining genomic profiling and clinical data through machine learning modeling for the prediction of coronary artery disease severity: insights from the Genetic SYNTAX Score (GESS) trial. Circulation. 2023;148:A17570. doi:10.1161/circ.148.suppl_1.17570.

1. 1. Barber JL et al. Phenotypic manifestations of polygenic risk for ischemic stroke in adults free from disease. Circulation. 2023;148:A18928. doi:10.1161/circ.148.suppl_1.18928.

1. 1. Mishra A et al. Stroke genetics informs drug discovery and risk prediction across ancestries. Nature. 2022;611(7934):115–123. doi:10.1038/s41586-022-05165-3.

1. 1. Essa H et al. Atrial fibrillation and stroke. Card Electrophysiol Clin. 2021;13(1):243–255. doi:10.1016/j.ccep.2020.11.003.

1. 1. Park H et al. Polygenic Risk Scores Predict Future Atrial Fibrillation-Related Stroke/Systemic Embolism. Circulation. 2023;148:A16390. doi:10.1161/circ.148.suppl_1.16390.

1. 1. Alonso A et al. Simple risk model predicts incidence of atrial fibrillation in a racially and geographically diverse population: the CHARGE-AF consortium. J Am Heart Assoc. 2013;2(2):e000102. doi:10.1161/JAHA.112.000102.

1. 1. Hingorani AD et al. Performance of polygenic risk scores in screening, prediction, and risk stratification: secondary analysis of data in the Polygenic Score Catalog. BMJ Med. 2023;2(1):e000554. doi:10.1136/bmjmed-2023-000554.

1. 1. University College London. Genetic risk scores not useful in predicting disease. October 16, 2023. https://www.ucl.ac.uk/news/2023/oct/genetic-risk-scores-not-useful-predicting-disease.