Abstract
Clopidogrel is first‐line for secondary stroke prevention, but its effectiveness is affected by CYP2C19 genetic variation. An American Heart Association (AHA) scientific statement concluded that CYP2C19‐guided P2Y12 inhibitor (P2Y12i) therapy is evidence‐supported, though real‐world application depends on implementation factors (e.g., timely, understandable results). This prospective observational cohort quality improvement project used data from three comprehensive stroke centers within one health system, each offering rapid CYP2C19 testing. Eligible patients had CYP2C19 testing ordered, a P2Y12i prescribed (clopidogrel or ticagrelor), or a P2Y12i indication. Consensus recommendations were clopidogrel for CYP2C19 normal/rapid/ultrarapid metabolizers (NM/RM/UM) and ticagrelor for intermediate/poor metabolizers (IM/PM). The primary target population was patients with acute ischemic stroke or TIA (AIS/TIA) prescribed a P2Y12i. The primary outcome was the proportion of discharge ticagrelor prescriptions (among P2Y12i users) with AIS/TIA, compared between IM/PM and NM/RM/UM groups. Secondary endpoints included test turnaround time (TAT), result availability before discharge, and alignment of discharge P2Y12i with consensus recommendations (PGx‐aligned prescribing). The evaluation period was September 2024–August 2025. Results were available for 467 patients; 451 (97%) had results before discharge, with a median TAT of 5.2 hours (interquartile range 3.3–9.6). Of 337 patients prescribed a P2Y12i at discharge, 296 had AIS/TIA. Clinicians more often selected ticagrelor over clopidogrel for IM/PMs than NM/RM/UMs (89 of 102 [87%] vs. 11 of 194 [5.7%]; RR 15.4; 95% CI: 8.6–27.4; P < 0.0001). PGx‐aligned prescribing occurred in 272 (92%) of 296 patients. Rapid CYP2C19 testing efficiently translates into tailored P2Y12i prescribing in clinical practice.
Study Highlights.
WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?
CYP2C19 poor and intermediate metabolism is associated with reduced bioactivation of clopidogrel. CYP2C19‐guided antiplatelet therapy reduces the risk of secondary stroke within 90 days.
WHAT QUESTION DID THIS STUDY ADDRESS?
For patients presenting with acute ischemic stroke or transient ischemic attack, is it feasible to implement rapid CYP2C19 genotyping that supports integration of genotype into antiplatelet prescribing decisions?
WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?
The results demonstrated PGx‐aligned antiplatelet prescribing is feasible in the real‐world and significantly influenced clinical decision making.
HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?
This project offers a practical example for other health systems seeking to operationalize guideline‐recommended use of CYP2C19 results to support patients with acute ischemic stroke or transient ischemic attack.
Acute ischemic stroke (AIS) and transient ischemic attack (TIA) occur in 795,000 and 240,000 patients, respectively, each year in the United States. 1 Approximately 5–10% of these patients experience secondary AIS or TIA, depending on patient‐specific risk factors and secondary‐prevention treatment strategies. American Heart Association (AHA) guidelines recommend clopidogrel and aspirin (dual antiplatelet therapy; DAPT) for patients based on event etiology and severity, including those with minor non‐cardioembolic AIS or high‐risk TIA. 2
Clopidogrel, a P2Y12 inhibitor (P2Y12i), is a prodrug that requires bioactivation by the liver enzyme CYP2C19 to exert its antiplatelet effect. 3 Therefore, clopidogrel's effectiveness in preventing recurrent stroke is compromised in patients who carry CYP2C19 no‐function alleles (also called loss‐of‐function alleles). Carriers of CYP2C19*2, and *3 alleles (i.e., intermediate or poor metabolizers [IM/PMs]) have a significantly increased risk of recurrent stroke compared to noncarriers (normal, rapid, or ultra‐rapid metabolizers [NM/RM/UMs]). 4 Ticagrelor is an alternative to clopidogrel. Ticagrelor, a more potent and faster‐acting P2Y12 inhibitor, is administered twice a day orally and is a reversible P2Y12 receptor inhibitor. Unlike clopidogrel, ticagrelor exerts its antiplatelet effect without requiring hepatic activation through liver enzymes. Ticagrelor has been shown to achieve greater platelet inhibition than clopidogrel, which may enhance ischemic protection at the cost of additional bleeding risk. 5 Dual ticagrelor and aspirin has demonstrated greater efficacy than aspirin alone in lowering the risk of stroke or death in patients with acute mild‐to‐moderate AIS or high‐risk TIA. 6
Given the effect of CYP2C19 genetic variability on clopidogrel response, the CHANCE‐2 trial investigated whether ticagrelor provides superior protection against recurrent stroke in patients with CYP2C19 IM/PM phenotypes and minor AIS or high‐risk TIA. This trial randomized 6,412 Chinese patients at the point of care and found ticagrelor was associated with a lower risk of recurrent stroke at 90 days compared to once‐daily clopidogrel (6.0% vs. 7.6%; HR 0.77; 95% CI: 0.61–0.94). Although patients on ticagrelor had similar rates of severe and moderate bleeding compared to clopidogrel, those on ticagrelor had a higher incidence of minor bleeding events (5.3% vs. 2.5%; HR 2.41; 95% CI: 1.81–3.20). 7 Genotyping provides a targeted approach to balance this trade‐off, allowing clinicians to tailor antiplatelet therapy and optimize ischemic protection while minimizing bleeding and pill burden.
