Coronary revascularisation by coronary artery bypass graft (CABG) surgery is associated with better long-term survival rates in patients with severe coronary artery disease (CAD) compared with an initial strategy of anti-anginal medical therapy alone.1 Since its inception in 1968,2 CABG has proved to be an effective tool for the treatment of symptomatic CAD, conferring a degree of survival advantage directly proportional to the severity of CAD.3 In addition, CABG is more beneficial in patients with extensive severe CAD, such as multivessel disease with involvement of either the proximal left anterior descending or the left main coronary artery (LMCA).3 Indeed, current guidelines recommend CABG as the standard of care for patients with severe CAD;4 however, improvements in the field of interventional cardiology, particularly following the introduction of drug-eluting stents (DES), have made it feasible to utilise percutaneous coronary intervention (PCI) in more complex coronary lesions, despite a lack of evidence to support its suitability for patients with severe CAD.
The SYNTAX Trial
The Synergy Between Percutaneous Coronary Intervention with Taxus and Cardiac Surgery (SYNTAX) trial was a large prospective, randomised trial in which the investigators sought to establish the optimal revascularisation strategy (CABG versus PCI with DES implantation) in patients with de novo three-vessel and/or LM CAD in the presence of either symptomatic or asymptomatic myocardial ischaemia.5 The clinical relevance of this trial was enhanced by its robust design as a multicentre, international study (85 centres and 17 countries) with the use of contemporary tools in terms of DES (paclitaxel-eluting stent, Taxus, Boston Scientific, Natick, MA, US), PCI techniques and the use of arterial grafts where possible in the surgical arm. Additionally, a novel score was developed (the SYNTAX score) to prospectively characterise disease severity within the coronary vasculature in terms of lesion frequency, location and angiographic complexity.6
Patients eligible for randomisation were those for whom a consensus opinion that both therapeutic options could attain equivalent revascularisation was reached by an interventional cardiologist and a cardiothoracic surgeon. Subjects were prospectively stratified by diabetic status and the presence or absence of LM CAD. Of 4,337 patients enrolled in the study, 1,262 (29%) were ineligible for randomisation and were excluded, while only 1,800 patients (41%) were deemed suitable for entry; the latter group was randomly assigned to undergo PCI with paclitaxel-eluting stent (PES) implantation or CABG.
Patients deemed suitable for only one therapeutic option and not eligible for randomisation either because of the presence of complex anatomy precluding PCI (1,077 patients) or as a result of co-morbidities precluding open heart surgery (198 patients) were enrolled into a parallel nested registry. Given that this cohort of patients is frequently seen in clinical practice, this approach made it possible to include a high percentage of screened patients (71%) into either the randomisation or registry arms of the trial (see Figure 1).
Aspirin was prescribed indefinitely for all patients who underwent randomisation. In most centres, thienopyridines were continued for more than six months following PCI, with 71.1% of patients receiving dual-antiplatelet therapy (DAT) at 12 months, while patients who underwent CABG received significantly less pharmacological treatment (only 15% of patients were on DAT at 12 months; p<0.001).
The pre-defined primary clinical end-point was non-inferiority of PCI compared with CABG in the incidence of major adverse cardiac or cerebrovascular events (MACCE, i.e. death from any cause, stroke, myocardial infarction [MI] or repeat revascularisation) at 12 months. Notably, a non-inferiority margin of 6.6% was considered to be clinically significant for the trial, and was derived from historical data (weighted average of 5% for three-vessel disease and 9% for LM CAD), resulting in the failure of the study to reach its primary end-point.
Attempts have subsequently been made to describe the differences in a number of subgroups within the SYNTAX trial cohort, particularly in terms of the LM CAD subgroup. However, despite the fact that patients were stratified according to the presence of unprotected LM CAD, the study was not statistically powered to detect differences between the treatment options in the pre-specified subgroups and therefore the results in this subset of lesions can only be interpreted as hypothesis-generating. Moreover, as the primary study end-point was not met, no definitive conclusions can be drawn from any subgroup analysis. Nevertheless, the results in the LM CAD subset are of particular interest and will be tested in larger future studies that are sufficiently powered to investigate outcomes in this group.
Results at 12 Months in the Overall Population
The incidence of MACCE was significantly higher among PCI-treated patients compared with patients randomised to CABG (17.8 versus 12.4%; p=0.002). PCI with PES implantation failed to meet statistical non-inferiority criteria for the primary study end-point; this result was primarily driven by the higher rates of repeat revascularisation following PCI compared with CABG (13.5 versus 5.9%; p<0.0001).
