Author + information
- Received March 1, 2013
- Revision received June 3, 2013
- Accepted July 2, 2013
- Published online December 24, 2013.
- Vasim Farooq, MBChB∗,
- Patrick W. Serruys, MD, PhD∗∗ (, )
- Yaojun Zhang, MD∗,
- Michael Mack, MD†,
- Elisabeth Ståhle, MD‡,
- David R. Holmes, MD§,
- Ted Feldman, MD‖,
- Marie-Claude Morice, MD¶,
- Antonio Colombo, MD#,
- Christos V. Bourantas, MD∗,
- Ton de Vries, MSc∗∗,
- Marie-angèle Morel, BSc∗∗,
- Keith D. Dawkins, MD††,
- Arie Pieter Kappetein, MD, PhD‡‡ and
- Friedrich W. Mohr, MD§§
- ∗Department of Interventional Cardiology, Erasmus University Medical Centre, Thoraxcenter, Rotterdam, the Netherlands
- †Medical City Dallas Hospital, Dallas, Texas
- ‡University Hospital Uppsala, Uppsala, Sweden
- §Mayo Clinic, Rochester, Minnesota
- ‖Evanston Hospital, Evanston, Illinois
- ¶Institut Jacques Cartier, Massy, France
- #San Raffaele Scientific Institute, Milan, Italy
- ∗∗Cardialysis BV, Rotterdam, the Netherlands
- ††Boston Scientific Corporation, Natick, Massachusetts
- ‡‡Department of Cardiothoracic Surgery, Erasmus University Medical Centre, Thoraxcenter, Rotterdam, the Netherlands
- §§Herzzentrum, Leipzig, Germany
- ↵∗Reprint requests and correspondence:
Dr. Patrick W. Serruys, Interventional Cardiology Department, Erasmus Medical Center, 's-Gravendijkwal 230, 3015 CE Rotterdam, Zuid Holland 3015 GD, the Netherlands.
Objectives The aim of this study was to report the short-term and long-term clinical impact of stent thrombosis (ST) and graft occlusion (GO) in the final 5-year outcomes of the SYNTAX (SYNergy Between PCI With TAXUS and Cardiac Surgery) trial.
Background The clinical effect of newer-generation drug-eluting stents and operative factors in complex coronary artery disease is uncertain.
Methods The incidence of 5-year ST and GO, and their association with clinical outcomes, were analyzed in the randomized percutaneous coronary intervention and coronary artery bypass graft cohorts. ST and GO were defined by the SYNTAX protocol definitions (clinical presentation with acute coronary syndrome and angiographic/pathological evidence), the Academic Research Consortium (ARC) definition for ST, and the newly devised “ARC-like” definition of GO (i.e., definite, probable, or possible GO).
Results At 5 years, 871 of 903 patients (96.5%) in the percutaneous coronary intervention cohort and 805 of 897 patients (89.7%) in the coronary artery bypass graft cohort completed follow-up. As compared with other vessel locations, protocol ST (72 lesions) occurred more frequently in the left main (14 of 72; 19%) and proximal coronary vasculature (37 of 72; 51%) and protocol GO (41 lesions) with grafts anastomosed to the distal right coronary artery (17 of 41; 42%). The incidence of 5-year ARC definite ST and ARC-like definite GO did not significantly differ (7% [n = 48] vs. 6% [n = 32], log rank p = 0.34); landmark analyses indicated significantly increased ARC definite ST within 30 days (3% [n = 19] vs. 1% [n = 6], log rank p = 0.033) but not >30 days to 5 years (4.2% [n = 29] vs. 4.5% [n = 26], log rank p = 0.78). At presentation, ARC definite ST (n = 48) and ARC-like definite GO (n = 32) were adjudicated to be linked to 4 (8%) and 0 deaths, respectively. At 5 years, ARC definite ST (n = 48) and ARC definite/probable ST (n = 75) were associated with 17 (17 of 48, 35.4%; median days to death: 0 days; interquartile range: 0 to 16 days; maximum: 321 days) and 31 (31 of 75, 41.3%; median: 0 days; interquartile range: 0 to 9 days; maximum: 721 days) cardiac deaths, respectively. At 5 years, ARC-like definite GO (n = 32) and ARC-like definite/probable GO (n = 53) were associated with 0 and 12 (12 of 52, 23.1%; median: 0 days; interquartile range: 0 to 14 days; maximum: 257 days) cardiac deaths, respectively.
