Heart failure in diabetes

Abstract

Heart failure and cardiovascular disorders represent the leading cause of death in diabetic patients. Here we present a systematic review of the main mechanisms underlying the development of diabetic cardiomyopathy. We also provide an excursus on the relative contribution of cardiomyocytes, fibroblasts, endothelial and smooth muscle cells to the pathophysiology of heart failure in diabetes. After having described the preclinical tools currently available to dissect the mechanisms of this complex disease, we conclude with a section on the most recent updates of the literature on clinical management.

1. Introduction

The heart represents an organ with a high metabolic activity, consuming >400 kcal/kg/day at rest. The processes of excitation-contraction coupling within cardiomyocytes heavily rely on ATP-dependent reactions: approximately 60–70% of ATP hydrolysis in cardiomyocytes is spent on the contractile shortening and most of the remaining ATP fuels the functioning of sarco/endoplasmic reticulum Ca²⁺-ATPase 2 (SERCA) and other ion pumps [1]. A thorough evaluation of ATP, ADP and Pi concentrations in vivo revealed that their levels remain mostly unchanged throughout the cardiac cycle and during physiological intensifications in cardiac work and oxygen consumption [2], suggesting that an increase in ATP production is the main mechanism responsible for fulfilling cardiac energy demands. Nearly 95% of ATP in cardiomyocytes is produced by oxidative phosphorylation in mitochondria. This process is fueled by NADH and to a lesser extent by succinate, both produced in the tricarboxylic acid cycle (TCA, also known as citric acid cycle or Krebs cycle) with acetyl coenzyme A (acetyl-CoA) being indispensable for keeping the TCA cycle running [3,4]. Three main sources of acetyl-CoA production exist in the cell: β-oxidation of fatty acids, decarboxylation of pyruvate provided by glycolysis, and ketone bodies oxidation [5,6].

A fundamental feature of cardiac metabolism is its flexibility; indeed, the heart is able to switch between different substrates to fulfill its energetic needs with the nutrients that are more available at the moment [7]. The adult heart under non-ischemic conditions produces 60% to 90% of acetyl-CoA by means of fatty acid oxidation and only 10–40% by pyruvate carboxylation [8]. Production of ketone bodies (e.g. β-hydroxybutyrate, BHB) is quite low in normal conditions and their contribution to myocardial bioenergetics has been estimated to be <10% [9-11]. It is important to note that the heart ability to store energy substrates in form of glycogen and triglycerides is limited, hence the increase in heart work relies on its ability to increase the uptake and utilization of glycogen or fatty acids [12-14].

2. Diabetes mellitus (DM) impairs cardiac metabolic flexibility

Heart disease remains the leading cause of death for both Type 1 and Type 2 diabetes mellitus (T1DM and T2DM) patients [15]. In diabetic patients, the availability and the usage of substrates by the heart is drastically disturbed (Fig. 1). Impaired insulin signaling causes both hyperglycemia and hyperlipidemia due to the decreased glucose uptake by the skeletal muscle and increased free FA secretion by the adipose tissue [16-22]. Despite the excessive supply of both types of energy substrates, the overall contribution of glycolysis to cardiac ATP production is drastically decreased and heart reliance on fatty acid oxidation becomes higher [23-28]. The mechanisms underlying this shift are complex and multifaceted [7].

Fig. 1. Myocardial bioenergetics in the healthy and in the diabetic heart.

6-P-gluconolactone: 6-phospho-gluconolactone; Acetyl-CoA: acetyl coenzyme A; ADP: Adenosine diphosphate; ATP: Adenosine triphosphate; CD36/FAT: Cluster of differentiation 36/fatty acid translocase; CPT1: Carnitine palmitoyl transferase I; CPT2: Carnitine palmitoyl transferase II; ETC: Electron transport chain; Fructose-6-P: fructose-6-phosphate; G6PDH: Glucose-6-phosphate dehydrogenase; GAPDH: Glyceraldehyde 3-phosphate dehydrogenase; GFAT: Glutamine fructose-6-phosphate aminotransferase; Glucosamine-6-P: glucosamine-6-phosphate; Glucose-6-P: glucose-6-phosphate; GLUT1: Glucose transporter type 1; GLUT4: Glucose transporter type 4; GPI: Glucose-6-phosphate isomerase; HBP: hexosamine biosynthetic pathway; HK2: hexokinase 2; IMM: inner mitochondrial membrane; NAD: Nicotinamide adenine dinucleotide; NOX: NADPH oxidase; O-GlcNAcylation: O-linked-N-acetylglucosaminylation; OGT: O-linked N-acetylglucosaminyltransferase; OMM: outer mitochondrial membrane; PDH: pyruvate dehydrogenase; RNS: reactive nitrogen species; ROS: reactive oxygen species; TCA: tricarboxylic acid cycle; UDP-GlcNAc: Uridine diphosphate N-acetylglucosamine.

In T1DM, the main mechanisms plausibly relate to the vanishing of insulin signaling. Insulin promotes cardiac glucose uptake inducing glucose transporter 4 (GLUT4) expression and translocation to cell membrane [29-32], and GLUT4 abundance is markedly diminished in T1DM [33]. Insulin signaling prevents degradation of phosphofructokinase 2 (PFK2), an enzyme that leads to the generation of fructose-2,6-biphosphate (F-2,6-BP), a potent activator of glycolysis [34-36]. Insulin signaling also removes inhibition of pyruvate dehydrogenase (PDH), which converts the pyruvate into the TCA substrate Acetyl-CoA [25,37]. So, insulin signaling has a complex role in regulation of glycolysis and its loss diminishes glucose uptake, glucose conversion into pyruvate, and pyruvate incorporation into TCA. Moreover, insulin regulates cardiomyocyte usage of fatty acid. Insulin-dependent downregulation of Forkhead box protein O1 (FOXO1) dampens fatty acid cellular and mitochondrial uptake [38,39], while insulin-dependent phosphorylation of protein kinase B (AKT) and protein kinase C δ (PKCδ) downregulates the oxidation of fatty acids [25,37,40,41]. All these data indicate that insulin plays an imperative role in determining the substrate choice in the heart.

The exact mechanisms underlying the substrate switch in T2DM remain mostly unclear, as the development of cardiac insulin resistance in T2DM is quite elusive [42]. A minimal cardiac responsiveness to insulin has been shown in numerous in vivo assays [28,43-45] and a diminished cardiac glucose uptake in T2DM patients was reported in a clinical trial [46]. Insulin treatment was shown to result in reductions in fatty oxidation, measured using positron emission tomography (PET), in both control subjects and T2DM patients but the achieved rates of myocardial fatty oxidation remained elevated in T2DM compared with controls [24]. Another clinical study demonstrated that insulin effectively improved heart bioenergetics in T2DM patients [47].

The metabolic switch observed in T2DM may be explained by the inhibitory action of fatty acids accumulation on glycolysis. As depicted in Fig. 1, cardiomyocytes become heavily overloaded with fatty acids in T2DM [48-52]. Animal studies point on peroxisome proliferator-activated receptors (PPAR) α and γ as culprits of the metabolic switch [51-55]. PPARα and PPARγ are directly activated by fatty acids and then upregulate the expression of pyruvate dehydrogenase kinase 4 (PDK4), which inhibits PDH, thus preventing the oxidizing of pyruvate [52,55]. Since data from T2DM human hearts confirmed the upregulation of PDK4 but argued against PPAR activation [56], further research is required in this direction. An alternative mechanism of fatty acid dependent downregulation of glycolysis could be the hyperacetylation of its key enzymes [26,57].

