Much has been written, but little is known, about the long-term outcomes after drug-eluting stent (DES). The introduction of DES has extended the use of stents from a passive platform for vessel wall scaffolding to an active biotechnology vehicle for local delivery of antiproliferative pharmacotherapy. Like any other therapy or medical technology, no treatment is perfect and free of potential adverse effects or complications. Nonetheless, DES have been the most significant progress in the field of interventional cardiology, drastically reducing the rates of restenosis and need for revascularizations.1–6
Recent publications of registries and meta-analyses have raised concerns among healthcare providers and the lay public regarding the long-term safety of DES.7–13 However, these publications have been criticized as being inconclusive because of an insufficient number of patients, the absence of concurrent controls, a limited duration of follow-up, or a lack of access to original source data.14 Some common terms need to be clarified for better understanding of the subject, including the new expanded definitions of stent thrombosis, randomized clinical trials (RCTs) versus registries, on- and off-label use of stents, and the role of long-term antiplatelet therapy.
To address the limitations of the trial-level meta-analyses, Stone et al. performed a patient-level pooled meta-analysis of data from nine double-blind trials in which patients were randomly assigned to receive DES or bare-metal stents (BMS) in single, previously untreated coronary lesions through four years of follow-up. Data were used from the original databases, as defined and adjudicated by the clinical events committees for each study. Both stent types were associated with marked reductions in revascularization, an advantage that was maintained through four years of follow-up. Compared with BMS, target lesion revascularization (TLR) rates were 7.8% versus 23.6% (p<0.001) with Cyper, and 10.1% versus 20.0% (p<0.001) with Taxus. The rates of death or myocardial infarction (MI) were not significantly different between the groups with DES at any point during the four years of follow-up, and correspond to previously published natural histories of patients with coronary artery disease (CAD) treated by percutaneous coronary revascularization. Stone’s findings differ from those of some other analyses mentioned above, which suggested that overall rates of stent thrombosis and death are higher with DES than with BMS.10,11
These discrepancies may be explained by the fact that Stone et al. included only randomized double-blind studies and had full access to the complete patient-level data of the trials compared with the trial-level meta-analyses, and thus their results are more accurate. The trial-level meta-analyses relied on estimations of event rates from limited published results, abstracts, and online summaries, and therefore are not as accurate as patient-level analyses.
Stent thrombosis is a dramatic clinical event associated with MI and high rates of mortality. Recent publications have suggested that DES may be associated with increased rates of late stent thrombosis and increased mortality compared with BMS. The protocols of all the RCTs with DES had a stringent definition for late stent thrombosis, demanding MI attributable to the target vessel, and angiographic documentation (site-reported or by quantitative coronary angiography) of thrombus or total occlusion of the target site, and freedom from an interim revascularization of the target vessel. It has been suggested that with this narrow definition some stent thrombosis cases may be missed, given the difficulties in defining stent thrombosis in the absence of angiographic confirmation or autopsy.4,15 Based on these principles, the Academic Research Consortium (ARC)—assembled by interventional cardiologists, investigators of DES studies, US Food and Drug Administration (FDA), and industry representatives—suggested a set of consensus definitions that will be implemented in clinical trials of DES.16 The ARC definition considers distinct reportable time-points: acute stent thrombosis (0–24 hours post-stent implantation); sub-acute stent thrombosis (>24 hours up to 30 days); late stent thrombosis (30 days to one year); and very late stent thrombosis (more than one year post-stent implantation). The ARC recognized three categories of evidence: confirmed/definite, probable, and possible. Confirmed/definite stent thrombosis was defined as the clinical event of acute coronary syndrome symptom, combined with either angiographic or pathological confirmation of stent thrombosis. Probable late stent thrombosis was defined as any MI in the territory of the implanted stent in the absence of any other obvious cause. Possible stent thrombosis included all unexplained deaths occurring at least 30 days after the procedure.
