The Arterial Reascularization Therapy Study Part I (ARTS I) and the Stent or Surgery (SoS) trials have compared bare metal stent based PCI with CABG [39,40]. In the ARTS I trial, death and myocardial infarction were similar with both treatment strategies, but, as one might expect, complete revascularization was less often achieved and repeat revascularization was more frequent with percutaneous revascularization. The results of the SoS trial were similar but, for reasons that are unclear, there was an unexpected higher mortality associated with PCI (5% vs 2%; p = 0.01), likely unrelated to treatment selection as the difference was entirely a result of non‐cardiac deaths. The long‐term results (5–6 years) of the ARTS, ERACI II, and SoS trials, and a large (7812 patients) meta‐analysis including 10 randomized trials of balloon angioplasty or stenting vs surgery with a median follow‐up of 5.9 years has been published [41–44]. In the meta‐analysis, long‐term mortality was similar after CABG and PCI in patients with multivessel CAD (hazard ratio 0.91, 95% CI 0.82–1.02; p = 0.12). CABG was associated with a significantly lower mortality in patients with diabetes (HR 0.70, 0.56–0.87; p = 0.014 for interaction) and patients aged 65 years or older (0.82, 95% CI 0.70–0.97; p = 0.002 for interaction).
DES account for vast majority of stents used for contemporary PCI around the world, but there are limited data comparing the outcomes between PCI using DES and CABG. There are three randomized clinical trials (SYNTAX, CARDia, and FREEDOM) to date [45–47]. The Synergy Between PCI with TAXUS and Cardiac Surgery (SYNTAX) trial randomized 1800 patients with three vessel and/or left main disease to either CABG or PCI after a local interventional cardiologist and cardiac surgeon at each site prospectively evaluated eligible patients and determined that equivalent anatomic revascularization could be achieved with either strategy [45]. A score of anatomic complexity (SYNTAX score: www.syntaxscore. com) was prospectively calculated. Unlike most previous trials where a small minority of patients who were screened were ultimately enrolled, almost half of the 4337 screened were randomized, with the majority of those excluded were unsuitable for PCI. The rates of death and myocardial infarction at 12 months were similar in the two groups. Stroke was significantly more frequent in the CABG group (2.2% vs 0.6%; p = 0.003) while the incidence of stent thrombosis and symptomatic graft occlusion at 12 months were similar in the two groups (3.3 and 3.4%, respectively). An increased rate of repeat revascularization in the PCI group (13.5% vs 5.9%; p < 0.001) led to an excess of major cardiac and cardiovascular events in the PCI group, with the endpoint of the trial (non‐inferiority of PCI using DES) not being met. Subgroup analysis based on the predetermined SYNTAX score showed that the negative outcome was confined to patients with high scores (>33, 10.9% in the surgical arm vs 23.4% in the PCI arm). The patients with left main disease had similar incidence of 12 month MACE in the surgical and PCI groups (13.7 vs 15.8%; p = 0.44). At 5‐year follow‐up, the composite primary endpoint (death from any cause, stroke, MI, or repeat revascularization) remained significantly higher (37.3% vs 26.9%) in the PCI group due primarily to the higher rates of repeat revascularization and myocardial infarction. The rates of all‐cause death and stroke were not different [48]. When analyzed by SYNTAX score, those with a score of <23, inclusive of left main patients, there was no difference in the composite primary endpoint (32.1% vs 28.6%; p = 0.43). In those with a SYNTAX score of 23–32 with unprotected left main disease, the outcomes were similar (32.7% vs 32.3%; p = 0.88); but not in patients with three vessel disease in whom PCI had higher event rates for the primary composite endpoint (37.9% vs 22.6%; p = 0.0008). For patients with a SYNTAX score ≥33, the rates were higher in patients with left main disease (46.5% vs 29.7%; p = 0.003) and those with three vessel disease without (41.9% vs 24.1%; p = 0.0005) unprotected left main disease. At 10 years follow‐up, CABG was associated with a survival benefit in patients with three‐vessel disease [49]. The results of the SYNTAX trial indicate that CABG remains the standard for patients with complex three vessel disease. However, in patients with less complex disease (i.e., left main coronary disease with low and possibly intermediate SYNTAX scores, or three vessel disease with low SYNTAX scores), PCI is a reasonable alternative treatment to CABG. In patients with a high SYNTAX score, the potential advantages of surgery should be stressed, but PCI should not be denied to patients who have a strong preference or a very high surgical risk. However, these conclusions must be interpreted in the context of the trial’s limitations, such as the use of a first generation (paclitaxel‐eluting) stent, which has a higher rate of restenosis than current second generation drug‐eluting stents. Also, the analyses of subgroups by SYNTAX score was not pre‐specified or adequately powered, and therefore the findings should be considered hypothesis generating.