A recent scientific statement from the AHA reviewed the use of CYP2C19 genetic testing to guide P2Y12i therapy, emphasizing its potential to optimize treatment for patients with conditions like coronary artery disease, peripheral arterial disease, and stroke. 8 This scientific statement determined that the “evidence to date supports CYP2C19 genetic testing before oral P2Y12i therapy,” though it also highlights significant implementation challenges that hinder the implementation of CYP2C19 genotyping in practice. For example, the ambitious methodology of the CHANCE‐2 trial featured a 1‐hour genotype turnaround time (TAT) with a device unavailable in the United States. The rapid TAT may be critical as the difference between ticagrelor and clopidogrel response appears to emerge within the first several days. 7 As noted by the AHA statement, this rapid test turnaround may not be feasible outside controlled settings, posing challenges to replicating its success in the real‐world. 4 , 7
Here, we describe the implementation of rapid CYP2C19 genetic testing as part of standard care across three comprehensive stroke centers within MedStar Health, a large health system in the United States. This implementation project was focused on feasibility and implementation over the first year. It aimed to evaluate whether CYP2C19 genotyping influences antiplatelet prescribing, hypothesizing IM/PMs are more likely to receive ticagrelor than NM/RM/UMs.
METHODS
Design and setting
This prospective observational cohort consisted of three comprehensive stroke centers within the MedStar Health system: MedStar Georgetown University Hospital (Site A; academic medical center), MedStar Franklin Square Medical Center (Site B; community medical center), and MedStar Washington Hospital Center (Site C; academic medical center). The study evaluated the implementation of rapid CYP2C19 testing and its potential effect on antiplatelet therapy over 1 year, September 2024–August 2025.
Implementation and patient population
The MedStar Health Pharmacogenomics (PGx) Initiative integrated rapid CYP2C19 testing into standard clinical care. The PGx Initiative includes an interdisciplinary group that aims to operationalize PGx testing across MedStar Health. 9 The project included patients presenting to a site that offers rapid CYP2C19 genotyping, who have CYP2C19 tests ordered, are prescribed a P2Y12i, or have an indication for a P2Y12i.
The neurology service's involvement in stroke care differed by site: at Sites B and C they acted as consultative services, and at Site A they served as the primary team. At all hospitals, the stroke team (e.g., neurologists, residents, advanced practice providers) ordered rapid CYP2C19 genotyping as part of routine clinical practice for eligible patients. Within the electronic health record (EHR), CYP2C19 genotype was available as a standalone order and was integrated into the AIS/TIA order sets. After the order was placed, trained nursing staff collected three buccal swab samples from the patient, following the test kit manufacturer's directions. The required buccal swab was unique to the assay and required a brief training prior to use. Sites universally opted to train superusers for sample collection, but specific strategies varied between sites. Site A used two workflows based on code stroke activation. For code strokes, a stroke responder nurse carried a backpack with swabs and collected samples when genotyping was ordered. For non‐code stroke orders, the bedside nurse contacted the stroke responder nurse (label listed phone number) to collect the sample. Site B used a unit charge nurse model where charge nurses on projected high‐utilization units (e.g., emergency department, stroke unit, neuro intensive care unit) were trained. Site C, like the non‐code stroke Site A workflow, instructed bedside nurses to call a stroke nurse trained to collect the sample when a test was ordered.
Once the sample was collected and delivered (e.g., via pneumatic tube) to the on‐site laboratory, the assay took ~1 hour to perform. The rapid CYP2C19 test used in this project was the Genomadix Cube®, a Class II, FDA 510(k)‐cleared assay with a reported PCR accuracy of 99.1%. 10 The Genomadix Cube is a thermal cycling instrument that ejects three assay‐specific buccal swab tips into a single‐use cartridge. The instrument then automatically performs DNA extraction, PCR amplification, and fluorescence‐based genotyping. The assay targeted all Association for Molecular Pathology (AMP) tier 1 CYP2C19 alleles, including *2, *3, and *17, to determine the patient's metabolizer status. 11 Prior to going live, site‐specific validation was performed in accordance with College of American Pathologists (CAP) accreditation and Clinical Laboratory Improvement Amendments (CLIA) standards, and on‐site training was conducted for laboratory personnel.
Results were uploaded to the EHR in a genetics section within laboratory results and automatically sent to the ordering provider. Antiplatelet therapy selection was based on Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines and internal experts (e.g., neurologists, PGx specialist, neurology pharmacy specialist). Consensus recommendations for antiplatelet therapy were based on CYP2C19 phenotype per CPIC guidelines, with clopidogrel for patients with NM/RM/UM phenotypes and ticagrelor for those with IM/PM phenotypes. 7 , 12 , 13 Patients with P2Y12i therapy matching these recommendations were considered to be prescribed PGx‐aligned therapy. P2Y12i therapy was initiated empirically; Sites B and C followed clinician discretion, while Site A recommended ticagrelor if results were not yet available.
Multiple strategies used to support the incorporation of CYP2C19 results into antiplatelet prescribing decisions included stakeholder engagement (e.g., pharmacy, nursing, lab, pathology), identification of physician champions, presentations at faculty meetings, integration of recommendations into treatment protocols, pharmacist interventions, and electronic clinical decision support (CDS; e.g., order set integration, pharmacist interventions). See Supplemental Text 1 for details on pharmacist interventions.
