At the University of Exeter, researchers have cracked open a puzzle that has frustrated doctors for decades: who will actually suffer serious side effects from steroid medication. By studying genetic data from nearly 38,000 UK patients, Dr. Deniz Turkmen and her team discovered that adding a person's genetic blueprint to steroid prescribing decisions could predict adverse reactions far more accurately than age, sex, or dosage alone—a finding that could transform how millions receive treatment for arthritis, asthma, and autoimmune diseases.
Oral corticosteroids are miracle workers for chronic inflammation, but they come with a steep cost. More than one in ten patients develop side effects, particularly those who need steroids long-term. Until now, doctors have had no reliable way to know who would be vulnerable, so they've defaulted to defensive strategies: keeping courses short, using the lowest possible doses, or switching patients to expensive alternative treatments like biologics. These workarounds often fail for people with genuinely chronic conditions that demand sustained steroid treatment.
Turkmen's breakthrough came from examining how much steroid each patient took over time and whether genetic variations could explain individual risk. The team identified specific genetic variants that signal danger: the CYP3A4 variant increases osteoporosis risk, while CTLA4 raises the likelihood of stroke and cataracts. They also confirmed the intuitive finding that higher steroid doses correlate with more side effects—but the real surprise was what happened when they layered genetic information on top of traditional factors.
When the researchers incorporated polygenic risk scores for bone mineral density into their assessment model, the prediction accuracy jumped dramatically. This improvement was not marginal; it was particularly striking in younger patients facing their first steroid prescription. A 25-year-old with a genetic predisposition to weak bones could now be identified before starting treatment, allowing doctors to intensify monitoring or switch them to safer alternatives early.
The implications ripple across healthcare systems worldwide. Steroids are prescribed constantly, in vast numbers, for conditions that affect hundreds of millions of people. Yet implementing genetic screening on that scale presents formidable challenges. The researchers acknowledge that the most practical near-term application would target those at highest risk—particularly people who need steroids for months or years rather than weeks. They also note that the study's findings need validation in other populations, ideally more ethnically diverse than the predominantly European UK Biobank cohort.
Turkmen envisions a future where genetic data becomes as routine in prescribing as taking a patient's blood pressure. "We hope that, in time, greater availability of genetic data at population level will mean that it will be possible to integrate genomics into everyday health care," she said. That shift could mean identifying high-risk patients early enough to give them steroid-sparing treatments or intensify monitoring for complications. For patients with severe lupus, relapsing polyarteritis, or severe asthma—conditions where steroids remain essential—this could be the difference between a manageable course of treatment and years of preventable damage.
The challenge now is turning research into routine clinical practice. But the blueprint is emerging, rooted in the genes we all carry.