Introduction

Heart failure represents the end result of different pathophysiologic processes, which culminate in functional impairment. Regardless of its aetiology, the presentation of heart failure usually involves symptoms of pump failure and congestion, which forms the basis for clinical diagnosis [1]. Contemporary practice has categorised heart failure into several distinct groups: acute vs. chronic, left sided vs. right sided and preserved vs. reduced ejection fraction. During acute presentations, the schema of ‘wet-dry’ and ‘warm-cold’ often aids in appropriate management. However, unlike the dichotomous nature of classification systems, clinical manifestations of heart failure can be subtle and lie on a spectrum between asymptomatic dysfunction to severe end-stage disease. Nevertheless, the distinction between different classes of heart failure is important as they represent divergent pathophysiology and treatment. Heart failure with preserved ejection fraction (HFpEF) remains a poorly understood entity, with a lack of effective therapy, reflecting our lack of insight into the disease process. Increasingly, HFpEF is being viewed not as a primary cardiac disease but as a disease driven by complex and heterogeneous comorbidities [2].

On the other hand, pathophysiologic descriptions of heart failure with reduced ejection fraction (HFrEF) are better established [3]. Most commonly, HFrEF is centred on a reactive model where a significant initial insult leads to reduced cardiac output, further triggering a cascade of maladaptive processes [4, 5]. Predisposing factors include myocardial injury of any cause, chronically abnormal loading due to hypertension, valvular disease or tachyarrhythmias. The pathophysiologic processes behind remodelling in heart failure are complex and reflect systemic neurohormonal activation, peripheral vascular effects and localised changes affecting the cardiac substrate. These abnormalities have been the subject of intense research.

Much of the translational successes in HFrEF have come from targeting neurohormonal responses to reduced cardiac output, with blockade of the renin-angiotensin-aldosterone system (RAAS) and beta-adrenergic blockade being particularly fruitful. However, mortality and morbidity associated with heart failure remains high [6]. Although systemic neurohormonal blockade slows disease progression, localised ventricular remodelling still adversely affects contractile function.

Historically, a direct approach to stimulating cardiac contraction through the use of inotropic agents was employed. However, an association with poor clinical outcomes lead to re-evaluation in their routine use. Mechanistically, multiple aspects of the cardiac contractile apparatus are dysregulated in HFrEF and subject to adverse re-modelling. Bluntly increasing contractile force via influencing cardiac calcium levels did not address the underlying derangement in electrochemical coupling and energetic inefficiencies that are now well established. Novel therapy targeted at improving cardiac contractile mechanics and metabolic efficiency in HFrEF hold the promise of alleviating heart failure at its source, yet so far none has found success, i.e. been incorporated into current heart failure management guidelines [7, 8]. Nevertheless, there are increasing calls for a proximal, ‘cardiocentric’ approach to therapy [5, 9, 10]. In this review, we summarise translational progress aimed at improving cardiac pump function, focusing particularly on excitation-contraction coupling, novel inotropes and cardiac energetics.

Excitation contraction coupling

The mechanical action of the heart depends on the coordinated action of individual cardiomyocytes, which are composed mainly of contractile proteins and mitochondria. Myofibrils mainly contain actin and myosin and span the entire length of the cardiomyocyte. They are subdivided into individual sarcomeric elements identifiable by electron microscopy. The actin filaments are anchored at Z-lines found on either end of each sarcomere, whilst myosin filaments in the A band are interconnected at the central M line and to the Z-disc by the titin filaments. Invaginations of sarcolemma in the form a complex network of transverse tubules (T-tubules) are also anchored to the Z-discs. The T-tubules are rich in ion channels and serve to regulate excitation contraction coupling within the myocyte.

Contraction begins with an action potential that causes Ca2+ release from voltage gated L type Ca2+ channels (LTCCs) in the sarcolemma and within the T-tubules. The resulting increase in Ca2+ concentration triggers Ca2+-induced Ca2+ release (CICR) from ryanodine receptors (RyR) located on the closely apposed sarcoplasmic reticulum (SR). The propagating Ca2+ binds to cardiac troponin C and induces conformational changes between tropomyosin and actin on the thin filament. This exposes the myosin-binding sites on actin, enabling myosin to bind to it, thus activating the cross-bridge cycle (systole). Cardiac relaxation (diastole) requires the active uptake of Ca2+ into the SR through sarcoplasmic reticulum Ca2+ ATPase 2a (SERCA-2a) and also active Ca2+ efflux through the sodium-calcium exchanger (NCX) [Fig. 1]. Significant derangements in excitation-contraction (EC) coupling are found in HFrEF and correspond with abnormalities in the systems involved in Ca2+ handling. These include remodelling of the T-tubules, decoupling of RyRs and reduced activity of SERCA. The outcomes correspond to reduced amplitudes of the Ca2+ transients and reduced SR Ca2+ concentration, which negatively impact on both systolic contractility and diastolic relaxation [11]. Importantly, elevation of intracellular Ca2+ and Ca2+ leak from the SR predisposes to malignant arrhythmias and myocyte death [12].

