Phenotyping atrial myopathy in HFpEF patients with and without clinical atrial fibrillation

Background

Heart failure with preserved ejection fraction (HFpEF) is a highly heterogeneous syndrome frequently accompanied by atrial remodelling (electrically, mechanical and biochemical). Clinical and subclinical atrial fibrillation (AF) are frequent and exercise levels are often reduced. Moreover, data on LA function during exercise or stress testing are scarce, although dynamic assessment may better reflect functional reserve and early atrial dysfunction.

Purpose

This study aims to phenotype HFpEF patients (pts) using multimodal atrial parameters integrating electrical, structural, functional, and biochemical markers, and to explore whether distinct atrial remodelling profiles can be identified. We also assessed exercise performance in patients with and without clinical atrial fibrillation.

Methods

A total of 153 HFpEF pts (clinical AF (n=61); no AF (n=91); 73±8y; 45% women) were included. We analysed LAVI (left atrial volume index), LA emptying fraction (LAEF), LA reservoir strain (LASr) at baseline and during stress echo (n=69, 2CH, 4CH), age-predicted workload in cardiopulmonary exercise testing ², electrical markers of atrial remodeling (P-wave amplitude, duration, dispersion, P-wave terminal force in lead V1 (PTFV1)) in SR ECGs (n=99), and NT-proBNP. Cluster analysis, correlation matrices, and principal component analysis (PCA) were performed to identify patterns of atrial remodelling.

Results

Pts with clinical AF had significantly lower LAEF and reduced LASr (both p < 0.001) than those without AF. However, even in SR, approximately half of HFpEF pts showed pathological ³ LASr (52%) or LAEF (40%), indicating widespread subclinical atrial dysfunction. During stress echo, LASr failed to augment with increasing workloads. Despite these differences, exercise workload (% predicted) was similar between groups (p = 0.4), yet, exercise capacity was markedly reduced and few pts achieved age-predicted targets.

Cluster analysis based on ECG, LA, and NT-proBNP parameters did not yield distinct subgroups (Cronbach’s α = 0.018). PCA explained ~39% of variance (PC1 = 22%, PC2 = 17%). PC1 reflected functional–electrical coupling (P-wave amplitude, LAEF, LASr), whereas PC2 captured structural and load-related aspects (LA volume, NT-proBNP, P-wave dispersion). LAEF and LASr correlated positively, while NT-proBNP and P-wave amplitude correlated inversely with LA function.

Conclusion

HFpEF pts exhibit substantial atrial remodelling beyond overt AF, with potential thromboembolic implications. Exercise capacity is markedly reduced irrespective of AF, and exercise atrial strain is similarly depressed, indicating severely limited atrial reserve. Electrical, mechanical, and biochemical abnormalities are interrelated but non-redundant. The absence of discrete clusters and a multidimensional principal-component structure underscore heterogeneity, supporting integrated multimodal assessment for finer phenotyping and risk stratification.Figure 1Table 1 and Table 2