Implementation of clopidogrel pharmacogenetic clinical decision support for a preemptive return of results program

Abstract

Purpose

To describe our experiences implementing and iterating CYP2C19 genotype–guided clopidogrel pharmacogenetic clinical decision support (CDS) tools over time in the setting of a large health system–wide, preemptive pharmacogenomics program.

Summary

Clopidogrel-treated patients who are genetically predicted cytochrome P450 isozyme 2C19 (CYP2C19) intermediate or poor metabolizers have an increased risk of atherothrombotic events, some of which can be life-threatening. The Clinical Pharmacogenetics Implementation Consortium provides guidance for the use of clopidogrel based on CYP2C19 genotype in patients with cardiovascular and cerebrovascular diseases. Our multidisciplinary team implemented an automated, interruptive alert that fires when clopidogrel is ordered or refilled for biobank participants with structured CYP2C19 intermediate or poor metabolizer genomic indicators in the electronic health record. The implementation began with a narrow cardiovascular indication and setting and was then scaled in 4 primary dimensions: (1) clinical indication; (2) availability across health-system locations; (3) care venue (e.g., inpatient vs outpatient); and (4) provider groups (eg, cardiology and neurology). We iterated our approach over time based on evolving clinical evidence and proactive strategies to optimize CDS maintenance and sustainability. A key facilitator of expansion was socialization of the broader pharmacogenomics initiative among our academic medical center community, accompanied by clinician acceptance of pharmacogenetic alerts in practice.

Conclusion

A multidisciplinary collaboration is recommended to facilitate the use of CYP2C19 genotype–guided antiplatelet therapy in patients with cardiovascular and cerebrovascular diseases. Evolving clopidogrel pharmacogenetic evidence necessitates thoughtful iteration of implementation efforts and strategies to optimize long-term maintenance and sustainability.

KEY POINTS.

• A multidisciplinary approach for preemptive return of CYP2C19 genotyping results and design of clopidogrel pharmacogenetic clinical decision support tools was utilized for implementation at UCHealth.

• Clopidogrel pharmacogenetic evidence has evolved over time; therefore, a multidisciplinary effort is recommended to facilitate CYP2C19 genotype–guided antiplatelet therapy in patients with cardiovascular and cerebrovascular diseases.

• Keys to success for clopidogrel pharmacogenetic implementation initiatives include early and iterative engagement of clinicians, prioritization of resources required for ongoing maintenance, and sharing of successes across the organization.

Clopidogrel is an antiplatelet agent used to treat patients with acute coronary syndrome (ACS) (with or without percutaneous coronary intervention [PCI]), neurovascular indications (eg, acute ischemic stroke), and peripheral arterial disease. Clopidogrel is metabolized, in part, by cytochrome P450 isozyme 2C19 (CYP2C19). Genetically predicted CYP2C19 intermediate metabolizers (IMs) and poor metabolizers (PMs) have decreased conversion of clopidogrel to its active metabolite, thereby reducing efficacy.¹ As a result, clopidogrel-treated patients who are genetically predicted CYP2C19 IMs or PMs have an increased risk of atherothrombotic events, some of which can be life-threatening.^2,3 Studies have demonstrated that CYP2C19 genotype–guided antiplatelet strategies lower the risk of major adverse cardiovascular events when compared with non–genotype-guided antiplatelet therapy.^4-7

The Clinical Pharmacogenetics Implementation Consortium (CPIC) issued a guideline for CYP2C19 genotype and clopidogrel therapy in 2011 and an update in 2013, both of which focused on patients undergoing PCI for ACS.^8,9 The clinical recommendation was to avoid clopidogrel in CYP2C19 IMs and PMs and use an alternative antiplatelet agent (eg, ticagrelor, prasugrel) if no contraindications were present. The CPIC guideline was updated again in 2022, with the inclusion of expanded cardiovascular and neurovascular indications.¹⁰ The evolving pharmacogenomic knowledge base has changed the scope of implementation initiatives, expanding the reach and potential benefits of genotype-guided interventions such as those observed with CYP2C19 variants and antiplatelet therapy in patients with cardiovascular and cerebrovascular diseases.

