Two people can take the same dose of the same medicine and have very different results. One may improve, while the other gets little benefit or develops serious side effects. Part of the explanation often lies in their genes.
Author
- Rebecca Moore
General Practitioner and postgraduate researcher in pharmacogenomics, University of East Anglia
Pharmacogenomics uses genetic information to predict how someone may respond to certain medicines. Some genes make proteins that help the body process medicines. Others influence how medicines act. Variants in these genes can change how well a treatment works or the risk of side effects.
Genes are only part of the picture. Age, kidney and liver function, other medicines and the condition being treated can all affect a person’s response. Allergic reactions also have different causes.
Almost everyone carries a relevant variant. An analysis of nearly 500,000 UK Biobank participants found that 99.5% were predicted to have an unusual response to at least one drug because of variants in 14 well-studied genes. Almost one-quarter had already been prescribed a medicine affected by one of those genes.
The immediate value of this information is narrower than that figure might suggest. A variant becomes useful when someone is taking, or may be prescribed, a relevant medicine and there is good evidence about what a doctor should do differently.
In those circumstances, testing may prevent harmful reactions or ineffective treatment. Genetic differences account for only a proportion of adverse drug reactions, which can range from nausea to life-threatening bleeding. A 2022 study of one month’s adult medical admissions at a Liverpool NHS trust found that an adverse reaction caused or contributed to 16.5% of admissions.
The researchers estimated that such admissions could cost NHS England £2.21 billion a year. This was extrapolated from one NHS trust over a single month, so it is an estimate rather than a national count.
Older people taking several medicines face a particularly high risk of adverse reactions. As the population ages, reducing avoidable harm will become increasingly important.
There is evidence that testing can make prescribing safer. A trial involving 6,944 patients in seven European countries tested variants in 12 genes and gave prescribers drug-specific advice.
Among patients with a result that could affect their treatment, clinically relevant adverse reactions occurred within 12 weeks in 21% of those whose care was guided by the test, compared with 27.7% of those receiving standard care.
Testing is already used for selected medicines. An NHS resource updated in 2025 lists five drug-gene pairs for which testing occurs relatively routinely in England. These include testing the DPYD gene before certain chemotherapy drugs, because some variants reduce the body’s ability to break them down. Clinicians also test the immune-system gene HLA-B before prescribing the HIV medicine abacavir, because one variant increases the risk of a serious hypersensitivity reaction.
Wider use is being explored through an NHS England-funded study examining how a national pharmacogenomic testing service could work. The project is also investigating how results and prescribing advice could be incorporated into clinical computer systems.
The Netherlands offers a useful comparison. Advice from the Dutch Pharmacogenetics Working Group is incorporated into the national medicines database used by prescribers and pharmacists. This makes guidance available at the point of care when a patient’s genetic result is known. Access to testing remains a separate issue.
How could a genetic test change treatment?
Consider someone leaving hospital after a heart attack. They may be prescribed clopidogrel to prevent blood clots, a statin to lower cholesterol and omeprazole to protect their stomach.
Some people have CYP2C19 variants that reduce their ability to activate clopidogrel. In certain heart conditions, clinical guidance recommends considering another antiplatelet medicine for people with particular results.
Other variants can influence the risk of muscle symptoms from some statins . CYP2C19 also affects levels of omeprazole , which in turn affects how useful it is in preventing stomach problems caused by other medications.
Each medicine would still require a separate clinical assessment. A result could give the prescriber additional evidence when choosing a drug or dose.
The same caution applies to antidepressants. CYP2D6 and CYP2C19 variants can influence levels of some antidepressants, and guidelines explain how results may inform prescribing . A result may help a clinician choose a starting dose or consider an alternative. Genes explain only part of someone’s response to treatment.
A sample for testing can come from blood or a cheek swab. Because inherited variants do not change, the result could be stored in a medical record for a patient’s lifetime, and consulted when other medicines are prescribed. Its interpretation may develop as evidence and clinical guidance are updated.
Who could benefit?
People taking several medicines have more opportunities to encounter a relevant drug-gene combination. Fair access will need careful planning, however, to ensure that testing is accessed by those who need it most.
Pharmacogenomics can reduce some of the trial and error involved in prescribing when a drug-gene relationship is backed by strong evidence. The NHS already has parts of the system in place. Wider use will require reliable testing and current clinical guidance. Prescribers will also need suitable training.
For both clinicians and patients, the practical question is simple: it is not so much a question of can or should we test, but how can we make testing routine?