Treating clinicians retained discretion over final prescribing decisions. This pragmatic approach enabled prescribers to incorporate CYP2C19 genotype into a comprehensive set of clinical factors to consider for each patient (e.g., concomitant and prior medications, bleeding risk).
Analysis
The primary dataset was derived from an automated EHR report for patients with an ordered CYP2C19 test. These data were supplemented by local site data reported to the national Get With the Guidelines® (GWTG; NCT02693223) registry and manual EHR abstraction. 14 , 15 As directed by the GWTG abstraction guidelines, sites reported data to GWTG for all patients with a primary discharge diagnosis coded as stroke or TIA. For this work, site GWTG data were limited to those prescribed a P2Y12i at discharge to limit the population for whom CYP2C19 testing would be informative. Manual EHR abstraction was used to validate P2Y12i discharge prescriptions, collect missing data (e.g., ABCD2 score), and collect data for patients absent from the GWTG dataset (e.g., CYP2C19 ordered for non‐AIS/TIA indication, AIS/TIA was not the primary discharge diagnosis).
The primary outcome was the proportion of discharge ticagrelor prescriptions (among antiplatelet users) for AIS/TIA, compared across IM/PM vs. NM/RM/UM groups. The study was not powered for post‐discharge clinical outcomes, but data were collected as secondary outcomes, including secondary stroke, TIA, death, or severe bleeding, and a composite of all of the above (i.e., any significant clinical event). Mild bleeding, emergency department (ED) and hospitalizations were also recorded. Data were abstracted from the EHR. Consistent with the CHANCE‐2 trial, a 90‐day follow‐up period was used, and bleeding events were classified using GUSTO criteria. 7 , 16 Lastly, secondary hospitalizations within 30 days were recorded.
Several implementation outcomes were also assessed. 17 Fidelity was defined as the proportion of patients with AIS/TIA and CYP2C19 results with PGx‐aligned therapy; feasibility as the proportion with results returned before discharge. The test order‐to‐result TAT was an additional measure of feasibility. Overutilization and underutilization of CYP2C19 testing were also measured. These determinations were made retrospectively and may not reflect real‐time appropriateness, since evolving diagnostics (e.g., subsequent imaging) can alter the clinical picture. Overutilization was defined as the proportion of patients with CYP2C19 results not prescribed a P2Y12i.
Of patients with AIS/TIA discharged on a P2Y12i, reach (those with a CYP2C19 result) and underutilization (those without a CYP2C19 result) were also calculated. Because the primary cohort was restricted to patients with a CYP2C19 test ordered, there is a risk of selection bias if this same approach were used to estimate testing uptake. Therefore, to ensure a consistent denominator that includes both tested and untested patients, reach and underutilization were assessed using patients with an AIS/TIA treated with a P2Y12i per the GWTG dataset.
Statistical analysis
For the primary endpoint, the percentage of patients prescribed ticagrelor at discharge was compared between the IM/PM and NM/RM/UM groups using a chi‐square test. A relative risk with a 95% confidence interval was calculated, and statistical significance was defined as P < 0.05. The study was adequately powered (Supplemental Text 2 ). For the secondary endpoints, descriptive statistics, including counts and percentages, were used to summarize antiplatelet prescribing patterns, indications for ordering tests, and previously mentioned implementation metrics. PGx‐aligned prescribing was compared between the approaches to pharmacy intervention: initial consult‐based phase vs. the later pharmacy clinical surveillance (PCS) alert phase (see Supplemental Text 1 for details).
Ethics statement
The MedStar Health Research Institute Institutional Review Board reviewed and approved this project as a quality improvement initiative.
RESULTS
Primary patient population
Testing was live for 1 year (Site A; September 1, 2024–August 31, 2025), 36 weeks (Site B; December 23, 2024–August 31, 2025), and 23 weeks (Site C; March 25, 2025–August 31, 2025) at the data cutoff point. The primary patient population was 296 patients with CYP2C19 results, AIS/TIA, and a P2Y12i prescribed at discharge (Figure 1 ). Table 1 describes the baseline characteristics of this population, including by site. Almost half (47%) of patients identified as Black. A history of prior stroke or TIA was present in 39% of patients and hypertension in 75%. The population predominantly exhibited minor (NIHSS score ≤ 5) non‐cardioembolic AIS or high‐risk TIA (ABCD2 score ≥ 4).
Figure 1.
Consort diagram. ACS, acute coronary syndromes; CAD, coronary artery disease; CVA, cerebrovascular accident; IM, intermediate metabolizer; LAMA, left against medical advice; NM, normal metabolizer; PE, pulmonary embolism; PM, poor metabolizer; RM, rapid metabolizer; TIA, transient ischemic attack; UM, ultrarapid metabolizer.
Table 1.
