Author + information
- Received February 16, 2007
- Revision received June 27, 2007
- Accepted July 2, 2007
- Published online November 20, 2007.
- J. Dawn Abbott, MD, FACC⁎,*,⁎ (, )
- Matthew R. Voss, MD⁎,*,
- Mamoo Nakamura, MD†,*,
- Howard A. Cohen, MD, FACC‡,*,
- Faith Selzer, PhD†,*,
- Kevin E. Kip, PhD*,
- Helen A. Vlachos, MSc†,*,
- Robert L. Wilensky, MD, FACC§,* and
- David O. Williams, MD, FACC⁎,*
- ↵⁎Reprint requests and correspondence:
Dr. J. Dawn Abbott, Rhode Island Hospital, 814 APC, 593 Eddy Street, Providence, Rhode Island 02903.
Objectives We investigated the effectiveness and safety of drug-eluting stents (DES) as used in routine clinical practice.
Background Randomized trials have shown that DES prevent target vessel revascularization in selected patients, but whether this translates into superior outcomes, compared with bare-metal stents (BMS), for the full spectrum of patients treated with DES in North America is unknown.
Methods Patients in the National Heart, Lung, and Blood Institute Dynamic Registry enrolled in 2004 who received at least 1 DES (n = 1,460) were compared with 1,763 patients enrolled in the recruitment period immediately preceding the approval of DES (2001 to 2002) who received at least 1 BMS.
Results Patients receiving DES more often had diabetes mellitus and less often presented with an acute myocardial infarction (MI). At 1 year, cumulative death and MI was 7.6% in DES- and 8.7% in BMS-treated patients (adjusted hazard ratio [HR] 0.88, 95% confidence interval [CI] 0.68 to 1.15; p = 0.34). The 1-year rate of target vessel revascularization was 5.0% in DES and 9.2% in BMS patients (p < 0.001), and the risk of any repeat revascularization by percutaneous coronary intervention or coronary bypass was lower in DES patients (adjusted HR 0.38, 95% CI 0.25 to 0.60; p < 0.001). Patients with both simple and complex lesion characteristics benefited from DES with lower risk of repeat target vessel revascularization by percutaneous coronary intervention compared with BMS (any complex lesion: adjusted HR 0.57, 95% CI 0.39 to 0.83; absence of any complex lesion: adjusted HR 0.44, 95% CI 0.28 to 0.71). The 1-year incidence of stent thrombosis was 1.0% in DES patients.
Conclusions The generalized use of DES resulted in better outcomes than BMS, with fewer clinically driven revascularization procedures and similar rates of death and MI at 1 year.
Although the effectiveness of drug-eluting stents (DES) in reducing in-stent restenosis and target vessel revascularization has been confirmed in many patient and lesion subsets (1–11), questions remain regarding their generalized use. Even in clinically stable patients with de novo lesions enrolled in randomized trials comparing DES with bare-metal stents (BMS), there is concern regarding late stent thrombosis rates (12–17). The risk of stent thrombosis may be exaggerated in unselected populations treated with DES and may be related to adjunctive antiplatelet therapy, but whether this risk is different from similar patients treated with BMS is unclear (18–20). Importantly, since their commercialization DES have been universally accepted as superior to BMS and are being used for most patient and lesion subsets and in as many as 95% of stent procedures (21–22). To this point, data comparing such use of DES with BMS in North America are lacking. Therefore, we compared 2 patient cohorts from the multicenter National Heart, Lung, and Blood Institute Dynamic Registry, one consisting of consecutive patients treated with at least 1 DES and the other of patients treated with a BMS just before approval of DES.
Study design and patient population
The Dynamic Registry is a prospective observational study of consecutive patients undergoing percutaneous coronary intervention (PCI) at selected clinical centers in North America during prespecified time intervals or “waves.” Five enrollment waves of approximately 2,000 consecutive patients each have been collected over the past 10 years. Each wave is intended to capture an experience with specific new device technologies and examine trends in PCI. Enrollment for wave 3 was completed between October 2001 and March 2002 and represented the last wave before the routine availability of DES. Wave 4 extended from February 2004 to May 2004, a time when both BMS and DES were routinely available. Preliminary review of wave 4 findings, however, indicated a substantial bias in stent selection with a predominance of higher-risk patients, such as those with acute MI, more often receiving BMS. Accordingly, for the present BMS and DES comparison, BMS patients were selected from wave 3, in which patient selection was more similar to that of the DES patients in wave 4. For this analysis, only patients who received a stent are included. From a total of 2,047 patients recruited in wave 3, 1,763 received 1 or more BMS. Wave 4 recruited 2,112 patients of which 1,460 received at least 1 DES.
