Author + information
- Received August 4, 2009
- Revision received October 6, 2009
- Accepted October 8, 2009
- Published online December 1, 2009.
- Stephen G. Ellis, MD⁎,⁎ (, )
- Gregg W. Stone, MD†,
- David A. Cox, MD‡,
- James Hermiller, MD§,
- Charles O'Shaughnessy, MD∥,
- Tift Mann, MD¶,
- Mark Turco, MD#,
- Ronald Caputo, MD⁎⁎,
- Patrick J. Bergin, MD††,
- Thomas S. Bowman, MD, MPH‡‡,
- Donald S. Baim, MD‡‡,
- TAXUS IV Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Stephen G. Ellis, Cleveland Clinic, 9500 Euclid Avenue, Department of Cardiology/F25, Cleveland, Ohio 44195
Objectives The pivotal TAXUS IV (TAXUS IV-SR: Treatment of De Novo Coronary Disease Using a Single Paclitaxel-Eluting Stent) trial evaluated the long-term safety and effectiveness of the paclitaxel-eluting stent (PES) compared with an otherwise identical bare-metal stent (BMS) in a relatively uncomplicated population of patients with a single de novo lesion in a native coronary vessel, treated between March and July 2002.
Background Long-term follow-up is required to determine whether the early safety and efficacy of drug-eluting stents are maintained.
Methods The primary end point of this prospective, randomized, double-blind trial was 9-month ischemia-driven target vessel revascularization (TVR) for PES versus the BMS control. Follow-up was complete in 1,230 (95.1%) of 1,294 randomized evaluable patients at 5 years.
Results Compared with BMS, PES significantly reduced TVR at 9 months (12.1% vs. 4.7%; p < 0.0001); this benefit was maintained through 5 years (27.4% vs. 16.9%; p < 0.0001), given comparable TVR rates for BMS and PES between years 1 and 5 (4.1%/year vs. 3.3%/year; respectively, p = 0.16). Similar patterns were observed for composite major adverse cardiac events (MACE) (32.8% BMS vs. 24.0% PES, p = 0.0001 at 5 years). Stent thrombosis was comparable for PES and BMS at 9 months (0.8% BMS vs. 0.8% PES; p = 0.98) and at 5 years (2.1% BMS vs. 2.2% PES, p = 0.87). The overall revascularization benefits of PES were consistent across multiple subgroups, including sex, diabetes, left anterior descending artery lesion location, reference vessel diameter, lesion length, and multiple stents.
Conclusions These 5-year results demonstrate the long-term safety and sustained efficacy of PES compared with BMS in patients with noncomplex lesions. (TAXUS IV-SR: Treatment of De Novo Coronary Disease Using a Single Paclitaxel-Eluting Stent; NCT00292474)
The TAXUS IV (TAXUS IV-SR: Treatment of De Novo Coronary Disease Using a Single Paclitaxel-Eluting Stent) prospective randomized clinical trial compared the Taxus Express2 (Boston Scientific Corp., Natick, Massachusetts) paclitaxel-eluting stent (PES) to a visually indistinguishable bare-metal stent (BMS) in 1,326 randomized patients with noncomplex de novo lesions in a single native coronary artery, representing the pivotal trial for U.S. approval. The primary end point analysis found that PES significantly reduced the rate of ischemia-driven target vessel revascularization (TVR) at 9 months compared with BMS, with a comparable safety profile (1). These favorable early results, however, do not address concerns regarding subsequent potential safety risks (e.g., increased mortality or stent thrombosis [ST]) with drug-eluting stents over longer-term follow-up (2–4).