Fig. 1
figure 1

Excitation contraction coupling in the cardiomyocyte. The cardiac action potential [1] is initiated by influx of Na+ ions via Na+ channels which brings the net cellular potential from negative to positive. This triggers Ca2+ release from LTCC [2] found within the invagination of the T-tubule. The localised elevation of Ca2+ concentration induces CICR from the closely opposed SR via RyR receptors [3]. Contraction [4] and relaxation [5] of myofilaments are dependent on binding and dissociation of Ca2+ from troponin. Diastole is therefore dependent on efflux of Ca2+ to either the SR via SERCA or extracellularly via NCX. Return to net negative cellular potential occurs during repolarisation and involves efflux of K+ ions via K+ channels and NKX. LTCC, L-type Ca2+ channels; CICR, calcium induced calcium release; SR, sarcoplasmic reticulum; RyR, ryanodine receptor; SERCA, sarcoplasmic reticulum Ca2+ ATPase; NCX, sodium-calcium exchanger; NKX, sodium-potassium exchanger

Full size image

T-tubule and ryanodine receptor dyad

In animal and (some) human studies, T-tubule disruption is a feature of established heart failure and hampers cardiac contractility and synchrony [13,14,15,16,17]. Impaired intracellular Ca2+ cycling is evident long before the development of clinical heart failure, coinciding with T-tubule disorganisation [18]. Migration of the T-tubules away from the Z-line alters LTCC distribution and leaves behind ‘orphaned RyRs’ [19], which exhibit ‘leaky’ SR Ca2+ release. This leads to asynchronous and regionally heterogeneous CICR, which correlates strongly with poor contractility in human HFrEF [20] and is postulated to be arrhythmogenic [21]. There is probably a linear reduction in ejection fraction with increasing disorganisation [22].

The underlying cause of T-tubule remodelling has not been defined, but in small animal studies, it has been linked to reduced expression and altered distribution of the anchor protein junctophilin-2 (JPH-2), possibly in response to increased wall stress [23, 24]. The T-tubule/SR triad is held together by JPH-2, which is essential for stabilisation of local ion channels including RyR and LTCC [25]. Cardiac-specific knock-down of JPH-2 in a mouse model led to the development of acute heart failure, with loss of the junctional membrane complexes and reduced CICR [26]. However, contradictory studies exist which show no correlation of JPH-2 expression with T-tubule distribution in both rats and sheep [27]. Whilst JPH-2 mutations has been found in individuals with hypertrophic cardiomyopathy (HCM), they remain at a case study level without a definitive link to HFrEF [28].

Recovery of tubule morphology has been reported in animal models with cardiac resynchronisation [29], sildenafil [30] and beta-blockade [31]; however, the efficacy of these treatments is based on unloading of the myocardium. Novel therapeutic approaches include gene therapy and miRNA inhibition. Overexpression of JPH-2 via viral-mediated gene therapy halted the progression of heart failure in a mouse model, with preservation of T-tubule structures [32]. Suppression of miRNA-24, which is known to target JPH-2, rescued LTCC signalling and stabilised JPH-2 expression in a mouse model of heart failure secondary to aortic constriction. It has been recently shown that JPH-2 fragments found in stressed hearts localise to the nucleus and activate cardioprotective transcriptional programming, attenuating hypertrophic remodelling in mice [33]. However, given the uncertain role of JPH-2 in human HFrEF, preclinical studies using human heart failure samples will be necessary prior to clinical trials.

SERCA

The Ca2+ re-uptake channel SERCA-2a has long been implicated in the pathogenesis of heart failure and has been the target of several clinical trials in HFrEF [34]. Normally, SERCA is the principal determinant of the rate of Ca2+ efflux, determining SR levels. Hence, SERCA activity greatly impacts on both the rate of diastole and the subsequent force of systolic contraction induced by Ca2+ release from the SR. SERCA is required for complete diastolic relaxation and minimises delayed after-depolarisations. The primary regulator of SERCA is phospholamban (PLB), which reduces the activity of SERCA by reducing its affinity for Ca2+. Phosphorylation of PLB by protein kinase A (PKA) or Ca-dependent protein kinase II (CaMKII) leads to disinhibition of SERCA resulting in both increased lusitropy and inotropy.

Animal and human studies in heart failure have variously confirmed the reduced expression of SERCA-2a at mRNA and protein levels [34,35,36]. SERCA activity is reduced by a relative decrease in SERCA/PLB levels and by reduced phosphorylation of PLB [37, 38]. SERCA KO mice exhibit inefficient Ca2+ handling, reduced contractile efficiency and eventual heart failure [39]. These and other experiments highlight the causal relationship between SERCA loss and heart failure, identifying it as a target for therapy.

