To our knowledge, this is the first study to evaluate and confirm the direct effects of SGLT2i on myocardial contractility using LMS in a biomimetic system. The addition of empagliflozin to LMS from end-stage HF patients increased total CD, resulting from an increase in the duration of both the systolic and diastolic phase of the contraction, without diminishing maximum force production or altering force displacement.

SGLT2i have been developed to treat hyperglycemia in patients with type II diabetes mellitus, which act by inhibiting glucose reabsorption in the proximal tubule of the kidney [10]. Ensuing studies have highlighted important effects outside the kidney, suggesting potential new pharmacological targets for HF patients [1,2,3,4,5,6]. Of note, it is important to make a clear distinction between specific myocardial effects and systemic effects that alter cardiac function when investigating the mechanistic properties of SGLT2i [17]. In-vivo studies are therefore, despite their great value in multidirectional analyses, less suited to study the direct effects of SGLT2i on myocardial contraction. LMS in an advanced setup, like this biomimetic system, allow for an isolated and controlled setting [27] to assess the biomechanical effects of SGLT2i. In this study, culture conditions were equal before and after drug administration and each LMS served as its own control, which is why the effects of empagliflozin addition are likely to be a direct effect of empagliflozin on the myocardium.

In our study, both contraction and relaxation duration were prolonged while maintaining maximum force generation per beat, consequently seen in all patients regardless of underlying etiology of heart failure. In order to comprehend these results, a number of biomechanical parameters need to be addressed. LMS in culture show isotonic contractions where the same tension is maintained while the muscle slice shortens. As presented in Fig. 2, there is a shift and alteration of the form of the curve due to the increase in time needed to contract (TTP) and an even longer time needed to relax (TTR), yet, without altering the Fmax or dF/dt nor a change in the AUC. In other words, the effect seen can best be described as a more gradual build-up and off of force over time, and thus more efficient handling of contraction and relaxation kinetics.

Furthermore, the observed prolongation of CD appeared irrespective of the type of underlying cardiomyopathy, although the sample size was low and this was not statistically tested. The longer median CD of left ventricular LMS compared to right ventricular LMS is possibly explained by micro-architectural and physiological differences between the left and right ventricle originating from dissimilarities in their development, anatomy, and function [32]. The fact that 7 LMS from different patients did not react to empagliflozin administration whilst exhibiting a normal contractile profile might be based on inter-layer variability of the myocardium with LMS production.

The observed shift in diastolic function corroborates with findings from previous studies [17, 21]. Pabel et al. showed a ~24.2% decrease in diastolic tension of human and murine twitching ventricular trabeculae after empagliflozin administration, while diastolic tension remained unaffected in controls treated with DMSO [17]. The systolic force did not change in their study, being similar to our results. However, in our study, the constant maximum contractile force was accompanied by a 7% increase in TTP. Yet, Azam et al. also assessed the direct effect of empagliflozin on cardiac contractility of Langendorff-perfused rabbit hearts subjected to global ischemia–reperfusion and showed improved left ventricular developed pressure without altering left ventricular end-diastolic pressure [33]. The differences between their study and our study might be explained by the different nature of both studies, as they worked with a rabbit model of ischemia–reperfusion and we used human tissue from patients with end-stage HF. Pabel et al. showed that the observed contractile effects can be attributed to a reduction in the myofilament stiffness of cardiomyocytes due to increased phosphorylation of myofilament regulatory proteins [17]. These results were confirmed by in-vivo echocardiographic measurements in rats, showing a shortened isovolumetric relaxation time and increased E/A diastolic filling velocities ratio [17]. Similar results with regard to diastolic function were obtained by animal studies from other groups [21, 34, 35] and in-human magnetic resonance imaging trials [22]. Our study, together with the study from Pabel et al. [17], shows that these effects are independent of systemic effects that may indirectly influence cardiac contractility, since the study was performed on isolated myocardium.

