A coronary artery disease polygenic risk score estimates inherited susceptibility by combining effects from hundreds, thousands, or millions of common DNA variants. Unlike a pathogenic LDLR variant that can diagnose familial hypercholesterolemia, a PRS is probabilistic: it shifts risk upward or downward but does not establish that plaque is present, predict the date of a heart attack, or guarantee protection. Its usefulness depends on the model, reference population, ancestry, quality control, and how the score is integrated with age, blood pressure, cholesterol, diabetes, smoking, kidney disease, family history, lipoprotein(a), and coronary imaging. High polygenic risk can be present long before conventional risk factors become abnormal, which creates an opportunity for earlier prevention. Yet different commercial scores can place the same person in different percentiles, and evidence that testing improves outcomes is still developing. The most responsible interpretation converts the result into an absolute-risk discussion and an actionable prevention plan rather than treating a percentile as a diagnosis.
- A PRS measures relative inherited susceptibility, not current coronary blockage.
- Percentile, relative risk, and absolute risk are different quantities and should not be interchanged.
- Performance can decline when the tested person differs from the ancestry or setting used to build the score.
- A high score may strengthen prevention efforts but does not replace standard risk assessment.
- A low score cannot cancel smoking, high LDL cholesterol, diabetes, hypertension, or symptoms.
Table of Contents
- How a coronary PRS is built
- What the report numbers mean
- Who might consider testing
- Combining genetics with clinical risk
- Ancestry, model, and laboratory limitations
- How results may affect prevention
- What PRS cannot tell you
- Questions to ask before acting
How a coronary PRS is built
Coronary artery disease is highly polygenic. Most inherited susceptibility does not come from one rare, high-impact variant but from a large number of common variants, each contributing a very small effect. Genome-wide association studies compare people with and without disease to identify statistical associations across the genome. A PRS algorithm selects variants, assigns weights based on estimated effects, and sums the risk alleles carried by an individual.
Modern scores may include millions of variants because nearby markers can capture information through linkage disequilibrium even when they are not causal. Statistical methods account for correlation between variants, shrink uncertain effect estimates, and combine data from one or more studies. The final raw number has no universal meaning. It becomes interpretable only after comparison with a reference population and, ideally, calibration within a relevant clinical cohort.
A score is generally stable throughout life because germline DNA does not change. The interpretation, however, changes as age and clinical risk factors change. A 25-year-old and a 70-year-old with the same percentile do not have the same near-term absolute risk. The younger person has more time for genetically influenced risk to accumulate, while the older person’s observed health history and measured plaque become increasingly informative.
PRS is different from testing for monogenic familial hypercholesterolemia. A pathogenic variant in LDLR, APOB, or PCSK9 can have a large effect and supports cascade testing under Mendelian inheritance. A coronary PRS usually does not identify a single causal pathway or generate a simple 50% transmission probability. Children inherit a random half of each parent’s genome, so their score cannot be predicted precisely from a parent’s percentile.
The score may capture pathways related to lipids, blood pressure, inflammation, vessel-wall biology, thrombosis, and other processes. It can also contain indirect associations shaped by study design and population structure. A high PRS does not tell a clinician which pathway dominates in one person or which medicine will work best.
Genotyping arrays, imputation, or sequencing can supply the DNA data. The laboratory must ensure strand alignment, genome-build consistency, allele matching, and quality control before calculating the score. Small technical errors across a very large number of variants can materially alter a percentile, which is one reason raw consumer data should not automatically be uploaded into unvalidated calculators.
What the report numbers mean
Reports may present a percentile, standard-deviation score, odds ratio, hazard ratio, relative risk category, or estimated absolute risk. These are not interchangeable. The percentile answers how the score compares with a specified reference group. A result at the 95th percentile means the score is higher than about 95% of that reference population; it does not mean a 95% chance of developing coronary disease.
A relative-risk estimate compares groups. If the top category has twice the risk of an average category, an individual’s actual probability still depends strongly on baseline risk. Doubling a 2% ten-year risk produces a different clinical situation from doubling a 20% risk. Age, sex, country, competing mortality, and conventional risk factors determine the baseline against which genetic susceptibility operates.
Absolute risk estimates are more useful for decisions but more demanding to calculate. The model must be calibrated to disease rates and integrated with clinical variables. A report that converts a PRS directly into a lifetime percentage without explaining population, assumptions, and uncertainty can appear more precise than the evidence supports.
