aimosti

CYP2D6: the pharmacogene a genotyping chip cannot honestly call

CYP2D6 acts on more prescription drugs than any other single enzyme, and it is the hardest pharmacogene to read from consumer data. Both of those things are true at once, which is why this page spends as much space on what the data cannot say as on what it can.

Also known as: cytochrome P450 2D6, debrisoquine 4-hydroxylase.

What CYP2D6 does

The enzyme sits in the liver and either activates a drug or clears it. What makes CYP2D6 unusual is its neighbourhood: two near-identical pseudogenes, CYP2D7 and CYP2D8, sit beside it, and the region is prone to whole-gene deletions, duplications and hybrid copies that splice real gene and pseudogene together. Reading CYP2D6 therefore means counting copies and resolving which sequence belongs to which gene, rather than checking a handful of positions.

The metabolizer groups

CPIC gives each CYP2D6 allele an activity value, from 0 for a dead allele up to 1 for a fully working one, then adds the two values a person carries. A duplication multiplies its allele's contribution, which is how a score can exceed 2. The total falls in a band: 0 is a poor metabolizer, 0.25 to 1.0 intermediate, 1.25 to 2.25 normal, and 2.5 and above ultrarapid. For drugs the enzyme activates, the two ends of that scale fail in opposite directions, and both appear in prescribing guidance.

Medications where it matters

Codeine / tramadol

Both are prodrugs. Codeine does very little until CYP2D6 converts a fraction of it to morphine, and tramadol depends on the same enzyme for its stronger metabolite. Poor metabolizers get weak pain relief from a standard dose. Ultrarapid metabolizers make the active opioid faster than expected, which is the safety end of the problem rather than the efficacy end. Both ends have published guidance, shown below.

If your result isPublished CPIC guidance
Poor MetabolizerCPIC: the guideline suggests an analgesic not metabolised by CYP2D6 (e.g. morphine or a non-opioid), avoiding codeine/tramadol.
Intermediate MetabolizerCPIC: the guideline notes monitoring response; an alternative analgesic may be considered if response is inadequate.
Normal MetabolizerCPIC: standard use per label.
Ultrarapid MetabolizerCPIC: the guideline suggests avoiding codeine/tramadol and using an analgesic not metabolised by CYP2D6.

Tamoxifen

Tamoxifen is also a prodrug: CYP2D6 converts it to endoxifen, the metabolite that does most of the work in hormone-receptor-positive breast cancer. Less enzyme activity means less endoxifen. This is oncology territory, and the guidance below is something to raise with the treating team, never a reason to alter a cancer treatment on your own.

If your result isPublished CPIC guidance
Poor MetabolizerCPIC: the guideline discusses considering an aromatase inhibitor where appropriate; confirm with your oncologist.
Intermediate MetabolizerCPIC: the guideline notes that intermediate metabolisers form somewhat less endoxifen than normal metabolisers; your oncologist can weigh whether this affects treatment choice.
Normal MetabolizerCPIC: standard use per label.
Ultrarapid MetabolizerCPIC: standard use per label.

Is this in your raw DNA file?

This is the gene where a genotyping chip runs out of road. A 23andMe or AncestryDNA file may carry a few CYP2D6 positions, but a chip cannot count gene copies, cannot see a whole-gene deletion, and cannot tell a real CYP2D6 sequence from its pseudogene neighbour. Those are precisely the events that set the activity score. So Aimosti does not report CYP2D6 from a chip file at all: the gene belongs to the Deep Read panel, which runs from aligned reads in a BAM or CRAM. Even there, a diplotype the reads do not resolve is reported as Indeterminate rather than rounded to the nearest plausible answer.

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.

See how Aimosti reports it

Common questions

Is CYP2D6 in my 23andMe or AncestryDNA raw data?

A few CYP2D6 positions may be, but the variation that decides the result is structural: whole-gene deletions, duplications and pseudogene hybrids. A chip file has no way to see any of those, so a CYP2D6 call from chip data alone is a guess dressed up as a result. Aimosti does not make it.

What kind of file does Aimosti need to read CYP2D6?

Aligned reads, meaning a BAM or CRAM file. That is the input for the Deep Read panel, and CYP2D6 is one of the genes in it. If a genotyping-chip export is all you have, this gene is out of reach, and no amount of processing changes that.

Other services report a CYP2D6 result from my chip file. Are they wrong?

They are reporting something the file cannot support. A chip reads pre-selected single positions; CYP2D6 is defined by copy number and by sequence that a chip never looks at. A result produced that way may happen to be right, but the data behind it cannot tell you when it is wrong.

Does CYP2D6 affect antidepressants as well?

Yes, and CPIC publishes separate guidance for several SSRIs and tricyclics that involves this gene. Aimosti's Deep Read currently restates the codeine, tramadol and tamoxifen guidance for CYP2D6, which is what the tables above show. The antidepressant guideline is linked in the sources if you want to read it directly.

Sources

Written by Raine Laurila. Last reviewed 2026-07-22.

This page is educational and is not medical advice. It restates published CPIC guidance and does not replace a conversation with your prescriber.