2.1. Interplay between dysregulated glycolysis and accessory metabolic pathways

Despite the downregulation of glucose transporters in the diabetic heart, glucose uptake is not fully abrogated; besides, the multistep process of glucose conversion into pyruvate is retarded. This situation eventually results in the accumulation of intermediate products of glycolysis feeding accessory metabolic pathways (Fig. 1).

One example of these processes is the activation of the hexosamine biosynthetic pathway (HBP) in the diabetic heart [58], which most likely is triggered by the loss of insulin-dependent stimulation of PFK1 activity and the subsequent retarded conversion of fructose-6-phospahte (F-6-P) into F-1,6-BP. F-6-P also serves as precursor for glucosamine synthesis, an initial step of HBP [59]. In the HBP, uridine diphosphate N-acetylglucosamine (UDP-GlcNAc) is produced, which is further involved in intracellular signaling as a substrate for O-linked N-acetylglucosamine transferases (OGTs). OGTs are responsible for O-GlcNAc posttranslational modifications on a number of proteins involved in cardiac function [58]. For instance the massive protein hyper-O-GlcNAcylation was observed in experimental models of DM [60,61].

Another consequence of dysbalanced glycolysis is the decreased NAD⁺/NADH ratio, which may be attributable to an excessive availability of glyceraldehyde-3-phosphate, converted into 1,3-bisphophoglycerate with concomitant reduction of NAD⁺ to NADH [62]. In normal conditions, NAD⁺ pull is replenished by the activity of lactate dehydrogenase making lactate from pyruvate; however, under conditions of lactic acidosis, the oversupply of lactate slows down this reaction [63]. NAD⁺ levels are of great importance as NAD⁺ serves as cofactor for a class of lysin deacetylases called sirtuins [64]. The decrease in NAD⁺ dampens their activity resulting in decreased activity of sirtuins [50,65-67] and subsequent augmentation of protein acetylation [28,68].

Glycolysis retardation may also result in glucose-6-phosphate (G-6-P) accumulation [69]. Besides being utilized in glycolysis, G-6-P is also converted into 6-phosphogluconolactone (6-PGL) by glucose-6-phosphate dehydrogenase (G6PDH) to be further metabolized in the pentose phosphate pathway [70-73]. Unlike glycolysis and glucose aerobic oxidation, the pentose phosphate pathway does not lead to the production of adenosine 5′-triphosphate (ATP) but supplies NADPH and ribose 5-phosphate (R5P). R5P is a building block for nucleic acid synthesis. NADPH provides the reducing power required for the synthesis of fatty acids, sterols, nucleotides and non-essential amino acids. NADPH also serves as substrate for NADPH-oxidases (NOXs), enzymes involved in the production of superoxide from oxygen, initiating a cascade of reactive oxygen species (ROS) and reactive nitrogen species (RNS) generation [74-79]. DM has been also shown to alter the esterified lipid biosynthesis (the glycerolipid biosynthetic pathway, a.k.a. Kennedy pathway [80]), a mechanism that can further contribute to diabetic cardiomyopathy [81-84].

2.2. Metabolic inefficiency of the diabetic heart

The shift in cardiac metabolism profoundly affects the function of the heart. Mounting evidence suggests that the contractile performance of the heart at a given level of oxygen consumption is greater when the heart is oxidizing more glucose and lactate, and less fatty acids [85-88]. A plausible explanation for this phenomenon is that fatty acids can mediate the uncoupling of mitochondria [89]; the mechanisms of fatty acids mediated uncoupling are based on the ability of fatty acids to cross the mitochondrial membrane not only via carnitine-palmitoyl-transferase, but also by a ‘flip-flop’ mechanism. In the mitochondrial matrix, fatty acids become deprotonated, thus ‘smuggling’ protons across the inner mitochondrial membrane and diminishing the proton gradient across it. Importantly, these ‘stowaway’ fatty acids cannot be utilized in β-oxidation, as they enter mitochondria in their unesterified form and do not contribute to energy production. Moreover they are exported out of mitochondria by uncoupling protein 3 (UCP3) and then can again enter mitochondria by a ‘flip-flop’ mechanism, creating an uncoupling circuit [90,91]. As a result, a significant portion of protons pumped across the mitochondrial membrane by electron transport chain enzymes is ‘wasted’ instead of being used for ATP production.

Taken together, these data indicate that cardiac efficacy of the diabetic myocardium is drastically decreased, meaning that for accomplishing the same work the diabetic heart requires much more oxygen compared to the healthy one [61]. Indeed, clinical studies demonstrated ATP deficit and poor oxygenation of the heart in T2DM patients subjected to exercise [92].

An exacerbation of the diabetic heart metabolic inefficiency is caused by protein hyperacetylation due to decreased NAD⁺/NADH ratio and subsequent deactivation of sirtuins, as mentioned above. Lysin acetylation plays essential roles in modulating the activity of a number of enzymes involved in fatty acid metabolism [93,94]. Indeed, mitochondria from diabetic hearts exhibit a worsened ability to utilize fatty acids for respiration compared to mitochondria from healthy hearts both in animal and human studies [68,95,96]. Restoration of NAD⁺/NADH ratio via dietary NAD⁺ supplementation or pharmacological activation of sirtuins improved both mitochondrial function and cardiac performance in animal models of DM [65,96].

2.3. Role of dysregulated metabolism in the functional changes observed in the diabetic heart

Experimental evidence indicates that glycolytic enzymes can form a subdomain adjacent to the endo/sarcoplasmic reticulum and plasma membrane [97,98]. Furthermore, ATP produced by glycolysis is primarily consumed by SERCA2 and plasmalemmal ion pumps removing Ca²⁺ from the cytosol [1]. Indeed, glycolysis inhibition does not affect the overall concentration of ATP in the cell, neither cardiac contraction [99-102]; nevertheless, left ventricular end-diastolic pressure was shown to be increased alongside an augmentation in Ca²⁺ and K⁺ concentrations [102,103]. Transgenic mice with PFK2 deficiency exhibited prolonged time of myocardial relaxation [35]. All these alterations compromise cardiomyocyte relaxation and provide mechanistical explanations for some hallmarks of diabetic cardiomyopathy, namely increased myocardial stiffness, impaired relaxation, and decreased left ventricular filling [104].

Metabolic dysregulation in diabetic heart may also contribute to impaired relaxation via fatty acid mediated uncoupling. In this sense, mitochondrial uncoupling following UCP2 overexpression was shown to inhibit Ca²⁺ uptake by mitochondria and to prolong cytosolic Ca²⁺ clearance [105]. Plausibly, fatty acid mediated uncoupling may have similar effects on mitochondrial Ca²⁺ handling.

