Wei, Eric Q. and Beyer, Martin and Brown, Kemar J. and Bansbach, Alexander J. and Gorham, Joshua M. and McDonough, Barbara and Chen, Huachen and Khoramjoo, Mobin and Zhang, Anran and Bishop, Brian and Ahmad, Ferhaan and del Rio, Carlos and Chang, Ching-Pin and Ryba, David M. and Day, Sharlene M. and Fatkin, Diane and Oudit, Gavin Y. and Seidman, Christine E. and Seidman, Jonathan G. (2026) Deep-learning analysis of 3D microarchitectural remodeling in hypertrophic cardiomyopathy. Science, 391 (6782). ISSN 0036-8075
Full text not available from this repository.Abstract
INTRODUCTION
Hypertrophic cardiomyopathy (HCM), an inherited heart disease affecting ~1 in 200 to 500 people worldwide, is a leading cause of sudden cardiac death. Hallmark features include left ventricular (LV) wall thickening, cardiomyocyte disarray, and fibrotic remodeling. Pathogenic sarcomere variants—most commonly in MYH7 or MYBPC3 (PVpos)—account for the majority of familial cases and show worse clinical outcomes compared to patients without identifiable variants (PVneg). Although cardiac hypertrophy has long been recognized as a defining feature of HCM, the relative contributions of cardiomyocyte hypertrophy and extracellular volume (ECV) to LV wall thickening—and their dependence on genotype and disease stage—remain unresolved.
RATIONALE
Characterizing the structural basis of LV wall thickening is critical to understanding HCM pathogenesis and the genotype-specific differences in clinical outcomes observed in patients. Conventional histology, restricted to thin two-dimensional (2D) sections, cannot accurately quantify cardiomyocyte or extracellular volumes in situ. To overcome these limitations, we developed CaMVIA-3D, a volumetric imaging and deep-learning pipeline for high-resolution analysis of cardiac microarchitecture, including 3D cardiomyocyte morphology and extracellular volume.
RESULTS
We applied CaMVIA-3D to cardiac tissues obtained from patients with HCM who underwent myectomy or transplant and to a longitudinal pig model carrying the endogenous MYH7 R403Q variant. Human PVpos hearts showed marked concentric cellular hypertrophy with disarray, most pronounced in MYH7 , whereas PVneg hearts maintained normal cell volumes but exhibited substantial ECV expansion. Modeling the relationship between microstructural features and LV wall thickness showed that PVpos thickening reflected nearly equal contributions of cardiomyocyte hypertrophy and ECV expansion, whereas PVneg thickening was driven primarily by ECV. Longitudinal analysis of MYH7 R403Q pigs recapitulated the human phenotype, with progressive concentric cellular hypertrophy and disarray. Notably, fibrotic remodeling—including replacement fibrosis and interstitial fibrosis—preceded overt cardiomyocyte hypertrophy, highlighting fibrotic remodeling as an early driver of HCM. To connect structural remodeling with underlying molecular mechanisms, we integrated CaMVIA-3D data with single-nucleus RNA sequencing (snRNA-seq). Using elastic-net regression, we identified gene sets in cardiomyocytes and fibroblasts most strongly associated with microstructural features, providing a reference for future studies.
CONCLUSION
Leveraging advanced microscopy and deep learning–based 3D analysis of human and pig HCM hearts across multiple disease stages, we uncovered genotype-specific differences in cardiomyocyte morphology and extracellular matrix composition. These findings highlight distinct underlying biological mechanisms and underscore the need for targeted therapies based on genotype and remodeling patterns.
Genotype-specific mechanisms of wall thickening in hypertrophic cardiomyopathy.
Volumetric imaging and deep-learning analysis (CaMVIA-3D) of human and pig hearts revealed distinct remodeling patterns: PVpos (pathogenic variant–positive; including MYH7 and MYBPC3 ) hearts showed concentric cardiomyocyte hypertrophy, whereas PVneg (pathogenic variant–negative) hearts exhibited extracellular volume (ECV) expansion as the primary driver. Integration with snRNA-seq linked microstructural features to gene expression. [Figure created in part with BioRender.com]
| Item Type: | Article |
|---|---|
| Subjects: | R Medicine > R Medicine (General) |
| Depositing User: | Repository Administrator |
| Date Deposited: | 29 Aug 2026 04:47 |
| Last Modified: | 29 Aug 2026 04:47 |
| URI: | http://eprints.victorchang.edu.au/id/eprint/1816 |
Actions (login required)
 |
View Item |