Pharmacogenomic-guided prescribing

1. Introduction

Version followed: CPIC guideline for CYP2D6, CYP2C19, CYP2B6, SLC6A4 and HTR2A and serotonin reuptake inhibitor antidepressants (Bousman et al., 2023); FDA Table of Pharmacogenetic Associations (content current as of 10 September 2026, checked October 2026); FDA safety communication on pharmacogenetic test claims (November 2018, as reported); ISPG genetic testing statement (updated 11 March 2019).

Pharmacogenomic (PGx)-guided prescribing means that a prescriber uses a person's genotype for selected genes, mostly drug-metabolising enzymes such as CYP2D6 and CYP2C19 and, for some drugs, immune genes such as HLA-B, as one input when choosing a medicine or adjusting how it is used. The idea of change is pharmacological, not psychological: some people break a drug down faster or slower than average, or carry a variant linked to a serious reaction, and knowing this in advance may reduce side effects or failed trials. It is ordered, interpreted and acted on by the prescriber. It is not a psychotherapy and it does not diagnose depression or predict whether a person "will respond to" therapy or to antidepressants in general. [source: CPIC 2023; FDA Table; ISPG statement]

This guide is written for the psychotherapist working alongside the prescriber. It contains no doses or dosing adjustments; any question about starting, stopping, switching or changing a medicine belongs to the prescriber. [synthesis]

2. Framework

PGx-guided prescribing is a clinical decision-support process rather than a staged treatment. As the sources describe it, it runs through these steps, all owned by the prescriber:

Deciding whether to test. No consensus says who should be tested. The ISPG states that current guidelines give no advice on when, or to whom, testing should be offered, and that CYP2C19 and CYP2D6 information is likely most useful for people who had an inadequate response to, or adverse reaction from, a previous antidepressant or antipsychotic. Routine testing of everyone is not recommended by ISPG. CPIC explicitly leaves the questions of whom to test, test choice and cost-effectiveness outside its guideline. [source: ISPG statement 2019; CPIC 2023]
Testing and reporting. A laboratory genotypes selected variants and translates them into predicted phenotypes (for CYP2D6: ultrarapid, normal, intermediate or poor metaboliser, based on an activity score; for CYP2C19 and CYP2B6 also a "rapid" category and "likely" or "indeterminate" calls). Different laboratories test different variants, so two reports on the same person can differ. [source: CPIC 2023]
Interpreting with guidelines and labelling. The prescriber looks up the drug-gene pair in CPIC guidelines and in FDA-approved labelling and the FDA Table of Pharmacogenetic Associations. [source: ISPG statement; FDA Table; Rubinstein & Pacanowski 2021]
Integrating with everything else. Genotype is one factor among age, other medicines (which can shift a person's actual metaboliser status, called phenoconversion), smoking, pregnancy, liver and kidney function, diet and comorbidity. People doing well on a current medicine may gain nothing from genotype-based changes. [source: CPIC 2023; ISPG statement]
Monitoring. Clinical follow-up continues as usual. For carbamazepine the FDA states that genotyping does not replace clinical vigilance. [source: FDA Table; Rubinstein & Pacanowski 2021]

Session count and format. None applies; a test is usually ordered once and the result kept in the record. In therapy, PGx work typically occupies parts of a few sessions: when testing is being considered, when results arrive, and when a medication change is under discussion. [synthesis]

Regulatory status (as of October 2026). In the US most PGx tests have not been reviewed by the FDA or another regulator for their clinical claims; CLIA oversight does not evaluate those claims before tests reach the market. In November 2018 the FDA warned that many tests claiming to predict response to specific medicines had not been reviewed, and that acting on them could lead to inappropriate treatment decisions. In October 2018 the FDA authorised 23andMe's Personal Genome Service Pharmacogenetic Reports as a direct-to-consumer test (de novo pathway, 33 variants), stating that the test does not determine whether a medication is appropriate. Inclusion of a drug-gene pair in the FDA table does not mean the FDA advocates testing before prescribing, except where a companion diagnostic is required. ISPG does not recommend direct-to-consumer genetic testing for medical purposes in people with psychiatric illness or their families. [source: Rubinstein & Pacanowski 2021; RAPS 2018; FDA Law Blog 2018; FDA Table; ISPG statement] UK and EU regulatory status and NHS availability were not checked for this guide. [uncertain]

Evidence status (as sources state it).

