Mapping myocarditis in three dimensions

This editorial refers to ‘Whole-heart 3D reconstruction of mouse CVB3 myocarditis reveals spatial and transcriptomic heterogeneity of immune foci’ by A. Daoud et al., https://doi.org/10.1093/cvr/cvaf209. Myocarditis remains one of the most enigmatic inflammatory diseases of the heart.1 Despite decades of research, clinicians continue to face major challenges in its diagnosis, risk stratification, and treatment.1–5 The clinical spectrum ranges from fulminant presentations with high early mortality to insidious progression towards dilated cardiomyopathy (DCM).1,4 Myocarditis continues to be one of the most common causes of sudden cardiac death.6 While cardiac imaging has advanced our ability to detect myocardial inflammation, and endomyocardial biopsy (EMB) remains the diagnostic gold standard, both approaches suffer from critical limitations (Figure 1).3 EMB in particular is hampered by a variable diagnostic yield, in part due to the focal distribution of inflammatory lesions.3 Current tools available to assess myocarditis. Of the many tools that are currently available to assess myocardial inflammation, the novel method presented here combines immunohistochemistry (IHC) with artificial intelligence (AI)-directed spatial transcriptomics to provide three-dimensional data about the molecular drivers of myocarditis with the potential to transform our understanding of disease. Created in BioRender. Fairweather, D. (2025) https://BioRender.com/0d8ft6o. Against this backdrop, the work of Daoud and colleagues, published in this issue of Cardiovascular Research, represents an important methodological and conceptual advance.7 The authors apply a novel AI-assisted workflow for histological reconstruction—CODA—to generate whole-heart three-dimensional reconstructions of acute Coxsackievirus B3 (CVB3) myocarditis in mice, integrating immunohistochemistry and spatial transcriptomics. This method offers several advantages over traditional mapping using single-cell transcriptomics8 by providing detailed spatial relationships. This study not only establishes a new standard for whole-organ spatial immunology but also yields insights that challenge long-standing assumptions about the nature of myocarditis lesions. Daoud et al. demonstrate that acute viral myocarditis does not consist of diffuse, uniformly distributed immune infiltrates. Instead, the disease is dominated by a relatively small number of large, elongated, and highly branched inflammatory foci. Importantly, these foci preferentially occupy the free left ventricular wall, a region rarely sampled by routine EMB, which typically targets the interventricular septum for safety reasons. This anatomic distribution provides a compelling explanation for the historically poor sensitivity of EMB in suspected myocarditis. Equally striking is the discovery of spatial immunological heterogeneity within and between foci. Using CD3 and CD64 staining in combination with spatial transcriptomics, the investigators reveal distinct immune niches: T cell–rich hotspots localizing predominantly to anterior regions of the heart, and macrophage-rich areas in posterior regions. At higher resolution, T cell hotspots are shown to coincide with zones of high vascularization, suggesting that vascular access may shape immune cell recruitment and retention. Moreover, transcriptomic analyses reveal that anterior, T cell–rich foci upregulate chemokines such as Cxcl10, Ccl7, and Ccl8, along with type I collagen, while surrounding homogeneous foci preferentially express adhesion and retention molecules including Cd93, Adgre5, and S1pr4. These findings are not only internally validated by flow cytometry and quantitative PCR across a larger cohort of mice but also supported by human data. Explanted hearts from patients with inflammatory DCM exhibited a similar anterior–posterior gradient in T cell vs. macrophage content, underscoring the translational relevance of the mouse model. Several implications arise from this study. First, the anatomic distribution of immune foci highlights the intrinsic limitations of current biopsy practice. If the majority of inflammatory lesions lie in the free left ventricular wall, blind septal biopsy will inevitably miss pathology in a substantial subset of patients. CMR guidance as suggested in prior literature may mitigate that risk.4 Second, the recognition of spatial immune niches introduces a new dimension to the prognostic assessment of myocarditis (Figure 1). Prior studies have established the prognostic relevance of total CD3+ T cell counts in DCM. Daoud et al. now suggest that the ratio of T cells to macrophages, and their anterior localization, may provide even stronger prognostic information. The association of T cell hotspots with collagen I expression further links immune heterogeneity to fibrotic remodelling, a major determinant of adverse outcomes. Collagen expression may not only reflect damage to the cardiac infrastructure caused by viral infection but, in this autoimmune model of heart-passaged viral myocarditis,9,10 collagen may contribute to autostimulation of T cells. Third, the methodological innovation cannot be overstated. CODA overcomes the limitations of tissue-clearing techniques by leveraging AI-assisted reconstruction of serial histological sections, thereby enabling multimodal integration at single-cell resolution across the entire organ. The ability to map morphological, immunological, and transcriptional features in three dimensions opens the door to anatomic contextualization of virtually any cardiac pathology. While the study represents an important milestone, several avenues for future work emerge. Longitudinal studies are needed to track the evolution of immune niches from acute to chronic stages, particularly to understand whether anterior T cell hotspots persist, resolve, or transform into fibrotic scars. That will require the use of BALB/c mice, which progress from acute myocarditis to DCM in this model while male C57BL/6 mice do not.11,12 Importantly in this model of viral myocarditis, male C57BL/6 mice develop a fulminant form of myocarditis while male BALB/c develop lymphocytic myocarditis- the most common form of myocarditis in the US and Europe. Additionally, perimyocardial inflammation is a common feature in clinical cases of myocarditis and is found in BALB/c but largely absent in C57BL/6 mice using this model.11,12 Thus, applying CODA to other aetiologies of myocarditis—autoimmune, lymphocytic, giant cell, and vaccine-related—will test the generalizability of the anterior—posterior immune gradient. Integration with electrophysiological mapping could further elucidate whether localized immune foci contribute to arrhythmogenesis, a major cause of sudden cardiac death in myocarditis. The role of B cells and other immune subsets, while less prominent in this study, warrants deeper exploration in chronic disease stages. Finally, the therapeutic implications of spatial immunology must be explored. If chemokine gradients and vascular niches drive preferential T cell recruitment, could targeted modulation of these pathways alter disease progression or remodelling? Could the T cell/macrophage ratio itself become a therapeutic target? Such questions exemplify how spatial insights can move from description to intervention. Increasing advances in AI-driven analysis tools will reveal key mechanisms of disease and one day be seen not only at the bench but in the clinic. The study by Daoud and colleagues delivers the first three-dimensional, multimodal atlas of myocarditis, revealing unexpected anatomic and immunologic heterogeneity. It reframes the limitations of current diagnostic tools, proposes new prognostic markers, and highlights mechanistic links between immune localization and fibrosis. More broadly, it exemplifies how spatial biology can transform cardiovascular research. As CODA and similar platforms are applied across disease entities, we move closer to a spatially informed cardiology- one in which the ‘where’ of pathology is as important as the ‘what.’ BH receives funding from the Deutsches Herzzentrum Berlin (DHZB) Foundation. This study was supported by funding from the National Institutes of Health R01 HL164520 to DF.