The promise is compelling: a cheek swab, a few weeks, and a personalised list of which medications will work for you and which to avoid. Pharmacogenomic (PGx) testing has been heavily marketed to psychiatrists and patients for over a decade. For some specific gene-drug pairs, the evidence is real and the results are useful. For most of what's on the test panels, the evidence is much thinner than the marketing implies.
This is an honest tour of what's known, what's overpromised, and what to do if a clinician or test company offers PGx testing.
Some PGx findings are genuinely clinically useful — particularly CYP2D6 and CYP2C19 variants for antidepressants and a handful of antipsychotics — but most large commercial test panels include genes whose clinical utility in schizophrenia is weak or unproven.
What pharmacogenomics actually tests
Most PGx panels look at variants in two categories:
- Pharmacokinetic genes — variants in liver enzymes (CYP2D6, CYP2C19, CYP1A2, CYP3A4) that affect how fast a drug is metabolised. People can be ultra-rapid metabolisers, normal, intermediate, or poor metabolisers depending on their genotype.
- Pharmacodynamic genes — variants in drug targets (serotonin receptors, dopamine receptors, transporters) that may affect response or side effects.
Pharmacokinetic findings have stronger evidence. Pharmacodynamic findings, especially for psychiatric drugs, are mostly weaker associations from small studies.
Where the evidence is solid
CYP2D6 and risperidone, aripiprazole, perphenazine, haloperidol
CYP2D6 is the main metaboliser for several antipsychotics. CYP2D6 poor metabolisers can have significantly higher drug levels and more side effects at standard doses. The FDA labels for several drugs (including aripiprazole and pimozide) specifically mention CYP2D6 status. The Clinical Pharmacogenetics Implementation Consortium (CPIC, cpicpgx.org) publishes evidence-graded dosing guidance.
CYP2C19 and antidepressants
For people with schizophrenia who also need an antidepressant (depression in schizophrenia is common), CYP2C19 status affects the metabolism of citalopram, escitalopram, and several other SSRIs. Dosing recommendations exist.
CYP1A2 and clozapine, olanzapine
CYP1A2 is the main metaboliser for clozapine and olanzapine. Genetic variation matters less here than smoking status (which strongly induces CYP1A2). Genetic testing is less informative than just measuring the plasma level — see plasma-level monitoring.
HLA-B*15:02 and carbamazepine
For people of certain Asian ancestries, HLA-B*15:02 testing before starting carbamazepine is recommended because of severe skin reaction risk. Carbamazepine isn't a primary schizophrenia drug but is sometimes used as augmentation.
Where the evidence is much weaker
- Most pharmacodynamic gene variants on commercial panels (5-HTTLPR, BDNF, COMT, DRD2 variants, MTHFR) have small or inconsistent effects in schizophrenia trials.
- "Combinatorial" reports that traffic-light colour-code dozens of drugs as green/yellow/red are usually based on aggregating weak signals into a single recommendation. They look authoritative but the evidence behind individual recommendations varies wildly.
- Several large randomised trials (notably GUIDED) have shown only small or null benefits of PGx-guided prescribing on outcomes compared with usual care.
The big trials
The GUIDED trial (Greden et al., 2019), the largest randomised study of combinatorial PGx-guided antidepressant prescribing, showed a small benefit on response rate at 8 weeks but no significant benefit on the primary outcome of symptom reduction. Smaller schizophrenia-focused PGx trials have been similarly underwhelming. The evidence simply isn't strong enough to justify the marketing claims of most commercial tests.
What about specific antipsychotic choice?
If a PGx test tells you that one antipsychotic is "preferred" over another based on combined gene findings, ask:
- Which specific gene-drug pair is driving that recommendation?
- Is that pair on the CPIC list with high evidence grade?
- Or is it a pharmacodynamic association extrapolated from small studies?
For real metabolic findings (CYP2D6 poor metaboliser status), starting at a lower dose of a CYP2D6-metabolised antipsychotic is reasonable. For most other recommendations, treat them as one input among many — not as a verdict.
Cost and coverage
PGx testing in the US ranges from a few hundred to over a thousand dollars depending on the panel. Insurance coverage is patchy. Some labs bill aggressively whether or not the result is clinically used. Before agreeing to a test, ask about cost, coverage, and what specific decisions the test will inform.
What this means in practice
Reasonable uses of PGx in schizophrenia
- You've had unusual side effects at low doses of multiple antipsychotics — CYP2D6 testing may explain it.
- You're being considered for a drug with FDA labelling that includes a PGx recommendation (e.g., aripiprazole at low doses in CYP2D6 poor metabolisers).
- You need carbamazepine and have East Asian ancestry — HLA-B*15:02 testing is recommended.
Less helpful uses
- Routine testing of every patient before starting any antipsychotic.
- Using a combinatorial colour-coded report as the primary basis for choosing a medication.
- Switching a stable, well-tolerated medication because a test result colour-coded it as "yellow."
Companies that market PGx tests directly to patients with strong personalisation claims. The marketing usually outpaces the evidence, and a confidently-presented report can override the more important clinical conversation.
The role of plasma levels vs PGx
For some questions, just measuring the drug level (plasma TDM) is more informative than predicting metabolism from genes. Genes are one input; smoking, diet, drug interactions, and adherence are others. The level integrates all of them. For clozapine especially, TDM is more useful than CYP1A2 genotyping in routine care.
The bottom line
Pharmacogenomics in schizophrenia is real but narrower than the marketing. A handful of gene-drug pairs warrant testing in specific situations. Most commercial panels go far beyond what the evidence supports. The right framing: useful tool in selected cases, not a replacement for careful clinical reasoning.
This article is for educational purposes only and is not medical advice, diagnosis, or treatment. Always consult a qualified mental health professional. If you or someone you know is in crisis, call or text 988 in the US, or your local emergency number.