Some laboratories define “high” as the top 20%, 10%, 5%, or 1%. These cutoffs are not biologic boundaries. Risk changes continuously across the score distribution, and a result just below a threshold is not meaningfully different from one just above it. The chosen category may reflect a study, commercial reporting convention, or clinical workflow rather than a universal standard.
Confidence intervals and model version matter. Scores evolve as genome-wide studies grow, diverse datasets improve, and statistical methods change. A person’s DNA is stable, but recalculation with a new model can move the percentile. This does not mean the first sample was wrong; it means the model and reference distribution changed.
A report should name the score, variant set or method, target outcome, validation populations, ancestry approach, and whether clinical variables are included. It should distinguish coronary artery disease from myocardial infarction, stroke, or broad atherosclerotic cardiovascular disease. Related outcomes overlap but are not identical.
The result should be translated into plain language: “This score is higher than most people in the reference group and may increase lifetime coronary risk,” not “You will have a heart attack.” A low result should be framed similarly: it may reduce inherited susceptibility relative to the reference, but it does not show that arteries are clear or that acquired risks are harmless.
Who might consider testing
Potential use is greatest when the result could resolve a prevention decision rather than merely satisfy curiosity. A younger adult with a strong family history of premature coronary disease but normal current risk factors may gain information about inherited susceptibility that conventional ten-year calculators understate. A person at borderline or intermediate clinical risk may use a validated PRS as one element in a shared decision about earlier or more intensive prevention.
Testing may also be considered in health systems that have defined protocols for combining PRS with clinical risk tools, lipid management, and follow-up. Implementation is stronger when the score has been validated in the population served, clinicians understand its limitations, and patients receive a specific action plan.
A PRS is less likely to add value when treatment is already clearly indicated. Someone with established coronary artery disease, very high LDL cholesterol, diabetes with high cardiovascular risk, or a pathogenic familial-hypercholesterolemia variant usually needs intensive prevention regardless of polygenic percentile. Testing should not delay therapy.
It is also less useful for evaluating chest pain or acute symptoms. A PRS does not diagnose an obstructed artery, unstable plaque, or myocardial infarction. Symptoms require clinical assessment, electrocardiography, biomarkers, and imaging as appropriate. A low score must never be used to dismiss exertional chest pressure, shortness of breath, or other warning signs.
Before ordering, clarify the decision the result could change. Would high risk lead to a stronger recommendation for statin therapy, earlier LDL lowering, more frequent monitoring, or coronary calcium imaging? Would low risk truly change management, and would doing so be supported by guidelines? If no result changes the plan, the test has limited clinical utility.
People seeking direct-to-consumer testing should consider privacy, data reuse, laboratory quality, and whether the report provides medically interpretable calibration. A wellness score generated from a subset of array markers may differ from a clinically validated assay. Genetic counseling can be useful when the result causes anxiety or conflicts with family history.
Children are a special case. Coronary PRS can technically be calculated at birth, but evidence-based pediatric management pathways are limited. Familial hypercholesterolemia screening and treatment rely on measured lipids and monogenic testing when indicated. Predictive genomic testing in minors should have a clear health benefit during childhood and include discussion of autonomy and privacy.
Combining genetics with clinical risk
The PRS is one layer in a larger risk profile. Clinical tools estimate ten-year or longer-term risk using variables such as age, sex, cholesterol, blood pressure, diabetes, smoking, kidney function, and medication use. Family history, inflammatory disorders, pregnancy-related history, and social determinants may further refine the picture. Genetics should be integrated with these factors rather than placed above them.
Measured LDL cholesterol remains central because cumulative exposure is causal and treatable. A high PRS with low LDL can still indicate elevated susceptibility, but lowering LDL may reduce risk across genetic strata. A low PRS does not make severe LDL elevation benign. If familial hypercholesterolemia is suspected, a PCSK9 and familial cholesterol panel addresses a different question from a genome-wide score.
Lipoprotein(a) is another largely inherited risk factor that is measured directly in blood. Its concentration can be high even when standard cholesterol and PRS are reassuring. A lipoprotein(a) assessment and PRS should not be substituted for each other.
Coronary artery calcium, measured by noncontrast computed tomography, provides evidence of calcified plaque. PRS estimates susceptibility before disease appears; calcium reflects accumulated disease at the time of scanning. In selected adults whose preventive decision remains uncertain, the two may provide complementary information. A calcium score of zero can lower near-term risk in many middle-aged adults, but it does not guarantee absence of noncalcified plaque or erase high lifetime genetic risk.