PLB ablation improves EC coupling and Ca2+ handling in small and large animal models of heart failure, a result which has been replicated in humans [40,41,42]. A recent murine model of heart failure demonstrated reduced mortality following PLB ablation [43]. Modulation of PLB function via increased phosphorylation (inhibition of protein phosphatase 1 or enhancing PKA activity) has also been studied with some success in animals [44]. However, there are significant concerns with downstream effects on other signalling pathways and non-specific induction or inhibition of phosphorylation may have unforeseen outcomes. Importantly, mutations in PLB have been linked to hereditary dilated cardiomyopathy [45] and arrhythmogenic cardiomyopathy [46]. Multiple binding partners underlie the complexity of PLB signalling, and more preclinical work on human models is needed to translate these findings [47].

The direct SERCA activator, istaroxime, was investigated in the HORIZON-HF trial for acute heart failure (AHF). Istaroxime is a novel non-glycoside inhibitor of Na+/K+ ATPase, which also directly stimulates SERCA-2a activity, likely by disrupting the SERCA-PLB complex [48]. It exerts a dual inotropic/lusitropic effect by increasing SR Ca2+ sequestration in diastole and increasing cytosolic Ca2+ in systole. Increased SERCA activity had been demonstrated in guinea pigs and human cardiomyocyte preparations [49]. Istaroxime improved systolic and diastolic functions in a canine chronic heart failure model without increased myocardial oxygen consumption or an increase in heart rate [50]. In the HORIZON-HF trial randomising 120 AHF patients to istaroxime infusion or placebo, istaroxime decreased pulmonary capillary wedge pressure and diastolic stiffness whilst improving contractility without an increase in adverse events. Compared with other commonly used inotropes, istaroxime has a better safety profile than digoxin and does not increase energy consumption as is the case with dobutamine [51]. Further phases I and II trials of istaroxime in AHF are underway (ClinicalTrials.gov Identifier NCT02617446 and NCT02477449).

SERCA gene therapy has garnered much attention in preclinical research with extensive modelling both in vitro and in vivo. Transgenic murine models overexpressing SERCA demonstrate improved cardiac function and protection against developing heart failure [52]. Human cardiomyocytes from HFrEF patients [53] transfected with SERCA-2a carrying adenovirus have improved contractile characteristics in vitro [53]. Multiple small animal models provide proof of concept for gene therapy, with reports of restoration of contractile function in heart failure [54], reversal of negative remodelling [55], protection against malignant arrhythmias [56] and improved survival [57]. Large animal studies of canine, ovine and porcine models followed suit, with promising outcomes despite heterogeneity in the choice of the heart failure model [58]. Proposed mechanisms to explain the beneficial effect of SERCA gene therapy include improvement in mechano-energetic efficiency [59], modulation of apoptotic signalling and more recently altered miRNA expression [60]. Evidence for SERCA gene therapy has been described as ‘overwhelming’, years prior to the initiation of its first human trial [61].

The phase 2 CUPID trial enrolled 39 patients with advanced heart failure and utilised intracoronary delivery of SERCA-2a via adeno-associated virus (AAV) vector. The authors reported promising results across multiple domains to a prespecified p value < 0.2, assessing the likelihood of a false-positive result to be 2.7% [62]. A 3-year follow-up of CUPID participants reported an 82% reduction in recurrent cardiovascular events and a favourable safety profile [63]. Nevertheless, as is often the case for promising heart failure therapies, the phase 2b CUPID II trial in 250 patients failed to meet its primary endpoint of time to recurrent cardiovascular events, nor its secondary outcome of time to all-cause of death [64]. Whilst no substantial differences in patient characteristics exist between the two trials to explain these results, the study authors point to a possible reduction in viral transduction efficacy in the CUPID II trial. However, significant weaknesses exist in the design of the CUPID trial, including less stringent p values, subdivision of an already small sample size and the lack of dose response to treatment, all of which bias to a false-positive result. Following the disappointing result of the CUPID II trial, enthusiasm for SERCA gene therapy waned, with withdrawal of the companion phase II AGENT-HF trial examining its effect on ventricular remodelling [65]. Preliminary data favoured placebo, although the study at termination was too underpowered to detect a difference between the two arms. Despite the generally favourable safety profile of gene therapy for HFrEF, challenges remain in optimising its delivery and establishing its efficacy [66].

Post-translational regulation of SERCA2a is now recognised. Glutathionylation increases SERCA activity, whereas glycosylation decreases it [67]. Specifically, O-GlcNAcylation refers to the addition of O-linked N-acetylglucosamine and has been shown to reduce SERCA expression [68]. SUMOylation of SERCA2a has gained the lion’s share of attention in heart failure research and has been identified as essential for normal cardiac development [69]. SUMOylation refers to attachment of small ubiquitin-like modifier proteins which modify the function of the targeted protein [70]. SUMO1 gene transfer rescues decreased levels of SUMO1, increases SERCA2a expression and potentiates contractile function in mice with heart failure [71]. This result was replicated in a swine model of ischaemic heart failure [72]. A small molecule activator of SUMOylation has been discovered, which acts as a dose-dependent inotrope and improves left ventricular function in vivo [73]. Better elucidation of Ca2+ handling and its effect on cardiac function continues to supplement an ever-expanding menu of treatment options. Given its essential role not only in EC coupling but also its critical function in gene transcription [74], mechano-energetics, growth and apoptosis [11], it is likely that Ca2+ signalling will remain an active area of translational research.

