Reading a Polygenic Risk Score for Coronary Artery Disease

Drew, a patient handed me a consumer genomics report. One line is red: coronary artery disease polygenic risk score, 94th percentile. He's 41, LDL-C 128, BP fine, non-smoker, dad had a stent at 58. He wants to know if he's having a heart attack and whether he needs a statin today.

The first trap is in the label. Ninety-fourth percentile means he scores higher than 94 percent of the reference population — it is a rank, not a probability. It does not say his lifetime risk is 94 percent.

So what does the score actually represent mechanically?

A polygenic risk score aggregates thousands of SNPs — single nucleotide polymorphisms, meaning positions in the genome where a single DNA letter commonly differs between people — each weighted by how strongly it associates with CAD in large genome-wide association studies. The weights are multiplied by which version of each letter the patient carries, then summed into one number, then ranked against a reference cohort.

And there's the second trap — the reference cohort?

Exactly. Most commercial CAD scores were trained predominantly on European-ancestry cohorts. If your patient has South Asian, African, or admixed ancestry, the weights may not apply and the percentile can be actively misleading. That's not a caveat — it's a reason to pause before acting.

Assuming the ancestry is appropriate, what does the 94th percentile actually convert to in terms of absolute risk?

You need to plug the percentile into a validated absolute-risk framework. The Khera et al. 2018 Nature Genetics data — which is guideline-referenced — showed that individuals above the 90th percentile have roughly a threefold higher relative risk than those at the median, translating to approximately the same 10-year absolute risk as traditional high-risk patients. But you still have to run his Pooled Cohort Equations to get his actual number.

And the third trap — the score says nothing about existing disease?

Correct. A high PRS tells you about predisposition, not about whether atherosclerosis is already present. If the clinical question is whether to start a statin right now, a coronary artery calcium score is what bridges that gap — guideline-endorsed by ACC/AHA for exactly this intermediate-risk reclassification.

So in practice for this patient, what do I actually do?

Confirm ancestry appropriateness of the score. Run his 10-year PCE risk. His LDL-C and family history already put him in a discussion zone. If you're uncertain whether to initiate a statin, a CAC score of zero lowers urgency considerably; a CAC above 100 Agatston units moves him to initiation regardless of the PRS.

What do I tell him about the PRS itself? He's treating that 94th percentile like a sentence.

Tell him it shifts his prior probability upward — it's one more reason to treat modifiable risk factors seriously. It does not diagnose disease and it does not override a CAC of zero. A high score usually changes timing and intensity, not the diagnosis.

One thing to remember?

The percentile is a rank within a population that may not be his population. Convert it to absolute risk, check the ancestry fit, and let the CAC score — not the headline label — drive the statin decision.