Clopidogrel is one of the most commonly prescribed heart medications in the world. Millions of patients take it every day to prevent blood clots, stroke, and heart attack. And yet, a new study from the Helix Research Network shows that nearly 1 in 3 of those prescriptions are genetically inappropriate, putting patients at measurably higher risk for the very events the drug is meant to prevent.
The findings, published in Circulation: Genomic and Precision Medicine, make a clear and quantifiable case for pharmacogenomic testing as a standard part of cardiovascular care. Here's what we found and why it matters.
Key Highlights
- 29% of individuals are inappropriately prescribed clopidogrel based on the patient's CYP2C19 genotype
- 38% excess incidence of adverse cardiovascular events in poor and intermediate metabolizers receiving inappropriate prescriptions
- 22 extra thrombosis events and 16 extra myocardial infarctions per 1,000 individuals taking clopidogrel without genotype-informed dosing
- ~1 preventable thrombosis for every 30 people prescribed clopidogrel
- $700K in estimated excess costs per 1,000 patients, attributable to preventable thromboses and MIs
- Study cohort: 101,883 individuals with paired Exome+® sequencing and EHR data across multiple U.S. health systems
What Is the CYP2C19–Clopidogrel Problem?
Clopidogrel is a prodrug: it doesn't work on its own. The body must convert it into its active form using the CYP2C19 enzyme. Patients who carry genetic variants that reduce or eliminate CYP2C19 function known as poor or intermediate metabolizers — cannot adequately activate clopidogrel, meaning the drug provides little to no protection against clotting events.
This isn't new science. The FDA, the American Heart Association, and the Clinical Pharmacogenomics Implementation Consortium (CPIC) have all issued guidance recommending CYP2C19 testing before prescribing clopidogrel. And yet, in practice, the vast majority of prescriptions are still written without a genetic test.
Our study set out to answer a straightforward but largely unaddressed question: What is the real-world cost of that gap?
How We Studied It: The Helix Research Network at Scale
To answer that question, we needed something most research studies don't have: a large, unselected, real-world population with both genomic data and longitudinal clinical records.
The Helix Research Network (HRN) provided exactly that. Our cohort included >100,000 consented individuals from multiple U.S. health systems, all with Clinical Exome+® sequencing linked to electronic health records and insurance claims data. CYP2C19 star alleles were derived from Exome+® data to classify each participant as a poor, intermediate, normal, or rapid metabolizer.
We then used EHR data to identify everyone who had received at least one prescription for clopidogrel, prasugrel, or ticagrelor and to track subsequent outcomes including thrombosis, myocardial infarction (MI), and a five-point measure of major adverse cardiac events (MACE) at 2 months, 6 months, and 1 year after the first prescription.
One methodological highlight: we used a large language model (LLM) to extract prescribed dosage from free-text prescription "sig" fields at scale — validated at 99% accuracy against a human-derived gold standard. This enabled us to apply CPIC guidelines at a level of specificity that would have been impractical through manual chart review alone.
What We Found: Inappropriate Prescriptions Are Both Common and Consequential
In our cohort, 26% of clopidogrel recipients had intermediate metabolic (IM) activity of CYP2C19 — and 97% of them were taking an unadjusted standard dose, despite CPIC guidelines recommending a higher dose or an alternative antiplatelet drug. An additional 3% were poor metabolizers (PM) who should not be prescribed clopidogrel at any dose.
In total, 29% of patients prescribed clopidogrel were doing so inappropriately relative to their genotype.
The consequences were significant and consistent. Poor and intermediate metabolizers receiving inappropriate prescriptions experienced:
- 38% more adverse events overall compared to appropriately prescribed patients
- Higher rates of thrombosis, MI, and MACE at every timepoint measured: 2 months, 6 months, and 1 year
- The majority of adverse events occurring within just 2 months of starting treatment
Critically, these trends held across genetic ancestry groups, were not explained by age or sex, and were consistent in patients with and without percutaneous coronary intervention (PCI), ruling out several potential confounders.
What Does This Mean in Human Terms?
The aggregate numbers translate into stark individual-level risk. Per 1,000 people prescribed clopidogrel:
- 22 thrombosis events are attributable to the lack of PGx-informed prescribing
- 16 myocardial infarctions could potentially be avoided with genotype-guided therapy
- The excess healthcare costs tied to those preventable events: approximately $700,000
Put simply: for every 30 patients prescribed clopidogrel without a genetic test, one person is likely suffering a blood clot that appropriate testing could have prevented.
Why This Study Is Different
Previous research on CYP2C19 and clopidogrel has largely been conducted in clinically ascertained populations, where patients are already known to have cardiac conditions and referred for specialized care. Our cohort is fundamentally different: it's an all-comers population drawn from routine clinical care across multiple health systems, representing the full spectrum of patients who end up on clopidogrel in the real world.
That distinction matters. It means our findings reflect what's actually happening in standard-of-care prescribing — not a curated high-risk population — making them directly applicable to health system decision-making about PGx testing implementation.
A Proof Point for the Accessible Genome
Beyond the clinical findings, this study demonstrates something important about what becomes possible when genomic data are captured at scale and paired with longitudinal health records.
The Helix Exome+® assay was designed to be reanalyzable, meaning that once a patient's genomic data are stored, they can be queried repeatedly for future applications without the need for an additional sample, allowing immediate return of results. CYP2C19 star allele calling is just one example. The same underlying data can be used for pharmacogenomic analysis across dozens of additional drug-gene pairs, rare disease screening, polygenic risk scoring, and more.
This is the core value proposition of population genomics done right: the investment in sequencing compounds over time, delivering new clinical and research value with every application layered on top.
What Comes Next
This work opens several possible directions for future research:
- Expanding PGx testing beyond CYP2C19 — the Exome+® data already exist for 100K+ individuals, enabling analysis of additional pharmacogenes without additional sequencing
- Prospective health system implementation — evaluating outcomes when genotype-informed prescribing is systematically deployed across a health network
- Broader cost-effectiveness modeling — extending the $700K-per-1,000-patients framework to other high-impact drug-gene pairs
Interested in Pharmacogenomic Research at Scale?
The Helix Research Network offers life sciences teams access to one of the largest real-world clinico-genomic cohorts available for research — with paired Exome+® sequencing, longitudinal EHR data, and standardized phenotyping across multiple health systems. Whether you're exploring pharmacogenomics, biomarker discovery, or real-world evidence generation, we'd love to talk.
Connect with us at https://www.helix.com/life-sciences