CYP450 interactions in psychiatric prescribing

How to think about this

CYP interactions get overtaught and undermanaged. Trainees memorize dozens of tables and then either freeze at the pharmacy line or ignore the whole thing. Neither is right. The working approach is this. When you add or change a medication, ask two questions. Is this drug a strong inhibitor or inducer of a CYP that anything else on the list uses? Is anything on the list a substrate with a narrow therapeutic window? If the answer to both is yes, you look it up. If not, you move on.

Substrate status alone rarely matters. Everyone metabolizes through these enzymes. What matters is when a substrate meets a potent modifier of its clearance, or when a patient's genotype makes them a poor metabolizer of a drug you're relying on.

Potency shorthand: strong inhibitors raise substrate AUC by five-fold or more, moderate by two to five-fold, weak by 1.25 to two-fold. Strong inducers drop AUC by 80 percent or more.

CYP2D6

This is the enzyme with the most clinically relevant psychiatric drug interactions, and the one with the most useful pharmacogenetic testing. About 5 to 10 percent of people of European ancestry are poor metabolizers. Ultrarapid metabolizers matter too, mostly on the codeine side.

CYP3A4

The busiest enzyme in the body. It handles a huge chunk of drug metabolism, which means both a lot of substrates and a lot of ways to knock the balance off. Most 3A4 interactions come from strong inhibitors or inducers rather than substrate competition.

CYP1A2

Small number of psychiatric drugs, but two of them (clozapine and olanzapine) sit at the top of the "levels matter" list. That makes 1A2 punch above its weight.

CYP2C19

Matters mostly for citalopram and escitalopram, and for the occasional clopidogrel interaction question.

CYP2C9

Less relevant in day-to-day psych, but two interactions matter: warfarin and phenytoin.

CYP2B6

Small but relevant, mostly for bupropion, methadone, and ketamine.

The five moves that catch people off guard

These are the specific interactions that actually cause harm on inpatient units and in outpatient clinics. Every psychiatrist should know them cold.

Fluvoxamine plus clozapine. Combined 1A2 and 2C19 inhibition can raise clozapine levels 5 to 10-fold. Some clinicians use it deliberately at very low fluvoxamine doses to reduce clozapine dosing burden. Fine, but only with levels, and only if you know what you're doing. Otherwise, don't combine them.

Fluoxetine, paroxetine, or bupropion plus tamoxifen. Tamoxifen needs 2D6 to become endoxifen, its active metabolite. Strong 2D6 inhibitors reduce endoxifen substantially, and the observational data suggest higher breast cancer recurrence rates. If your patient is on tamoxifen, use sertraline, escitalopram, citalopram, venlafaxine, or mirtazapine instead. Oncologists know this. Some psychiatrists don't.

Fluoxetine or paroxetine plus a TCA. 2D6 inhibition raises nortriptyline or desipramine levels several-fold. Anticholinergic and cardiac toxicity follow. If you need the combination (rare, but it happens for chronic pain), check a TCA level.

Carbamazepine and almost anything. It induces 3A4, 2C9, 2C19, 1A2, and UGTs. It drops OCP levels enough to cause pregnancy, drops quetiapine, aripiprazole, lamotrigine, and warfarin. It also autoinduces its own metabolism, so levels drift down over the first month. Someone stable on carbamazepine who suddenly runs out will re-induction differently on restart, and the level you get won't be the level you had.

Smoking status change plus clozapine or olanzapine. A patient admitted to a smoke-free unit stops smoking abruptly. 1A2 de-induces over about a week. Clozapine levels rise. Toxicity can appear before you've thought to check. Whenever smoking status changes, plan a clozapine level.

Bonus sixth move worth knowing: paroxetine or fluoxetine plus codeine or tramadol. 2D6 blockade prevents conversion to the active opioid metabolite. Your postop patient gets no pain relief and looks like they're drug-seeking. They aren't. Use hydromorphone, oxycodone (partial 2D6, but works better), or morphine instead.

