Decoding the cellular and paracrine landscape of epicardial adipose tissue–myocardial crosstalk in coronary artery disease

The epicardial adipose tissue (EAT), located between the myocardium and visceral pericardium, is uniquely positioned to communicate directly with the underlying heart, given the absence of a fascial layer between these two tissues.1,2 Under physiological conditions, the EAT serves metabolic, mechanical, and thermogenic functions that may support myocardial homeostasis. However, ageing and cardiometabolic disease can shift this protective phenotype towards a dysfunctional state, characterized by inflammation, fibrosis, altered adipokine secretion, and metabolic stress.1,2 Due to this phenotypic plasticity, interest in the EAT as a potential interface between systemic metabolic disease and local cardiovascular injury is growing. Clinically, the EAT has traditionally been assessed predominantly in terms of quantity. A greater EAT volume has been associated with coronary artery disease (CAD)3 and atrial fibrillation.4 However, the biological heterogeneity of EAT is increasingly recognized as being at least as relevant as its anatomical burden. Human epicardial adipocytes can undergo disease-related molecular remodelling,5 with a dynamically modifiable phenotype: empagliflozin suppresses epicardial preadipocyte differentiation and shifts the EAT secretome towards a less pro-inflammatory profile, with attenuation of cardiomyocyte oxidative stress.6 Conversely, glucose fluctuations induce broad transcriptional reprogramming of human epicardial adipocytes, strongly upregulating IL-1β and enhancing paracrine oxidative stress in cardiomyocytes.7 Cathelicidin (LL37) is another example of a biologically relevant EAT-associated inflammatory signal: its expression is increased in the EAT of patients with CAD, while its suppression attenuates atherosclerotic lesion development in experimental models.8 Together, these observations establish EAT as a metabolically responsive and pharmacologically modifiable signalling organ, rather than as a static fat depot. Against this background, the distinctive contribution of the study by Doppler et. al., titled ‘Beyond the vessel wall: single-cell dissection of epicardial fat–myocardial crosstalk in coronary artery disease’, was to resolve this functional heterogeneity at the cellular level and to connect specific inflammatory cell states with candidate signalling pathways and target-cell responses.9 An important aspect of this study was its focus on anatomy. Previous transcriptomic studies have investigated pericoronary EAT and their potential crosstalk with the adjacent coronary vasculature,10 whereas, in this study, the authors profiled myocardial-associated EAT (mEAT) collected from a non-pericoronary region overlying the left ventricular myocardium. This shift was more than a change in the sampling location; it expanded the conceptual framework of EAT biology from a predominantly adipose–coronary artery axis to an integrated adipose–vascular–myocardial network. In patients with CAD, myocardial remodelling may therefore not only reflect the consequences of epicardial coronary stenosis and ischaemia but also the influence of a local inflammatory and metabolic microenvironment immediately adjacent to the myocardium. Using single-nucleus RNA sequencing, the authors identified two CAD-enriched inflammatory cell types: inflammatory adipogenic cells (IAC) and inflammatory stromal cells (ISC). IAC are characterized by impaired adipocyte differentiation and maturation, along with activation of inflammatory and stress-response programmes, whereas ISC display a fibroinflammatory transcriptional profile. These observations are supported by histological evidence of adipocyte hypertrophy and fibrosis in CAD-associated mEAT. Importantly, these populations cannot be interpreted in isolation. By comparing IAC and ISC signatures across independent single-cell and single-nucleus datasets from cardiac adipose tissue, pericoronary EAT, and other white adipose depots, the authors demonstrated that IAC share features with previously described inflammatory adipocyte states, whereas ISC partially overlap with known fibroinflammatory stromal populations, albeit while retaining a distinct marker combination. This cross-dataset comparison strengthened the biological plausibility of newly defined cell states and placed them within the broader landscape of human adipose tissue remodelling.9 Another strength of the study was the attempt to move from a descriptive cellular taxonomy to functional intercellular signalling. Cell–cell communication analyses identified THBS1–CD36 and SEMA4A/SEMA3C–NRP1/PLXNA signalling as prominent CAD-associated