“The Gene Test Chose This Medication” Is Not a Prescribing Decision

The patient below is a fictional composite. No individual is depicted. The article does not reproduce a complete assessment or recommend ordering, continuing, stopping, or changing any test or treatment.

The referral says, "Pharmacogenomic testing recommends changing the antidepressant."

The patient is 41 and has recurrent major depression. She previously stopped one medication because of nausea and another because it intensified anxiety. A third helped for several years before the current episode.

A multigene report places the previously helpful medication in a caution category. A medication she has never taken appears in a favorable category. The report uses colors, footnotes, and a short explanation of predicted metabolism.

She reads the page as a verdict. The favorable medication is the one her genes say will work. The caution medication is now unsafe. The referral asks psychiatry to follow the result.

The report may contain useful information.

The inference is the error.

Pharmacogenomic testing can identify genetic variants associated with how a person metabolizes certain medications. For some gene-drug pairs, that information can support a change in dose, closer monitoring, or selection of an alternative.

That is not the same as predicting which medication will treat this patient's depression.

A prescribing decision has to integrate at least four kinds of evidence:

The diagnosis and treatment target. What condition is being treated, which symptoms and functions matter now, and what alternative explanations remain?

The patient's treatment history. What was tried, at what dose and duration, with what adherence, benefit, adverse effects, and reason for stopping?

The medication in this patient. What do comorbidities, concurrent medications, pregnancy status, organ function, age, substance use, cost, access, and preference change?

The gene-drug evidence. Does the result concern metabolism, blood concentration, adverse effects, dose, or clinical response, and how strong is that evidence for this exact drug and variant?

The report contributes to the fourth category. It does not fill in the first three.

A randomized comparative-effectiveness study published online August 10, 2026 tested whether a proprietary combinatorial pharmacogenomic report improved outcomes when added to guideline-informed depression treatment.

The study enrolled 201 people with major depressive disorder. Clinicians in one group received GeneSight results within two business days and used them alongside guideline-informed treatment. The comparison group received guideline-informed treatment without the report. Participants were followed for twelve months.

Both groups improved in average well-being. The pharmacogenomic-plus-guideline group was not superior to guideline-informed treatment alone on the primary well-being outcome or any reported secondary outcome. The result was not modified by baseline depression severity, number of previously failed depression medications, or the presence of a comorbid condition.

That finding matters. It does not prove that every pharmacogenomic test is useless, that no individual can benefit, or that every gene-drug association lacks clinical value. It tested one proprietary combinatorial approach against an active, guideline-informed comparator in one study population.

The authors offered two possible interpretations: guideline-informed treatment may have created a high comparator ceiling, or the pharmacogenomic report may not have produced better outcomes.

The disciplined conclusion is smaller than either slogan. In this trial, adding that report did not improve the measured outcomes beyond guideline-informed care.

The phrase "pharmacogenomic guidance" can conceal three different questions.

This is a pharmacokinetic question. A genetic variant may be associated with faster or slower enzyme activity and therefore with a different drug concentration at the same dose.

That can be clinically important. It still does not by itself establish whether the patient will respond, whether an adverse effect will occur, or whether another medication is the better treatment.

FDA's pharmacogenetic-associations table makes this distinction explicit. Some listed associations support therapeutic-management recommendations. Others indicate a possible effect on safety or response. A third group concerns pharmacokinetic properties only.

The categories are not interchangeable.

The Clinical Pharmacogenetics Implementation Consortium, or CPIC, publishes peer-reviewed guidance for interpreting existing genetic results. Its 2023 serotonin-reuptake-inhibitor guideline addresses CYP2D6, CYP2C19, CYP2B6, SLC6A4, and HTR2A in relation to selected antidepressants.

The scope matters. CPIC says the document is intended to interpret existing genotype results and guide dosing or selection where the evidence supports it. Deciding whom to test, which test to order, and whether testing is cost-effective lies outside that guideline.

An interpretation guideline is not automatically an ordering guideline.

This is a clinical-utility question. It cannot be answered by showing that a genotype affects metabolism. It requires evidence that using the result changes decisions in a way that improves outcomes that matter to patients.

The 2026 comparative-effectiveness trial belongs to this third question. The FDA table and CPIC guidance often address the first two. Treating them as one claim makes the evidence appear more complete than it is.

FDA states that including a gene-drug interaction in its pharmacogenetic table does not necessarily mean the agency advocates testing before prescribing that medication, unless the test is a companion diagnostic.

FDA also notes that most associations in the table have not been evaluated for whether genetic testing improves clinical outcomes. Some evidence links genetic variation only to pharmacokinetics, without establishing differences in efficacy or safety among genotype groups.

This does not make the table weak. It makes the table specific.

The mistake is asking a pharmacokinetic entry to answer an outcome question or reading an association as a mandate to order a panel.

The same caution applies when a patient brings a direct-to-consumer result. FDA's current consumer-testing page says it has not authorized a direct-to-consumer pharmacogenetic test to predict response or adverse reactions to a specific therapeutic drug. A consumer result can begin a clinical conversation. It should not be converted directly into a medication change.