Conclusions Although the incidence of ST and GO was similar at 5 years, the clinical impact of ST appeared greater, with a negative impact on short-term to long-term mortality.
The final 5-year reporting of the SYNTAX (SYNergy Between PCI With TAXUS and Cardiac Surgery) trial showed that coronary artery bypass graft (CABG) surgery remained the standard of care for patients with complex (left main or 3-vessel) coronary artery disease compared with percutaneous coronary intervention (PCI) with first-generation drug-eluting stents (1,2). In subjects with lesser complex coronary artery disease undergoing PCI, using the anatomic SYNTAX Score to assess coronary artery complexity (3) or the SYNTAX Score augmented with clinical factors (SYNTAX Score II), (4,5) were shown to be an acceptable alternative to CABG.
The clinical impact of stent thrombosis (ST) and graft occlusion (GO) in the SYNTAX trial is unreported. The purpose of the study was to report the incidence, timing, predictors, and clinical impact of ST and GO in the SYNTAX trial.
The SYNTAX trial is a randomized, prospective, multicenter, “all-comers” trial investigating subjects with unprotected left main coronary artery (ULMCA) disease (isolated or associated with 1-, 2-, or 3-vessel disease) or de novo 3-vessel disease and has previously been described (1,2). In total, 1,800 subjects were randomized on a 1:1 basis to either PCI with Taxus Express paclitaxel-eluting stents (Boston Scientific Corporation, Natick, Massachusetts) or CABG.
An independent clinical events committee (CEC), including cardiologists, cardiac surgeons, and a neurologist, reviewed all the primary clinical endpoints and ST/GO events. All deaths were subdivided into cardiovascular and noncardiovascular deaths by the CEC. Cardiac deaths were classified as related or unrelated to a cardiac procedure by the CEC.
ST and GO
Because the SYNTAX trial began before publication of the Academic Research Consortium (ARC) definition, (6) the SYNTAX trial instigated a protocol definition for ST and GO (Table 1). Subsequently, the ARC criteria (6) for ST were implemented and reported by a separate CEC (see the Acknowledgments section). To allow comparisons of ARC ST events to GO, “ARC-like” definitions were formulated for GO (i.e., definite, probable, or possible GO) using adapted ARC definitions (Table 1). ARC-like definitions for GO were retrospectively assessed in conjunction with the safety reporting data obtained by the CEC.
Continuous variables are expressed as mean ± SD and binary variables as counts and/or percentages. Time-to-event variables are presented as Kaplan-Meier (KM) estimates and compared using the log-rank test. Corrected KM curves were constructed by removing the ST adjudicated clinical event from the KM curves (7). Multivariable analyses (Cox regression) were conducted using the baseline characteristics (Online Appendix) and the forced enter method (entry criteria 0.05, no exit criteria). There was no departure from the proportionality of hazards assumption using the global proportional hazards test based on Schoenfeld residuals (8). Variables were screened for correlation before entry into the multivariable model, and none were sufficiently correlated to warrant removal. A 2-sided p value <0.05 was considered significant for all tests. All analyses were conducted using SPSS version 20.0 (SPSS Inc., Chicago, Illinois) and SAS system software version 9.2 (SAS Institute, Cary, North Carolina).
Figure 1 shows the study design and frequency of ST and GO using the SYNTAX protocol and ARC/ARC-like definitions. At 5 years, 871 of 903 patients (96.5%) in the PCI cohort and 805 of 897 patients (89.7%) in the CABG cohort completed follow-up.