Inotropes in heart failure

Inotropes were first introduced into clinical practice over 200 years ago with William Withering’s treatise ‘An Account of the Foxglove’, where digitalis was used to treat the ‘most hopeless and deplorable’ cases of heart failure [75]. Due to its significant symptomatic benefit in acute heart failure, digoxin use gained widespread acceptance but was plagued by significant toxicity. The search for a ‘digitalis replacement’ in heart failure took centre stage in the 1980s, with the development of adrenergic agonists and phosphodiesterase inhibitors as novel inotropes [76]. It soon became apparent, however, that their clinical benefit over digoxin was overstated.

Theoretically, augmenting systolic function should blunt maladaptive hormonal responses that are associated with remodelling. However, inotrope use was soon linked to increased arrhythmias, myocardial ischaemia and mortality. Current international clinical guidelines do not endorse the use of inotropes in chronic stable heart failure but advocate for its limited application to acute decompensated heart failure or as a bridge to destination therapy [1]. However, novel inotropes continue to be developed, which may improve our understanding of the pathophysiology of heart failure.

Calcium mobilisers

Medications that increase inotropy by elevating cytosolic Ca2+ as ‘calcium mobilisers’ include digoxin, milrinone and dobutamine [77]. Dobutamine is an adrenergic agonist and milrinone is a phosphodiesterase inhibitor. Both increase contractility by increasing cAMP whereas digoxin acts by blocking the Na+/K+ ATPase pump, thereby reducing the chemical gradient for Ca2+ efflux through NCX.

Neurohormonal hyperactivity in heart failure leads to desensitisation and down-regulation of adrenoceptors, reduction in cAMP levels and ultimately depleted intracellular Ca2+, all proportional to the severity of heart failure [78]. Medications that enhance cAMP rationally target the reduced cAMP levels found in heart failure. Whilst calcium mobilisers provide modest short-term haemodynamic benefit [79, 80], chronic administration leads to further loss of contractile reserve [81]. There was a 28% increase in all-cause mortality in the PROMISE trial of milrinone in severe chronic heart failure [82], a result echoed in studies involving dobutamine [83]. Undesirable pathophysiologic effects of increased cytosolic Ca2+ include increased myocardial oxygen demand, ventricular arrhythmias, reduced diastolic relaxation and acceleration of myocyte death [84]. Methods to circumvent these adverse effects such as reduced or intermittent dosing and the use of partial agonists have all failed to demonstrate a clinical benefit in HFrEF [85]. Digoxin itself did not improve all-cause mortality in heart failure as reported in the Digitalis Investigator Group Study [86]. Many began questioning the wisdom of stimulating rather than resting the failing heart, and long-term use of inotropes in stable HFrEF was abandoned soon after.

Intravenous inotropes are still used for acute heart failure (AHF) refractory to vasodilators and diuretics and/or accompanied by hypotension. Calcium mobilisers improve haemodynamics in the short-term, although evidence suggests that even brief use may lead to increased mortality. The prospective OPTIME-CHF trial of milrinone failed to show a clinical benefit in AHF, with higher rates of hypotension and ventricular arrhythmias [87]. Retrospective analysis of the FIRST trial identified dobutamine infusions as a strong independent risk factor for mortality [88]. Recent systematic reviews of dobutamine [89] and milrinone [90] continue to point to possible harm, although the reported trials tended to be small and of poor methodological quality, often utilising surrogate outcomes with significant clinical heterogeneity. Contrary with the mortality neutral results of the DIG trial, systematic review of nine large trials of digoxin in heart failure yielded a hazard ratio of 1.14 for all-cause mortality [91]. This has placed doubt on the role of the most established calcium mobilising agent. Given that large observational studies such as the ADHERE registry [92] echo the concern in controlled trials regarding the use of calcium mobilisers in acute heart failure, larger, better designed trials are needed to conclusively settle this question.

Calcium sensitisers

Calcium sensitisers enhance myocyte contractility by increasing myofilament Ca2+ sensitivity. They are thought to be superior to calcium mobilisers as they do not elevate cytosolic Ca2+ levels, in addition to avoiding down-regulation of the adrenergic signalling pathway in heart failure. The best studied calcium sensitisers are pimobendan and levosimendan, both of which augment troponin C binding to Ca2+.

In vitro animal studies of pimobendan using muscle preparations, demonstrated superior mechano-energetic efficiency compared with dobutamine [93]. In vivo, pimobendan exerts positive inotropic, lusitropic and vasodilatory effects [94]. Double-blinded RCTs in canine-dilated cardiomyopathy models suggested a strong mortality benefit, similar to those previously been observed with angiotensin converting enzyme (ACE) inhibitors [95]. However, this benefit never translated in the human trials. Like other inotropic agents of its time, pimobendan improved exercise tolerance at the expense of increased mortality [96]. Use of pimobendan in humans was subsequently abandoned, despite subsequent contradictory studies [97].