Pharmacogenetic testing: what actually changes prescribing

The clinical yield of routine 2D6/2C19 testing is modest but real. Where it actually changes decisions:

• 2C19 poor metabolizers on citalopram: the 20 mg cap is a real safety threshold. Escitalopram is a milder version of the same concern.
• 2D6 poor metabolizers on atomoxetine: dose 10-fold higher exposure, so titrate by side effects rather than the standard mg/kg approach.
• 2D6 poor metabolizers on TCAs: check levels early. Don't push doses blindly.
• 2D6 status for aripiprazole and pimozide (label-driven dose caps).
• 2D6 ultrarapid metabolizers and codeine: don't prescribe codeine in kids, and be cautious in adults.

Outside those situations, treat pharmacogenetic reports as one input among many. The Genesight-style broad panels overpromise. The CPIC guidelines are the honest reference for what the evidence supports.

Common questions

Does fluoxetine really need a 5-week washout before an MAOI? Yes. Fluoxetine's active metabolite norfluoxetine has a half-life of one to two weeks, so it hangs around for a month or more after the last dose. Starting an MAOI too soon risks serotonin syndrome, which is genuinely dangerous. Five weeks is the standard, and it isn't a suggestion. For the other SSRIs and SNRIs, two weeks is the usual washout. Going the other way (stopping an MAOI, starting an SSRI) is 14 days for phenelzine and tranylcypromine to let MAO enzyme regenerate.

Do I need to check smoking status before starting clozapine? Yes, and you need to keep checking it. Heavy smoking (usually 10 or more cigarettes a day) induces CYP1A2 enough to drop clozapine levels 30 to 50 percent. That affects the dose you'll end up at. More importantly, when a patient stops smoking (hospitalization, quit attempt, illness), levels can rise sharply within a week. That's when toxicity shows up. A level within a week of any smoking status change is reasonable practice.

What happens if a patient on carbamazepine misses doses? Carbamazepine autoinduces its own metabolism over about a month. Once induction is established, missing doses drops the level fast. Restarting at the previous dose after a lapse of a week or more can produce a bigger level than expected because the induction fades, but the exact kinetics vary. This is one reason mood stabilizer levels matter more than they get credit for. When restarting after a significant gap, start lower and retitrate.

Is grapefruit really a problem, or is that overstated? For most drugs, it's overstated. For a specific short list, it's a real problem. Buspirone AUC goes up about 9-fold with grapefruit juice. Simvastatin, felodipine, and cyclosporine all have big effects. In psychiatry, watch it with buspirone, quetiapine, some benzos, and carbamazepine. The intestinal 3A4 effect lasts roughly a day per dose of juice, so telling patients to just take it separately doesn't fix it. Better to switch fruit.

How much does CYP2D6 genotype actually change my prescribing? For the average patient starting an SSRI, not much. For atomoxetine, TCAs, and pimozide, quite a bit. For codeine and tramadol, a lot. The honest answer is that I check pharmacogenetics selectively, when the patient hasn't tolerated typical doses of multiple 2D6 substrates, when I'm about to use a narrow-window drug, or when the patient asks and it might change the decision. Routine broad-panel testing at first visit hasn't earned its keep in outcomes trials.

Sources

1. FDA Prescribing Information for citalopram (Celexa), aripiprazole (Abilify), atomoxetine (Strattera), pimozide (Orap), vortioxetine (Trintellix), lurasidone (Latuda), cariprazine (Vraylar), clozapine (Clozaril).
2. Flockhart DA. Drug Interactions: Cytochrome P450 Drug Interaction Table. Indiana University School of Medicine, Division of Clinical Pharmacology.
3. DailyMed (National Library of Medicine) for full current labels.
4. Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines for CYP2D6, CYP2C19, and CYP2B6 substrates.
5. Cipriani A, et al. Comparative efficacy and acceptability of 21 antidepressants for the acute treatment of adults with major depressive disorder. Lancet 2018.
6. American Psychiatric Association. Practice Guideline for the Treatment of Patients with Major Depressive Disorder, third edition.
7. National Institute for Health and Care Excellence. Depression in adults: treatment and management. NICE guideline NG222.
8. de Leon J. Clozapine and smoking. Multiple review articles, especially those addressing 1A2 induction and abrupt smoking cessation.

Reviewed against current guidelines as of June 8, 2026. This is not medical advice.