Abstract 9396: Bayesian Methods Affirm the Use of Percutaneous Coronary Intervention to Improve Survival in Patients with Unprotected Left Main Coronary Artery Disease

Introduction: Several randomized clinical trials (RCTs) have supported the Class I recommendation for coronary artery bypass grafting (CABG) for unprotected left main coronary artery disease (ULMCAD). In the absence of RCTs directly comparing percutaneous coronary intervention (PCI) with medical therapy (MT) for this indication, the new Class IIa recommendation for PCI to improve survival in selected patients with ULMCAD has been statistically inferred from several studies comparing CABG with PCI. Hypothesis: We tested whether Bayesian approaches supported the recommendation for PCI to improve survival in patients with ULMCAD. Methods: We performed a Bayesian cross-design and network meta-analysis of 12 studies (4 RCTs, 4 observational matched studies and 4 other cohort studies) comparing CABG with PCI (N=4,574 patients) and of 7 studies (2 RCTs and 5 observational studies) comparing CABG with MT (N=3,224 patients). Results: The odds ratios (ORs) for 1-year mortality after PCI versus CABG using Bayesian cross-design meta-analysis were not different among RCTs (OR 0.99, 95% Bayesian credible interval [BCI] 0.67-1.43), matched cohort studies (OR 1.10, 95% BCI 0.76-1.73), and other types of cohort studies (OR 0.93, 95% BCI 0.58-1.35). A network meta-analysis suggested that MT alone is associated with higher 1-year mortality than the use of PCI for patients with ULMCAD (Figure). Conclusions: Bayesian methods support the current revascularization guidelines and suggest that PCI may improve survival over MT alone for patients with ULMCAD. An integrated approach analyzing data from both observational and randomized studies using Bayesian methods may yield new insights to enhance the translation of trial data into clinical practice.