Levosimendan is routinely used in clinical practice, demonstrating several pharmacologic effects that act in concert to improve myocardial function [98]. Inotropy is sustained via stabilisation of the Ca2+-troponin C complex [99], with possible contribution from highly selective PDE III inhibition. Levosimendan has prominent vasodilatory effects brought about through activation of ATP-dependent potassium channels (KATP), which leads to relaxation of smooth muscles and after-load reduction. Cardioprotection is mediated by improvement of coronary flow and opening of mitochondrial KATP, which may reduce ischaemia-related cell damage [100]. Advantages of levosimendan over other agents include its potency despite β-blockade, lack of propensity for tachyphlaxis and sustained duration of action due to active metabolites [77].

Preclinical and clinical models of levosimendan indicate a beneficial effect in heart failure. Preservation of myocardial contractile efficiency despite inotropy has been demonstrated in guinea pig hearts [101] and human volunteers [102]. In vivo, levosimendan reduces infarct size in rat and pig models of LAD ligation [103, 104]. Open-label human studies of levosimendan in acute heart failure secondary to myocardial infarction suggest preservation of contractile function, improvements in coronary perfusion and mortality [105]. Double-blinded RCTs of levosimendan have been performed with inconsistent results reflecting heterogeneous study designs. The LIDO study of 203 patients with severe heart failure found levosimendan to be superior to dobutamine with haemodynamic improvement and lower 6-month mortality [106]. The RUSSLAN study randomised 504 patients with acute heart failure post-myocardial infarction to levosimendan or placebo, with safety being the primary outcome. Again, levosimendan correlated with lower mortality without an increase in hypotension or ischaemia [107].

However, larger follow-up trials SURVIVE [108] and REVIVE-II failed to demonstrate superiority of levosimendan in terms of safety and mortality. A meta-analysis of 45 trials involving levosimendan found a significant reduction in mortality despite the negative mortality benefit in the two largest studies [109]. However, the authors excluded singular studies at a time in their sensitivity analysis, which is not as rigorous as the exclusion of all low quality or unblinded studies [110]. Only smaller studies exist for the intermittent use of levosimendan in the chronic heart failure setting, such as in the LevoREP study [111]. No difference was found in the primary outcome of functional capacity or quality of life [112].

The jury is still out on whether calcium sensitisers are beneficial in both acute and chronic heart failure. However, there seems to be increasing clinical equipoise for larger trials of levosimendan therapy [113]. It remains controversial whether the inotropic effects of levosimendan can be partially or fully explained by phosphodiesterase inhibition [99, 114, 115]. Pleotropic effects of levosimendan extend to multiple organ systems, improving circulation in pulmonary, hepatic and renal vasculature, whilst protecting against reperfusion injury [116]. More research is needed to further elucidate the role of calcium sensitisers at both preclinical and clinical levels in heart failure.

Direct myosin activators

The actin-myosin cycle is where chemical energy is converted to mechanical energy in myocytes. Improvements in contractility can be gained by pharmacologically targeting actin-myosin kinetics. The steps of the cycle are well described. Hydrolysis of ATP bound to the myosin head allows weak interaction between actin and myosin, which is strengthened by the release of phosphate. Conformational change of the myosin head follows the release of phosphate and leads to a power stroke of roughly 10 nm on the actin fibres. The conversion from a weak to a strong bond between actin and myosin, also termed cross-bridge formation, is the rate-limiting step of this cycle. If the release of ADP and phosphate occurs before cross-bridge formation, no power stroke occurs. Direct myosin activators were first discovered via high-throughput screening of agents catalysing cross-bridge formation without increasing cytosolic Ca2+ concentrations [117]. Of the many myosin and actin modulators currently being investigated [118], the standout agent is omecamtiv mecarbyl (OM), a selective activator of cardiac myosin [119].

Cardiomyocyte preparations exposed to OM exhibit an increased duration of contraction without an acceleration of contraction velocity. This inotropic effect has been confirmed in a rat model of heart failure and also canine models with both tachycardia and ischaemia induced heart failure [117]. The investigators report a 20–30% increase in contractile efficiency. Early human studies were consistent with animal modelling, whereby measures of left ventricular function improved with administration of OM in a dose-dependent manner in healthy volunteers and patients with chronic heart failure. [120, 121]. Increases in systolic duration and stroke volume were noted without an increase in heart rate.

The ATOMIC-HF and COSMIC-HF trials were phase 2b trials published in 2016 that investigated OM in acute and chronic heart failure, respectively. The former did not meet the primary endpoint of relieving dyspnoea nor any of the secondary endpoints including length of hospitalisation and short term mortality [122]. Nevertheless, the authors point to a possible benefit in a subgroup using the highest OM concentration (425 ± 173 ng/mL).

