Aims: Obstructive sleep apnoea (OSA) associates with atrial fibrillation (AF), but the relationship of OSA severity and AF phenotype with the atrial substrate remains poorly defined. We sought to define the atrial substrate across the spectrum of OSA severity utilizing high-density mapping.
Methods and results: Sixty-six consecutive patients (male 71%, age 61 ± 9) having AF ablation (paroxysmal AF 36, persistent AF 30) were recruited. All patents underwent formal overnight polysomnography and high-density left atrial (LA) mapping (mean 2351 ± 1244 points) in paced rhythm. Apnoea-hypopnoea index (AHI) (mean 21 ± 18) associated with lower voltage (-0.34, P = 0.005), increased complex points (r = 0.43, P < 0.001), more low-voltage areas (r = 0.42, P < 0.001), and greater voltage heterogeneity (r = 0.39, P = 0.001), and persisted after multivariable adjustment. Atrial conduction heterogeneity (r = 0.24, P = 0.025) but not conduction velocity (r = -0.09, P = 0.50) associated with AHI. Patchy regions of low voltage that co-localized with slowed conduction defined the atrial substrate in paroxysmal AF, while a diffuse atrial substrate predominated in persistent AF. The association of AHI with remodelling was most apparent among paroxysmal AF [LA voltage: paroxysmal AF -0.015 (-0.025, -0.005), P = 0.004 vs. persistent AF -0.006 (-0.017, 0.005), P = 0.30]. Furthermore, in paroxysmal AF an AHI ≥ 30 defined a threshold at which atrial remodelling became most evident (nil-mild vs. moderate vs. severe: 1.92 ± 0.42 mV vs. 1.84 ± 0.28 mV vs. 1.34 ± 0.41 mV, P = 0.006). In contrast, significant remodelling was observed across all OSA categories in persistent AF (1.67 ± 0.55 mV vs. 1.50 ± 0.66 mV vs. 1.55 ± 0.67 mV, P = 0.82).
Conclusion: High-density mapping observed that OSA associates with marked atrial remodelling, predominantly among paroxysmal AF cohorts with severe OSA. This may facilitate the identification of AF patients that stand to derive the greatest benefit from OSA management.