The current study presents the final (5-year) follow-up assessment of safety and efficacy data for the TAXUS IV study. Although subsequent trials have evaluated the Taxus Express stent in more complex anatomy (TAXUS V) (5), in “real world” use (ARRIVE [TAXUS Peri-Approval Registry: A Multi-center Safety Surveillance] 1 and 2) (6), in left main and/or 3-vessel disease (SYNTAX [SYNergy Between PCI With TAXUS and Cardiac Surgery]) (7), in acute myocardial infarction (HORIZONS-AMI [Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial Infarction]) (8), as well as the same drug and polymer on other stent geometries (Taxus Liberte–ATLAS , and Taxus Element–PERSEUS), the final 5-year follow-up of TAXUS IV represents the longest and most complete study of this drug-eluting stent that has been implanted in more than 5 million patients worldwide.
The TAXUS IV was a prospective, randomized, double-blind, controlled trial designed to evaluate the efficacy and safety of the TAXUS Express PES by 1:1 randomization against an otherwise identical BMS. Detailed methods for this study have been previously published (1) and are briefly summarized here.
Eligible patients were ≥18 years with stable or unstable angina or provokable ischemia due to a single de novo lesion 10 to 28 mm in length and ≥50% stenosis in a native coronary artery of diameter 2.5 to 3.75 mm that could be covered by a single stent. A total of 1,326 patients were enrolled in the trial, including 154 patients with lesions >24 mm who received a 32-mm stent when it became available in the latter part of the study. This study was conducted in accordance with the principles delineated by the Declaration of Helsinki and all local regulations. The study protocol was approved by the institutional review board at each site before patient enrollment. All patients provided written informed consent.
Randomization and study treatment
Eligible patients were randomized in a double-blind fashion, stratified by the presence or absence of medically treated diabetes (patients being treated with insulin and/or oral hypoglycemic medications) and vessel size (<3.0 mm vs. ≥3.0 mm), to either a slow-release PES (TAXUS Express, Boston Scientific Corp.) or an otherwise identical uncoated BMS (Express, Boston Scientific). Patients were pretreated with aspirin 325 mg and clopidogrel 300 mg before catheterization and were treated after the procedure with aspirin 325 mg daily indefinitely and clopidogrel 75 mg daily for at least 6 months per the TAXUS Directions For Use (10). Clinical follow-up was scheduled for 1, 4, and 9 months and yearly thereafter for a total of 5 years.
The primary end point of the trial was the incidence of ischemia-driven TVR at 9 months. Secondary end points included target lesion revascularization (TLR; stented segment ±5 mm margins); non-target lesion target vessel revascularization (non-TL TVR), myocardial infarction (MI); death; major adverse cardiac event (MACE), defined as a composite of cardiac death, MI, and TVR; and ST. This trial was conducted before the publication of the Academic Research Consortium (ARC) consensus definitions of ST (11), but all potential ST events were subsequently retrospectively adjudicated by a clinical events committee masked to treatment allocation according to both the per-protocol (1) and ARC definitions. The per-protocol definition of ST was nearly identical to the ARC definite plus probable definition and included angiographic (but not pathologic) confirmation of ST presenting in concordance with acute coronary syndrome, acute MI in the distribution of the treated vessel, or unexplained death within 30 days after procedure (1). The timing of ST was defined as acute (0 to 24 h), subacute (24 h to 30 days), late (31 to 365 days), or very late (>365 days). In a subset of patients prospectively assigned to receive 9-month angiography, quantitative coronary angiography was used to measure absolute lesion length, reference vessel diameter (RVD), minimal lumen diameter, and percent diameter stenosis.
The statistical analysis plan prespecified an intent-to-treat analysis for all end points through the first 2 years of follow-up. Patients who did not receive a study stent (e.g., because of inability to cross the lesion with the stent due to calcification, tortuosity, and so forth) were not followed beyond 2 years, per protocol; therefore, all long-term follow-up analyses were performed on the safety population (i.e., those who received either a PES or BMS). Binary or categorical variables are presented as percentages and were compared with the chi-square test or Fisher exact test. Continuous variables are presented as mean ± SD and were compared with Student t test.
