Diabetes mellitus (DM) is a group of metabolic disorders characterized by chronic hyperglycemia and perturbed metabolism of carbohydrates, lipids, and proteins, resulting from defects in insulin secretion and action. An estimated 9.3% of the world population (463 million aged 20-79 years) is affected by DM, and this number is projected to reach 10.9% (700 million people) by 2045 (1). Macrovascular complications, such as coronary artery disease (CAD) and ischemic cardiomyopathy, are the leading causes of cardiac death in DM patients. In addition, DM raises the risk of heart failure (HF) and cardiac dysfunction unaided by other risk factors, such as CAD and hypertension (2). Also, microvascular disease and cardiac capillary rarefaction contribute to severe cardiovascular morbidity and mortality in DM patients (3, 4). Diabetes-related cardiomyopathy (DCMP) represents DM-induced morphofunctional cardiac abnormality after the presence of valvular, atherosclerotic, congenital, or hypertensive heart disease is excluded (5–7). Clinically, DCMP can be presented as two distinctive phenotypes, restrictive (heart failure with preserved ejection fraction, HFpEF) and dilated (heart failure with reduced ejection fraction, HFrEF) (8, 9). Transitioning from HFpEF to HFrEF is not mandatory (3, 9). In HFpEF and HFrEF, the presence of DM increases the risk of hospitalization for HF or even death (10, 11). However, the difficulty in identifying HF is due to the asymptomatic presentation in the early stages of DCMP (12). Furthermore, DCMP worsens DM patients’ prognoses and raises their chance for overt HF (13, 14).

The present challenges in the definitive diagnosis of DCMP are the absence of specific circulating or histological biomarkers of the disease (3, 15, 16) and insufficient guidance for managing patients suffering from both DM and HF (15, 17). Currently, DCMP diagnosis is most widely determined using echocardiography; to detect changes in the myocardium structure and function (9, 18–20). However, due to its economic costs, it is not well-suited for routine screening of DCMP. Thus, there is an urgent need to identify and develop novel blood-based biomarkers to identify patients with an increased risk of developing DCMP (18).

Dysregulation of long non-coding RNA (lncRNA) and microRNA (miRNA) regulatory networks is emerging as an important mechanism in the pathophysiology of DCMP (3, 21–23). miRNAs are small, non-coding RNAs (ncRNA) that regulate the expression of numerous genes involved in physiological processes such as metabolism, apoptosis, differentiation, and cell proliferation. Increasing evidence points to miRNAs’ role in the regulation of pathophysiological alterations associated with DCMP, such as cardiac hypertrophy (24), myocardial fibrosis (25), oxidative stress (OS) and apoptosis (26), mitochondrial dysfunction (27), epigenetic modification (28), cardiac electrical remodeling (29). lncRNAs are long, non-translated transcripts with more than 200 nucleotides involved in regulating the activity and abundance of miRNAs through base-pairing interactions (22, 30). lncRNAs mediate numerous physiological processes, such as transcription regulation, RNA splicing, nuclear architecture and compartmentalization, and nuclear-cytoplasmic trafficking (31–34). Recent reports implicate the role of lncRNA in DM pathogenesis and associated cardiovascular complications, such as DCMP (35, 36).

In this review, we provide a systematic overview of DCMP pathogenesis and progression, focusing on the specific roles of miRNAs and lncRNAs in the pathophysiology of DCMP. Also, we discuss novel approaches based on the use of miRNAs and lncRNAs as targets for potential therapeutic interventions.

The pathogenesis of DCMP is based on joined metabolic conditions (hyperglycemia, hyperinsulinemia, and dyslipidemia) that promote OS, inflammation, the formation and deposition of advanced glycation end products (AGEs), damage and dysfunction of mitochondria, unbalanced Ca²⁺ homeostasis, endoplasmic reticulum stress (ERS), autonomic neuropathy, the renin-angiotensin system (RAS) activation, microvascular myocardial rarefaction, changes in gene regulation (microRNAs), and cardiomyocyte apoptosis (2, 5, 16, 37–39).

