Pharmacogenomic-Guided Antidepressant Selection
A combinatorial pharmacogenomic panel came back with a clean answer for why two drugs failed on side effects — the harder question is how far that same result should reach into which drug gets tried next.
T.B., a 44-year-old warehouse operations supervisor, has been trying to treat a moderate-to-severe depressive episode for the past seven months and has struggled through two medication trials that each ended on side effects rather than lack of response. Paroxetine, started first, left him so sedated and cognitively foggy at the standard 20mg dose that he nearly lost his forklift certification for failing an alertness check at work; he stopped it after three weeks. Venlafaxine, tried next, produced nausea and dizziness severe enough at 75mg that he couldn't titrate past it. He has no other chronic illness, takes no other regular medications, and has never had a genuinely inadequate trial — both drugs were stopped for tolerability, not for a documented lack of effect.
Given the pattern — two different drug classes, both failing on what looked like exaggerated dose-dependent side effects rather than lack of efficacy — his psychiatrist ordered a combinatorial pharmacogenomic panel. The result: T.B. is a CYP2D6 poor metabolizer and a normal CYP2C19 metabolizer. Paroxetine and venlafaxine are both substantially CYP2D6-dependent for clearance, which means a poor metabolizer accumulates meaningfully higher plasma levels than a standard dose is calibrated for — a genuinely plausible explanation for why both drugs produced side effects at doses most patients tolerate without difficulty.
The panel's report goes further than that pharmacokinetic explanation, though, sorting a long list of antidepressants into color-coded 'use as directed,' 'moderate gene-drug interaction,' and 'significant gene-drug interaction' categories, implicitly presenting the whole list as a guide to which drug will work for him — not just which doses are likely to be tolerated. The evidence behind that broader claim is genuinely mixed: the pragmatic trials of combinatorial pharmacogenomic testing have shown modest benefit, concentrated mainly in avoiding drugs the report flags red, without strong evidence the color categories reliably predict which green-bin drug will actually treat his depression.
At the pharmacogenomic-results review
The CYP2D6 result actually explains something concrete here — both paroxetine and venlafaxine lean heavily on CYP2D6 for clearance, and a poor metabolizer predictably runs higher plasma concentrations at standard doses, which lines up with side effects appearing before any therapeutic ceiling was reached. That's a real, mechanistically grounded finding.
What the genotype does not do is tell us which drug will treat his depression. Metabolizer status is a pharmacokinetics story — how much drug reaches his bloodstream at a given dose — not a pharmacodynamics story about whether serotonergic or noradrenergic augmentation will actually resolve his symptoms. Those are different questions, and the report's color-coding blurs them into one.
I hear that distinction, but in practice this report is what I have in front of me, and its green-bin list is genuinely useful as a starting point when I don't have anything better to sequence trials by — especially for a patient who has already burned two trials on side effects he didn't need to have.
Then let's use it for exactly the part it's actually validated for. Sertraline is metabolized less dependently on CYP2D6 than paroxetine or venlafaxine, so a poor-metabolizer result doesn't predict the same accumulation problem — that's a real, defensible reason to try it next, independent of the report's broader efficacy claim. What I wouldn't do is treat the green-versus-red bin as evidence sertraline will work better than any other green-listed drug; it only tells us it's less likely to fail the same way the last two did.
Sertraline was started at 50mg, chosen specifically for its comparatively CYP2D6-independent clearance rather than its position on the report's color-coded list, with a plan to titrate on the standard schedule and reassess at four weeks.
The panel's broader claim — that its color categories predict which drug will actually be effective for him, not just which doses are likely to be tolerated — was explicitly not treated as settled, and nothing about today's choice depended on it being true.
If sertraline also fails on efficacy rather than tolerability, the team agreed that would say something real about the limits of what this genotype can predict for him — and would not, on its own, be read as license to keep working down the report's list drug by drug.