Abstract / Summary
Abstract Background: Persistent atrial fibrillation (AF) is phenotypically heterogeneous. Existing phenotyping approaches rely predominantly on costly imaging or biomarker panels, limiting routine clinical applicability. Pulse rate deficit (PRD), an easily measurable clinical sign, is associated with exercise intolerance and impaired hemodynamics. Whether PRD can classify AF phenotypes and predict the response to catheter ablation remains unknown.
Methods: Consecutive patients with persistent AF undergoing first catheter ablation were prospectively enrolled at a single center; 103 formed the final clustering cohort. PRD was measured as |heart rate - pulse rate| (bpm) by a single investigator using sequential cardiac auscultation (60 s) and radial artery palpation (60 s), with measurement order randomized per patient. Pearson correlation analysis assessed the clinical correlates of PRD. Unsupervised K-means clustering (K=5) was applied to 7 standardized features (PRD, heart rate, pulse rate, LVEF, NT-proBNP, eGFR, age; random_state=42). Clustering stability was evaluated across 10 random seeds using the adjusted Rand index (ARI) and reclassification rate.
Results: Of 108 enrolled patients, 103 (mean age 63.2 ± 9.8 years, 22.3% female) formed the final clustering cohort. PRD correlated significantly with heart rate (r=+0.570, P<0.001), LVEF (r=-0.382, P<0.001), heart failure (r=+0.342, P<0.001) and WBC (r=+0.400, P<0.001). Five clusters were identified: C0 (markedly elevated PRD, preserved LVEF, n=16), C1 (low PRD, young, n=45), C2 (extreme phenotype: youngest, highest PRD, lowest LVEF, 100% heart failure, n=4), C3 (oldest, low PRD, n=29) and C4 (highest NT-proBNP, n=9). The silhouette score at K=5 was 0.196. C2 remained stable in 9/10 seeds (mean ARI=0.724, mean reclassification rate=11.4%). In exploratory analyses, recurrence rates did not differ significantly across clusters (P=0.386).
Conclusions: PRD is a valid, bedside-available phenotypic marker that identifies distinct AF subgroups, including a reproducible extreme subtype (C2: young, high PRD, reduced LVEF, 100% heart failure). Its primary value lies in cross-sectional phenotyping rather than prognosis prediction. Larger studies are needed to validate these phenotypes and their prognostic implications.