Phenotype-guided risk stratification in anticoagulated atrial fibrillation patients: Cluster analysis of the Murcia AF Project III (MAFP-III)

Abstract

BACKGROUND: Atrial fibrillation (AF) is characterised by clinical heterogeneity owing to diverse risk factors and comorbidities. OBJECTIVE: We aimed to identify phenotypic AF clusters and their association with outcomes. METHODS: We prospectively included patients with anticoagulated AF between January 2016 and November 2021. Phenotypic clusters were identified using hierarchical clustering. Outcomes included thromboembolic events, major bleeding, major adverse cardiovascular events (MACE), cardiovascular and all-cause death, over a 2-year follow-up. RESULTS: Overall, 3259 patients with AF (median age 77; interquartile range [IQR] 70-83 years; 52.8% women) were included. Five phenotypic clusters were identified: Cluster 1: patients ?75 years with few comorbidities; Cluster 2: male patients ?75 years with high-risk lifestyle and metabolic profile; Cluster 3: patients >75 years with stroke/transient ischaemic attack/thromboembolism history; Cluster 4: patients >75 years with cancer history; and Cluster 5: patients >75 years with multimorbidity. Cluster 5 consistently exhibited the highest relative risk across all outcomes. Compared with Cluster 1, adjusted Cox models showed significantly increased risks for: thromboembolic events (adjusted hazard ratio [aHR] range 2.37-3.31; highest in Cluster 5: aHR 3.31 [95% confidence interval [CI] 1.96-5.57]); major bleeding (aHR range 3.34-4.73; highest in Cluster 5: aHR 4.73 [95% CI 2.51-8.91]); MACE (aHR range 2.64-4.13; highest in Cluster 5: aHR 4.13 [95% CI 2.62-6.51]); cardiovascular death (aHR range 4.29-6.82; highest in Cluster 5: aHR 6.82 [95% CI 3.05-15.27]); and all-cause death (aHR range 2.47-4.18; highest in Cluster 5: aHR 4.18 [95% CI 2.77-6.31]). CONCLUSION: Cluster analysis revealed distinct phenotypic profiles among patients with AF, each associated with differential risks of adverse outcomes.