Development and Validation of a Clinical Polygenic Risk Report in U.S.-Based Health Systems for 8 Cardiovascular Conditions

Abstract

Polygenic risk scores (PRS) offer a powerful method to stratify inherited cardiovascular risk, yet their path to routine clinical use has been unclear. This study aimed to develop and validate integrated PRS for eight cardiovascular conditions and establish a practical framework for clinical reporting. Researchers analyzed genetic and clinical data from 245,394 participants in the All of Us Research Program. Publicly available PRS for coronary artery disease, atrial fibrillation, type 2 diabetes, venous thromboembolism (VTE), thoracic aortic aneurysm (TAA), extreme hypertension, severe hypercholesterolemia, and elevated lipoprotein(a) were combined using an elastic-net approach called PRSmix. These integrated PRS were then rigorously validated in an external cohort of 53,306 participants from the Mass General Brigham Biobank. The results demonstrated robust discrimination and appropriate calibration across all eight traits. Individuals in the high genetic risk group (top 10% of PRS distribution, or top 20% for the rarer TAA and VTE) showed significantly increased odds compared to those at average risk. Incorporating these integrated PRS into clinical models enhanced risk classification, and prospective analyses confirmed strong associations with future cardiovascular events. This work provides an actionable framework for genetic risk reporting in clinical practice, with the test now available for clinical ordering. Further prospective validation studies are warranted to solidify its utility.

Introduction

Cardiovascular diseases (CVD) remain a leading cause of global mortality and morbidity. While traditional risk factors like hypertension, smoking, and high cholesterol are well-established, inherited genetic susceptibility plays a substantial, often underappreciated role. Polygenic risk scores (PRS) aggregate the effects of thousands of common genetic variants across the genome to estimate an individual’s genetic predisposition to a specific disease. For cardiovascular conditions, PRS have shown promise in identifying individuals at high risk long before traditional symptoms or risk factors manifest. However, translating this promise into routine clinical practice has faced hurdles, including the need for robust validation across diverse populations and the challenge of integrating multiple condition-specific scores into a cohesive, clinically interpretable report. This study directly addresses these gaps by developing and validating integrated PRS for eight major cardiovascular conditions within large, real-world U.S. health system populations.

Methods

The research team leveraged a massive dataset from the All of Us Research Program, encompassing genotype data and linked electronic health records from 245,394 participants. This diverse cohort provided the foundation for developing the integrated PRS. The study focused on eight key cardiovascular traits: coronary artery disease (CAD), atrial fibrillation (AFib), type 2 diabetes (T2D), venous thromboembolism (VTE), thoracic aortic aneurysm (TAA), extreme hypertension, severe hypercholesterolemia, and elevated lipoprotein(a). For each condition, the team utilized publicly available, well-established PRS. The critical innovation was the integration of these individual condition-specific scores into a unified framework using PRSmix, a sophisticated statistical method based on elastic-net regression. This approach optimally combines the predictive power of each individual PRS while managing potential correlations between them. To ensure the findings were generalizable and not specific to the development cohort, the integrated PRS underwent rigorous external validation. This involved 53,306 genotyped participants from the Mass General Brigham Biobank (55.6% women, mean age 53 ± 17 years). The validation process employed logistic regression models, carefully adjusting for essential covariates like age, sex, and genetic ancestry to isolate the effect of the genetic risk score. Performance was assessed through discrimination (how well the score separates cases from controls, often measured by area under the curve) and calibration (how accurately predicted risks match observed event rates). The team also evaluated the improvement in risk classification when adding the integrated PRS to standard clinical risk models and conducted prospective analyses linking the PRS to future incident cardiovascular outcomes.