Both types of DM are characterized by decreased insulin signaling and changes in other signaling cascades, such as reduced AMPK and increased PKC and MAPK signaling, with resultant deleterious and maladaptive effects (3). DCMP’s clinical presentation may be preceded by myocardial structure changes and disturbed Ca²⁺ signaling and metabolism (2, 4, 7, 22). The myocardial structure changes, i.e., myocardial fibrosis, are favoured by increased collagen deposition and variations in extracellular matrix (ECM) protein structure (40). The imbalance between profibrotic factors, such as connective tissue growth factor and transforming growth factor β1, and the inactivity of the ECM-degrading enzyme metalloproteinase can lead to ECM accumulation (41, 42). Among numerous mechanisms that favour DM-induced cardiac fibrosis, the intriguing one is the endothelial-to-mesenchymal transition (EndMT). EndMT is known to be promoted by hyperglycemic conditions, and it evolves gradually, acquiring a fibroblastic phenotype while simultaneously losing the original phenotype of the endothelial cells (ECs). This phenotypic change is accompanied by a progressive decline in EC activity and the cells’ mesenchymal characteristics, such as increased ECM protein production, becoming more pronounced. In injured tissue, the EndMT-derived cells act as immature fibroblasts and promote the fibrosis process (43).

Cardiomyocytes with abnormal metabolism are susceptible to increased free fatty acid (FFA) uptake and oxidation. Increased lipids may promote cardiomyocyte death induced by lipotoxicity due to limited FFA oxidation (3, 44). In addition, reactive oxygen species (ROS) and reactive nitrogen species (RNS) are produced more frequently as a result of increased intracellular fatty acid content and mitochondrial malfunction, which in turn increases OS and ERS and inhibits autophagy (20, 45, 46). The interaction of these effects causes ECM remodeling and fibrosis, along with cardiomyocyte loss, cardiac enlargement, and inflammation (47). Heart stiffness, poor cardiac relaxation, and diastolic dysfunction are early signs of DCMP caused by pathophysiological anomalies (46). In addition, accumulated lipids in ECs may decrease nitric oxide (NO) bioavailability, promoting endothelial dysfunction and accelerating atherosclerosis (9).

Diastolic or systolic dysfunction is encouraged by left ventricular (LV) hypertrophy and perivascular and interstitial cardiac fibrosis (3, 48). On echocardiograms, LV hypertrophy presents increased thickness in the posterior and septal walls (49). Myocyte hypertrophy, thickening of the myocardial capillary basement membrane, and increased interstitial and perivascular fibrosis are confounding factors contributing to the development of LV hypertrophy (50, 51).

DCMP occurs in approximately 12% of patients with DM (52). The prevalence of HF varies between 19 and 26% in both types of DM (3, 4). The link between type 2 DM (DMT2) and HF is bidirectional: HF is highly prevalent in DMT2 patients, and HF increases the risk of DMT2 (3, 53, 54). Sometimes, HF is the first cardiovascular presentation in patients with DMT2 (9, 55). In DM patients, the risk of a negative HF outcome is greater (56). The risks of HF in diabetic patients are in close relation to the quality of retrograde glycemic control, as the patients with type 1 (DMT1) and DMT2 have a 30% and 8% increase in HF risk for each 1% increase in glycated haemoglobin (HbA1c) level, respectively (3, 57).

DCMP’s natural course is determined by phenotype. In DCMP with HFpEF phenotype, the LV is hypertrophied, stiff, and of normal size. In DCMP with HFrEF phenotype, the LV is dilated with reduced ejection fraction (9, 58). In humans, diastolic dysfunction almost always precedes the development of systolic dysfunction (59, 60). Metabolic abnormalities in DMT2 predispose to the development of HFpEF DCMP, while the autoimmune abnormalities in DMT1 favour HFrEF DCMP (9).

