Thrombosis in Coronary Drug-Eluting Stents
Report From the Meeting of the Circulatory System Medical Devices Advisory Panel of the Food and Drug Administration Center for Devices and Radiologic Health, December 7–8, 2006
Coronary drug-eluting stents (DES) were first approved for use in the United States in 2003 (Cypher Coronary Sirolimus-Eluting Stent [SES], Cordis Corporation, Miami Lakes, Fla) and 2004 (Taxus Express Paclitaxel-Eluting Stent [PES], Boston Scientific Corporation, Natick, Mass) on the basis of randomized, blinded, controlled studies that demonstrated reduced target vessel failure (Cypher) and target vessel revascularization (TVR) (Taxus) at 9 months. Major adverse cardiac events were also significantly reduced for both DES when compared with bare metal stents (BMS).
Presentations at national and international meetings in 2006 suggested that DES use may be associated with an increase in late (≥1 year after implantation) stent thrombosis (ST). Other studies suggested that there may be a significant increase in the combined end point of death and myocardial infarction (MI) compared with BMS. In response to these safety concerns, the US Food and Drug Administration (FDA) convened a meeting of the Circulatory System Medical Devices Advisory Panel. Advisory Panels make nonbinding recommendations to the FDA.
Panelists recognized the need for a method to assess the true frequency of ST. However, one of the challenges in the assessment of DES ST risk is that the ST definition varies from study to study (Table 1). Differences in these on-protocol definitions of ST make it difficult to pool studies for analysis and to compare one stent to another (sirolimus-eluting versus paclitaxel-eluting). Furthermore, it is important to note that the on-protocol definition of ST for some of the clinical trials that support DES approval in the United States censored events that occurred after target lesion and/or target vessel revascularization (TLR/TVR). This means that an ST that occurs after TLR/TVR may not be counted as an ST by the on-protocol definition. Because BMS patients are more likely to experience restenosis and TLR/TVR than DES patients, more BMS patients (and therefore BMS-late ST events) are censored. This potentially introduces a bias against DES with regard to the on-protocol definitions of ST.
To address the shortcomings of the ST definitions, an Academic Research Consortium (ARC) composed of clinical investigators, industry representatives, and regulatory authorities (including the FDA) has proposed new definitions for ST (Table 1) in an attempt to establish uniformity in the industry and to improve sensitivity for the diagnosis of ST. For the purposes of the panel meeting, the FDA believed that the ARC definitions were acceptable and requested that sponsors and investigators apply these definitions to their data sets when possible. The FDA also recognized that other definitions of ST may be appropriate and proposed for use in clinical studies. A majority of the panel agreed that, among the ARC categories, “definite” and “probable” are most useful, although the true sensitivity and specificity have not been determined; the “possible” category, which lacks specificity, was least useful and was likely to lead to overestimates of the true ST frequency. Not all panel members, however, felt that the ARC definitions should be used for future studies, and several felt it was inappropriate to readjudicate (even blindly) results of prior studies. A number of panel members felt that it was more appropriate to use the “on-protocol” ST definitions. Some felt that a process independent of industry input would be preferable and that ST definitions should be readdressed by previously established entities, such as the American College of Cardiology or American Heart Association.
On-Label Drug-Eluting Stent Use
The FDA approves medical devices when a reasonable assurance exists that the device can be used safely and effectively in a specific patient population. These patient populations are typically defined by the inclusion and exclusion criteria of the pivotal clinical trials submitted to support device approval and are described in the products’ instructions for use or “label.” Use within these well-defined criteria is termed “on-label.” On-label use for the 2 US-approved DES is described in Table 2.
