Clopidogrel Pharmacogenomics: Next Steps - JACC: Cardiovascular ...

2 downloads 0 Views 508KB Size Report
cogenomics in drug development and views it as a pathway to ..... function allele or proton pump inhibitor coadministration: a systematic meta-analysis.
JACC: CARDIOVASCULAR INTERVENTIONS

VOL. 3, NO. 10, 2010

© 2010 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER INC.

ISSN 1936-8798/$36.00 DOI: 10.1016/j.jcin.2010.08.012

POINT OF VIEW

Clopidogrel Pharmacogenomics: Next Steps A Clinical Algorithm, Gene–Gene Interactions, and an Elusive Outcomes Trial Patrick Gladding, MBCHB,*§ Laura Panattoni, PHD,† Mark Webster, MBCHB,‡ Leslie Cho, MD,* Stephen Ellis, MD* Cleveland, Ohio; and Auckland, New Zealand

Clopidogrel pharmacogenomics has received significant attention since a black box warning was announced by the Food and Drug Administration in March. This has left clinicians in a difficult situation where many questions remain unanswered. In this brief viewpoint article, we ask some pointed questions of our own and outline the pathway that needs to be taken for clinical translation to occur. (J Am Coll Cardiol Intv 2010;3:995–1000) © 2010 by the American College of Cardiology Foundation

On March 12, 2010, the U.S. Food and Drug Administration (FDA) released a black box warning on the antiplatelet drug clopidogrel (1). This warning alerted clinicians of the heterogeneity of response to clopidogrel and advised pharmacogenetic testing with consideration of alternative antiplatelet agents in nonresponders. This elicited a strong reaction from clinicians as no prospective randomized trial has shown that this treatment strategy improves clinical outcomes. Clinical trials testing this hypothesis are underway, but several questions remain to be answered. One unaddressed question is why this has not been explored further by pharmaceutical companies involved with clopidogrel manufacture. It is common practice that pharmaceutical compounds are screened at an early stage against cytochrome P450 enzymes. This assesses whether a drug is metabolized by polymorphic cytochrome P450 enzymes (i.e., pathways that are known to vary markedly in a population). Compounds that are metabolized by a single pathway are generally discontinued from further development. Although this knowledge might not have

From the *Cleveland Clinic, Cleveland, Ohio; †School of Population Health, University of Auckland, Auckland, New Zealand; ‡Green Lane, Cardiovascular Service, Auckland City Hospital, Auckland, New Zealand; and the §Theranostics Laboratory, Auckland, New Zealand. Drs. Gladding and Webster have a patent pending regarding clopidogrel pharmacogenomics, and Dr. Gladding has founded a nonprofit company, Theranostics Laboratory. All other authors have reported that they have no relationships to disclose. Manuscript received June 4, 2010; revised manuscript received August 24, 2010, accepted August 25, 2010.

been available for clopidogrel during its early development, it does not preclude laboratory work being done in phase IV post-marketing surveillance. Eli Lilly and Company first reported a genetic basis to the response to clopidogrel in 2006 (2) and has continued to investigate this. However, no attempt was made to explore this finding by other pharmaceutical companies involved with the licensing or sales of clopidogrel. Because of these findings, a biomarker program was built into the Phase I/II trials of the Lilly drug development program for prasugrel. The subsequent TRITON (TRial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet InhibitioN with Prasugrel) (3) had a genomic component that showed that the CYP2C19*2 polymorphism conferred a risk to carriers taking clopidogrel but not to those receiving prasugrel (4). This result supported the earlier strong scientific evidence of in vivo and ex vivo experiments showing the importance of the CYP2C19 gene and enzyme in clopidogrel metabolism (5–9). Despite this finding, in 1 of the largest pharmacogenetic trials performed, the message of using genetic testing has not been part of Lilly’s marketing strategy. The FDA decision overall, however, is not completely unexpected. The FDA Critical Path Initiative has long supported the role of pharmacogenomics in drug development and views it as a pathway to improving patient responses to medication and reducing the cost of drug development (10). In assessing new treatments the FDA uses

