CYP2C9: why two people need very different warfarin doses
Warfarin dose requirements vary enormously between individuals, and CYP2C9 is one of the two genes that explains a large part of that spread. The other is VKORC1, and neither is much use without the other.
Also known as: cytochrome P450 2C9.
What CYP2C9 does
The enzyme clears drugs out of the body by oxidising them, and warfarin is its best-known passenger. Warfarin is given as a mixture of two mirror-image forms, and CYP2C9 handles the more potent one. Slower clearance means the drug accumulates, so the dose that produces a stable anticoagulation level is lower. The same enzyme also clears several non-steroidal anti-inflammatories and phenytoin, where the consequence of slow clearance is higher exposure rather than a dosing algorithm.
The metabolizer groups
CYP2C9 uses an activity score. Each allele is worth 1 if it works normally, 0.5 if it is decreased-function, and 0 if it has no function, and the two are added. A score of 2 is a normal metabolizer, 1 to 1.5 is intermediate, and 0 to 0.5 is a poor metabolizer. The two common reduced-function alleles in European populations are *2 and *3, with *3 the more severe of the pair.
Medications where it matters
Warfarin (Marevan, Coumadin)
Warfarin has a narrow margin between too little and too much, and the first weeks of treatment are where most of the bleeding happens. CYP2C9 changes how fast the drug clears, VKORC1 changes how sensitive the target is, and the published dosing algorithms use both alongside age, weight and the reason for treatment. A genotype on its own does not produce a dose.
| If your result is | Published CPIC guidance |
|---|---|
| Normal metabolizer | CPIC: warfarin dosing follows the genotype-guided algorithm together with VKORC1 and clinical factors; no CYP2C9-driven adjustment beyond it. |
| Intermediate metabolizer | CPIC: the genotype-guided warfarin algorithm (CYP2C9 + VKORC1 + clinical factors) applies; reduced function typically lowers the predicted dose. |
| Poor metabolizer | CPIC: the genotype-guided warfarin algorithm applies; poor metabolism predicts a markedly lower dose and closer INR monitoring. |
NSAIDs (e.g. celecoxib, ibuprofen)
The anti-inflammatories cleared by this enzyme sit in a different risk category from warfarin: nobody titrates ibuprofen to a blood test. What a poor metabolizer result changes is the exposure from a standard dose, which matters most for the gut and kidney side effects that come with sustained high levels.
| If your result is | Published CPIC guidance |
|---|---|
| Intermediate metabolizer | CPIC: the guideline discusses starting at the lowest effective dose for affected NSAIDs and titrating to response. |
| Poor metabolizer | CPIC: the guideline discusses a reduced starting dose (or an alternative not cleared by CYP2C9) for affected NSAIDs. |
Is this in your raw DNA file?
CYP2C9 is one of the pharmacogenes Aimosti reads from a plain consumer file. The *2 and *3 alleles are each defined by a single position, rs1799853 and rs1057910, and consumer arrays commonly include both, so this gene is in the standard report rather than Deep Read only. Rarer CYP2C9 alleles, including several that matter more in African and Asian populations, are not on consumer arrays and will read as reference. If a defining position is missing from your export, the report says the genotype is unconfirmed rather than calling it normal.
A genotyping chip reads only a few hundred thousand pre-selected positions, about 0.02% of your genome, chosen for common variation. It does not sequence the rest, so it cannot find the rare or novel pathogenic variants the clinical and carrier modules look for: a “no finding” from chip data means “this chip never looked”, far more so than with whole-genome sequencing. Chip data suits common-variant traits, pharmacogenomic tag SNPs, and haplogroup ancestry, not clinical or carrier screening.
Common questions
Is CYP2C9 in my 23andMe or AncestryDNA raw data?
The two common European reduced-function alleles usually are, and Aimosti reads them from the file you already have. The array does not cover the rarer alleles, so a normal result means the tested positions were normal rather than that the gene is proven normal.
Can I work out my warfarin dose from this?
No. Genotype-guided warfarin dosing uses a published algorithm that combines CYP2C9 and VKORC1 with clinical factors, and the resulting number still has to be checked against INR measurements. This result is an input a prescriber can use, not an answer.
I take a direct oral anticoagulant instead of warfarin. Does this apply?
No. Apixaban, rivaroxaban and the rest are not dosed on CYP2C9 and are not covered by this guidance. The gene matters here only for warfarin and for the drugs listed above.
Why is VKORC1 mentioned but not given its own page?
Because on its own VKORC1 is a genotype without published dosing guidance attached to it. Aimosti reports it as part of the warfarin picture rather than as a standalone finding, which is also why it stays on the coverage list rather than getting a page here.