Clopidogrel remains widely prescribed in many settings worldwide, even with the availability of more potent antiplatelet agents (ie, ticagrelor and prasugrel). This is primarily due to low cost, decreased bleeding risk, and absence of dyspneic adverse effects. As a result, clopidogrel is one of the most frequently prescribed medications with pharmacogenetic guidance, making it an attractive use-case for implementation.^11,12 Accordingly, a study of 15 institutions implementing pharmacogenetic testing showed that preemptive use of CYP2C19 genotyping to inform drug prescribing was common, with clopidogrel being one of the top medications for which clinical guidance was provided.¹³ Additional work demonstrated that using CYP2C19 genotyping to guide antiplatelet therapy also has implications beyond the cardiovascular setting, with over half of genotyped patients being prescribed another medication impacted by CYP2C19 in the following year.¹⁴ Thus, while CYP2C19-targeted implementation efforts often initially focus on clopidogrel, they must also consider the future implications for other clinically actionable drugs affected by CYP2C19 that a patient may receive long after their initial pharmacogenetic test.

Here, we describe the process used to return CYP2C19 genotype results for a preemptive pharmacogenomics program and our experience implementing and iterating clopidogrel pharmacogenetic clinical decision support (CDS) tools, particularly as evidence evolved over time.

Overview of the pharmacogenomics program and setting

Our preemptive return of clinical pharmacogenetic results initiative is focused on individuals who have participated in a health system–wide research biobank.¹⁵ The Biobank at the Colorado Center for Personalized Medicine (CCPM Biobank) was established at the University of Colorado Anschutz Medical Campus in partnership with UCHealth, with the goal of advancing genomics research and personalized medicine across the Rocky Mountain region.^15,16 UCHealth is comprised of 12 hospitals, more than 2,000 inpatient beds, and over 150 clinics across Colorado, with affiliated hospitals in Colorado, Wyoming, and Nebraska. The academic medical center is headquartered at University of Colorado Hospital on the Anschutz Medical Campus in Aurora, CO.

For the initiative, UCHealth patients 18 years of age or older may enroll in the biobank via an electronic self-consent model through their UCHealth online patient portal, My Health Connection (Epic Systems Corporation, Verona, WI). A single consent form approach has been used since 2019 to authorize consent for research, participant recontact, use in industry partnerships, and return of clinical genetic test results to the electronic health record (EHR). Following consent, a blood sample is collected at the patient’s next routine clinical blood draw. Saliva samples were added as a collection method in 2021 at select clinic locations and events. As of July 2023, over 227,000 participants have enrolled in the CCPM Biobank, close to 129,000 have provided a blood or saliva sample for genotyping, and more than 84,000 are eligible for clinical return of results based on their consent status.

Collected biospecimens are sent to the CCPM Biobank Laboratory at the University of Colorado Anschutz Medical Campus. The CCPM Biobank Laboratory is College of American Pathologists (CAP) accredited and Clinical Laboratory Improvement Amendments (CLIA) certified for high-complexity testing. At the laboratory, genomic DNA is extracted and samples are genotyped using customized versions of genome-wide genotyping arrays, either the Infinium Expanded Multi-Ethnic Genotyping Array (Illumina, Inc., San Diego, CA), used before October 2022, or Illumina’s Infinium Global Diversity Array with Enhanced PGx (used after October 2022), which includes over 44,000 genome-wide pharmacogenetic markers spanning more than 2,000 pharmacogenetic targets.^16,17 A panel of selected pharmacogenetic variants has been validated for clinical use on the microarrays directly, permitting moderate- to high-throughput clinical testing. Subsequently, clinical pharmacogenetic test results and associated quality control metrics are evaluated and approved by a board-certified clinical molecular geneticist. Results are returned to UCHealth’s Epic EHR (Epic Systems Corporation) via a semiautomated process using the CCPM Translational Informatics Service, a results translation engine (BC Platforms, Zurich, Switzerland), and a custom interface with the Epic system. Results are filed as structured (discrete) data and used to drive downstream, drug-gene–specific CDS tools.

