CYP2D6 Genotype

Somewhere between one in twenty and one in ten people with European ancestry carry two copies of this gene that don't work, and in some European populations the figure runs above one in ten. About a quarter of commonly used prescription drugs touch this enzyme. If you are one of those people, codeine gives little pain relief, and a standard dose of an older antidepressant can rise to a much higher blood level than expected.

The result is set at conception and does not change. Most people find out only after a drug has already failed them or hit harder than it should have.

Reading a Result Like 1/4

CYP2D6 is short for cytochrome P450 2D6. You inherit one copy from each parent, so the result arrives as a pair. Each version gets a star number. 1 is the reference version that works normally, 4 is the most common no-function version in people with European ancestry, and more than a hundred others sit in between, some working partway.

The lab adds the predicted activity of the two copies into an activity score and sorts you into one of four groups.

• Poor metabolizer: both copies have no function, so the enzyme is effectively absent. Drugs it clears can pile up. Drugs it has to switch on may barely work.
• Intermediate metabolizer: one working copy with one no-function copy, or other combinations that leave reduced activity. 1/4 usually lands here under newer scoring. Older reports sometimes called the same pair normal.
• Normal metabolizer: enough predicted activity that standard starting doses usually apply. Many normal results have two working copies, but the activity score matters more than the label.
• Ultrarapid metabolizer: one or more extra working copies, usually from a duplication. This pattern is more common in parts of North Africa and the Middle East and uncommon in East Asia.

Keep the star alleles and the copy-number call, not just the group name. The categories have changed before. The inherited variants underneath them do not.

Codeine, Tramadol, and the Two Ways This Goes Wrong

Codeine is mostly a prodrug. Most of its pain effect comes after the enzyme turns part of it into morphine. A poor metabolizer may still feel side effects, but gets little pain relief, which is how people can spend years being told they have an unusually high pain tolerance.

The other direction carries a boxed warning. Children have died after being given codeine following tonsil surgery. FDA labeling now contraindicates codeine and tramadol in children younger than 12. It also bars both drugs after tonsil or adenoid surgery in people younger than 18, warns against their use in adolescents 12 to 18 with obesity or breathing problems, and says breastfeeding is not recommended while taking either drug. An ultrarapid mother can convert enough codeine to morphine that her milk carries a real dose to a nursing infant.

This is where the result stops behaving like a good number and a bad number. Fast is not better. For codeine and tramadol, fast can be hazardous and slow can make the drug weak. For tamoxifen, slow is the main concern because less enzyme means less endoxifen. For nortriptyline, paroxetine, or metoprolol, slow means the dose can stack up and fast means it may wash out before it works. The same words on your report mean different things depending on the drug.

Antidepressants and Antipsychotics

The older tricyclic antidepressants are the cleanest case, because the gap between the dose that helps and the dose that can disturb heart rhythm is narrow. CPIC turns gene results into prescribing advice. For intermediate metabolizers, its guidance is to cut the starting dose by about a quarter. For poor metabolizers, the advice is to pick a different drug or halve the dose and measure blood levels. The Dutch pharmacogenetics group lands in a similar place by a different route, setting reductions drug by drug rather than as one flat percentage.

Among non-tricyclic antidepressants, paroxetine has the most data behind it. Drug-level studies find poor metabolizers running roughly four times the serum concentration of normal metabolizers, and intermediate metabolizers about twice as high, while an ultrarapid metabolizer can clear it fast enough that a normal dose never reaches a working level. The prescribing advice here is still moving. The Dutch guideline long advised no dose change for paroxetine, and the US label notes that poor metabolizer pharmacokinetics were never formally studied. Fluvoxamine, venlafaxine, and vortioxetine also have genotype-based guidance, though the evidence is less clean than it is for tricyclics.

Several antipsychotics travel the same path, including risperidone, aripiprazole, and haloperidol. In poor metabolizers, ordinary doses can mean higher drug exposure and a higher chance of dose-related side effects, including movement symptoms.

Tamoxifen After Breast Cancer

Tamoxifen is active, but much of its anti-estrogen effect depends on endoxifen. This enzyme helps make endoxifen, and poor metabolizers make less of it. The drug label puts average endoxifen in poor metabolizers at roughly a third of the level seen in normal metabolizers. CPIC guidance recommends an aromatase inhibitor instead for postmenopausal poor metabolizers, and an aromatase inhibitor with ovarian function suppression for premenopausal poor metabolizers when that fits the cancer plan. It also advises avoiding strong and moderate inhibitors of the enzyme when a substitute exists. Paroxetine and fluoxetine are common examples.

