Author + information
- Received November 4, 2008
- Revision received April 6, 2009
- Accepted April 7, 2009
- Published online September 1, 2009.
- Adriano Caixeta, MD, PhD⁎,
- Martin B. Leon, MD⁎,
- Alexandra J. Lansky, MD⁎,
- Eugenia Nikolsky, MD, PhD⁎,
- Jiro Aoki, MD, PhD⁎,
- Jeffrey W. Moses, MD⁎,
- Joachim Schofer, MD†,
- Marie-Claude Morice, MD‡,
- Erick Schampaert, MD§,
- Ajay J. Kirtane, MD, SM⁎,
- Jeffrey J. Popma, MD∥,
- Helen Parise, DSc⁎,
- Martin Fahy, MSc⁎ and
- Roxana Mehran, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Roxana Mehran, 161 Fort Washington Avenue, 5th Floor, New York, New York 10032
Objectives Five-year clinical follow-up has been scheduled per protocol by the 4 Cypher (Cordis/Johnson & Johnson, Warren, New Jersey) sirolimus-eluting stent (SES) versus bare-metal stent (BMS) randomized trials.
Background A delayed arterial healing response after drug-eluting stent implantation has raised concerns about the long-term safety of drug-eluting stents.
Methods In a pooled analysis of 4 randomized trials, 1,748 patients were assigned to receive either an SES (n = 878) or BMS (n = 870).
Results At 5 years, there was no significant difference in the rate of death, myocardial infarction (MI), or the composite of death/MI between the 2 groups (15.1% in the SES group vs. 13.6% in the BMS group; p = 0.36). The 5-year incidence of stent thrombosis by the Academic Research Consortium definition did not differ between SES and BMS (definite/probable stent thrombosis, 2.1% vs. 2.0%; p = 0.99). The incidence of very late stent thrombosis was also similar between the SES and BMS groups (1.4% vs. 0.7%; p = 0.22). The annualized rates of definite/probable stent thrombosis after 1 year were 0.4% for SES and 0.2% for BMS. The 5-year incidence of target vessel revascularization was significantly lower in the SES group (15.2% vs. 30.1%; p < 0.0001).
Conclusions In this patient-level pooled analysis, overall use of SES compared with BMS demonstrated persistent superior efficacy at 5 years in terms of a reduction in target vessel revascularization, without an increase in rates of death, MI, or stent thrombosis. (The Initial Double-Blind Drug-Eluting Stent vs Bare-Metal Stent Study, NCT00233805; The Study of the BX Velocity Stent in the Treatment of De Novo Artery Lesions, NCT00381420; Study of Sirolimus-Coated BX VELOCITY Balloon-Expandable Stent in Treatment of de Novo Native Coronary Artery Lesions [SIRIUS], NCT00232765; The Study of the BX VELOCITY Stent In Patients With De Novo Coronary Artery Lesions, NCT00235144)
Sirolimus-eluting stents (SES) dramatically reduce the incidence of restenosis and rates of target lesion revascularization (TLR) (1). Soon after approval, SES were enthusiastically adopted even beyond the on-label, U.S. Food and Drug Administration-approved indications. However, the theoretical thrombogenicity of drug-eluting stents (DES) and the delayed arterial healing response seen with DES use raised concerns that these devices might be associated with an increase in the incidence of very late stent thrombosis (2,3).
Although 4-year clinical outcomes in a pooled analysis of data from the 4 double-blind SES versus bare-metal stent (BMS) trials have been published recently (4), the SES manufacturer was required by the U.S. Food and Drug Administration to follow patients for 5 years (5). To further address the issue of long-term safety and efficacy of SES, we investigated 5-year clinical outcomes in a pooled analysis of the 4 SES versus BMS randomized trials.
Patient population and study procedure
We performed a patient-level pooled analysis of the 4 multicenter, double-blinded, randomized SES versus BMS trials, including the RAVEL (Randomized Study With the Sirolimus-Coated Bx Velocity Balloon-Expandable Stent in the Treatment of Patients With De Novo Native Coronary Artery Lesions), SIRIUS (Sirolimus-Eluting Stent in De Novo Native Coronary Lesions), E-SIRIUS (European–SIRIUS), and C-SIRIUS (Canadian–SIRIUS) trials, in which patients with single de novo lesions were treated with either a Cypher SES or an uncoated Bx Velocity BMS of identical design (both Cordis/Johnson & Johnson, Warren, New Jersey). The protocol and principal results of each trial have been published elsewhere (1,6–8).
Definitions and clinical end points
The primary safety end point of this analysis was death or myocardial infarction (MI). The secondary safety end point was stent thrombosis using the Academic Research Consortium (ARC) definition up to 5 years (9). The primary efficacy end point was target vessel revascularization (TVR) up to 5 years. Follow-up information was collected by the investigating sites, including telephone contact at 1 year and annually for 5 years thereafter. Five-year follow-up was completed in 87.1% of patients. To be included in the analysis, patients must have had at least 1,800 days of follow-up.
We pre-specified that we would compare the clinical outcomes between the SES and BMS groups in patients with diabetes, because the mortality rate was significantly higher in the SES group at 4 years in the published report (4).
