Quick Answer
A polygenic risk score (PRS) is a number that summarises the cumulative effect of thousands to millions of common genetic variants (SNPs) across your genome to estimate your inherited risk for a specific condition such as coronary artery disease, type 2 diabetes, or breast cancer. Unlike single-gene tests (e.g. BRCA), which test for rare, high-penetrance mutations, PRS captures the additive effect of many common variants each with small individual effects. A high PRS does not mean you will develop a disease — it means your baseline genetic risk is elevated, which should inform prevention strategies.
Polygenic risk scores are built on genome-wide association studies (GWAS) — large-scale investigations that compare the genomes of hundreds of thousands to millions of people with and without a given disease. By scanning millions of single nucleotide polymorphisms (SNPs) — positions in the genome where one DNA letter differs between individuals — researchers identify which variants appear more frequently in people who develop the disease. Each association is quantified as an odds ratio or effect size, representing how much that specific variant shifts population-level disease probability.
To construct a PRS, each SNP identified in GWAS is assigned a weight proportional to its effect size. For every individual, the number of risk alleles at each SNP position is multiplied by that weight, and the products are summed across the entire genome. The resulting score places you on a continuous distribution relative to the reference population — typically expressed as a percentile. Someone in the 95th percentile carries a higher burden of risk variants than 95% of the reference cohort.
The predictive accuracy of a PRS is commonly measured using the area under the receiver operating characteristic curve (AUC). A score of 0.5 indicates no discrimination (no better than chance), while 1.0 represents perfect classification. Current best-in-class PRS for coronary artery disease — built using UK Biobank data and validated in independent cohorts — achieve AUC values of approximately 0.80 to 0.83, meaning they correctly rank cases above controls about 80–83% of the time. This is clinically meaningful but far from deterministic. To learn more about how polygenic risk scores are applied in clinical medicine, including recent validation studies, see our in-depth explainer.
Not all polygenic risk scores are equal in clinical utility. Coronary artery disease (CAD) PRS is currently the most mature application. A landmark 2018 study in Nature Genetics, using data from over 480,000 UK Biobank participants, demonstrated that individuals in the top 8% of CAD PRS have more than triple the average population risk — a magnitude comparable to familial hypercholesterolaemia, a condition that already triggers aggressive clinical intervention. This finding has prompted cardiology societies to explore integrating PRS into statin initiation guidelines, especially for borderline-risk patients where clinical risk calculators (e.g. PCE, SCORE2) provide ambiguous results.
Breast cancer PRS has strong clinical traction, particularly in the context of BRCA gene testing. While BRCA1/2 testing identifies individuals with up to 70–80% lifetime risk, roughly 85% of hereditary breast cancers arise in women without BRCA mutations. A well-validated breast cancer PRS can stratify these non-BRCA women meaningfully — those in the top decile have approximately 2x average risk, warranting consideration of supplemental MRI screening and earlier mammography initiation. For type 2 diabetes, PRS can identify individuals who benefit from more aggressive lifestyle intervention or earlier metformin consideration. For atrial fibrillation, PRS contributes to risk stratification ahead of cardiovascular procedures. Prostate cancer PRS, validated in cohorts exceeding 200,000 men, is beginning to inform PSA screening start-age recommendations in high-risk men.
Psychiatric polygenic risk scores for conditions like major depressive disorder, schizophrenia, and bipolar disorder are statistically robust — derived from studies of over a million participants by the Psychiatric Genomics Consortium — but currently remain less clinically actionable. They are used predominantly in research settings to understand genetic architecture rather than to guide individual treatment decisions. The clinical standard of care has not yet incorporated psychiatric PRS into diagnostic or treatment protocols, though this is an active area of translational research.
The most consequential limitation of current PRS is their reduced accuracy in non-European populations. An analysis published in Cell in 2019 demonstrated that a PRS trained in European cohorts explained roughly 4.5-fold less variance in people of African ancestry. The root cause is linkage disequilibrium — the correlation structure between nearby SNPs differs between ancestral populations. Because most large GWAS have been conducted in populations of European descent (over 78% of participants in GWAS Catalog as of recent audits), the SNP weights embedded in most commercial PRS are calibrated to European genomic architecture. Clinicians using PRS in diverse patient populations must treat these estimates with particular caution and ideally wait for ancestry-specific validation.
PRS predict population-level statistical risk, not individual biological destiny. A person in the 97th percentile of CAD PRS still has a substantial probability of never developing coronary artery disease within their lifetime — the score reflects excess risk relative to a reference population, not certainty. Equally important: PRS does not capture rare variants. A whole genome or exome sequencing approach may detect rare pathogenic variants (e.g. LDLR mutations causing familial hypercholesterolaemia, or Lynch syndrome mutations) that a GWAS-derived PRS is blind to. A low PRS therefore cannot and should not be used to override a strong family history.