Use of aspirin therapy was significantly lower after CABG compared with PCI post-procedurally (88.5% vs. 96.3%, p < 0.001) and at 1 year (84.3% vs. 91.2%, p < 0.001) but not at 5 years (85.0% vs. 87.1%, p = 0.24). Use of thienopyridine therapy was significantly lower for CABG post-procedurally (19.4% vs. 96.7%, p < 0.001) and at 1 year (15.0% vs. 71.1%, p < 0.001) and 5 years (12.1% vs. 32.0%, p < 0.001). Dual antiplatelet use at 5 years was significantly higher with PCI compared to CABG (9.1% vs. 27.4%, p < 0.001).
Anatomic location of ST and GO at 5 years of follow-up
Protocol ST was confirmed in 47 subjects with 72 lesions (Fig. 2). Forty-three of 47 subjects had documented protocol ST confirmed by angiography (62 lesions), 3 subjects had documented protocol ST confirmed by autopsy (7 lesions), and this was not recorded in one subject (3 lesions). Protocol ST occurred more frequently in the left main (14 of 72; 19.4%) and proximal (37 of 72; 51.4%) coronary vasculature.
Protocol GO was confirmed in 32 subjects with 41 lesions (Fig. 2). All 32 subjects had documented protocol GO confirmed by angiography (41 lesions). Protocol GO occurred more frequently with grafts anastomosed to the distal right coronary artery (17 of 41; 42%) compared with other vessel locations.
ARC/ARC-like definitions of ST and GO: incidence and timing
The incidence and timing of ARC/ARC-like ST and ARC/ARC-like GO are shown in Figures 3 and 4. At 5 years (KM estimates), there was no significant difference between ARC definite ST and ARC-like definite GO (ARC ST: 6.8% [n = 48]; ARC-like GO: 5.5% [n = 32]; p = 0.34), ARC definite/probable ST and ARC-like definite/probable GO (p = 0.48), and ARC definite/probable/possible ST and ARC-like definite/ probable/possible GO (p = 0.69).
Landmark analyses (Fig. 4) indicated significantly more ARC definite ST within 30 days (ARC definite ST: 2.7% [n = 19]; ARC-like definite GO: 1.0% [n = 6]; p = 0.033) but not >30 days to 5 years (ARC definite ST: 4.2% [n = 29]; ARC-like definite GO: 4.5% [n = 26]; p = 0.78).
Comparisons of ARC definite/probable ST and ARC-like definite/probable GO showed no significant differences at intervals of 0 to 30 days (p = 0.38), 30 days to 5 years (p = 0.12), and 5 years (p = 0.48). Comparisons of ARC definite/probable/ possible ST and ARC-like definite/probable/possible GO showed no significant differences at 30 days to 5 years (p = 0.33) and 5 years (p = 0.69).
ARC/ARC-like definitions of ST and GO: adjudicated clinical outcomes
ARC definite ST (n = 48) was adjudicated by the CEC to be linked to 4 deaths (4 of 48; 8.3%), 11 myocardial infarctions (MIs) (11 of 48; 22.9%), 23 cases of all-cause revascularization (23 of 48; 47.9%), and 10 cases with no clinical event (10 of 48; 20.8%). Corrected KM curves with the clinical events, adjudicated to be associated with the ARC definition of ST by the CEC, removed are shown (Fig. 5). Notably, removal of ARC definite/probable ST-related events would have led to a reduction of 5.1% in cardiac death/MI/all-cause revascularization and a 1.5% reduction in cardiac death at 5 years.
ARC-like definite GO (n = 32) was adjudicated by the CEC to be linked to 0 deaths, 7 MIs (7 of 32; 21.9%), 21 cases of all-cause revascularization (21 of 32; 65.6%), and 4 cases with no clinical event (4 of 32; 12.5%).