GUIDED (Greden et al., 2019): 1,167 outpatients with depression and at least one inadequate antidepressant response; the primary outcome, mean symptom improvement at week 8, was not significantly different (27.2% vs 24.4%), while response and remission were significantly higher in the guided arm. The test was a commercial combinatorial panel. [source: Greden et al. 2019 abstract] Sponsor details are not drawn on. [uncertain]
PRIME Care (Oslin et al., 2022, JAMA): 1,944 veterans at 22 VA centres starting or switching an antidepressant. Testing reduced prescriptions of drugs with predicted drug-gene interactions; remission was higher across the 24 weeks, but not significantly different at week 24. The authors described the effects on remission as small and non-persistent, and the lead investigator said the results do not support testing all patients. [source: US Medicine report of Oslin et al. 2022]
CPIC rates the evidence for CYP2D6, CYP2C19 and CYP2B6 effects on drug levels as mostly moderate to high, with some recommendations only optional. Evidence for the serotonin transporter (SLC6A4) and receptor (HTR2A) genes is mixed and insufficient; CPIC makes no recommendation for them. Proprietary multi-gene algorithms were outside CPIC's evaluation. [source: CPIC 2023]
The FDA has stated that a link between DNA variation and antidepressant effectiveness has not been established. [source: RAPS 2018; Rubinstein & Pacanowski 2021]

3. Guiding principles

The prescriber owns the medicine. Every question about starting, stopping, switching or changing a drug goes to the prescriber; you never interpret a result into a medication action. [source: FDA Table, which tells patients not to change medicines without their prescriber]
Genes inform drug handling, not whether help is possible. Report language such as "poor metaboliser" describes enzyme activity, not the person or their prognosis; say so when clients read it as a verdict. [source: CPIC 2023; synthesis]
One input among many. A result supplements, never replaces, clinical, demographic and lifestyle information. In practice: ask what else the prescriber is weighing before treating a report as decisive. [source: ISPG statement; CPIC 2023]
Know the evidence grades. Some drug-gene pairs carry official labelling and strong guideline recommendations; others are pharmacokinetic only; serotonin-gene results are not actionable. In practice: do not treat every coloured box on a commercial report as equal. [source: FDA Table sections 1–3; CPIC 2023]
Avoid overclaiming in both directions. Neither promise that a test will find "the right drug" nor dismiss a result that may matter (for example HLA-B results before carbamazepine). [source: GUIDED; PRIME Care; Equetro label; synthesis]
Direct-to-consumer results are a starting point for a conversation with the prescriber, not a prescription. [source: FDA 2018 communication as reported; ISPG statement]
Genetic data are sensitive and shared with relatives. Treat results as protected health information; discuss privacy before a client buys or uploads a test. [source: genome.gov GINA page; Minnesota AG 2025; synthesis]
Stay within competence. Psychotherapists do not interpret genotype reports; route interpretation to the prescriber, a clinical pharmacist or a genetic counsellor. ISPG recommends counselling from a professional skilled in mental health and genetic interpretation for diagnostic testing. [source: ISPG statement; synthesis for the PGx extension]
Psychotherapy continues regardless. Testing does not pause or replace evidence-based psychological treatment. [synthesis]

4. Intervention inventory

Part A describes how PGx-guided prescribing is delivered, at a descriptive level, so you can follow what the prescriber is doing. Part B gives the psychotherapist's own interventions. Nothing here tells a client to start, stop or change a medicine. [synthesis]

A1. The test and the report (prescriber or laboratory)

Purpose: to know what a client's report actually contains.
Steps (as delivered): the prescriber or the consumer orders a test; a DNA sample is collected [synthesis: collection method varies by laboratory and is not described in the sources]; the laboratory genotypes a fixed panel of variants and converts them into predicted phenotypes. For CYP2D6, CPIC assigns an activity score and groups people as ultrarapid, normal, intermediate or poor metabolisers; CYP2C19 and CYP2B6 add a rapid category, "likely" calls and an "indeterminate" category when alleles are uncertain. Rare variants may not be on standard panels, so the predicted phenotype can be wrong, and results from different laboratories can disagree. [source: CPIC 2023]
Example language (illustrative): "This report predicts how fast your body clears certain medicines. It doesn't measure your depression or say what will help it."
Watch for: reports from consumer services or multi-gene "psychotropic" panels that combine genes into colour-coded lists; CPIC did not evaluate these proprietary algorithms. [source: CPIC 2023]