Data collection and protocol
For each wave, demographic, clinical history, baseline angiographic, and procedural data during the index PCI hospitalization and 1-year follow-up events were collected by trained research coordinators who used standardized report forms and were guided by a manual of operations and definitions. Information about in-hospital outcomes was obtained from review of hospital records and coronary angiograms. All information was site determined. Follow-up data were obtained at 1 month, 6 months, and 1 year. Medication use was recorded on the follow-up forms. In wave 3, clopidogrel was reported under cardiac medication-other, whereas in wave 4, patients were specifically queried for clopidogrel use. In waves 3 and 4, 89.4% and 99.6%, respectively, of enrolled patients consented to follow-up. One-year follow-up data was available in 93% of wave 3 patients and 96% of wave 4 patients that consented to follow-up. Medical records were reviewed whenever possible for patients requiring repeat hospitalization. In addition to patient telephone interviews, referring physicians and institutions were contacted whenever necessary for additional information.
During follow-up, coronary angiography was obtained as clinically indicated by symptoms or documentation of myocardial ischemia. Planned staged PCI was not considered a repeat PCI. No angiographic reevaluation was performed unless clinically driven. Lesion-specific data was collected for repeat PCI if available to determine target vessel revascularization. In wave 3, 183 patients reported a need for a repeat PCI, and repeat lesion data was available in 172 patients. In wave 4, 124 underwent a repeat PCI, and repeat lesion forms were available for 111 patients.
Death is defined as all-cause mortality. The end point of major adverse cardiac events (MACE) includes death, MI, and any repeat revascularization, and repeat PCI includes both target and nontarget vessel interventions. Target vessel revascularization was defined as a repeat revascularization procedure involving the initially treated artery. Specific data on subacute (0 to 30 days) and late (31 days to 1 year) stent thrombosis was collected in wave 4 (n = 1,460 DES patients) but not in wave 3. Only definite stent thrombosis, defined as angiographically confirmed cases, is reported in this study.
Patient characteristics pertaining to the index PCI, including demographics, medical history, cardiac presentation, periprocedural medications, procedural characteristics, and in-hospital outcomes were compared between stent types by Student ttests or Wilcoxon nonparametric tests for continuous variables and by chi-square test or Fisher exact test for categoric variables. Similar methods were used for lesion-level analyses. One-year event rates were calculated using the Kaplan-Meier approach, and unadjusted comparisons of survival curves were performed using the log-rank test. Adjusted survival curves for the composite outcomes death/MI and need for repeat revascularization were plotted and compared using inverse probability weights (23). Cox proportional hazards modeling was used to estimate 1-year hazard ratios (HRs) for adverse clinical events in relation to stent type and 1-year death/MI and target lesion revascularization by PCI in relation to both lesion morphology and stent type. Covariate adjustment was performed such that clinical, demographic, and procedural variables were entered individually into outcome-specific models that included an indicator variable for stent type. Confounding variables were assessed in a forward stepwise manner to determine the final adjusted model. The screened variables with a p value of <0.20 were included in the forward selection process, and those where the p value was <0.10 remained in the model. All models were age adjusted regardless of the p value. The variables remaining in the final models for adverse events by stent type are listed in the Online Appendix. Covariate adjustment for outcomes based on lesion and stent type were done in a similar fashion. Patients who did not experience the outcome of interest were censored at the last known date of contact or at 1 year if contact extended beyond 1 year. Proportional hazards assumptions were evaluated and met. Interactions between stent type and preselected variables (those thought to be biologically plausible) were formally evaluated. For all analyses, a 2-sided p value of <0.05 was considered to be statistically significant.