Baseline characteristics
| Baseline characteristics of those with IS/TIA & P2Y12i at DC | Total (n = 296) | Hospital | ||
|---|---|---|---|---|
| #1 (Site A) (n = 130) | #2 (Site B) (n = 118) | #3 (Site C) (n = 48) | ||
| Age, years, median (IQR) | 67 (59, 76) | 66 (58, 75) | 70 (61, 76) | 67 (57, 73) |
| Sex, female | 155 (52) | 70 (54) | 60 (51) | 25 (52) |
| Race | ||||
| Black | 139 (47) | 66 (51) | 40 (34) | 33 (69) |
| White | 111 (38) | 42 (32) | 67 (57) | 2 (4.2) |
| Unknown | 20 (6.8) | 7 (5.4) | 6 (5.1) | 7 (15) |
| Other | 14 (4.7) | 10 (7.7) | 1 (0.85) | 3 (6.3) |
| Multiple races | 12 (4.1) | 5 (3.9) | 4 (3.4) | 3 (6.3) |
| Ethnicity | ||||
| Hispanic or Latinx | 9 (3.0) | 6 (4.6) | 1 (0.85) | 2 (4.2) |
| Non‐Hispanic or Latinx | 208 (70) | 92 (71) | 79 (67) | 37 (77) |
| Unknown | 79 (27) | 32 (25) | 38 (32) | 9 (19) |
| Past medical history | ||||
| Diabetes | 121 (41) | 47 (36) | 49 (42) | 25 (52) |
| Dyslipidemia | 156 (53) | 47 (36) | 83 (70) | 26 (54) |
| Hypertension | 222 (75) | 85 (65) | 96 (81) | 41 (85) |
| Prior stroke or TIA | 114 (39) | 49 (38) | 46 (39) | 19 (40) |
| Prior to admission medications | ||||
| Aspirin | 108 (36) | 32 (25) | 58 (49) | 18 (38) |
| Clopidogrel | 43 (15) | 17 (13) | 20 (17) | 6 (13) |
| Ticagrelor | 5 (1.7) | 2 (1.5) | 1 (0.85) | 2 (4.2) |
| Anticoagulant (e.g., apixaban, rivaroxaban, heparin) | 9 (3.0) | 3 (2.3) | 6 (5.1) | 0 (0) |
| Other antiplateleta | 2 (0.34) | 2 (1.5) | 0 (0) | 0 (0) |
| P2Y12i Indication | ||||
| AIS | 244 (82) | 97 (75) | 102 (86) | 45 (94) |
| TIA | 52 (18) | 33 (25) | 16 (14) | 3 (6.3) |
| NIHSSb | 2 (1, 6) | 2 (1, 6) | 2 (1, 4) | 3 (1, 7) |
| ABCD2 Scorec | 5 (4, 5) | 5 (4, 6) | 5 (4, 5) | 4 (4, 5) |
| Length of stay, days | 3.1 (1.8, 6.7) | 3.1 (1.9, 7.5) | 2.6 (1.3, 4.9) | 4.9 (2.1, 9.8) |
Percentages are reported with two significant figures and, due to rounding, may not add to 100%.
NIHSS, National Institutes of Health Stroke Scale; P2Y12i, P2Y12 inhibitor.
Dipyridamole (n = 1); prasugrel (n = 1).
For 243 of 244 patients with AIS.
For 52 of 52 patients with a TIA.
The P2Y12i at discharge for the primary patient population is reported in Table 2 . As hypothesized for the primary outcome, ticagrelor was more likely to be prescribed to IM/PMs compared to NM/RM/UMs (89 of 102 [87%] vs. 11 of 194 [5.7%]; RR 15.4; 95% confidence interval 8.6–27.4; P < 0.0001). PGx‐aligned prescribing occurred in 272 (92%) of 296 patients. The proportion of patients receiving PGx‐aligned prescribing was 94% (183 of 194) in NM/RM/UMs and 87% (89 of 102) in IM/PMs. Exploratory analyses on the impact of unaligned care on clinical outcomes are in Supplemental Text 3 and Tables S1 and S2 .
Table 2.
CYP2C19 phenotype and P2Y12i at discharge
| CYP2C19 phenotype | Total (n = 296), N (%) | P2Y12i at discharge | |
|---|---|---|---|
| Clopidogrel (%) | Ticagrelor (%) | ||
| Ultrarapid metabolizer | 14 (4.7) | 14 (100) | 0 (0) |
| Rapid metabolizer | 73 (25) | 68 (93) | 5 (6.9) |
| Normal metabolizer | 107 (36) | 101 (94) | 6 (5.6) |
| Intermediate metabolizer | 89 (30) | 11 (12) | 78 (88) |
| Poor metabolizer | 13 (4.4) | 2 (15) | 11 (85) |
Percentages are reported with two significant figures and, due to rounding, may not add to 100%. P2Y12i: P2Y12i.
Post‐discharge follow‐up data were available for 235 (79%) of 296 patients in the primary patient population. Of the 235 patients, significant clinical events within 90 days of discharge occurred in 13 (5.5%) patients. These significant clinical events included: 11 (4.7%) with a secondary stroke or TIA (secondary stroke [n = 9; 3.8%] or TIA [n = 2; 0.85%]), one death (unknown cause), and one major bleeding event (CYP2C19 NM prescribed clopidogrel). Mild bleeding was reported in 9 (3.8%) patients. Clinical outcomes by PGx‐alignment are presented in Table S1 . Acute healthcare utilization within 90 days occurred in 76 (32%) patients, which included hospitalizations (42, 18%) or emergency room visits (34, 14%). Hospitalizations within 30 days of discharge of the index encounter occurred in 21 (8.9%) patients (Table S3 ).
Implementation metrics
Expectations prior to go‐live were (i) test volume: four to five tests per week per site, (ii) test TAT: 8 to 12 hours, and (iii) CYP2C19 IM/PM phenotype frequency: 30%. Two sites met expected test volume metrics (Table 3 ). Site C met the expected volume after the cutoff point for this report (data not shown). CYP2C19 IM/PM phenotypes were slightly more common than projected (162 of 467; 35%). Results were returned before discharge for 451 (97%) of 467 patients. The median test TAT was between 3 and 5 hours. However, as shown in Table S4 , 94 (20%) of the 467 tests took more than the projected 12 hours to result, with variation in performance by site: Site A 188 of 235 (80%), Site B 117 of 158 (74%), and Site C 68 of 74 (92%).
