Author + information
- Jason B. Lindsey, MD and
- Steven P. Marso, MD, FACC⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Steven P. Marso, Clinical Scholar, Mid America Heart Institute, Associate Professor of Medicine, University of Missouri-Kansas City, 4401 Wornall Road, Kansas City, Missouri 64111
Diabetes mellitus (DM) has long been recognized as a risk factor for cardiovascular disease including coronary artery disease (CAD) and heart failure (HF) (1). Three decades ago, Rubler et al. (2) first described a cohort of diabetic patients with HF but normal coronary arteries, later termed “diabetic cardiomyopathy.” Diabetic cardiomyopathy (DCM) can be defined as ventricular dysfunction that is present independent of CAD and hypertension (3). The existence of DCM is supported by population-based studies (4) and by observations that in comparison with age-matched individuals, diabetic subjects have an increased frequency of HF that persists even after adjustment for CAD (1). Although many advocate DCM as a clinical phenotype, there is no specific mention of it in the most recent American College of Cardiology/American Heart Association heart failure guidelines (5), and there are no established criteria for its diagnosis. However, the fact remains that those with DM have an increased risk for development of HF and a worse prognosis from HF compared with those without DM, yet controversy remains as to the pathobiologic mechanisms that contribute to the observed myocardial dysfunction (3).
Myriad hypotheses have been proposed to explain the pathophysiologic derangements that may lead to myocardial dysfunction in those with DM (3). Candidate mechanisms include altered myocardial calcium homeostasis, mitochondrial dysfunction, oxidative stress, direct glucotoxicity, myocardial fibrosis, perturbed myocardial substrate use, and myocardial lipid accumulation (4). All are biologically plausible and most are supported by animal models. Much recent attention has focused on the role of myocardial steatosis in diabetic myocardial dysfunction (6). During normal physiologic conditions, a normal cardiomyocyte generates 70% of its adenosine triphosphate (ATP) from the oxidation of fatty acids (FAs). In patients with DM, glucose uptake, glycolysis, and insulin function are deranged, resulting in enhanced lipolysis and release of FAs from adipose tissue. Excess myocardial FAs, which remain after cardiomyocyte energy production, are sequestered as triglycerides (TGs) to provide cardiomyocytes a reserve of FAs. In the normal heart, lipolysis and lipogenesis are balanced, resulting in constant levels of intracellular TGs. Under conditions of excess FA production, such as in obesity or DM, the balance is in favor of excess intracellular TG accumulation. The hypothesis of myocardial TG accumulation has been supported by animal models that disrupt normal myocardial lipid metabolism via several pathways and result in TG accumulation and diastolic dysfunction (7). In humans, there are emerging yet limited data for myocardial lipid accumulation. For example, histopathology of explanted hearts among diabetic transplant recipients has shown myocardial lipid accumulation (8). Recently, using magnetic resonance spectroscopy (MRS), McGavock et al. (6) provided provocative evidence that, indeed, myocardial lipid accumulation occurs in vivo in humans, specifically among individuals with DM and in those with insulin resistance.
In this issue of the Journal, Rijzewijk et al. (9) provide further compelling evidence that myocardial lipid accumulation occurs in diabetic subjects and that it is independently associated with impaired left ventricular (LV) diastolic function. The study consisted of 38 male diabetic subjects and 28 age- and body mass index (BMI)-comparable controls. Overall, subjects had uncomplicated DM; all had hemoglobin A1c values <8.5%, with no evidence of cardiovascular disease or microvascular complications. Importantly, the authors excluded those taking lipid-lowering therapies and those with evidence of myocardial ischemia (as detected by dobutamine stress echocardiography). The investigators then performed proton MRS on all subjects using a previously validated protocol, similar to the MRS methods used by McGavock et al (6). MRS allows for in vivo quantification of myocardial TG content by the generation of distinct spectral peaks (corresponding to protons in lipids), which are easily separated from the spectral peaks of water and other tissue metabolites. Myocardial TG content was then expressed as a percentage relative to myocardial water content.