A2. Applying CPIC guidelines (prescriber)

Purpose: CPIC guidelines tell clinicians how to use genotype results that are already available; they do not say whom to test. [source: CPIC 2023]
Steps (as delivered): the prescriber identifies the drug-gene pair, reads the phenotype-based recommendation and its strength (strong, moderate or optional), then weighs it against co-prescribed drugs that may cause phenoconversion and against age, smoking, pregnancy, organ function and how well the person is currently doing. The 2023 antidepressant guideline covers SSRIs, SNRIs, vilazodone, vortioxetine and others for CYP2D6, CYP2C19 and CYP2B6; it gives no recommendation for SLC6A4 or HTR2A. Recommendations may concern drug choice or dosing; their content is a prescribing matter and is not reproduced here. [source: CPIC 2023]
Children and adolescents: citalopram, escitalopram and sertraline have the most paediatric PGx data; for other drugs young people are under-represented and applicability is uncertain. [source: CPIC 2023]
Ancestry: most study participants were of European or East Asian ancestry; more research is needed elsewhere. [source: CPIC 2023]
Example language (illustrative): "Your prescriber will check the result against published guidelines and your other medicines; I'd bring the report to that appointment."
Watch for: clients assuming a result means a drug is "banned" or "guaranteed"; CPIC states stable patients may not benefit from changes. [source: CPIC 2023]

A3. FDA labelling and the Table of Pharmacogenetic Associations (prescriber)

Purpose: to show which drug-gene pairs the FDA considers supported, and how strongly.
What the table says: Section 1 lists associations where data support management recommendations (such as monitoring, avoiding a drug or considering alternatives); Section 2 associations with potential impact on safety or response; Section 3 associations with effects on drug levels only, where effects on safety or response are not established. The table is not comprehensive, most associations have not been tested for whether testing improves outcomes, and patients are told not to change medicines without their prescriber. [source: FDA Table, as of 10 September 2026]
Psychiatric examples listed (no management detail): CYP2D6 with aripiprazole, brexpiprazole, atomoxetine, vortioxetine and pimozide (Section 1); CYP2C19 with citalopram (Section 1, including QT-prolongation risk in poor metabolisers); CYP2D6 with paroxetine, fluvoxamine and several tricyclics, and CYP2C19 with escitalopram (Section 3, drug-level effects only); thioridazine is listed as contraindicated in CYP2D6 poor metabolisers. [source: FDA Table, as summarised; check the live table for current entries] [uncertain: entries change with updates]
Labelling example, carbamazepine: the Equetro (carbamazepine, indicated for acute manic or mixed episodes in bipolar I disorder) label carries a boxed warning for serious, sometimes fatal skin reactions (Stevens-Johnson syndrome, toxic epidermal necrolysis), with risk roughly ten times higher in patients of Asian descent; it directs testing for HLA-B*1502 before starting in patients likely to carry the allele and avoiding use in positive patients unless benefit clearly outweighs risk. A second boxed warning covers aplastic anaemia and agranulocytosis, with blood counts before and during treatment. ISPG recommends HLA-A and HLA-B testing before carbamazepine and oxcarbazepine. [source: Equetro label via Guideline Central; ISPG statement] The FDA notes that genotyping does not replace clinical vigilance. [source: FDA Table]
Watch for: any rash, blistering, mouth sores, fever or unusual bleeding or bruising reported by a client taking carbamazepine: pass on the same day (see B6). [source: Equetro label; synthesis for the routing]

A4. Commercial combinatorial panels (prescriber or consumer)

Purpose: to understand the reports clients most often bring.
What sources say: GUIDED tested a combinatorial panel in difficult-to-treat depression: no significant difference on the primary symptom outcome, significant differences on response and remission. PRIME Care found testing shifted prescribing away from drugs with predicted interactions, with small, non-persistent remission effects, not supporting testing for all patients. CPIC did not evaluate proprietary algorithms. [source: Greden et al. 2019; US Medicine report of Oslin et al. 2022; CPIC 2023]
Example language (illustrative): "Studies suggest these reports can help some prescribers avoid medicines that may cause problems, but the effects on recovery were modest. It's worth discussing, not a final answer."
Watch for: marketing claims stronger than the trials; the FDA has said that a link between DNA variation and antidepressant effectiveness has not been established. [source: RAPS 2018]