At present, echocardiography represents an indicative diagnostic tool for assessing a patient with suspected DCMP (16, 61). The first echocardiographic signs of DCMP are LV diastolic dysfunction and mechanical changes leading to HFpEF and, ultimately, HFrEF (19). In the early stage or the restrictive HFpEF form, the echo findings show normal LV diameters and volumes with concentric hypertrophy, preserved systolic function (EF ≥50%), and indications of diastolic dysfunction (9). Systolic dysfunction is a later manifestation, sometimes misdiagnosed using standard two-dimensional echocardiography (2). Rarely, T1 cardiac MRI mapping is an initial diagnostic procedure in detecting DCMP, as myocardial ECM in DM patients and non-DM controls exhibit significant differences (62). In addition, the increased levels of natriuretic peptide, inflammatory markers, and cardiac fibrosis markers are linked to diastolic dysfunction in DCMP (16, 63, 64). In the advanced stages of DCMP, or the dilated/HFrEF form, systolic dysfunction (ejection fraction <50%) occurs and an increase in LV volume (9).

Continuous inflammatory stimulation appears to be one of the most critical factors of DM pathogenesis (65). In the acute phase of inflammation, cytokines and acute-phase proteins (APPs) mitigate the effects of transient inflammatory processes (66, 67). However, prolonged inflammation results in a chronic condition where immune response leads to tissue damage contributing to the pathogenesis of many diseases, including atherosclerosis, cardiomyopathy, and DM (68). Nevertheless, diagnostics of DCMP based on measurements of circulating markers of inflammation, such as complement compounds, C-reactive protein (CRP) and alpha-macroglobulin (α2M), and amyloid A and P, is not sufficiently reliable, thus requiring identification of more specific biomarkers that enable early detection of DCMP (18, 68).

Stringent control of DM and the treatment of HFpEF or HFrEF is the cornerstone of DCMP management. DCMP is not a rare cardiovascular complication of DM (16). Using tissue Doppler strain analysis and measurements of peak systolic velocity, almost every fifth patient with DM was diagnosed with systolic dysfunction after excluding CAD or hypertension (49). Novel oral agents currently used in DM management (i.e., sodium-glucose cotransporter 2 (SGLT2) inhibitors, glucagon-like peptide 1 receptor agonists (GLP1-RAs)) enable a reduction in hospitalization rates for HF in DM patients independently of the presence of HF at baseline (84, 85). SGLT2 inhibitors exert antioxidative, antiapoptotic, and anti-inflammatory effects and decelerate atherosclerosis (86).

miRNAs represent small (17-25 nucleotides), single-stranded non-coding RNA molecules that regulate gene expression (87). Theoretically, a single miRNA could bind to over 1000 target mRNAs, and various miRNAs could regulate the expression of the same target transcript (88, 89). Since each miRNA may target several mRNAs, it has been estimated that miRNAs may regulate the expression of up to 60% of protein-coding genes in humans (90). Until 2019, the miRBase database (miRBase Release 22.1, https://www.mirbase.org/) reported entries of 38 589 miRNAs in 271 species, including 2654 mature human miRNAs (91). Increasing evidence supports the significant roles of miRNAs in regulating the mechanisms responsible for the pathophysiology of numerous diseases, including cardiovascular diseases, obesity, different types of cancer, and diabetes (73, 92–98).

miRNAs biogenesis is a multistep process that starts with primary miRNA (pri-miRNA) transcription by RNA polymerases II and III in the nucleus, which is subsequently processed by the nuclear endoribonuclease DROSHA or by components of the splicing machinery (99) to approximately 70 nucleotides long precursor (pre-miRNA) molecules that are exported to the cytoplasm by exportin 5 and Ran-GTPase. Additional processing by type III endoribonuclease DICER associated with RNA-binding proteins yields mature double-stranded miRNAs. The guide strand of mature miRNAs associates with Argonaute (AGO) proteins or chaperones HSC70/HSP90 to form the minimal miRNA-induced silencing complex (miRISC) that binds to the target mRNA’s complementary sequences called miRNA response elements (MREs). MiRNAs mainly interact with the target mRNAs’ 3′ untranslated regions (UTR) to induce translational repression and mRNA deadenylation (100–102), but interactions of miRNAs with 5′ UTR, gene promoters, and coding sequences have also been observed (103). It is generally assumed that the interaction of miRNAs with coding regions and 5′ UTR silence gene expression (104, 105), while binding of miRNAs to promoter regions can trigger transcription (106).