Long-Term Follow-Up of Pivotal DES Trials
The panel reviewed long-term follow-up data from the pivotal US trials as presented by the DES manufacturers. Data from the Sirolimus-Coated Stent in Treatment of Patients With De Novo Coronary Artery Lesions (SIRIUS) trial, which compared SES to BMS, included ≈4 years of follow-up in 94% of the 1058 study patients. Long-term effectiveness of SES compared with BMS was demonstrated with significant sustained reductions in the rate of target vessel failure. No significant differences were observed in mortality, MI, or on-protocol ST between SES and BMS (Table 3). Similarly, data from TAXUS IV compares PES to BMS with >95% follow-up available at 3 years for the 1314 patients. Significant sustained reductions in TVR were observed in PES patients compared with BMS patients with no significant difference in cardiac mortality, MI, or on-protocol stent thrombosis (Table 3).
Pooled analyses and meta-analyses of patient-level data from both the pivotal and additional randomized, blinded, controlled DES trials were also presented. These analyses demonstrated no significant differences in the rate of death, MI, or death/nonfatal MI for either SES (follow-up ≈4 years) or PES (mean follow-up 3.2 years) when compared with BMS.
Similarly, the cumulative incidence of ST at 4 years was not significantly different between SES and BMS either by the protocol-defined ST definition (SES 1.2% versus BMS 0.6%, P=NS) or by the ARC ST definition (with categories of definite and probable) (SES 1.5% versus BMS 1.8%, P=NS) (Table 4). Patient-level pooled analysis of the PES trial experience similarly confirmed the absence of a statistically significant difference in the cumulative ST rate at 4 years (protocol definition: PES 1.3% versus BMS, 0.9%, P=NS; ARC definition categories definite and probable: PES 1.9% versus BMS 1.5%, P=NS) (Table 4). The panel, however, recognized that the studies were underpowered to detect even moderate clinically significant differences in the rate of rare events like ST.
Despite the absence of a significant overall difference in the rate of ST, the time distribution of events appeared different for BMS compared with DES. Specifically, there were numerically more BMS ST in the 30-day to 1-year time period (late), and numerically more DES ST in the >1-year time period (very late). The pooled analysis of the 4 randomized, blinded, controlled, SES stent trials demonstrated a protocol-defined ST rate beyond 1 year of 0.6% (5 of 848 cases) in the SES arm and 0% (0 of 843 cases) in the BMS control arm (P<0.05), and a numerical, non–statistically significant increase in the number of ARC definite and probable ST in the SES group (Table 4). A similar analysis of PES randomized controlled trials also demonstrated a significant increase in the protocol-defined rate of very late ST (>1 year) in the PES group compared with BMS and a numerical, non–statistically significant increase in the ARC rate of definite and probable ST in the PES group (Table 4). An additional, patient-level, pooled analyses of the PES randomized controlled trial data set reported a small but statistically significant increased risk (0.21% per patient per year) of late (>6 months) ST with the ARC definitions of definite and probable (DES 0.25 per 100 patient-years of follow-up versus BMS 0.04 per 100 patient-years, P=0.007).
For on-label use, the panel felt that, in total, the data were consistent with a numerical increase in very late (>1 year after implant) ST associated with DES use compared with BMS use, but that the true magnitude of the risk and the duration of the risk were uncertain. Given the convincing and persistent reduction in target vessel failure and TVR with DES, as well as evidence that indicates that mortality and MI rates are not different between DES and BMS patients, the panel unanimously agreed that, when used in accordance with their labeled indications, both the Cypher SES and the Taxus PES are safe and effective.
Recommendations for antiplatelet therapy as indicated in the DES product label are based on the clinical studies that led to device approval (Table 2). For the Cypher stent, clopidogrel or ticlopidine were administered for 2 to 3 months postprocedure. For the Taxus stent, 6 months of clopidogrel or ticlopidine therapy were administered in the pivotal clinical trial. Aspirin is recommended indefinitely for both stents.
The ideal duration of dual antiplatelet therapy for patients who receive on-label DES is unknown and the panel acknowledged that currently available data are insufficient to support a specific duration of dual antiplatelet therapy for all patients. The panel discussed the current (2005) American College of Cardiology/American Heart Association/Society for Cardiovascular Angiography and Interventions Guidelines for the Treatment of Patients Undergoing Percutaneous Coronary Intervention, which recommend thienopyridine treatment (75 mg clopidogrel daily), in addition to aspirin, ideally for up to 12 months after DES in patients at low risk for bleeding. Ultimately, the panel recommended addition of a synopsis of these guidelines to the current product label.
