NEW YORK (GenomeWeb) – The Mayo Clinic's Center for Individualized Medicine has designed a system for integrating pharmacogenomic testing into patients' electronic medical records in such a way as to provide real-time alerts to physicians who are prescribing medications.
The researchers described the system in a publication in Genetics in Medicine last month. Pedro Caraballo, lead author of the study, told GenomeWeb that advances in genomic technology has led to many new discoveries about how genomics relates to health and disease and decreasing costs have made it possible to use next-generation sequencing in clinical settings. "However, translating all that information to patient care has been extremely difficult," Caraballo, also a clinical informaticist and internist at the Mayo Clinic, told GenomeWeb.
In exploring ways in which they could apply genomics to patient care, the Mayo researchers thought that pharmacogenomics would be a good place to start. "It is an area that is more ready to be implemented," Caraballo said. "It is relatively straightforward and the information is there."
Researchers at the Mayo have been studying how to implement PGx testing in a clinical setting for a number of years now. The institution has been collaborating with Baylor College of Medicine to sequence 84 genes related to drug metabolism for more than 10,000 patient samples in the Mayo's biobank from patients who have consented to research.
The Mayo Clinic is also part of the Electronic Medical Records and Genomics (eMERGE) consortium, which is studying best practices for implementing NGS-based PGx testing into electronic medical records.
This recent Genetics in Medicine publication, however, describes how the Mayo was able to implement its alert system for specific drugs and pharmacogenetic variants at the point of care.
The Mayo team wanted to design a system that could be easily implemented by other laboratories that wanted to do PGx testing, Caraballo said. "We tried to develop a model of implementation that other institutions could use, even those with less resources," he said.
The researchers approved the use of 20 different gene-drug interactions to integrate into medical records for such drugs like clopidigrel, tamoxifen, codeine, and simvastatin. Between August 2012 and June 2015, 18 were implemented at the point of care.
During the study period, 1,247 unique providers, including physicians, residents/fellows, physician assistants, nurse practitioners, and pharmacists from multiple clinical areas interacted with PGx clinical decision support interventions, which were triggered for more than 3,000 patients. Alerts were delivered to physicians either when he or she attempted to order a drug for a patient or following an actionable PGx test result.
The researchers also developed eleven different educational resources to accompany selected drug-gene interactions in order to describe the interaction and what actions should be taken for specific genotypes. In addition, the team developed models specifically for pharmacists, including a general pharmacogenomics module, and four specific to various drugs or disease areas. Out of 500 pharmacists at the Mayo, 422 completed the general pharmacogenomics competency-based module, while between 247 and 415 completed each of the other programs.
Overall, Caraballo said that the model was successful, although there were a number of challenges and there will continue to be challenges as additional drug-gene interactions are discovered.
For one thing, the system had to be accessible and understandable by a broad range of providers — including general practitioners, specialists, nurse practitioners, and pharmacists.
In addition, the group designed the model to accommodate both preemptive testing, as well as reactive testing. Preemptive testing is the model being explored by the eMERGE consortium, which is analyzing PGx genes in patients who are deemed likely to be prescribed a drug for which there are genetic variants that would impact treatment. However, most PGx testing currently performed is done only when a patient needs a certain drug, if it is done at all.
In designing the model, the researchers first put together a multidisciplinary team of experts. The team decided on the initial 20 drug-gene interactions to implement and was also responsible for overseeing the implementation and evaluating the clinical impact of each drug-gene interaction.
One challenge in scaling the model up is that the process of choosing the drug-gene interactions and creating the alert system is time consuming and labor intensive, the authors wrote. In addition, there were often disagreements regarding the impact of PGx testing and using PGx results to influence patient management.
The authors wrote that there is a need for a "national consensus between PGx experts and medical societies in charge of the clinical guidelines to widely disseminate standardized PGx knowledge that can be easily accepted by clinicians and quickly implemented in clinical practice."
One challenge in applying this same model to other institutions will be in the lack of standards between different labs in how PGx results are communicated and in the nomenclature used.
Most commercial electronic health records are also not designed to handle genomic information, particularly over the long term. The Mayo team created a model that would enable both electronic and manual data entry.
Another future challenge, Caraballo added, will be in updating the database and alert system with newly discovered gene-drug interactions. Currently, he said, it is a manual process and likely will be for some time, he said.
Caraballo said he is now working on a follow up publication that describes the "implementation of the clinical support decision rules into the electronic medical record," he said. "There are a lot of technical issues that we had to deal with and hope that others could benefit by learning from our mistakes and successes," he said.