The excessive O-GlcNAcylation of proteins within the cardiomyocyte, resulting from upregulated HBP, as discussed above, also contributes to the impaired myocardial function. Ergo, O-GlcNAcylation of Ca²⁺/calmodulin-dependent protein kinase II (CaMKII), phospholamban (PLN), L-type Ca²⁺ channels (Ca_v), and transcriptional factor specificity protein 1 (Sp1), was found in animal models of diabetic cardiomyopathy as well as in diabetic patients [60,106,107]. Supporting this view, a downregulation of HBP, providing a substrate for O-GlcNAcylation or ablation of OGTs genes, specifically responsible for this posttranslational modification, resulted in partial functional recovery of the heart [60,107].

3. Cellular and molecular impact of DM on the cardiovascular system

DM has profound effects on signaling pathways in different cell types of the cardiovascular system. For instance, insulin resistance is known to predispose patients to inflammation, increase mitochondrial dysfunction, and cause the activation of the renin-angiotensin-aldosterone system. Insulin resistance in diabetic patients reduces glucose uptake in peripheral tissues [108]. Dysregulation of cardiac insulin signaling has been shown to impair phosphoinositide-3-kinase (PI3K)-protein kinase B (Akt) pathway and to increase mTOR/S6K1 signaling. Impairment of PI3K-Akt signaling can suppress GLUT4 receptor, reducing glucose uptake in cardiomyocytes. Inactivation of PI3K-Akt decreases nitric oxide (NO) production; such a reduction in NO levels leads to the inhibition of cyclic guanosine monophosphate (cGMP), protein kinase G (PKG), and eventually cell death and cardiac dysfunction. All of these processes have a detrimental impact on myocardial energetics and eventually promote cardiac hypertrophy and interstitial fibrosis [109].

In the next paragraphs of this section, we will examine in detail the impact of DM on adrenergic signaling and the specific effects on cardiac fibroblasts cardiomyocytes, endothelial cells and vascular smooth muscle cells (VSMCs).

3.1. Effects of DM on cardiac adrenergic signaling

DM has been shown to affect both the expression and the responsiveness of the subtypes of beta adrenergic receptors (β-ARs), crucial in cardiac contractility [110]. A seminal study had demonstrated that the number of β-AR was decreased by 28% in the ventricular rat tissue 8 weeks after streptozotocin (STZ) injection [111]; such decrease in β-AR density in the heart was taken into account to explain the depressed contractile responsiveness to adrenergic stimuli in diabetic cardiomyopathy [112].

However, these findings were not confirmed by other investigators [113-117]; controversial results were also reported when using other animal models, like rabbits [118,119] and pigs [120,121]. These discrepancies could be attributed to the different expression of the three β-AR subtypes, namely β₁, β₂, and β₃. Indeed, subsequent investigations have demonstrated a reduction of β₁-AR [122-125] and an increase of β₃-AR [122,123,126,127], whereas results on β₂-AR remained controversial [122,124,127,128], most likely because of the different methods used to determine their expression (immunoblots, radioligand binding assays, RT-qPCR).

When exploring the β-AR signaling pathway instead of the mere β-AR expression, isoproterenol-stimulated PKA activity was shown to be decreased in T2DM mice and T1DM rats [129,130], although phosphorylation of PKA was increased when the duration of DM reached 12 weeks [131]. Alterations in PKA-dependent phosphorylation of proteins involved in excitation-contraction coupling have been evidenced in diabetic hearts, with increased phosphorylation of RyR expression at both Ser²⁸¹⁴ and Ser²⁸⁰⁸, which are known to enhance pathologic intracellular Ca²⁺ leak [132], in T1DM rats [133], and decreased phosphorylation of phospholamban in both T1DM and T2DM animal models [125,129,131,134]. Levels of GRK2, essential in cardiac β-AR desensitization [135-139], were found to be significantly increased in the left ventricle of 12-week-old db/db mice compared to age-matched non-diabetic littermates as well as in peripheral blood mononuclear cells (PBMCs) of T2DM patients compared to non-diabetic controls [140]. Consistent with these findings, inhibiting GRK2 was shown to have anti-inflammatory effects and to reduce cardiac fibrosis and oxidative stress [141,142].

3.2. Effects of DM on cardiomyocytes and cardiac fibroblasts

Hyperglycemia is one of the main mechanisms that can trigger cardiomyocyte apoptosis through means of the activation of the cytochrome c stimulated caspase-3 pathway [143]. Besides, hyperglycemia causes mitochondrial dysfunction, prompting an enhanced production of ROS, and restricts the ability of cardiomyocytes to use glucose as an energy source (Fig. 2), leading to increased FFA oxidation, whose uptake occurs by the cluster of differentiation 36 - fatty acid translocase (CD36/FAT) [144-147].

Fig. 2. Main molecular mechanisms underlying cardiac dysfunction in diabetes.

AGEs: Advanced glycation end products; Ang II: Angiotensin II; ASC: Apoptosis-associated speck-like protein containing a caspase-recruitment domain; AT1: Angiotensin II type 1 receptor; ATPase: Adenosine triphosphatase; [Ca²⁺]_i: Intracellular calcium; Casp-1: Caspase-1; CD36/FAT: Cluster of differentiation 36/fatty acid translocase; cGMP: Cyclic guanosine monophosphate; FFA: Free fatty acid; GLUT4: Glucose transporter type 4; ICAM-1: Intercellular adhesion molecule; IL-18: Interleukin 18; IL-1β: Interleukin 1 beta; IL-6: Interleukin 6; IR: Insulin receptor; MMPs: Matrix metalloproteinases; mTOR/S6K: Mechanistic target of rapamycin (mTOR)-ribosomal S6 kinase (S6K) pathway; NF-kB: Nuclear factor kappa-light-chain-enhancer of activated B cells; NLRP3: Nucleotide-binding oligomerization domain like receptor (NLR) pyrin domain containing 3; NO: Nitric oxide; PI3K/Akt: Phosphoinositide-3-kinase (PI3K)-protein kinase B (Akt); PKC: Protein kinase C; PKG: Protein kinase G; Pro-casp1: Pro-caspase-1; PTP: Permeability transition pore; ROS: Reactive oxygen species; SERCA2a: Sarcoplasmic reticulum Ca²⁺-ATPase 2a; TGF-β: Transforming Growth Factor beta; TNF-α: Tumor Necrosis Factor alpha; VCAM-1: Vascular cell adhesion molecule 1.

Fibrosis, defined as the excessive and/or inappropriate deposition of extracellular matrix proteins, is a typical characteristic of diabetic cardiomyopathy. ROS generation and oxidative stress have been specifically linked with the pathogenesis of cardiac fibrosis and cardiomyopathy [148,149] and several investigators have reported elevated oxidative stress in experimental models of T2DM [150,151]. Whereas during normal conditions a little amount of oxygen is transformed in ROS, in a diabetic status an excessive amount of ROS is generated. ROS can activate transcription factors such as nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), which can in turn regulate the transcription of multiple pro-inflammatory genes including Tumor Necrosis Factor α (TNF-α), Transforming Growth Factor β1 (TGF-β1), and Interleukins (IL-1β, IL-6, IL-18) that are known to be involved in the pathogenesis of DM-associated heart failure (HF) [152]. TNF-α stimulation stimulates collagen synthesis while IL-1β promotes the pro-inflammatory fibroblast phenotype [153].