Table 3.
Implementation metrics among patients with CYP2C19 results
| Sites | Weeks live | Tests ordered per week | Tests completed | Tests completed before discharge | Turnaround Time (h), Order to Result, Median (IQR1, IQR3) | CYP2C19, IM & PMs |
|---|---|---|---|---|---|---|
| Site A | 52 | 4.5 | 235 | 228 (97%) | 5.8 (3.9, 11) | 73 (31%) |
| Site B | 36 | 4.4 | 158 | 151 (96%) | 4.8 (3.0, 12) | 56 (35%) |
| Site C | 23 | 3.2 | 74 | 72 (97%) | 3.6 (2.5, 6.0) | 33 (45%) |
| Overall | N/A | 4.3 | 467 | 451 (97%) | 5.2 (3.3, 9.6) | 162 (35%) |
Percentages are reported with two significant figures and, due to rounding, may not add to 100%.
IM, intermediate metabolizer; IQR, interquartile; N/A, not applicable; PM, poor metabolizer.
Figure 1 shows 526 tests were ordered, and 467 (88.8%) of those tests were completed. Incomplete tests were either canceled (52) or had indeterminate results (7). Reasons for cancellation were alternative diagnosis identified (e.g., stroke ruled out; 19), P2Y12i not indicated for stroke etiology (11), patient discharged before result (6), with additional reasons available in Table S5 . Indeterminate results were initially returned for 17 patients, of whom 10 received a conclusive result after a second sample was collected then processed; 7 did not have additional testing attempted.
Patients prescribed a P2Y12i at discharge
Of the 467 with test results, 337 (72%) were prescribed a P2Y12i at discharge (Figure 1 ). The 130 (28%) patients not prescribed a P2Y12i at discharge were considered overutilization (Table 4 ). Overutilization test proportions were similar across sites: 76 of 235 (32%) Site A, 32 of 158 (20%) Site B, and 22 of 74 (30%) Site C. Corollary analyses examined underutilization of tests and reach among the 364 patients with a primary discharge diagnosis of stroke or TIA (Table 4 ). Overall, test reach was 245 (67%), and underutilization was 119 (33%). Test reach was highest in the first site launched, Site A, which tested 117 (94%) of 125 AIS/TIA patients prescribed a P2Y12i at discharge. Expansion sites, Site B (59%, 97 of 165) and Site C (42%, 31 of 74), had lower test reach.
Table 4.
System‐wide implementation metrics
| Implementation outcome | Analysis population | Definition | Result, n (%) | |
|---|---|---|---|---|
| Feasibility | All with CYP2C19 results (regardless of indication) | 467 | Proportion of patients with CYP2C19 results returned prior to discharge | 451 (97%) |
| Overutilization | Proportion of patients with CYP2C19 results who were not prescribed a P2Y12i | 130 (28%) | ||
| Fidelity | All with CYP2C19 results & P2Y12i at discharge | 337 | Proportion of patients with AIS/TIA and CYP2C19 genotype results who received PGx‐aligned therapy | 310 (92%) |
|
AIS/TIA & CYP2C19 results & P2Y12i at discharge (primary study cohort) |
296 | 272 (92%) | ||
| Underutilization | Primary discharge diagnosis was AIS/TIA (GWTG cohort) & P2Y12i at discharge | 364 | Proportion of patients without CYP2C19 genotype results | 119 (33%) |
| Reach | Proportion of patients with CYP2C19 genotype results | 245 (67%) | ||
PGx‐aligned prescribing occurred in 310 (92%) of the 337 patients prescribed a P2Y12i at discharge. The most common reasons for unaligned care were: decision per the primary, non‐neurology, team (7, 29%), operations (e.g., result released near/after discharge; 5, 21%), and medication cost (5, 21%; Table S6 ).
Pharmacist involvement
During the first 6 months, inpatient consultation notes from pharmacists with PGx training were generated for 139 patients. The median TAT between test result and consult note completion was 7.8 hours (interquartile range [IQR] 1.4–19.0). The proportion of unaligned care ~doubled when these consults were not provided: consult (4 of 92 [4.9%]) vs. no consult (23 of 245 [9.3%]), and a similar effect was seen in those with AIS/TIA prescribed a P2Y12i at discharge (Table S7 ). Descriptive data suggested a possible site effect; PGx‐aligned prescribing appeared higher at Site B when consult notes were present (Table S8 ), but small sample size limited further analysis. PCS alerts triggered for 99 patients later discharged on a P2Y12i, prompting a change to PGx‐aligned therapy in 81% overall, with higher alignment among NM/RM/UMs than IM/PMs (Table S9 ). Inter‐site alignment with PCS alerts was similar across sites (Table S8 ).
DISCUSSION
This project describes the real‐world feasibility and prescribing impact of implementing rapid CYP2C19 genotyping to guide antiplatelet therapy across a multisite health system. This is the largest known cohort of patients with AIS/TIA receiving real‐world implementation of CYP2C19 genotyping. 18 The use of the in‐house CYP2C19 genotyping was feasible as shown by result availability prior to discharge in 97% of patients and providers incorporating the results for PGx‐aligned prescribing in over 90% of patients. This translates to patients with AIS/TIA being 15× more likely to receive ticagrelor if they have a CYP2C19 IM/PM phenotype than if they have an NM/RM/UM phenotype.
