Phenotypes and outcomes in patients with atrial fibrillation: a global unsupervised cluster analysis from Phase II/III of the GLORIA-AF registry

Abstract

Abstract Background Atrial fibrillation (AF) patients are characterized by a great clinical heterogeneity and complexity across the world. Usual classifications may not adequately characterize this population. Data-driven cluster analysis reveals different possible classifications of patients. Purpose To identify different clusters of AF patients who share similar clinical phenotypes and evaluate the association between identified clusters and clinical outcomes using cluster analysis. Methods An agglomerative hierarchical cluster analysis was performed in patients from phase II/III of the GLORIA-AF registry. Associations between clusters and a composite outcome composed of all-cause death, stroke and major bleeding were evaluated using Cox regression analyses. Results The study included 26,045 AF patients (mean±SD age 70.1±10; 44.8% female) with a median [IQR] follow-up of 3 years [2.3-3.1]. Five clusters were identified: Cluster 1 was composed of young males from North America with morbid obesity and cardiovascular (CV) risk factors; Cluster 2 included older Asian patients with lower body mass index (BMI), fewer CV risk factors, lower anticoagulation rate but the highest AF ablation rate; Cluster 3 identified European males with CV comorbidities and behavioral disorders (smoking, alcohol abuse); Cluster 4 was characterized by patients with the highest rate of permanent AF, cancer, anticoagulant therapy prescription ; Cluster 5 included older women from Europe with high CHA2DS2-VASc and HAS-BLED scores, the highest heart rate, lower BMI and high rate of previous stroke and transient ischemic attack. The incidence of the composite outcome increased a stepwise fashion among clusters from 3.2% patient-years (95% confidence interval (CI): 2.4-4.1) for Cluster 1 to 6% patient-years (95% CI: 5.6-6.4) for Cluster 5 (p<0.0001): Cluster 1 versus 2 hazard ratio (HR): 0.85 (95% CI 0.65–1.11); 2 versus 3 HR: 0.84 (95% CI 0.76–0.93); 3 versus 4 HR: 0.9 (95% CI 0.82–0.99) and 4 versus 5 HR: 0.82 (95% CI 0.74–0.91). Conclusion Cluster analysis identified five statistically-driven groups of AF patients, with distinct phenotype characteristics and associated with different risks for major clinical adverse events.Baseline characteristics.Dendrogram and Kaplan Meier-curves.