A polygenic, multi-condition risk report validated in US health systems could help clinicians identify inherited cardiovascular risk earlier, improve prevention strategies, and guide more personalized care.
Study: Development and Validation of a Clinical Polygenic Risk Report in US-Based Health Systems for 8 Cardiovascular Diseases. Image source: ArtemisDiana / Shutterstock
In a recent study published in Journal of the American College of Cardiology (JACC)investigators described the development and validation of integrated polygenic risk scores (PRS) for eight cardiovascular diseases using data from 245,394 All of Us participants (AOU) Research Program and 53,306 participants from the Mass General Brigham Biobank (MGBB).
The complete PRS The platform demonstrated robust risk stratification and clinically structured reporting that generally matched or exceeded individual input models, offering a transparent framework for identifying high genetic risk individuals that may be missed by traditional clinical markers.
Background of cardiovascular polygenic risk score
cardiovascular diseases (CVD) remain a major cause of global mortality, but their incidence is characterized by a complex genetic architecture involving significant heritability and pleiotropy. While some heart diseases are caused by rare, high-impact mutations in a single gene (monogenic), decades of research have shown that the vast majority of cases arise from thousands of common genetic variations in the genome, each with a subtle individual effect.
Traditional clinical risk models, exemplified by Aggregate Cohort Equations (PCE), estimate risk using demographic and phenotypic markers such as blood pressure and cholesterol, whereas PRS quantification of inherited risk from common genetic variants. However, a holistic methodology for risk stratification is still lacking. Systematic reviews and meta-analyses of available PRS Approaches show that they often fail to capture the full range of inherited risk, particularly in younger or “intermediate risk” populations.
Consequently, there is a pressing need for a standardized, “consensus” approach that could aggregate these scores into a single, reliable reference across multiple conditions.
Design and validation of a comprehensive PRS study
The present study aimed to address these knowledge gaps by creating a transparent line to bring genetic risk stratification into routine preventive care. The whole project included a multi-phase development and validation study in three large-scale biobanks:
The training dataset was derived from a genomic and electronic health record (EHR) data from 245,394 All of us (AOU) participants (mean age = 51.7 ± 17.0 years). Seven feature models were trained using this dataset, while the elevated lipoprotein(a) model was trained on United Kingdom Biobank because standardized measurements of Lp(a) were not available. AOU. The training methodology focused on eight clinical conditions, namely atrial fibrillation (AF), coronary artery disease (VILLAIN), type 2 diabetes mellitus (T2DM), thoracic aortic aneurysm (Yay), extreme hypertension, venous thromboembolism (VTE), severe hypercholesterolemia and elevated lipoprotein (a).
The PRSmix software package was used for the publicly available integration PRS from the PGS Directory. A stratified 80/20 split was applied to AOU cohort for internal model testing prior to external validation, with an equil United Kingdom Biobank-based approach used for Lp(a).
Next, external validation of model performance was performed in an independent cohort of 53,306 participants from the Mass General Brigham Biobank (MGBB). The study adjusted for age, sex, and genetic background using imputed primary data (PC; derived from a pool of 1000 genomes PC interval) to account for genetic diversity.
Specifically, discrimination was assessed using C-statistics, and model calibration was assessed across subgroups of age, sex, and ethnicity.
PRS risk stratification by cardiovascular characteristics
The novel embedded PRS The platform demonstrated consistent risk stratification, which generally matched or exceeded the performance of the individual input scores across the eight attributes. However, predictive performance varied by condition, with more modest discrimination for some outcomes, including VTE, Yayand extreme hypertension.
The most striking results of the study were those in elevated levels of Lipoprotein (a), where individuals in the high genetic risk category (the top 10%) had a significant increase of 41.0-fold (95% CI: 27.0-62.2) to have elevated levels compared to those with average genetic risk (P < 0.0001).
Although not as dramatic, people at high risk (top 10%) for severe hypercholesterolemia (odds ratio [OR] = 4.1), VILLAIN (OR = 3.73), T2DM (OR = 3.1), AF (OR = 3.0) and extreme hypertension (OR = 2.1) demonstrated multiple risks of the corresponding average risk. The study also showed that increased genetic risk was common in this biobank-based analysis, with 71.2% of MGBB population that has at least one PRS-defined threshold corresponding to at least a 3-fold increased relative genetic risk for one or more of the eight traits.
Most importantly, the study found that the addn PRS to existing clinical tools such as pooled cohort equations (PCE), significantly improved “net reclassification”. In VILLAINincorporation of the genetic score improved risk classification by 17% (P < 0.0001) among patients previously considered "marginal" or "intermediate" risk. Prospective follow-up (mean 7.6 years) confirmed that a high PRS was associated with an incident VILLAIN, AF, T2DM, VTEand Yayeven in participants under 50 years of age.
Clinical implications of the multi-condition PRS test
The present study marks an important step towards a clinically ordered multistate cardiovascular system PRS essay. By validating a complete PRS in eight states, the new study approach has provided a scalable framework that identifies individuals who may harbor previously unrecognized inherited genetic risks despite having normal traditional biomarkers.
However, the authors emphasize that currently limitations remain. While scores were performed across ancestry groups, predictive power remained strongest in European populations, highlighting the need for more diverse research data. The authors also noted that broader prospective validation and further evidence of clinical utility are needed beforehand PRS-Guided care pathways can be fully established.
Moving forward, this report is now available as a clinically ordered test, allowing doctors to use genetic ‘risk enhancers’ to inform preventive discussions, targeted screening, lifestyle advice and medication decisions where clinically appropriate for their patients.
Polygenic risk scores for 8 cardiovascular traits in both @MassGenBrigham and @AllofUsResearch – exaggerated – strongly suggest danger. It is still not applied in clinical practice pic.twitter.com/PkFkQ4dLvf
– Eric Topol (@EricTopol) April 29, 2026