Baseline and procedural characteristics
Baseline characteristics comparing BMS- and DES-treated patients are presented in Table 1.Mean age (64.4 vs. 63.7 years) and female gender (35.8% vs. 33.3%) were similar between BMS and DES patients. The BMS-treated patients were less likely to have diabetes (29.1% vs. 34.3%) and other cardiovascular risk factors, prior percutaneous coronary revascularization procedures, and comorbidities. The extent of coronary artery disease did not differ, but the procedural indication was more often an acute MI in BMS patients (29.8% vs. 23.8%). Medications used before the procedure differed between groups, with the DES group receiving more thienopyridines and BMS group more glycoprotein IIb/IIIa inhibitors.
Lesion characteristics and procedural outcomes
A total of 2,551 lesions in the BMS group and 1,995 lesions in the DES group were attempted (Table 2).Lesion location and reference vessel diameter were similar between groups. There were several differences in the types of lesions treated, with more total occlusions, thrombus, and bifurcation lesions in patients treated with BMS and more calcification and complex lesion types by American College of Cardiology/American Heart Association criteria in DES. Although the prevalence of these lesion features differed between the 2 groups, the magnitudes of the differences were small. Rates of complications were low and angiographic success high in both groups.
All patients in the present analysis received at least 1 stent, but not all attempted lesions per patient were stented. A higher proportion of attempted lesions among DES patients received a stent than in the BMS group (95% vs. 91%, respectively; p < 0.001). In the DES group, sirolimus-eluting stents were used in 59.8% of treated lesions and paclitaxel-eluting stents in 31.5%. Some lesions in the patients in whom DES was attempted were finally treated with BMS only (4.4%) or a combination of DES and BMS (2.3%). The mean number of stents per patient was slightly higher in the DES patients compared with those receiving only BMS (1.6 ± 0.9 vs. 1.5 ± 0.9; p = 0.004).
Failure of stent delivery was evaluated in the DES group. There were 21 lesions where a DES was unable to be delivered. For 9 lesions, a BMS was successfully deployed when a DES failed. In 12 lesions, neither a DES nor BMS could be delivered, but the lesion was treated successfully with balloon angioplasty.
Rates of in-hospital mortality (DES 0.5% vs. BMS 1.1%; p = 0.06), MI (DES 2.2% vs. BMS 1.9%; p = 0.60), and coronary bypass (DES 0.1% vs. BMS 0.3%; p = 0.10) were low and similar in both groups. At discharge, more patients in the DES group were prescribed aspirin (DES 96.8% vs. BMS 94.0%; p < 0.001), or a thienopyridine (DES 99.0% vs. BMS 95.5%; p < 0.001).
Clinical outcomes at 1 year
At 1 year, the observed cumulative incidences of death and MI were similar between the DES and BMS groups (Tables 3 and 4).⇓Patients treated with DES had significantly less revascularization than BMS, including target vessel revascularization by PCI (5.0% vs. 9.2%, respectively; p < 0.001), any PCI (9.0% vs. 12.7%, respectively, adjusted HR 0.65, 95% confidence interval [CI] 0.52 to 0.82; p < 0.001), and coronary bypass surgery (1.2% vs. 3.1%, respectively, adjusted HR 0.34, 95% CI 0.19 to 0.60; p < 0.001). Overall MACE was 15.5% in the DES patients and 20.9% in BMS (adjusted HR 0.67, 95% CI 0.56 to 0.81), with the reduced risk in DES-treated patients driven by lower repeat revascularization. The 1-year freedoms from repeat revascularizations for both BMS and DES patients are depicted in Figure 1,indicating a substantial divergence at approximately 4 months after PCI that is sustained to 1 year. For 1-year freedom from death/MI, no divergence of events was noted between BMS and DES patients.
In subgroup analyses, the overall comparable safety profile of DES versus BMS and the overall reduced need for repeat revascularization with DES were consistently observed. The only differential effect noted (p = 0.04) was for the outcome need for repeat revascularization in which the adjusted HR comparing DES to BMS was statistically lower (but in the same direction) in single-vessel disease patients (adjusted HR 0.35, 95% CI 0.22 to 0.57) than in multivessel disease patients (adjusted HR 0.66, 95% CI 0.51 to 0.85).
