The following are key points to remember from this American Heart Association (AHA) scientific statement on polygenic risk scores (PRSs) for cardiovascular disease (CVD):

1. Despite advances, risk prediction to inform risk mitigation of CVDs by synthesizing multiple clinical risk factors into a single risk estimate for coronary artery disease (CAD) and stroke remains imprecise (e.g., Pooled Cohort Equations [PCEs]) with persistently high rates of incident CVD. And the PCE systematically underperforms in some groups (e.g., higher socioeconomic and well-educated adults) and particularly in younger and older adults. Genetic characterization has been proposed as an approach to enable earlier and potentially tailored prevention.
2. This AHA scientific statement reviews the value of PRSs for CVD including five cardiometabolic diseases: CAD, hypercholesterolemia, type 2 diabetes (T2DM), atrial fibrillation (AF), and venous thromboembolic disease. It reviews the current science, clinical implementation considerations (e.g., efficacy and cost) across stakeholders (health care professionals, patients, and health care administrators), and outstanding questions about the clinical use of PRS for selected CVDs.
3. Incorporation of genetics into risk prediction frameworks offers the opportunity to refine risks and creation of earlier and tailored risk reduction strategies. A simple supporting example is that a family history of premature CVD is associated with a 50% higher odds of CVD in middle-aged adults independent of clinical risk factors. That genetics may be additive in risk prediction is supported by twin studies (comparing monozygotic twins with dizygotic twins), which have shown that variation in the development of CAD, AF, and diabetes are attributable to common genetic variations.
4. Monogenic pathogenic risk variants (minor allele frequency <1%) that explain a disease are rare. For example, a high risk of atherosclerotic CVD (ASCVD) is associated with an autosomal dominant mutation in the low-density lipoprotein receptor (LDLR). Apolipoprotein B (APOB) and proprotein convertase subtilisin/kexin type 9 (PCSK9) genes with life-long high and very high LDL cholesterol (LDL-C) known as familial hypercholesterolemia. The mutation resulting in heterozygous familial hypercholesterolemia (HeFH) occurs in about 1/250 persons and is associated with a small contribution to overall risk of cardiometabolic disease but very high risk for early-onset myocardial infarction. Other examples of monogenic risk variants include variants in GCK for diabetes, KCNQ1 for AF, and F5 for venous thromboembolic disease.
5. In contrast, very large genome-wide association studies (GWAS) have confirmed that single nucleotide variants (SNVs; a variation in a nucleotide base pair compared to what is expected, i.e., an A instead of G) within the genome each of which results in a small risk, but collectively may account for substantial CVD risk known as “polygenic risk score (PRS).” Common polygenic risk variants are denoted as minor allele frequency ≥1%. PRSs are the weighted sum of risk conferred by multiple disease associated SNVs across the genome. PRSs are proposed as tools to improve the prediction of common complex CVDs.
6. GWAS PRSs for common diseases can identify persons with risk equivalent to monogenic mutations, and for multiple CVDs, the PRSs are independently associated with respective CVD. Constructing a PRS requires a list of SNVs with their accompanying effect sizes (a quantification of the association of the SNV with the disease) from an external data set, typically acquired from a GWAS. Historically SNVs were excluded, which may have had a negative influence but are now all included with reweighted effect sizes based on at least p and r² value.
7. Large biorepositories such as the UK Biobank, Million Veteran Program, and Electronic Medical Records and Genomics Network have aided in the performance of high-throughput and high-sample-size genome-wide discovery efforts of a wide range of cardiometabolic diseases. In addition, biobanks located in areas with higher levels of ancestry diversity have contributed a critical view of genetic association in understudied populations.
8. The risk of developing a disease is influenced by both monogenic and polygenic risk variants. In persons with HeFH monogenic risk variants, LDL-C concentrations varied, as did their risk of developing CAD. Those with a low LDL-C PRS and monogenic FH had, on average, lower LDL-C and a lower risk of CAD compared with those with a high LDL-C PRS and monogenic FH. A similar pattern is seen in persons with a monogenic risk variant for hypertrophic cardiomyopathy (e.g., MYH7) and high risk PRS for hypertrophic cardiomyopathy. These observations support the “liability threshold model,” that is, the notion that multiple factors—monogenic, polygenic, and nongenetic—may each contribute to a threshold necessary for disease development.
9. Three broad criteria should be considered before implementation of PRSs for clinical cardiovascular care: (1) efficacy, (2) harm, and (3) logistics. The clinical efficacy of PRS is likely appropriate when either of the following is achieved: (1) the integration of PRS into clinical risk tools substantially improves their accuracy, or (2) PRS risk tools can identify participants at a risk at least equivalent to that of individuals with monogenic risk variants. Regarding the first point, for most of the disease examples in this scientific statement (AF, CAD, and T2DM), the predictive accuracy of established clinical risk factor models is improved with the addition of PRS (i.e., PRS improves prediction when incorporated into the PCE for ASCVD, the CHARGE-AF risk model [AF], and the American Diabetes Association risk model [T2DM]). Among potential harms include the potential of exacerbating existing racial disparities in health care because of the limited inclusion of those with non-European ancestry. And when PRSs infer a higher-risk cohort, such as for venous thromboembolism (VTE) and AF, prospective randomized controlled studies addressing safety and efficacy are necessary. There are several logistical and educational considerations including inability to utilize research study data for clinical use and the limitations of direct-to-consumer testing products. Both physicians and patients have very limited knowledge of population genetics and PRS, which potentially hinders putative benefits and may exacerbate harms. Finally, current PRS studies are presented in percentiles according to the cohort studied, which limits the ability to generate universal raw scores. The ultimate goal for application of PRSs should be their representation in absolute risk, not percentiles.
10. Other areas covered included PRSs for pharmacogenomics, and consideration for payers and commercial genetics organizations. The most interesting is the value of PRS in predicting drug response and toxicity. The incremental absolute risk reduction across the spectrum of PRS for statins and PCSK9 inhibitors was as expected in primary and secondary prevention. Other potential utility includes response to hypoglycemic agents in T2DM, antiplatelet agents, and use of beta-blockers and angiotensin-converting enzyme inhibitors in chronic heart failure, each of which has been reported but require further placebo-controlled studies with continual validation of PRSs.