Two important mechanisms are present in the heart to modulate its ability to increase strength and rate of contraction, based on the β-adrenergic pathway and length-dependent activation [36]. Our study consistently showed no effect of empagliflozin on force generation or maximum force attained throughout all patient samples. This implies that secondary empagliflozin effects are not related to the β-adrenergic pathway, as increases in contractile force would then be expected. However, this was not assessed directly in this study. Second, myofilament properties play a central role in cardiac relaxation where protein kinase A (PKA)-mediated myofilament phosphorylation affects length-dependent activation, and the decline of intracellular Ca2+ is necessary [36]. Phosphorylation of cardiomyocyte binding protein C has been shown to play a vital role in cardiac diastolic function [37] and recent studies on molecular docking suggested glucose transporters 1 and 4 as potential binding target sites for empagliflozin [38]. This could potentially explain the altered biomechanical kinetics presented in our study where direct binding of empagliflozin with those glucose transporters restores the coupling between glycolysis and oxidative phosphorylation [39]. Nevertheless, such a restored coupling would also show a resultant enhanced Ca2+ transient with increased contractility, a feature that was absent in our study. Hence, varying results have been reported on the contractile effects of empagliflozin which could be attributed to the model used. Up-to-date, most studies either used isolated cardiomyocytes thereby lacking three-dimensional microarchitecture and native extracellular matrix, or animal models sometimes lacking extrapolation to the human setting. In addition, studies that used cardiac tissue biopsies [17, 19] mainly depended on isometric twitching of papillary muscles where papillary muscles are known to poorly project ventricular free wall kinetics and isometric testing does not reflect near-physiological human cardiac function [27].

As such, in our study, we opted to use human end-stage heart failure patient-specific tissue for biomimetic electromechanical stimulation with physiological preload that allowed for constant monitoring of the contractile capacity which provided crucial information related to tissue function in real-time. LMS can be kept in culture for several weeks [28], while still presenting a high degree of in-vivo representativeness [27]. Biochemical analysis of the culture medium and slices will hopefully contribute to a better understanding of the molecular mechanisms underlying SGLT2i therapy for patients with HF in future studies.

The current study presents some limitations. Firstly, it assessed the direct effect of empagliflozin on cardiac contractility based on mechanical force measurements, but without molecular analyses. This would be of added value for future studies. Secondly, the effects of DMSO cannot be completely ruled out, since control experiments were not performed. Yet, the final DMSO concentrations were very low. Thirdly, the use of different drugs in individual patients prior to surgery was not taken into account, which could have influenced the contractile response to empagliflozin administration in LMS. However, the setup used the same LMS as controls, and effects were seen in all patients. Furthermore, given the extended handling of LMS (wash-out in Tyrode buffer and culture medium) and the half-time of most HF medications, it is unlikely that receptors would still have been blocked by prior medication. Fourthly, the majority of tissue in this study was obtained from non-diabetic patients (n = 6/7), which does not allow for comparison between diabetic and non-diabetic patients. Yet, although only n = 1, the increase in CD was also observed in the LMS from the patient with diabetes mellitus. Lastly, no patients with heart failure and preserved ejection fraction (HFpEF) were present in this study and the question rises whether the presented effects could be extrapolated to this population. Yet, the inclusion of patients with HFpEF is very difficult since they do not undergo cardiac surgery very often.

Data from this study indicates that SGLT2i directly affect cardiac contractility, apart from volume regulation, cardiorenal mechanisms, and metabolic effects, aiding our understanding of the observed beneficial effects in diabetic and non-diabetic patients with HF.

Recently, it has been demonstrated that a shortened systolic ejection time (SET) is independently associated with an increased risk of cardiovascular morbidity and mortality in patients with HF [40, 41]. This supports a potential role for normalizing cardiac time intervals in patients with HF. In the current study, total contraction duration was prolonged after empagliflozin administration, including time-to-peak, which represents the systolic phase of the cardiac contraction. Hence, our data suggest that SGLT2i can directly improve SET intervals, which could correlate to the significant reductions in hospitalizations and mortality as seen in clinical studies of SGLT2i therapy. Yet, more in-vivo imaging studies are needed to confirm whether empagliflozin and other SGLT2i indeed improve cardiac time intervals as suggested by the current study.

In conclusion, contraction duration of isolated LMS from end-stage HF patients increased after the addition of empagliflozin without diminishing maximum force production. Here, we present convincing evidence that SGLT2i directly act on the myocardium, independent of renal and other systemic influences. Yet, the exact pathway behind this cardioprotective mechanism needs to be investigated in future studies.