Abstract Background Pressure-volume (PV) loop analyses recently identified two different hemodynamic phenotypes in response to exercise in heart failure with preserved ejection fraction (HFpEF). In addition to a rightward shift (RS) along the end-diastolic PV relation, a cohort of HFpEF patients characterised by a straight upward shift (US) of the PV relation with markedly increased systolic and diastolic stiffness and impaired left ventricular (LV) stroke volume (SV) was identified. Whether this is an effect of RV-LV interdependence or an alternative mechanism remains elusive. Purpose To invasively study biventricular exercise physiology in patients with HFpEF to decode cardiac mechanisms underlying exercise intolerance. Methods HFpEF was diagnosed according to the 2016 ESC guidelines. Patients underwent cardiac magnetic resonance imaging prior to invasive hemodynamic characterisation including sequential biventricular PV loop analysis during preload reduction and handgrip exercise. HFpEF patients were grouped according to LV hemodynamic phenotypes into RS cohort (change in indexed end-diastolic volume, ∆EDVi >0ml/m²) and US cohort (∆EDVi ≤0 ml/m²). Results A total of 37 HFpEF patients (n=20 RS and n=17 US, median age 71 years [68, 78], 70% female) were included. Clinical presentation did not differ between cohorts regarding age (p=0.22), functional NYHA class (p=0.52) and NTproBNP (p=0.57). In PV loop analysis, diastolic stiffness (EDP/EDVi ratio) was higher in the US cohort as compared to the RS cohort in both the LV (0.31 vs. 0.23 mmHg/ml, p<0.01) and the RV (0.14 vs. 0.10 mmHg/ml, p=0.03). This difference was even more pronounced during handgrip exercise (LV: 0.45 vs. 0.29 mmHg/ml/m², p<0.01; RV: 0.18 vs. 0.12 mmHg/ml/m², p=0.02). Similarly, exertional systolic stiffness was increased in US cohort compared to the RS cohort in both the LV (V120i 21.1 vs. 30.6 ml/m², p=0.04) and RV (V40i 27.8 vs. 40.5 ml/m², p<0.01). LV and RV volumes were closely correlated at rest (EDVi r=0.77, p<0.01; end-systolic volume, ESVi r=0.54, p<0.01; SVi r=0.95 p<0.01). The US of the LV PV relations could not be explained by ventricular interaction since exercise-induced increases (∆) in RV-EDVi did not correlate with increases in LV diastolic stiffness (∆LV-EDP/EDVi, p=0.58). However, the change in LV afterload from baseline to peak exercise (∆LV-Ea) did correlate with ∆LV-EDP/EDVi (r=0.61, p<0.01). A similar correlation was also found for RV afterload and RV diastolic stiffness (r=0.50, p<0.01). Conclusion Biventricular hemodynamic profiles closely resemble one another. The pathological LV US of the PV relations in response to exercise could not be explained by ventricular interdependence but was rather related to excess of biventricular afterload. Therefore, these findings indicate HFpEF to be a biventricular disease, which should be incorporated in future considerations about pathomechansims and potential therapeutic targets.Schematic biventricular PV-Loops