Metabolic and Molecular Imaging of the Diabetic Cardiomyopathy

Abstract

The term “diabetic cardiomyopathy” is defined as the presence of abnormalities in myocardial structure and function that occur in the absence of, or in addition to, well-established cardiovascular risk factors. A key contributor to this abnormal structural-functional relation is the complex interplay of myocardial metabolic remodeling, defined as the loss the flexibility in myocardial substrate metabolism and its downstream detrimental effects such as mitochondrial dysfunction, inflammation and fibrosis. In parallel with the growth in understanding of these biological underpinnings has been developmental advances in imaging tools such as positron emission tomography and magnetic resonance imaging and spectroscopy that permit the detection and in many cases quantification, of the processes that typifies the myocardial metabolic remodeling in diabetic cardiomyopathy. The imaging readouts can be obtained in both pre-clinical models of diabetes mellitus and diabetic patients facilitating the bi-directional movement of information between bench and bedside. Moreover, imaging biomarkers provided by these tools are now being used to enhance discovery and development of therapies designed to reduce the myocardial effects of diabetes mellitus through metabolic modulation. In this review the use of these imaging tools in the diabetic patient from a mechanistic, therapeutic effect and clinical management perspective will be discussed.

Introduction

The term “diabetic cardiomyopathy” is defined as the presence of abnormalities in myocardial structure and function that occur in the absence of well-established cardiovascular (CV) risk factors such as coronary artery disease or hypertension in patients with diabetes mellitus (DM). It was originally described nearly a half a century ago based on postmortem pathological findings in a small cohort of diabetic patients who manifested heart failure symptoms without evidence of coronary artery or valvular heart disease.(1) It was confirmed in several larger population-based studies that demonstrated a higher incidence of heart failure in diabetic patients after adjustment for well-established heart failure risk factors.(2) As a consequence, diabetic cardiomyopathy has now been codified in various heart failure guidelines.(3, 4)

However, it is also clear that the presence of DM alters the myocardial structural-functional relation increasing the susceptibility of diabetics to the effects of common forms of CV disease. For example, the presence of DM is an independent risk factor for developing left ventricular (LV) hypertrophy (LVH) and congestive heart failure and worsens the prognosis for those who develop these conditions.(2, 5–7) Although plasma inflammatory biomarkers such as interleukin-6 and C-reactive protein, as well as macroalbuminuria have been shown to be predictors of heart failure, their ability to predict these conditions in individual diabetic patients is limited.(8) Thus, there is an urgent need for diagnostic approaches that incorporate biomarkers unique to the diabetic cardiomyopathic process. Ominously, the poorer prognosis engendered by DM compared with non-diabetics does not appear to be ameliorated by the institution of appropriate therapies designed to reduce LVH or treat LV dysfunction.(5, 9–11) Of note, diabetic women also exhibit a greater prevalence of LVH and heart failure compared with male diabetics.(10, 12, 13) In addition, the risk of death is greater in diabetic women with heart failure compared with their male counterparts even in the setting of optimal medical therapies.(14) These sex differences appear to be independent of the type (e.g., reduced versus preserved LV ejection fraction) or etiology (e.g., ischemic versus non-ischemic) of the heart failure.

Exquisite studies in various pharmacological and genetic pre-clinical models of DM have demonstrated the detrimental effects of DM on the myocardium extend well beyond the consequences of the presence of hyperglycemia. A key contributor is the complex interplay of myocardial metabolic remodeling and its downstream detrimental effects such as mitochondrial dysfunction, inflammation and fibrosis which when taken in sum, may adversely affect myocardial structure and function.

In parallel with the growth in understanding of these biological underpinnings has been the explosive development of imaging tools such as positron emission tomography (PET) and magnetic resonance imaging (MRI) and spectroscopy (MRS) that permit the detection and in many cases quantification, of the processes that typifies the myocardial metabolic remodeling in diabetic cardiomyopathy. Indeed, in many cases the imaging readouts can be obtained in both pre-clinical DM models and diabetic patients facilitating the bi-directional movement of information between bench and bedside. Moreover, imaging biomarkers provided by these tools are now being used to enhance discovery and development of therapies designed to reduce the myocardial effects of DM through metabolic modulation. In this review the use of these imaging tools in the diabetic patient from a mechanistic, therapeutic effect and clinical management perspective will be discussed.

Metabolic Remodeling in Diabetic Cardiomyopathy – Potential Imaging Targets

A detailed discussion of the mechanistic basis for the detrimental effects of DM on the myocardium, even limited to the remodeling of myocardial substrate metabolism and the subsequent downstream detrimental effects, is beyond the scope of this review. Comprehensive descriptions of this topic can be found in these reviews.(15, 16) In brief the myocardial metabolic response to DM can viewed as a local response to systemic conditions. This is because of the critical need of the normal heart to oxidize a variety substrates via mitochondrial oxidative phosphorylation to generate nearly all of the adenosine triphosphate (ATP) (~98%) to meet its energy needs.(17, 18) Under normal conditions fatty acid (FA) oxidation accounts for 60–90% of mitochondrial ATP generation while the balance is provided by the metabolism of carbohydrates (glucose and lactate) to pyruvate. The dynamic and oftentimes uniquely reciprocal control of substrate flux through myocardial FA β-oxidation and glycolytic pathways is dictated by ever changing physiologic conditions such as the plasma substrate environment, neurohumoral milieu and level of cardiac work. These acute adaptations in substrate selection and metabolism are central to cardiac myocyte health.(17, 18)