996

Gladding et al. Clopidogrel Pharmacogenomics: Next Steps

JACC: CARDIOVASCULAR INTERVENTIONS, VOL. 3, NO. 10, 2010 OCTOBER 2010:995–1000

advanced population modeling tools, such as pharmacometrics, which assesses the impact of polymorphic variability in a population (1,11). From a population basis, there is an argument that individualizing treatment will not only benefit patients but also save money for health care providers. In countries such as New Zealand and Germany, where clopidogrel is generic and inexpensive, there stands a clear advantage of targeting expensive agents to nonresponders. With a Monte Carlo simulation, we have modeled the cost-effectiveness of using pharmacogenomics to target high-risk patients for treatment with prasugrel (12). Although our method did not include clinical risk factors or platelet testing, genomics alone was sufficient to target subpopulations in a cost-effective manner. The reason this is of importance to the providers of health care resources is that clopidogrel is the second-most-prescribed drug in the world, with global sales of over $6 billion. There will soon be a generic version of clopidogrel in many countries which will substantially reduce that cost (13). However, substituting clopidogrel for a new patented medication will lead to ongoing cost, for the duration of that patent life. Of further interest in our cost-effectiveness study was the result that ethnic groups, with a higher frequency of the CYP2C19*2 allele, benefited the most from a Abbreviations targeted strategy. In the U.S., and Acronyms African Americans and Asians carry this single nucleotide polyFDA ⴝ Food and Drug morphism (SNP) at a disproporAdministration tionately higher rate than CauSNP ⴝ single nucleotide polymorphism casians (3). The cost of testing in our simulation was modeled at $175 and needs to be performed once in a lifetime. This makes it unattractive to industries that benefit from repeated diagnostic testing (12).

The black box decision of the FDA was based on a number of clinical trials and meta-analyses but also on a sponsor-funded trial of 40 healthy volunteers (Table 1) (2,14 –21). These volunteers received 75 mg and 150 mg of clopidogrel in a crossover study, with platelet function as an outcome, and showed that a marginal benefit was obtained in dose escalation (1). Although alternative antiplatelet treatments were recommended in the warning, specific agents were not mentioned, leaving clinicians with minimal guidance on what to do next. Several recent studies have investigated the benefits of alternative treatments for nonresponders, but these have focused on platelet function to guide treatment strategies (22–25). Of concern is that comparative outcome studies have shown that only a small handful of platelet function analyzers actually predict events. Definitions of nonresponse with these functional tests are also often lacking. Despite these disadvantages, it seems attractive to test the pharmacodynamic outcome, that is, what the drug does to the body. Although recent studies have suggested that platelet function testing outperforms genotyping, the pharmacogenomics of clopidogrel is not yet fully elucidated (26,27). The known CYP2C19 polymorphisms are thought to only contribute 12% to 20% of response variability, and it seems evident that other genes are involved (28). The result from a recent genome-wide association study has shown that the response to clopidogrel is highly heritable (70%) (16). Because CYP2C19*2 only explains 12% to 20% of variability, it is probable that other genes or rare variants within the 2C19 gene will explain this high heritability. There are a number of further factors to consider with the CYP2C19 gene. First, each individual carries 2 copies of the gene, CYP2C19*2 heterozygotes still have 1 functional copy and this means that a higher dose of clopidogrel might be effective. However, with saturable

Table 1. Clinical Outcome Studies Evaluating CYP2C19*2 First Author (Ref. #)

n, Trial Name

Clopidogrel Dose

Clinical Outcome

Shuldiner et al. (16) Mega et al. (2)

n ⫽ 429

300/75 mg

CV event or death

HR: 2.42 (95% CI: 1.18–4.99; p ⫽ 0.02)

n ⫽ 1,477 TRITON

300/75 mg

MACE ST

AR between *2 carriers vs. Wt ⫽ 4.1% HR: 1.5 (95% CI: 1.07–2.2, p ⫽ 0.01) HR: 3.09 (95% CI: 1.19 – 8.00, p ⫽ 0.02)

Simon et al. (17)

n ⫽ 2,208 FAST-MI

300/75 mg

MACE

HR: 1.98 (95% CI: 1.10–3.58, p ⫽ 0.05)

Collet et al. (18)

n ⫽ 259

MACE ⫹ death ST

HR: 3.69 (95% CI: 1.69 – 8.05, p ⫽ 0.0005) HR: 6.02 (95% CI: 1.81–20.04, p ⫽ 0.0009)

