Left atrial parameters predict incident atrial fibrillation and flutter in patients with heart failure with reduced ejection fraction

Abstract Background Left atrial (LA) function parameters have shown prognostic value in relation to incidence of atrial fibrillation in the general population, but this remains to be investigated in patients with heart failure with reduced ejection fraction (HFrEF). Purpose This study aimed to investigate the relationship between LA function parameters and the risk of incident atrial fibrillation/atrial flutter (AF/AFL) in patients with HFrEF. Methods Subjects free of AF/AFL were retrospectively included from a HF clinic and followed using the Danish National Patient Registry. Cox proportional hazards regression was used to assess the prognostic value of LA function parameters. These included LA ejection fraction (LAEF), peak atrial longitudinal strain (PALS), peak atrial contractile strain (PACS) and LA conduit strain (LACS). In addition, multivariable Cox regression was performed to adjust for demographic and echocardiographic parameters, including LA volume index (LAVi). The LA function parameters that remained significant in adjusted analysis were examined using cubic splines, and cutoff values corresponding to a hazard ratio (HR) equal to 1 were derived from these. Results The total study population consisted of 435 subjects with a left ventricular ejection fraction (LVEF) ≤45% (mean age 65.6±11.7 years, male sex 70.1%) free from AF/AFL at baseline. During a median follow-up of 11.4 years [IQR: 10.8, 14.2], 111 (25.5%) developed AF/AFL. Subjects who met the outcome were generally older (69.0±8.8 years vs. 64.4±12.4, p<0.001), had lower baseline heart rate, and a higher prevalence of moderate or severe valve disease (12.6% vs. 3.1%, p<0.001). They also had higher LV mass index and LV internal diastolic diameter index, but similar LVEF. These subjects had larger LAVi (30 ml/m2 [IQR: 24, 38] vs. 26 ml/m2 [IQR: 20, 32], p<0.001), while LAEF (33±14.8% vs. 39±14.9%, p<0.001), PALS (20.3% [IQR: 16.2, 26.5] vs. 23.2% [IQR: 18.6, 29.6], p=0.001), and PACS (9.7% [IQR: 6.9, 13.5] vs. 12.0% [IQR: 7.8, 16.4], p=0.004) were all significantly lower. LACS was similar between groups. In univariable Cox regression, all LA parameters were significant predictors of the AF/AFL (LAEF: HR 0.98 per 1% increase, 95% CI 0.96-0.99, p=0.0002, PALS: HR 0.96 per 1% increase, 95% CI 0.94–0.98, p=0.0008, PACS: HR 0.95 per 1%, 95% CI 0.93–0.98, p=0.0024, LACS: HR 0.96, 95% per 1% increase CI 0.92–0.998, p=0.039. In adjusted analysis, however, only LAEF (HR 0.98, 95% CI 0.976-0.995, p=0.0087) and PACS (HR 0.96, 95% CI 0.993–0.999, p=0.044) remained significant. Splines were constructed for LAEF and PACS (Figure 1), and a cutoff for increased risk of AF/AFL at 38% for LAEF and 12% for PACS, respectively, were derived (Figure 2). Conclusion LAEF and PACS are associated with risk of incident AF/AFL in patients with HFrEF, independent of clinical and echocardiographic risk factors. Our results suggest an increased risk of AF/AFL at LAEF below 38% and PACS below 12%.Figure 1 SplinesFigure 2 Cumulative incident curves