Mroz’s team rolls out a system-wide advanced pharmacogenomics (PGx) platform
Bad reactions to medications are more common than most people think. They constitute a major public health concern because they can lead to treatment failure, hospitalization, and sometimes death. More than a million serious adverse drug reactions (ADRs) are reported in the U.S. every year.
Right now, the most common way to prevent ADRs due to prescribing drugs is trial-and-error. Yet it is estimated that four out of five patients are likely to carry a genetic variant that could alter the effectiveness of commonly prescribed drugs. Wouldn’t it be a good idea to have that information in advance?
LMP associate professor Pawel Mroz and his PGx team think it would. PGx stands for pharmacogenomics, the study of how genes affect a person’s response to drugs. PGx seeks to optimize the therapeutic benefit of a medication while reducing the risk of an ADR based on how someone’s genes may affect how that person metabolizes a drug. That metabolic process is concentrated in the liver. That’s where the chemical makeup of the drug encounters the metabolizing enzymes of the patient. Drug metabolizing enzymes are influenced in their behavior by genes and genetic variants. Drugs are standardized, but no two genetic endowments are exactly alike.
We introduced Mroz’s PGx initiative in 2023 with “Mroz shepherds pharmacogenomics (PGx) into the clinic,” a collaborative effort aiming to take PGx—using the right drug at the right dosage for the right patient at the right time—statewide. Now the initiative, involving LMP, M Health Fairview, the Institute for Health Informatics, and the College of Pharmacy, is in prime time with the launch of its next-generation sequencing (NGS)-based Go4PGx testing platform within M Health Fairview. “NGS will enable a major expansion of our program,” Mroz said, adding that NGS is much more comprehensive than targeted genotyping, which most PGx testing panels at other laboratories still employ.
NGS Go4PGx testing platform takes the lead
As noted in the 2023 PGx story, Mroz teamed up with LMP professors Bharat Thyagarajan and Sophia Yohe in the M Health Fairview Molecular Diagnostics Laboratory (MDL) to create and validate a clinical-grade multiplex PGx genotyping panel based on single nucleotide polymorphisms (SNPs) and other genetic variants. The assay consisted of nine genes that affect how 20 medications are broken down by the liver's drug-metabolizing enzymes. The new NGS-based PGx assay consists of 19 genes that affect how more than 65 medications are metabolized, based on gene-drug guidelines from the Clinical Pharmacogenetics Implementation Consortium (CPIC). Plus, the NGS approach has revealed thousands of additional variants, some of which could influence drug metabolism to such an extent that, with further study, they become clinically relevant.
“CPIC is the main source of the pharmacogenomic guidelines,” Mroz said, adding that CPIC provides a “level of evidence” for each guideline issued. “We’ve limited the gene-drug pairs to the highest level of CPIC’s specific evidence, which is based on the results of clinical trials showing drug-dose relationships.”
Among the potentially PGx “actionable” drugs based on NGS testing results are anesthetics, antibiotics, anticoagulants, anticonvulsants, antidepressants, antiemetics, antifungals, anti-inflammatories, antimalarials, anti-reflux agents, antivirals, chemotherapeutic drugs, pain medications, and statins used to lower blood cholesterol. Some of these drugs are among the most commonly implicated in adverse drug reactions.
Full-scale interactive CDS implemented system wide
But establishing an advanced Go4PGx testing platform is only part way to reaching the ultimate goal Mroz and his colleagues have envisioned since the project was launched in 2022. The testing opportunity needs to be integrated into an interactive support system within a health information system that can implement testing services upon request, report results, send alerts, and adjust recommendations as CPIC guidelines are extended with more clinical data. In brief, a clinical decision support (CDS) system needs to be implemented. Mroz and his colleagues have done just that. They’ve built a “full-scale CDS with genomic indicators and best practices alerts,” Mroz said.
Two key systems are now operational for Go4PGx patient testing with its in-house CDS: Fairview’s Epic electronic medical record (EMR) system with its patient friendly MyChart, and HL7 or Health Level Seven, a set of international standards for the exchange, integration, sharing, and retrieval of electronic health information. The not-for-profit HL7 enables what has long eluded health information sharing in the U.S.: the “interoperability” of different health IT systems. HL7 has been dubbed the universal language of health technology.