Off-Label Drug-Eluting Stent Use
DES were rapidly embraced by the interventional cardiology community in the United States and are used in >75% of percutaneous coronary interventions. Frequently, however, DES use occurs outside the relatively narrow anatomic substrates studied in the pivotal trials. It is estimated that as much as 60% of DES use may be off-label and occurs in more complex patients and with more complex coronary artery lesions such as saphenous vein bypass grafts, bifurcation lesions, chronic total occlusions, smaller diameter vessels, and patients with multivessel disease. It may be expected that the rate of serious adverse events associated with DES would be higher in these off-label populations compared with their less-complicated on-label counterparts.
The panel reviewed data from a number of “real-world” stent registries (periapproval, single-center, multicenter, United States–based, and international) to assess the safety and effectiveness of DES. A summary of selected registries is displayed in Table 5. Data from FDA-mandated, postapproval, single-arm DES registries from Cordis (1 year follow-up) and Boston Scientific (2 year follow-up) were presented. Other registries were used to make outcome comparisons between DES and BMS, although it should be noted that the patients in these registries were not randomized, and the choice of stent was at the discretion of the treating physician. Therefore, it is uncertain whether observed differences between the 2 types of stents were caused by a true difference in the performance of the stents, or caused by other confounding factors (such as physician preference in the selection of a specific stent).
Nevertheless, certain observations were consistent among the registries presented: 1) DES use results in a reduction in the need for TLR and TVR, and this benefit appears to persist for at least several years; 2) compared with on-label use, off-label DES use (like off-label BMS use) is associated with an increased risk of adverse events, such as death or the combined end point of death or nonfatal MI, which likely reflects the increased complexity of the patients and the lesions as noted above; 3) data are currently inadequate to assess the relative benefit of DES compared with BMS or DES compared with coronary artery bypass surgery in the off-label population. In particular, it is uncertain whether multivessel DES or coronary artery bypass grafting is the preferred approach for patients with multivessel disease. Data from several large-scale nonrandomized registries suggest that historical long-term survival with coronary artery bypass grafting may be better than multivessel stenting in selected patients, but again confounding factors may have contributed to these observations. The preferred treatment approach for these patients is unlikely to be established until the completion of prospective, randomized, controlled trials of cardiac surgery versus DES in the setting of multivessel disease.
The majority of registries suggested that no significant mortality difference existed between patients who received DES and those who received BMS. Similarly, data from the >19 000 patients in the Swedish Coronary Angiography and Angioplasty Registry with complete long-term follow-up for death and MI were presented, and the data demonstrated no significant difference between DES and BMS for death/MI (relative risk, 1.04; 95% confidence interval, 0.95 to 1.15) at 2.5 years of follow-up. Further analysis demonstrated an increased mortality risk after 6 months for DES compared with BMS (relative risk, 1.2; 95% confidence interval, 1.05 to 1.37), which was not observed in the first 6 months postimplantation (relative risk, 0.94; 95% confidence interval, 0.83 to 1.06). Importantly, this was a post hoc analysis not prespecified by design. In addition, because of the lack of randomization and potential differences in the DES versus BMS patient populations, the validity of these findings is uncertain.
Stent Thrombosis and Dual Antiplatelet Therapy in the Off-Label Population
Assessment of the incidence of DES ST in the off-label population is challenging because of varying definitions, patient populations, antiplatelet regimens, and durations of treatment. In total, the registry data suggested that: 1) the off-label population, like the on-label population, is affected by both early, late, and very late (>1 year after implant) ST; 2) the rate of DES ST in the off-label population appears higher than in the DES on-label population (the same could be said for BMS ST); 3) the relationship of very late ST to the cessation of antiplatelet therapy is uncertain and highlighted by anecdotal reports of DES ST in patients compliant with dual antiplatelet regimens; 4) the risk of ST in the first year after stent implantation is increased with premature discontinuation of antiplatelet therapy when compared with uninterrupted treatment.