The MACE and its components were evaluated with descriptive statistics, Kaplan-Meier estimates, and the log-rank test. Annualized hazard rates were also calculated separately for MACE and its components with the person-time method and expressed as the event rate/100 patient-years for the time periods 0 to 1 year and between 1 and 5 years (12). In this analysis, patients with multiple events occurring in both time periods were considered to have the event in each time period. Baseline predictors of outcomes at 5 years were evaluated with logistic regression with Wald's chi-square test and included all baseline and clinical variables, treatment assignments, and procedural variables. The results are expressed as odds ratios with 95% confidence intervals. All p values are 2-sided. All statistical analyses were performed with SAS System software, version 8.2 or above (SAS Institute, Inc., Cary, North Carolina).
Patients and antiplatelet therapy
A total of 1,326 patients were initially enrolled in this trial (Fig. 1). Twelve patients were deregistered, and 20 patients did not receive a study stent; these 32 patients were not followed beyond 2 years, per the protocol design. Deregistered patients did not have a study stent placement attempted; reasons for deregistration included appropriate study stent size not available (n = 3), inability to pass lesion with balloon (n = 3), pre-stenting complications (n = 3), inclusion/exclusion criteria violation (n = 2), and patient withdrew consent (n = 1). Reasons for nonplacement of study stents included inability of the stent to cross the lesion (n = 12), lesion characteristics did not allow for placement of a study stent (e.g., vessel diameter increased to 4.0 mm after pre-dilation) (n = 4), nondeployed stent due to vasospasm caused by guidewire (n = 1), and patients found not to meet inclusion criteria after randomization (n = 3). Therefore, the predefined safety population used to analyze long-term results consisted of 1,294 patients (651 PES and 643 BMS). Clinical follow-up at 5 years was 95.1% (1,230 of 1,294) complete and did not differ significantly between groups (Fig. 1).
Among the safety population, baseline patient and lesion characteristics were similar between those who received a PES and those who received a BMS (Table 1). Immediately after procedure, nearly all patients were treated with dual antiplatelet therapy, and this was continued through at least the first 4 months of the study (Fig. 2). At 9 months after procedure, >95% were taking aspirin and approximately 40% of patients in each group were still receiving dual antiplatelet therapy. Given the double-blind nature of the study, there was no significant difference in the percentage of patients receiving dual antiplatelet therapy between groups at any point during study follow-up, although rates of thienopyridine therapy at 4 months and of aspirin therapy at 3 years were briefly different. At 5 years, >90% of patients in both groups were taking aspirin, but dual antiplatelet therapy had fallen to 32% of the BMS patients and 31% of the PES patients.
At the 9-month primary end point time point, the use of PES compared with BMS significantly decreased the rate of ischemia-driven TVR (p < 0.001) (1). This difference was then maintained throughout follow-up, with 16.9% of PES patients versus 27.4% of BMS patients having undergone TVR by 5 years (hazard ratio: 0.56, 95% confidence interval: 0.44 to 0.71; p < 0.0001) (Fig. 3A). Similar results were observed for TLR, with a significant benefit for PES at 9 months, followed by a low and relatively stable annualized hazard rate of 2.0%/year for BMS and 1.5%/year for PES between years 1 and 5 (Fig. 3B, Table 2). The annual TVR rates were slightly higher than those of TLR between years 1 and 5 (4.1%/year for BMS and 3.3%/year for PES), due to similar ongoing disease progression in the remainder of the stented vessel (non-TL TVR) for both PES and BMS (Fig. 3C, Table 2) (13). The Kaplan-Meier curves for the safety end points are nearly superimposable for PES and BMS, with no significant difference for either cardiac death (p = 0.85) or MI (p = 0.87) (Figs. 3D and 3E). Therefore, the sustained first-year reductions in TLR and TVR translated into an ongoing significant (p < 0.0001) reduction in the composite MACE end point through 5 years for PES treatment (Fig. 3F, Table 2), with no evidence of late “catch-up” in adverse events by the PES group.
