Scientists propose new blueprint to model and reverse atrial fibrosis in AF

Atrial fibrillation, the most common sustained heart-rhythm disorder worldwide, is becoming steadily more prevalent as populations age and obesity, hypertension, diabetes, valvular disease and heart failure grow more widespread. Anticoagulation, rate and rhythm control, and catheter ablation can reduce serious clinical risks, but none of these interventions reliably eliminates the structural substrate that keeps persistent or recurrent atrial fibrillation alive. A major new review in Materials Today Bio argues that this therapeutic stalemate reflects a deeper translational problem: the field still lacks models that connect molecular regulation and fibroblast state with extracellular matrix organization, tissue mechanics, electrical conduction, contractility and safety. Authors Jing Ni, Yi Luo and Rong Guo of the review’s institution contend that atrial fibrosis is not a single measurable entity but a multiscale component of a broader condition called atrial cardiomyopathy, and that only a carefully staged, evidence-bound integration of patient-specific stem cells, engineered heart tissue platforms and nanomedicine can close the gap between laboratory promise and clinical benefit.

The review’s central message is one of disciplined realism. Atrial fibrosis involves interlinked but distinct changes in stromal-cell state, extracellular matrix turnover, matrix architecture and tissue function. Fibroblast activation and fibroblast-to-myofibroblast transition are prominent features, but the authors stress that neither should be treated as synonymous with established tissue fibrosis, which additionally depends on the amount, composition, organization and persistence of deposited matrix. Fibroblast phenotype is better understood as a continuum, with quiescent, activated, matrix-producing, inflammatory and highly contractile states coexisting and shifting with stimulus, disease stage and culture conditions. Signaling pathways such as TGF-beta/Smad, Wnt/beta-catenin, the renin-angiotensin-aldosterone axis, inflammatory cascades, oxidative stress and Hippo-YAP/TAZ reinforce one another in context-dependent ways, which helps explain why suppressing a single molecular node rarely produces durable matrix or functional recovery.

Equally important, the authors warn that molecular regulation of atrial fibrosis is tissue- and chamber-dependent. Atrial fibroblasts differ from their ventricular counterparts in electrical coupling, mechanosensitivity, metabolic state and ion-channel expression, meaning that findings from ventricular, post-infarction or extracardiac fibrosis models cannot simply be transplanted into the atrial setting. At the tissue level, reactive interstitial fibrosis must be distinguished from replacement fibrosis associated with myocyte loss, and total collagen abundance, fiber orientation, crosslinking and patch size can have very different effects on conduction and mechanical compliance. Clinical imaging adds further ambiguity: late-gadolinium cardiac magnetic resonance and low-voltage mapping identify remodeling-related abnormalities, but the thin atrial wall, spatial resolution limits and threshold-dependent quantification mean imaging-defined fibrosis should be treated as a biomarker, not as a direct histological measurement of collagen or matrix reversal.

To organize the field’s often overclaimed evidence, the review classifies every claim into three categories: directly atrial and AF-specific evidence from human atrial tissue or chamber-validated cells; cardiac but non-atrial evidence from ventricular or generic fibrosis systems; and extrapolated evidence from extracardiac fibrosis, oncology or delivery engineering. The authors also introduce three recurring sources of model-related uncertainty. A mechanism mismatch arises when the experimental stimulus or cell composition fails to reproduce the pathway under study. An endpoint mismatch occurs when reduced collagen staining or molecular suppression is interpreted as tissue-level functional recovery. A safety-context mismatch appears when electrophysiological liability is assessed in healthy or ventricular tissue while the intended therapy targets a remodeled atrial substrate. These categories describe sources of uncertainty rather than proven causes of clinical trial failures, but they give researchers a shared vocabulary for judging whether a model has the capabilities its conclusions require.

Within this framework, patient-derived induced pluripotent stem cells emerge as a powerful but bounded tool. iPSC-derived atrial cardiomyocytes retain the donor’s germline genetic background and can be directed toward atrial identity through timed retinoic-acid exposure and controlled Wnt signaling, with chamber identity confirmed through convergent markers such as NPPA, NR2F2, KCNA5 and GJA5 rather than any single indicator. Yet the review is blunt about limitations: reprogramming and culture attenuate age-related and acquired epigenetic states, and differentiated cells do not automatically reproduce AF duration, obesity, diabetes, inflammation or medication exposure. iPSC-derived atrial cardiomyocytes also remain fetal-like in metabolism, structure and electrophysiology, and maturation interventions can trade adult-like features against variability and throughput. The strongest direct evidence, including work by Seibertz and colleagues showing that tachypacing chamber-validated atrial cardiomyocytes reproduces selected features of AF-associated electrical remodeling, validates only specific claims, not an integrated fibrosis-on-chip platform or patient-specific anti-fibrotic prediction.