In DM, this flexibility in substrate usage is lost as the systemic impairment in insulin production and signaling in DM results in elevated plasma FAs leading to a sustained increase in myocardial FA uptake. This activates key transcriptional pathways such as the PPARα/PGC-1 signaling network resulting in a further increase in myocardial FA uptake and oxidation. Concomitantly, glucose uptake, glycolysis and glucose oxidation are reduced in diabetic hearts, most likely mediated by decreased GLUT4 expression and translocation and impaired pyruvate decarboxylation.(19–21) Recently, there has been growing interest in the role ketone metabolism in diabetic cardiomyopathy given the heart is one of the main organs to metabolize ketone bodies.(22, 23) Ketone metabolism also offers the advantage of producing ATP more efficiently per molecule of oxygen consumed than glucose or FAs, thus may be attractive alternative fuel. Driving this interest is the documented improvement in cardiac function in DM patients in response to treatment with the sodium/glucose cotransporter 2 (SGLT2) inhibitors that is paralleled by increase in plasma ketone body levels.(24) However, definitive evidence that the diabetic heart exhibits increased ketone body metabolism is lacking and if present, whether such an increase is adaptive or maladaptive.

These chronic metabolic adaptations, particularly when superimposed upon systemic activation of the renin–angiotensin–aldosterone system, sympathetic nervous system and low-grade inflammation that characterize DM, stimulate a host of effectors that both further exacerbate the upstream perturbations in substrate metabolism and accelerate downstream processes that contribute to cardiomyocyte dysfunction and death (Figure 1). Examples include:

Figure. 1.

Myocardial metabolic remodeling – beyond substrate flux. Several processes are modulated by metabolic perturbations, such as those that occur with obesity, DM and inborn errors of metabolism. Among these are shown clockwise from upper left:

1) Lipid accumulation. Lipid droplets within the myocardium are shown on this electron micrograph (reproduced with permission)(133);

2) Energetic abnormalities: ³¹P spectra from a normal control (left) and a patient with Barth syndrome (right) (reproduced with permission) (134);

3) Oxidative stress: Increased hydrogen peroxide production in the myocardium of wild type diabetic (streptozotocin [WT/STZ]-treated) animals as compared with wild type controls (WT/C);

4) Inflammation: CD68 tissue levels in rat heart induced by high fat feeding (HFD);

5) Apoptosis: Increased caspase-3 levels in hyperglycemic (HG) conditions compared with normoglycemic (NG) and osmotic control (OSM) conditions, and with treatment with liraglutide + HG. (reproduced with permission)(135);

6) Infarct size: Infarction/necrosis in human heart;

7) Fibrosis: Fibrosis in human heart;

8) Gene expression: Increased mRNA expression of acyl-CoA accentuated by fasting in mice that overexpress PPARα in mouse heart.

Center) A global longitudinal strain map showing diminished (less negative) strain in a patient with decreased LV function. Normal strain is depicted by dark red; lighter pink = lower, less negative strain, and blue indicates dyskinesis.

Increased Oxidative Stress

An increase in myocardial FA uptake can stimulate reactive oxygen species production either directly through accelerated mitochondrial electron transport chain flux due to increased FA oxidation or indirectly due to insulin resistance leading to intracellular hyperglycemia a potent stimulus for nicotinamide adenine dinucleotide phosphate oxidase.(25, 26) The increase in oxidative stress appears to limit FA oxidative capacity in relation to continued enhanced FA uptake.(27) The resultant lipid FA accumulation further worsens oxidative stress via inducible form of nitric oxide synthase stimulation which leads to reactive nitrogen species production.(28) Potential downstream mechanisms by which oxidative stress contributes to the diabetic cardiomyopathic process include activating cell death pathways via caspase-3 and poly (ADP-ribose) polymerase 1, increased formation of advanced glycation end-products, stimulating hyperglycemia induced activation of protein kinase C isoforms and increasing mitochondrial uncoupling.(29–33)

Steatosis

As mentioned above, FA oxidative capacity is limited in relation to continued enhanced FA uptake leading to FA accumulation or steatosis.(27, 34) In addition to increasing reactive nitrogen species, steatosis can potentially impair LV function through a variety of mechanisms including ceramide mediated impairment of insulin signaling and activation of cell death pathways via ceramide production and increasing mitochondrial coupling via impairment mitochondrial oxidative phosphorylation and ATP synthesis that occurs in the setting of preserved or increased myocardial oxygen consumption (MVO₂).(28, 35) The presence of mitochondrial uncoupling results in reduced LV efficiency.(29, 36, 37)

Mitochondrial Dysfunction and Impaired Energetics

Mitochondrial dysfunction and impaired energetics is characteristic of the diabetic heart. Beyond FA-induced mitochondrial uncoupling other detrimental mitochondrial effects of metabolic remodeling include oxidative damage, transcriptional and translational alterations of oxidative phosphorylation subunit expression, impaired mitochondrial calcium handling and changes in cardiac insulin signaling.(38, 39)

Inflammation

In DM, the myocardium already subjected to low-grade systemic inflammation, is further exposed to local cellular inflammation due to the consequences of metabolic remodeling, through a variety of different mechanisms.(16) Most notable is the activation nuclear factor-κB from diverse stimuli such as high cellular levels of glucose and FAs, reactive oxygen species and advanced glycation end products. The activation of this family of transcription factors either directly or indirectly (via induction in the nucleotide-binding oligomerization domain like receptor pyrin domain containing 3 inflammasome) stimulates pro-inflammatory cytokine production and innate immune system responses.(40–42)

Fibrosis

Myocardial fibrosis, which typifies the diabetic cardiomyopathy represents the confluence of these perturbations in cellular metabolism. For example, hyperglycemia, advanced glycation end-product production and increased oxidative stress directly activate resident cardiac fibroblasts and elicit a proliferative response through the induction and activation of fibrogenic growth factors (such as TGF-β and connective tissue growth factor).(38, 43, 44) Moreover, various chemokines may stimulate fibrosis either through direct actions on fibroblasts or through recruitment of fibrogenic monocyte subsets.(40)

Discussed in the sections that follow are how current and newly developed imaging approaches can help characterize these chronic adaptations in diabetic cardiomyopathy.

