Characterisation of exercise responses in HFpEF stratified by ejection fraction: implications for therapeutic myosin modulation

Background Patients with heart failure and preserved ejection fraction (HFpEF) and higher LVEF >60% derive less benefit from neurohormonal therapies compared to those with LVEF 50-60% suggesting fundamental differences requiring precision therapeutic approaches. Purpose We sought to determine whether mechanisms of exercise intolerance differ between HFpEF patients with resting LVEF>60% and those with LVEF 50-60% by characterising haemodynamics using cardiac magnetic resonance (CMR) and modelling the potential effects of myosin inhibition using electro-mechanical digital twins. Methods 64 patients with HFpEF and 16 controls were prospectively recruited and underwent free-breathing, real-time CMR imaging at rest and following 5 minutes of supine steady-state exercise at 35W using an ergometer. Pulmonary congestion was quantified using a proton density sequence. Non-invasive pressure-volume (PV) loops were generated using a time-varying elastance function incorporating brachial blood pressures coupled with CMR data. These were used to create patient-specific electromechanical digital twin models to simulate predicted effects of therapeutic myosin modulation. Results In comparison to HF-EF50-60% (n=18), patients with LVEF>60% (n=46) had smaller ventricles at rest (LVEDVi 63.8 [54.0, 73.4] vs 89.7 [80.5, 111.4] ml/m2, p< 0.001) with diminished exercise cardiac index (4.3 [3.7, 5.4] vs 4.9 [4.2, 6.4] L/min/m2, p=0.048) and shorter 6-minute walk distance (6MWD 364 [285, 420] vs 390 [353, 454] m, p=0.03) despite lower developed pulmonary congestion (∆proton density -0.2 [-3.8, 2.5] vs 3.6 [-0.1, 9.1]%, p=0.005). PV loop analysis showed left-shifted curves with pathologically elevated contractility compared to HF-EF50-60% at rest (2.9 [2.3, 3.5] vs 1.4 [1.2, 1.6] mmHg/mL, p<0.01), during exercise (3.6 [2.7, 4.1] vs 1.7 [1.3, 2.2] mmHg/mL, p<0.01) and relative to controls at rest (2.3 [2.0, 2.9] mmHg/mL, p=0.04). Exercise cardiac index was strongly correlated with 6MWD across all groups (ρ=0.47, p<0.001). In silico digital twin simulation of mavacamten in a representative patient from each group at a therapeutic concentration of 500nM predicted augmentation of cardiac index at rest (+0.4L/min/m2) and exercise (+0.7L/min/m2) in HF-EF>60% mediated by improved diastolic compliance without deleteriously reducing LVEF. However, HF-EF50-60% had a narrow therapeutic range with declining LVEF. Conclusion HFpEF with LVEF >60% represents a physiologically distinct hypercontractile endotype. Cardiac myosin inhibition is predicted to normalise pathological left-shifted pressure-volume relationships seen in HF-EF>60%, restoring stroke volume augmentation and improving exercise cardiac index, and with higher therapeutic index than in HF-EF50-60%. The findings therefore provide a mechanistic rationale for LVEF-stratified trials of cardiac myosin modulators to improve exercise haemodynamics, and thereby potentially symptoms in this subgroup.