A5. Direct-to-consumer (DTC) PGx tests

Purpose: to respond accurately when clients buy tests themselves.
What sources say: the FDA's November 2018 communication covered both clinician-ordered and DTC tests, warning that many medication-response claims had not been reviewed and that changes based on them could harm patients; it said it would take compliance action where appropriate. The one FDA-authorised DTC PGx report (23andMe, October 2018) was described by the FDA as not determining whether a medication is appropriate. ISPG does not recommend DTC genetic testing for medical purposes in psychiatric patients or their families, citing missing informed consent and misinterpretation. [source: FDA Law Blog 2018; RAPS 2018; ISPG statement] Whether the FDA required clinical confirmation before acting on the 23andMe report was not verified in a source opened. [uncertain]
Example language (illustrative): "Some consumer results are accurate for the variants they test, but they aren't designed to choose your medicine. Your prescriber may want a clinical test before using it."
Watch for: clients who have already changed or stopped a medicine on the basis of a DTC result: this needs prompt prescriber contact (see B4).

B1. Recognising when to suggest a medication review

Purpose: to route clients to the prescriber when PGx might be relevant, without recommending a test yourself.
When to use: repeated antidepressant or antipsychotic trials with poor response or troublesome side effects (the group ISPG describes as most likely to benefit); a client asking about "DNA tests for meds"; a planned start of carbamazepine or oxcarbazepine (HLA testing is guideline- and label-relevant). [source: ISPG statement; Equetro label]
Steps: (1) Note the pattern in the client's own words (how many medicines, what happened). (2) Ask whether they have discussed it with the prescriber. (3) Suggest a review and, with consent, offer to write a brief note (B7). (4) Leave the decision about testing to the prescriber. [synthesis]
Example language (illustrative): "You've had a hard run with side effects. Would it help to ask your prescriber whether anything, including genetic testing, could inform the next step?"
Watch for: framing testing as a solution; a review is the intervention, the test is the prescriber's call. [synthesis]

B2. When a client brings test results

Purpose: to receive results respectfully, prevent misinterpretation and route them.
Steps: (1) Ask what the client understands the report to say and how they feel about it. (2) Clarify your role: you do not interpret genotypes or advise on medicines. (3) Correct obvious misreadings at a general level (A1, A4): enzyme speed, not diagnosis; one input among many. (4) Ask whether the prescriber has the report; if not, encourage sharing it and offer to help prepare questions. (5) Ask whether they have changed anything already; if yes, encourage same-day or next-day contact with the prescriber, and treat any withdrawal, worsening mood or suicidal thinking as a safety issue (B9). (6) Record only what is clinically needed (B8). [synthesis, consistent with FDA Table patient advice and ISPG statement]
Example language (illustrative): "Thank you for showing me. I can't tell you what it means for your medicines, but I can help you get it to the person who can, and help you think through your questions."
Watch for: relief ("finally an explanation") or despair ("my body rejects antidepressants"); both deserve exploration (B5). [synthesis]

B3. Psychoeducation without overclaiming

Purpose: to give accurate, proportionate information.
Key points you can safely convey (attributed): PGx results mainly predict how fast certain drugs are processed; the FDA has said a DNA link to antidepressant effectiveness is not established; serotonin-gene results are not used in CPIC guidance; large trials found modest effects of testing on recovery; other medicines, smoking and health can change how drugs are processed; some results (for example HLA-B before carbamazepine) are named in official labelling. [source: CPIC 2023; RAPS 2018; GUIDED; PRIME Care; Equetro label]
Steps: (1) Ask what they have heard. (2) Offer one or two points, not a lecture. (3) Check understanding by asking them to say it back. (4) Point them to the prescriber for anything specific to their medicines. [synthesis]
Example language (illustrative): "The test is about how your body handles certain medicines, not about whether you can get better. People with every result recover."
Watch for: your own enthusiasm or scepticism leaking in; stay with what the sources say. [synthesis]