Definitions of major adverse cardiac events were consistent across the trials (1,6–8). Members of the independent Clinical Events Committee retrospectively readjudicated all clinical and angiographic data based on the ARC definition of stent thrombosis (9).
Patient-level data were pooled from the 4 randomized trials comparing SES and BMS. Interactions between trial and stent on 5-year death, MI, TLR, TVR, protocol thrombosis, and ARC-defined thrombosis were not statistically significant, justifying pooling of the 4 studies. Binary variables are summarized as counts and percentages and compared using chi-square tests or the Fisher exact test where appropriate. Continuous variables are summarized as means and standard deviations and compared using ttests. Five-year outcomes are summarized as Kaplan-Meier estimates and compared using log-rank tests and hazard ratios. Kaplan-Meier event curves are presented and compared using log-rank tests. To assess events occurring between years 4 and 5, a landmark analysis was performed. Cox proportional hazards models using stepwise selection were used to determine multivariate predictors of clinical events. All statistical tests were 2-tailed. A p value <0.05 denoted significance.
Baseline and procedural characteristics
Between August 2000 and April 2002, 1,748 patients at 115 international centers were assigned to either SES (n = 878) or BMS (n = 870). The 2 groups were well matched for all baseline and procedural characteristics except for a lower prevalence of diabetes in patients randomized to SES and a lower rate of type C lesions in patients randomized to BMS (Table 1).
Clinical outcomes up to 5 years
The 5-year rates of all-cause death, cardiac death, and MI were similar between the 2 groups (Table 2).There were no significant differences in the composite end point of death or MI at 5 years and from 1 to 5 years between the 2 groups. The 5-year cumulative incidence of TVR was nearly doubled in patients randomized to BMS. The striking difference in the TVR rates peaked at approximately 1 year and then remained through 5 years (Fig. 1).
The incidence of any stent thrombosis did not differ between the 2 groups at 5 years (Table 3,Fig. 2).The rates of very late definite stent thrombosis (1.1% in the SES group vs. 0.5% in the BMS group; p = 0.16) and very late definite or probable stent thrombosis (1.4% in the SES group vs. 0.7% in the BMS group; p = 0.22) were also similar between the 2 groups.
Late definite or probable stent thrombosis was more frequent in the BMS group during the first year (0.2% in the SES group vs. 0.9% in the BMS group; p = 0.05), whereas very late definite or probable stent thrombosis tended to be more frequent in the SES group (1.4% vs. 0.7%; p = 0.22) (Table 3). The annual definite or probable stent thrombosis rate from 1 to 5 years was low in both groups and did not differ significantly between SES and BMS (0.4% vs. 0.2% per year) (Table 4).The annualized rates of TLR, TVR, and nontarget lesion TVR (remote TVR) after 1 year were similar for SES and BMS. In the landmark analysis, there were no significant differences in rates of the clinical end points, including stent thrombosis between years 4 and 5 in patients treated with SES versus BMS (Fig. 3).
Clinical outcomes in patients with diabetes
At 5 years, diabetic patients treated with SES versus BMS had significantly higher rates of mortality (15.9% vs. 9.0%; p = 0.03) and Q-wave MI (3.3% vs. 0.4%; p = 0.03) and had no significant differences in rates of definite, definite or probable, or any stent thrombosis either at 5 years or from 1 to 5 years (Table 5,Fig. 4).At 5 years, the use of SES compared with BMS reduced the rate of TVR from 37.1% to 17.7% (p < 0.0001).
Independent predictors of cardiac death or MI for the entire population were smoking (hazard ratio [HR]: 1.85; 95% confidence interval [CI]: 1.34 to 2.57; p = 0.0002), congestive heart failure (HR: 1.79; 95% CI: 1.13 to 2.83; p = 0.01), diabetes (HR: 1.37; 95% CI: 1.01 to 1.86; p = 0.041), prior MI (HR: 1.49; 95% CI: 1.12 to 1.97; p = 0.006), older age (HR: 1.05; 95% CI: 1.03 to 1.06; p < 0.0001), and total number of stents implanted (HR: 1.02; 95% CI: 1.01 to 1.03; p = 0.001). In the diabetes subset, predictors of cardiac death or MI included congestive heart failure (HR: 2.33; 95% CI: 1.27 to 4.29; p = 0.006), smoking (HR: 2.06; 95% CI: 1.11 to 3.81; p = 0.02), and older age (HR: 1.05; 95% CI: 1.02 to 1.08; p = 0.0005).
The main findings of this analysis comparing 5-year clinical outcomes in patients treated with either SES or BMS for on-label indications were as follows: 1) the incidences of the composite of death or MI as well as ARC-defined stent thrombosis were similar up to 5 years and from 1 to 5 years between the 2 groups; 2) a significant difference in the incidence of TVR favoring the SES group versus the BMS group within the first year was maintained up to 5 years; and 3) in diabetic patients, the rates of mortality and the composite of death or Q-wave MI were significantly higher in patients receiving SES versus BMS.