One-year event rates according to the complexity of attempted lesions are shown in Table 5.The incidence of death/MI is greater among patients with one or more complex lesion characteristic than in those with none, and no differences (p > 0.05) were noted between BMS and DES patient subsets. Rates for repeat target vessel revascularization by PCI trended or were significantly lower in patients receiving DES compared with BMS regardless of the presence or absence of a specified complex lesion. After adjustment (Fig. 2),these relationships remained, with a lower risk of repeat target vessel revascularization by PCI in DES patients irrespective of the presence (HR 0.57, 95% CI 0.39 to 0.83) or absence (HR 0.44, 95% CI 0.28 to 0.71) of any complex lesion type and no appreciable difference in death/MI according to lesion or stent type.
Information regarding stent thrombosis was obtained for patients enrolled in wave 4 but was not available for previous waves. Therefore, findings are limited to the DES group only. The overall rate of definite stent thrombosis in DES-treated patients was 1.0% (n = 14), with 11 events occurring before day 31, and 3 between day 31 and 1 year. All 3 late thrombosis events presented as acute MI and were nonfatal. One occurred at day 37 in a single Taxus stent (Boston Scientific, Natick, Massachusetts) in the left anterior descending artery. The second was at day 91 in a bifurcation lesion involving the left anterior descending treated with 2 Cypher stents (Cordis, Miami Lakes, Florida) using the crush technique. The third was at day 344 in 2 Taxus stents in the left circumflex artery, 2 weeks after discontinuing aspirin and clopidogrel in preparation for a surgical procedure.
For patients that consented to follow-up after discharge, who did not die, and had a 1-year follow-up form, 15.6% in wave 3 reported using clopidogrel and 61.3% in wave 4 at 1 year (p < 0.001). Aspirin use was 85.8% in wave 3 and 90.7% in wave 4 (p < 0.001). Baseline characteristics and selected outcomes were assessed according to clopidogrel use at 1 year in wave 4 patients. Compared with nonclopidogrel users, clopidogrel users were more likely to have prior PCI (37.3% vs. 25.2%; p <.0001), previous treatment with a stent, prior MI (28.8% vs. 23.4%; p = 0.03), and multivessel disease. At 1 year, MIs were reported in 3.1% of non-clopidogrel users and 3.9% of clopidogrel users (p = 0.50), and the rate of repeat PCI was higher in the clopidogrel users (3.9% vs. 11.6%, respectively; p < 0.001).
Despite the expanded use of DES in patient and lesion subsets that were more complex than in randomized controlled trials, we found that patients treated with DES had a substantial reduction in clinically driven target vessel revascularization compared with BMS-treated patients. The overall risk of repeat revascularization by PCI or coronary bypass after adjustment was 43% lower in DES patients, confirming the important contribution of target vessel revascularization, presumably from in-stent restenosis, to this end point. Importantly, these favorable treatment effects were achieved without excess hazard of death or MI; moreover, we observed a low rate of late stent thrombosis in DES patients.
In this investigation, we compared outcomes for DES patients with similar types of patients receiving BMS when they were used broadly before commercialization of DES. Although separated in time by a few years, no other technologic advances other than DES occurred between recruitment of the BMS and DES patients. We believed that the patients who received BMS in this investigation would have likely received a DES had they been available. A large proportion of both BMS and DES patients had high-risk clinical features, including diabetes mellitus and acute coronary syndromes, and lesion characteristics, including ostial, bifurcation, calcified, and total occlusion lesions. Although there were statistical differences in baseline and lesion characteristics that may limit comparisons, some favored BMS and others DES, and adjustment had no meaningful influence on any of the outcomes. Finally, when baseline differences were observed, the absolute magnitude of these differences was small.
