Pharmacogenomic testing for psych meds

GeneSight and the other pharmacogenomic tests (cheek-swab panels that look at your liver-enzyme genetics to suggest which psych medications you’ll handle well or badly) promise that a single test can tell us which psychiatric medication will work for you. The marketing is everywhere. Primary care doctors order it. Patients ask for it by name after watching a TikTok about it. The evidence base does not actually support most of the claims being made, and I’d rather be honest with you about that before anyone takes your money or your insurance company’s money for the test.

The basic science underneath is real. Some of us have genetic variants that affect how we metabolize medications, meaning how fast or slow our liver breaks them down. CYP2D6 and CYP2C19 (two of the liver enzymes that process most psychiatric drugs) are the big ones. If you’re a slow metabolizer of CYP2D6, you can build up high blood levels of certain medications even at what looks like a normal dose, which means you feel side effects that other people don’t because you’re effectively running double the dose. If you’re an ultra-rapid metabolizer, you can clear medications so fast that the standard dose barely registers and the prescriber is wondering why nothing’s working. All of this is real. The tests measure it accurately.

The gap between what’s measured and what helps

The problem is that knowing your CYP2D6 status doesn’t predict whether a specific medication will work for your depression. It only predicts how your body processes that medication. Whether the medication actually helps depends on a hundred other things we can’t measure with a cheek swab… what your receptors look like, what’s already going on chemically in your brain, what your specific symptom picture is, what your life situation is, what kind of depression you actually have. The test is reading one chapter of a book that has fifteen chapters.

The studies that tried to show pharmacogenomic-guided prescribing led to better outcomes have been mixed at best. The big ones funded by the testing companies themselves showed modest benefit, which is the kind of result you’d expect from a study funded by the company that profits from the conclusion. The independent ones, including the largest randomized trial in the field, showed no clinically meaningful advantage over treatment as usual. Which is roughly what most of the prescribers I respect have been saying out loud for years, but you can’t sell a cheek swab on “the evidence is mixed and your prescriber is probably fine without it.”

The most defensible use is in patients who have already had multiple bad reactions to medications or unusual responses at standard doses. There, knowing your metabolizer status can actually save you from being prescribed the wrong dose of the next one. For most people walking in with a first episode of depression, the test doesn’t change what the prescriber would do.

Where the test earns its keep

Three scenarios where the swab actually contributes something.

First, a patient has tried three or four antidepressants and either didn’t respond at standard doses or had weird side effects that didn’t fit the usual profile. The test sometimes shows a metabolizer pattern that explains the whole history… ultra-rapid metabolism of CYP2D6 means the SSRIs at standard doses were running blood levels half what they should have been, which is why nothing worked. That changes which medication we pick next, or at least changes the dose we start it at.

Second, a patient has had a clear bad reaction to a specific medication, particularly something like nortriptyline (an older antidepressant where blood levels really matter for both safety and effectiveness) or paroxetine, and we want to know if the whole CYP2D6-dependent class needs to come off the table or whether the problem was just that one drug.

Third, the patient is genuinely anxious about taking any medication and wants any data they can get to commit to the treatment plan. Sometimes the test results, even if they don’t dramatically change the clinical choice, give the patient enough of a story about their own body that they’re willing to actually take the next prescription. That’s not nothing. The medication that the patient actually takes works better than the one they don’t.

What it doesn’t do

First-line prescribing. The test doesn’t tell anyone whether Zoloft or Lexapro is more likely to lift your depression. It tells you whether you’ll metabolize either of them at a normal or weird rate. Most patients metabolize both fine. The test color-codes them green, yellow, or red based on metabolism, not effectiveness, and the green-yellow-red layout fools people into thinking the green options are the ones that will work and the red ones won’t, which is just not what the test is measuring.

Predicting whether a specific drug will work. A green drug can fail you. A red drug, if your previous prescriber was actually paying attention to dosing and side effects, might have been completely fine. The categories are not “effective for you” and “not effective for you.” They’re “your body processes this at a normal rate” and “your body processes this at a weird rate,” which is a much narrower claim than the report makes it look.

Justifying a particular medication choice to insurance. That’s not how this works. Insurance is not going to approve a $1,200 antidepressant because a swab said it was green.