In the Coronary Artery Revascularization in Diabetes (CARDia) trial, 510 diabetic patients with multivessel or complex single vessel coronary disease were randomized to PCI using a stent (and routine abciximab) or CABG. The trial was underpowered for the primary endpoint (composite of all‐cause mortality, myocardial infarction, and stroke), and at one year there was no difference between CABG and the 69% of patients who received a DES (12.4% and 11.6%; p = 0.82) [46]. In the Future Revascularization Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multivessel Disease (FREEDOM) trial, 1900 patients with diabetes and multivessel (83% with three vessel) disease (median SYNTAX score of 26) were randomized to either PCI with a first generation (paclitaxel or sirolimus‐eluting) stent or CABG. Both groups received optimal medical therapy [47]. At a median follow‐up of 3.8 years, the primary composite endpoint death from any cause, non‐fatal myocardial infarction, or non‐fatal stroke was more frequent in the PCI group (26.6% vs 18.7%; p = 0.005). The benefit of CABG was predominantly because of lower rates of myocardial infarction and death from any cause while stroke was more frequent in the CABG group (2.4% vs 5.2%; p = 0.03 at 5 years). The findings of FREEDOM suggest that in patients with diabetes and advanced CAD, CABG is superior to PCI using first generation stents.
Comparison of coronary artery bypass surgery with medical therapy for stable angina
The European Coronary Surgery Study (ECSS), Coronary Artery Surgery Study (CASS), and Veterans Administration Cooperative Study (VA Study) are relatively small, randomized trials that have compared CABG with medical therapy among patients with mild to moderate angina [50–52]. They were conduced over 35 years ago in an era when there was no effective medical therapy for CAD. The consistent finding from these studies was that surgical revascularization provides better symptomatic relief from angina, but the benefit is lost over time, most likely because of vein graft failure and subsequent crossover to CABG in the medical treatment arm. The randomized trials and a meta‐analysis [53] indicate that an initial strategy of surgical revascularization does not improve survival in the general population of CAD, but that there are specific subsets that either have a large amount of ischemic myocardium or significant LV dysfunction. Thus, patients with three vessel disease (especially in those with abnormal LV function), two or three vessels disease with >75% stenosis of the LAD or a markedly positive stress test derive prognostic benefit from CABG. In general, patients with severe symptoms have been excluded from the trials, but an analysis from registry data of the CASS study indicates that surgical revascularization probably improves prognosis in patients with severe angina who have multivessel disease, even in the absence of LV dysfunction or proximal LAD stenosis [54]. It is important to be aware that this evidence, which has been used to craft current guidelines, is limited by the fact that the randomized trials were all conducted in the early years of bypass surgery, and are not representative of the contemporary surgical techniques such as the routine use of internal mammary grafts or minimally invasive and off‐pump surgery. Conversely, the medical group did not benefit from the aggressive preventive measures which are now routine nor did they consistently receive beta‐blockers or angiotensin‐converting enzyme (ACE) inhibitors. Furthermore, the general applicability of these trials is limited by the fact that they did not enrol many women or patients over 65 years old.
The STICH trial has reported outcomes in stable multivessel CAD (excluding significant left main or CCS III/IV angina) in the presence of reduced left ventricular function (EF≤35%). The intial data showed no benefit of CABG compared to medical therapy alone,