The present prospective observational study, carried out at multiple centers in North America and inclusive of both sirolimus- and paclitaxel-eluting stents, demonstrated a clinically driven target vessel revascularization rate of 5.0% in DES- and 9.2% in BMS-treated patients. The BMS patients had lower rates of target vessel revascularization than observed in earlier randomized trials of BMS versus DES that included protocol-driven angiographic follow-up as part of the trial design (2,3,24). In the SIRIUS (Sirolimus-Eluting Stent in Coronary Lesions) trial, the BMS arm had a high target lesion revascularization rate of 20% at 12 months compared with 4.9% for the SES arm (24). Similarly, in the TAXUS (Paclitaxel-Eluting Stents in the Treatment of Longer Lesions) IV trial, the BMS arm had target vessel revascularization rate of 17.1% compared to 7.1% for the PES arm (2). Whether BMS stent design, need for angiographic follow-up, or both contributed to those higher rates of target lesion revascularization is unclear. Patients in the randomized TAXUS V trial, which included complex lesions (type B2/C lesions) in 78% of patients, had target vessel revascularization rates of 12.1% and 17.3% for the PES and BMS groups respectively (3). Accounting for increased rates in both groups in the setting of angiographic follow-up in this trial, the absolute magnitude of reduction we observed for unselected patients in routine practice were similar.
Compared with other registries, such as the RESEARCH (Rapamycin-Eluting Stent Evaluated at Rotterdam Cardiology Hospital) registry, we observed higher absolute event rates but similar trends in outcomes (25). Both registries showed that compared with BMS patients, DES patients had lower rates of target vessel revascularization and MACE and similar rates of death and MI.
In the SCAAR (Swedish Coronary Angioplasty Registry), at up to 3 years of follow-up, the adjusted rate of any repeat PCI or revascularization was lower in DES patients but the absolute rates were similar (26). A repeat revascularization procedure was performed in 15.2% of patients initially treated with DES and in 16.5% with BMS. Clinically driven restenosis, however, was 60% lower among DES patients compared with BMS patients, with an absolute reduction of approximately 3%. The composite risk of death and MI was similar in patients with DES and BMS, but a higher risk of mortality was suggested after 6 months in the DES patients (26). Whether a longer duration of dual antiplatelet therapy or the shorter follow-up period in the Dynamic Registry is responsible for the different findings is unclear. Dynamic Registry patients overall, however, were older and more often had diabetes mellitus and acute coronary syndromes than in the RESEARCH registry and had more diabetes mellitus, hypertension, and prior PCI and coronary bypass than the SCAAR, reflecting different risk profiles in the studies. These differences may reflect variations between multicenter versus single-center practice patterns within and outside the U.S. The characteristics of patients in the Dynamic Registry and the DEScover registries, both carried out primarily in U.S. centers, are similar as are the 1-year event rates (21). Unlike the DEScover registry, the present report contains far more BMS patients with baseline features more similar to the comparative DES cohort.
In terms of safety, we observed similar low in-hospital complications in BMS and DES treated patients. As in randomized trials and the RESEARCH registry, no significant difference was observed in the observed rates of death and MI at 1 year (1,2,25). Over 1 year, we observed stent thrombosis in 1.0% of DES patients, with 0.2% occurring after 30 days. In the RESEARCH registry, angiographically documented stent thrombosis occurred in 0.4% of SES within 30 days and no additional cases were observed out to 1 year. Using a similar definition as the Dynamic Registry, DES patients in DEScover had an adjudicated stent thrombosis rate of 0.6% at 1 year. In a recent meta-analysis of 14 randomized trials comparing DES and BMS, the rate of stent thrombosis for DES after 30 days was 0.5% (12). Although stent thrombosis may be underestimated in registries, owing to less rigorous follow-up than some randomized clinical trials, the low overall rate in the Dynamic Registry is reassuring. Longer follow-up of the registry cohort is needed to address the issue of stent thrombosis occurring after 1 year (i.e., very late stent thrombosis). Although very late stent thrombosis seems to occur more frequently in DES- than in BMS-treated patients, no excess in cardiac death or MI has been demonstrated (12,13,15,27).