“What we built with HL7 allows us to send discrete data for pharmacogenomic results that include everything from the data that are related to genomic variants that we have detected, to the phenotype interpretation [the process of translating an individual's genetic variations into predicted drug response], and then the gene-drug recommendation,” Mroz said. “That all gets built into Epic and into MyChart. With HL7, the system can be easily activated and alerts sent if new CPIC guidelines are released or if new relevant variants are discovered.” That’s because CPIC and HL7 are interoperable, which means that Epic and MyChart are interoperable with CPIC through HL7.
Michelle Stoffel, LMP assistant professor, medical director of clinical informatics, and associate chief medical information officer of LMP for M Health Fairview, is one of about 30 faculty and staff helping drive the success of Go4PGx.
“It's been a massive project,” Stoffel said. “It brings together lab teams, pharmacy teams, molecular pathology, and Epic specialists in a unique collaboration that’s pushed the boundaries of what we consider possible in lab testing.”
Go4PGx reports provide clinicians with clear, actionable guidance, such as “prescribe this, avoid that", all based on a patient’s genetic profile. “What’s especially innovative,” Stoffel added, “is that the system is designed to update recommendations over time. As CPIC guidelines evolve or new actionable genes are identified, the patient’s record can be updated accordingly.”
Traditional lab results are static snapshots, Stoffel explained. “We usually think of them as ‘one and done.’ But with sequencing, a PGx test done 10 years ago could yield a different report today. It's very novel to have the potential for lab results for a given specimen to evolve over time.”
From “one and done” to dynamic PGx testing reports
That’s the power of the NGS Go4PGx assay, Mroz said. “Before our interactive CDS, a patient typically had one provider, one disease, one medication, and they wanted to see what would work for them. A static CDS report was issued, and the only people who knew about it were the provider and the patient. Now, we’ve built an interactive CDS and are testing comprehensive [NGS] panels, so even though we started with one patient, one provider, and one medication, now, because all the results are connected to CDS, and the patient goes to different providers -- as they usually do because a lot of patients have more than one complaint -- the Rx and medications will be available for other providers across the entire healthcare system. We’ve already seen CDS firing for individual patients across different locations within the Fairview system even though it was initiated by a single provider.”
As system-wide NGS Go4PGx gene-drug data accumulates, prediction models based on machine learning and artificial intelligence (AI) might indicate which patients with certain co-morbidities would most likely benefit from additional PGx testing, Mroz said.
What about reimbursement, a familiar and often major hurdle for the adoption of new healthcare technologies? Would a sharp reduction in the risk of an adverse drug reaction persuade insurance companies to cover the cost of the test? That remains to be seen, but Mroz observed that some genetic testing introduced a decade ago is now fully covered.
“We tell the patient it’s going to cost out of pocket,” Mroz said. “We try to keep the cost as low as possible. We do attempt to send it for reimbursement, and occasionally with success. I expect that the reimbursement rate will continue to grow as the testing benefits become more apparent and PGx testing becomes the standard of care across several specialties.”
Out of the initial thousand patients for whom Go4PGx testing has been offered (most by genotyping, but several dozen by the new NGS approach), fewer than ten have refused based on the cost. “The acceptance of payment is quite high, and patients are quite motivated,” Mroz said. “Patients are becoming more savvy and more informed about the role of pharmacogenomics in treatment. They are starting to demand testing. The oncology patients in particular are very motivated in pushing for pharmacogenomic testing, for advocating the testing.”
The Go4PGx test is available to anyone in the Fairview system. Through its development phase the testing has been driven by primary care and mental health patients. Mroz expects pediatric and adult oncology to participate soon, based on discussions he’s had. “The strength of the CPIC guidelines and the strength of our relationship with different departments is what matters,” he said. “That’s the future.”
What about the state-wide effort as first described in the 2017 UMN Grand Challenge State-wide Pharmacogenomics (PGx) Initiative? Will other Minnesota health systems participate?
“I hope so,” Mroz said. “I do think we have one of the best assays and one of the best clinical incident portals in the state. That’s without question. We are the only ones doing NGS at scale connected to a CDS system at scale. With HL7 – CDS integration, Go4GPx can send gene-drug recommendations supported by discrete pharmacogenomic data to any system.”
Image at the top of the story: National Human Genome Research Institute, NHGRI.gov