However, when repeat procedures were excluded from the combined end-point, the rate of all-cause mortality, cerebrovascular events (CVE) and MI was almost identical between the two groups (7.7% PCI group versus 7.6% CABG group; p=0.98); indeed, the incidence of CVE was higher in CABG-treated patients (0.6% PCI group versus 2.2% CABG group; p=0.003). Interestingly, the incidence of stent thrombosis (ST) in the PCI group was similar to the symptomatic graft occlusion rate in the CABG arm (3.4 versus 3.3%; p=0.89).
Subgroup Analysis at 12 Months – Left Main Coronary Artery Disease
The 12-month MACCE rate for patients in the LM CAD subgroup (n=357 in the PCI group, n=348 in the CABG group) was similar in both the PCI and CABG groups (15.8 versus 13.7%; p=0.44). Although the rate of repeat revascularisation among patients with LM CAD was significantly higher in the PCI group than in the CABG group (11.8 versus 6.5%; p=0.02), the absolute difference of 5.3% was small. Furthermore, the limited benefit in terms of revascularisation rates in the CABG arm came at a cost of a nine-fold increase in rate of stroke (0.3% PCI subgroup versus 2.7% CABG subgroup; p=0.01).7 A total of 36.6% of the patients with LM CAD also had concomitant three-vessel disease. A post hoc analysis of the subgroups of patients with LM CA revealed a higher rate of MACCE among those patients with concurrent two- or three-vessel disease compared with the patients with isolated LMCA lesions or associated disease in only one other vessel.
Subgroup Analysis at 12 Months – Diabetic Status
An exploratory subgroup analysis was also pre-specified per study protocol for patients with medically treated diabetes, which was defined as treatment with oral hypoglycaemic agents or insulin at the time of enrolment, in accordance with prior studies.8,9 Of the 1,800 patients allocated in the SYNTAX trial, 452 (n=231 in the PCI subgroup, n=221 in the CABG subgroup) had medically treated diabetes. The overall one-year MACCE rate in diabetics was higher among patients treated with PCI compared with CABG (26 versus 14.2%, relative risk [RR] 1.83, 95% confidence interval [CI] 1.22–2.73; p=0.03), mainly driven by the higher repeat revascularisation rates in the PCI arm (20.3% PCI group versus 6.4% CABG group; p=0.001). Furthermore, the incidence of repeat revascularisation was significantly higher in diabetic than in non-diabetic patients after PCI treatment (p=0.001), but this difference was not evident following CABG (p=0.74).10
Notably, the number needed to treat by CABG in order to avoid one MACCE event in diabetic patients is nine, while this number is 31 for non-diabetic patients. However, no statistically significant differences were found between the CABG and PCI groups in the composite safety end-point of death/CVE/MI for either diabetic (10.1% PCI group versus 10.3% CABG group; p=0.96) or non-diabetic patients (6.8% PCI group versus 6.8% CABG group; p=0.97).10
Results at 24 Months in the Overall Population
Two-year outcome results were presented by Dr AP Kappetein as a late-breaking clinical trial at the European Society of Cardiology Congress 2009.11 In the analysis, MACCE (analysed in a time-to-event manner) was significantly increased in PCI patients (23.4% PCI group versus 16.3% CABG group; p=0.0002); however, the composite safety end-point of death/CVE/MI was similar between the two groups (10.8% PCI group versus 9.6% CABG group; p=0.44). The increase in MACCE at two years was mainly due to the rate of repeat revascularisation in PCI-treated patients (17.4% PCI group versus 8.6% CABG group; p<0.0001), although it is notable that the majority of repeat revascularisations were required within the first year (see Table 1).
The incidence of CVE, significantly higher in the CABG group in the first year (0.6% PCI group versus 2.2% CABG group; p=0.003), appeared to have evened out during the second year (0.7% PCI group versus 0.6% CABG group); however, the cumulative stroke rate remained significantly higher in CABG patients (1.4% PCI group versus 2.8% CABG group; p=0.03). Interestingly, the MI rate (which was previously equal between the two groups at one year) was significantly increased in PCI patients (5.9% PCI group versus 3.3% CABG group; p=0.01) (see Table 2).
