- Design
- observational hospital cohort with Cox regression adjusted for clinical covariates and CYP2D6 inhibitors
- Population
- 996 patients receiving metoprolol tartrate, median follow-up 101.4 months
- Primary outcome
- time to death from enrolment
- Effect
- higher metabolizer status HR 0.82 (95% CI 0.67-0.99, p=0.04); drug concentration not associated after adjustment
Nine hundred and ninety-six patients taking metoprolol tartrate in the Montreal Heart Institute cohort had a single random plasma concentration measured by mass spectrometry and CYP2D6 metabolizer status assigned by standard classification. Over a median 101.4 months, 24.3% died.
Higher metabolizer status was associated with lower mortality, hazard ratio 0.82 (95% CI 0.67-0.99, p = 0.04), after adjustment for age, sex, cardiovascular history, concomitant drugs and CYP2D6 inhibitors. The measured metoprolol concentration, by contrast, was no longer associated with mortality once confounders were included.
That dissociation is the whole interest of the paper and the reason to be careful with it. If the genotype predicted survival through drug exposure, the concentration should have carried the signal too - and it did not. Either the single random level is too noisy a measure of exposure to detect it, which is likely, or CYP2D6 status is marking something other than metoprolol handling. The result is a hazard ratio whose upper confidence limit is 0.99, from one cohort, and the authors call for larger studies. Nothing here supports genotyping before prescribing metoprolol. What it does support is remembering that a poor metaboliser on a CYP2D6 inhibitor - fluoxetine, paroxetine, bupropion, terbinafine - is getting considerably more beta blockade than the dose suggests.
- Do not genotype before prescribing metoprolol; this evidence does not support it.
- Check for CYP2D6 inhibitors - fluoxetine, paroxetine, bupropion, terbinafine - when metoprolol is poorly tolerated.
- Bradycardia or fatigue after adding an antidepressant is a drug interaction until proven otherwise.
- Consider a beta blocker cleared differently, such as bisoprolol or atenolol, where an interacting drug cannot be changed.
- A single random plasma level is a weak measure of metoprolol exposure and should not guide dosing.
Why it matters
It puts a survival number on a pharmacogene most prescribers treat as a pharmacokinetic curiosity rather than a clinical variable.
Don't overread it
An observational association at the edge of significance, unsupported by the drug level it should have acted through.
The statistics, in plain English
A hazard ratio of 0.82 with an upper confidence limit of 0.99 sits right at the edge of conventional significance, in a single observational cohort - the kind of result that often shrinks on replication. The more informative finding is negative: the drug concentration lost its association after adjustment, which undercuts the pharmacokinetic explanation the genotype result seems to offer. When a proposed mechanism does not carry the signal, the association needs another explanation before it changes practice.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for clinical pharmacology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free