Numerous studies have shown that oxidative stress induces cardiac fibrosis by stimulating TGF-β1 expression [154,155]. TGF-β1 acts as a central mediator in the pathogenesis of tissue fibrosis [148]. Hyperglycemia and hyperinsulinemia activate angiotensin II, TGF-β1/SMAD signaling, and elevate protein kinase C (PKC) activity in fibroblasts [156,157]. These processes in turn elicit interstitial collagen deposition and fibrosis, which is associated with increased expression of TGF-β1. TGF-β1 mediates myofibroblast trans-differentiation and promotes matrix preservation [158-160]. Oxidative stress is also triggered by chronic angiotensin II, which is linked to a pro-fibrogenic phenotype in the heart, and causes modifications in the extracellular matrix by reducing matrix metalloproteinases (MMPs) [161]. Additionally, NF-κB increases the expression of intercellular adhesion molecule (ICAM-1) and vascular cell adhesion molecule 1 (VCAM-1) [145].

NLRP3 inflammasome is a relatively novel molecular marker in diabetic cardiomyopathy [162]. Some of the factors that activate the NLRP3 inflammasome are high FFA levels, impaired insulin metabolic signaling and hyperglycemia. The NLRP3 inflammasome stimulates myocardial dysfunction through the activation of IL-1β. Activated caspase-1 cleaves IL-1β and IL-18 precursors and promotes multiple proinflammatory pathways involving NF-κB, chemokines, and ROS [163]. Furthermore, NF-κB increases NLRP3 inflammasome assembly and activates pro-caspase-1 [164,165].

In diabetic cardiomyopathy, reduced SERCA2 is associated with the decrease of recaptured Ca²⁺ and lowers the overall Ca²⁺ content in the sarcoplasmic reticulum [166]. Since cardiac contractility is controlled by changes in intracellular Ca²⁺ concentrations [1,167,168], reduced Ca²⁺ for the next contraction leads to an impaired cardiomyocyte contractility. In fact, some studies have shown that diastolic dysfunction is linked to an impaired myocardial Ca²⁺ handling [169,170].

Mitochondrial dysfunction is another strategic player in the development of diabetic cardiomyopathy and HF [75,171-174]. Generally, ~90% of intracellular ATP production in cardiomyocytes is generated by mitochondrial oxidative phosphorylation. However, in T2DM, mitochondria substitute glucose with FFA oxidation for ATP production which in turn causes increased production of mitochondrial ROS and reduced oxidative phosphorylation [7,145]. Altered mitochondrial Ca²⁺ handling further induces mitochondrial respiratory dysfunction and ultimately causes cell death [172,175].

3.3. Effects of DM on the endothelium

Both endothelial cells and VSMCs play key roles in controlling the myogenic tone of the vessels as well as blood pressure, fundamental aspects in the pathogenesis of HF [176,177].

Three decades ago, Torsten Deckert of the Steno Memorial Hospital of Copenhagen formulated the “Steno hypothesis”, the revolutionary idea, for that time, ascribing to endothelial dysfunction most of the clinical manifestations of DM [178]. Years later, the pivotal role of vascular endothelium in the development of microalbuminuria and hypertension associated to DM was confirmed [179,180]. It is clear now that if macrovascular complications in DM are similar to the processes occurring during atherosclerosis of non-diabetic origin, the microvascular alterations with a predominant participation of the endothelium represent a DM signature [181]. In fact, ultrastructural alterations of the glomerular endothelium have been detected in patients with T2DM before the development of microalbuminuria [182].

One of the main factors altered in the diabetic endothelium is permeability, and different mechanisms have been suggested as responsible for the increased endothelial permeability in DM [183-192]. VEGF signaling could be involved, as this factor is one of the main agents able to induce endothelial permeabilization [193]. Several studies focused on the assessment of VEGF expression in DM, however the results were controversial; data were ambiguous not only for the results on the overexpression or downregulation of VEGF in diabetic patients, but also concerning the efficacy of VEGF targeting as therapeutic strategy [194].

An additional mechanism suggested as responsible for the increased endothelial permeability is inflammation. Macrophages are able to secrete cathepsin-S, activating protease-activated receptor-2 (PAR2) on endothelial cells [195]. The interference with PAR-2 pathway determines modifications of endothelial barrier integrity, with increased microvascular permeability [196].

The impaired ability to induce vasorelaxation is another hallmark of the diabetic endothelium [197]. The relaxation response to endothelium-dependent vasodilators, including ACh and BK, is impaired in both the early phase and in the late stage of DM, and the responses to ACh are significantly improved following pre-treatment with indomethacin [197].

Mechanistically, the impaired endothelial dependent vasodilatation occurs in DM as a consequence of 1) decreased NO availability, 2) oxidative stress, 3) increased of harmful metabolites, and 4) inflammation.

Brodsky and colleagues demonstrated the relationship between high glucose levels and reduced NO bioavailability. Specifically, they verified that NO-dependent responses of endothelium to bradykinin were negatively affected by high glucose levels [198]. Mass spectrometry studies identified the formation of a covalent unitary addition of NO to glucose, so that glucose itself could act as scavenger for NO [199]. Moreover, hyperglycemia induces uncoupling of mitochondrial NOS (mtNOS) in endothelial cells, reducing the production of NO and favoring hydrogen peroxide generation [200]. Consistent with these findings, patients with T2DM and nephropathy, show decreased intravascular NO synthesis from L-arginine [201].

This switch from NO to superoxide production, alongside with the metabolic abnormalities of DM, may cause mitochondrial superoxide overproduction and oxidative stress in endothelial cells [202]. In turn, oxidative stress exacerbates the metabolic alterations and NO production, generating a vicious cycle. The use of a mitochondrial-targeted superoxide dismutase mimetic was able to correct the metabolic disarrangement of endothelial cells in diabetic rats, supporting the central role of oxidative stress in the development and progression of the disease [203].

The eNOS uncoupling alongside with superoxide overproduction can generate RNS, such as peroxynitrite (ONOO−), a toxic metabolite derived from NO, and superoxide anion (O2*−) [204]. The mechanisms by which RNS can modify vascular function are poorly understood. However, it is known that endothelial peroxynitrite accumulation induces nitration and inactivation of prostacyclin synthase (PGIS), which inhibits vasodilator, antithrombotic, and antiadhesive effects of prostacyclin (PGI2), increasing the release of a potent vasoconstrictor and prothrombotic factor like thromboxane A2 (TxA2) [205]. The functional contribution of peroxynitrite in the clinical manifestations in DM is supported by a study using diabetic rats with signs of peroxynitrite formation; the use of ebselen, a peroxynitrite scavenger, was able to prevent endothelial damage and nephropathy [203].

Another mechanism of endothelial damage in DM is mediated by Advanced Glycation End-Products (AGEs), which have been shown to induce premature endothelial senescence by activating the receptor for advanced glycation end products (RAGE) [206,207]. AGEs are derived from proteins or lipids that become glycated after prolonged exposure to aldose sugars [202,208,209]. They represent a sort of record of the cumulative prevalence and duration of the hyperglycemic status. An accelerated buildup of AGEs in both the intracellular and extracellular space mediates inflammation and cardiac fibrosis through multiple pathways. AGEs can trigger upregulation of the RAGE and increase oxidative stress [181]. AGE-RAGE interaction increases the production of matrix protein and connective tissue through the activation of Janus Kinase (JAK) and mitogen-activated protein kinase (MAPK) pathways. Additionally, activation of RAGE involves NF-αB and causes reduction of myocardial contractility during HF. Some of the actions of AGEs may be also mediated by angiotensin II [210] or by the inhibition of NO activity in the endothelium [145,211].

