Genetic data could identify high-risk patients before damage accumulates
For the millions who rely on corticosteroids to manage chronic illness, the drugs have always carried a quiet gamble — relief for some, serious harm for others, with little way to know in advance which fate awaited. Researchers at the University of Exeter, drawing on nearly 38,000 participants from the UK Biobank, have now shown that genetic data — including specific variants and polygenic risk scores — can meaningfully sharpen that prediction. The work, presented at the European Society of Human Genetics conference, points toward a future where prescribing decisions are shaped not only by symptoms and age, but by the biological architecture each patient carries into the clinic.
- Between one in ten and one in five long-term steroid patients develop serious complications — weakened bones, stroke, cataracts — yet physicians have had almost no tools to foresee who is most vulnerable.
- Two genetic variants, CYP3A4 and CTLA4, emerged as specific, biologically coherent risk signals for osteoporosis, stroke, and cataracts — not statistical noise, but pieces of a known mechanism.
- Polygenic risk scores, combining hundreds of small genetic influences on bone density, dramatically improved side-effect prediction beyond standard factors like age and sex, with the sharpest gains seen in younger first-time steroid users.
- Biologic alternatives exist but remain expensive and inaccessible for many, making early genetic identification of high-risk patients a potentially transformative — and cost-reducing — intervention.
- Researchers urge caution: the study drew primarily from European-ancestry populations, and scaling genetic screening to clinical practice would require validation across diverse groups and significant logistical investment.
Millions of people take oral corticosteroids daily for conditions like arthritis, asthma, and lupus. The drugs are effective — but somewhere between one in ten and one in five long-term users will develop serious complications: brittle bones, cataracts, stroke, unexpected infections. Until recently, doctors had almost no way to predict who would be spared and who would suffer.
A study presented at the European Society of Human Genetics conference suggests genetics could change that. Researchers at the University of Exeter analyzed nearly 38,000 UK Biobank participants prescribed steroids, tracking doses and complications while searching for genetic patterns that might explain the divergence in outcomes. Two variants stood out: CYP3A4, linked to higher osteoporosis risk, and CTLA4, associated with stroke and cataracts — both connected to known biological pathways for steroid metabolism and immune regulation.
The more powerful finding came from polygenic risk scores, which aggregate the effects of hundreds of common variants on bone density. When layered onto traditional risk factors like age and sex, these scores substantially improved prediction — most strikingly in younger patients receiving their first prescription.
The practical stakes are real. Biologic drugs can reduce steroid dependence, but they are costly and not universally accessible. Routine genetic screening could allow clinicians to identify high-risk patients early, enabling closer monitoring or earlier transitions to steroid-sparing treatments before harm accumulates.
The researchers are measured about the road ahead. The current findings are drawn largely from people of European ancestry, and genetic effects can vary across populations. Scaling this approach would require broad validation and significant infrastructure. Still, the vision animating the work is clear: a clinical future where the biology beneath a patient's symptoms informs the prescription — personalized medicine moving quietly from aspiration into everyday practice.
Millions of people take oral corticosteroids every day—for arthritis, asthma, lupus, and dozens of other chronic inflammatory conditions that won't quit. The drugs work. They tamp down inflammation, ease pain, quiet an overactive immune system. But somewhere between one in ten and one in five patients who take them long-term will develop serious side effects: weakened bones, cataracts, stroke, infections that shouldn't happen. Until now, doctors have had almost no way to know in advance who will be hit and who will sail through.
A study presented this month at the European Society of Human Genetics conference suggests that genetic data could change that calculation. Researchers at the University of Exeter examined nearly 38,000 people in the UK Biobank who had been prescribed steroids, mapping their doses over time and tracking which ones developed complications. Then they looked for genetic variants that might explain why some patients suffered while others didn't.
What they found was specific and actionable. Two genetic variants stood out: a version of the CYP3A4 gene that made osteoporosis more likely, and variations in CTLA4 that increased the risk of stroke and cataracts. But the real breakthrough came when the researchers went further. Instead of looking at single genes in isolation, they built what's called a polygenic risk score—a calculation that combines the effects of hundreds of common genetic variants, each with a tiny influence on bone density. When they added this score to the traditional risk factors doctors already use—age, sex, family history—the prediction got dramatically better. The improvement was especially striking in younger patients getting their first steroid prescription.
Dr. Deniz Turkmen, the postdoctoral researcher who led the work, noted something reassuring: the genetic findings made biological sense. CYP3A4 and CTLA4 are known to play roles in how the body metabolizes steroids and regulates immune response. The genes weren't random associations pulled from noise. They were pieces of a mechanism.
Right now, when doctors want to minimize steroid side effects, they have limited tools. They prescribe the shortest course possible, use the lowest effective dose, or switch patients to biologic drugs—newer, more targeted treatments that work around the need for steroids altogether. But biologics are expensive, often hard to access, and not always an option for people with conditions that need long-term steroid treatment. If genetic screening became routine, doctors could identify high-risk patients early and either monitor them more closely or move them to steroid-sparing alternatives before damage accumulates.
The researchers are careful about the next steps. Implementing genetic testing at scale would be a massive undertaking. The most practical approach would be to target people most likely to need steroids for months or years—those with severe asthma, autoimmune disease, chronic rejection after transplant. The findings also need to be tested in other populations to make sure they hold up. The current study was done primarily in people of European ancestry; genetic effects can vary across populations, and a test built on one group might not work as well in another.
Turkmen and her colleagues see this as a step toward a larger vision: a future where genetic information is woven into routine medical care, where prescribing decisions are informed not just by symptoms and age but by the biology underneath. It's personalized medicine moving from the margins into everyday practice. For the millions taking steroids, it could mean the difference between a drug that heals and one that harms.
Citações Notáveis
Genetic information could mean that in the future, patients at high risk could be identified and given earlier steroid-sparing treatments, or have closer monitoring for side effects.— Dr. Deniz Turkmen, University of Exeter
The improvement in prediction of osteoporosis when incorporating polygenic risk score data was remarkable, especially in younger patients.— Dr. Deniz Turkmen
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter that we can predict steroid side effects genetically? Doctors already know steroids can cause problems.
They do know it in theory. But right now they can't tell which individual patient will be the one in ten who gets osteoporosis or stroke. So they either restrict steroids for everyone—which hurts people who actually need them—or they prescribe freely and hope. Genetic data changes that from guessing to knowing.
So you're saying this is about precision, not just safety.
Exactly. A young person with severe asthma might need steroids for years. Right now, a doctor might push them toward an expensive biologic drug out of caution, even if genetically they're low-risk. With this test, you could say: your genes suggest you'll tolerate this well, let's monitor you carefully. Or the opposite: your risk profile is high, let's try something else first.
What's the catch? Why isn't this already standard practice?
Scale and diversity, mainly. Testing 38,000 people is one thing. Rolling it out to millions is another—it requires infrastructure, cost, training. And the study was mostly white Europeans. The same genetic variants might not predict risk the same way in other populations. You need proof it works everywhere before you make it routine.
Does this mean steroids are going to become safer?
Not safer in themselves. But smarter. The drug doesn't change. What changes is that doctors can match the right patient to the right treatment, and catch problems earlier in people who are actually at risk. That's a meaningful shift.