Imaging Tools to Evaluate Diabetic Cardiomyopathy

PET and MR-based approaches (MRI, MRS and hyperpolarized MRS [HP-MRS]) are the most commonly used imaging tools to measure the perturbations in myocardial metabolism and their downstream consequences. In the next few sections each method is discussed in a bit more detail.

PET

A major advantage of PET is the inherent high sensitivity of radionuclide methods that permit the detection of low density processes with radiotracers exhibiting a mass with nano-picomolar concentrations which are too low to alter the metabolic, biological or molecular processes of interest. Indeed, PET is the most widely used radionuclide approach to measure myocardial metabolism and to perform molecular imaging because of its flexibility in radiotracer design and inherently quantitative capabilities.(45, 46) Several of the radionuclides used to form the radiotracers are biologically ubiquitous elements such as oxygen (¹⁵O), carbon (¹¹C), and nitrogen (¹³N). Fluorine (¹⁸F) that can be substituted for hydroxyl groups which allow for incorporation into a wide variety of substrates or substrate analogues that participate in diverse biochemical pathways. Moreover, these radiotracers are administered at doses that do not alter the metabolic processes of interest. However, because PET the signal reflects the integration of all pathways involved in the metabolism of the radiotracer used, metabolic measurements must be performed with radiotracers designed for targeting specific pathways (e.g., uptake or oxidation) or with the assistance of relatively complex modeling of myocardial kinetics.(46) In terms of molecular imaging, the flexibility in PET radiotracer design utilizes various targeting strategies (e.g., small molecules, peptides and antibodies) radiolabeled with a host of radionuclides (e.g., ⁶⁸Ga, ¹⁸F and ⁶⁴Cu) that permit tailoring radiotracer performance to detect specific processes (e.g., surface receptor versus enzymatic activation) in different locations (e.g., vascular endothelium versus cardiac myocyte).(47, 48) Listed in Table 1 are many of the currently available PET radiotracers for performing myocardial metabolic and molecular imaging. It should be noted that numerous metabolic radiotracers have been used to characterize the diabetic cardiomyopathy. However, of the many molecular imaging radiotracers, only those evaluating neuronal function have been used to study diabetic cardiomyopathy. Although cardiac autonomic neuropathy is a key contributor to the diabetic cardiomyopathy it appears less directly related to myocardial metabolic remodeling, so the use of these radiotracers will not be discussed further. Excellent summaries of their use in diabetic cardiomyopathy can be found elsewhere.(49, 50)

Table 1.

Examples of PET Radiotracers of Myocardial Metabolism and Molecular Imaging

Metabolic or Molecular Process of Interest	Radiotracers
Oxygen consumption	15O2(51), 1-11C-acetate(52)
Long chain fatty acid metabolism
Uptake, oxidation, and storage	1-11C-palmitate(53)
Uptake and oxidation	18F-FTHA(54), 18F-FTP(55),18F-FTO(56)
Ketone body metabolism	1-11C-acetoacetate(57)
Glucose metabolism
Uptake	FDG
Uptake, glycolysis, oxidation	1-11C-glucose(58)
Lactate metabolism	L-3-11C-Lactate(59)
Neuronal function	11C- hydroxyephedrine(60)
Inflammation
Metabolism	FDG, 11C-methionine(61)
SSTR2 expression	68Ga-Dotatate(62)
TSPO	11C-PK11195(63)
Chemokines	64Cu-DOTA-ECL1i(64), 68Ga-Pentoxifer(65)
Oxidative Stress	18F-F12(66), 18F-FHMT(67)
Fibrosis	68Ga-FAPIs(68)

FTHA, 14-(R,S)-¹⁸F-fluoro-6-thia-heptadecanoic acid; FTO, 18-¹⁸F-fluoro-4-thia-oleate; FTP, 16-¹⁸F-fluoro-4-thia-palmitate, FDG, fluorodeoxyglucose; STTR2, somatostatin receptor 2, TSPO, translocator protein, FAPI, fibroblast activating protein inhibitor

The PET detection scheme permits quantification of radioactivity and thus, the process being imaged within the field of view. As a consequence, metabolic processes of interest can be quantified in nmol/g/min. This quantitative capability combined with high sensitivity are major drivers for PET as a molecular imaging tool. Another advantage of the PET approach is that many of the radiotracers employ radionuclides that have a physical half-life (¹⁸F – ~110min and ⁶⁴Cu-~13hrs) and exhibit kinetics that permit whole body imaging to detect systemically relevant targets and the potential for radiotracer delivery to imaging sites that are remote from the site of production significantly easing the complexity and expense of performing PET.(47, 48)

The major physical limitation of PET is low spatial resolution. However, this is mitigated by hybrid technologies such as PET/CT and PET/MR where more accurate radiotracer localization can be achieved. Other limitations include the relatively high cost of radiotracer production and imaging and the use of, albeit very low, ionizing radiation.

MR Methods

MRI

The ability of MRI to provide exquisite detail of cardiac morphology and function is well known and its role in evaluating patients with various forms of cardiomyopathy has been extensively reviewed so this aspect will not be further discussed.(69, 70) Given the importance of myocardial fibrosis and inflammation in diabetic cardiomyopathy and that they are an outcome of the metabolic remodeling in the diabetic heart, MR methods to measure these processes are discussed below.

