Six multicenter, RCTs were included after eligibility assessment (Fig. 1)[7–13]. The overall quality of evidence using the GRADE system for the primary endpoints was moderate (Supplementary table 1), mainly due to some concerns regarding risk of bias and indirectness (Supplementary Figs. 1 and 2). Median follow-up was 6.25 years (IQR: 5- 2.5). The weighted median age of the patients at the time of inclusion was 65.1 years (IQR: 63.2–66) and the weighted median perioperative risk estimated with the EuroSCORE was 2.8% (IQR: 2.7–3.8). First generation DES were employed in the SYNTAX, PRECOMBAT and FREEDOM trials, and second-generation DES in the BEST, NOBLE and EXCEL trials. The median of arterial grafts employed was 1 (IQR 1–2). Characteristics of included studies are resumed in Table 1. Different definitions of events were employed for each RCT, which are summarized in supplementary table 2.
The SYNTAX trial compared the composite of death, stroke, MI and repeat revascularization. 1800 patients were included, 903 in the PCI arm and 897 in the CABG arm; of those, 23.6% (n = 213) and 21.1% (n = 189) were women, respectively[7]. After completion of 5-year follow up, the study was continued as the SYNTAXES trial, which evaluated survival at 10-year follow-up[8].
The PRECOMBAT trial included patients with LM disease, and evaluated the composite of death, stroke, MI and repeat revascularization up to 10 years of follow-up. 300 patients were included in each arm, with 24% (n = 72) and 23% (n = 69) female patients for PCI and CABG, respectively[9].
The FREEDOM was designed to compare the composite of death, stroke and MI in patients with diabetes and multivessel coronary artery disease (CAD), and reported outcomes at 5 years of follow-up. 953 patients were included in the PCI arm and 947 in the CABG arm. 26.8% (n = 255) and 30.5% (n = 289) were women in each arm, respectively[10]. After study completion, the study continued as the FREEDOM Follow-On study, evaluating survival at 7.5 years of follow-up[14].
The BEST trial included 880 patients with multivessel CAD and compared the composite of death, MI and repeat revascularization up to 5-year follow-up. 30.6% (n = 134) were female patients in the PCI arm and 26.5 (n = 117) in the CABG arm[11].
The NOBLE and EXCEL trials included 1184 and 1905 patients with LM CAD, respectively, and reported outcomes at 5 years of follow up. The primary endpoint of NOBLE was the composite of death, stroke, MI and repeat revascularization while EXCEL evaluated a primary endpoint of death, stroke and MI. In the NOBLE trial, 20% (n = 116) and 24% (n = 140) were women in the PCI and CABG arms respectively. In the EXCEL trial 23.8% (n = 226) were women in the PCI arm and 22.5% (n = 215) in the CABG arm[12, 13].
3.1 Death, stroke, MI
Five studies reported the composite endpoint of death, stroke or MI[7, 9–11, 13]. 25.1% (n = 1779) were female and 74.9% (n = 5306) male patients. In the pooled analysis, we observed a significant benefit favoring CABG over PCI (HR = 1.24; 95% CI 1.01–1.52; p = 0.037; I2 = 0.0%) (Fig. 2). In the sensitivity analysis, the SYNTAX, FREEDOM and EXCEL studies had a significant influence in the composite outcome (Supplementary Fig. 3). In the subgroup analyses stratified by LM or multivessel RCT, no significant differences were found when evaluating only LM dedicated RCTs (Supplementary Fig. 4).
3.2 Death, stroke, MI, repeat revascularization
The composite of death, stroke, MI or repeat revascularization was reported in 4 studies[7, 9, 11, 12]. 23.5% (n = 1050) and 76.5% (n = 3414) were women and men, respectively. A significant benefit of CABG over PCI was observed in the pooled analysis (HR = 1.60; 95% CI 1.25–2.03; p < 0.000; I2 = 25.7%) (Fig. 3). In the sensitivity analysis, the estimate from the SYNTAX trial significantly influenced the composite outcome (Supplementary Fig. 5). Results were consistent in the subgroup analysis for LM RCT’s (Supplementary Fig. 6).
3.3 Individual components of the primary outcomes
Death was reported in four studies (n = 4277)[8, 9, 11, 14], 26.3% (n = 1125) were women and 73.7% (n = 3152) men. In the pooled analysis, no significant difference between PCI and CABG was observed (HR = 1.02; 95% CI 0.82–1.25; p = 0.884; I2 = 0.0%) (Fig. 4A). Three studies reported the incidence of MI, repeat revascularization and stroke (n = 3280)[7, 9, 11]. 24.2% (n = 794) were women and 75,8% (n = 2486) were men. A significant benefit of CABG over PCI was observed in the pooled analysis for MI and repeat revascularization (MI HR = 2.15; 95% CI 1.06–4.35; p = 0.034; I2 = 0.0%) (Repeat revascularization HR = 2.55; 95% CI 1.69–3.86; p < 0.000; I2 = 0.0%), respectively (Fig. 4B, 4C). Regarding stroke, we observed no significant differences between treatments in the pooled analysis (HR = 0.55; 95% CI 0.25–1.21; p = 0.137; I2 = 0.0%) (Fig. 4D).