Environmental and lifestyle factors commonly dwarf genetic risk for most common diseases. Physical activity, diet quality, smoking status, sleep duration, chronic stress, and environmental toxin exposure each exert influences on disease risk that often exceed the effect of the polygenic component. Epidemiological modelling suggests that transitioning from an unfavourable to a favourable lifestyle can offset approximately 50% of high genetic risk for coronary artery disease. There are also psychosocial risks to consider: disclosure of a high PRS without adequate genetic counselling can induce significant health anxiety without necessarily triggering productive behaviour change, a concern raised by clinical psychologists and geneticists in multiple published commentaries.
Single-gene (monogenic) tests and polygenic risk scores operate on fundamentally different principles and answer different clinical questions. Monogenic tests — such as BRCA1/2 sequencing, Lynch syndrome panel testing (MLH1, MSH2, MSH6, PMS2), LDLR/APOB/PCSK9 testing for familial hypercholesterolaemia, or RET testing for multiple endocrine neoplasia — detect rare variants with high penetrance. These variants are individually rare in the population (often less than 1 in 500 carriers) but carry large, well-characterised effects on disease risk. A positive result in one of these tests is directly actionable at the individual level: it changes screening protocols, informs surgical options (e.g. risk-reducing salpingo-oophorectomy for BRCA1 carriers), and has implications for first-degree relatives who may share the same variant. Insurance coverage in most healthcare systems reflects this clinical utility.
Polygenic risk scores, by contrast, capture the aggregate effect of many common variants, each with individually trivial effects that only become meaningful in aggregate. They are primarily tools for population-level risk stratification — identifying the subset of ostensibly average-risk individuals who harbour a disproportionate burden of common risk alleles and who therefore warrant more intensive screening or earlier preventive intervention. PRS does not replace monogenic testing; the two are complementary layers. A useful clinical heuristic: if a patient presents with strong family history suggestive of a hereditary syndrome (multiple first-degree relatives, early-onset disease, bilateral tumours), the appropriate first step is referral to a clinical geneticist and targeted panel testing, not PRS. If a patient is interested in proactive, population-level risk stratification in the absence of a specific hereditary concern, PRS combined with clinical risk factors (cholesterol, blood pressure, lifestyle) provides meaningful added information.
The distinction also matters for genetic counselling. Monogenic results carry implications for family members and often require cascade testing — an entire framework of clinical genetics infrastructure exists for this. PRS results, while informative for the individual, do not carry the same family-cascade mandate, since each family member will have their own independent polygenic profile. See our guide comparing 23andMe vs clinical genetic testing for a practical breakdown of test types and appropriate use cases.
The clinical-grade route involves ordering through a physician with genetic counselling support. Companies including Genomics PLC (UK-based, partnered with NHS England for the Generation Study), Invitae (US, offers comprehensive hereditary cancer and cardiac panels with optional PRS interpretation), and Color Genomics (US, employer and health system partnerships) provide physician-ordered genomic testing that includes hereditary variant detection alongside polygenic risk assessment. These tests typically sequence or genotype 10–15 million variants after imputation, providing substantially higher resolution than consumer arrays. Prices range from roughly $250–$600 out-of-pocket; insurance coverage varies widely depending on clinical indication.
Consumer-grade options via 23andMe or AncestryDNA use SNP genotyping arrays measuring approximately 600,000–900,000 variants. While insufficient for clinical-grade PRS on their own, raw genotype data from these platforms can be downloaded and re-analysed through third-party calculators such as Impute.me (a free, open-source academic tool hosted by the University of Copenhagen) or GenePlaza, which offer PRS estimates across dozens of diseases using published GWAS weights. These results carry important caveats — imputation introduces noise, array coverage misses many GWAS-significant variants, and the tools are research-grade rather than clinically validated. They are appropriate for curiosity and preliminary personal health education, not for clinical decision-making.
Emerging clinical PRS products from companies including Ambry Genetics and Genomic Life are integrating PRS alongside comprehensive hereditary panel testing, aiming to bridge the gap between consumer and clinical-grade genomics. When you receive a PRS result, bring the following to your doctor: the name of the score and the disease it covers, the company that generated it and the GWAS it was derived from, your percentile ranking relative to the reference population, and any ancestry information relevant to assessing score applicability. This gives your clinician enough context to assess whether the score warrants changes to your screening schedule or preventive treatment plan.