Summary of pertinent clinical points and considerations requiring further evidence via clinical trials:

1. CAD heritability, or the proportion of phenotype explained by the additive sum of genetic factors, is estimated to be 40-60%. Persons with monogenic FH variants have a greater relative and absolute clinical benefit from statins for the prevention of incident CAD. Among middle-aged adults, a CAD PRS performs similarly to conventional risk factors and provides additional prognostic information, but the clinical significance of this improvement is contentious. PRS provides early identification for the value of lifestyle therapies and statins and potentially improved compliance. Also, earlier screening for subclinical atherosclerosis to time initiation of pharmacotherapies, and use as a risk-enhancing factor for primary prevention in middle-aged persons at borderline-intermediate 10-year ASCVD risk.
2. PRSs for AF have consistently shown incremental predictive capabilities in addition to clinical risk factors. Proposed utility has been to refine the identification of individuals meriting close surveillance for AF, including earlier detection and resultant prophylactic anticoagulation, potentially with monitoring devices, as well as rigorous control of additive clinical risk factors for AF.
3. High PRS for T2DM would support earlier lifestyle modification, potential consideration for prophylactic hypoglycemic medications with concomitant additional aggressive treatment of T2DM, and other cardiometabolic risk factors. Genomic stratification may help optimize hypoglycemic choice.
4. VTE has inherited monogenic thrombophilia (e.g., factor V Leiden and prothrombin G20210A) that increase the relative risk of VTE by approximately 3- to 5-fold. The role of genetic testing for informing therapy is limited because of uncertainties about the effects of inherited thrombophilias on recurrent VTE risk, the lack of data demonstrating that thrombophilia testing improves outcomes, and the risks of prolonged anticoagulation. Women in the 95th percentile of polygenic risk had a 2.5-fold risk of incident VTE (hazard ratio [HR], 2.5; 95% confidence interval [CI], 2.0–3.2); comparable in effect size to the risk conferred by factor V Leiden (HR, 2.4; 95% CI, 1.9–3.4) and prothrombin G20210A (HR, 3.3; 95% CI, 1.1–10.2). This suggests the very high VTE PRS could be candidates for prophylactic anticoagulant in high-risk scenarios (prolonged travel, major surgery, hospitalization, peripartum, and post-partum).
5. PCE can be utilized for personalized drug therapy regimens that increase drug efficacy and decrease toxicities (e.g., personalized beta-blocker target dose in patients with heart failure with reduced ejection fraction, selection of antiplatelet agents, or the prevention of drug-induced QT prolongation).

Clinical Topics: Anticoagulation Management, Arrhythmias and Clinical EP, Diabetes and Cardiometabolic Disease, Dyslipidemia, Heart Failure and Cardiomyopathies, Prevention, Pulmonary Hypertension and Venous Thromboembolism, Atherosclerotic Disease (CAD/PAD), Anticoagulation Management and Atrial Fibrillation, Anticoagulation Management and Venothromboembolism, Atrial Fibrillation/Supraventricular Arrhythmias, Homozygous Familial Hypercholesterolemia, Lipid Metabolism, Nonstatins, Novel Agents, Primary Hyperlipidemia, Acute Heart Failure

Keywords: Anticoagulants, Apolipoproteins, Atherosclerosis, Atrial Fibrillation, Cardiometabolic Risk Factors, Cardiomyopathy, Hypertrophic, Cholesterol, LDL, Coronary Artery Disease, Diabetes Mellitus, Type 2, Drug Therapy, Genetic Testing, Genetic Variation, Genome-Wide Association Study, Heart Failure, Hypercholesterolemia, Hyperlipoproteinemia Type II, Metabolic Syndrome, Multifactorial Inheritance, Platelet Aggregation Inhibitors, Proprotein Convertase 9, Primary Prevention, Risk Factors, Secondary Prevention, Stroke, Thrombophilia, Vascular Diseases, Venous Thromboembolism

< Back to Listings