Sibbing et al. (19)

n ⫽ 2,485

600/75 mg

ST

HR: 3.81 (95% CI: 1.45–10.02, p ⫽ 0.007) Wt/Wt HR: 0.4% *2/Wt HR: 1.5% *2/*2 HR: 2.1%

Giusti et al. (20)

n ⫽ 772

600/75 mg

CV death

OR: 2.7 (95% CI: 1.00 – 8.42, p ⫽ 0.05) (ADP) OR: 2.9 (95% CI: 1.08 –12.98, p ⫽ 0.08) (*2/Wt) ⫹ ADP 11.45 (95% CI: 1.84 –71.27, p ⫽ 0.009)

Trenk et al. (21)

n ⫽ 797

600/75 mg

Mortality

OR: 3.0 (95% CI: 1.4–6.8; p ⫽ 0.004)

75 mg

Risk for CYP2C19*2 Carriers

ADP ⫽ platelet function using ADP agonist; CI ⫽ confidence interval; CV ⫽ cardiovascular; FAST-MI ⫽ Registry on Acute ST–Elevation Myocardial Infarction; HR ⫽ hazard ratio; MACE ⫽ major adverse cardiac event; ST ⫽ stent thrombosis; TRITON ⫽ TRial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet InhibitioN with Prasugrel; Wt ⫽ wild type.

JACC: CARDIOVASCULAR INTERVENTIONS, VOL. 3, NO. 10, 2010 OCTOBER 2010:995–1000

enzyme kinetics means this benefit might be marginal (22). The CYP2C19*2 homozygotes, with 2 null alleles, might not respond to higher doses and might require an alternative antiplatelet drug (19). The field of pharmacogenomics is rich with examples where multiple interacting genes act in combination to influence drug response. Warfarin response is influenced by the VKORC1, CYP2C9, GGCX, and CYP4F2 genes (29 –31). Clopidogrel absorption and metabolism is complex and involves multiple biological bottlenecks, essentially efflux pumps and enzymes that have polymorphic genes. Biological bottlenecks are prone to significant influence from outside perturbations and are particularly affected by “multiple hits.” The genes for each step in clopidogrel absorption and metabolism have already been well-characterized (Fig. 1), and new evidence is emerging that testing other SNPs—such as a SNP in the ABCB1 gene—in addition to CYP2C19*2 identifies more nonresponders (32). Looking at the oncology literature is also helpful as CYP2C19 is involved with the response to some chemotherapeutics. The science here is more evolved and shows that

Gladding et al. Clopidogrel Pharmacogenomics: Next Steps

997

the gene is induced by inflammatory disease states and influenced by master gene regulators. The main master regulator is the human pregnane receptor gene, which also effects CYP3A4 (33–35). The inducible nature of gene expression for 2C19 and the multifactorial nature of platelet activation and aggregation argue for the addition of platelet function testing in personalizing antiplatelet treatment. Clinical context and timing are vitally important in determining the correct treatment for patients. Platelets are activated by a number of agonists, and many platelet function assays measure global responses, influenced by additional factors, such as acutely generated thrombin that are not influenced by antiplatelet agents. Therefore it makes sense that testing both genetic factors and platelet aggregation will yield more information than testing either alone (Fig. 2) (20). Platelet reactivity is higher around the time of an acute coronary syndrome, and this might explain the short-term benefits of prasugrel in acute coronary syndrome patients. Stratifying a patient to a long-term treatment regimen, when platelet reactivity is elevated during this acute period is nonintuitive. A practical situation worth

Figure 1. Antiplatelet Drug Clopidogrel Pathway Genes are shown in blue ovals. Adapted, with permission, from PharmGKB and Stanford University–Sangkuhl K, Klein TE, Altman RB. Clopidogrel pathway. Pharmacogenet Genomics 2010;20:463–5.