A high PRS is a call to action, not a diagnosis or a sentence. For cardiovascular disease, individuals in the top decile of CAD PRS should discuss with their cardiologist whether their LDL-C treatment threshold should be lower than standard guidelines suggest for their age group. A 45-year-old with LDL of 130 mg/dL might be considered borderline-risk by conventional calculators but high-risk when PRS is factored in — a distinction that can justify earlier statin initiation. Annual or biennial coronary artery calcium (CAC) scoring from age 40 in high-PRS individuals is supported by emerging evidence; a CAC score of zero in a high-PRS individual provides meaningful reassurance and can support a watchful-waiting approach to statin therapy. Aggressive lifestyle optimisation — regular aerobic exercise (150+ minutes per week of moderate-intensity), Mediterranean-pattern diet, smoking cessation, and blood pressure control — is supported by strong evidence for reducing actual CAD event rates even in genetically high-risk individuals.
For breast cancer PRS, the threshold for supplemental MRI screening (alongside mammography) is typically a 20% or greater lifetime risk estimate. Some models incorporating PRS alongside age, family history, and hormone exposure can push borderline-risk women above this threshold, qualifying them for enhanced surveillance that would otherwise not be recommended. Women with elevated breast cancer PRS who are BRCA-negative should discuss this with a breast oncologist or clinical geneticist, as the risk may influence decisions around chemoprevention (tamoxifen or raloxifene for postmenopausal women with sufficient risk elevation).
The general principle across all disease areas is to use PRS as one input into a personalised screening and prevention strategy, not as a standalone oracle. Clinical risk scores (e.g. Framingham for cardiovascular risk, Tyrer-Cuzick for breast cancer risk) already incorporate many of the strongest predictors; PRS adds independent discriminatory power particularly for individuals in whom those conventional tools are most uncertain. Work with a physician who understands genomics, or ask for a referral to a preventive medicine specialist or clinical geneticist who can contextualise your score within your full clinical picture — family history, lifestyle, biomarkers, and imaging findings included.
Free Tool
Take our precision medicine quiz to understand how genetic information, including polygenic risk scores, fits into a comprehensive personalised health strategy.
Open Tool →A polygenic risk score (PRS) is a numerical summary of your inherited genetic risk for a specific disease, calculated by aggregating the effects of thousands to millions of common genetic variants (single nucleotide polymorphisms, or SNPs) across your genome. Each SNP associated with a disease in large-scale studies is weighted by the size of its effect, and these weighted values are summed to produce your score. A higher score means a higher proportion of risk variants are present in your genome.
Accuracy depends on the disease, the size of the GWAS used to build the score, and your ancestry. The best-validated PRS (for coronary artery disease, type 2 diabetes, and breast cancer) can identify roughly 5–8% of the population at 3x or greater average risk. However, most PRS were built predominantly from European ancestry data, making them significantly less accurate for people of African, East Asian, or South Asian ancestry. PRS also does not capture rare high-penetrance variants, so a low score does not rule out a hereditary syndrome.
BRCA1 and BRCA2 testing looks for rare mutations in single genes that dramatically elevate lifetime breast and ovarian cancer risk (BRCA1 mutation carriers have ~70% lifetime breast cancer risk vs ~12% average). These are monogenic, high-penetrance variants. A polygenic risk score for breast cancer instead captures the combined effect of hundreds of common variants, each with tiny individual effects. The two tests are complementary — BRCA-negative women can still have elevated PRS-based risk, and high-PRS women without BRCA mutations may warrant enhanced surveillance.
A high PRS is useful information for health planning, not cause for panic. Your actual disease risk depends on your lifestyle, environment, and other health factors — often more than your genetics. A person with a high cardiovascular PRS who exercises regularly, does not smoke, maintains a healthy weight, and has well-controlled blood pressure and cholesterol will likely have lower actual risk than a sedentary, smoking person with a low PRS. Use a high PRS as motivation to discuss personalised screening and prevention strategies with your doctor.
Yes, 23andMe offers PRS-based health predisposition reports for conditions including type 2 diabetes, coronary artery disease, and some cancers. However, 23andMe uses a chip array (genotyping array) that measures roughly 600,000–900,000 SNPs rather than the 10–15 million imputed in research-grade PRS. This limits accuracy compared to research PRS built on larger variant sets. For clinical decision-making, a physician-ordered clinical-grade genetic test with genetic counselling is preferable. 23andMe is a reasonable entry point for general awareness.
No — this is a significant and acknowledged limitation. Most large-scale GWAS studies have enrolled predominantly European-ancestry participants, so PRS built from these studies perform worse in non-European populations. A PRS developed in Europeans may misclassify risk in people of African, South Asian, East Asian, or mixed ancestry. Research to build ancestry-diverse PRS (e.g. the PAGE study, the Global Biobank Meta-analysis Initiative, and the Million Veteran Program) is ongoing. When interpreting a PRS, ask whether the reference population matches your ancestry.
© 2026 QuanMed - All rights reserved