June 11, 2025 | After the human genome was sequenced in 2003, expectations were high that it would drive direct and immediate benefits for the everyday healthcare of the general population. While genomics has fundamentally altered the clinical pathways for individuals with rare diseases and cancer, it has yet to become part of routine clinical practice, according to John McDermott, Ph.D., NIHR Academic Clinical Lecturer at the University of Manchester (UK).
But pharmacogenomics, which deals with the role of the genome in drug response, has the potential to become the leading example of how an understanding of a person’s genetic makeup can improve patient outcomes and potentially reduce overall healthcare costs, he says. A growing number of funders, commissioners, and healthcare systems have all come to the same conclusion.
Different types of pharmacogenomic initiatives have popped up around the world, most notably in the U.S., The Netherlands, and Spain, says McDermott. The UK set out two years ago to do likewise, using the learnings from global colleagues to inform how to implement a program aligned to the peculiarities of the National Health Service (NHS).
“We know a large proportion of medical treatments that are issued in routine clinical care are ineffective or don’t work at all,” he continues. “You see that borne out in the proportion of people who don’t respond to the first-line antidepressants, or... go on to have secondary strokes.”
Genetics is just one of the reasons why people don’t do well on certain medicines, McDermott emphasizes. Drug responses are influenced by a variety of interacting factors—e.g., age, body size, other medications, food, and unrecognized comorbidities—so this “needs to be combined with other patient data to help select the best medicines.”
That said, “a significant proportion of individuals’ poor response or adverse reaction is driven by their pharmacogenomic variation,” he adds, so knowing what that is ensures patients are put on medicines they have the best chance of responding to. In addition to people hopefully feeling better sooner and having a better quality of life, this could mean big cost savings for health systems because patients require fewer future appointments and hospitalizations.
A case in point is when clopidogrel is given to patients with a CYP2C19 poor or intermediate metabolizer phenotype who have had a prior ischemic stroke, which can increase the risk of a secondary, more debilitating stroke, McDermott says, highlighting the importance of getting individuals on the right antiplatelet drugs. Between 25% and 30% of the population in the UK have a reduced response to clopidogrel based on common genetic changes in their CYP2C19 gene, he notes.
This is why the National Institute for Health and Care Excellence has guidelines recommending that all patients who have had a stroke or are experiencing a transient ischemic attack (TIA) receive pharmacogenomic testing. The associated health savings over five years have been estimated at “potentially hundreds of millions of pounds in savings,” he reports.
Similarly, testing for variation in the DPYD gene is mandated in England for all patients before starting fluoropyrimidine-based chemotherapy. As part of routine clinical practice, about 40,000 patients get tested each year, says McDermott.
“Things can move slowly in the NHS at times, but when they do move, they move at scale, with initiatives impacting the whole population and health system,” he explains. “This contrasts to the U.S. where you have pockets of real excellence in pharmacogenomics, but other centers where access to those initiatives may not be possible.” The NHS is in fact a “world leader” as it pertains to incorporating DPYD-based pharmacogenomic data into the clinical care of patients, he says.
At the annual congress of the European Society of Human Genetics, held recently in Milan, Italy, McDermott described a pioneering new approach to integrating genomic data into electronic health records (EHR) in both primary care practices and hospitals. It was the brainchild of the NHS-England Network of Excellence for Pharmacogenomics & Medicines Optimization, which McDermott co-leads.
The presentation included promising interim results from the ongoing PROGRESS (Pharmacogenetics Roll Out – Gauging Response to Service) project of the NHS that aims to implement pharmacogenomic-guided prescribing into routine clinical practice. This is a pragmatic implementation trial testing whether pharmacogenomic testing results can be delivered back to clinicians in a clinically relevant format and timeframe, he says.
The PROGRESS trial recruited patients from 20 sites across England following prescription of common medicines—specifically, statins, opioids, antidepressants, and proton pump inhibitors—and pharmacogenomic guidance was returned via EHRs. Based on findings from the first 500 study participants, guidance was provided on all of them with a median turnaround time of seven days.
Tellingly, 95% of patients had any actionable pharmacogenomic variant, and just over one in four participants had their prescription adjusted to achieve safer or more effective treatment. In essence, the EHR-embedded popup warnings were working as intended.
The initial learning phase of the PROGRESS trial involved a few hundred participants and focused on understanding the different EHR systems being used by general practitioners and how patients were interacting with those care providers, McDermott says. At that point, pharmacogenomic testing results were being returned through an external portal.
Last fall, the trial entered its second and final implementation phase that will wrap up early next year. Prior work found that when returning pharmacogenomics data via traditional methods, either an external system or PDF report, clinicians were less likely to act on the data, says McDermott, which is why full EHR integration became the end goal.
“Our attitude is that pharmacogenomic data should be treated like any other biomarker,” he says, referencing EHR popup alerts about low renal function when physicians go to prescribe the antibiotic nitrofurantoin, so that they choose an appropriate alternative medicine for those patients. “What we want to do with pharmacogenomics is exactly the same thing.”
PROGRESS is being funded by NHS England and the Genomic Medicine Service Alliance, in addition to a few other organizations, including the National Institute for Health and Care Research. The rationale for their support is tied to the longstanding promise of genomics to improve the health and well-being of the general population, McDermott says.
