Modeling heart failure in a dish: functional signatures in human ventricular slices

Abstract Background Living myocardial slices (LMS) are ultrathin, living sections of cardiac tissue that preserve human-specific structure and function ex vivo. Unlike traditional models, LMS maintain the patient’s genotype and phenotype, allowing disease mechanisms to be studied in a controlled environment while directly reflecting human biology. This challenges reliance on animal models, which require cross-species extrapolation and carry risks of pathological and pharmacological mismatch, as well as on engineered cardiac models that fail to recapitulate myocardial complexity. Establishing large-scale, patient-specific human LMS can therefore advance mechanistic and translational understanding of the failing human heart. Aim To establish a patient-specific framework for human ventricular LMS by characterizing long-term biomechanical function across health and disease, including diverse heart failure (HF) phenotypes in vitro. Methods Ventricular tissue was obtained from donors, explanted hearts of transplant patients, or residual tissue from cardiac surgery. LMS were created following our optimized protocol and maintained in biomimetic chambers under isotonic loading and continuous electrical pacing to mimic in vivo conditions. Contraction metrics, including peak force (Fmax), time to peak (TTP), time to relaxation (TTR), and refractory period (RP), were measured using MyoDish software. Statistical analysis was performed using linear mixed-effects models. Results In total, 889 LMS from 126 individuals were analyzed. Fmax stabilized during the initial culture period (Day 0-3) and remained consistent from Day 4 through Day 14. Across the cohort, right and left LMS exhibited similar profiles. Healthy LMS generated higher Fmax than diseased tissue (1531.1 vs 944.7 µN, p=0.04) and showed trends toward shorter timing parameters but longer RP. Among HF subtypes, contraction profiles were comparable in ischemic (ICM, Fmax 1022.5 µN) and dilated cardiomyopathy (DCM, Fmax 1030.4 µN), with reduced Fmax in hypertrophic (HCM, 850.8 µN), lowest in arrhythmogenic (ACM, 489.6 µN), and highest in myocarditis-associated cardiomyopathy (MCM, 2381.2 µN). In transplantation patients, LMS from individuals with a left ventricular assist device (LVAD) exhibited higher Fmax (p=0.04) and prolonged TTP (p=0.04) compared with those without LVAD support prior to transplantation. Conclusion Human ventricular LMS sustain contractility in long-term culture and capture patient- and disease-specific biomechanical signatures across HF phenotypes. Reduced Fmax in most HF types suggests contractile impairment, whereas enhanced force in myocarditis may reflect reactive hypercontractility. Mechanical unloading through LVAD support may promote reverse remodeling and improve contractile performance. Collectively, these findings establish LMS as a translationally relevant platform for mechanistic HF research and drug testing, bridging bench and bedside.For image description, please refer to the figure legend and surrounding text. For image description, please refer to the figure legend and surrounding text.