Long-term cardiac mortality associated with the index ARC/ARC-like ST/GO events
The 5-year relationship of cardiac mortality to the index ARC ST or ARC-like GO events is shown (Table 2). At 5 years, ARC definite ST was associated with 17 cardiac deaths (17 of 48; 35.4%) (median number of days to cardiac death: 0; interquartile range: 0 to 16 days; maximum: 321 days) and ARC definite/probable ST (n = 75) with 31 cardiac deaths (31 of 75; 41.3%) (median number of days to death: 0; interquartile range: 0 to 9 days; maximum: 721 days).
Comparatively, at 5 years, ARC-like definite GO was not associated with any deaths and ARC-like definite/probable GO (n = 53) with 12 cardiac deaths (12 of 52; 23.1%) (median number of days to death: 0; interquartile range: 0 to 14 days; maximum: 257 days).
Multivariable predictors of the ARC definition of ST
Table 3 shows multivariable analyses of ARC definite/probable ST. Model 1 incorporated components of the SYNTAX Score and clinical- and procedural-related variables. Model 2 incorporated the SYNTAX Score and EuroSCORE. Stent length (in millimeters) and number of stents implanted were not univariable predictors of early or late/very late ST.
Early ARC ST (Within 30 Days)
The presence of ULMCA disease was not a univariable predictor of early ARC ST (p = 0.84). Lack of post-procedural antiplatelet therapy was the strongest independent predictor of early ST (p < 0.001), followed by peripheral vascular disease (p = 0.054), heavy calcification (p = 0.054), and prior MI (p = 0.037). Both the SYNTAX Score (p = 0.086) and additive EuroSCORE (p = 0.006) were predictors of early ST.
Late/Very Late ARC ST (Beyond 30 Days)
The presence of ULMCA disease showed a trend toward being protective against late/very late ARC ST on univariable analyses (hazard ratio: 0.61; 95% confidence interval: 0.33 to 1.12; p = 0.10). The SYNTAX Score was not a univariable predictor of late/very late ARC ST (p = 0.71). Any baseline angiographically visible thrombus (p = 0.003) and trifurcation lesion (p = 0.048) were the strongest independent predictors.
In total, 23 of 903 subjects (2.5%) in the PCI cohort were reported to have a baseline angiographically visible thrombus, of which 7 of 23 subjects had ARC ST beyond 30 days (ARC definite ST: n = 3; ARC possible ST: n = 3; ARC probable ST: n = 1). Sixty-six of 903 subjects (7.3%) in the PCI cohort were reported to have trifurcation lesions, of which 12 of these 66 subjects had ARC ST (ARC definite ST: n = 5; ARC probable ST: n = 5; ARC possible ST: n = 2). Three of the 12 cases of ARC ST occurred in <30 days (2 in subjects with ULMCA disease), and 7 occurred in >30 days (6 in subjects with ULMCA disease).
Multivariable predictors of ARC-like definite GO
In the CABG cohort, double left internal mammary artery (LIMA)/right internal mammary artery (RIMA) and complete arterial revascularization were reported to be used in 236 of 897 subjects (26.3%) and 161 of 897 subjects (17.9%), respectively. On univariable analyses, double LIMA/RIMA (hazard ratio: 0.86; 95% confidence interval: 0.35 to 2.10; p = 0.74), complete arterial revascularization (hazard ratio: 0.81; 95% confidence interval: 0.31 to 2.11; p = 0.67), and ULMCA disease (p = 0.66) were not statistically associated with a reduction in ARC-like definite GO. Any baseline angiographically visible thrombus (n = 31) (p = 0.41) was not statistically associated with a reduction in ARC-like definite GO.
Multivariable analyses indicated that other arterial/venous conduits (i.e., no LIMA) to the left anterior descending artery (LAD) (p = 0.050) and the number of grafts (p = 0.017) were independent predictors of ARC-like definite GO (Table 3). Of subjects with ARC-like definite GO (n = 32), 8 had no LIMA to the LAD and 25 were not on pre-procedural aspirin therapy.