Fibrosis:

MR with late gadolinium enhancement is the gold standard for the non-invasive measurement of replacement fibrosis that typifies scar formation resulting from myocardial infarction or injury. However, this approach may miss diffuse interstitial fibrosis that characterizes a host of cardiac diseases such as diabetic cardiomyopathy and which is particularly relevant because the process may be reversible.(71–73) Myocardial T1-mapping methods are used to measure tissue fibrosis using native (without gadolinium-based contrast agents) or post-contrast approaches. Native myocardial T1 reflects a composite signal from both the intracellular (predominantly myocytes) and the extracellular compartment (edema and fibrosis), thus its specificity for fibrosis is limited. Furthermore, the measurement is sensitive to many scanner parameters, so its standardization with healthy volunteers is recommended for each individual scanner and each specific imaging protocol. In addition, to these limitations, post-contrast native T1 measurements are further confounded by the type and dosage of gadolinium contrast used and physiological factors such as the hematocrit and renal clearance that alter the pharmacodynamics of the contrast agent. To circumvent these problems, the most widely accepted approach to determine interstitial fibrosis is measure the extracellular volume (ECV) based on the knowledge of the myocardial and blood T1 before and after administration of contrast agents as well as the patient’s hematocrit value. The ECV represents the entire interstitial space so it will contain not just the key fibrotic components but also non fibrotic components such as non-fibrotic proteins and other enzymes. In addition, there are small contribution from the microvasculature. Despite, these and other non-fibrotic components, measurements of ECV have been shown to be reasonable estimates of interstitial myocardial fibrosis.(74)

Inflammation:

T2-weighted imaging is the most commonly used MR approach to measure tissue inflammation. The imaging is an indicator of tissue water content which is increased during the inflammatory process. However, because inflammation frequently induces vasodilation and that the increase water content may reflect necrosis, additional imaging early (vasodilation) and delayed (necrosis) after gadolinium injection is frequently performed to further enhance the detection of inflammation.(75)

Common limitations of these various MR methods, include the standard contraindications to MRI, the risks of gadolinium based contrast agents in renal failure and the increasing awareness of gadolinium deposition in the brain.(76)

MRS

¹³C-MRS allows direct measurement of various metabolic pathways involved with in vivo processes. Unlike PET, it provides exquisite detail of the relevant pathways underlying metabolic processes of interest. However, at biological temperatures and field strengths attainable in clinical MR scanners, the difference in these populations is small, leading to a low sensitivity. Compared with radionuclide methods, ¹³C-MRS detects millimolar as opposed to nanomolar concentrations. In addition, the method suffers from concomitant limited spatial resolution, intravoxel signal contamination and long acquisition times. Thus, although recent studies suggest imaging of cardiac metabolism using ¹³C labeled agents is possible in intact animals, studies in humans are still not possible. Currently, only ³¹P and ¹H have been widely used for in vivo clinical cardiac examinations focusing on myocardial energetics (³¹P) and lipid accumulation (¹H).(77, 78)

HP-MRS

Hyperpolarizarion was developed to overcome the low sensitivity of conventional ¹³C-MRS. In brief, the principle of hyperpolarization is based on the concept that low temperature and high magnetic field can temporarily align all nuclei in the same direction, increasing the polarization and thus the signal of a particular compound. This high level of polarization can be transferred to ¹³C-labeled tracers, increasing their MRS signals by ~10,000 fold. Hyperpolarized ¹³C technology then takes advantage of the chemical specificity of ¹³C-MRS to probe multiple specific pathways plus the improved sensitivity afforded by hyperpolarization.(79, 80) Examples are summarized in Table 2. Currently, 1-¹³C-pyruvate has been used to image the human heart.(81) Exploiting the acceleration glycolysis by pro-inflammatory cells, the same method has been shown to detect activation and resolution of the innate immune system following myocardial infarction in relevant pre-clinical models.(82)

Table 2:

Examples of ¹³C-labeled Compounds for Cardiac Metabolism.

Metabolic Process	13C-labeled compound
Oxygen Consumption	[1-13C]acetate, [2-13C]pyruvate(83)
Short Chain Fatty Acid Metabolism
Acetyl-Carnitine metabolism	[1-13C]acetate(84)
Butyrate oxidation	[1-13C]butyrate(85)
pH Determination
CO2 – bicarbonate ratio	[1-13C]pyruvate(86)

However, there are several challenges to the routine use of this technology for either pre-clinical or human studies.(87, 88) For example, the T1 of the various metabolic tracers tends be quite short (e.g., T1 for 1-¹³C-pyruvate is ~40 sec) which means useful imaging times of only 1–2 minutes are possible. This greatly limits the depth of interrogation of metabolic pathways and places significant demands on the synthesis/imaging process to ensure a successful study. Because the method does not provide quantification of the delivery of the metabolic tracers (a function of blood flow and concentration), quantification (in nmol/g/min) of the various metabolic pathways is not possible. Finally, because super-physiological concentrations of the various HP-MRS tracers are used, their administration will result in studies at non-steady state metabolic conditions. These high concentrations also increases the safety concerns for human use.

New Insights on Diabetic Cardiomyopathy Gleaned From Metabolic and Molecular Imaging

Detailed below are the potential research uses of imaging technologies to study diabetic cardiomyopathy in both pre-clinical disease models and in humans with diabetic cardiomyopathy. In addition, the potential of these various imaging biomarkers to risk stratify patients with diabetic cardiomyopathy is discussed.

