From microstructure to arrhythmia: Structural drivers of re‐entrant activity in atrial fibrillation

Atrial fibrillation remains the most prevalent sustained cardiac arrhythmia worldwide, and its burden continues to escalate as populations age (Van Gelder et al., 2024). Characterised by rapid, irregular atrial activation and impaired mechanical function, atrial fibrillation increases the risk of stroke, heart failure and cognitive decline. At the electrophysiological level atrial fibrillation is sustained by disorganised wavefront propagation and chaotic electrical signalling within the atrial myocardium. Although the mechanisms behind this disarray have been long debated, they are increasingly recognised to be driven by localised, self-perpetuating sources of electrical activity that are embedded in atrial tissue and have been termed re-entrant drivers (Zhao et al., 2023). These drivers represent an appealing mechanistic target, and the ablation of them has been proposed as a rational strategy for atrial fibrillation termination. However despite major advances in mapping technologies, catheter design and ablation energy delivery, clinical outcomes have plateaued, particularly for patients with persistent atrial fibrillation. Contemporary data indicate that approximately only 50%–60% of patients with persistent atrial fibrillation remain free from recurrent arrhythmia following a single catheter ablation procedure, with outcomes substantially lower for those with long-standing persistent atrial fibrillation (Baqal et al., 2024). This modest efficacy reflects not a failure of technology as such, but a fundamental limitation in the ability to identify reliable, patient-specific targets. As a result, in recent years, a sharper research focus has been on the mechanisms that sustain atrial fibrillation and on understanding why ablation strategies fail to deliver lasting rhythm control. The majority of this research has consisted of in vivo imaging or animal models, and although these methods have their advantages, they fail to provide a detailed understanding of the three-dimensional structural substrates that anchor and stabilise re-entrant drivers within the human atria. Animal models do not fully replicate human atrial anatomy and in vivo human imaging lacks the spatial resolution necessary to fully understand the three-dimensional organisation of the atrial wall. In a recent issue of The Journal of Physiology, Kulathilaka and colleagues tackled these challenges by combining high-resolution ex vivo structural magnetic resonance imaging with validated functional mapping to generate heart-specific digital twins of the human atria. Using explanted human hearts from five donors with cardiac comorbidities, the authors constructed personalised atrial models in which key structural substrates could be selectively included or excluded, allowing unparalleled mechanistic dissection of the role of re-entrant drivers in atrial fibrillation. This framework enabled the creation of five models which spanned from simplified atrial geometries with uniform wall thickness to fully integrated representations of the atria that incorporated patient-specific wall thickness, myofibre architecture and fibrosis. Re-entrant activity was identified at multiple sites, and driver behaviour was quantified using three-dimensional phase singularity tracking. Although previous work has implicated atrial wall thickness, myocardial fibre architecture and fibrosis as key determinants of atrial arrhythmogenesis, it was unknown how these structural features interact within the human atria and what their relative contributions were to re-entrant driver initiation, drift and anchoring. The approach used by Kulathilaka et al. (2025) allowed them to directly investigate how individual structural features and their interactions shape the dynamics of re-entrant drivers in the human atria. Several important findings were noted from this work. Most strikingly atrial wall thickness emerged as a dominant and previously underappreciated determinant of re-entrant driver behaviour. Across simulations drivers consistently drifted towards and stabilised within regions of reduced atrial wall thickness, an effect that was more pronounced in the right atrium than in the left. Meanwhile myofibre orientation and fibrosis exerted additional region-specific influences. In the right atrium drivers preferentially localised to regions of coherent fibre alignment bordering areas of fibre disorganisation. In contrast, fibrosis played a more prominent role in the left atrium, where drivers clustered near fibrotic boundaries, underscoring chamber-specific mechanisms of re-entry stabilisation. These findings carry important conceptual and translational implications. They challenge the prevailing fibrosis-centric view of atrial fibrillation substrates and argue for a broader structural perspective in which atrial geometry and fibre architecture play central roles in sustaining re-entrant activity. This is important as many current ablation strategies focus on fibrotic regions (Zhao et al., 2017). From an ablation standpoint the work suggests that wall thickness gradients and fibre organisation may represent actionable targets, either independently or in combination with fibrosis, opening the door to more sophisticated and mechanism-based intervention strategies. Additionally the findings demonstrate that the dominant structural determinant for atrial fibrillation differs between the right atrium and the left atrium. Taking these findings into consideration novel predictive approaches which target atrial fibrillation re-entrant drivers could be developed. A key strength of this study is its use of explanted human atrial tissue combined with direct functional validation. The use of ex vivo optical mapping enabled definitive localisation of re-entrant drivers, providing a robust basis for model calibration and selection. Also the high spatial resolution of the imaging data permitted detailed characterisation of atrial wall thickness, myofibre architecture and fibrosis at a level not currently achievable with clinical magnetic resonance imaging. Furthermore the systematic comparison of five model variants further allowed the isolation of individual substrate effects and their interactions. Nonetheless several limitations merit consideration. The researchers only examined five donor hearts, four of which were from male donors, and all donors had cardiac comorbidities. This may have introduced a potential selection bias. Additionally only atria exhibiting optically mapped re-entrant drivers were modelled, which may limit the generalisability of the dataset across the heterogeneous population of patients with atrial fibrillation. Moreover the ex vivo setting differs from the normal in vivo physiology of patients, with changes in mechanical loading, autonomic influences and tissue viability affecting arrhythmia behaviour. Fibrosis was modelled in a simplified way as regions of reduced electrical conductivity. However in living hearts, fibrotic remodelling is highly heterogeneous, with different cellular compositions and spatial patterns influencing electrical activity through mechanisms beyond conduction slowing alone. Thus further advances in imaging technology are needed to fully comprehend the mechanisms behind atrial fibrillation patterns. The models also assumed uniform ionic properties and did not include mechano-electrical coupling, both of which may shape atrial fibrillation dynamics in vivo. Finally the high computational costs and the need to down sample imaging data raise important questions about the scalability of this research and future clinical applications. Nevertheless this work represents a timely and significant advance in mechanistic atrial fibrillation research. By integrating high-resolution human structural data with validated computational modelling, Kulathilaka and colleagues provide a compelling framework for rethinking how atrial fibrillation substrates are defined and targeted. At a time when clinical ablation strategies require renewed mechanistic grounding, this study offers both conceptual clarity and a path forward. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. None declared. Sole author. None.