The AHA and American Stroke Association (ASA) recommend DAPT for patients with high‐risk AIS/TIA. 19 After data collection for this work was completed, the updated AHA/ASA Guidelines were published and included a new recommendation that ticagrelor is preferred over clopidogrel in patients with a CYP2C19 no‐function allele (also called loss‐of‐function). However, this guideline limited the recommendation to the CHANCE‐2 eligibility criteria (i.e., AIS with NIHSS ≤ 3 or TIA with ABCD2 ≥ 4) rather than all AIS/TIA who receive a P2Y12i. 7 These real‐world (RW) data demonstrate clinician interest in using additional evidence to expand testing criteria beyond those defined by CHANCE‐2. 12 Clinicians applied CYP2C19‐guided care in a broader population, including in NIHSS 4–5. This has important RW implications, as clinicians routinely prescribe P2Y12i for AIS with an NIHSS of 4–5 based on AHA guidelines and the INSPIRES and THALES trials. 19 , 20 , 21 Given the preponderance of evidence from clinical pharmacology, observational evidence, and clinical trials, the evidence supports the role for CYP2C19‐guided therapy in patients beyond the CHANCE‐2 population. 12
A prior AHA scientific statement described implementation barriers for CYP2C19 testing, such as limited clinician education and integration of results into the EHR. 8 Given the short length of stay for patients with AIS/TIA, it was unclear whether rapid CYP2C19 testing could return results before discharge. The CHANCE‐2 trial notably achieved test TAT under 80 minutes for nearly all participants. 7 These RW results suggest in‐house testing was a feasible approach, with a median TAT of 5 hours. This contrasts with 7–10‐day TAT for standard send‐out tests, which result after patients are discharged (median length of stay 3 days) and miss the most important window for benefit.
The TAT data (Table S4 ) offer several lessons. The rare instances of post‐discharge result availability (3%) likely reflect a short order‐to‐discharge window for these patients (median: 4.3 hours). Nights, holidays, and weekends had a marginal effect on TAT, likely clinically insignificant. While most patients had results before discharge, TAT exceeded the 12‐hour threshold in 20% of patients, with delays mainly due to sample collection, likely because bedside nurses relied on superusers. After the one‐year period, Site B began training all nurses on high‐volume units to address this, with future analyses to assess impact.
These data are also notable for the high utilization of results, as evidenced by 310 (92%) patients receiving a P2Y12i aligned with CYP2C19 results. This high fidelity is likely driven by physician champions (i.e., site stroke directors) advocacy for the intervention, coupled with integration of recommendations into treatment protocols for AIS/TIA. This support enabled dedicated time for stakeholder engagement and training. Additional support was provided through multiple forms of electronic CDS. Support was predominantly passive, relying on non‐interruptive CDS such as inclusion of CYP2C19 genotype test ordering within AIS/TIA order sets and pharmacist services.
During the project's design, the extent of pharmacist services required to support integration of CYP2C19 genotyping into care was unclear. Support evolved from the more labor‐intensive approach of providing individual consults within 1 business day by a PGx expert to a population‐monitoring model that integrated PGx services into existing clinical pharmacists' roles that routinely monitored the PCS tool for other clinical tasks (e.g., renal dosing). Overall, PCS alerts were triggered for 99 patients with initially unaligned care, but 81 (81%) were changed to PGx‐aligned therapy. This is a well‐performing intervention compared to other population management pharmacist tools. 22 Some of the unaligned therapy may be outside of the control of pharmacists, as 10 (42%) instances of unaligned care would not have been addressable by pharmacist intervention (e.g., result released near/after discharge; medication cost; Table S6 ). Small sample sizes limit definitive conclusions, but PGx expert consult notes may be associated with slightly higher PGx‐aligned prescribing than the population monitoring tool, an effect that may be more pronounced at community health centers, where neurology serves a consultative role for AIS/TIA patients.
Interruptive alerts are commonly used in PGx and other factors affecting prescribing. 23 , 24 This implementation is unique as it did not utilize prescriber‐facing interruptive alerts. The antiplatelet prescribed at discharge aligned with PGx recommendations for 310 patients (92%). However, there may be a role for interruptive alerts to marginally enhance PGx‐aligned prescribing or facilitate faster time to PGx‐aligned prescribing. After the 1‐year evaluation period described in this report, interruptive alerts were activated to target providers ordering clopidogrel for a CYP2C19 IM/PM (Figure S2 ).
This report characterizes testing under‐ and overutilization. Overutilization is hard to define since differential diagnoses evolve with new clinical details (e.g., imaging), and a natural tradeoff exists between over‐ and under‐testing depending on when in the workflow tests are ordered (e.g., stroke workup negative vs. already discharged). Delaying testing decisions may also cause missed orders, worsening undertesting.
For reach (proportion of AIS/TIA patients with CYP2C19 results), the top‐performing site (Site A, 94%) had neurology as the primary service, whereas Sites B (59%) and C (42%) used consultative models. Similarly, unaligned care was most often driven by the non‐neurology primary team's decision. Site A also had the most pharmacist consult notes (74 vs. 18 at Site B, 0 at Site C), likely improving PGx‐aligned prescribing and possibly having an unintended effect of reminding clinicians to order testing on future patients.