Pre-clinical imaging

Small animal imaging has helped clarify the mechanisms responsible for the metabolic alterations that occur in DM. For example, PET studies with 1-¹¹C-palmitate and FDG in genetic models demonstrate that PPARα and PPARβ/σ drive different metabolic regulatory programs in the diabetic heart.(89, 90) In a more clinically relevant model of T2DM, the Zucker-Diabetic-Fat rat, myocardial glucose uptake correlates directly and closely with GLUT 4 gene expression, demonstrating the quantitative capability of the technique.(91) PET measurements in the same model demonstrated a decline in myocardial glucose uptake and an increase in FA uptake and oxidation and that the metabolic changes were driven predominantly elevated plasma FA levels. The metabolic adaptations were associated with a decline in insulin-mediated phosphorylation of Akt which is indicative of reduced insulin action and an increase in abundance at the sarcolemma of the fatty acid transporter CD36, respectively.(92) Small-animal PET has been used to assess changes in myocardial substrate metabolism after monotherapy with either metformin or rosiglitazone (Figure 2). Metformin had no effects on myocardial glucose utilization and FA oxidation. In contrast, the PPAR-γ agonist rosiglitazone not only increased myocardial glucose utilization but reduced myocardial FA oxidation, thus reversing the metabolic phenotype of the diabetic heart and resulting in significant fractional net gain in glucose utilization.(93) Similarly, PET with FDG in the Goto-Kakizaki (GK) rat model of nonobese T2DM, demonstrated that in addition to reducing hyperglycemia, the GLP-1 agonist sitagliptin increased myocardial glucose utilization that was paralleled by an improvement in function. The changes in the PET-derived measurements in myocardial glucose uptake appeared to reflect sitagliptin stimulation of sarcolemmal translocation of the glucose transporter-4 and attenuation of the fatty acyl translocase/CD36.(94)

Figure 2:

Myocardial FA utilization measurements. (A) Myocardial FA utilization uptake rate (MFAU_UpR), (B) Myocardial FA esterification uptake rate (MFAE_UpR) and (C) Myocardial FA oxidation uptake rate (MFAO_UpR) in untreated (ZDF; N=6), Metformin-treated (ZDF+MET; N=6) and Rosiglitazone-treated (ZDF+ROSI; N=6) ZDF rats at week 14 (W14) and week 19 (W14). MFAE_UpR, MFAO_UpR and MFAU_UpR represents the intrinsic capacity of the heart to oxidize and utilize FAs, respectively, independent of the concentration of free FAs in plasma ([FFA]_P) while MFAO is derived by the relation MFAO=MFAO_UpR*[FFA]_P. *, denotes that the treatment is significantly better than no treatment; †, denotes that the Rosiglitazone treatment is significantly better than Metformin treatment. A P<.05 was considered significant. All resulted are presented as mean ± 1 SD. Reproduced with permission.(93) Illustration credit: Ben Smith.

Studies with HP-MRS have provided clarity on the perturbations in metabolic pathways more downstream to those detected with PET. Pyruvate dehydrogenase (PDH) activity is central to maintaining flexibility in myocardial substrate usage and is well-documented to be reduced in DM. In models of both T1DM (streptozotocin) and T2DM (low dose streptozotocin with high-fat feeding), HP-MRS using 1-¹³C-pyruvate confirms this observation demonstrating the potential to now perform such measurements noninvasively.(95, 96) Indeed, these studies have demonstrated the decline in PDH activity is paralleled by a decline LV diastolic function. Of interest, in a mouse model of DM with a phenotype of hyperglycemia and hypoinsulinemia without dyslipidemia (T1DM-like), the reduction in PDH activity was paralleled by a decrease in proteins involved in glucose oxidation and increase in various proteins involved in FA metabolism.(97) These data suggest the metabolic remodeling in the diabetic cardiomyopathy could be driven primarily by chronic hyperglycemia and hypoinsulinemia and occur independently of an increase in plasma FA levels. Moreover, serial HP-MRS confirms PDH as a potential therapeutic target where its stimulation with dichloroacetate increases metabolic flexibility that is accompanied by improved LV diastolic function.(95)

In addition to carbohydrate and long-chain FA usage, ketone body metabolism is now being evaluated in the diabetic heart. HP-MRS using 3-¹³C-acetoacetate in the GK rate, demonstrated increased ketone body oxidation (Figure 3). Consistent with this finding was the demonstration of increased activity of succinyl-CoA:3-ketoacid-CoA transferase activity, the rate-limiting enzyme of ketone body utilization. Moreover, the increase in ketone body oxidation was associated with myocardial hypertrophy and systolic dysfunction suggesting the enhanced ketone body oxidation may be a contributing factor to the diabetic cardiomyopathic process.(98) Interestingly, six months of treatment of diabetic obese rats with spontaneously hypertensive heart failure with the SGLT2 antagonist, empagliflozin, decreased obesity, lowered blood pressure, reduced blood glucose and insulin levels, and increased circulating FA and fasting blood β-hydroxybutyrate levels. However, despite the increase in plasma ketone body levels, empagliflozin lowered myocardial ketone body utilization while not affecting myocardial PDH flux, the level of LV hypertrophy or fibrosis and cardiac function.(99) These data suggest there is still significant uncertainty on the role of ketone body metabolism in diabetic cardiomyopathy and the mechanisms by which SGLT2 antagonists elicit their beneficial effects.

Figure 3:

Myocardial ketone body utilization in GK rats. The detection of [3-¹³C]acetoacetate, [1-¹³C]acetoacetate, [5-¹³C]glutamate, and [1-¹³C]acetylcarnitine over 2 minutes upon [3-¹³C]acetoacetate injection in (a) controls and (b) GK rats. (c) Representative cardiac ¹³C MR spectra from a control and a GK rat. The quantification of (d) [5-¹³C]glutamate + [1-¹³C]acetylcarnitine, (e) [5-¹³C]glutamate, (f) [1-¹³C]acetylcarnitine, and (g) [3-¹³C]β-OHB. Metabolic conversion rates for (h) [3-¹³C]acetoacetate to [5-¹³C]glutamate exchange and (i) [3-¹³C]acetoacetate to [1-¹³C]acetylcarnitine exchange. Data are means ± SD (except for (a) and (b): mean ± SEM), and normalized to [3-¹³C]acetoacetate (controls n = 10, GK n = 9; except for (i) GK n = 8). AcAc: acetoacetate, acc: acetylcarnitine, cit: citrate, glu: glutamate. *P < 0.05, **P < .01, ***P < .001 vs. controls. Reproduced with permission.(98)