The 5-year cumulative incidence of all-cause death was 11.2% for BMS and 10.0% for PES (Kaplan-Meier estimates; log-rank p value = 0.49), corresponding to an annualized mortality of approximately 2%/year for both PES and BMS. At 9 months, total ARC ST (definite/probable) was 0.8% in both groups (p = 0.98). The observed rates of ARC ST (either definite/probable or definite only) were not statistically significantly different between BMS and PES at any point during follow-up (Figs. 3G and 3H). Acute ARC definite/probable ST (binary rates) occurred in 0.3% (2 of 643) of BMS patients versus 0% (0 of 651) of PES patients (p = 0.25); subacute ST occurred in 0.5% (3 of 641) of BMS patients and 0.3% (2 of 650) of PES patients (p = 0.68); late ST in 0.5% (3 of 639) of BMS patients and 0.5% (3 of 648) of PES patients (p > 0.99); and very late ARC definite/probable ST (VLST) occurred in 1.1% (7 of 625) of BMS versus 1.4% (9 of 632) of PES patients (p = 0.63). The relative risk for VLST (ARC definite/probable) in PES versus BMS was 1.27 (95% confidence interval: 0.48 to 3.39; p = 0.63), corresponding to an annualized hazard rate for VLST of 0.3%/year for BMS and of 0.4%/year for PES (p = 0.45) between years 1 and 5. No PES-related ARC definite/probable ST occurred after day 1,334; the last ST in the BMS arm occurred on day 1,835.
The decreased risk of TLR at 5 years for PES compared with BMS was consistent across multiple subgroups, including sex, diabetes, left anterior descending artery lesion location, baseline RVD, lesion length, and single versus multiple stents, with comparable and significant reductions in TLR for each of the subgroups (Fig. 4).
Effect of routine angiographic follow-up on clinical outcomes
In this study, 56% (722 of 1,294) of patients were assigned to receive routine angiographic follow-up at 9 months. There were no significant differences for either the PES or BMS groups in terms of MACE or ARC ST according to the performance of routine angiographic follow-up (Fig. 5). However, as previously reported at 1 year (14), the performance of routine angiographic follow-up did lead to comparable trends for increased TLR in both groups at 5 years (PES: 10.7% angiographic vs. 6.9% nonangiographic; p = 0.09; BMS: 23.4% angiographic vs. 16.9% nonangiographic; p = 0.07; multivariate p value for interaction = 0.77).
Baseline factors associated with MACE and TVR through 5 years
Multivariable analysis showed that PES use compared with BMS use was associated with a significant reduction in MACE over 5 years of follow-up (p = 0.002) (Table 3). The risk of MACE strongly correlated with the number of stents implanted (p = 0.03), medically treated diabetes (p = 0.001), and smaller baseline RVD (p = 0.005). Table 3 also displays the independent correlates of MACE at 5 years in the BMS and PES groups separately. In the overall population, treatment with PES, chronic heart failure, and acute gain were strongly and significantly correlated with decreased TVR through 5 years, whereas a family history of coronary artery disease and diabetes requiring medication were significantly associated with increased TVR at 5 years (Table 4).Table 4 also displays the independent correlates of TVR at 5 years in the BMS and PES groups separately.
Patients with medically treated diabetes mellitus
A total of 163 patients in the BMS group and 152 patients in the PES group had medically treated diabetes. Higher rates of MACE were noted in diabetic compared with non-diabetic patients, in both the PES and BMS groups. Patients with diabetes tended to have higher 5-year all-cause mortality for both BMS (13.1% diabetic vs. 10.5% nondiabetic; p = 0.31) and PES (12.7% diabetic vs. 9.1% nondiabetic; p = 0.18), with similar rates of MI and ARC definite/probable ST (Fig. 6). Patients treated with PES had a comparable and sustained approximately 50% reduction in TLR at 5 years compared with BMS, in both diabetic (27% BMS vs. 13% PES; p = 0.002) and nondiabetic patients (18% BMS vs. 8% PES; p < 0.0001). Although diabetic patients treated with PES had significantly lower rates of TLR compared with BMS, PES- and BMS-treated patients exhibited similar increases in the absolute rates of non-TL TVR, reflecting more aggressive CAD beyond the stent and its margins (Fig. 7). In the PES group, non-TL TVRs represented one-half of all TVRs at 5-year follow-up, compared with only 35% of the 5-year TVRs in the BMS group.
