CYP2C9 Genotype

If a prescriber ever starts you on warfarin, phenytoin, siponimod, or a CYP2C9-cleared anti-inflammatory pain medicine, the first dose is often built from clinical factors and population averages. Your liver may not be average. Roughly one in three people of European ancestry carries a reduced-function version of this enzyme; a much smaller group has very low predicted activity and clears some drugs far more slowly.

CYP2C9 is short for cytochrome P450 2C9. It is one of the handful of liver enzymes that break down roughly 15% to 20% of clinically used drugs. This result does not tell you whether your liver is healthy today. It tells you how a short list of drugs should be prescribed when they come up.

This is one of the better-established pharmacogenomic results. The part that still needs care is scope: it only matters for specific drugs, and a normal call is only as good as the alleles the panel covered.

What the Enzyme Actually Does

CYP2C9 is made mostly in liver cells. It chemically changes drug molecules so the body can remove them through urine or bile. It is one of the most abundant drug-metabolizing enzymes in the liver, and it handles many widely prescribed medications.

The gene comes in variant forms, labeled with a star and a number. 1 is the normal-function reference version. 2 and 3 are the two best-characterized reduced-function variants in people of European ancestry, with 3 causing a larger loss of activity than 2. 5, 6, 8, and 11 are more common in people of African ancestry and are missed entirely by panels that only test 2 and *3.

Your two copies combine into an activity score, which gets translated into a prescribing label: normal metabolizer, intermediate metabolizer, poor metabolizer, or indeterminate. Intermediate usually means mildly to moderately reduced clearance. Poor means very low predicted activity. The lower the activity, the longer some standard doses linger and the higher your exposure can be.

Warfarin

This is the association the test is best known for, and the one with some of the strongest evidence behind it. Warfarin's more potent S-warfarin form is cleared mostly by CYP2C9. If the enzyme runs slow, the drug can accumulate, your blood can thin further than intended, and the risk of a serious bleed goes up during the first weeks of treatment.

The practical effect is on dose. People with reduced activity often need less warfarin than normal metabolizers to reach the same anticoagulation target, and they may take longer to settle into a stable dose. Guideline-backed dosing algorithms combine CYP2C9 genotype with VKORC1, and sometimes CYP4F2 and rs12777823, plus clinical factors like age, weight, and interacting drugs.

CYP2C9 inactivates S-warfarin. A slow enzyme leaves more active drug around, which is the opposite of how some other pharmacogenetic pairings work.

If warfarin is likely, or if a close relative needed unusually low doses or had an early bleeding scare on it, put this genotype in the chart before dosing decisions start.

Siponimod

Siponimod, an oral multiple sclerosis drug, is one of the clearest cases because the drug label calls for CYP2C9 genotyping before treatment. People with a 3/3 genotype should not use it because exposure can become too high. People with 1/3 or 2/3 are usually dosed lower.

NSAIDs and Bleeding Risk

Several common anti-inflammatory pain medicines run through this enzyme, including celecoxib, ibuprofen, meloxicam, piroxicam, and flurbiprofen. Reduced activity can raise drug exposure. With NSAIDs, higher exposure can add to the baseline risk of stomach ulcers, bleeding, kidney strain, and blood-pressure rise.

Prescribing guidance is drug-specific. For short-acting drugs, lower starting doses may be used. For meloxicam, a lower dose or another drug may be chosen, and in poor metabolizers guidance favors a different drug. For piroxicam, guidance often favors another drug in people with low activity because it can build between doses. The pharmacology here is solid; the direct evidence tying genotype to actual NSAID side effects is still thin.

Phenytoin and Seizure Medication

Phenytoin has a narrow space between a dose that controls seizures and a dose that causes harm, and CYP2C9 is its main clearance route. Carriers of reduced-function variants can reach toxic levels at doses that are fine for others. Signs of that toxicity include unsteadiness, involuntary eye movement, and confusion.

Phenytoin dosing guidance built on CYP2C9 genotype recommends lower maintenance doses for intermediate and poor metabolizers, on the order of a quarter less for one reduced-function copy and about half for two. Slow clearance is not the only reason the genotype matters with this drug: *3 carries its own association with rare but severe skin reactions to phenytoin, separate from any effect on drug levels.

For people with ancestry in groups where HLA-B15:02 is more common, such as many Southeast Asian and some East or South Asian populations, HLA-B15:02 is checked because it predicts a rare severe skin reaction rather than a dosing problem.

Diabetes Drugs and Other Substrates

The older sulfonylurea diabetes drugs, including glipizide, glyburide, and glimepiride, are cleared in part by CYP2C9. The pharmacology is clear: slower clearance can raise drug levels. The link to real-world low blood sugar is weaker and less consistent than it is for warfarin or phenytoin, with the clearest signal for *2, so this result is a risk flag, not a dosing rule by itself.