The review’s most ambitious proposal is a fit-for-purpose, donor-informed atrial fibrosis microphysiological system built from validated modules. Microengineering can supply anisotropic structure through micropatterned scaffolds, controlled mechanical loading through programmable stretch, and fluidic exposure through endothelialized perfusion, but each capability must be justified against a specific clinical scenario such as pressure-dominant hypertension, volume-dominant valve disease or tachycardia-driven remodeling. Fibroblast provenance is a central design decision: adult primary atrial fibroblasts, iPSC-derived cardiac fibroblasts and commercial primary cardiac fibroblasts are not interchangeable, and the review argues that source, activation state and cardiomyocyte-to-fibroblast ratio must be documented alongside donor metadata including sex, age, AF subtype, comorbidities and clone identity. Representative studies, from Brown and colleagues’ patterned atrial cardiomyocyte-fibroblast cocultures to Reyat and colleagues’ vascularized chamber-specific microtissues showing TGF-beta-driven myofibroblast activation and collagen deposition that responds to receptor inhibition, demonstrate real progress while illustrating how far a fully integrated, mechanically actuated, patient-specific platform still remains.

On the evaluation side, the authors propose tiered high-content phenotyping in which candidates must clear successive gates: molecular target engagement, fibroblast-state modulation, extracellular matrix synthesis and crosslinking, tissue structure and mechanics, electrophysiological function and safety. Microelectrode arrays, voltage and calcium optical mapping, and calibrated force measurements can each provide directly measured endpoints, but higher-level labels such as rhythm stability, reentry suppression or fibrosis reversal require explicit definitions and should never substitute for the underlying data. The review also calls for a shared quality-control matrix across laboratories, specifying minimum reporting items for donor information, differentiation efficiency, device geometry and materials, perfusion and stretch settings, reference pro-fibrotic and anti-fibrotic compounds, and batch-to-batch coefficients of variation. Nominal dose, the authors note, must be distinguished from actual exposure, particularly because device materials such as polydimethylsiloxane can adsorb both small molecules and nanocarriers.

Nanomedicine receives similarly careful treatment. The authors argue that no biologically active payload can modify atrial fibrosis unless adequate unbound exposure reaches the relevant atrial compartment for a sufficient duration, yet direct quantitative evidence for nanoparticle biodistribution in fibrotic human atria is sparse. Delivery route fundamentally changes both opportunity and risk, from intravenous administration that exposes liver, spleen and immune system first, to intracoronary, intracardiac, epicardial or pericardial approaches that improve local concentration while introducing procedural constraints. Particle size, surface chemistry, charge, protein-corona formation and release kinetics jointly shape circulation, penetration and uptake, and dense crosslinked matrix can restrict diffusion, though most quantitative evidence for this barrier comes from hepatic, pulmonary, tumor or cartilage models. Stimulus-responsive release systems and fibroblast-targeting ligands remain promising design principles rather than validated atrial solutions, and one atrial-selective adeno-associated viral vector stands as the clearest demonstration that chamber-biased delivery is biologically achievable.

The review closes with a proposed closed-loop workflow in which delivery measurements inform formulation changes, biological and functional endpoints test mechanistic hypotheses, and safety and uncertainty drive go or no-go decisions, all documented with full traceability. The authors position these platforms as experimental decision-support models, explicitly rejecting the phrase clinical trial in a dish, since no chip reproduces whole-body pharmacokinetics, neurohumoral regulation or clinical outcomes. Artificial intelligence may assist with segmentation, feature extraction and dose-response modeling, but the authors insist on donor-level data splitting, external validation and prospective testing in newly manufactured devices before closed-loop optimization can be considered demonstrated. Their priority list includes multicenter benchmarking with shared controls, direct comparison of fibroblast sources, loading histories mapped to clinical pressure and volume overload, and blinded evaluation of reference compounds. The central opportunity, they conclude, is not a single all-inclusive platform but a transparent evidence chain in which reductionist assays, chamber-specific human atrial models, microphysiological systems, in vivo studies and clinical observations each address a defined uncertainty, moving the field toward therapies that might one day genuinely reverse the structural substrate of atrial fibrillation.

Subject of Research: Patient-specific iPSC models, atrial microphysiological systems and nanomedicine for modeling AF-associated atrial fibrosis

Article Title: Reconstructing AF-associated atrial fibrosis: Patient-specific iPSC models, fit-for-purpose atrial microphysiological systems, and nanomedicine

Article References: Ni, J., Luo, Y., & Guo, R. (2026). Reconstructing AF-associated atrial fibrosis: Patient-specific iPSC models, fit-for-purpose atrial microphysiological systems, and nanomedicine. Materials Today Bio, 40, Article 103633. https://doi.org/10.1016/j.mtbio.2026.103633

Image Credits: AI Generated

DOI: 10.1016/j.mtbio.2026.103633

Keywords: atrial fibrillation, atrial fibrosis, iPSC, organ-on-a-chip, microphysiological systems, cardiac fibroblasts, extracellular matrix, nanomedicine, drug delivery, TGF-beta signaling, electrophysiology, atrial cardiomyopathy