Does Clopidogrel Each Day Keep Stent Thrombosis Away?

New medicines and new methods of cure always work miracles . . . for a while.
William Heberden, MD

The polymer-based elution of antiproliferative medications from a metal stent prosthesis has been demonstrated to suppress neointimal proliferative response to stent-vessel injury and to reduce consequent late arterial lumen loss and restenosis. In randomized controlled trials, drug-eluting stents compared with bare-metal stents caused a 70% to 80% reduction in binary (>50%) angiographic restenosis and a 50% to 70% reduction in repeat target lesion or vessel revascularization.^{1 - 2} Furthermore, comprehensive follow-up to 1 year demonstrated no evidence for any excess in the incidence of death, myocardial infarction (MI), or stent thrombosis for either treatment strategy.^{3 - 4} Thus, based on pivotal clinical trial evidence for both the relative safety and efficacy of drug-eluting stents, the US Food and Drug Administration approved both Cypher, the sirolimus-eluting stent, in 2002, as well as TAXUS, the paclitaxel-eluting stent device, in 2004. Subsequently, multiple other drug-eluting stent platforms have become available outside of the United States.

Following approval of the sirolimus-eluting stent, sweeping adoption of this promising new technology into clinical practice occurred as it appeared that restenosis, the Achilles heel of bare-metal stenting, had finally been cured.⁵ Indeed, the pace of drug-eluting stent utilization outstripped any available clinical trial–based evidence for support, and drug-eluting stent deployment “off label” soon accounted for up to half of all use.^{6 - 7} However, some leaders questioned, “Should standard practice patterns undergo marked immediate changes when fewer than 2000 patients have been studied in 3 clinical trials?”⁸

In the initial US Food and Drug Administration panel discussions preceding approval of the sirolimus-eluting stent, however, a probing question was posed by a panel member, “What is the β error level for a 1000 patient randomized study? Could we miss a 1% adverse outcome?”^{9 - 10} Indeed, with longer duration clinical follow-up in larger numbers of patients, an important observation has been made. The aggregate follow-up of multiple randomized comparative trials of drug-eluting stents vs bare-metal stents demonstrates a low frequency (approximately 0.5% absolute) relative increase in the incidence of very late stent thrombosis, which becomes evident only after 9 to 12 months and appears to extend to 3 to 4 years.^{11 - 12} The relative incremental risk approximates 0.2% per year (total increment 0.46%-0.57% through 3 years) for the types of patients included in these trials and appears similar for both drug-eluting stent devices.

Several postulates regarding pathogenesis of late stent thrombosis were proposed, with the most plausible being that the same drug/polymer combinations that effectively reduce in-stent smooth muscle cell hyperplasia may also cause slower, less complete arterial healing and endothelial coverage, as has been noted on both angioscopic¹³ and pathological examination.¹⁴ Beyond this, delayed endothelialization and healing due to polymer-related hypersensitivity, inflammation, or both, as well as acquired late incomplete stent apposition due to vessel positive remodeling in the stented arterial segment, have all been incriminated as potential causes of late drug-eluting stent thrombosis, particularly once combination antiplatelet therapy (aspirin plus thienopyridine) is discontinued. Indeed, discontinuation of clopidogrel within 30 days¹⁵ and even within 6 months^{16 - 17} following drug-eluting stent deployment has been associated with stent thrombosis, death, or both. The benefit of extending clopidogrel therapy beyond 6 months has been debated.^{16 - 17} News media reports have increased public awareness and generated serious concern about late drug-eluting stent thrombosis.

In this issue of JAMA, the article by Eisenstein and colleagues¹⁸ provides constructive information valuable to current and future patients treated with drug-eluting stents by examining the optimal duration of clopidogrel therapy following implantation of drug-eluting and bare-metal stents. The authors provide observations from consecutive patients who received either bare-metal stents (n = 3165) or drug-eluting stents (n = 1501) between January 1, 2000, and July 31, 2005, at Duke Heart Center. Landmark analyses were performed among patients who were event-free (no death, MI, or revascularization) at 6- and 12-month follow-up, stratified by stent type (bare-metal stent vs drug-eluting stent) and patient report of clopidogrel use (yes vs no). The end points of death, nonfatal MI, and the composite of death or MI were ascertained at 24-month follow-up.

Eisenstein et al concluded that patients treated with clopidogrel at 6 and 12 months following drug-eluting stent deployment had significantly lower subsequent rates of death or MI than patients treated with drug-eluting stents not receiving clopidogrel at these time intervals. Furthermore, patients with drug-eluting stents who received extended clopidogrel treatment to at least 1 year had significantly lower rates of death and death or MI when compared with patients with bare-metal stents, regardless of their clopidogrel therapy status. The salutary effects of clopidogrel were maintained to at least 2 years of follow-up.

The optimal therapeutic strategy from the patient's perspective would be drug-eluting stents plus clopidogrel. The fact that clopidogrel benefit was apparent only among patients treated with drug-eluting stents (not bare-metal stents) effectively argues the premise that this benefit derives from a specific drug (clopidogrel)–device (drug-eluting stents) interaction, which is most likely expressed at the drug-eluting stent treatment sites rather than through a salutary systemic effect of clopidogrel on background (nontarget lesion) events in patients treated with drug-eluting stents. Importantly, amid sensational hypothetical extrapolation (“drug-eluting stents may cause 4500 heart attacks every year, 2000 of them fatal”¹⁹ ) and hyperbole (“drug-eluting stents manufacturers could end up rivaling Vioxx maker, Merck, as targets of lawsuits from people who suffer heart attacks”²⁰ ) in the media, the article by Eisenstein et al¹⁸ injects real-world observations that should constructively modify current therapeutic algorithms and maintain societal focus on the net clinical benefit provided by this novel, revolutionary technology (drug-eluting stents).