The effects of specific AGEs, such as glycosylated-collagen, on premature senescence of the endothelium could involve the sirtuins SIRT1, one of the most important factors in regulating cell survival and longevity [212]. SIRT1 expression was downregulated in diabetic mice [207]; glycosylated-collagen mediated cell apoptosis and senescence were attenuated by SIRT1 overexpression, while further SIRT1-downregulation enhanced endothelial cell death [213]. Furthermore, the exposure of endothelial cells to glycosylated-collagen induces lysosomal permeabilization [214-216]. This evidence can explain the mechanism of SIRT1 regulation by AGEs. Indeed, by promoting lysosomal permeabilization, glycosylated-collagen can induce cathepsins release from the organelle, as observed by the Goligorsky's research group [217]. Interestingly, cathepsins are able to cleave SIRT1, so this mechanism could be responsible for SIRT1 degradation in diabetic endothelium with increased senescence [207].

Endothelin (ET-1) is a potent vasoconstrictor secreted by vascular endothelial cells in response to specific cytokines, growth factors, thrombin, or hypoxia. ET-1 has also complex effects on the renin-angiotensin system. In diabetic mice, ET-1 is decisive in DM-associated vascular remodeling and promotes cardiac hypertrophy, which is also due to overstimulation of the IGF-1 receptor on cardiomyocytes [211,218,219].

3.4. Effects of DM on VSMCs

DM is an independent risk factor for atherosclerosis [220]. Indeed, in diabetic patients the phenomenon of macrovascular atherosclerosis is very frequent making them the most common subjects undergoing peripheral vascular and coronary revascularization procedures [221,222]. This association is mainly attributable to the direct impact of DM on the vasculature, affecting different compartments, from VSMCs to the endothelium. VSMCs play strategic roles in the pathophysiology of atherosclerosis. Specifically, a phenotypic switch of these cells is crucial for atheroma formation and progression [223]. In fact, the switch from contractile (differentiated, quiescent, non-migratory) to synthetic (dedifferentiated, proliferative, migratory) phenotypes typically supports plaque deposition and progression. The proatherogenic phenotype of VSMCs seems to be more aggressive in diabetic patients, as suggested by their increased intimal hyperplastic activity [224]. An elegant study by Faries and colleagues reported intrinsic differences in VSMCs obtained from diabetic patients compared to healthy control cells; specifically, VSMCs of diabetic origin displayed abnormal morphology and increased proliferation, migration, and adhesion rates [225]. These features are consistent with the increased rate of atherosclerotic events and the augmented risk of restenosis reported in diabetic patients [226] and, more recently, in non-diabetic patients with hyperglycemia [227]. Interestingly, those phenotypes were observed in vitro, using an isolated system without other cofactors that could have affected the vascular behavior in vivo [225]. Therefore, the intrinsic alterations of VSMC in diabetic patients should be considered also in designing new therapies.

The exact factors promoting the atherogenic transformation of diabetic VSMCs and the underling mechanisms are still mostlyunknown. Among these factors, microRNAs (miRNAs) produced by diabetic VSMCs have been recently proposed; miRNAs are a class of non-coding RNAs that can play vital roles in gene expression as well as cardiac remodeling. Increased oxidative stress downregulates some miRNAs including miR-1, miR-499, miR-133a, and miR-133b [228]. The up-regulation of miR-221 and miR-212 has been reported in hypertrophy and autophagic responses [145]. In a recent study, a high throughput RNA sequencing identified >100 miRNAs upregulated in VSMCs from db/db mice [229], an established animal model of DM (see below the section on preclinical models). In particular, miR-504 was highly expressed and new partners regulated by miR-504 in diabetic VSMCs were identified, like Egr2 and Grb10, molecules involved in gene transcription regulation and pro-proliferative pathways [229]. Conversely, miR-132 was significantly decreased in VSMCs from diabetic rats or after hyperglycemia exposure in vitro. Accordingly, its direct target E2F5 was upregulated, determining an increased migration and proliferation capacity of diabetic VSMCs [230]. Other miRNAs have been linked to the dysfunction of VSMCs behavior during disease development. miR-24 is able to regulate VSMCs phenotyping switch induced by high-glucose conditions [231]; the inhibitory effect of miR-24 on VSMCs proliferation and migration is due to its ability to target HMGB1-box, reducing NF-κB nuclear translocation and DNA binding and blocking proinflammatory signals like TNF-α and IL-6 production. Despite the extensive progress in understanding the molecular mechanisms that regulate VSMCs phenotypic plasticity, most of the story is still unknown. Lately, the involvement of epigenetic mechanisms has emerged in the reactivation of embryonic and stem-cell related genes [232].

One of the major factors regulating epigenetic phenomena during VSMCs phenotypic switch is KLF4, which interacts with phosphorylated Elk-1 (pElk-1) and histone deacetylase 2 (HDAC2) to repress the expression of contractile and cytoskeletal proteins [233,234]. In a similar way, KLF4 is able to induce the activation of pluripotency related genes like Sox2 and Oct4. Hyperglycemia has been shown to cause the down-regulation of KLF4 in VSMCs in vitro, and reduced KLF4 levels have been detected in arteries from diabetic patients and mice [235]. Interestingly, KLF4 downregulation seems to be mediated by miR-29c, whose expression is induced by Foxa2.

Alterations of Ca²⁺ signaling in VSMCs are elemental in the development and progression of HF and other vascular complications in DM [236]. A defective expression/activity of several partners belonging to the VSMC-Ca²⁺ signalosome, including Ca²⁺ channels, pumps, and regulators has been reported in DM [236]. For instance, both expression and distribution of the intracellular Ca²⁺ release channel ryanodine receptor (RyR) change in response to high glucose [237]. The different subcellular localization and clusterization of RyR may affect the entity of Ca²⁺ sparks which normally generate spontaneous transient outward currents (STOCs) by activating the big conductance Ca²⁺ sensitive K⁺ channels (BK) [238,239]. Ca²⁺ sparks and STOCs in VSMCs constitute the functional unit that regulates arterial tone; indeed, the generated STOCs reduce the membrane potential limiting Ca²⁺ influx through VDCCs, consequently diminishing cytosolic Ca²⁺ concentration and vasoconstriction. Losing the appropriate STOCs generation in DM due to redistribution of RyR can be responsible for a decline in vasorelaxation [237]. Consistent with this scenario, several experimental models of DM show reduced expression of RyR in VSMCs [240,241] and Ca²⁺ spark amplitude and duration were reduced in VSMCs freshly isolated from cerebral arteries of male db/db mice [241]. Interestingly, this behavior was reported in male mice but not in diabetic females, which exhibited an unchanged amplitude of spontaneous Ca²⁺ sparks [236]. These data are in line with the clinical observation of a higher occurrence of microvascular alterations in diabetic men [242]. Alongside the reduction in RyR expression, levels of IP3R, the other main intracellular Ca²⁺ release channel, are reduced as well in diabetic VSMCs [240] and in VSMCs chronically exposed to hyperglycemia [237]. IP3R Ca²⁺ signaling in DM seems to be also regulated at the protein-protein interaction level; specifically, the interaction of the receptor with anti-apoptotic proteins like Bcl-2 has been shown to have an instrumental role [243,244]. Indeed, IP3-induced Ca²⁺ transients were shown to be significantly larger in diabetic VSMCs due to enhanced IP3R excitability via Bcl2-IP3R protein interactions [245]. This phenomenon could have important implications in the diabetic VSMCs phenotype, especially in terms of proliferation/apoptosis balance.