lncRNAs are heterogenous RNA transcripts with more than 200 nucleotides that are not translated into proteins (167) but can interact with DNA, RNA and proteins via base pairing or chemical interactions, thus exhibiting more versatile roles compared to miRNAs. RNA polymerase II transcribes lncRNAs from exonic, intergenic, or distal protein-coding regions of the genome into pre-mature lncRNAs that are polyadenylated at the 3’-end and capped on the 5’-end with methyl-guanosine (168). The precursor lncRNA undergoes alternative splicing either by interacting with specific splicing factors or forming RNA-RNA duplexes with pre-mRNA molecules (169). lncRNAs regulate gene expression at the transcriptional, translational and post-translational levels (30, 170) by binding to DNA-binding proteins (171), recruiting epigenetic complexes during DNA methylation (172), and serving as precursors of miRNAs (173). Their function depends on the cellular location; lncRNAs expressed in the nucleus regulate gene expression via recruitment of transcription factors or epigenetic complexes (174) whereas cytoplasmic lncRNAs participate in modulation of the mRNA stability and translation and post-translational modifications (175–177). lncRNAs are further classified as signal, guide, decoy, and scaffold lncRNAs depending on their cellular function. Signal lncRNAs respond to specific stimuli at distinct subcellular locations whereas guide lncRNAs direct ribonucleoprotein complexes to specific targets (33). Decoy lncRNAs bind and sequester regulatory proteins such as transcription factors (178), while scaffold lncRNAs have a structural role in chromatin organization as platforms for assembling ribonucleoprotein complexes (179). It has been estimated that the human genome contains over 16000 lncRNAs (Gencode-Human Release 27, https://www.gencodegenes.org/human/ ) (180). However, despite this remarkable number, the number of functional lncRNAs remains questionable, although they express valuable cellular properties (168, 181).

Although there are fewer reports in the literature regarding lncRNAs’ connection to DCMP compared to miRNAs, recent evidence strongly supports the equally important emerging role of lncRNAs in DCMP pathophysiology ( Figure 1 ). Levels of several lncRNAs are perturbed in serum and myocardial biopsy of patients with DCMP (182, 183). For instance, the plasma level of the lncRNA, the steroid receptor RNA activator (SRA), is decreased in DM patients with CVDcompared to DM patients without any associated complications and healthy subjects. Furthermore, a 5-year follow-up study demonstrated that perturbed levels of SRA correlate with an increased incidence of cardiovascular disease in DM patients (184). Accumulating evidence shows that lncRNAs participate in the modulation of multiple pathways associated with OS and inflammation, which are implicated as important factors in DCMP development and progression, myocardial injury, cardiac hypertrophy, and diabetic vascular complications (142). HOX transcript antisense RNA (HOTAIR) has a crucial role in the CVD pathophysiology (185), and its expression is significantly downregulated in myocardial tissues and serum of patients with DCMP compared to DM patients and healthy controls (146). HOTAIR expression was also decreased in the hearts of streptozotocin-treated mice, whereas its overexpression decreased OS and inflammation and improved cardiac function (147). HOTAIR was reported to serve as a molecular sponge of miR‐34a, which targets Sirt1 ( Table 1 ) (147, 186). HOTAIR was also shown to ameliorate DCMP by increasing the viability of cardiomyocytes via PI3K/Akt pathway activation (146).

In a study by Yu et al. (145), differentially expressed lncRNAs during cardiomyocytes’ OS and apoptosis induced by high glucose were identified by RNA sequencing. Consequent functional studies showed that inhibition of lncRNA NONRATT007560.2 reduces ROS generation and apoptosis, suggesting its important role in developing cardiomyopathy. In addition, it was observed that NON-RATT007560.2 have binding sites for miR-208a (145), which was previously associated with the perturbed cardiac remodeling in the myocardium of DMT2 patients (144). Xu et al. found that NONRATT021972 siRNA treatment of DM rats decreased the elevated TNF-α expression and abolished serine phosphorylation of IRS-1 in superior cervical ganglion cells, whereas downregulation of NONRATT021972 restored decreased heart rate variability in diabetic rats ( Table 1 ) (143).