Oral administration of OM over 20 weeks produced the more encouraging result of increased stroke volume and a reduction of NT-proBNP [123]. The main safety concern with OM is ischaemia due to a reduction in the duration of diastole in healthy individuals, which emerge at serum concentrations greater than 1200 ng/mL [120]. However, in patients with ischaemic cardiomyopathy, administration of OM at standard doses did not result in signs and symptoms of ischaemia with exercise induced stress [124]. Despite being generally well-tolerated with no increase in adverse event rates, both phase 2b studies associated OM administration with a small, concentration independent increase in troponin levels. It is likely that dosing of OM will need to be within a narrow therapeutic range.

The underlying mechanisms of action of OM may be more complex than first thought. The specificity of OM for cardiac myosin is under question due to a study reporting an increase in slow twitch skeletal muscle fibre contractility with OM [125]. More recently, the idea that OM increases contractile efficiency has been challenged, with reports of impaired contractile efficiency and increased myocardial demand in porcine models, attributed to increased resting myosin ATPase activity [126]. In the same paper, it was suggested that OM may also have calcium sensitising effects due to improved interaction between the thick and thin filaments. Regardless of these lingering questions, enthusiasm for myosin activators remains high. GALACTIC-HF is a large phase III trial of OM in chronic HFrEF currently under way, which seeks to answer the question of whether supporting cardiac contractility with direct myosin activators will lead to reduced cardiovascular mortality (ClinicalTrials.gov Identifier: NCT02929329) (Table 1).

Table 1 Selected trials of pharmacologics affecting myocardial contractility in heart failure
Full size table

Energetics in heart failure

Consideration of myocardial energetics is essential for any therapy aimed at improving contractile function. Under normal circumstances, oxidative metabolism in the mitochondria supplies 95% of cardiac energy with 5% derived from anaerobic glycolysis [133]. In terms of substrate utilisation, fatty acid oxidation (FAO) accounts for 70% of cardiac metabolic requirements, with the remainder being derived from oxidation of glucose, ketone bodies, lactate and other amino acids [134]. Substrate metabolism in the heart responds to the physiologic environment, hence fatty acid oxidation is suppressed in presence of excess glucose and vice versa [135]. The majority of energy consumption is for maintenance of EC coupling, including cross-bridge cycling and powering ion fluxes. The efficient turnover of metabolic substrates is therefore a prerequisite for normal contractile function and energetic deficits are strongly linked to both cellular oxidative stress and contractile failure.

Heart failure represents impairment of energetic reserve and substrate utilisation. This is corroborated by declining stores of both ATP and phosphocreatine (PCr) with progressive left ventricular dysfunction in failing human hearts [136, 137]. Even in the absence of lower concentrations of adenosine nucleotides, reduced shuttling of the creatine kinase system contributes to failure to meet energy demands [138]. In particular, a reduction in PCr/ATP ratio is predictive of adverse outcomes in heart failure [139]. Modification of substrate use is seen in both cardiac hypertrophy and heart failure, with increasing dependence on glucose metabolism and reduced FAO. The physiological impetus for this change is not a reduction in availability of fatty acids but from alterations in transcription signalling [140]. Peroxisome proliferator-activated receptor-⍺ (PPAR-⍺) and its co-activator peroxisome proliferator-activated receptor gamma .co-activator-1 (PGC-1) induce fatty acid metabolism and mitochondrial biogenesis. They are both down-regulated in human heart failure [141, 142]. Although oxidative metabolism of glucose is more efficient than FAO, there is uncoupling of glucose oxidation from glycolysis in heart failure [134, 143]. Products of glycolysis such as pyruvate and lactate can be channelled to the Krebs cycle via accessory or ‘anaplerotic’ pathways. However, efflux of substrate and loss to other non-productive pathways such as protein glycosylation and polyol formation results in reduced oxidative efficiency, as observed in animal models of heart failure [143, 144]. Given the paradigm of energetic exhaustion in the failing heart and the trepidation for the current use of inotropes (‘flog a dead horse’), it is plausible that enhancing cardiac energetics may improve contractile function.

Metabolic substrate modulation

An approach to improve energetic balance in the failing heart is to focus on inducing glucose oxidation. Amongst the investigated agents, anti-anginals target the imbalance in myocardial oxygen supply and demand, with perhexiline and trimetazidine demonstrating promising results. Trimetazidine promotes glucose oxidation by competitively inhibiting long chain 3-ketoacyl-coenzyme A thiolase (3-KAT), the enzyme responsible for the last step in beta-oxidation of FA. Administration of trimetazidine to heart failure patients positively impacted on cardiac PCr/ATP ratio with additional beneficial effects including improved endothelial function, restoration of cytosolic Ca2+ levels and protection against free radical injury and fibrosis [145]. Recent interest in extending the use of trimetazidine to heart failure led to a series of small trials of variable methodological quality, which demonstrated an improvement in LVEF and NYHA class. A systematic review and meta-analysis performed by Gao et al. in 2011 was significant for an improvement in cardiovascular events, hospitalisation and overall mortality [146]. The results are echoed by a more recent retrospective cohort study [147]. However, larger double-blinded RCTs are needed to clarify the benefits of trimetazidine in HFrEF.