Before embracing the apparent logical conclusion that “clopidogrel each day keeps thrombosis away,” it is important to carefully examine potential strengths and weaknesses of these data as well as their implications for current practice. First, the number of patients from which the conclusions are drawn is relatively small. Indeed, the total number of patients with drug-eluting stents at risk at 24 months is less than 250 in the treatment groups either receiving or not receiving clopidogrel. Second, reliance on patients' self-reporting of clopidogrel usage does not account for noncompliance with clopidogrel. This limitation would be best overcome in the context of a randomized clinical trial comparing different durations of clopidogrel therapy.

Third, the observation that clopidogrel did not influence the occurrence of events in patients treated with bare-metal stents may in part be explained by the paucity of prescription beyond 30 days or 6 months. Indeed, patients with bare-metal stents receiving clopidogrel (n = 417) was the smallest subgroup in the 6-month landmark analysis. Nevertheless, the 12-month landmark analysis included more patients with bare-metal stents receiving clopidogrel (n = 346) than patients with drug-eluting stents either receiving (n = 252) or not receiving (n = 276) clopidogrel. Thus, if the effect of clopidogrel use in patients with bare-metal stents is similar to that in patients with drug-eluting stents, these sample sizes should be sufficient to detect differences. Also pertinent is the concern of how likely (or unlikely) would be the scenario that a patient would resume clopidogrel therapy after stopping it (between landmark analyses). Could either an error in 6-month self-reporting or intercurrent procedure/event have occurred?

Fourth, the major limitation of this study (nonrandomized allocation of clopidogrel use) is addressed by the authors through the use of a rather extensive propensity score model that includes demographics, comorbid conditions, procedural, and even socioeconomic (ZIP code–level median income and average house value) variables. Thus, obvious imbalances between analysis groups that could bias their conclusions are unlikely.

Fifth, these data provide no measure of interindividual variability in responsiveness to clopidogrel therapy, which has been implicated in late adverse events following stenting.^{21 - 22} Clopidogrel adherence should not be equated with responsiveness (adequate platelet inhibition). Sixth, the events that occurred during follow-up were not specifically attributable to the treated vessel. This analysis cannot discern target vs nontarget vessel distribution of events and whether there were differences in these attributions between stent types. Nevertheless, the authors have ostensibly captured the outcomes of greatest interest to both patients and physicians (death and MI) in a comprehensive database inclusive of all clinically recognized events, including those events associated with restenosis and its treatment. The absence of lesion-specific data does not detract from their overall message and should stimulate interrogation of other data sets as well as prospective studies to better delineate the mechanisms that underlie their observations.

And sixth, the observation regarding “net clinical benefit” of extended clopidogrel use in patients with drug-eluting stents is limited by the absence of data on major bleeding events with or without transfusion, which might be associated with such therapy. For example, in a patient cohort with asymptomatic cardiovascular disease and/or risk factors for atherothrombosis enrolled into the Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance (CHARISMA) trial,²³ bleeding events requiring transfusion were observed in 2.1% of patients followed up for a similar duration (approximately 2 years) to that of the current study. Acknowledging the epidemiological association if not pathogenetic link between bleeding events and ischemic events, future observations and certainly randomized studies should include measures of bleeding for analysis.

How should the results of the study by Eisenstein et al influence current practice? Clearly, the choice of drug-eluting or bare-metal stents should be individualized based on patient demographics and the logistics of extended clopidogrel use. For patients without contraindication, clopidogrel therapy should be continued through at least 1 year and possibly indefinitely until the time course of vulnerability for stent thrombosis in patients treated with drug-eluting stents is better defined. Patients with drug-eluting stents treated in this fashion appear to have survival without MI to a greater extent than either patients treated with drug-eluting stents who received more limited duration clopidogrel therapy or patients treated with bare-metal stents regardless of clopidogrel treatment status. No data were provided by Eisenstein et al on the optimal strategy for interrupting clopidogrel therapy should interim operative procedures, bleeding events, or both occur. The current American Heart Association/American College of Cardiology/Society of Cardiac Angiography and Intervention clinical practice guidelines recommend discontinuation of clopidogrel for 5 days in advance of coronary artery bypass graft surgery.²⁴ This recommendation, in the absence of readily available point-of-care platelet function testing, should be sufficient for other surgical procedures as well. In these cases, it also appears reasonable to consider resuming clopidogrel therapy soon (≤48 hours) following the procedure. Ideally, aspirin therapy (81 mg/d) should not be interrupted. However, the safety and efficacy of these recommendations with respect to limiting the “window of vulnerability” for stent thrombosis should be formally tested.

In addition, novel drug-eluting stent design iterations to limit polymer surface area exposure (bioresorbable) over time as well as to passively or actively promote healing/endothelial stent coverage are currently being investigated in clinical trials. Indeed, the optimal strategy to eliminate late stent thrombosis risk may be to eliminate the permanent metal prosthesis altogether (bioresorbable platform). Furthermore, next generation platelet P2Y12 receptor inhibitors currently in clinical trials may circumvent issues of variability in platelet response, resistance associated with clopidogrel therapy, or both. Moreover, the global process of premarket medical device evaluation and approval deserves scrutiny. Any requirement for larger numbers of patients who are treated to be followed up for longer periods of time before device approval could further delay access to potentially lifesaving technological advances and, no doubt, will further prolong the current 2- to 3-year time delay between when a new medical device becomes available for use in Europe compared with in the United States. Alternatively, strategies aimed at enhanced and more rigorous postmarket surveillance in both Europe and the United States could provide earlier, more accurate insights into low-frequency adverse event occurrences and more credible information on device safety and efficacy in patient cohorts excluded from participation in pivotal clinical trials.²⁵