In the pivotal TAXUS IV clinical trial, the use of PES provided a significant reduction in repeat revascularization of both the target lesion and target vessel compared with BMS, which was evident at the 9-month primary time point and was sustained throughout 5-year follow-up. There were comparable, low, and constant annualized hazard rates for new TLR and TVR events in both PES and BMS patients between years 1 and 5, with no evidence of late “catch-up” in patients who received PES.
Although this study enrolled relatively low-risk patients (e.g., no recent MI, 34% enzyme-negative acute coronary syndrome), comprising approximately 35% of the current stent-treated population (6), and was underpowered to detect differences in low-frequency events between treatment groups, the careful long-term follow-up does allow for some important safety conclusions. Critically, safety profiles—including rates of death, MI, and ARC ST—remained low and comparable between the PES and BMS groups throughout the 5-year follow-up. Results from this trial are generally similar to those from the recently reported 5-year follow-up of the pivotal randomized trial comparing the sirolimus-eluting stent with a BMS control (15). Despite earlier concerns about the incidence of VLST (events occurring after 1 year) with drug-eluting stents compared with BMS (4,16,17), and acknowledging previously noted limitations in statistical power, this randomized, double-blind study did not identify significant differences in ST rates between PES and BMS at any time point (acute, subacute, late, or VLST). Over 5 years, the cumulative incidence of ARC definite/probable ST was 2.2% for PES versus 2.1% for BMS (p = 0.87). Annualized hazard rates for VLST were 0.4%/year for PES compared with 0.3%/year for BMS (p = 0.45) between years 1 and 5 (7 of 625 BMS patients, and 9 of 632 PES patients). A pooled analysis of 8 randomized, controlled trials comparing either PES or sirolimus-eluting stent with BMS also found similar cumulative rates of ARC definite or probable VLST of 0.9% for sirolimus-eluting stent versus 0.4% for BMS among 1,748 patients and 0.9% for PES versus 0.6% for BMS among 2,797 patients, at 4 years of follow-up (16). Of course, even these findings do not exclude a small potential increased risk of VLST for drug-eluting stents compared with BMS, but no such difference between PES and BMS was observed in the relatively noncomplex TAXUS IV patient population. Long-term follow-up from additional studies are required to verify the late safety profile of PES in more complex patient cohorts.
Several baseline factors were significantly associated with an increased risk of MACE at 5 years for the overall population, including smaller baseline RVD, medically treated diabetes, the total number of stents implanted, and use of BMS rather than PES. In particular, the presence of diabetes and greater number of stents implanted can be considered as surrogate markers for more diffuse and rapidly progressive coronary artery disease, which might be expected to translate into an increased annual rate of late adverse events (13). A pooled analysis of the TAXUS trials has demonstrated that approximately one-half of all late revascularizations after 1 year occur in nonstented segments of the target vessels in both PES and BMS, highlighting the importance of aggressive risk-factor control in reducing this ongoing manifestation of progressive atherosclerosis (13).
Routine angiographic follow-up has previously been shown to increase the rate of revascularization compared with clinically driven follow-up. In this study, routine angiographic follow-up produced significant increases in repeat revascularization at 1-year follow-up in both the PES and BMS patients (14), but those differences no longer were statistically significant at 5 years, given the ongoing accumulation of non-TL TVR in both the routine angiographic and clinical follow-up only groups.