Perhexiline is an anti-anginal agent, which inhibits carnitine palmitoyl transferase-1 (CPT-1) responsible for the transport of FA into mitochondria. Evidence for use of perhexiline in heart failure lies mainly in one trial conducted by Lee et al., which demonstrated improvements in contractile function and heart failure symptoms in patients with advanced heart failure [148]. A more recent study of perhexiline in patients with non-ischaemic heart failure demonstrated improvements in PCr/ATP ratio and patient symptoms, without evidence for change in LVEF or substrate utilisation [149]. This is in agreement with an earlier murine study by the same group [150]. At present, multiple mechanisms have been proposed for perhexiline without consensus [151]. Concerns with hepatotoxicity and neurotoxicity places an increased burden of proof for benefit in heart failure before perhexiline can be used for this indication [152].

Ranolazine is an inhibitor of the late Na+ channel current with expanding roles in management of angina, cardiac arrhythmias and diastolic dysfunction [153]. Perhaps reflective of the complexities in extrapolating animal studies to human heart failure, earlier suggestions of enhanced glucose oxidation by ranolazine in murine models did not translate to humans [154]. Agents that directly increase glucose oxidation have also been explored. Dichloroacetate (DCA) inactivates mitochondrial pyruvate dehydrogenase kinase to release inhibition of pyruvate dehydrogenase, the rate-limiting enzyme, which converts pyruvate into acetylCoA for entry into the Krebs cycle. Early clinical experience with DCA suggested an improvement in cardiac function and efficiency [155], but this was not always replicated [156]. Poor pharmacokinetic properties of DCA also make it unsuitable for chronic use, despite benefits found in animal models [157].

The benefit of repressing FAO to induce glucose oxidation has been questioned. FA remains the most important metabolic substrate in heart failure, and acute depletion of FA levels actually impairs contractile function [158]. Inhibition of FAO may also lead to accumulation of lipid metabolites, which are a reversible cause of contractile dysfunction and leads to structural damage in a process termed cardiac lipotoxicity [159]. A newer approach to managing heart failure energetics has been to induce FAO, which in addition to boosting supply of ATP, may clear FA derivatives that accumulate in heart failure. Induction of PPAR-⍺, the main regulator of FAO, can have positive effects on the contractility of the failing heart. Overexpression of PPAR-⍺ under normal circumstances can repress glucose utilisation, resulting in a diabetic cardiomyopathy in murine models [160]. However, in the context of heart failure, PPAR-⍺ activation preserves the level of high-energy phosphates and in particular, exerts a cardioprotective effect in vivo [161, 162]. Subgroup analyses of the VA-HIT trial of fibrates, which act via PPAR-⍺ agonism, suggest a benefit in heart failure [163]. However, this may reflect enhanced endothelial function and anti-inflammatory effects of PPAR-⍺ agonists rather than the promotion of FA metabolism [164]. In fact, fibrates may reduce cardiac FAO by inducing FA consumption in the periphery [165]. It remains to be demonstrated that enhancement of cardiac FAO occurs with fibrates in humans as no RCTs have directly proven their benefit in heart failure [166].

Restoring mitochondrial activity

Defects in multiple domains of mitochondrial function occur in heart failure and result in abnormal ion handling, oxidative stress and programmed cell death amongst other pathophysiologic effects [134]. Directly relevant to the issue of energetics is the disruption of the mitochondria electron transport chain (ETC), noted in both animal and human studies of heart failure [167]. The ETC facilitates transformation of free energy released from oxidative reactions within the mitochondria into ATP through the action of a series of enzyme complexes. Decreased expression of mitochondrial complexes has been described in failing human hearts of multiple aetiologies [168]. On the other hand, mitochondrial myopathies are also associated with contractile dysfunction and defective oxidation [169]. Inefficient ETC function additionally increases reactive oxygen species (ROS), which results in myocyte apoptosis, pathological remodelling and progressive cardiac contractile dysfunction [170]. Modulation of ETC function is seen with proven therapies in HFrEF, including cardiac resynchronisation and neprilysin inhibition [171]. Direct modulation of mitochondrial ETC activity may be the target for future treatments in heart failure.

Coenzyme Q10 (CoQ10) is an over-the-counter supplement with limited but promising data for HFrEF treatment. CoQ10 primarily participates in the ETC as an electron shuttle, facilitating the production of ATP. However, it also has anti-oxidant and endothelial protective effects [172]. Reduced levels of CoQ10 are found in HFrEF and correlate with lower LVEF and increased mortality [173]. In agreement with animal studies, clinical trials of CoQ10 supplementation have shown improvements in LVEF and the recent Q-SYMBIO study found a 50% relative reduction in major cardiovascular events and all-cause mortality. Nevertheless, significant weaknesses exist with the existing trials, including small patient numbers, heterogeneous study populations and protocols and large margins of error for outcome measures.