Family history overlaps with genetics but is not redundant. It captures shared variants, rare variants, environment, access to care, and sometimes inaccurate reporting. PRS may identify high inherited risk without a family history, particularly in small families or when relatives are young. Conversely, a strong family history can remain important despite an average score because the PRS may omit relevant rare or population-specific variants.
Clinical interpretation should distinguish risk prediction from treatment-effect prediction. High polygenic risk often identifies a group with higher baseline event rates, so the absolute benefit of effective prevention may be larger. That does not mean the score proves a unique response to one drug. Statins, blood-pressure control, smoking cessation, activity, and diabetes prevention remain evidence-based across many genetic profiles.
An integrated discussion should produce an absolute-risk range, identify modifiable drivers, and state what action is recommended now. The score has failed clinically if it remains an isolated percentile in the chart with no connection to prevention.
Ancestry, model, and laboratory limitations
The most important equity limitation is portability across ancestry groups. Many discovery datasets have been dominated by people of European genetic ancestry. Differences in allele frequencies, linkage disequilibrium, environmental context, and baseline disease rates can reduce accuracy in other populations. A score may preserve some ranking ability yet be poorly calibrated, meaning the percentile or absolute risk is misleading.
Genetic ancestry is continuous and mixed, not identical to race or ethnicity. Laboratories use different methods to assign ancestry, select weights, or combine ancestry-specific models. A multi-ancestry score can improve performance, but it does not automatically eliminate disparities. The report should describe validation in populations relevant to the tested person and acknowledge when evidence is limited.
Model instability is another limitation. Multiple CAD scores can have similar population-level performance yet classify particular individuals differently. Variant selection, weighting method, imputation panel, phenotype definition, and training cohort all contribute. There is no single universal CAD PRS comparable across all laboratories.
Calibration can drift when a model moves from a research biobank to routine care. Disease prevalence, statin use, smoking patterns, age distribution, and healthcare access may differ. A score validated for lifetime coronary disease may not accurately estimate ten-year myocardial-infarction risk in another country.
Laboratory quality extends beyond genotyping accuracy. The calculation pipeline, reference population, missing-variant handling, software version, and report generation must be validated. Consumer raw data can have false-positive calls or missing markers, and third-party calculators may silently substitute proxies. A clinically consequential result should come from a transparent, quality-controlled process.
Selection bias can inflate apparent performance. Biobank participants may be healthier or less diverse than the general population, and case definitions may rely on electronic records. Prospective studies and randomized implementation trials are needed to show not only better prediction but improved behavior, treatment uptake, outcomes, and equity.
Privacy risks are durable because germline data are identifying and shared with relatives. Consent should cover data storage, research use, recontact, and possible access by third parties. Legal protections against genetic discrimination vary by country and may not cover life, disability, or long-term-care insurance.
These limitations do not make PRS useless. They define the conditions under which it should be used: validated models, transparent reporting, integration with conventional risk, and explicit uncertainty.
How results may affect prevention
A high PRS can support earlier attention to LDL cholesterol, blood pressure, smoking, glucose, sleep, activity, and diet. It may strengthen a decision to begin statin therapy in someone whose clinical risk is otherwise borderline, especially when current guidelines treat high polygenic risk as a risk-enhancing factor. The final decision should still consider age, absolute benefit, adverse effects, preferences, and alternative risk markers.
The result may motivate prevention, but communication matters. Fatalistic messages can reduce engagement, while deterministic labels can cause anxiety. Genetic risk is not destiny: coronary disease develops through an interaction between inherited susceptibility and modifiable exposures over time. People at high polygenic risk can have much lower event rates when favorable lifestyle and clinical risk factors are maintained.
A low score should not be used to de-intensify clearly indicated treatment. Statin recommendations based on established disease, very high LDL, diabetes, or high absolute risk remain valid. The score’s relative protection may be overwhelmed by smoking, hypertension, or prolonged hypercholesterolemia.
Medication choices are not currently assigned by the specific variants contributing to a routine CAD PRS. The score may influence how strongly treatment is recommended, not which statin molecule or dose is genetically optimal. Pharmacogenomic testing addresses selected medication-response questions and is a separate discipline.
Some clinicians may consider coronary calcium testing when PRS and conventional risk disagree. For example, a high PRS in an intermediate-risk adult could support imaging if the result would alter treatment. Routine imaging of every high-score young person is not established and exposes the patient to cost and a small amount of radiation.
