In the December 28, 2011 issue of the Journal of the American Medical Association, Holmes and colleagues publish what they label as a “systematic review and meta-analysis” of clopidogrel CYP2C19 genotyping, platelet function, and cardiovascular events [1]. There has been no shortage of meta-analyses of clopidogrel platelet-function testing and genotyping and their relationship with clinical outcomes [1,2,3].

But the current one is remarkably misleading. While it reviews data from 32 studies in over 42 000 patients, the conclusion is that “there was no significant association of genotype with cardiovascular events.” This conclusion is directly contradicted by the data presented, showing a significant and highly consistent excess of stent thrombosis in patients with a loss-of-function CYP2C19 allele. The absolute increase of stent thrombosis was 14 per 1000 individuals. With over one million patients undergoing coronary stenting per year in the US, this extrapolates to over 14 000 stent-thrombosis events per year.

Stent thrombosis results in either death or heart attack. In a prior meta-analysis by Mega and colleagues who studied 9685 patients undergoing coronary stenting, the stent thrombosis hazard ratio among carriers of a loss-of-function CYP2C19 allele was 2.67 and for homozygotes it was 3.97. Beyond stent thrombosis, there was also a significant excess of death, MI, and stroke for either carriers or homozygotes [2]. Similarly, in an analysis by Hulot and colleagues among 4905 patients who had coronary intervention, there was an odds ratio of 3.45 for stent thrombosis for individuals carrying a loss-of-function CYP2C19*2 allele [3].

Accordingly, a critical flaw of the analysis by Holmes and colleagues was the lack of testing for heterogeneity among patients who underwent stenting as compared with those treated medically. No data for a genotype-by-treatment (medical vs stenting) interaction was provided. The analysis includes a large number of patients from trials that had nothing to do with coronary stenting (eg, atrial fibrillation or STEMI patients treated with thrombolytics) and assess outcomes where the benefit of clopidogrel itself is dubious (eg, target vessel revascularization). Instead of acknowledging the prior extensive body of work that unequivocally documents the stent thrombosis liability for patients who do not metabolize clopidogrel normally, the authors ignored this and homogenized the data, resulting in an erroneous and sweeping conclusion. In fact, all of the prior work on clopidogrel pharmacogenomics has demonstrated the CYP2C19 loss-of-function variants to be important only in coronary stenting patients and does not show up in those medically treated. It makes perfect sense that a metal implant in a coronary artery would pose a particular vulnerability to inadequate platelet suppression.

At Scripps, Vanderbilt, Pitié-Salpetrière in Paris, and other centers, the use of genotyping and platelet function testing for patients receiving clopidogrel is restricted to those undergoing stenting. We and others in the genomic medicine community consider this practice to represent the prototype of pharmacogenomics for several key reasons. First, we have long known that the pharmacologic response to clopidogrel, as reflected by platelet suppression, is particularly heterogeneous. Second, the only consistent DNA sequence variants that have been demonstrated to be associated with the heterogeneous response are localized to the principal cytochrome involved in clopidogrel’s metabolism—CYP2C19. Third, the risk of a stent thrombosis with even one loss-of-function variant is high, consistently two- to threefold across all of the studies that have focused on this end point; similar risks have been observed with platelet-function testing. Fourth, since almost all patients with stent thrombosis will either die or have a heart attack as a result, it is vital to prevent these events. Fifth, the genotypic data are particularly actionable—using alternative dosing regimens of clopidogrel [4] or alternative P2Y12-receptor antagonists, either prasugrel or ticagrelor, both of which are commercially available and bypass the CYP2C19 metabolic–pathway concerns. And sixth, the response of clopidogrel for individualizing the treatment regimen can be corroborated and titrated with bedside platelet-function testing.

The accompanying editorial to the paper by Holmes et al concludes that “physicians should use CYP2C19 or platelet-reactivity testing rarely, if ever” and that there has been “overzealous adoption” [5]. The editorialist advocates a large randomized controlled study. While such trials would always be useful, it unfortunately is a fantasy: it would require tens of thousands of patients, there is no entity that would support such a trial, and the reality is that we have both compelling data and millions of patients who are undergoing coronary stenting around the world each year. Moreover, the era of individualized medicine needs to transcend megatrials of populations, which are not only unsustainable but also crowd out the benefits that can be exceptionally important and validated for the individual patient [6]. All best efforts to prevent stent thrombosis need to be implemented in clinical practice. Refinements that are actively being pursued include point-of-care genotyping to make this much more rapid and inexpensive, along with more complete definition of genomic markers defined by exome and whole-genome sequencing.

Eric J Topol MD
Paul S Teirstein MD
Matthew Price MD
Scripps Clinic, La Jolla, CA
Dan Roden MD
Vanderbilt University, Nashville, TN
Gilles Montalescot MD
Hôpital Pitié-Salpetrière, Paris, France

References:

  1. Holmes MV, Perel P, Shah T, et al. CYP2C19 genotype, clopidogrel metabolism, platelet function, and cardiovascular events. JAMA 2011; 306:2704-2714.
  2. Mega JI, Simon T, Collet JP et al. Reduced-function CYP2C19 genotype and risk of adverse clinical outcomes among patients treated with clopidogrel predominantly for PCI. JAMA 2010; 304:1821-1830. Available here.
  3. Hulot JS, Collet JP, Silvain J, et al. Cardiovascular risk of clopidogrel treated patients according to cytochrome P450 2C19*2 loss-of-function allele or proton pimp inhibitor coadministration: A systematic meta-analysis. J Am Coll Cardiol 2010; 56:134-143. Available here.
  4. Mega JL, Hochholzer W, Frelinger AL, et al. Dosing clopidogrel based on CYP2C19 genotype and the effect of platelet reactivity in patients with stable cardiovascular disease. JAMA 2011; 306:2221-2228. Available here.
  5. Nissen SE. Pharmacogenomics and clopidogrel. JAMA 2011; 306:2727-2728.
  6. Topol EJ. The Creative Destruction of Medicine; New York: Basic Books, 2012.