For clopidogrel, the initial CDS tool evaluated CYP2C19 diplotype (eg, *1/*3, denoting one normal and one no-function allele) based on the pattern of the string of text in the laboratory results record. However, in May 2022, UCHealth implemented Epic’s Genomics module. This involved transitioning in December 2021 from a standard results–type record to a more specific Molecular Result Type record in Epic’s laboratory information system, Beaker (Epic Systems Corporation). The system’s Molecular Result Type record format supports filing and mapping rich structured data (eg, detailed variant information) associated with genetic results. It also supports the use of patient genomic indicators. Genomic indicators are tags in distinct sections of the Epic EHR that provide summary-level information about a patient’s genetic results (Figure 1). Genomic indicators serve multiple functions and can be (1) used as criteria or structured data fields to trigger CDS alerts, (2) updated by the health system as knowledge and clinical workflows evolve, and (3) accompanied by links to clinician-friendly educational resources.¹⁸ Today, the clopidogrel pharmacogenetic alert criteria evaluate the presence or absence of structured genomic indicators (eg, CYP2C19 PM).

Figure 1.

Screenshot of patient genomic indicators displayed in the UCHealth electronic health record.

A key nonlaboratory facilitator of our overarching initiative is the CCPM Pharmacogenomics Implementation Committee Colorado (PICColo), a multidisciplinary committee with several working groups that carries out the clinical pharmacogenomic implementation process.¹⁵ Core functions include literature evaluation and prioritization of new drug-gene pairs, preimplementation clinical partner engagement (described below), standardized design and scalable integration of CDS tools, development and dissemination of clinician and patient educational materials, weekly monitoring of synchronous and asynchronous CDS, and annual evaluation of drug-gene pairs that are live across the health system.

As of July 2023, pharmacogenetic CDS tools, including both interruptive and noninterruptive alerts, were live across the UCHealth system for 30 medications affected by variants in CYP2C19, SLCO1B1, DPYD, CYP2C9, TPMT, NUDT15, and ABCG2. These medications include the following: for CYP2C19, clopidogrel, voriconazole, citalopram, escitalopram, proton pump inhibitors (dexlansoprazole, lansoprazole, omeprazole, and pantoprazole), brivaracetam, and clobazam; for SLCO1B1, atorvastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, and simvastatin; for DPYD, systemic 5-fluorouracil, topical 5-fluorouracil, and capecitabine; for CYP2C9, celecoxib, ibuprofen, meloxicam, piroxicam, phenytoin, fosphenytoin, and siponimod; for TPMT/NUDT15, azathioprine, 6-mercaptopurine, thioguanine; and for ABCG2, rosuvastatin.

Implementation considerations

The CCPM Biobank initiative is fully preemptive, such that a UCHealth patient is genotyped and clinical pharmacogenetic results are returned to the EHR asynchronously, ie, without any association with the patient’s current or past medication therapy. Importantly, a physician affiliated with CCPM and UCHealth, rather than the patient’s clinician, is the ordering provider for the pharmacogenetic test in our model. Patients’ clinicians are generally unaware of when results are filed in the EHR or that testing has occurred. Thus, it is critical that automated CDS tools are in place to notify clinicians of relevant drug-gene interactions and provide guidance at the point of prescribing.

CYP2C19 was selected as the first gene for return of results to the EHR. It was selected due to the known functional significance of the variants (ie, CYP2C19 alleles *2 [no function], *3 [no function], and *17 [increased function]), performance of the variants on the genotyping array, likelihood of detecting variant alleles in our population, large number of drugs affected by CYP2C19 variation, and corresponding prescribing frequencies of the affected drugs at our health system.¹⁵ In general, we also align our selection of variants with those designated by the Association for Molecular Pathology as tier 1, which for CYP2C19 includes alleles *2, *3, and *17.¹⁹ As with all genomic testing panels, the absence of a variant does not exclude the possibility that a rare variant exists. This is noted as such in the laboratory report, consistent with commonly accepted best practices in the field.^20,21 Clopidogrel was selected as the first medication for which a pharmacogenetic CDS tool would be surfaced to providers for CCPM Biobank participants. This decision was primarily driven by the strength of clopidogrel pharmacogenetic evidence, the availability of CPIC guidelines, the boxed warning on the US Food and Drug Administration (FDA)–approved label regarding CYP2C19 variants, and the high-risk nature of the clinical scenario.¹⁵