The cost picture

GeneSight bills insurance up to about $2,000. Many plans cover it, often with significant out-of-pocket. Medicare covers it under specific circumstances. Some patients end up with surprise bills in the hundreds because the lab’s coverage agreement with their plan wasn’t what the patient was told it would be. Always check coverage in writing before you swab, because “your insurance will probably cover it” is not the same as “your insurance will definitely cover it” and the difference shows up on a bill three months later.

The other tests in this space (Genecept, GenoMind, a handful of others) are similar in price and roughly comparable on how much they actually contribute to a decent prescriber’s decision, which is modest in both directions. They’re not scams, and they’re also not the magic decoder ring the marketing makes them look like.

The pattern this is for

Say you’ve got a patient where the test sometimes actually changes the plan – somebody who’s been through three or four antidepressant trials, none of them worked, several of them caused side effects that the prescriber chalked up to “well, some people just can’t tolerate SSRIs,” and the patient walks in convinced he’s somehow broken differently from everyone else. Run the test, find out he’s a slow metabolizer of CYP2C19, which means his Lexapro and Zoloft were running blood levels probably twice what a typical metabolizer would have. The “standard dose” he was on was effectively a double dose for him. That history makes sense now. We pick something processed through a different pathway (Wellbutrin, for instance), titrate carefully, and the next trial has a real shot.

The test probably wasn’t needed to figure out Wellbutrin was the next move. A careful history and a prescriber paying attention would have gotten to the same place. What the test did was give the patient a story about his body that helped him stop blaming himself for the previous failures. For that patient, the story-about-the-body piece is actually doing real work, because the next attempt requires him to commit to taking the medication consistently for six to eight weeks, and somebody who feels broken doesn’t commit to anything for six to eight weeks.

The test categorizes drugs by metabolism, not effectiveness. A green drug can fail you. A red drug might have been fine. The report’s design fools patients into thinking otherwise.

The marketing problem

The testing companies have done aggressive marketing to primary care doctors, who order these tests for patients who don’t really need them. The reports come back with color-coded recommendations that look like a definitive answer but aren’t, and patients end up convinced certain medications “won’t work for them” when the test only said their metabolism for those drugs is on the unusual end. The patient then refuses a perfectly reasonable medication because of a yellow flag that just meant “use a different dose.”

That’s not the patient’s fault. That’s the report’s fault for designing itself to look more authoritative than it is. A traffic-light layout signals “do this, don’t do that” to the human brain, and the brain reads it that way even when the fine print says something more modest.

Where the autonomy stance lands

If you’ve heard the trade-offs and you want the test, you get the test. I’m a provider, not a parent. If your insurance covers it and you’re prepared for the result to be less decisive than the marketing promised, fine. If your insurance won’t cover it and the out-of-pocket is hundreds of dollars and you’ve been on one antidepressant that didn’t dramatically work in eight weeks, my honest take is that the money would do more for you spent on a real second medication trial than on the swab. But that’s the honest take, not a rule.

Bottom line

Pharmacogenomic testing is a useful tool in a narrow subset of patients, mostly people who’ve had multiple medication failures or weird reactions. It’s an oversold tool for most other patients, where it doesn’t really change what the prescriber would do anyway. The science underneath is real but limited. The marketing has run way ahead of the evidence. If your prescriber is recommending it after your first or second medication trial, ask what specifically they expect the result to change in your treatment plan. If you’ve been through five medications and nothing’s worked, it might be worth running. Like most tests in medicine, the question is whether the result will change the plan, and most of the time, with this one, it won’t.

Sources

1. Greden JF, Parikh SV, Rothschild AJ, et al. Impact of pharmacogenomics on clinical outcomes in major depressive disorder in the GUIDED trial. J Psychiatr Res. 2019;111:59-67. PMID 30677646.
2. Oslin DW, Lynch KG, Shih MC, et al. Effect of Pharmacogenomic Testing for Drug-Gene Interactions on Medication Selection and Remission of Symptoms in Major Depressive Disorder. JAMA. 2022;328(2):151-161. PMID 35819423.
3. Bousman CA, Bengesser SA, Aitchison KJ, et al. Review and Consensus on Pharmacogenomic Testing in Psychiatry. Pharmacopsychiatry. 2021;54(1):5-17. PMID 33147643.