It should be remembered, though, that these extended definitions are liberal and, while trying to avoid missed stent thrombosis events, they include patients who had MI or sudden death unrelated to the DES implemented years earlier. Another weakness of the ARC definitions is not excluding patients who had interim revascularization between the index DES and the qualifying event; and, indeed, probable and possible stent thrombosis were found to be more frequent after treatment for in-stent restenosis.16 Furthermore, since revascularization procedures are much more common with BMS, thromboses after interim procedures would be expected to be more common with these stents.14 With this perspective in mind the reader can understand why the overall four-year cumulative rates of stent thrombosis increased from 0.6–1.3% (protocol definition) to 3.2–3.6% (ARC definition). Whatever definition is used, the rate of stent thrombosis at long follow-up was similar, without any statistically significant differences between DES and BMS in the RCTs, 3.6% versus 3.3%, respectively, in the Cypher RCTs, and 3.2% versus 3.5%, respectively, in the Taxus RCTs.16
One single-center randomized trial needs a distinct discussion. The Basel Stent Kosten Effektivitats Trial (BASKET) Late trial extended the follow-up of the BASKET trial beyond the original six months, and evaluated the comparison of DES with BMS.9 Although there were no significant differences between BMS and DES in the rates of death and MI, a trend toward a higher rate of late stent thrombosis with DES than with BMS was noted (2.6% versus 1.3%, respectively, at 7–18 months). However, all patients with major adverse cardiac events—including death, MI, TLR, and stent thrombosis—in the first six months after stent placement were excluded from the BASKET Late analysis. This is a major flaw in the study, since the vast majority of adverse events with BMS occur in the first six months,17 thus the results are biased in favor of BMS. The major advantage of DES over BMS is in the first six months, and when combining the first six months with the following year, no significant differences between DES and BMS are noted.
RCTs have the advantage of careful planning and execution with a valid control group that can lead to scientifically reliable results and conclusions. Usually, the operators in these studies are experienced interventionalists from selected high-volume centers. The study protocols adhere to relatively narrow enrolling criteria, excluding more complex clinical scenarios and anatomical conditions such as multi-vessel disease, small vessels, long segments, ostial lesions, bifurcations, bypass grafts, or restenosis lesions that are frequent in real-life interventional practice.
Thus, RCTs, by enrolling relatively simple lesions in less sick patients, may underestimate adverse outcomes. Registries, despite lacking the scientifically designed control group, may report a wide range of clinical and anatomical presentations, especially if including multiple centers and a wide geographical area. On the other hand, registries should be evaluated with caution because of their frequent retrospective profile, and being aware that the data processing and adjudication of events is less comprehensive in many large-scale multicenter registries than in RCTs.
Reports from outside the US presented larger patient cohorts. In a short- to intermediate-term follow-up report, including 2,229 patients from Italy and Germany treated with DES (Cypher or Taxus), the rate of stent thrombosis was 1.3%.18 The e-Cypher registry, holding more than 20,000 Cypher stent implantations in 41 countries, showed a short- and intermediate-term safety profile similar to the randomized trials, with one-year cumulative major adverse cardiac events of 5.8% and a stent thrombosis rate of 0.9%.19 Daemen et al. reported the three-year follow-up of 8,146 patients who were treated with either Cypher or Taxus stents in two large academic institutions in The Netherlands and Switzerland.20 They observed that the incidence of early stent thrombosis was similar for Cypher (1.1%) and Taxus (1.3%), but late stent thrombosis was more frequent with Taxus (1.8%) than with Cypher (1.4%; p=0.031). Late stent thrombosis occurred steadily with no evidence of diminution up to three years, at a constant rate of 0.6% per year. The largest reported registry came from Sweden, comparing 6,033 patients treated with DES and 13,738 patients treated with BMS. At three-year follow-up, DES were associated with an increased rate of death compared with BMS (adjusted relative risk of 1.18). This difference emerged after six months, when the risk of death was 0.5% higher and a composite of death or MI was 0.5–1% higher per year.13
Can the dichotomy in the outcomes of the reports with DES be explained? The answer emerges from the differences between RCTs and registries. The RCTs are the pivotal trials, providing the basic safety and efficacy data for the public health authorities (such as the FDA) to approve the use of the studied device. Based on the relatively narrow inclusion and exclusion criteria in the pivotal Sirolimus-Eluting Stent in de Novo Native Coronary Lesions (SIRIUS) and TAXUS randomized trials, the FDA has granted approval of the Cypher and Taxus stents, and labeled their use corresponding to the patients enrolled in these studies—relatively focal or short simple single lesions in native coronary arteries with diameter of 2.5–3.75mm.
