Variant in a Taste Receptor Locus Tied to Changes in the Use of Insomnia Medication

I have a new first-author paper out in Biological Psychiatry Global Open Science: Variant in a taste receptor locus tied to changes in the use of insomnia medication. This one took a long time to get right, and it’s the most useful thing I’ve worked on. A clinician could act on the results today.

The problem with zopiclone

Zopiclone is one of the most commonly prescribed drugs for insomnia. It works, but it comes with a side effect patients frequently complain about: a persistent bitter or metallic taste, sometimes described as a taste of copper that lingers long after the pill is taken. For many patients this is tolerable. For others it’s bad enough that they stop taking the drug or ask for something different.

Clinicians haven’t been sure why some patients are so much more bothered by it than others. The taste experience is subjective, hard to measure systematically, and easy to dismiss as individual variation. But variation in biology usually has a genetic cause, and that’s where this study starts.

A genome-wide look at medication switching

We used Icelandic prescription data from deCODE to identify people who had been prescribed zopiclone and then switched to a different sleep medication, specifically zolpidem, which doesn’t carry the same taste side effect. Switching is a noisy signal, but at the population scale of Icelandic data it becomes informative. If there’s a genetic variant that reliably predicts who stops taking zopiclone and moves on, that’s telling you something real about drug tolerability.

We found exactly that. A variant near a taste receptor gene is associated with switching away from zopiclone. Taste receptors vary across individuals, and those differences appear to translate into real differences in how unpleasant the drug’s aftertaste is. People who carry the variant are more likely to find it intolerable enough to switch.

Why this one matters

Most pharmacogenomics findings live in a gray area. They tell you about risk or metabolism, but the clinical implications tend to be indirect. This one isn’t. If a patient carries a variant that predicts poor tolerability of zopiclone, you have a clear alternative right away: prescribe zolpidem instead. There’s no need for a trial period of suffering through a side effect that was always going to be a problem, no waiting months for the switch to happen on its own.

For patients who are already dealing with something hard, this makes a bigger difference. Cancer patients come to mind. Chemotherapy already brings nausea, metallic tastes, sensory distortions. For someone going through that, a better-tolerated sleep medication is not a minor footnote. If a genetic test could tell the prescribing physician upfront which patients will find zopiclone unpleasant, those patients could start on zolpidem from day one.

Routine pharmacogenomic prescribing isn’t here yet, but groups across the field are building the infrastructure. Results like this push it closer.

A long time coming

This paper went through many iterations. Prescription switching is a messy endpoint: people switch medications for all kinds of reasons, and filtering the taste-related signal from the noise required a lot of careful work on the phenotype definition. I’m proud of how the analysis came together, and I think it’s a good example of what large-scale population data, the kind deCODE has been building for decades, makes possible. The variant would have been nearly invisible in a smaller cohort.

The mechanism is simple: a gene that shapes how something tastes, a drug with a distinctive aftertaste, and a patient who has to decide whether to keep taking it.