Other agents that target respiratory chain function include flavonoids and melatonin. Flavonoids refer to naturally occurring pigments found in a variety of plants whose consumption has been associated with reduced incidence of heart failure. Members of the flavonoid family have been shown to improve ETC activity and it has been proposed for long term preventative therapy for heart failure [174]. In animal models, melatonin stabilises the inner mitochondrial membrane leading to improved ETC function, whilst also exerting anti-oxidant effects [175]. Both classes of compounds have limited evidence in human heart failure and require further study.

Although there is consensus that alterations to ETC expression and function occur in HFrEF, the reported defects are highly variable between different heart failure models and human heart failure of differing aetiologies [167]. ETC phosphorylation status, respiratory complex assembly and regional distribution of mitochondrial defects are also increasingly recognised as important aspects of a complex metabolic network. The fundamental question of whether heart failure progression is a direct result of metabolic dysfunction or whether energetic remodelling is a consequence of adapting to a cardiac insult remains unanswered. A recent study of failing vs. non-failing human hearts found no difference in oxidative capacity for fatty acid or glucose [176] compared with controls; whilst a separate study on fresh human heart failure samples found preserved in vitro oxidative capacity of cardiac mitochondria [177]. The complex physiology of cardiac metabolism has proven difficult to capture with current experimental models, especially those based on animals.

Towards human models of heart failure

Cardiac contractility relies on complex and multifaceted physiology, which is modulated at the level of ion fluxes, myofibrillar interaction and energetics. As can be seen by our brief exploration of current therapies (summarised in Fig. 2), many gaps remain in our understanding of the pathophysiology of human heart failure. In adopting a cardiocentric approach to heart failure, we continue to rely on fragmented approaches, which are often incompletely characterised. Manipulation of one aspect of the system may simultaneously be counteracted by another or have unintended consequences, and the increased risk of arrhythmias with calcium mobilising inotropes is a case in point. Even in the case of OM, which was rationally screened and designed as a cardiac selective inotrope, it also has pleotropic effects.

Fig. 2
figure 2

Summary diagram of pharmacologic agents modulating cardiomyocyte function and their molecular targets. Contractile function can be enhanced via optimisation of excitation contraction coupling through influencing the activity of SERCA and Junctophilin-2. Calcium-mobilising inotropes increase contractility by increasing cytosolic calcium either via action of mediators such as cAMP or directly acting on ion channels and exchangers. Calcium-sensitising inotropes increase sensitivity to the action of calcium whereas myofilament activators directly increase cross-bridge cycling. Contractile efficiency may be appropriately targeted via modulating the metabolic substrate as in the case of the anti-anginal agents or through manipulation of the electron transport chain within mitochondria

Full size image

There is a great discrepancy between successful animal studies and lacklustre performance in larger human trials. Although this can be partly attributed to small sample sizes and biases in publication and study design [178], even in methodologically robust animal trials, results lack external validity when applied to humans. Preclinical animal studies often fail to replicate the complex patient phenotype. On a macroscopic level, animal models generally do not capture the progressive and degenerative nature of human disease nor are they able to replicate the multiple comorbidities commonly found in the HFrEF population. Significant differences on a genetic level are further confounded by heterogeneity in the results of inter- and even intraspecies studies [179].

Research utilising human tissue can help shed light on the intertwining factors contributing to a poorly contractile heart without needing to rely on rough disease models. A vogue area of research that may prove insightful is that of cardiac recovery, specifically recovery of failing human hearts requiring mechanical unloading via left ventricular assist devices (LVADs). It is now well established that following a period of support and pharmacological therapy, significant structural reverse remodelling occurs which allow successful explant of the supporting device in a minority (1.3%) of patients [180]. Importantly, success has been reported not just for acute cardiomyopathy but also for chronic heart failure. Whilst clinical research has emphasised on establishing robust clinical criteria to identify those most suitable for device explantation, availability of human cardiac tissue at inplantation, explantation and transplantation has proven a boon for understanding underlying pathophysiology and its reversibility. With LVAD support, remodelling occurs across multiple domains, with improvement in Ca2+ homeostasis, T-tubular structure, mitochondrial function and sarcomeric contraction [181]. Nevertheless, there remains a tendency for patient phenotype to relapse, and current mechanical and pharmacologic support may be better described to provide remission rather than full recovery [182, 183]. Currently, supportive pharmacotherapy to enable weaning of mechanical support relies heavily on systemic neurohormonal approach such as RAAS blockade and beta-blockers—although some localised beneficial effects on the myocardium may also exist [184]. To achieve durable recovery, targeting aetiology specific findings within the cardiac substrate will likely be necessary.

In an era of increasingly personalised medicine, specific, human relevant molecular targets are more important than ever. Although fresh human cardiac tissue samples such as those utilised in the cardiac recovery studies represent the gold standard for in vitro research, there is generally very limited availability. The key to improving access will be the development of reliable, collaborative tissue banking [185]. Whether at the scale of elucidating transcriptional modification in heart failure or in large clinical trials of therapeutics, ultimately, all animal studies require validation with human studies.