Long-term safety of SES
Due to safety concerns (5), 5-year follow-up was required from the manufacturer for patients enrolled in randomized DES versus BMS trials. In 1 community-based registry, very late stent thrombosis occurred at a constant rate of 0.6% per year up to 3 years after DES implantation (2). Higher rates of very late stent thrombosis with SES rather than BMS (1.4% vs. 0%; p = 0.02) have been also shown at 4-year follow-up in another large-scale registry (10). However, despite concern initially raised about DES safety based on 3-year outcomes from the large Swedish registry, repeat analysis performed at later follow-up (at 4 years) demonstrated no differences in hard clinical end points between patients treated with DES or BMS (11). These latter results are consistent with the present study, in which rates of death and MI were similar for SES- and BMS-treated patients during 1 to 5 years of follow-up. Even though the annual incidence of definite or probable stent thrombosis after the first year was 2-fold higher in the SES group compared with the BMS group (0.4% vs. 0.2%), the difference was not statistically significant. Still, the lack of an observed difference between hard clinical end points, including stent thrombosis, for SES- and BMS-treated patients in this study may be due to inadequate statistical power.
The current study provides information on the longest (5-year) follow-up, including an update on clinical events occurring between 4 and 5 years. After the completion of 4-year follow-up (4), patients treated with SES versus BMS had 8 versus 9 additional cases of cardiac death, 3 versus 2 cases of Q-wave MI, and 4 versus 2 cases of very late definite or probable stent thrombosis, respectively.
Clinical outcomes in diabetic patients
Mortality in diabetic patients was significantly higher in the SES group than the BMS group, mainly due to a higher rate of cardiac death beyond 1 year. Among 19 diabetic patients who experienced cardiac death in the SES group, 11 (57.9%) died of unknown causes; these cases were adjudicated as cardiac death and possible stent thrombosis. However, it is not certain that this higher mortality is in fact related to late stent thrombosis. Furthermore, the small number of diabetic patients in the current study makes it underpowered to detect differences between the 2 groups for rare events such as death and stent thrombosis. In addition, the surprisingly low event rates in the BMS group beyond 1 year may have biased the outcomes in favor of BMS. Therefore, the difference in SES versus BMS mortality seen at 5 years in patients with diabetes may be due to chance alone. Larger, more recent studies provide evidence of superior outcomes (12,13), including lower mortality and fewer MIs (12), in diabetic patients treated with DES compared with BMS.
Long-term efficacy of SES
In this pooled analysis, TVR at 5-year follow-up was reduced nearly 2-fold in the SES versus BMS group, mainly due to remarkably lower rates of TLR during the first year. The highly significant differences in rates of TLR and TVR between patients randomized to SES versus BMS persisted from 1 to 5 years. Thus, it is unlikely that the use of SES is associated with a late catch-up phenomenon.
The results of this analysis are hypothesis-generating. Information about patient adherence to antiplatelet therapy at follow-up was not collected. Insufficient antiplatelet therapy as a significant predictor of late stent thrombosis was not a concern in 2002, when patient enrollment in the trials was completed. Therefore, a relationship between duration of dual antiplatelet therapy and stent thrombosis cannot be established. Finally, the results of this study cannot be generalized for the use of SES beyond the approved (off-label) indications.
In this pooled analysis, the use of SES compared with BMS demonstrated persistent superior efficacy in terms of a reduction in TVR without an increase in rates of death, MI, or stent thrombosis for up to 5 years and from 1 to 5 years.
The authors thank Jason Kahn for the editing of this paper.
Dr. Leon, within the last 12 months, was a member of Cordis/Johnson & Johnson Science Advisory Board. Dr. Lansky received a research grant from Cordis/Johnson & Johnson. Dr. Moses has served as a speaker for Cordis/Johnson & Johnson (minor). Dr. Schampaert has served as a speaker (honoraria <$10,000) with Cordis Canada/Johnson & Johnson and Boston Scientific, and was on the Advisory Board of Cordis/Johnson & Johnson, Boston Scientific, Eli Lilly, Sanofi-Aventis, and GlaxoSmithKline. Dr. Kirtane was a consultant (advisory board/lecture fees) for Abbott Vascular and Medtronic Cardiovascular. Dr. Mehran received research grants (significant) from Abbott Vascular, Cordis/Johnson & Johnson, Boston Scientific, and The Medicines Company, and has served as a speaker/consultant (modest) for Abbott Vascular, Cordis/Johnson & Johnson, Medtronic, Lilly/Daiichi Sankyo, and The Medicines Company.
- Abbreviations and Acronyms
- Academic Research Consortium
- bare-metal stent(s)
- drug-eluting stent(s)
- myocardial infarction
- sirolimus-eluting stent(s)
- target lesion revascularization
- target vessel revascularization
- Received November 4, 2008.
- Revision received April 6, 2009.
- Accepted April 7, 2009.
- American College of Cardiology Foundation
- Laskey W.K.,
- Yancy C.W.,
- Maisel W.H.
- Schampaert E.,
- Cohen E.A.,
- Schluter M.,
- et al.
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.
- Garg P.,
- Normand S.L.,
- Silbaugh T.S.,
- et al.
- Stettler C.,
- Allemann S.,
- Wandel S.,
- et al.