Pharmacogenetic testing: what it can tell you about your medicines, and what it can’t

A test looks for specific variants in one or more genes. Depending on the gene-drug pair, these variants are associated with a drug being cleared more slowly or more quickly, a higher risk of side effects, or a different level of effectiveness. Not all of these associations are equally solid.

The US Food and Drug Administration (FDA) keeps a Table of Pharmacogenetic Associations that it has reviewed, sorted into three tiers: associations where the data support therapeutic management recommendations, those that point to a possible impact on safety or response, and those that only show an effect on how the drug moves through the body. The first tier includes, for example, the CYP2C19 gene and clopidogrel, a blood thinner; CYP2D6 and codeine; and DPYD and fluorouracil, a chemotherapy drug.

The FDA spells out two limits itself. Being listed does not mean the agency advocates a test before prescribing, unless the test is a companion diagnostic, essential to using the drug safely and effectively. And for most listed associations, the impact of testing on clinical outcomes has not been evaluated.

In Europe, the European Medicines Agency has a guideline on good pharmacogenomic practice, in effect since 2018, which covers the choice of genomic methods during a drug’s development and life cycle. On the clinical side, the Clinical Pharmacogenetics Implementation Consortium (CPIC) publishes guidelines for specific gene-drug pairs. As its published development process explains, they tell clinicians how to use a test result that is already available, not whether a test should be ordered. Each recommendation is graded strong, moderate or optional.

A documented gene-drug pair

The result has to concern the medicine actually being considered, and a variant the test can reliably detect. A very broad panel can produce plenty of information with no bearing on your treatment. The useful question is a precise one: should a drug be avoided, a dose adjusted, monitoring stepped up or an alternative chosen?

A recommendation that applies

Genetic information becomes valuable when it connects to an evidence-based course of action: the drug’s official labeling, a regulatory recommendation or a guideline such as CPIC’s. The level matters: a strong recommendation does not carry the weight of an association still under study.

The rest of the picture

A genotype does not sum up how your body handles a drug. In the FDA’s words, a patient’s genetic makeup is only one of many factors that can affect drug concentrations and response. Other medicines taken at the same time are one of them: the FDA table includes cases where caution applies only when a poor metabolizer is also taking another drug that blocks the same enzyme. A result is therefore read together with your health history, your current medicines and the goal of the treatment.

Pharmacogenetics has a very concrete use in cancer care. Fluoropyrimidines, 5-fluorouracil and capecitabine, are cleared mainly by an enzyme called DPD. In patients who lack it, these drugs can cause severe, even fatal, toxicity. That is why the FDA table lists DPYD, the gene behind the enzyme.

France offers a telling case. In December 2018, the French National Authority for Health (HAS) and National Cancer Institute recommended screening for DPD deficiency before treatment. Since 2019, the French medicines agency (ANSM) has made it mandatory: the prescriber notes on the prescription that the result was taken into account, and the pharmacist checks for that note before dispensing. According to the ANSM, partial deficiency affects 3 to 8% of the population, complete deficiency 0.05 to 0.1%.

The detail worth noting: the test France chose is not a genetic one. It is a blood measurement, uracil levels, which HAS confirmed in 2023 as the most suitable option. HAS also points out that screening will not prevent every severe toxicity, since not all of them are caused by DPD deficiency. Knowing a gene is one route to the right decision; it is not the only one.

A pharmacogenetic test does not predict every side effect, does not guarantee that a drug will work and does not replace medical follow-up. It only looks for a defined set of variants. For genetic tests sold directly to consumers, the FDA notes that companies do not all test for the same variants, that some tests do not report every known variant, and that a negative result therefore does not rule everything out. How a variant is classified can also vary, which adds uncertainty.

Consumer tests call for particular caution. In 2018, the FDA warned about pharmacogenetic tests making claims it had not reviewed, as two FDA experts recall in a 2021 review for family physicians. Their example is antidepressants: it has not been established that their effectiveness is related to DNA variations. Relying on such tests could lead to avoiding first-line drugs in favor of less proven alternatives, or to staying on a medicine that does not help. And the consumer pharmacogenetic report that the FDA lists among authorized tests does not say whether a person will respond to a given drug; its information should not be used to start, stop or change any treatment.

Rules differ from country to country

Where you live changes what is allowed. In France, for instance, examining a person’s genetic characteristics is permitted only for medical or scientific research purposes, with their prior written consent, under Article 16-10 of the French Civil Code, which also bans advertising for such tests. In the European Union, genetic data are one of the special categories of personal data that Article 9 of the GDPR protects more strictly. Wherever you are, genetic data do not change and also say something about your relatives: before handing a raw result to any service, check who keeps it, for how long, for what purposes, and how to have it deleted.

Say a report flags a variant in an enzyme that processes a medicine you take. The first step is not to change the dose. First check that the report is reliable, that the variant is actually covered by the test, that a recommendation applies to this specific drug, and that your other medicines do not change the reading. Then come the expected benefit and the alternatives.

The decision may be to change nothing, to adjust the treatment, to monitor more closely or to confirm the result. It belongs to your prescriber, with you. Skip these steps, and the apparent precision of a result can give false reassurance.

Bring the full report, not a screenshot or a color. Note the laboratory, the date, the genes and variants tested, and the interpretation provided. Add an up-to-date medication list: it shows which medicines the result may concern, and which ones might interfere.

A few useful questions:

• Does this result concern the medicine being considered?
• Which recommendation supports this interpretation?
• Does the test cover the variants that matter in my case?
• Could my other medicines change how the result should be read?
• Does the result change the dose, the drug, or only the monitoring?
• Should it be confirmed by a clinical laboratory?
• Where will it be recorded, so it is not misread later?

Your prescriber can draw on a pharmacist, a clinical pharmacologist, a geneticist or a specialized laboratory.

A genetic variant does not change; its interpretation can, as knowledge moves on. Keep the original report, its date and the decisions that followed, along with the rest of your personal health record. Avoid turning a one-off conclusion into a permanent label, “contraindicated” or “compatible”, without the drug, dose and context it referred to.

With MyTwin for patients, your genetic tests sit in your record alongside your lab results, your medical reports and your medicines, and you can talk them through with a doctor or another healthcare professional. The decision to start, change or stop a medicine stays with the prescriber.