In this post-hoc study of the randomized, all-comers SYNTAX trial, the main findings were as follows. First, although the incidence of ST and GO at 5 years was similar using the ARC/ARC-like definitions, the clinical impact of ST appeared greater, with a negative impact on short-term to longer-term mortality. Second, independent predictors of ST and GO were diverse and related to anatomic or procedural- or clinical-related factors.
The findings of ST in the SYNTAX trial are notable in that early (<30 days) and late/very late ARC definite ST (>30 days) were associated with short-term and long-term mortality. It is important to emphasize that the actual causes of death are multifactorial and not easily directly attributed to ST. For example, although lack of post-procedural antiplatelet therapy was the strongest independent predictor of ST (Table 3), a recent substudy of the SYNTAX trial showed that this was of multifactorial origin (9). Factors precluding antiplatelet administration were reported to include gastrointestinal bleeding, retroperitoneal bleeding from procedure-related femoral vascular access, coronary perforation, or following surgical bailout for PCI-related complications. Consequently, directly attributing ST to mortality is difficult, although it is clear there was a strong causal link.
In addition, ST most frequently occurred in the left main and proximal vasculature (>70% of cases). Accordingly, the impact of ST on mortality is not only related to the acute consequences of abrupt closure of the treated vessel but also the short-term to longer-term complications of MI, including heart failure, arrhythmias, or mechanical complications (10). Indeed, previously reported analyses of PCI-treated patients who had post-procedural creatine kinase levels >2 times the upper limit of normal (with corroborating elevation of creatine kinase MB fraction) in the SYNTAX trial were shown to be associated with short-term (<3 months) and longer-term (2 to 4 years) mortality (11). Even selecting a time window (e.g., 1 day, 1 week, or 1 month) associating mortality with ST is arbitrary because, as shown in this study, the interquartile range of cardiac death after ARC definite ST ranged from 0 to 16 days, with a maximum duration of 321 days, and cardiac death was often of multifactorial origin. Consequently, determining the potential impact on mortality of newer-generation drug-eluting stents with proven reductions in ST (12,13) is difficult to determine in the SYNTAX trial. Comparatively, potential reductions in ST-related MI and all-cause revascularization were easier to determine, as highlighted in Figure 5.
The present study confirms the established supremacy of the long-term durability of LIMA to the LAD (14). Notably, double LIMA/RIMA was not associated with a reduced risk of GO on univariable analyses, despite being undertaken in more than one-fourth (27.6%) of randomized patients undergoing CABG. It is important to emphasize that the data related to double LIMA/RIMA were underpowered to draw any firm conclusions and are in contrast to the data in published reports (14,15). Ten-year follow-up data from a large ongoing randomized trial are awaited (16).
The finding of the number of grafts to be an independent predictor of GO is consistent with reported studies in that too extensive surgical revascularization has been associated with the occurrence of major perioperative complications and acute MI (17,18). Conversely, the 15-month angiographic substudy of the SYNTAX trial (19) showed that more than one-fourth (27.2%) of subjects undergoing CABG had at least one significantly diseased (≥50% to <100%) or occluded bypass graft at 15 months. Notably, the reported angiographic loss of the bypass grafts was not significantly associated with early clinical events (19), in contrast to GO in the current study (defined in the context of clinical presentation with acute coronary syndrome). Reasons to account for the “clinically silent” loss of bypass grafts have included that they were anastomosed to functionally insignificant lesions, with resultant competitive filling to the native vessel, and were therefore unnecessary (20).