The clopidogrel pharmacogenetic CDS tool is an automated interruptive (“pop-up”) alert using the Epic system’s native CDS software (BestPractice Advisory [BPA]). The alert is synchronous in nature, meaning that it fires when clopidogrel is ordered or refilled for a patient with a CYP2C19 IM or PM phenotype in the EHR. The clinician must acknowledge the alert to complete the order. Over time, we have expanded our clopidogrel pharmacogenetic CDS tool based on several factors, including (1) indication, (2) geographic location, (3) setting (ie, inpatient vs outpatient), (4) evolving knowledge, and (5) maintenance/sustainability, as described below and summarized in Table 1.

Table 1.

Overview of Clopidogrel Pharmacogenetic CDS Tool Implementation and Iteration at UCHealth

Year	Indication	UCHealth region(s)	Care venue/setting	Provider group(s)	Lessons learned
2018	Elective ACS/PCI	Metro	Inpatient	Interventional cardiology	Starting with a small group of highly engaged clinical partners promoted acceptance of PGx and facilitated thoughtful, user-centered CDS tool design.
2019	Acute ACS/PCI	Metro	Inpatient	Cardiology	Post-PCI order sets allowed for indication-specific alert triggers, but the tradeoff was increased CDS tool maintenance across the health system.
	Elective and acute ACS/PCI	North, South	Inpatient	Cardiology
2020	Post PCI	Metro, North, South	Outpatient cardiology	Cardiology	Socialization and clinician acceptance of the initiative prompted expansion to outpatient cardiology settings.
2020	Any indication	Metro, North, South	Inpatient and outpatient cardiology	Cardiology, vascular surgery	Evolving clopidogrel PGx knowledge and continued program socialization facilitated indication expansion. At the same time, inpatient CDS builds were simplified (ie, order set dependency was removed), thus improving long-term maintenance and sustainability.
2021	Any indication	Metro, North, South	Inpatient and ambulatory care	Any provider	Evolving clopidogrel PGx knowledge and continued program socialization facilitated indication, care venue, and provider expansion.

Abbreviations: ACS/PCI, acute coronary syndrome/percutaneous coronary intervention; CDS, clinical decision support; PGx, pharmacogenomics.

Synchronous CDS

We began work on the clopidogrel pharmacogenetic CDS tool in 2018 with a small pilot implementation for a narrow indication and setting—elective ACS/PCI cases at one UCHealth Metro region cardiac catheterization laboratory. This approach allowed us to conduct preimplementation clinical partner engagement with a small group of 6 highly engaged cardiology clinicians. The group provided critical input on the design of the alert’s user interface (eg, default options for actionable orders, removal of empty space so the alert fits on one screen when accessing the EHR from a tablet). They provided guidance on the timing of clopidogrel prescribing within standard workflows, such as upon arrival to the inpatient floor following PCI or on discharge from the hospital. They also suggested the types of ancillary tools that would most efficiently communicate genetic results to other clinicians and patients. These tools included 3 Epic SmartPhrases (ie, abbreviations or words used to pull long phrases or paragraphs into a note) for clinicians to (1) add pharmacogenetic results to their clinical documentation, (2) notify primary care providers of pharmacogenetic results and reasons for using alternative antiplatelet therapy via the discharge summary, and (3) educate patients about pharmacogenetic results and any changes to antiplatelet therapy via the after-visit summary at discharge. A link to an internal website was added to the alert’s user interface to provide “just-in-time” clinician education.

Following the initial implementation, we proceeded to scale in 4 primary dimensions: (1) clinical indications (eg, elective vs acute PCI); (2) availability across clinical locations (eg, regions of the health system); (3) care venue (eg, inpatient vs outpatient); and (4) provider groups (eg, cardiology vs neurology).