The substrate list is long, so it helps to think in groups rather than memorize names:

• Anticoagulants: warfarin, where genotype affects dose most sharply.
• Multiple sclerosis drugs: siponimod, where genotype affects whether and how the drug is used.
• Anti-inflammatories: celecoxib, ibuprofen, meloxicam, piroxicam, flurbiprofen.
• Seizure drugs: phenytoin, where the safe dose range is narrow.
• Diabetes drugs: older sulfonylureas, where slow clearance can raise the risk of low blood sugar.

Two other things can change enzyme activity from the outside. Inhibitors, including fluconazole, amiodarone, metronidazole, and trimethoprim-sulfamethoxazole, slow the enzyme down temporarily, so a normal metabolizer can behave more like an intermediate one while on them. Inducers, including rifampin, carbamazepine, and phenobarbital, speed it up and can push drug levels too low. Your genotype is the fixed baseline; drug interactions move you around it.

Ancestry Determines What a Negative Result Means

This is the main limitation of the test, and it is easy to miss. The commonly tested variants, 2 and 3, are relatively frequent in people of European ancestry and much rarer elsewhere. In one analysis of global population data, about 35% of Europeans carried a reduced-function or low-function phenotype from these two variants, compared with roughly 27% of South Asians, about 7% of East Asians, and only about 2% of people of African ancestry. The reduced-function variants that matter more in people of African ancestry, such as 5, 6, 8, and 11, are on some panels and not others.

So a normal metabolizer call means different things depending on both your ancestry and which variants the lab looked for. If you have African ancestry and the panel tested only 2 and 3, a normal result does not rule out reduced function. Check which alleles were covered. A panel that does not name the alleles relevant to your background has not ruled them out.

When a Genetic Result Can Still Mislead

• Panel coverage: the assay detects only the variants it was designed to detect. Rare variants outside the panel may be reported as normal by default.
• Ancestry: a normal result is less reassuring when the panel skips variants that are more common in your ancestry group.
• Rare or uncertain variants: sequencing may find a rare change the lab cannot interpret. Many star-allele panels skip rare changes and report only the alleles they were built to call.
• Somatic versus germline reports: cancer tumor sequencing and some blood-based tests after a stem-cell or bone-marrow transplant may not reflect the DNA you inherited.
• Drug interactions and organ function: genotype predicts enzyme capacity, not current drug levels. People with the same genotype can still clear a drug at noticeably different rates, and liver disease, kidney disease, inhibitors, and inducers can change exposure on top of genotype.
• Consumer-grade reports: a direct-to-consumer genotype may not have the same allele coverage or confirmation standard as a clinical pharmacogenomic panel.

A One-Time Result You Use When Prescriptions Change

Your genotype does not change, so this is not a marker you track. There is no trend, no retest, no seasonal drift. The value works the other way: one result, then years of prescriptions written with it in mind.

What does need ongoing monitoring is the phenotype downstream of any drug you take. On warfarin, that is your INR, checked frequently while a dose is being established and then at a stable interval after. On phenytoin, it is drug levels plus liver enzymes. On sulfonylureas, it is glucose. On siponimod, it is the drug-specific safety monitoring such as blood counts and infection watch. The genotype sets the starting point; those numbers confirm the drug is landing where it should.

Get this once, then make sure it is somewhere a prescriber will see it. A result buried in an old email does nothing. Put it in your medical record, keep a copy on your phone, and bring it up when any new prescription is being written.

What to Do With a Reduced-Function Result

An intermediate or poor metabolizer result is not a diagnosis and does not require treatment. It is a flag on a list of specific drugs. The first move is to check whether anything you currently take is on that list.

If you are already on warfarin, phenytoin, siponimod, a sulfonylurea, or a regular NSAID, bring the result to whoever prescribes it and ask whether the dose or drug choice should be reconsidered. Do not change a dose yourself, particularly with warfarin or a seizure medication, where both too much and too little carry real risk. For warfarin specifically, pairing this with VKORC1 gives a far better dose estimate than either alone.

If a direct-to-consumer report gave the result, confirm it with a clinical pharmacogenomic test before it drives a high-risk prescription. If a clinical result clashes with a strong drug history, such as needing tiny warfarin doses despite a normal call, the issue is usually allele coverage or a drug interaction, not that the gene changed.

If you take none of these drugs, there is nothing to do now, which is the point of testing early. The result waits in your record until it matters. Consider also checking CYP2C19 and CYP2D6, which cover a different and largely non-overlapping set of common drugs including clopidogrel, some antidepressants, and several pain medications. Together those three genes account for a large share of the prescriptions where genotype changes dosing.

Family members share your genes. A poor metabolizer result in you means siblings have a meaningful chance of carrying the same variants, and children carry one of your two copies. That is worth mentioning to them, especially if anyone in the family is facing anticoagulation, phenytoin, siponimod, or regular NSAID use.