Results

The validation in the Mass General Brigham Biobank cohort yielded highly significant results. The integrated PRS demonstrated robust performance across all eight cardiovascular traits, exhibiting strong discrimination and appropriate calibration, indicating their reliability for clinical application. Comparing individuals at high genetic risk (defined as the top 10% of the PRS distribution for most conditions, or the top 20% for the less common TAA and VTE) against those at average genetic risk (26th-75th percentiles for most, or 21st-80th percentiles for TAA and VTE) revealed substantial increases in disease odds. The Odds Ratios (ORs) quantify this increased risk: coronary artery disease (OR = 3.7, 95% CI: 3.4-4.1), type 2 diabetes (OR = 3.1, 95% CI: 2.8-3.3), atrial fibrillation (OR = 3.0, 95% CI: 2.7-3.3), venous thromboembolism (OR = 1.9, 95% CI: 1.6-2.0), thoracic aortic aneurysm (OR = 1.7, 95% CI: 1.5-1.9), extreme hypertension (OR = 2.1, 95% CI: 1.8-2.3), severe hypercholesterolemia (OR = 4.1, 95% CI: 3.7-4.5), and notably, elevated lipoprotein(a) (OR = 41.0, 95% CI: 27.0-62.2). The exceptionally high OR for lipoprotein(a) underscores the profound genetic influence on this specific lipid marker, which is a major independent risk factor for atherosclerosis and valvular heart disease. Adding the integrated PRS to existing clinical risk models significantly improved the ability to correctly classify individuals into higher or lower risk categories (net reclassification improvement). Importantly, prospective analyses confirmed that these integrated PRS were strongly associated with the actual development of new (incident) cardiovascular events over time, solidifying their predictive value beyond baseline risk assessment.

Discussion

This study marks a significant advance in the field of cardiovascular genetics and preventive cardiology. By successfully integrating PRS for eight major conditions and validating them in large, real-world populations representative of U.S. health systems, the researchers have overcome a key barrier to clinical implementation. The results demonstrate that these integrated scores provide substantial, complementary risk information beyond traditional clinical factors. Individuals identified in the high genetic risk group face markedly increased odds of developing these conditions – often 2 to 4 times higher, and even 41 times higher for elevated lipoprotein(a) – compared to those with average genetic risk. This level of risk stratification has profound implications for personalized prevention. For instance, identifying a young adult with a high PRS for coronary artery disease could justify earlier, more intensive lifestyle interventions and potentially earlier initiation of lipid-lowering therapies like statins, long before traditional risk scores like the Pooled Cohort Equations would flag them as high risk. Similarly, identifying high genetic risk for atrial fibrillation could prompt more vigilant monitoring for arrhythmias. The establishment of a clear, practical framework for reporting these scores is a crucial step. The report is now available as a clinically orderable test, potentially enabling broader access. However, challenges remain. Integrating genetic risk information effectively into clinical workflows requires physician education, clear guidelines on how to act on results, and careful consideration of ethical aspects like genetic counseling and avoiding unnecessary anxiety. The cost-effectiveness of widespread PRS testing also needs evaluation. Furthermore, while the study populations are large, ongoing efforts are needed to ensure the scores perform equally well across diverse racial and ethnic groups, as genetic risk architectures can vary. The critical next step highlighted by the authors is broader prospective validation studies. These studies will track large groups of individuals over many years to definitively prove that using integrated PRS in clinical practice leads to earlier interventions, prevents more cardiovascular events, and ultimately improves patient outcomes compared to standard care alone.

Conclusion

This research provides robust evidence for the clinical validity and utility of integrated polygenic risk scores for eight major cardiovascular conditions. The development and validation framework using data from the All of Us and Mass General Brigham Biobank cohorts demonstrates that these scores offer powerful discrimination and can significantly improve upon traditional risk classification methods. The successful creation of a clinically orderable report based on this framework represents a tangible step towards the integration of polygenic risk information into routine cardiovascular care in U.S. health systems. By identifying individuals at high inherited risk long before symptoms appear, these integrated PRS hold immense potential to shift the paradigm towards truly personalized, preventive cardiology, enabling earlier and more targeted interventions. The immediate next phase requires focused prospective studies to conclusively demonstrate that acting on this genetic risk information translates into measurable reductions in cardiovascular disease burden. These studies will be pivotal in establishing the long-term value and shaping the future implementation of polygenic risk assessment in clinical practice.