In July 2019, we expanded the CDS indication to include acute ACS/PCI within the UCHealth Metro region and widened the system-level coverage by activating the CDS tool in the UCHealth North region for both elective and acute ACS/PCI. We subsequently expanded to the UCHealth South region several months later. Our initial clopidogrel BPA was triggered in part by post-PCI order sets. Although the workflow was the same in all 3 locations, the post-PCI order sets and identifiers were different, thus requiring separate CDS triggers for each region’s order sets. From this experience, we learned that although order sets can increase the specificity of CDS tools, particularly for indication-specific drug-gene examples, they came with the trade-off of requiring more build maintenance for our team, particularly across a large health system.

In May 2020, we implemented the CYP2C19-clopidogrel CDS tool in outpatient cardiology settings, specifically cardiology clinics, for the first time. Given the literature evidence at the time, we modified the language in the alert user interface to specify that the information applied to post-PCI cases only and did not provide recommendations for other indications. Later that year, our vascular surgery colleagues expressed interest in the potential utility of CYP2C19 genotype for clopidogrel therapy in peripheral arterial disease.²² Implementation for this indication was further supported by the FDA boxed warning in clopidogrel prescribing information, which recommends alternative antiplatelet therapy in CYP2C19 PMs, regardless of indication.^23,24 At the same time, several post-PCI order sets were being retired across the UCHealth system, necessitating that we revise our inpatient clopidogrel pharmacogenetic CDS build. To simplify our approach and improve long-term maintenance and sustainability, we removed our dependency on inpatient clopidogrel order sets and expanded the alert to display for any clopidogrel order, irrespective of indication, in the inpatient setting. The alert user interface language was modified to address the peripheral arterial disease indication and to acknowledge that data were not yet applicable for stroke/transient ischemic attack (TIA). This change simplified the build and maintenance while expanding the implementation of CYP2C19-clopidogrel CDS.

In September 2021, it became clear that evidence supported CYP2C19 genotype–guided clopidogrel prescribing in the setting of stroke/TIA.²⁵ Thus, we modified the alert user interface to address the increased risk of cardiovascular and cerebrovascular events in altered CYP2C19 metabolizers prescribed clopidogrel (Figure 2). While prasugrel is on formulary, it is not widely used in our health system and is contraindicated in stroke and TIA. Therefore, it is not listed as an actionable order in the BPA. At the same time, organizational readiness for more widespread pharmacogenetics implementation had grown, and we removed restrictions based on setting. The alert now fires for any clopidogrel order in any inpatient or ambulatory setting for any provider, thus completing our scaling efforts from a narrow indication involving a very small provider group in one center to a pharmacogenetic CDS application available to our entire health system.

Figure 2.

Clinical decision support alert for CYP2C19 and clopidogrel displayed to prescribers at UCHealth.

With our move to using Epic’s Genomics module in May 2022, we converted our existing clopidogrel CDS to activate based on CYP2C19 genomic indicators, which are filed for CCPM Biobank participants when a CYP2C19 result is entered into the EHR. We retroactively filed CYP2C19 genomic indicators for participants who had CYP2C19 results in their EHR prior to our Genomics module go-live. The decision to implement the Genomics module promoted the program’s long-term sustainability, as the indicators can be updated when new genomic knowledge emerges (eg, if a genotype-to-phenotype translation changes) and capitalized on new functionality for reporting genetic results. Along the same lines, in April 2022 we transitioned our clinician and patient educational materials for all aspects of the CCPM Biobank program to an externally facing website to further streamline our maintenance efforts (Figure 3).

Figure 3.

Screenshot of page on the CCPM Biobank’s clinician education website providing information on clopidogrel and CYP2C19.

