SNP Panels in Clinic: What to Act On and What to Ignore

So a patient comes in holding a consumer genetics printout. MTHFR C677T, heterozygous, highlighted in yellow. They want to know what it means for their health. Where do I even start?

Start by asking what the test was built to answer. A SNP — single-nucleotide polymorphism, meaning one DNA letter that varies between people and is common enough in the population to be considered normal variation — tells you almost nothing in isolation. MTHFR C677T heterozygous is carried by roughly forty percent of people. It is not a diagnosis.

But the report flagged it. Doesn't flagging mean something?

Flagging reflects panel design, not clinical weight. For MTHFR, there is no CPIC — Clinical Pharmacogenomics Implementation Consortium, the main guideline body here — recommendation to change any drug or supplement based on heterozygous C677T alone. The American College of Medical Genetics explicitly advises against reflex folate or B12 supplementation from this result without an elevated plasma homocysteine to confirm functional impact.

So the actionable rows are the pharmacogenomic ones — like CYP2C19?

Exactly. CYP2C19 intermediate metaboliser status has CPIC level A evidence — their highest tier — for adjusting clopidogrel, some SSRIs, and voriconazole. That result changes prescribing today, in guideline-supported ways. MTHFR does not.

What about the polygenic bars — cardiovascular risk, cognitive resilience? Those look more comprehensive.

Those are polygenic scores — weighted sums of thousands of SNPs that together correlate with population-level disease risk. The cardiovascular ones have the best evidence; a high-percentile score for coronary artery disease does independently predict events in large cohort studies. But we don't yet have trial evidence that acting on a polygenic score improves outcomes beyond acting on the same patient's LDL, blood pressure, and smoking status.

So I shouldn't intensify a statin based on a high polygenic score?

Not on that score alone, not yet. The score adds information; it doesn't yet have a treatment threshold attached to it in any major guideline. Use it to support a conversation about risk, not to override a clinical decision you'd make differently.

And if the panel is negative — no flags at all — can I tell the patient they're clear?

No, and this is the thing most patients misread. A curated panel checks perhaps a few hundred pre-selected positions. It will miss a rare pathogenic variant at the same gene if that variant isn't on the panel. A negative result means the panel didn't find what it looked for, not that nothing is there.

So a symptomatic patient with a negative panel still needs a clinical workup?

Completely. If someone has unexplained thrombosis and their panel shows no Factor V Leiden, they need formal thrombophilia testing, not reassurance from a consumer report.

What's the single thing you'd want me to carry out of this?

Ask three questions about any row: is there a CPIC or equivalent guideline behind it, does the zygosity and haplotype match the actionable pattern, and what was the panel actually designed to detect. If you can't answer all three, the row is background noise until proven otherwise.