Another aspect of vascular remodeling in diabetic patients is the perturbation of the balance between cell proliferation and cell death. VSMCs isolated from the internal mammary arteries of diabetic patients show resistance to apoptosis induced by several stimuli [246]. This phenomenon could be attributed to the over-expression of Bcl-2 in diabetic VSMCs, as its downregulation by RNA-silencing restored the sensitivity to pro-apoptotic signals. The different apoptotic/proliferative rate is exacerbated by high-glucose conditions, inducing activation of protein kinase C (PKC) [247]. Other evidence supports the impairment of the Fas/Fas-ligand pathway in DM, contributing to the down-regulation of apoptosis [248]. The altered balance between proliferation and apoptosis is considered decisive in determining the intima/media thickening in DM, with particular clinical relevance for T2DM patients [227,249-251].

Another point to take in consideration regarding VSMCs behavior in DM is the perturbation of insulin signaling. Physiologically, insulin can maintain VSMC quiescence and, on the other hand, can promote VSMC migration [252]. Indeed, insulin keeps VSMCs quiescent by activating PI3K pathway, while it promotes their migration via the activation of the MAPK pathway [252]. Impaired PI3 kinase signaling and intact MAPK signaling characterize insulin-resistance; therefore, in diabetic conditions insulin seems to lose its ability to maintain VSMCs quiescent, pathologically promoting their migration.

4. Animal models of diabetic cardiomyopathy

Due to the complex pathophysiology of the disease, research in DM relies on the use of in vivo models of the disease. Both genetically induced spontaneous and experimentally induced non-spontaneous DM models are commonly used as animal models in preclinical research on DM [253,254]. Various rodent models of DM that are useful for studying mechanisms and pharmacological therapies are currently available (Fig. 3). Undoubtedly, research on animal models has significantly expanded our knowledge of the molecular mechanisms responsible for diabetic cardiomyopathy.

Fig. 3. Animal models of diabetic cardiomyopathy.

4.1. High Fat Diet/Streptozotocin (HFD/STZ)-induced rodents

The high-fat diet/STZ (HFD/STZ)-treated rodents represent a widely used experimentally induced diabetic model [255,256]. The HFD/STZ model mimics the early or late stages of T2DM [256,257]. To develop the corresponding animal pathology model, high fat diet (HFD) is often fed to the mice to induce insulin resistance, followed by low-dose STZ injection to cause β-cell impairment without completely compromising insulin secretion, thereby mimicking the characteristics of T2DM [257-259]. Generally, HFD induces metabolic changes that are commonly observed in T2DM patients, such as development of obesity and decrease in insulin sensitivity [260,261]. Several studies have been conducted to establish the best HFD feeding time and the ideal STZ doses to induce the T2DM models [256,259,262]. The diabetic condition is assessed by routine tests such as food and water intake, body weight gain, glucose tolerance test (GTT), and insulin tolerance test (ITT) among others. The interaction of AGEs with RAGE causes progression of diabetic complications and renal injuries [263,264]. Primarily, HFD/STZ-induced model can be beneficial in evaluating the effectiveness of anti-diabetic and anti-obesity drugs in T2DM. More recently, the combination of HFD and sucrose (known as HFS) has been shown to be a reliable model to induce DM in small rodents [265-267].

4.2. OVE26 mouse

The OVE26 mouse was generated by Epstein and collaborators in 1989 on a FVB background [268,269]. These mice have lower insulin levels and elevated serum triglyceride levels. They exhibit severe early-onset T1DM and can survive for up to 2 years without insulin administration [270]. They develop severe hyperglycemia after they reach 2–3 weeks due to overexpression of calmodulin in pancreatic β cells [271-273].

Increased cardiac mitochondrial biogenesis and oxidative stress have been reported in these mice [274]. However, one of the studies performed on these rodents revealed that the contractility of cardiomyocytes isolated from diabetic OVE26 mice remained stable during culture and that OVE26 hearts in young animals have quite normal values of ejection fraction and fractional shortening [274]. More recent experimental assays have shown that the development of diabetic cardiomyopathy in these mice is both age and sex dependent and is mainly attributable to fibrosis, ROS, and inflammation [275].

This model is one of the simplest ones for studying DM-associated complications as it exerts direct effects on pancreatic β cells, displays early onset DM, which can last for up to 2 years without any insulin treatment [270].

4.3. The ob/ob mouse

The ob/ob mouse has a recessive mutation in leptin (Lep^ob) [276-279] and the severity of DM is dependent on the strain's genetic background. They are most used to model DM and obesity. The ob/ob mice exhibit mild DM with marked obesity, hyperphagia, hyperinsulinemia, and transient hyperglycemia [280,281].

Hyperglycemia develops in these mice between 8 and 15 weeks. The maximum blood glucose levels have been recorded when these mice reach 3–5 months; then, blood glucose starts decreasing and normalizes as they grow older. On the BL/6 inbred background, ob/ob mice develop cardiac hypertrophy and hyperplasia [281]. There is only mild or no impairment in systemic dysfunction in ob/ob mice, however, they have diastolic dysfunction, most likely due to cardiac lipid accumulation, which may trigger inflammation and oxidative stress, as also validated in clinical studies [282]; ob/ob mice exhibit increased left ventricular (LV) systolic pressure (LVSP) and LV end-diastolic pressure (LVEDP) as they get old; dP/dt is elevated in ob/ob mice at 4 weeks of age whereas it decreases when the mice are 8-weeks old [283].

Isolated perfused hearts from ob/ob mice exhibited decreased LV function [284]. Cardiac metabolism studies have shown that there is a reduction in glucose oxidation whereas the rates of fatty acid oxidation [285] and myocardial triglyceride storage [286] increase. Moreover, mitochondrial dysfunction and oxidative stress have been reported in ob/ob hearts [287].

4.4. The db/db mouse

The genetically diabetic-obese (db/db) mice are homozygous for a mutation in the leptin receptor (Lepr^db) [276] and are widely used to model T2DM and obesity. In this type of mice, leptin action is impaired due to the defect of the leptin receptor. In these mice, obesity starts at three to four weeks of age [288]. On the BL/Ks background, these rodents exhibit elevated levels of blood sugar, transient high plasma insulin levels, atrophy of insulin-producing pancreatic β-cells, peripheral neuropathy, and myocardial disease [280]. Hyperglycemia develops in db/db mice between 4 and 8 weeks of age [284].