The ideal duration of dual antiplatelet therapy for off-label DES patients is unknown. For example, 1 report from the Northern California and Colorado Kaiser Permanente organization suggested that no additional protective effect of clopidogrel existed beyond 9 months of treatment for SES or 12 months of treatment for PES. Premature discontinuation, however, of dual antiplatelet therapy after DES implantation does appear to be associated with an increased risk of ST, death, and MI. These risks may be even higher in the off-label compared with the on-label use of DES. The panel highlighted the need for improved physician and patient education about this important issue and discussed methods to better inform the public about the risks of premature discontinuation of antiplatelet therapy.
In weighing the risks and benefits of extended dual antiplatelet therapy, the panel recognized that for some patients the increased risk of bleeding may be of the same or greater magnitude as the risk of stent thrombosis itself (≈1% to 2%). Issues that affect medical compliance (eg, psychosocial, economic, and demographic) were also discussed. Most panelists agreed that dual antiplatelet therapy for at least 12 months should be advised for patients who receive DES in an off-label setting and who are not at high risk of bleeding, which is consistent with the current American College of Cardiology/American Heart Association/Society for Cardiovascular Angiography and Interventions Guidelines for the Treatment of Patients Undergoing Percutaneous Coronary Intervention. Patients who are unable to take dual antiplatelet therapy or who are not expected to adhere to uninterrupted dual antiplatelet therapy should be considered for alternative treatment strategies. Additional clinical studies to determine the ideal duration of antiplatelet therapy after DES implantation should be performed.
Limitations of Registries
The panel discussed the inherent limitations of registry data both with regard to the interpretation of the data presented and with regard to the design of future registries for DES after market evaluation. Specifically, the panel discussed the need for: 1) a prespecified comparison group (ie, BMS or coronary artery bypass grafting) depending on the patient population to be studied; 2) complete or nearly complete ascertainment; 3) sample sizes adequate to provide reliable estimates of relevant clinical outcomes; and 4) prespecified subgroup analyses (such as patients with diabetes mellitus who are more likely to have complex and multiple lesions). The panel agreed that post hoc subgroup analysis of current registries generated hypotheses but was not sufficient to identify particular subsets of patients that may benefit from one treatment or another.
Drug-eluting stents are associated with a significant sustained reduction in the need for repeat coronary revascularization procedures compared with bare metal stents. Although there appears to be a numerical increase in the frequency of very late stent thrombosis associated with DES use, the magnitude of this risk is uncertain. When used according to labeled indications, DES use appears safe. Off-label use of DES, like off-label use of BMS, is associated with a greater risk of adverse events such as death, MI, and stent thrombosis compared with on-label use. Dual antiplatelet therapy is recommended in patients at low risk for bleeding for at least 1 year. Additional research is needed to clarify the safety and effectiveness of DES use in specific off-label populations and to define the ideal duration of dual antiplatelet therapy.
Dr Maisel is the chair and Dr Yancy is a member of the Circulatory System Devices Advisory Panel. Dr Laskey served as the previous chair of the Circulatory System Devices Advisory Panel and is a US Food and Drug Administration consultant. Dr Yancy has received research support from GlaxoSmithKline, Medtronic, and Scios Inc; and honoraria from GlaxoSmithKline, Novartis, Nitromed, and Scios Inc. He has served on the Speakers’ Bureau for GlaxoSmithKline and Novartis and as a consultant or advisory board member for GlaxoSmithKline and Nitromed.
The opinions expressed in this article are the personal views of the authors and do not necessarily represent the policies, practices, positions, or opinions of the US Food and Drug Administration.
↵*A full transcript of the panel meeting may be found at http://www.fda.gov/ohrms/dockets/ac/cdrh06.html#circulatory.