A person's inherited genotype does not change when the medication list changes. The clinical meaning of the result can.

CPIC notes that predicted metabolizer status and drug exposure may also be affected by age, diet, comorbidities, smoking, pregnancy, and concurrent medications. Different laboratories may test different variants and can therefore produce different predicted phenotypes. Rare variants may not be included.

The report cannot see that a strong enzyme inhibitor was added last month unless someone brings the medication list to the interpretation. It cannot know that the patient previously improved on a medication placed in a caution category. It cannot decide whether that prior benefit outweighs a manageable exposure concern. It cannot know whether the favorable alternative is affordable or available.

"The genes do not change" is true.

"The report settles the decision forever" is not.

Commercial reports often compress several inputs into categories designed to be quickly read. That can be useful decision support. It can also make unlike claims look equivalent.

A favorable category may mean that the tested variants did not trigger a specific warning in the report's algorithm. It does not mean the medication is likely to work, has no important adverse effects, is appropriate for the diagnosis, or is preferable to every medication in another category.

A caution category does not necessarily mean "never use." Depending on the evidence and the product, it may point toward dose adjustment, monitoring, a potential interaction, or consideration of an alternative.

The color is the beginning of interpretation, not the end of prescribing.

CPIC also draws a boundary around proprietary combinatorial tests. It notes that some commercial products use opaque algorithms combining multiple genotypes. Because those methods are proprietary, CPIC says they fall outside its standardized guideline-evaluation process.

That does not establish that the algorithms are invalid. It means the clinician should not treat a CPIC gene-drug recommendation and a proprietary multigene color category as though they were the same kind of evidence.

The 2022 VA/DoD major-depression guideline found insufficient evidence to recommend for or against pharmacogenetic testing to guide antidepressant selection. That is neither an endorsement nor a prohibition.

The recommendation reflected the evidence reviewed at that time. The 2026 trial adds information, but one later study does not silently rewrite the guideline.

"Insufficient evidence" should not become "the test is useless." It should also not become "use clinical judgment" as a way to avoid describing what evidence is missing.

The useful task is narrower: identify what the result establishes, what it does not establish, and what independent clinical evidence still has to carry the decision.

The chart can say that a pharmacogenomic report was reviewed and what finding influenced the plan. It should not convert the report into a fictional prescriber.

A pharmacogenomic result can inform a prescribing decision.

It cannot make one.

Primary and authoritative sources consulted for this article, with the scope used and the date reviewed.

1. Nierenberg AA, Kamali M, Rabideau DJ, et al. "Comparative Effectiveness of Pharmacogenomics for Treatment of Depression." Journal of Clinical Psychopharmacology. 2026;46(5):571-583. doi: 10.1097/JCP.0000000000002175. PMID: 42572938. Used: randomized sample of 201 participants, intervention and comparator, twelve-month follow-up, outcomes, and authors' stated interpretations. PubMed abstract reviewed August 30, 2026.

2. Bousman CA, Stevenson JM, Ramsey LB, et al. "Clinical Pharmacogenetics Implementation Consortium Guideline for CYP2D6, CYP2C19, CYP2B6, SLC6A4, and HTR2A Genotypes and Serotonin Reuptake Inhibitor Antidepressants." Clinical Pharmacology & Therapeutics. 2023;114(1):51-68. doi: 10.1002/cpt.2903. Used: scope as interpretation of existing results; factors beyond genotype; limits of proprietary combinatorial algorithms; genetic results as one part of clinical information. Full text reviewed August 30, 2026.

3. U.S. Food and Drug Administration. "Table of Pharmacogenetic Associations." Used: distinction among therapeutic-management, safety or response, and pharmacokinetic associations; inclusion not constituting a recommendation to test; limits concerning clinical outcomes. Reviewed August 30, 2026.

4. Department of Veterans Affairs and Department of Defense. VA/DoD Clinical Practice Guideline for the Management of Major Depressive Disorder. Version 4.0, February 2022. Used: Recommendation 13 and its discussion, finding insufficient evidence to recommend for or against pharmacogenetic testing to guide antidepressant selection. Full official PDF reviewed August 30, 2026.

5. U.S. Food and Drug Administration. "Direct-to-Consumer Tests." Used: current FDA statement that no direct-to-consumer pharmacogenetic test has been authorized to predict response or adverse reactions to a specific therapeutic drug, plus consumer-test limitations. Reviewed August 30, 2026.

Educational Disclaimer: The Psychiatric Record provides general educational information for psychiatric and mental-health professionals. Content does not constitute medical, legal, regulatory, compliance, billing, genetic-counseling, or other professional advice; does not establish a standard of care; and is not a substitute for independent professional judgment. Appropriate testing and treatment depend on the patient, laboratory, medication, evidence, setting, and current authoritative guidance. This article does not recommend ordering, continuing, stopping, or changing any test or medication. The patient described is a fictional composite; no individual is depicted.