Genetic information could identify high-risk patients earlier
For generations, physicians have prescribed corticosteroids to millions of patients with chronic inflammatory conditions, knowing that serious harm might follow but lacking the means to foresee it. A study presented at the European Society of Human Genetics conference now suggests that the human genome itself may hold the key to that foresight — that by reading the subtle language of polygenic risk scores alongside dosing patterns, clinicians can begin to distinguish, before damage is done, who is most vulnerable. The work, drawing on nearly 38,000 patients, does not promise to end the era of steroids, but to make their use wiser, more humane, and more precisely matched to the biology of each individual who needs them.
- More than one in ten long-term steroid users develop serious complications — osteoporosis, stroke, cataracts — yet doctors have had no reliable way to identify who is at risk before harm occurs.
- Researchers at the University of Exeter identified specific gene variants, CYP3A4 and CTLA4, that significantly predict vulnerability to particular side effects, and found that heavier, longer steroid use compounds that risk in a clear dose-response pattern.
- The most significant advance came when polygenic risk scores — aggregating hundreds of small genetic signals — were layered onto dose data, dramatically improving predictive power beyond what age and sex alone could offer, especially for younger patients facing decades of treatment.
- Current clinical tools are blunt: shortest courses, lowest doses, or costly biologic alternatives that remain out of reach for many patients with severe chronic disease.
- Targeted genetic screening of high-risk patients could open a third path — earlier monitoring, preventive intervention, and more rational prescribing — though large-scale implementation and validation across diverse populations remain significant hurdles.
Doctors have long confronted a stubborn uncertainty at the heart of steroid prescribing: these drugs are powerful and often necessary, yet more than one in ten patients who take them long-term develop serious complications — osteoporosis, stroke, cataracts — with no reliable way to predict who will be harmed. Research presented this week at the European Society of Human Genetics conference suggests that genetic testing could finally change that calculus.
Led by Dr. Deniz Turkmen at the University of Exeter, the study analyzed medical records and genetic data from nearly 38,000 UK Biobank patients prescribed corticosteroids. The team confirmed a clear dose-response relationship — longer and heavier use meant greater risk — and identified specific genetic variants, including CYP3A4 and CTLA4, that predicted vulnerability to particular complications. The more transformative finding came when polygenic risk scores, which aggregate hundreds of small genetic signals into a single measure of biological predisposition, were added to the model. Predictive accuracy improved substantially beyond what age and sex alone could provide, with the gains most striking in younger patients facing potential decades of treatment.
Today's clinical tools are blunt by comparison: shortest possible courses, lowest effective doses, or expensive biologic alternatives that are not universally accessible. Genetic screening could offer a third path — identifying high-risk patients early enough to monitor them closely or offer preventive treatment before damage accumulates.
The path to widespread implementation is not straightforward. Screening large populations is logistically and financially demanding, and the genetic associations observed will need validation in more diverse populations before they can be confidently applied across clinical settings. The researchers suggest beginning with targeted screening of patients most likely to require long-term steroid therapy. Dr. Turkmen framed the ambition carefully: not to eliminate steroid use, but to make it safer and more rational — giving each patient the dose their own biology can tolerate, guided by the information already written in their genome.
Doctors have long faced a stubborn problem: when they prescribe oral corticosteroids to treat arthritis, asthma, or autoimmune disease, they cannot easily predict which patients will suffer serious side effects. More than one in ten people who take these drugs long-term develop complications—osteoporosis, stroke, cataracts—but until now, there has been no reliable way to identify who is at risk before damage occurs. A study presented this week at the European Society of Human Genetics conference suggests that genetic testing could change that calculation, allowing physicians to prescribe these powerful anti-inflammatory drugs with far greater precision.
Researchers at the University of Exeter, led by Dr. Deniz Turkmen, analyzed medical records and genetic data from nearly 38,000 patients in the UK Biobank who had been prescribed steroids. They tracked how much medication each person received over time, whether higher doses correlated with more adverse effects, and whether certain genetic variants could predict who would suffer complications. The work revealed a clear dose-response relationship: the longer and heavier the steroid use, the greater the risk of harm. More importantly, they identified specific genetic markers—variants in genes called CYP3A4 and CTLA4, among others—that significantly increased vulnerability to particular side effects. CYP3A4 variants predicted osteoporosis risk; CTLA4 variants predicted stroke and cataract risk.