Another lncRNA, KCNQ1OT1, whose expression is increased in the serum of diabetic patients, as well as in high glucose-induced cardiomyocytes in vitro, and cardiac tissue of T1DM streptozotocin-induced diabetic mice, has been associated with pathophysiological mechanisms leading to cardiac dysfunction (138, 139, 183). A study by Coto et al. revealed that the increased levels of KCNQ1OT1 induce TGF-β1, p-Smad2 and p-Smad3 expression and are accompanied by collagen deposition, activation of fibrotic formation and cardiac remodeling, ultimately resulting in deterioration of LV function. The inhibition of KCNQ1OT1 expression significantly ameliorated cardiac function and reduced remodeling via TGF-β1/Smads pathway (138). Another study showed that KCNQ1OT1 silencing improves cardiac function by decreasing apoptosis via targeting miR-214-3p and caspase-1 gene, which leads to reduced cell death and abnormalities in cytoskeletal structure as decreased calcium overload ( Table 1 ) (183).

lncRNA H19 also regulated cardiomyocyte apoptosis in diabetic cardiomyopathy (134). Li et al. reported that expression of H19 was significantly downregulated in the myocardium of diabetic rats, whereas its overexpression reduced OS, inflammation and apoptosis, leading to an improvement of LV function (134). In cultured cardiomyocytes transfected with H19 siRNA, decreased expression of H19-derived miR-675 was observed. VDAC1 gene, involved in cardiomyocyte apoptosis and the progression of cardiac muscle dysfunction, was identified as a target of H19/miR-675-mediated downregulation (134). Another study reported that overexpression of H19 epigenetically silences DIRAS3 (DIRAS Family GTPase 3), promotes mTOR (mammalian target of rapamycin) phosphorylation, and inhibits autophagy in cardiomyocytes exposed to high glucose ( Table 1 ) (140).

lncRNAs are also implicated in cardiomyocyte injury via activation of NF-κB and TNF signaling pathways. In obesity, DM and other metabolic disorders, excessive amounts of saturated fatty acids, such as palmitic acid (PA), may be deposited in cardiomyocytes causing lipotoxic damage (2, 187). Upregulation of inflammatory factors TNFα and IL-1β and lncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1), which plays a crucial role in cardiomyocytes ischemia-reperfusion damage, was shown in PA-treated cardiomyocytes (82). MALAT1 knockdown increased the viability of PA-treated cardiomyocytes and reduced TNF-α, IL-1β, myocardial damage markers such as lactate dehydrogenase (LDH) and CK-MB, and apoptosis (142). MALAT1 specifically binds to miR-26a, inhibiting the inflammatory signaling pathway Toll-like receptor 4 (TLR4)/NF-κB by binding to its target gene, HMGB1. Thus, MALAT1 inhibition alleviates lipotoxic myocardial injury via the miR-26a/HMGB1/TLR4/NF-κB axis (142). Downregulation of MALAT-1 also reduces inflammation under high glucose conditions. A study by Puthanveetil et al. reports significant upregulation of MALAT1 in endothelial cells exposed to high glucose levels (141). Increased MALAT1 levels were associated with a parallel increase in TNF-α, interleukin 6 (IL-6) and serum amyloid antigen 3 (SAA3), an inflammatory ligand and target of MALAT1. These findings suggest that MALAT1 regulates glucose-induced upregulation of inflammatory mediators IL-6 and TNF-α by activating SAA3 (141).

The level of lncRNA Antisense Non-coding RNA in the INK4 Locus (ANRIL) is increased in peripheral venous blood from DMT2 patients with acute myocardial infarction (188). ANRIL was shown to regulate the expression of HBEGF and CDH5 genes involved in vascular permeability, leukocyte migration, and associated inflammation (148). ANRIL level is increased in the hearts of diabetic rats, and its silencing is associated with reduced levels of LDH, CK-MB, and inflammatory cytokines TNFα, IL-6, and IL-1β, suggesting that ANRIL inhibition improves cardiac function ( Table 1 ) (149).