"New and Improved” Genetic Risk Score Predicts Coronary Artery Disease

A collaborative team of researchers from the Center for Genomic Medicine (CGM) has created a new and “significantly improved” score for predicting coronary artery disease. The score, called GPS_Mult, is published in Nature Medicine.

Coronary artery disease – a multifactorial condition

CAD is the most common heart disease in the United States and a major cause of death worldwide. Caused by a buildup of cholesterol deposits in the coronary arteries, it can lead to reduced blood flow to the heart muscle and in some cases a heart attack.

Identifying CAD risk before its onset is important for preventing heart disease, however, this can be challenging. A multifactorial disease, the development of CAD is influenced both by environmental factors – such as diet and physical activity – and genetics; its heritability is estimated to sit between 40–60%.

The researchers, including joint lead authors Dr. Amit V. Khera, a physician-scientist and associate director of the precision medicine unit at CGM, and Dr. Pradeep Natarajan, director of preventive cardiology and the Paul and Phyllis Fireman Endowed Chair in vascular medicine at Massachusetts General Hospital, sought to address these issues.

They developed a new polygenic risk score for CAD, utilizing participant data from several global resources, including the UK Biobank, the Million Veteran Program and Genes & Health.

Meet GPS_Mult

The research team was motivated by evidence suggesting polygenic scores, which integrate GWAS data from individuals that have diverse ancestries, can trump the accuracy of disease risk predictions developed using GWAS data from a single ancestry source.  “We used information from ancestrally diverse 269,000 CAD cases, over 1,178,000 controls and data from related traits in over 2 million individuals along with methods leveraging commonalities in mechanistic pathways to develop a new polygenic risk score for CAD,” they explain. 

The polygenic scores were created using GWAS data from individuals with CAD and control participants from mixed ancestries. “These scores were trained within the UK Biobank cohort in 116,649 individuals of European ancestry and then validated in the remaining independent study population of 325,991 individuals (54.3% female, 7,281 African, 1,464 East Asian, 308,264 European and 8,982 South Asian ancestry),” the researchers say.

Khera and colleagues named their score GPS_Mult, which they say “demonstrated significantly improved performance” for predicting CAD risk across multiple ancestries compared to risk scores published previously.

The research team says that GPS_Mult could – in some cases – identify healthy individuals that carry a risk of CAD to a similar degree of accuracy as identifying CAD in people with pre-existing disease. “When added to risk scores used in current clinical practice, GPS_Mult significantly improved discrimination and reclassification relevant to clinically important decision thresholds, such as the decision to initiate statin therapy,” Khera and colleagues describe.

In the paper, Khera and colleagues call for further research studies to explore the utility of GPS_Mult, as the work carries some limitations. For example, individuals in research studies can often be healthier than the general public, which can impact the validity of disease risk models created using these cohorts. They also emphasize that more GWAS studies in diverse populations are required in order to predict disease risk in wider ancestry groups.

However, they conclude that “incorporating GWAS data for CAD and related traits from multiple ancestries on a large scale leads to significantly improved performance of GPS_Mult in external validation among diverse ancestry populations when compared with previously published scores.” 

Reference: Patel AP, Wang M, Ruan Y, et al. A multi-ancestry polygenic risk score improves risk prediction for coronary artery disease. Nat Med. 2023. doi: 10.1038/s41591-023-02429-x