Follow-up should be based on risk factors and treatment, not repeated DNA testing. The genotype does not need annual measurement. Recalculation might be offered if the laboratory releases a materially improved model, but repeated model changes can confuse patients unless the implications are explained.
A useful prevention plan includes measurable targets: smoking abstinence, blood-pressure goals, LDL reduction, diabetes prevention or control, physical activity, and adherence. The PRS is best viewed as a reason to take established prevention seriously, not as a substitute for it.
What PRS cannot tell you
A PRS cannot diagnose coronary artery disease. It does not show plaque, stenosis, ischemia, myocardial injury, or vulnerable lesions. A person with a high score may have no current disease, and a person with a low score may already have advanced atherosclerosis.
It cannot predict the timing or certainty of a heart attack. Risk estimates describe groups, and individual outcomes remain uncertain. The score also cannot specify whether a future event would result from plaque rupture, spasm, embolism, spontaneous coronary dissection, or another mechanism.
It cannot replace testing for rare monogenic disorders. Very high LDL, tendon xanthomas, or a striking dominant family history calls for familial-hypercholesterolemia evaluation. Cardiomyopathy, arrhythmia syndromes, and aortic disease require phenotype-specific genetic testing, not a CAD PRS.
It cannot fully represent ancestry, environment, or social conditions. Access to healthy food, preventive care, safe exercise, medication, and freedom from chronic stress can shape cardiovascular outcomes. A genetic score should not be used to imply that disparities are biologically predetermined.
It cannot determine that a family member shares the same category. Siblings inherit different combinations of common variants and can have substantially different scores. Testing one person does not screen the family in the way a known pathogenic variant does.
It cannot justify ignoring symptoms. New chest pressure, shortness of breath, sweating, nausea, faintness, or pain radiating to the arm, jaw, or back requires urgent clinical assessment regardless of genetic risk category. Acute care decisions depend on symptoms and objective testing.
Finally, it cannot resolve every prevention debate. Even a well-validated PRS may shift estimated risk only modestly. The question is not whether the score is statistically associated with disease, but whether it changes a decision in a way likely to improve health.
Questions to ask before acting
Start with the model: What exact PRS was used, how many variants or what method does it include, and what outcome was predicted? Ask which population established the percentile and whether the score was validated in people with similar ancestry, age, sex, and clinical setting.
Clarify the reported metric. Is the number a percentile, standard-deviation value, relative risk, or absolute risk? What is the comparison group? If an absolute percentage is provided, which clinical variables and time horizon were used, and how well is the model calibrated locally?
Ask how the result changes care. Does it alter a statin discussion, LDL goal, blood-pressure strategy, or decision about coronary calcium? What would the recommendation be without the PRS? A result that does not change management should not dominate the medical record.
Review all major risks in parallel: untreated and current LDL, lipoprotein(a), blood pressure, diabetes status, kidney disease, smoking, family history, inflammatory disease, pregnancy-related factors, and symptoms. Determine whether a monogenic test is indicated because a PRS cannot exclude rare high-impact variants.
For a high result, request an absolute-risk explanation and a concrete prevention plan. For a low result, ask which treatments and screening remain indicated despite the score. For an ancestry-mismatched or poorly validated result, consider treating the percentile as uncertain rather than precise.
Ask about privacy, data retention, model updates, and whether the laboratory will recalculate the score. Preserve the full report and date because the algorithm may change. A good consultation ends with proportionate action and uncertainty understood—not with genetic determinism or false reassurance.
References
- Polygenic Risk Scores for Cardiovascular Disease: A Scientific Statement From the American Heart Association. 2022. Professional scientific statement.
- Clinical utility and implementation of polygenic risk scores for predicting cardiovascular disease. 2025. European clinical consensus statement.
- Polygenic Risk Score Implementation into Clinical Practice for Cardiometabolic Disease. 2024. Peer-reviewed review.
- 2026 ACC/AHA/AACVPR/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Dyslipidemia. 2026. Multisociety clinical guideline.
- A multi-ancestry polygenic risk score improves risk prediction for coronary artery disease. 2023. Peer-reviewed validation study.
- Clinical use of polygenic risk scores: current status, barriers and future directions. 2025. Peer-reviewed review.
Disclaimer
This article is for general education and does not replace cardiovascular evaluation, genetic counseling, or individualized preventive treatment. Do not start or stop a statin or other medicine solely because of a PRS report. Seek emergency care for chest pain, stroke symptoms, severe shortness of breath, or another possible cardiovascular emergency.