Subgroup Analysis at 24 Months – Left Main Coronary Artery Disease
These results were presented by Dr MC Morice at the Trans-Catheter Cardiovascular Therapeutics Conference 2009.12 The rate of MACCE among patients in the LM CAD subgroup was similar between the two study groups (22.9% PCI and 19.3% CABG; p=0.27). No difference was observed in all-cause mortality (5.6% PCI group versus 6.2% CABG group; p=0.77) or MI (4.1% PCI versus 5.5% CABG; p=0.45). A higher rate of CVE was seen in the CABG cohort (0.9% PCI group versus 3.7% CABG group; p=0.01), while an increased rate of re-intervention was observed in the PCI group (17.3% PCI group versus 10.4% CABG group; p=0.01).
These results strongly suggest that PCI is a safe alternative to CABG for patients with LM CAD, particularly in patients with a SYNTAX score <33 (i.e. two-thirds of the patients with LM CAD randomised within the trial). In addition, the two-year rates of MACCE were similar among the two groups in the subgroups of patients with low and intermediate SYNTAX score with no significant statistical difference, while these rates were higher in the PCI group in a specific analysis of the subgroup of patients with LM CAD and high SYNTAX score (>33: 29.7% PCI group versus 17.8% CABG group; p=0.02).
The SYNTAX Score
The SYNTAX score provided insights into the optimum selection of patients with three-vessel disease or LM CAD who can be managed by PCI. The score was designed to reflect the complexity of CAD based mainly on anatomical characteristics present on the angiogram, with higher scores indicating more complex disease.6 Unlike previous scoring systems, the SYNTAX score is applied to the complete coronary artery tree, and in the SYNTAX trial it was calculated and agreed a priori by the heart team (consisting of a cardiac surgeon and an interventional cardiologist) before the revascularisation procedure.5 It is important to remember that the SYNTAX score only reflects the coronary anatomy and not the treatment strategy.
For the purpose of analysis, the investigators divided the enrolled patients into three groups in terms of their SYNTAX score (low ≤22, intermediate 23–32 and high ≥33). At one-year follow-up, PCI outcomes were reflected in the SYNTAX score.5 In patients with low scores (0–22), MACCE rates were similar between the two groups (13.6% PCI group versus 14.7% CABG group; p=0.71) and in the mid-score tertile (23-32) MACCE rates were not statistically different between treatment strategies (16.7% PCI group versus 12.0% CABG group; p=0.10), whereas in the most complex patients with the highest score (>33), the outcome in the CABG group was superior to that in the PCI group (23.4% PCI group versus 10.9% CABG group; p<0.001).
Importantly, the impact of lesion complexity (reflected by the SYNTAX score) on clinical outcomes was consistent at two-year follow-up.11 The two-year rates of MACCE were not significantly different between patients with low SYNTAX scores treated with either PCI or CABG (19.4 versus 17.4%; p=0.63).
In patients with intermediate SYNTAX scores, there was a trend towards increased MACCE with PCI (22.8% PCI group versus 16.4% CABG group; p=0.06). In the most complex patients (SYNTAX scores >33), MACCE was significantly increased in those treated with PCI (28.2% PCI group versus 15.4% CABG group; p=0.0001).11 Thus, not only was the SYNTAX score a useful descriptive tool, it was also predictive of outcomes at one and two years following PCI in patients with LM CAD and/or three-vessel disease and/or diabetes in the SYNTAX trial.5,10–12 By contrast, an association between the SYNTAX score and clinical outcome was not observed in the CABG arm, in that patients with a high SYNTAX score did just as well as patients with a low score. This is not surprising as surgical morbidity and mortality are determined to a greater extent by co-morbid factors (i.e. chronic obstructive pulmonary disease, renal impairment, prior stroke, etc.) than by the coronary anatomy per se. In surgical patients, risk is more reliably reflected in a scoring system that includes concurrent disease such as the European System for Cardiac Operative Risk Evaluation (EuroSCORE)13 or the Parsonnet Score.14
Although the SYNTAX score could be used as a stand-alone tool for the pre-assessment and selection of strategy in patients with CAD, its performance may be improved by the parallel use of clinical scores, such as EuroSCORE. In this setting, a recent study15 demonstrated that merging the angiographic and clinical information contained in the SYNTAX score and the EuroSCORE allowed the development of a combined risk model (Global Risk Classification), with a significant improvement in the prediction of cardiac mortality in patients undergoing PCI for unprotected LM CAD.15 To this end, an incorporation of clinical risk and co-morbid factors into the existing estimation system of the SYNTAX score may refine its prognostic ability and may reduce the gap between the art and science of decision-making when choosing the most appropriate coronary revascularisation strategy for patients with CAD. This development in the use of the SYNTAX score promises to ensure that it continues to have an important role in future clinical practice.