Elevated serum concentrations of triglycerides in db/db mice have been described by various investigators [289,290]. The levels of triglycerides were shown to be ~2.2-times greater in db/db mice of 15 weeks old versus ob/ob mice of the same age [284]. Isolated perfused hearts from 10 to 14 weeks old db/db mice exhibited the characteristics of diabetic cardiomyopathy, including a decrease in contractile performance and alterations in cardiac metabolism [291,292]. A thorough evaluation by cardiac ultrasound evidenced an age-dependent diastolic dysfunction, recapitulating features of human HF with preserved ejection fraction (HFpEF), more pronounced in female mice [293]. Invasive assessment of LV function showed an increased LVSP, increased LVEDP as early as 4-weeks of age and an increase in dP/dt over time. The rates of fatty acid oxidation and myocardial lipid storage are increased in db/db hearts [284–286,294] alongside a reduction of glucose uptake and oxidation and an impaired mitochondrial function [295].

4.5. The Zucker fatty rat and Zucker diabetic fatty rat

Zucker fatty rats have a homozygous missense mutation (fatty, Fa) in the gene encoding for the rat leptin receptor (Ob-R). The mutation occurs in a simple autosomal recessive gene, Fa, that is common to all leptin receptor isoforms [296,297]. Zucker fatty rats are obese, hyperphagic, hyperinsulinemic, but are not often hyperglycemic. Henceforth, they are an excellent model for studying the long-term effects of obesity. They have increased serum levels of triglycerides, fatty acids, and insulin. These rats start becoming obese at 3–5 weeks and their body is composed of >40% lipid by the time they reach 14 weeks [298].

The Zucker diabetic fatty (ZFD) rat was derived by selective breeding of the Zucker fatty rats with high serum glucose levels to obtain a model with a DM trait. These rats exhibit obesity with DM and are most commonly used for T2DM research. Nonfunctioning leptin receptors is believed to be the main cause of their obesity [299]. ZDF rats are obese, hyperinsulinemic, hyperglycemic, hyperleptinemic, and have high levels of serum fatty acids and triglycerides [300,301]. Until they are 6 weeks old, ZDF rats are insulin resistant but euglycemic; they start to develop hyperglycemia at around 6 weeks of age and they usually develop stable hyperglycemia at 10–12 weeks, displaying reduced insulin levels as a result of pancreatic β-cell deficiency [302].

Oltman and colleagues [303] reported that the development of vascular and neural dysfunction was slow in obese Zucker rats as compared to the ZDF T2DM rat model; however, since ZDF rats are inbred, it is complicated to directly compare the obese and diabetic Zucker rats because of the differences in the genetic background [302]. Other studies demonstrated that there were negligible changes in substrate oxidation and irregularities in cardiac contractility in Zucker fatty rats, as compared to diabetic ZDF rats; Zucker fatty rats develop oxidative stress, cardiac hypertrophy, and interstitial fibrosis [304,305]. Various studies conducted in Zucker fatty rats have shown that there is a decrease in cardiac function in isolated hearts whereas there is an increased rate pressure-product [306,307]. However, cardiac efficiency remains mostly unchanged [294,307]. Young and collaborators demonstrated that there is a decrease in carbohydrate oxidation whereas oleate oxidation and expression of genes involved in fatty acid oxidation do not rise in Zucker fatty rats [307]. Studies carried out by Luiken and colleagues revealed that palmitate uptake and myocardial lipid content are increased in cardiomyocytes isolated from Zucker fatty rats [308].

ZFD rats are not as much obese than Zucker fatty rats but are more insulin resistant [309]. The male ZDF rat rapidly progresses to T2DM because of the inability to compensate adequately for insulin resistance. According to the data obtained from echocardiography, ZDF rats have a more depressed LV fractional shortening in comparison to ZF rats [310]. A reduced LV insulin-mediated glucose utilization in ZDF rats has also been reported [310]. ZDF rats exhibit left ventricular hypertrophy and increased myocardial lipid storage [311-313]. Moreover, in ZDF rats there is an increase in the rates of fatty acids oxidation and expression of genes involved in this process, whereas carbohydrate oxidation and Glut4 expression are both decreased [312-314].

5. The heart in DM: clinical insights

For years, population-based studies have evidenced a significantly increased risk of HF in diabetic patients unexplained by traditional risk factors (obesity, hypertension, age, coronary artery disease, dyslipidemia, and significant valvular disease) [315-322]. In fact, even after adjustment for these factors, diabetic patients have up to a two-times relative risk of developing HF [323,324]. This ventricular dysfunction is deemed diabetic cardiomyopathy and is most commonly associated with diastolic dysfunction and HFpEF — through left ventricular hypertrophy, AGE deposition, and interstitial fibrosis early on in its development. Of note, ~30%–40% of individuals with HFpEF have DM [325,326]. Albeit several parameters of diastolic function were reported to be similar in diabetic and non-diabetic HFpEF patients, a trend toward higher left atrial filling pressures was noted in HFpEF patients with DM [327]. Furthermore, increased cardiac fibrosis, cardiomyocyte hypertrophy, endothelial dysfunction, and AGE deposition, have been reported to be major components of HFpEF in DM patients [328,329]. Coronary microvascular inflammation with endothelial dysfunction can be considered a common denominator among HFpEF, DM, and obesity, that may explain the pathophysiology of HFpEF and its synergistic relationship with DM and obesity [320,330].

A principal involvement of endothelial cells in the cardiac remodeling process could represent the core mechanism underlying HFpEF, opposed to a main involvement of cardiomyocytes in HFrEF [330-333]

Diabetic cardiomyopathy is often asymptomatic in its early stages [334]. However, left atrial enlargement and higher end-diastolic pressure will eventually result from the thickened left ventricle and impaired diastolic filling. The prevalence of diastolic dysfunction is high in diabetic patients, reported to be 23%–75% in various studies, and increases markedly with age [335-337]. As the disease progresses, systolic dysfunction may arise in the form of worsening ventricular dilation through eccentric hypertrophy and HF with reduced ejection fraction (HFrEF, ejection fraction ≤ 40%) [338]. At this point, the classic signs of congestive HF should be observed, including pulmonary rales, orthopnea, fatigue, and nocturia. The mechanisms underlying the change in phenotype are not fully understood but, as mentioned above in this review, most likely involve impaired Ca²⁺ handling in the mitochondria and sarcoplasmic reticulum, insulin resistance, autonomic neuropathy, and endothelial/microvascular dysfunction — influenced in part by the patient's glycemic control [334,339]. It is interesting to note that data from the Framingham Heart Study revealed that diabetic patients develop greater age-associated increases in left ventricular wall thickness [340]; markers of insulin resistance, including augmented serum insulin levels, have been also shown to be associated with greater left ventricular mass and concentric remodeling [341,342].