The present study represents a post-hoc analysis, and the results should be considered hypothesis generating. Because ST and GO were defined in the context of clinical presentation, the true occurrence is likely to be higher, particularly if the ST occurs in a distal vessel (19). In addition, saphenous vein grafts have been shown to be protective in the first 7 years; thereafter, mortality and repeat revascularization have been reported to increase significantly, secondary to a gradual loss of graft patency (21). Given that operative factors were linked to ARC-like probable GO, ARC-like definite GO was used for the multivariable analyses to identify long-term predictors of GO. As exists with ARC possible ST, ARC-like possible GO is likely to have a low specificity in identifying GO. The exact anatomic segment number of the ST using the ARC definite criteria is not available, because this was not recorded by the CEC. There was limited statistical power for further subanalyses in patients with ST or GO. Although multivariable adjustments were performed for significant confounders (p < 0.1), the possibility of other unmeasured confounders cannot be excluded. The possibility of overfitting in the multivariable analyses cannot be excluded. Because the confidence intervals of these correlates were relatively narrow, the risk would have been limited (22). Given the unavoidable differences in follow-up between the CABG and PCI cohorts, we cannot exclude the possibility that this may have affected the results.
Although the incidence of ST and GO was similar at 5 years, the clinical impact of ST appeared greater, with a negative impact on short-term to longer-term mortality.
The authors thank the study centers whose work made this study possible; Kristine Roy, Jian Huang, and Vicki Houle of Boston Scientific Corporation for their invaluable technical support in accessing the study database; and the members of the separate CEC: Dr. E. W. L. Jansen (Universitair Medisch Centrum Utrecht, Utrecht, the Netherlands), Dr. C. Hanet (Clinique Universitaire de Saint-Luc, Brussels, Belgium), the late Prof. Dr. O. M. Hess (Swiss Cardiovascular Center, University Hospital, Bern, Switzerland), and his replacement, Dr. E. P. McFadden (Cork University Hospital, Cork, Ireland).
For an expanded Methods section as well as supplemental tables and figures, please see the online version of this article.
The SYNTAX trial was funded by Boston Scientific. Dr. Mack has served on the Speakers’ Bureau of Boston Scientific, Cordis, and Medtronic. Dr. Feldman has served on the Speakers’ Bureau of Boston Scientific; received grant support from Abbott Laboratories, Atritech, Boston Scientific, Edwards, and Evalve; and served as a consultant for Abbott Laboratories, Boston Scientific, Coherex, Edwards, InterValve, Square One, and W. L. Gore and Associates. Dr. Morice’s institution has received a research grant from Boston Scientific. Dr. Dawkins is a full-time employee and stockholder of Boston Scientific. All other authors have reported that they have no relationships relevant to the content of this paper to disclose.
- Abbreviations and Acronyms
- Academic Research Consortium
- coronary artery bypass graft
- clinical events committee
- graft occlusion
- left anterior descending artery
- left internal mammary artery
- myocardial infarction
- percutaneous coronary intervention
- right internal mammary artery
- stent thrombosis
- unprotected left main coronary artery
- Received March 1, 2013.
- Revision received June 3, 2013.
- Accepted July 2, 2013.
- American College of Cardiology Foundation
- Mohr F.W.,
- Morice M.C.,
- Kappetein A.P.,
- et al.
- Farooq V.,
- van Klaveren D.,
- Steyerberg E.W.,
- et al.
- Farooq V.,
- van Klaveren D.,
- Steyerberg E.W.,
- Serruys P.W.
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.
- Serruys P.W.,
- Onuma Y.,
- Garg S.,
- et al.
- Schoenfeld D.
- Farooq V.,
- Serruys P.W.,
- Bourantas C.,
- et al.
- Windecker S.,
- Meier B.
- Farooq V.,
- Serruys P.W.,
- Vranckx P.,
- et al.
- Dorman M.J.,
- Kurlansky P.A.,
- Traad E.A.,
- Galbut D.L.,
- Zucker M.,
- Ebra G.
- Taggart D.P.,
- Altman D.G.,
- Gray A.M.,
- et al.
- Rastan A.J.,
- Walther T.,
- Falk V.,
- et al.
- Alamanni F.,
- Dainese L.,
- Naliato M.,
- et al.
- Morice M.C.,
- Feldman T.E.,
- Mack M.J.,
- et al.
- van Domburg R.T.,
- Kappetein A.P.,
- Bogers A.J.