Interventional cardiology and cardiac surgery are dynamic fields in continuous evolution with ongoing modern innovation and technical development. Since the advent of DES, interventional practice has considerably changed and has revolutionised the ability of the interventional cardiologist to approach more complex and varied lesions and patient subsets.
The population enrolled in the SYNTAX trial comprised the most complex cohort of patients with de novo, three-vessel or LM CAD ever studied in such a large clinical trial. More than 20% of patients had total occlusions and almost 75% had bifurcation lesions. An average of 4.6 DES were implanted in the PCI arm and at least one arterial conduit was grafted in 97.3% of the CABG group. The mean stent length was 86mm and one out of three patients had a stent length ≥100mm. Despite these factors, the differences in repeat revascularisation found in the SYNTAX trial between PCI with PES implantation compared with CABG were far smaller than those reported in other large, randomised trials in a less complicated coronary anatomical setting,16,17 which in itself represents a remarkable finding. Furthermore, the findings suggest that 14 bypass procedures are needed in order to prevent one repeat PCI, at a cost of four times the number of CVE.
The SYNTAX trial data support the benefit of DES in reducing the rate of restenosis and the need for repeat revascularisation after PCI. Nevertheless, stents remain imperfect solutions that are unable to completely eliminate restenosis in complex settings, such as bifurcation lesions and multivessel disease. Additionally, it must be noted that the SYNTAX trial was sponsored by Boston Scientific Corporation and, as a result, patients randomised to the PCI arm received only Taxus Express stents. Although the Taxus stent remains the market leader, it has been superseded by newer stent platforms with superior drug elution. Since the SYNTAX trial, additional DES have been approved in the US (ENDEAVOR Resolute, Medtronic Vascular, Santa Rosa, CA, US; XIENCE V, Abbott Vascular Santa Clara, CA, US) and in Europe (Promus Element, Boston Scientific, Natick, Mass; Nobori, Terumo Corp, Tokyo, Japan; Biomatrix Biosensors Interventional Technologies Pte Ltd, Singapore).18–20 The safety and efficacy of the everolimus-eluting stent (EES) platform (Xience V, Abbott Vascular, Santa Clara, CA, US) compared with the Taxus Express was demonstrated in A Randomized Comparison of a Durable Polymer Everolimus-Eluting Coronary Stent with a Bare Metal Coronary Stent (SPIRIT) First, SPIRIT II and SPIRIT III randomised trials.21–23 Both the recent Xience V versus Taxus Express (SPIRIT IV) trial24 and Xience V versus Taxus Liberte (COMPARE) trial25 demonstrated the superiority of Xience V compared with Taxus in the reduction of the primary end-point of one-year MACE (SPIRIT IV: 4.2% Xience group versus 6.9% Taxus group; p=0.0009; COMPARE: 6.2% Xience group versus 9.1% Taxus group; p=0.02). It may be that the use of second-generation DES led to more favourable outcomes in the PCI group compared with CABG.
It is also important to remember that PCI techniques are constantly improving and perhaps more advanced methods with the judicious use of intravascular ultrasound (IVUS) guidance and standardised post-dilatation with adequately sized balloons could have optimised stent deployment, leading to a reduction in the incidence of tricuspid valve regurgitation (TVR) in the PCI group of the SYNTAX trial.
Perhaps there is now a need for an adapted version of the SYNTAX trial in the modern era building on the encouraging results observed in the LM CAD subgroup and encompassing the technical developments in PCI (e.g. the use of ‘second-generation’ DES, increased use of IVUS, fractional flow reserve, new imaging techniques such as optical coherence tomography, assessment of clopidogrel responsiveness and the administration of more effective drugs such as prasugrel and ticagrelor) and CABG (e.g. higher percentage of the off-pump no-touch technique). Indeed, participants are being sought for the Evaluation of Xience Prime versus Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization (EXCEL) trial, a landmark trial randomising 2,500 patients with LM CAD and a SYNTAX score ≤32 to CABG versus second-generation DES (XIENCE Prime, Abbott Vascular Santa Clara, CA, US). The primary end-point of this study is the composite incidence of death, MI or stroke at a median follow-up of three years, powered for sequential non-inferiority and superiority testing. We await the results and hope that they will further enhance our clinical strategic decision-making.
In addition to the promise of better drugs and improved stent platforms and polymers, future developments in PCI with the introduction of bioabsorbable stents may further prevent restenosis by removing the potential endothelial irritation caused by the presence of metallic stents and thus promoting improved vascular healing.26 Given that the field of PCI continues to evolve and improve, the results of the SYNTAX trial are unlikely to be the final word on the best possible treatment for severe CAD.