These represent ‘on-label’ indication. Even before the use of DES, the common use of stents included treatment of a wider spectrum of lesions such as multi-vessel disease, small vessels, long segments, ostial lesions, bifurcations, bypass grafts, or restenosis lesions. It has been estimated that around 60% of stent use is in such ‘off-label’ indications.21 Data from the American College of Cardiology National Cardiovascular Data Registry (ACC-NCDR) database showed that the use of DES in off-label lesions such as acute MI, in-stent restenosis, bypass grafts, and chronic total occlusions comprise 25% of all the procedures done, and were associated with low rates of short-term adverse events. In registries, usually consecutive patients are enrolled, with no, or limited exclusion criteria. Such ‘all comers’ registries represent real-world practice and include a high proportion of patients with complex disease and coronary lesions. As discussed above, registries are descriptive, and lacking a control group weakens their scientific value. In some part, RCTs and ‘real-world’ registries may complement each other and illuminate the wide range of safety and efficacy of DES.
In December 2006, the Circulatory System Devices Advisory Panel of the FDA met in an effort to characterize the risks, timing, and incidence of DES thrombosis.21 The purposes of this meeting were to have a wide forum discussion on clinical data relevant to the issue of DES thrombosis and to address the appropriate duration of antiplatelet therapy after DES implantation. Based on the long-term follow-up of the randomized studies with the Cypher and Taxus stents, the panel agreed that compared with BMS, DES are not associated with an increased rate of all-cause mortality. Both approved DES are associated with a small increase in stent thrombosis compared with BMS that emerges one year post-stent implantation. However, based on the data available, this increased risk of stent thrombosis was not associated with an increased risk of death or MI compared with BMS.21
The FDA panel also addressed the broader use of DES in more complex patients and with coronary lesions—the use of a drug or device outside the FDA-approved indications. The panel stated that off-label use of DES is associated with an increased risk of stent thrombosis, death, or MI compared with on-label use. However, the panel agreed that the data on off-label use are limited, and additional studies are needed to determine optimal treatments for more complex patients. They also recommended that until more data are available, the DES labels should state that when DES are used off-label, patient outcomes may not be the same as the results observed in the clinical trials conducted to support marketing approval.21
Some parameters have been suggested as risk factors for stent thrombosis. These include stenting of complex lesions such as left main and bifurcations, chronic renal failure, prior brachytherapy, stent length, and premature discontinuation of the antiplatelet therapy.7–9,18,20,22 The latter has been clamed to be the most important risk factor, although late stent thrombosis events have been reported even in patients who were taking dual antiplatelet treatment. The occurrence of stent thrombosis in patients compliant with dual antiplatelet therapy may be related in part to antiplatelet non-responsiveness or inadequate inhibition of platelet function.23-25
There is no wide agreement about the optimal duration of dual antiplatelet therapy to prevent stent thrombosis. Patients in the initial pivotal RCT with DES received dual antiplatelet therapy for only two to six months, but for more complex lesions, the common practice in the US—as recommended also by the American College of Cardiology/American Heart Association and adopted by the FDA panel—is to continue them for at least 12 months. In addition, since even after one year there has been pathological documentation of delayed revascularization26 and very late stent thrombosis has been reported, longer treatment may be considered for patients at low risk for bleeding, especially after long segment or complex lesion stenting.
In summary, there are wide and convincing data on the long-term effectiveness of DES in significantly reducing the rates of in-stent restenosis and future revascularization. All pivotal RCTs, both individually and in patient-level meta-analysis, have demonstrated the short- and long-term safety of DES to be similar to that of BMS. However, some studies have suggested increased rates of very late stent thrombosis, especially in cases in which stents are used off-label, and with premature discontinuation of dual antiplatelet therapy.
Second-generation DES are going to be approved soon, and are expected to be less prone to stent thrombosis and thus safer than the first generation DES. However, their long-term safety and efficacy will have to be proved over the coming years in post-approval studies. Late stent thrombosis will continue to be a carefully scrutinized clinical complication after DES implantation, especially in an environment where more complex lesion subsets are being treated with increased numbers of DES.