This initiative has several limitations. It was confined to a single health system with a relatively small sample size. A stronger design would compare outcomes against a non‐genotyped control cohort, though this seems unjustified given CHANCE‐2's robust demonstrated benefit in over 6,000 randomized participants. 7 The real‐world, EHR‐based dataset also lacked adherence and out‐of‐region event data. Finally, this study doesn't address cost‐effectiveness, though this is well‐documented for CYP2C19‐guided P2Y12i therapy, and testing is covered by Medicare and other payers. 25 , 26 , 27 , 28 , 29 , 30 Given this evidence, the health system judged the clinical and readmission‐cost benefits outweighed the marginal cost of testing.
Several lessons emerged. Clinicians, especially neurologists, showed strong support for integrating rapid CYP2C19 testing into care. Despite shared health‐system infrastructure, performance varied across sites (e.g., intervention reach). This suggests the implementation method may help balance the inherent under‐ vs. over‐utilization tradeoff. Future studies should also explore the roles of neurology (primary vs. consultative) and pharmacists (expert consult vs. population monitoring). The initiative underscored the need for interdisciplinary engagement: physicians, pharmacists, pathology, lab staff, executive leadership, and nursing were all vital to its success. Further research should weigh the value of studying clinical and financial outcomes tied to PGx‐aligned P2Y12i prescribing.
CONCLUSION
These RW data have several strengths, including the pragmatic approach to implementation in a diverse population. Rapid CYP2C19 genotyping was successfully implemented across three comprehensive stroke centers within MedStar Health. The results demonstrated PGx‐aligned antiplatelet prescribing is feasible in the real‐world and significantly influenced clinical decision making. Future directions include expanding the study to assess the impact of CYP2C19‐guided therapy in other clinical settings and using different approaches to implementation.
FUNDING
No funding was received for this work.
CONFLICTS OF INTEREST
D.M.S., P.R.R., S.A., J.B., A.C.B., V.L.B., R.A.B., J.B., J.C., S.D., R.A.G., E.D., K.D., N.G., A.W.H., J.M., K.M., J.R., V.P.S.J., S.S., N.P.S., P.C.T., A.W., D.J.Z., J.C.Z., A.S.: Nothing to disclose. S.M.S.: Seagen Stock, Immunome Stock, Napo Pharmaceuticals Scientific Advisory Board, Honoraria/consulting TerSera, Roche/Genentech, AstraZeneca, Chugai Pharmaceuticals, MSD. In kind 3rd party writing AstraZeneca.
DISCLOSURE
This work represents the authors' independent analysis of local or multicenter data gathered using the AHA Get With The Guidelines® (GWTG) Patient Management Tool but is not an analysis of the national GWTG dataset and does not represent findings from the AHA GWTG National Program. The authors disclose that Genomadix provided instrument placement for 1 year prior to the institutional purchase of the equipment. The institution independently purchased all testing reagents throughout the study period. Genomadix had no role in the design or conduct of the study; collection, analysis, or interpretation of the data; preparation of the manuscript; or the decision to submit the manuscript for publication.
AI STATEMENT
ChatGPT and Grammarly were used to assess clarity and grammar.
AUTHOR CONTRIBUTIONS
D.M.S., P.R.R., and R.A.B. wrote the manuscript. D.M.S., P.R.R., A.C.B., R.A.B., S.D., and S.M.S. designed the research. S.A., J.B., V.L.B., R.A.B., J.B., J.C., R.A.G., E.D., K.D., N.G., A.W.H., J.M., K.M., J.R., V.P.S.J., S.S., N.P.S., P.C.T., A.W., D.J.Z., J.C.Z., and A.S. performed the research. D.M.S. and P.R.R. analyzed the data.
Supporting information
Data S1
ACKNOWLEDGMENTS
Beth N. Peshkin for thoughtful review of the manuscript.
References
- 1. Palaniappan, L.P. et al. 2026 heart disease and stroke statistics: a report of US and global data from the American Heart Association. Circulation 153, e275–e906 (2026). [DOI] [PubMed] [Google Scholar]
- 2. Kleindorfer, D.O. et al. 2021 guideline for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline from the American Heart Association/American Stroke Association. Stroke 52, e364–e467 (2021). [DOI] [PubMed] [Google Scholar]
- 3. Plavix(R) [package insert]. Morristown, NJ: Sanofi‐Aventis U.S. LLC; 2025. Accesssed March 11, 2026.
- 4. Wang, Y. et al. Association between CYP2C19 loss‐of‐function allele status and efficacy of clopidogrel for risk reduction among patients with minor stroke or transient ischemic attack. JAMA 316, 70–78 (2016). [DOI] [PubMed] [Google Scholar]
- 5. Storey, R.F. et al. Inhibition of platelet aggregation by AZD6140, a reversible oral P2Y12 receptor antagonist, compared with clopidogrel in patients with acute coronary syndromes. J. Am. Coll. Cardiol. 50, 1852–1856 (2007). [DOI] [PubMed] [Google Scholar]
- 6. Johnston, S.C. et al. Ticagrelor and aspirin or aspirin alone in acute ischemic stroke or TIA. N. Engl. J. Med. 383, 207–217 (2020). [DOI] [PubMed] [Google Scholar]
- 7. Wang, Y. et al. Ticagrelor versus clopidogrel in CYP2C19 loss‐of‐function carriers with stroke or TIA. N. Engl. J. Med. 385, 2520–2530 (2021). [DOI] [PubMed] [Google Scholar]
- 8. Pereira, N.L. et al. CYP2C19 genetic testing for Oral P2Y12 inhibitor therapy: a scientific statement from the American Heart Association. Circulation 150, e129–e150 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Smith, D.M. et al. Pharmacogenetics in practice: estimating the clinical actionability of pharmacogenetic testing in perioperative and ambulatory settings. Clin. Transl. Sci. 13, 618–627 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Genomadix Inc . 510(k) substantial equivalence determination decision summary: Genomadix Cube CYP2C19 System K220026. Silver Spring (MD): Food and Drug Administration; 2023. <https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K220026>. Accessed April 8, 2026.
