Polygenic Risk Scores: The Future of Personalized Disease Prediction
Dr. RP, MD — Board-Certified, Emergency Medicine & Critical Care Medicine — Founder, Analog Precision Medicine
I spent a year as an exchange student in Sweden at seventeen — living with a local family, learning the language as I went. One of the things that stays with me is how that experience demolished the assumption that people who look similar on the surface share a similar underlying biology. Two healthy teenagers, same rough age and build, could respond completely differently to the same food, the same illness, the same physical demand. It was an early lesson that the visible presentation of a person tells you very little about what is happening at a deeper level.
That observation turns out to be directly relevant to preventive medicine. Two patients who appear demographically indistinguishable — same age, same blood pressure, same cholesterol — can have coronary artery disease risk profiles that are miles apart because of differences in their inherited genomics that no standard clinical test will reveal. Polygenic risk scores are an attempt to bring some of that invisible biology into clinical view.
They are genuinely useful for a defined set of applications, and significantly overhyped for others. Getting specific about which is which matters.
The Concept
Most common diseases — coronary artery disease, type 2 diabetes, breast cancer, atrial fibrillation — are not caused by a single gene. They arise from the aggregate effect of hundreds or thousands of genetic variants, each individually too small to matter, but collectively creating real differences in lifetime risk between individuals. A polygenic risk score (PRS) sums these variants — weighted by their known effect sizes from genome-wide association studies — into a single numerical estimate of where a person sits on the inherited risk distribution for a given disease.
Someone in the top 5% of a coronary artery disease PRS has substantially higher lifetime risk than someone at median, even if today's clinical risk factors look identical. The score isn't measuring current disease. It's measuring the inherited predisposition that's been present since birth and will remain constant regardless of what that person does.
The Evidence That Makes the Case
Cardiovascular disease is where the PRS evidence is most mature. A 2023 Nature Medicine study built a multi-ancestry CAD polygenic score from data spanning more than 269,000 cases and 1.1 million controls across five ancestry groups. The resulting score identified roughly 20% of the population at 3-fold elevated CAD risk compared to median, with a comparable fraction at 3-fold reduced risk (Patel et al., Nat Med, 2023).[1] A 3-fold relative risk difference — without any symptoms, without any current lab abnormalities — is not a trivial signal.
A 2024 European Heart Journal analysis confirmed that adding a cardiovascular PRS to established clinical risk scores improved prediction in a real clinical setting, with the greatest benefit for patients in the borderline and intermediate risk ranges — exactly where the standard model is least decisive (Samani et al., Eur Heart J, 2024).[2] This is the practical case for PRS: not that it revolutionizes low-risk or high-risk patients, but that it helps resolve the ambiguous middle.
The American Heart Association published a scientific statement in 2022 endorsing PRS as a meaningful additive tool for cardiovascular risk assessment, identifying it as ready for clinical evaluation while noting that further validation work is needed (O'Sullivan et al., Circulation, 2022).[3]
“The practical case for PRS: not that it revolutionizes low-risk or high-risk patients, but that it helps resolve the ambiguous middle.”
For breast cancer, PRS has been incorporated into clinical hereditary cancer risk tools like BOADICEA, adding information beyond family history and BRCA status. For type 2 diabetes, high-PRS individuals develop disease years earlier than those at median genetic risk even when matched for lifestyle factors.
What PRS Doesn't Do
The scientific critique is real and deserves fair representation. A 2023 critical appraisal in the Journal of Community Genetics found that PRS alone has consistently low diagnostic or prognostic performance — and that adding PRS to existing clinical risk models produces moderate improvement at best, typically raising the AUC by 0.01–0.04 (Yin et al., 2023).[4] Statistically real; not transformative in isolation.
Ancestry bias. More than 90% of GWAS participants have been of European ancestry. PRS trained on these populations underperforms in individuals of non-European ancestry. Multi-ancestry efforts are actively underway — the NHGRI eMERGE consortium published validation work across diverse populations in 2024 — but the limitation remains significant and is an equity problem, not just a technical one.
PRS is one component of risk, not the whole story. A disease that is 50% heritable is 50% determined by non-genetic factors. A high PRS doesn't mean disease is inevitable. A low PRS isn't a guarantee against it. The genetic information is additive to — not a replacement for — traditional risk assessment, lifestyle, and biomarker data.
Communication context matters. A 2022 randomized study found that disclosing CAD PRS results increased information-seeking and shared decision-making around preventive treatment (Maamari et al., JACC Adv, 2022).[6] But delivering a high genetic risk result without adequate clinical context and counseling can generate anxiety without proportionate actionability. How the results are discussed matters as much as what they show.
Where the Clinical Value Is Clearest
The strongest current application is reclassification of borderline-risk patients — the large population that falls in the indeterminate middle of standard cardiovascular risk calculators. For someone at borderline clinical risk, a high PRS justifies escalation: earlier coronary calcium scoring, more aggressive lipid management, or more intensive lifestyle intervention. A low PRS in the same borderline clinical risk zone provides modest reassurance that allows a more conservative approach.
For cancer, the actionability is similar: a high breast or prostate cancer PRS modifies the timing and intensity of surveillance in ways that follow established screening logic, just personalized to that individual rather than applied at population-level age cutoffs.
The NHGRI eMERGE network's 2024 Nature Medicine paper, describing validated PRS implementation for ten conditions in 25,000 diverse patients across multiple health systems, represents the most serious current attempt to operationalize this clinically — using medical actionability as an inclusion criterion, not just statistical performance (Lennon et al., Nat Med, 2024).[5]
The Practical Bottom Line
Polygenic risk scores are a meaningful addition to comprehensive risk assessment for a defined set of conditions — particularly cardiovascular disease, breast cancer, and type 2 diabetes — when interpreted by a clinician who understands their limitations and integrates them with clinical and lifestyle context.
They are not standalone diagnostic tools. They do not predict individual fate. And their utility is substantially diminished when results are handed to a patient without adequate interpretation.
Used as one component of a multi-layer risk profile, alongside traditional risk factors, advanced biomarkers, and family history, a PRS provides genuine additive information that can change clinical decisions in the patients who need the most guidance: those sitting in the ambiguous middle of the risk distribution where standard models give the least traction.
References
- 1.Patel AP et al. A multi-ancestry polygenic risk score improves risk prediction for coronary artery disease. Nat Med. 2023;29:1793–1803.
- 2.Samani NJ et al. Polygenic risk score adds to a clinical risk score for cardiovascular disease. Eur Heart J. 2024;45(34):3152–3160.
- 3.O'Sullivan JW et al. Polygenic risk scores for cardiovascular disease: AHA scientific statement. Circulation. 2022;146:e93–e118.
- 4.Yin Z et al. Clinical utility of polygenic risk scores: a critical 2023 appraisal. J Community Genet. 2023;14:343–355.
- 5.Lennon NJ et al. Selection, optimization and validation of ten chronic disease polygenic risk scores. Nat Med. 2024;30:480–487.
- 6.Maamari DJ et al. Clinical implementation of combined monogenic and polygenic risk disclosure for CAD. JACC Adv. 2022;1(1):1–11.
Dr. RP, MD is dual board-certified in Emergency Medicine and Critical Care Medicine and is the founder of Analog Precision Medicine, a precision medicine practice in Southern California. This article is for educational purposes only and does not constitute medical advice or establish a physician-patient relationship.
← Back to Blog