CYP2C19 and Your SSRIs: When Citalopram, Escitalopram or Sertraline Needs a Different Dose

A report comes back: CYP2C19 poor metabolizer. Your patient is on escitalopram and doing fine. Do you change anything? Now picture the same patient on citalopram instead. CPIC’s advice for a poor metabolizer is much the same for both — but the dose ceilings are not: citalopram carries an FDA 20 mg/day limit in poor metabolizers that escitalopram’s label never sets, and the DPWG caps differ too. Same gene, same enzyme — yet the ceiling, and sometimes which body acts, change with the drug. A CYP2C19 result is only useful once you can say, for the exact drug in front of you, whether it calls for a lower dose or slower titration, an alternative, or no genotype-driven change — and who says so.

Why does CYP2C19 matter for these three drugs?

Because it strongly shapes drug exposure — alongside other metabolic pathways, interacting drugs and patient factors. Citalopram and escitalopram are extensively metabolized by CYP2C19 to much less potent metabolites, so a poor metabolizer tends to run higher levels and an ultrarapid metabolizer lower ones.

For citalopram and escitalopram, that higher exposure in poor metabolizers raises the risk of side effects — and, for citalopram, a QT concern specific enough that the FDA label sets a hard genotype-based dose limit. (Escitalopram’s label documents an exposure–QT relationship too, but sets no genotype-specific cap.) Lower exposure in ultrarapid metabolizers can lower the chance of benefit, though it doesn’t determine an individual’s response. Sertraline is handled by several enzymes — CYP2C19, CYP2B6, CYP2D6 and others — with CYP2C19 the major pathway and no evidence supporting a CYP2D6-guided dose adjustment, which is exactly why the sertraline recommendations hang on CYP2C19 and CYP2B6, not CYP2D6.

One caution before any number: the reported phenotype is a genotype-based prediction. Interacting medicines and other clinical factors can shift functional enzyme activity (phenoconversion), so a CYP2C19 call is read together with the current medication list, not in isolation.

Citalopram and escitalopram: what each body says

CPIC gives one shared table for both. The practical split is: poor metabolizers get a lower dose or a switch; intermediate metabolizers get a cautious standard start; ultrarapid metabolizers get a switch considered. Then the drug-specific label and DPWG caps layer on.

Note: for citalopram, the FDA 20 mg/day maximum also applies — regardless of CYP2C19 phenotype — to patients over 60, those with hepatic impairment, and those taking a CYP2C19 inhibitor. A DPWG ceiling never licenses exceeding a lower applicable FDA limit. The DPWG limits shown are for oral tablets, applying the lower dose from age 65; other formulations (e.g., oral drops) may carry different DPWG limits.

Two things to hold onto. First, the citalopram 20 mg/day ceiling in poor metabolizers is a QT-safety limit set by the FDA label — and DPWG independently caps citalopram poor metabolizers too — but it is a citalopram action. It does not transfer to escitalopram, whose label sets no genotype-specific cap. Second, for rapid metabolizers CPIC does not switch pre-emptively: it considers a higher maintenance dose or a switch only after an inadequate response, and any citalopram increase stays within the label ceiling.

Why is sertraline a different question?

Because it is the one of the three where a second gene, CYP2B6, carries its own recommendation — and where the two genes have to be read together, not separately.

CPIC addresses sertraline on two genes — and only CPIC gives the CYP2B6 recommendation. Here is each gene, by phenotype:

The sertraline FDA label carries no CYP2C19 or CYP2B6 dose action.

Where CPIC, DPWG and the label disagree

In three places that matter at the point of care — and merging them is the most common error.

When does a result not call for a change?

When no named body ties a genotype action to the combination, or when a stable, effective, well-tolerated regimen does not automatically require a genotype-driven change. “No genotype-driven change” is not “no reassessment”: dose, QT risk, interacting drugs and any label limit still apply. An explicit "standard starting dose" is itself an answer — read for context, don't invent an action.

The clearest example is sertraline in ultrarapid or normal metabolizers: CPIC advises the standard starting dose, DPWG no action, and the label nothing. The metabolizer call is read for context — and if a CYP2B6 result is also reported, read with the combined table — not acted on. And a stable, effective, well-tolerated regimen does not automatically require a genotype-driven change, though dose, QT risk, interacting drugs and any label limit are still reviewed — a citalopram poor metabolizer above 20 mg/day is over the FDA cap even if the regimen feels settled. Knowing when a result should change nothing is half of what it is good for. To close the loop on the patient at the top: a stable, well-tolerated escitalopram regimen with a new CYP2C19 poor-metabolizer result warrants a regimen review — exposure is higher, and escitalopram still carries a CPIC recommendation — but a working, well-tolerated treatment is not switched on the genotype alone.