Whether genotype predicts breast cancer recurrence has come out both ways across trials and retrospective analyses. A 2025 re-analysis of 33 trials that adjusted for incomplete genotyping and tumor DNA loss found higher recurrence risk in people with impaired metabolism, but that is one reading of a contested literature. The guidelines split on what to do about it: CPIC treats the genotype as actionable, while NCCN recommends against using CYP2D6 testing to choose tamoxifen. The chemistry is not in dispute: less enzyme means less endoxifen. If you are weighing tamoxifen against an aromatase inhibitor, this is one input among several, and one your oncology team should weigh with you rather than a decision the genotype makes on its own.

Atomoxetine for ADHD

Atomoxetine is used for ADHD. It is cleared mainly by this enzyme, and poor metabolizers reach blood levels roughly ten times higher than non-poor metabolizers on the same dose. That turns the result into a plan rather than a warning: start lower, raise slower, and judge response before adding more.

Drugs That Make a Normal Genotype Behave Poorly

Your genotype sets the ceiling on enzyme activity. Other drugs can push actual activity down. Bupropion, paroxetine, fluoxetine, quinidine, and terbinafine are strong inhibitors, and someone genetically normal who takes one of them can behave like a poor metabolizer for as long as the inhibitor is on board. The dangerous version is the invisible one, where the antidepressant comes from one prescriber and the painkiller from another.

Looking for a list of things that speed this enzyme up is a dead end. Unlike many drug-metabolizing enzymes, this one is hard to induce. Rifampin and St John's wort speed up other clearance routes, but they do not have a clinically meaningful effect here. Pregnancy is the main exception: activity often rises by late pregnancy, enough that dose needs for some drugs can shift.

What the Test Can Miss

• Which variants were looked for: a result showing nothing unusual means none of the tested variants were found. Rare versions outside the panel still exist.
• Extra and missing copies: whole-gene deletions, duplications, and hybrid genes formed with the neighboring CYP2D7 pseudogene need a method built to count copies. Testing that only reads single letters can miss an ultrarapid pattern.
• Ancestry: the common no-function and reduced-function versions differ by population. 4 dominates in Europe, 10 in East Asia, and 17 and 29 in parts of Africa. A panel built around European variants can misclassify someone from another background.
• Tumor-only sequencing: a cancer-panel result from tumor tissue is not the same thing as an inherited pharmacogenetic result. Use germline testing if a drug decision depends on it.
• Uncertain variants: sometimes a lab finds a rare version whose effect is not known yet. Treat that as unknown enzyme activity until the lab or a pharmacogenetics service interprets it.
• Consumer and raw-data reports: direct-to-consumer arrays and third-party raw-data tools may read only selected positions and may not resolve copy number or hybrid genes. A clinical result can legitimately disagree.

A negative result is only as broad as the method. For this gene, method matters.

One Result, Then Years of Use

You take this test once. The sequence does not drift, so there is no trend to follow and no reason to repeat it, unless the lab flags a call it could not resolve, the result came from consumer raw data, or you were typed years ago by a method that never counted gene copies.

The value accumulates instead. Every prescription over the next decades is a chance to use it, which is why the result belongs somewhere you will actually find it under pressure: in your chart, in your pharmacy record, and in a note on your phone with the star alleles spelled out.

There is trial evidence that acting on pharmacogenetic results pays off rather than merely sounding sensible. In a cluster-randomized study across seven European countries, prescribing guided by a twelve-gene panel that includes this one cut clinically relevant drug-related side effects from 27.7% to 21.0% among people with an actionable result, roughly a 30% relative reduction. That finding is about the panel as a whole, not proof that this gene alone caused the benefit.

What to Do With an Out-of-Pattern Result

If a high-stakes decision rests on a consumer report, an old report, or a result that does not count gene copies, confirm it with a clinical pharmacogenetic assay built for star alleles, deletions, duplications, and hybrids.

Then make an inventory. List everything you take, including over-the-counter items and supplements, and find out which ones run through this enzyme. A pharmacist is often the fastest person to do this well.

Look for the combinations that change a decision. A poor metabolizer already taking a strong inhibitor has very little activity left and should usually avoid starting a drug with a narrow safety window unless there is blood-level monitoring. An ultrarapid metabolizer whose antidepressant has never worked at any dose now has an explanation and a reasonable case for switching to something that leaves the body by a different route. An intermediate result with no affected drugs in play is worth filing and nothing more.

Use companion checks when the drug calls for them. For tricyclic antidepressants, that can mean a drug blood level. For metoprolol, it means pulse, blood pressure, symptoms, and dose response. A standard liver panel can show current liver injury or bilirubin problems, which your genotype cannot. It does not measure this enzyme directly.

Two other tests earn their place next to this one. CYP2C19 genotype covers a separate list of drugs with the same logic, including clopidogrel, several acid-reflux drugs, and some of the remaining antidepressants. For anything touching a cancer treatment decision, or a drug you have already reacted badly to, bring in a clinical pharmacist, oncology team, or pharmacogenetics service instead of working from a printout.

Tell your biological relatives. Each of your children inherits one of your two copies, and a full sibling has roughly a one in four chance of carrying the same pair you do.