The presence of diabetes increased the rate of revascularization in both the PES and BMS groups due to increased rates of both TLR and TVR outside the stent and its 5-mm margins (non-TL TVR) (Fig. 7), indicative of the more aggressive coronary disease in diabetic patients (18,19). However, the 5-year rate of revascularization remained lower in PES-treated diabetic patients, due to a significant reduction in TLR by PES. Although total mortality was slightly but not significantly increased in patients with diabetes compared with those without diabetes, the rates of MI and ST did not differ between the PES and BMS treatment groups according to diabetic status. These results are consistent with an earlier pooled analysis of 3,513 patients treated with PES or BMS in relatively noncomplex lesions (20).
The TAXUS IV study enrolled patients with relatively simple, uncomplicated lesions; as such, the results might not be applicable to more complex real-world anatomic or clinical populations. However, 35% of the real-world ARRIVE registry population met the enrollment criteria for TAXUS IV, suggesting that the findings of TAXUS IV do apply to a substantial number of real-world patients (6). The outcomes in this “TAXUS IV-like” subpopulation of ARRIVE, moreover, were essentially superimposable with those of the non-angiographic follow-up cohort in TAXUS IV through 2 years. Finally, as previously mentioned, despite the substantial size of this pivotal randomized controlled trial and the absence of signals of increased ST, death, or MI through 5 years of follow-up for PES versus BMS, it remains underpowered to study very low-frequency events.
The 5-year results from the pivotal TAXUS IV trial demonstrate sustained efficacy of PES in a diverse group of patients with non-complex lesions, with no increase in long-term safety events compared with the otherwise identical BMS control.
The authors thank Ruth M. Starzyk, PhD, and Vicki M. Houle, PhD (Boston Scientific Corporation), for assistance in manuscript preparation and Jian Huang, MD, MS, Peggy Pereda, MS, and Hong Wang, MS (Boston Scientific Corporation), for statistical analysis.
This work was supported by Boston Scientific Corporation. Dr. Ellis is a consultant/advisory board member for Boston Scientific, Abbott, and Cordis, and has received research support from Boston Scientific and Cordis. Dr. Stone is an advisory board member for Boston Scientific and Abbott Vascular. Drs. Cox and Hermiller are consultants/advisory board members for Boston Scientific. Dr. O'Shaughnessey is a consultant/advisory board member, a Speakers' Bureau member, and has received research support from Boston Scientific. Dr. Turco is a Speakers' Bureau member/advisory board member/consultant for Boston Scientific, Medtronic, and Cordis. Drs. Bowman and Baim are full-time employees of Boston Scientific Corporation, Inc. Dr. Donald Baim passed away unexpectedly on November 6, 2009.
- Abbreviations and Acronyms
- Academic Research Consortium
- bare-metal stent(s)
- major adverse cardiac events
- non-TL TVR
- non-target lesion target vessel revascularization
- paclitaxel-eluting stent(s)
- reference vessel diameter
- stent thrombosis
- target lesion revascularization
- target vessel revascularization
- very late (Academic Research Consortium definite/probable) stent thrombosis
- Received August 4, 2009.
- Revision received October 6, 2009.
- Accepted October 8, 2009.
- American College of Cardiology Foundation
- Ellis S.G.,
- Cannon L.,
- Mann T.,
- et al.
- Turco M.A.,
- Ormiston J.A.,
- Popma J.J.,
- et al.
- ↵(2007) TAXUS Express2 Paclitaxel-Eluting Coronary Stent System (Boston Scientific Corporation, Natick, MA) [Directions for Use].
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.
- Leon M.,
- Allocco D.,
- Dawkins K.,
- Baim D.
- Pinto D.S.,
- Stone G.W.,
- Ellis S.G.,
- et al.
- Weisz G.,
- Leon M.B.,
- Holmes D.R. Jr..,
- et al.
- Jadhav S.,
- Petrie J.,
- Ferrell W.,
- Cobbe S.,
- Sattar N.
- Kirtane A.J.,
- Ellis S.G.,
- Dawkins K.,
- et al.