998

Gladding et al. Clopidogrel Pharmacogenomics: Next Steps

Multiplex cytokine analysis

JACC: CARDIOVASCULAR INTERVENTIONS, VOL. 3, NO. 10, 2010 OCTOBER 2010:995–1000

Functional testing

Patient Phenotype e.g. CVA, age, weight, diabetes

Platelet Cytokines e.g. IL-6, TNFα

P2Y12 gene Active metabolite

CYP2C19 enzyme

Master Gene Regulators e.g. hPXR

Prodrug

Drug-drug interactions

CYP2C19 gene Pharmacogenomics

Metabolomics Gut absorption e.g. CES

CYP-gene interactions e.g ABCB1

Figure 2. Hypothesized Systems Approach to Clopidogrel Response Larger font items indicate important components in the system. Green font items indicate other forms of testing which may be valuable in assessing drug response. Reprinted, with permission, from Gladding et al. (22). ABCB1 ⫽ p-glycoprotein efflux pump; CES ⫽ carboxylesterase; CVA ⫽ stroke; hPXR ⫽ human pregnane X receptor; IL ⫽ interleukin; TNF ⫽ tumor necrosis factor; T2DM ⫽ type II diabetes; wt ⫽ weight.

considering is the individual who has already been treated with a potent irreversible antiplatelet agent, or efficacious agent with a long half-life, (i.e., prasugrel or glycoprotein IIb/IIIa inhibitor, respectively). In this circumstance it is not possible to identify responders to clopidogrel with functional testing. Genotyping in this circumstance is a logical option. Possible strategies that incorporate pharmacogenomics and platelet testing include the possibility of using platelet testing at the point of discharge or after discharge to modify treatment in the clinic. In addition to measuring drug efficacy, outpatient functional testing would provide an objective measure of compliance. The combination of testing genotype and phenotype has been valuable, in other examples of clinical pharmacogenomics. Thiopurine methyltransferase genotyping, for gastroenterology patients on azathioprine, is an example. Recommendations for this drug include initial genotyping, treatment stratification and subsequent phenotypic testing to monitor hematological dyscrasia (36). Clinicians are reserved about incorporating genotyping into clinical practice for a number of reasons. Lack of clinical trial evidence is perhaps the first issue, which might soon be resolved. A therapeutic window for antiplatelet drugs is sorely needed, and early demonstration of such a holy grail is alluring (37). Integrating multiple factors involved in drug response is also necessary. An algorithm for clopidogrel pharmacogenomic nonresponse has already been formulated by a group in Germany, though this requires validation in independent cohorts and use in a prospective trial (Fig. 3) (38). Arguing that complexity will

limit the clinical applicability of an algorithm undervalues the power of electronic medical records with decision support and electronic prescribing tools. Cost, reimbursement, and availability of testing are often raised as concerns; however, testing is available through a number of providers from both large laboratories and genotyping platform providers. Although costs of $400 are quoted with turnaround times of 1 week, the cost of testing can be provided at substantially less than this, with turnaround times of 2 to 8 h (22). The Nanosphere Verigene and Autogenomics INFINITI genotyping platforms are examples. A final point worth making is that financial incentives and the directions in which they bias treatment need to be addressed. Personalized medicine is not popular with industries that benefit from a 1-sizes-fits-all business strategy. However, individualization of drug treatment should be considered no different from stent sizing for a stenotic lesion, or choice of a drug-eluting stent based on lesion or patient profile. Is it justifiable to perform intravascular ultrasound or optical coherence tomography but not genotyping/phenotyping for drug treatment? These questions need addressing in costeffectiveness and comparative effectiveness studies. Current medical practice of providing all options to patients, without knowledge of their benefits, is not sustainable. Current practice will need to be replaced with a health care system delivered with greater refinement, targeted use of resources, and the ability to compare and monitor outcomes. Given the increasing complexity and volume of information in clinical medicine, software tools will need to be developed

Gladding et al. Clopidogrel Pharmacogenomics: Next Steps

JACC: CARDIOVASCULAR INTERVENTIONS, VOL. 3, NO. 10, 2010 OCTOBER 2010:995–1000

999

Clinical factors assessed used in the PREDICT score i. ii. iii. iv. v.

DM CRF Age >65 ACS LVF

+2 +2 +1 +1 +3

Authors suggest use of clinical risk factors, pharmacogenomic and platelet testing* vi. CYP2C19*2, *3 or other non-functional variant vii. CYP2C19*17 ultrametabolizer variant viii. ABCB1 C3435T efflux pump TT genotype

+7 -7 +7

1) Calculate Score: 7 = prasugrel 60/10 mg,