To demonstrate the potential of multi-modality imaging in pre-clinical models of DM, serial PET with FDG (glucose uptake), MRI (LV mass and function), ¹H-MRS (lipid accumulation) and ³¹P-MRS (energetics) was performed in non-diabetic (db/þ) and diabetic (db/db) mice before and over 12 weeks after the induction to pressure overload by transverse aortic constriction. The in vivo measurements were complemented with ex vivo high-resolution respirometry, proteomics, western blotting, and quantitative PCR to elucidate the underlying molecular pathways. They showed the development of heart failure in nondiabetic mice was correlated with an increase in LV mass and increased myocardial glucose uptake, which preceded the decrease in cardiac energy production. In contrast, at baseline, diabetic mice showed normal cardiac function, higher lipid content and mitochondrial capacity for FA oxidation, and lower glucose uptake, and energetics. Interestingly, aortic constriction resulted in only a mild reduction in LV cardiac function and a mild increase in LV mass, no change in myocardial lipid and normalization of glucose uptake, and cardiac energetics.(100) Taken in sum these observations suggest the cardiac metabolic adaptations in diabetic mice seem to prevent the heart from failing upon pressure overload, suggesting that restoring the balance between glucose and FA utilization may be beneficial for cardiac function.

In addition to measuring LV morphology/function, myocardial lipid and metabolism, MR tools have been used to clarify the role of fibrosis in diabetic cardiomyopathy. Close correlations were observed between native T₁ mapping, postcontrast T₁ mapping, and ECV values and myocardial fibrosis determined by Masson staining of tissue in alloxan treated rabbits.(101) In contrast, precontrast T₁ values showed no correlation with myocardial fibrosis measurements. Moreover, in the same model, the increase in ECV was paralleled by a decline in diastolic function. These observations were subsequently verified in a non-human primate model of spontaneous T2DM, presaging the use of MR to measure myocardial fibrosis in humans with diabetic cardiomyopathy.(102)

Human Imaging

By dint of the preceding discussion there are numerous imaging tools available to investigate key aspects of the diabetic cardiomyopathy. The need to perform these studies in humans becomes increasingly important given the uncertainty over the relevance of the various pre-clinical models of diabetic cardiomyopathy to faithfully recapitulate what is seen in diabetic patients. Moreover, the desire to develop and validate biomarkers that more accurately and precisely risk stratify diabetic cardiomyopathy patients or can serve as a valid endpoints to guide the development and rationale implementation of novel therapeutics further highlights the need for these imaging tools.

The results of several imaging studies in humans have generally reproduced the observations from pre-clinical studies and as a result greatly expanded our understanding of the chronic metabolic adaptations in the diabetic heart. In nonobese patients with T1DM, studies with PET and 1-¹¹C-palmitate and 1-¹¹C-glucose suggest the higher levels of in FA uptake and oxidation compared with non-diabetics are primarily due to increased plasma FA levels. In contrast, the reduced glucose uptake observed in the T1DM patients reflect decreased glucose transport mechanisms.(103) In addition, the metabolic fate of extracted glucose is impaired in nonobese T1DM with reduced rates of glycolysis and glucose oxidation which become more pronounced with increases in cardiac work induced by dobutamine.(104) However, the myocardium in T1DM patients is responsive to changes in plasma insulin and FA levels but at a cost. To achieve the same level of glucose uptake and glucose oxidation compared with non-diabetics, higher insulin levels are needed in T1DM patients, consistent with myocardial insulin resistance. Similarly, although myocardial FA uptake is increased in response to higher FA plasma levels, a greater fraction of the extracted FA is esterified potentially leading to increased lipid accumulation.(105)

Results of imaging studies in patients with T2DM that were composed of various combinations of PET, ¹H-MRS, ³¹P-MRS, MRI, and echocardiography have yielded remarkably consistent results that in general, parallel pre-clinical observations and support the mechanistic framework for diabetes-induced chronic metabolic-functional adaptations. It appears a systemic environment typified by reduced whole-body insulin sensitivity and increased plasma FA and triglyceride levels drives a PET-derived myocardial metabolic profile that includes increased FA uptake and oxidation and reduced glucose uptake, the magnitude of which is dependent upon whether the measurements are performed in the fasted state or during insulin clamp.(106–109) In general, results of ¹H-MRS studies demonstrate an increase in myocardial lipid content and ³¹P-MRS studies demonstrate a decline in energetics that are associated with LV diastolic and in some cases systolic dysfunction.(110–112) Indeed results from a recent study using HP-MRS with [1-¹³C-pyruvate in patients with T2DM and diastolic dysfunction would suggest that as is observed in experimental models of DM, these upstream abnormalities in myocardial substrate metabolism result in reduced PDH flux that is paralleled reduced myocardial energetics and increased lipid accumulation (Figure 4).(81)

Figure 4:

(Left) Representative examples of hyperpolarized MR spectra from both a healthy control and a subject with T2DM in both the fasted and fed states (Control; A & C, T2DM; D & F), with ¹³C containing downstream metabolites labelled (A). The 1-¹³C- bicarbonate resonance is visibly reduced in the subject with T2DM with increases seen during feeding in both controls and subjects with T2DM. Time courses of the normalized signal amplitudes of downstream ¹³C-labelled metabolic products of administered 1-¹³C-pyruvate (shown in blue), in both a control and a subject with T2DM are also shown (B & E). (Right) Plots of flux data for each metabolic product of administered 1-¹³C-pyruvate. Controls (Fasted (blue); N=5 and Fed (red); N=2) and T2DM (Fasted; N=5 and Fed; N=3). Flux through PDH (Bicarbonate) is reduced in the fasted subjects with T2DM (N=5) (p=.013, A), with increases seen during feeding (N=3) (p<.001, E). Levels of 1-¹³C-lactate were significantly higher in the hearts of people with T2DM (p<.001, B) with no change observed upon feeding (F). The ratio of bicarbonate and lactate was significantly lower in the subjects with T2DM (p<.001, C) and was elevated by feeding (p<.001, G). No significant differences in 1-¹³C-alanine were seen across all injections (D and H). ‘x’ indicates the data point excluded as an outlier. † p<.05 in subjects with T2DM vs. controls, * p<.05 in fasted subjects vs. fed, ‘x’ indicates the data point excluded as an outlier. Reproduced with permission.(81)

It appears changes in the myocardial metabolic-functional relationship parallels the progressive worsening in the systemic profile as one transitions from a lean condition, to obesity without diabetes, to concomitant obesity.(107, 109) It also appears the well-known sexual dimorphism in myocardial metabolism that exists in lean subjects and obese non-diabetics, is also present in patients with concomitant obesity and T2DM (Figure 5). In these patients, diabetic men appear to have a greater impairment in myocardial glucose uptake and oxidation compared with diabetic women. In contrast, diabetic women appear to have more pronounced augmentation in myocardial FA metabolism and a greater propensity for lipid accumulation compared with diabetic men.(107, 113) These results are intriguing given the greater susceptibility of women with T2DM to develop heart failure and exhibit a poorer prognosis when compared with male diabetics.(114–116)

Figure 5:

Sexual dimorphism in myocardial metabolism. A, Impact of sex and DM on myocardial FA oxidation, esterification, and %oxidation suggesting more pronounced effects in women. Obese men (N=10), obese women (N=29), T2DM men (N=12) and T2DM women (N=21). After adjustment for age, ^aP=NS, ^bP=.0060 and ^cP=.03. Subgroup analyses not significantly different for FA oxidation; for FA esterification, ^dP<.06 for diabetic men vs. women. Measurements of myocardial glucose utilization, glycogen deposition, glycolysis, and oxidation measured by PET with 1-¹¹C-glucose in (B) lean, obese, and T2DM men (N=4, 9 and 32, respectively) and (C) women (N=6, 17 and 40, respectively). Data suggest that men exhibit a greater decline in glucose metabolism compared with women as one transitions from lean to obese to T2DM. Reproduced with permission. (107, 113)

The effects of anti-diabetic therapies in humans on myocardial metabolism have been evaluated by PET. For example, in T2DM patients with LV systolic dysfunction the effects of the GLP-1 agonist, exenatide on LV function (MRI), MVO₂ (PET with 1-¹¹C-acetate) and myocardial blood flow (PET with ¹⁵O-water) were compared with insulin glargine and with non-diabetic controls.(117) Of note, LV myocardial efficiency defined as MVO₂/ stroke volume was not different between T2DM patients and controls. No differences in any of the parameters were observed with treatment. Although the results may have been obscured by a limited sample size, the data does suggest alterations in myocardial metabolism may not be contributing to any of the beneficial CV benefits of this class of anti-DM therapy. The differential effects of the PPARγ agonists pioglitazone and rosiglitazone, and the biguanide, metformin on myocardial glucose and FA metabolism have been assessed using PET with FDG or 1-¹¹C-glucose and 1-¹¹C-palmitate, respectively. Treatment with either rosiglitazone or pioglitazone results in an increase in insulin-stimulated myocardial glucose uptake. In contrast, insulin-stimulated myocardial glucose uptake is either unchanged or reduced with metformin therapy.(118, 119) In male diabetics, pioglitazone therapy does not have a significant effect on myocardial FA metabolism.(118) The lack of a decrease in FA metabolism and variable response in myocardial glucose uptake with these therapies may be explained by different responses in men and women. In men, metformin alone decreases whole-body FA clearance, which results in increased plasma FA levels, myocardial FA uptake and oxidation, and lower myocardial glucose uptake. In women, myocardial glucose uptake is increased. When metformin and rosiglitazone are combined, women exhibit increased whole-body FA clearance, which decreases plasma fatty acid levels and myocardial FA uptake. This effect is much less pronounced in men.(120) Thus, in diabetes, different therapeutic regimens impact myocardial metabolism in a sex-specific manner. Of note for all of the studies above, metabolic response could not be predicted by changes in the plasma glucose or HgBA1_C levels. Although requiring further evaluation in larger studies, these observations suggest metabolic imaging may be used to follow the effects of therapies designed to alter myocardial substrate metabolism in patients with diseases such as DM where more readily available clinical parameters are not predictive of a therapeutic response. Moreover, they suggest more personalized approaches that incorporate patient sex may be useful in designing diabetic therapies.

There is a rapidly growing experience with using MR measurements of fibrosis (both replacement/scar and diffuse interstitial) to study the diabetic cardiomyopathy in humans. For example, MR LGE measurements demonstrated a progressive increase in replacement/scar between healthy adult controls, pre-diabetics and those with T2DM with preserved LV function.(121) This increase in scar was paralleled by an increase in LV remodeling. Of interest there was not increase in ECV suggesting interstitial fibrosis was less important at this early stage. In contrast, in adolescents MR measurements of ECV progressively increased as one compared health controls to obese/non-diabetics to those with both obesity and T2DM.(122) Indexed measures of LV structure and function did not differ significantly between the groups suggesting fibrosis formation was preceding LV structure/function changes. This finding has been confirmed in adults with either T1DM or T2DM.(123, 124) Consistent across all of these studies as well as others, is the association of ECV values with plasma C-reactive protein and HgBA1_C levels suggest a relationship between fibrosis formation and systemic inflammation and glycemic control.(123–125)