candidate pathways that link inflammatory mEAT populations with vascular and myocardial cell types. These computational predictions were complemented by functional experiments: conditioned medium derived from CAD-associated adipocytes/IAC altered the contractile properties of human-induced pluripotent stem cell-derived cardiomyocyte microtissues and impaired tube formation and migration of primary human endothelial cells. Recombinant THBS1 and SEMA3C partially reproduced the selected phenotypes, whereas the combined blockade of CD36 and NRP1 partially attenuated these phenotypes. Taken together, these experiments provided functional support for the involvement of these signalling axes, stopping short of establishing in vivo causality.9 The observed incomplete rescue is biologically informative. Rather than supporting a single-mediator model, the partial effects of ligand supplementation and receptor blockade in this study indicated the complexity of the inflammatory EAT secretome, in which THBS1 and semaphorin signalling represent relevant, but not exclusive, components. This is an important conceptual point: dysfunctional mEAT may act less as a source of a single, dominant pathogenic molecule than as a multicellular signalling hub that reshapes endothelial, stromal, immune, and myocardial behaviour through a network of interacting mediators.9 Nevertheless, several caveats limit the interpretation of the study’s findings. Its cross-sectional design could not identify whether IAC and ISC precede CAD, emerge during CAD progression, or develop as a consequence of this cardiovascular milieu. Residual confounding factors related to age, systemic atherosclerosis, non-ischaemic cardiac remodelling, and local ischaemic stress may exist. Additionally, myocardial transcriptomic analyses used right atrial tissue rather than ventricular tissue, and endothelial assays relied on saphenous vein endothelial cells rather than coronary microvascular endothelial cells, limiting direct extrapolation to ventricular ischaemic remodelling and coronary microvascular biology. Finally, the imaging subset was small and exploratory. The exclusion of patients with diabetes improved the mechanistic interpretability, but reduced real-world generalisability. These limitations do not diminish the biological interest of the study; rather, they define the directions that should be pursued to determine whether inflammatory mEAT is a driver, amplifier, or consequence of CAD. This study provided a translational bridge between molecular biology and cardiovascular imaging. The EAT volume measured by computed tomography can identify an enlarged adipose depot, whereas the molecular signatures related to IAC and ISC may identify a biologically active depot. These features may overlap, but are unlikely to be equivalent. In this study, a greater IAC/ISC abundance was observed alongside a larger EAT volume and enlarged adipocyte lipid droplets, although the small imaging subset implies that these observations should be regarded as hypothesis-generating. The next challenge is to connect molecularly defined tissue states to non-invasive imaging phenotypes. This direction is already supported by published work showing that perivascular fat attenuation, captured by the Fat Attenuation Index, reflects coronary inflammatory activity.11 Future integration of EAT volume, attenuation, texture, or radiomic features with molecular signatures may allow a more informative form of non-invasive EAT phenotyping and, ultimately, may help to distinguish simply enlarged fat from biologically active fat. Ultimately, the importance of this study lies not in introducing a simple distinction between EAT quantity and quality, but in providing biological resolution to the concept of the EAT phenotype. By combining single-nucleus transcriptomics, histology, computational communication analysis, imaging, and functional validation, this work defined which cell states are altered in CAD, in which signalling circuits they engage, and how these signals may influence vascular and myocardial cells. Their findings position the EAT as a dynamic component of the cardiac microenvironment, the pathogenic relevance of which depends on the cellular composition, metabolic context, and intercellular communication. Whether the inflammatory cell states and signalling axes identified in this study represent therapeutic targets, imaging biomarkers, or mechanistic intermediates requires prospective and disease-specific validation. Therefore, the next step is not simply to characterize the EAT as biologically active, but also to determine which molecular and cellular programmes are clinically consequential, modifiable, and detectable in vivo.