Asynchronous CDS

An important consideration for the CYP2C19-clopidogrel implementation initiative was how to handle the preemptive return of results for patients who were already receiving clopidogrel but for whom a CDS alert had not fired yet (ie, the medication had not been reordered or refilled since the CYP2C19 result became available). In this setting, CDS would need to be asynchronous, ie, not timed with the order or refill of the medication. We developed an asynchronous monitoring protocol for these cases to identify high-risk scenarios, such as patients receiving clopidogrel who underwent PCI in the previous year. Specifically, a report is run following a large-scale results release to automatically screen for clopidogrel on the active medication lists of all patients who are CYP2C19 IMs or PMs. A CCPM-affiliated pharmacist evaluates the medical record of all identified cases, which are then reviewed and discussed by the pharmacist and a CCPM-affiliated cardiologist at weekly, virtual pharmacogenomics rounds. For patients who are deemed at high risk, the pharmacist sends the prescribing clinician an in-basket message notifying them of the CYP2C19 result and suggestions for alternative therapy. We are in the process of evaluating our asynchronous processes to determine which components can be further automated.

Postimplementation summary and lessons learned

From a programmatic perspective, as of July 2023, over 18,000 participants have received at least one clinical pharmacogenetic result, and over 55,000 patient genomic indicators have been filed in the EHR. Overall, participants in the CCPM Biobank are predominantly female (59%), white (85%), and non-Hispanic (88%); 51% are less than 50 years of age.¹⁶ The most common disorder among participants is hypertension.¹⁶ As of July 2023, over 3,300 distinct drug-gene interaction alerts have fired in clinical practice for 2,700 CCPM Biobank participants. The most common drug-gene interaction alerts are for proton pump inhibitors (55.2%), followed by statins (30.9%), escitalopram and citalopram (12.1%), clopidogrel (1.6%), fluoropyrimidines (<0.1%), voriconazole (<0.1%), and nonsteroidal anti-inflammatory drugs (<0.1%).

Of over 18,000 CYP2C19 genomic indicators filed in the EHR, the percentage of CCPM Biobank participants with CYP2C19 ultrarapid, rapid, normal, intermediate, and poor metabolizer phenotypes is 4.2%, 26.7%, 41.5%, 25.1%, and 2.6%, respectively. To date, 177 clopidogrel alerts have fired in clinical practice for 55 CCPM Biobank participants (93% CYP2C19 IMs and 7% CYP2C19 PMs). We have asynchronously screened 137 clopidogrel-treated CYP2C19 IMs and PMs and sent 8 in-basket messages to the prescribing clinicians.

Our CYP2C19-clopidogrel pharmacogenetic implementation started in a focused manner but became less specific over time, primarily driven by evolving evidence and strategies to optimize sustainability and long-term maintenance. A critical factor facilitating this transition was the increased socialization of the broader pharmacogenomics initiative, accompanied by end-user acceptance of drug-gene interaction alerts in clinical practice. Clinician acceptance was made possible through thoughtful CDS design across all drug-gene pairs, whereby only high-risk drug-gene pairs are targeted with interruptive alerts. In contrast, lower-risk drug-gene pairs are addressed with passive warnings. This has helped to minimize alert fatigue and cognitive load for busy clinicians. We have applied many of these lessons broadly across subsequent drug-gene alert implementations. One key practice is that we now aim from the outset to build a single alert that works for a given drug-gene pair across all indications, geographic locations, and settings. Indication-specific triggers are still sometimes warranted, but the maintenance was not scalable for our health system–wide implementation program. Instead, we have moved to surfacing indication-specific recommendations in the form of hyperlinked tip sheets within the user interface of our interruptive and passive CDS tools. Although the CDS alerts are less targeted, the centralized tip sheets can be updated easily with evolving knowledge or workflow changes.

Conclusion

As clopidogrel pharmacogenetic evidence evolves, so must implementation efforts. A multidisciplinary collaboration is recommended to facilitate the use of CYP2C19 genotype–guided antiplatelet therapy in patients with cardiovascular and cerebrovascular diseases. Clopidogrel pharmacogenetic CDS tools should be constructed in a manner that can accommodate changes recommended by multidisciplinary stakeholders.

An audio interview that supplements the information in this article is available on AJHP’s website at www.ajhpvoices.org.

Disclosures

This work was supported, in part, by National Institutes of Health (NIH) grants 1K23HL161352 (to K.E.T.) and 1K08HL125725 (to D.P.K.). The contents of this article are the authors’ sole responsibility and do not necessarily represent official NIH views. The authors have declared no potential conflicts of interest.