B4. Supporting informed consent and shared decision-making

Purpose: to help clients make and own decisions with their prescriber.
Steps: (1) Help the client list questions: why test now, which genes, what could change, is this a clinical or consumer test, cost and insurance coverage, who sees the data. (2) Rehearse raising them. (3) Explore values (wish for certainty, fear of more trials, privacy). (4) Afterwards, review what the prescriber said and any remaining confusion. [synthesis; ISPG notes counselling can help people decide whether to test]
Example language (illustrative): "Let's write down what you want to ask, so you leave the appointment with answers rather than more worries."
Watch for: decisions made under pressure from marketing or family; slow down and return to values. [synthesis]

B5. Meaning-making about genetic results

Purpose: to address beliefs and emotions a result can trigger.
Steps: (1) Elicit the story the client attaches to the result ("broken", "faulty genes", "proof it isn't my fault"). (2) Separate the factual content (drug handling) from the meaning. (3) Address hopelessness, stigma or self-blame with your usual modality (cognitive, narrative, acceptance-based). (4) If the client worries about relatives, note that results may be relevant to family members and suggest they ask the prescriber or a genetic counsellor. [synthesis]
Example language (illustrative): "It sounds as if 'poor metaboliser' landed as 'poor patient'. What it describes is an enzyme, not you."
Watch for: a result being used to justify giving up on all treatment, including therapy. [synthesis]

B6. Monitoring and passing on side effects and warning signs

Purpose: to make sure the prescriber hears about adverse effects promptly.
Steps: (1) Ask routinely about side effects after any medication change. (2) For carbamazepine, rash, blistering, mouth or eye soreness, fever, sore throat or unusual bleeding or bruising warrant urgent prescriber or medical contact, given the boxed warnings for severe skin reactions and blood disorders. (3) For any client, sudden worsening, agitation or new suicidal thinking after a change goes into your safety procedure and to the prescriber. (4) With consent, pass observations on in writing. [source: Equetro label for carbamazepine warnings; synthesis for the routing]
Example language (illustrative): "A new rash on this medicine is something your prescriber needs to know about today. Can we call now?"
Watch for: clients minimising side effects because "the test said this drug suits me"; the FDA says genotyping does not replace vigilance. [source: FDA Table]

B7. Collaborative communication with the prescriber

Purpose: to share relevant psychological observations without straying into prescribing.
Steps: (1) Get written consent. (2) Write briefly: what the client reports (medicines tried in their words, side effects, functioning, current risk), that the client has a PGx report and wishes to discuss it, and the client's questions. (3) Avoid recommendations about drugs or tests. (4) Agree how urgent issues will be communicated. [synthesis]
Example language (illustrative, letter): "Ms A has brought a pharmacogenomic report to therapy and would like to discuss it with you. She reports nausea and agitation on two previous medicines. Current risk is low; she has a safety plan."
Watch for: letters that read as advocacy for a test or a drug. [synthesis]

B8. Genetic privacy conversation and record-keeping

Purpose: to protect sensitive data the client may not realise they are sharing.
What sources say: in the US, GINA (2008) bars use of genetic information by health insurers and by employers with 15 or more employees, but not by life, disability or long-term care insurers, and its employment protections do not apply to the US military; under HIPAA genetic information is health information. Consumer data can change hands: 23andMe went through bankruptcy and was sold in 2025 to TTAM Research Institute, a nonprofit formed by its founder, with state attorneys general securing deletion rights and audits; the Minnesota Attorney General advised customers how to delete their data. [source: genome.gov GINA page; Minnesota AG 2025] Other countries' protections were not checked. [uncertain]
Steps: (1) Before a client buys a consumer test, invite them to read who can access, keep, sell or share the data. (2) Mention that coverage gaps exist (for example life insurance in the US). (3) Ask only for the information you need; do not file copies of full reports in therapy notes unless needed. (4) Do not send reports by insecure channels. [synthesis]
Example language (illustrative): "Genetic results tell something about your relatives too. Before you order a kit, it's worth checking what the company can do with your data."
Watch for: anxiety about past uploads; help the client find the company's deletion route rather than advising on legal matters. [synthesis]