The effects of new anti-diabetic therapies in humans on myocardial fibrosis have also been evaluated by MR. For example, in a double-blind, randomized controlled trial of T2DM patients of South Asian descent, with or without coronary artery disease, 6 months of treatment with the GLP-1 agonist, liraglutide had no effect on LV diastolic and systolic function, aortic stiffness, myocardial triglyceride content, or ECV when compared with placebo.(126) The results imply that liraglutide does not impact CV remodeling in diabetic cardiomyopathy and that perhaps it’s observed CV benefit may be related other mechanisms such as its effects on atherosclerosis. MR measurements of cardiac function, structure, myocardial adiposity, and diffuse fibrosis in patients with T2DM have also been performed before and after 6 months of empagliflozin.(127) On average no differences were evident after treatment. However, on multiple linear regression it appears the beneficial treatment effects are present when baseline conditions are considered. Although, further and larger studies are needed, it appears this imaging approach may useful in helping to unravel the reasons why this class of drugs exhibits CV benefit in patients with T2DM.

The presence of fibrosis appears to imaging biomarker of increased CV risk. For example in patients without documented coronary artery disease, the presence of LGE is a predictor of CV events in patients with impaired fasting glucose and when present confers the same survival as is seen in T2DM with LGE.(128) Indeed, in T2DM patients subclinical replacement/ scar fibrosis is fairly common (~30%) and when present portends a CV event rate similar to that observed in T2DM with documented myocardial infarction.(129) Similar observations have been made with respect to ECV. In a large, consecutive cohort of patients referred for CMR, DM is associated with higher ECV and the increased ECV is associated with mortality and/or incident heart failure admission in individuals with DM after adjusting for key baseline differences (Figure 6). (130) Of interest, interstitial fibrosis measured by ECV appeared to be reduced through inhibition of the renin-angiotensin system. Taken in sum, these data highlight the potential of fibrosis imaging for risk stratification and perhaps as means to identify viable therapeutic targets.

Figure 6:

Kaplan-Meier Curves for 231 individuals with diabetes and 945 individuals without T2DM. Extracellular matrix expansion in myocardium quantified by extracellular volume (ECV) fraction is associated with increased risks of: death or heart failure admission (top panel); heart failure admission ignoring or censoring for death (middle panel); or all-cause mortality (lower panel). Event rates were higher for those with T2DM. Reproduced with permission.(130)

Future Directions – The Need for Systems Biological Approach

As mentioned from the outset, diabetic cardiomyopathy represents a local response to systemic stimuli.(16, 38) Key stimuli include insulin resistance with changes to the plasma substrate environment, activation of the renin-angiotensin and sympathetic nervous systems and a pro-inflammatory state. The local or myocardial response is a loss in metabolic flexibility reflected in an overdependence on FA as the primary energy source limiting the hearts’ ability to alter substrate metabolism in response to varying physiological and metabolic conditions. These chronic adaptations stimulate a host of effectors that both further exacerbate the upstream perturbations in substrate metabolism and accelerate processes detrimental to cardiomyocyte health. There are numerous imaging tools that can characterize these various processes, however as evident in this review, many have not been employed to study diabetic cardiomyopathy. Of those that have been used, they have been frequently employed in relative biological, physiological or anatomical isolation that may not properly capture the multi-dimensional nature of diabetic cardiomyopathy process. Moving forward a multi-plexed imaging approach that combines imaging several key processes and contextualizes the readouts with the systemic environment will likely be needed. From an imaging perspective to successfully implement this approach will require the use of multi-modality imaging that utilizes the relative imaging strengths of each technology, optimizes imaging protocols and image co-registration, permits whole-body imaging and minimizes potential side effects. Two examples of new instrumentation to help in this regard PET/MR and the new total-body EXPLORER PET/CT system. (131, 132) In the case of the former, synergy would be facilitated in the acquisition of the different but complimentary information with PET and MR. For example, it might now be possible to determine the extent to which myocardial interstitial fibrosis (MR–ECV) is driven by inflammation (PET-⁶⁴Cu-DOTA-ECL1i) and is potentially reversible (PET-¹⁸F-FAPIs). In the case of the latter, it may now be possible to obtain quantitative PET information on metabolic and molecular processes relevant to DM and diabetic cardiomyopathy throughout the body in a short period of time at radiotracer doses that greatly reduce radiation exposure. These readouts can then be combined with measurements of whole-body substrate metabolism, -omic assays of blood, tissue and microbiome, pro-inflammatory cytokines and monocyte subset measurements and determinations of sympathetic nervous system activation. This integration will permit a more complete understanding of the mechanisms underlying diabetic cardiomyopathy and how these mechanisms can be exploited to improve the management of the diabetic patient.

Nonstandard Abbreviations and Acronyms:

DM

Diabetes Mellitus

T1DM

Type-1 Diabetes Mellitus

T2DM

Type-2 Diabetes Mellitus

PET

Positron Emission Tomography

MRI

Magnetic Resonance Imaging

MRS

Magnetic Resonance Spectroscopy

HP-MRS

Hyperpolarized Magnetic Resonance Spectroscopy

FA

Fatty Acid

PPARa

Peroxisome Proliferator-Activated Receptor Alpha

PPARg

Peroxisome Proliferator-Activated Receptor Gamma

PGC-1

PPAR Gamma Coactivator-1alpha

GLUT4

Glucose Transporter Type 4

SGLT2

Sodium/Glucose Cotransporter 2

LV

Left Ventricular

MVO₂

Myocardial Oxygen Consumption

TGF-β

Transforming Growth Factor Beta

ECV

Extracellular Volume

ZDF

Zucker Diabetic Fat Rat

GK

Goto-Kakizaki

PDH

Pyruvate Dehydrogenase