Unfortunately, there are no established clinical signs or histological markers that can definitively identify diabetic cardiomyopathy [318,323]. This disease has a long subclinical phase in which major histological and functional changes may occur, but clinical manifestations have not developed. Therefore, identifying cases within asymptomatic individuals with DM could be particularly useful. A study conducted in 101 patients with diabetic cardiomyopathy showed that screening using stress echocardiography and Doppler is useful in identifying asymptomatic DM patients with left ventricular hypertrophy, whereas measuring hypertrophic markers like brain natriuretic peptide (BNP) was not sufficiently sensitive to diagnose subclinical disease [343]. 2D/3D-Speckle Tracking Echocardiography is a relatively new technique that shows promise in detecting mechanical deformation and can definitely help stratify diabetic patients [344,345]. Other investigators have demonstrated that both magnetic resonance and nuclear imaging — including PET and single-positron emission computed tomography (SPECT) — are powerful tools in predicting cardiovascular mortality in asymptomatic diabetic patients [346]. Yet, the Detection of Ischemia in Asymptomatic Diabetics (DIAD) study, a trial of 1123 diabetic patients, revealed a non-significant decrease in cardiac events with screening via adenosine-stress radionuclide myocardial perfusion imaging (MPI) — bringing into question the utility of these screening methods in the clinic [347].

More recently, non-coding RNAs have been sought out as potential diagnostic criteria for their role in the pathogenesis of diabetic cardiomyopathy (via hypertrophy, fibrosis, and oxidative stress) [348-350]. If properly validated, anti-miRNA and miRNA mimic therapies could be an exciting new avenue for treatment, since miRNAs represent druggable therapeutic targets. A recent study demonstrated that decreased levels of miR-21 could be used as a reliable diagnostic biomarker of diabetic cardiomyopathy; the same study revealed that overexpressing miR-21 could improve mitochondrial biogenesis (including increased expression of PGC-1α), a process that was impaired in db/db mice [351]. Cardiomyocyte-derived exosomes could also serve as a specific biomarker for diabetic cardiomyopathy. Exosomes are membrane-derived nanovesicles that serve a role in extracellular communication between different cells and tissues through shuttling cellular components like DNA, non-coding RNA, and heat-shock proteins [352,353]. A 2014 preclinical study demonstrated that cardiomyocyte-derived exosomes containing miR-320 help mediate myocardial vascular deficiency present in diabetic cardiomyopathy [354]. However, this fascinating cross-talk between cardiomyocytes and endothelial cells has not been studied extensively as a diagnostic criterion or therapy in the clinical scenario.

5.1. Sex differences in diabetic cardiomyopathy

Diabetic cardiomyopathy, defined as ventricular dysfunction occurring in diabetic patients independent of a coronary heart disease or hypertension [355], has been shown to be more prevalent in diabetic women than in diabetic men [356]. Similarly, the Framingham Heart Study reported a two-fold increase in HFpEF in females vs males [357]. These findings are intriguing, because HFpEF has been shown to have a strong link with metabolic syndrome [358], which includes insulin resistance, lipid disorders, obesity, and hypertension.

The mechanisms underlying these differences are mostly unknown, but can include disparities in Ca²⁺ handling and other ion channels in cardiomyocytes [359-365], a sexually dimorphic function of contractile proteins [366,367], a dissimilar response to adrenergic stimuli [368,369], as well as a diverse remodeling response [370,371].

Some data from clinical studies can be very helpful in better understanding the relationship linking gender, HFpEF, and diabetic cardiomyopathy, although further dedicated studies are necessary: for instance, a functional role for estrogen in regulating insulin sensitivity has emerged in the Women's Health Initiative Hormone Trial [372], and metformin has been shown to decrease fatty acid clearance in men but not in women [373].

5.2. Notes on clinical management

As previously mentioned, glycemic control is a powerful way to reduce the risk of HF in patients with both T1DM and T2DM. Notably, a 2017 prospective study found a significant increase in mortality with higher admission blood glucose (ABG) in HF patients with both pre-existing DM (>200 mg/dL versus <110 mg/dL, HR = 1.20, p = 0.032) and those without DM (HR = 1.04 per 18 mg/dL increase in ABG, p = 0.014) over a follow-up period of 10 years [374]. Therefore, a constant glycemic therapy — including metformin, sulfonylureas, insulin, and glucagon-like peptide 1 (GLP-1) receptor agonists — is of utmost importance and has been shown in several prospective studies to ameliorate both systolic and diastolic dysfunction [375,376]. Diastolic dysfunction is a common characteristic in both T1DM and T2DM patients with diabetic cardiomyopathy [377], However, studies report fewer clinical symptoms of HF in T1DM patients compared to T2DM, possibly owing to T1DM patients being of younger age and having glycemia well-controlled with insulin [378,379].

SGLT2 inhibitors (SGLT2Is) are a promising new-line treatment that not only reduces hyperglycemia but also seems to have direct beneficial effects on cardiomyocytes — independent of glucose levels — including improved Ca²⁺ handling and myocardial contractility [380]. In patients with T2DM, treatment with empagliflozin decreased adverse cardiac remodeling [381,382]. Although there are limited reports regarding if SGLT2Is are beneficial in treating existing HF inpatients with DM, promising data are showing their effectiveness in preventing HF development across a broad spectrum of cardiovascular risk [383,384]. Additional putative benefits of SGLT2Is include improved cardiovascular energetics, reduced vascular tone and blood pressure [380], decreased renal dysfunction, increased circulating levels of ketone bodies [385], and overall reduced systemic inflammation [109].

Like SGLT2Is, GLP-1 agonists have also been shown to have cardioprotective properties, independent of their glucose-lowering effects, in patients with T2DM. However, the results of the 2016 Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial suggest that the benefits of GLP-1 agonists may be more closely tied to the mitigation of atherosclerosis rather than the hemodynamic benefits in the previously mentioned empagliflozin studies [334,386-388].

6. Conclusions and perspectives

At this moment, there are no proven therapies for reducing mortality in HFpEF [389-392], probably due to its heterogeneous nature. For example, as shown by the CHARM-Preserved (Candesartan in Heart Failure: Assessment of Reduction in Mortality and Morbidity in HFpEF] study, treatment using the angiotensin II receptor blocker (ARB) candesartan failed to reach statistical significance in mortality reduction (unadjusted HR 0.89 [95% Cl 0.77–1.03], p = 0.118) [393]. Instead, addressing any co-morbidities is the mainstay of HFpEF clinical management. Once symptoms of HFrEF begin to arise, patients should be treated with proven therapies such as angiotensin-converting enzyme (ACE) inhibitors, ARBs, and beta-blockers. In either case, compounds that promote diuresis, including the angiotensin receptor neprilysin inhibitor (ARNI), should be considered to relieve symptoms of volume overload with caution in avoiding hypotension [1,394]. For patients with arrhythmia, pharmacologic control of heart rate should be used; implantable cardiac defibrillators (ICDs) may be considered even in HF patients without arrhythmia as secondary prevention for sudden death [395,396].

Lifestyle changes are very effective in improving the quality of life in patients with asymptomatic diastolic dysfunction and HFpEF [397]; specific modifications that should be considered to reduce the burden of HFpEF include: stopping smoking, losing weight, getting active, eating better, managing blood pressure, controlling cholesterol, and reducing blood sugar [157,320,397]. Exercise training, in those who are able, has been shown to yield improvements in cardiorespiratory fitness and quality of life surveys, with mild, yet non-significant, changes in left ventricular diastolic function [22,398,399].

Understanding the molecular mechanisms underlying diabetic cardiomyopathyand HFpEF and finding new therapeutic strategies represent some of the main future challenges in cardiovascular endocrinology [400].