B9. Adherence, ambivalence and safety

Purpose: to explore ambivalence while never advising a change.
Steps: (1) Use reflective listening to explore what the result means for their willingness to take medicine. (2) If they are thinking of stopping or changing, acknowledge the reasons and route the decision to the prescriber, ideally before acting. (3) If they have stopped suddenly, ask about withdrawal-type symptoms and mood, and encourage prompt prescriber contact. (4) Update the safety plan when mood worsens during repeated medication trials; include means safety through storage or removal where indicated. [synthesis; FDA Table advises patients not to change medicines without their prescriber]
Example language (illustrative): "Part of you wants to stop because of the report, and part of you noticed it was helping. That's exactly what to take to your prescriber before anything changes."
Watch for: hopelessness after "failed" medicines; assess suicide risk directly and act on standard procedures. [synthesis]

B10. Integrating psychotherapy across the process

Purpose: to keep psychological treatment central whatever the medication plan.
Steps: (1) Before testing: continue the agreed therapy and process expectations. (2) While waiting for results or a medication change: keep goals active and avoid "on hold" thinking. (3) After a change: track mood and functioning with your usual measures and share with consent. (4) Over time: help the client weigh medication, therapy and lifestyle factors in their own recovery. [synthesis]
Example language (illustrative): "Whatever the result shows, the work we're doing on sleep and avoidance keeps going."
Watch for: attributing all progress or setbacks to the medicine choice. [synthesis]

5. Handling problems

The sources address the evidence and the regulatory position, not therapy-room difficulties; everything below is [synthesis] unless a source is named.

Client wants you to interpret the report or say which drug is "right". Restate your role kindly, name who can answer (prescriber, clinical pharmacist, genetic counsellor) and help them prepare questions (B4). Illustrative: "I'd be guessing, and with medicines guessing isn't safe. Let's get this to someone who can read it properly."
Client has stopped or changed a medicine because of a result. Do not endorse or reverse it. Explore what happened, check mood, withdrawal-type symptoms and risk, and encourage prompt prescriber contact; with consent, contact the prescriber yourself. The FDA advises patients not to change medicines without their prescriber. [source: FDA Table; synthesis]
Client is demoralised ("my genes mean nothing will work"). Correct the factual error (B3), then work with hopelessness using your modality; assess suicide risk and update the safety plan.
Client is over-invested ("the test will fix this"). Validate the hope, share the modest trial findings in plain words, and keep therapy goals active. [source: GUIDED; PRIME Care]
Prescriber declines testing, and the client feels dismissed. Normalise that guidelines do not set who should be tested and that a declined test is a clinical judgement; support the client to ask for the reasoning or a second opinion if they wish. [source: ISPG statement; CPIC 2023]
Two reports disagree, or a result changed. Explain that laboratories test different variants and that other medicines can change actual enzyme activity; route the discrepancy to the prescriber. [source: CPIC 2023]
Family conflict about the result (a parent wants to test a teenager, relatives fear implications). Hold a family conversation focused on consent, privacy and who will act on the result; note that paediatric evidence is limited outside a few drugs and route the decision to the prescriber. [source: CPIC 2023 for paediatric evidence; synthesis]
Privacy worry after a consumer upload. Help the client find the company's deletion route and the relevant consumer advice; do not give legal advice. [source: Minnesota AG 2025]
Non-adherence framed as "my genes". Explore other reasons (side effects, stigma, cost, beliefs) with motivational interviewing and route all of it to the prescriber. [synthesis]
Rupture: client feels you doubt their test. Acknowledge the impression, repair, and separate your respect for their effort from your limits of competence.
Your own competence drift. If you notice yourself giving drug-specific views, return to role and seek supervision; consider brief training from a clinical pharmacist or genetic counsellor on reading reports at a general level.
Missing ancestry representation. If a client asks whether results apply to them, say that most research participants were of European or East Asian ancestry and that the prescriber can weigh this. [source: CPIC 2023]

Sources

Sources:

Further reading (not drawn on): the FDA November 2018 safety communication itself (described here via the FDA Law Blog and RAPS reports); the ClinicalTrials.gov record for GUIDED (NCT02109939). [uncertain]

← Previous · IX-011Addiction pharmacotherapy Next · IX-013 →Anti-inflammatory and psychobiotic augmentation (experimental)