This study adds to our knowledge of how DES performs in complex lesion subsets. Rates of death and reinfarction were lower, but not statistically different, in patients that had 1 or more complex lesions treated with DES, including bifurcation, ostial, thrombotic, calcified, and total occlusion lesions. As expected, however, event rates were higher in both DES and BMS patients with 1 or more complex lesion type compared with those with no complex lesion type attempted. After adjustment for baseline characteristics, no appreciable differences were detected in the combined end point of death/MI by stent type according to the presence or absence of a complex lesion. Additionally, only 1 of the 3 late stent thrombosis events occurred in a complex lesion type. In terms of efficacy, target vessel revascularization was lower in DES patients compared with BMS patients independent of lesion complexity. These findings are reassuring and support the use of DES in lesions where the device is considered “untested,” including ostial, total occlusion, and bifurcation lesions.
Several limitations of the present study require comment. Stent selection was not randomized, and physician bias in choosing stent type cannot be assessed. We therefore compared DES patients to those treated with BMS before availability of DES. The DES patients, however, may be more “selected” than the BMS patients. Unmeasured confounding variables may also exist. Routine angiographic follow-up was not a component of the study design. As such, we were unable to determine rates of in-stent restenosis or target lesion revascularization. Evaluation of target vessel revascularization may underestimate the relative benefit of DES for preventing in-stent restenosis due to nonlesion disease progression; however, we did observe a significant reduction in this end point with DES. There were also lesions in both BMS and DES cohorts that were treated with balloon angioplasty alone, and some DES patients also received BMS, but the rates were low, and these lesions were excluded from the target vessel revascularization analysis.
We do not have data on the duration of dual antiplatelet therapy in all patients, and we did not assess whether use was continuous. In wave 4, however, those on clopidogrel at 1-year had more extensive coronary artery disease and prior PCI at baseline and were more likely to undergo repeat PCI during follow-up. We observed that cumulative rates of MI were similar to non-clopidogrel users but cannot comment on mortality and clopidogrel use. Lastly, we cannot comment on long-term outcomes, because the current analysis is limited to 1-year follow-up.
In summary, the present study provides additional information regarding the safety and effectiveness of DES. For at least up to 1 year, use of DES in standard clinical practice was not associated with any excess risk of death or MI compared with BMS, even in complex lesion subsets. Furthermore, the durability of the initial percutaneous revascularization was enhanced with DES.
For factors included in the adjusted model, please see the online version of this article.
Unrestricted Use of Drug Eluting Stents Compared With Bare Metal Stents in Routine Clinical Practice: Findings From the National Heart, Lung, and Blood Institute Dynamic Registry
Online Cardiosource Slide Set
The Dynamic Registry is supported by grant HL033292-21 from the National Heart, Lung, and Blood Institute of the National Institutes of Health. See accompanying online Cardiosource Slide Set.
- Abbreviations and Acronyms
- bare metal stent(s)
- drug-eluting stent(s)
- hazard ratio
- major adverse cardiac events
- myocardial infarction
- percutaneous coronary intervention
- Received February 16, 2007.
- Revision received June 27, 2007.
- Accepted July 2, 2007.
- American College of Cardiology Foundation
- Stone G.W.,
- Ellis S.G.,
- Cox D.A.,
- et al.
- Schampaert E.,
- Cohen E.A.,
- Schluter M.,
- et al.
- Vermeersch P.,
- Agostoni P.,
- Verheye S.,
- et al.
- Sabate M.,
- Jimenez-Quevedo P.,
- Angiolillo D.J.,
- et al.
- Suttorp M.J.,
- Laarman G.J.,
- Rahel B.M.,
- et al.
- Nordmann A.J.,
- Briel M.,
- Bucher H.C.
- Mauri L.,
- Hsieh W.,
- Massaro J.M.,
- Ho K.,
- D’Agostino R.,
- Cutlip D.E.
- Pfisterer M.E.,
- Brunner-La Rocca H.P.,
- Buser P.T.,
- et al.
- Kuchulakanti P.K.,
- Chu W.W.,
- Torguson R.,
- et al.
- Williams D.O.,
- Abbott J.D.,
- Kip K.E.,
- DEScover Investigators
- Holmes D.R. Jr..,
- Leon M.B.,
- Moses J.W.,
- et al.
- Lemos P.A.,
- Serruys P.W.,
- van Domburg R.T.,
- et al.
- Ong A.T.,
- van Domburg R.T.,
- Aoki J.,
- Sonnenschein K.,
- Lemos P.A.,
- Serruys P.W.