But the real breakthrough came when the researchers incorporated polygenic risk scores into their analysis. These scores combine information from hundreds of common genetic variants, each with a tiny individual effect, to create a more complete picture of a person's biological predisposition. When they added polygenic risk scores for bone mineral density to their steroid risk model, the predictive power improved substantially—far beyond what age and sex alone could tell them. The improvement was especially striking in younger patients receiving their first steroid prescription, a population where long-term treatment is often necessary but where side effects can accumulate over decades.
Currently, doctors manage steroid risk through blunt instruments: they prescribe the shortest possible course, use the lowest effective dose, or switch patients to more expensive biologic drugs that spare steroid use. These strategies work for some patients, but they fail those with chronic conditions requiring repeated or sustained treatment. Biologics, while effective, cost far more and are not universally accessible. A patient with severe rheumatoid arthritis or uncontrolled asthma cannot always wait for insurance approval or afford the price tag. Genetic screening could offer a third path: identify high-risk patients early, monitor them more closely for emerging problems, or offer them preventive treatments before damage occurs.
Implementing this approach at scale presents formidable challenges. Testing tens of millions of people for polygenic risk scores is logistically complex and expensive. The researchers suggest that the most practical near-term application would be targeted screening of patients likely to need long-term steroid therapy—those with chronic inflammatory diseases where repeated courses are inevitable. The findings also need validation in other populations, particularly those of different ethnic backgrounds, to ensure the genetic associations hold true across human diversity. The pharmacogenetic effects observed in this study align with what scientists already know about how genes influence steroid metabolism and immune regulation, which is encouraging. But larger, more diverse datasets will be needed before these tools can be confidently deployed in everyday clinical practice.
Dr. Turkmen emphasized that while individual genetic variants have modest effects on steroid side effect risk, the cumulative power of polygenic scores—layered on top of dose information—offers something genuinely new. The goal is not to eliminate steroid use but to make it safer and more rational. As genetic data becomes cheaper and more widely available, she suggests, it will become feasible to integrate genomic information into routine prescribing decisions. That shift would represent a meaningful step toward personalized medicine: giving each patient the drug they need at the dose they can tolerate, informed by their own biological blueprint.
Citações Notáveis
The routine use of genetic information could mean that 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
We hope that greater availability of genetic data at population level will mean it will be possible to integrate genomics into everyday healthcare and hence into prescribing decisions.— Dr. Deniz Turkmen
A Conversa do Hearth Outra perspectiva sobre a história
Why is it so hard right now to predict who will have problems with steroids?
Doctors have no good way to look at a patient and know whether their body will handle long-term steroid use well or poorly. They can see age and sex, but those are blunt tools. Two 55-year-old women with the same arthritis might respond completely differently to the same dose.
And this genetic testing changes that?
It adds another layer of information. Certain genetic variants make your body process steroids differently, or make your immune system more vulnerable to steroid side effects. When you combine that genetic data with what we know about dose and duration, the picture becomes much clearer.
So a doctor could test a patient before prescribing?
Theoretically, yes. But it's not quite that simple yet. The genetic variants they identified—like CYP3A4 and CTLA4—are real, and they matter. But the biggest improvement came from polygenic risk scores, which look at hundreds of tiny genetic effects all at once. That's more complex to implement in a busy clinic.
What's the practical barrier?
Cost and logistics, mainly. Testing millions of people is expensive. And the research was done on UK patients, mostly of European ancestry. Before doctors can confidently use this in practice, they need to know it works just as well in other populations.
If it works, what changes for patients?
Instead of just giving everyone the lowest dose for the shortest time, doctors could identify high-risk patients early and either monitor them more closely, switch them to alternatives sooner, or use preventive treatments. For someone with severe arthritis who needs steroids long-term, that could mean avoiding osteoporosis or a stroke.
But steroids would still be prescribed?
Absolutely. They're too effective for too many conditions to abandon. This is about using them smarter, not less. It's about matching the drug to the person.