- 11. Pratt, V.M. et al. Recommendations for clinical CYP2C19 genotyping allele selection: a report of the Association for Molecular Pathology. J. Mol. Diagn. 20, 269–276 (2018). [DOI] [PubMed] [Google Scholar]
- 12. Lee, C.R. et al. Clinical pharmacogenetics implementation consortium guideline for CYP2C19 genotype and clopidogrel therapy: 2022 update. Clin. Pharmacol. Ther. 112, 959–967 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Wang, Y. et al. Clopidogrel with aspirin in high‐risk patients with acute non‐disabling cerebrovascular events II (CHANCE‐2): rationale and design of a multicentre randomised trial. Stroke Vasc. Neurol. 6, 280–285 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Fonarow, G.C. et al. Characteristics, performance measures, and in‐hospital outcomes of the first one million stroke and transient ischemic attack admissions in get with the guidelines‐stroke. Circ. Cardiovasc. Qual. Outcomes 3, 291–302 (2010). [DOI] [PubMed] [Google Scholar]
- 15. American Heart Association . Get with the guidelines‐heart failure registry. NCT02693509. <https://clinicaltrials.gov/study/NCT02693509>. Accessed April 4, 2026.
- 16. Investigators G . An international randomized trial comparing four thrombolytic strategies for acute myocardial infarction. N. Engl. J. Med. 329, 673–682 (1993). [DOI] [PubMed] [Google Scholar]
- 17. Proctor, E. et al. Outcomes for implementation research: conceptual distinctions, measurement challenges, and research agenda. Adm. Policy Ment. Health 38, 65–76 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Patel, T.J. et al. Implementing CYP2C19‐guided clopidogrel therapy: a scoping review of pharmacogenomic testing services. Pharmacogenomics J. 25, 12 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Prabhakaran, S. et al. 2026 guideline for the early management of patients with acute ischemic stroke: a guideline from the American Heart Association/American Stroke Association. Stroke 57, e316–e436 (2026). 10.1161/STR.0000000000000513. [DOI] [PubMed] [Google Scholar]
- 20. Alhajala, H. et al. Expansion of Telestroke coverage in community hospitals: unifying stroke care and reducing transfer rate. Ann. Neurol. 95, 576–582 (2024). [DOI] [PubMed] [Google Scholar]
- 21. Zachrison, K.S. et al. Association of hospital telestroke adoption with changes in initial hospital presentation and transfers among patients with stroke and transient ischemic attacks. JAMA Netw. Open 4, e2126612 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Barnes, G.D. et al. Pharmacist use of a population management dashboard for safe anticoagulant prescribing: evaluation of a Nationwide implementation effort. J. Am. Heart Assoc. 13, e035859 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Smith, D.M. , Wake, D.T. & Dunnenberger, H.M. Pharmacogenomic clinical decision support: a scoping review. Clin. Pharmacol. Ther. 113, 803–815 (2022). [DOI] [PubMed] [Google Scholar]
- 24. Wake, D.T. , Smith, D.M. , Kazi, S. & Dunnenberger, H.M. Pharmacogenomic clinical decision support: a review, how‐to guide, and future vision. Clin. Pharmacol. Ther. 112, 44–57 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Claassens, D.M.F. et al. A genotype‐guided strategy for Oral P2Y(12) inhibitors in primary PCI. N. Engl. J. Med. 381, 1621–1631 (2019). [DOI] [PubMed] [Google Scholar]
- 26. Micieli, A. , Singh, N. , Jahn, B. , Siebert, U. , Menon, B.K. & Demchuk, A.M. Cost‐effectiveness of testing for CYP2C19 loss‐of‐function carriers following transient ischemic attack/minor stroke: a Canadian perspective. Int. J. Stroke 18, 416–425 (2023). [DOI] [PubMed] [Google Scholar]
- 27. Cai, Z. et al. Cost‐effectiveness of CYP2C19 genotyping to guide antiplatelet therapy for acute minor stroke and high‐risk transient ischemic attack. Sci. Rep. 11, 7383 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Carroll, J. et al. Clinical and cost‐effectiveness of clopidogrel resistance genotype testing after ischaemic stroke or transient ischaemic attack: a systematic review and economic model. Health Technol. Assess. 28, 1–194 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Empey, P.E. , Pratt, V.M. , Hoffman, J.M. , Caudle, K.E. & Klein, T.E. Expanding evidence leads to new pharmacogenomics payer coverage. Genet. Med. 23, 830–832 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Smith, D.M. et al. Progress in pharmacogenomics implementation in the United States: barrier erosion and remaining challenges. Clin. Pharmacol. Ther. 118, 778–789 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1