Creating Interoperability
That the NHS makes decisions nationally comes with opportunities and challenges when it comes to implementing changes of any sort, McDermott says. Unlike leading academic medical centers in the U.S.—McDermott has worked with Vanderbilt University Medical Center, Mayo Clinic, and St. Jude Children’s Research Hospital specifically—health systems in the UK are not using a single EHR but a variety of commercially supplied clinical software systems and the population is highly mobile when it comes to accessing care.
The implementation phase of PROGRESS has therefore required the development of an informatics solution for results-sharing that is agnostic to the EHR that has been adopted. Whether clinicians are working in Epic, EMIS Health, SystmOne, or some other system, the popups appear in the same way, says McDermott, and the response to that flexibility has to date been overwhelmingly positive.
A cloud-based data repository (“PROGRESS Rx”) receives the output files from labs running the pharmacogenetic tests about three times per week and automatically converts those into a standardized data format which his group developed, he explains. The platform interacts with the various EHRs so that results appear during prescribing moments along with relevant clinical alerts and recommendations.
It is more challenging in the UK than it is in the U.S. for individual centers to independently decide to implement a pharmacogenomics service because this type of intervention needs to happen at a national level to ensure equity across a health system, says McDermott. “It would be inappropriate if we had this testing in Manchester, but not in London, or available at one hospital and not another.
“That wouldn’t be fair and would also limit the usefulness of the intervention,” he continues. Furthermore, he stressed that “large-scale interventions of any type need to be justifiable from a health economic perspective.”
Similarities and variations between dozens of pharmacogenetic testing programs and multi-center initiatives—well over half based in the U.S. and nearly all of them in high-income countries—was the subject of a scoping review McDermott and his colleagues undertook a few years ago (Frontiers in Medicine, DOI: 10.3389/fmed.2022.945352). It highlighted the key facilitators that could be leveraged to promote successful implementation.
From the U.S., “we learned about the art of the possible with regard to integrating within electronic healthcare records,” McDermott says, and from The Netherlands how to develop laboratory testing at scale for pharmacogenomics. “We took what we thought were the best bits of each of those programs, recognizing which ones wouldn’t work in our health system,” and that became the starting point for subsequent efforts in the UK.
PROGRESS launched after a landmark randomized controlled trial demonstrating the potential value of a 12-gene pharmacogenetic panel was published in The Lancet (DOI: 10.1016/S0140-6736(22)01841-4). This so-called PREPARE trial showed an up to one-third reduction in adverse drug reactions across seven European countries.
The next question became how to deliver the same sort of results in a complex, real-world healthcare ecosystem like the NHS without all the hand-holding that happens in a trial setting, McDermott says. The interim analysis from PROGRESS suggests that beneficially changing on-the-ground prescribing behavior is indeed possible.
“We see that around one in eight participants have what we call a red flag recommendation and should have a change in their prescription based on their genetic data,” he adds. In the “vast majority” of cases, the recommended guidance is being followed.
The global pharmacogenomics community still needs to demonstrate if making panel-based pharmacogenomics part of routine clinical practice leads to improved healthcare outcomes for the overall population or health system in each of their respective jurisdictions. In the UK, the plan once PROGRESS concludes is to look at whether the program decreases healthcare utilization in real-world practice.
The potential benefits of pharmacogenomic testing are vast since patients often carry a genetic variant of interest when it comes to commonly prescribed drugs, including medicines used in most medical specialties, says McDermott. “If implemented correctly, it could benefit a large proportion of individuals at some point in their life.”
Currently, testing results in the UK are being returned to the care providers via the various EHR systems, McDermott says. But in an upcoming stage of the study protocol, it is expected that results will also be returned to participants. This could happen through the popular and widely used NHS App whereby users are already accessing a range of NHS services, viewing test results, and booking appointments.
The seven-day turnaround time for results is, admittedly, too long of a wait in some clinical circumstances, says McDermott. When answers are needed extremely quickly, that might happen by having the data in the patient’s record preemptively rather than “reactive with planned reviews.” At some set point in their lives, individuals would be tested, so the information would already reside within their records.
The other option is to develop point-of-care testing strategies. Two programs of work are underway in the UK, McDermott reports, one of which involves rapid swab-based testing in neonatal intensive care units around the country to learn if babies will develop deafness when they’re exposed to antibiotics. The other is rapid testing for the CYP2C19 phenotype to guide the prescribing of antiplatelet drugs to patients who have had a stroke or TIA.
For the more commonly prescribed drugs that are the subject of the PROGRESS trial, patients have generally felt comfortable with the idea of waiting seven days or so for their testing results. “If people wanted to be given their prescription [right away] and then have it adjusted at a later time point, that was permissible as well,” he says.
Academics around the world need to likewise be thinking about how to move not just from data to insights, but from insights to action, McDermott stresses. Unless pharmacogenomic data gets out of research settings and big laboratories and into the hands of clinicians working in emergency care, acute medical units, and general practice, it will have very little value.
“It is critical that we develop the pathways, that we develop the schemas, and that we develop the protocols to get that data out there so it can be used in clinical practice, and we can understand its value,” he says. “That’s what we’re trying to do here, and I think it’s often an under-recognized component of how we move forward in genomic medicine.”