Compared with controls, those with DM had significantly higher myocardial TG content. After adjustment for the presence of DM, myocardial TG content correlated with glucose concentrations, visceral adiposity, and hepatic TG content; however, plasma TG levels did not correlate with myocardial TG content.
In an effort to evaluate myocardial function in both groups, the authors used several magnetic resonance imaging indexes of diastolic function including the ratio of maximal LV early peak filling rate and the maximal LV atrial peak filling rate (E/A) and E peak deceleration. Iterative multivariable analysis using E/A and E peak deceleration as the outcome variables demonstrated that myocardial TG content was associated with these parameters of diastolic function.
Rijzewijk et al. (9) should be complimented on this important contribution, which expands the findings of McGavock et al. (6) and supports the concept of cardiac lipotoxicity. However, certain limitations to the present study should be considered. There is controversy as to the most valid qualitative or quantitative magnetic resonance imaging measures of diastolic function. In the current study some (such as E/A and E peak deceleration time), but not all, indexes of diastolic function were abnormal. Importantly, estimations of LV filling pressures (E/Ea) were not abnormal. These discrepancies highlight the difficulty in quantifying diastolic function. Validation by measurement of echocardiographic indexes of diastolic function or the use of myocardial tagging techniques with magnetic resonance imaging would have strengthened the current observations and could have been performed concurrently. Although the presence of diastolic dysfunction is a risk factor for developing diastolic HF, the presence of asymptomatic diastolic dysfunction should not be equated with the presence of DCM.
In this study, properly accounting for the effect of obesity on the exposure or the outcome variables is important. The authors attempted to adjust for the influence of obesity on myocardial TG content by using controls and diabetic subjects with comparable BMI. Although subjects were similar by 1 parameter, BMI, the striking difference between the groups in visceral adiposity cannot be discounted. Indeed, visceral fat correlated with myocardial TG content and BMI did not. In the models for diastolic function, visceral adiposity was independently associated with diastolic dysfunction but DM was not an important determinant. In total, this allows for some observations: 1) the presence of central adiposity may be more important than DM in determining myocardial TG content; and 2) measures of central adiposity (waist/hip ratio) could serve as an inexpensive and indirect surrogate of myocardial TG content. Importantly, these hypotheses are reinforced by the observation of McGavock et al. (6) that visceral fat content was the only independent determinant of myocardial TG content.
There is evidence from another chronic disease state that suggests visceral adiposity rather than diabetes per se is the important determinant of intramyocellular lipid accumulation. Compared with age-matched controls, nondiabetic human immunodeficiency virus (HIV)-lipodystrophy subjects had significantly increased skeletal TG content (10). Those with HIV-lipodystrophy had the same BMI as controls but had increased central adiposity and insulin resistance. An independent association between indexes of central adiposity and skeletal TG content was observed. Again, this emphasizes the importance of fat redistribution and insulin resistance in the process of intramyocellular TG accumulation, which appears to exist even in the absence of DM. Rijzewijk et al. (9) do not report the prevalence of insulin resistance in the DM subjects but, based on insulin levels, we estimate that a majority of these subjects had insulin resistance.
The findings from the current study suggest that myocardial TG accumulation only partially explains the observed diastolic dysfunction. Evaluation of the multivariable models reveals that although myocardial TG content was associated with diastolic dysfunction, the r2 of model #6 was only 0.47 and the individual contribution of myocardial TG content was modest. This suggests there are other unmeasured variables that contribute to diastolic dysfunction. Also, the correlation analysis between E/A and E peak deceleration and myocardial TG content show that myocardial TG content is a poor predictor of diastolic function. Furthermore, it is intriguing that a seemingly insignificant myocardial TG content of ∼1% (absolute difference compared with controls of 0.3%) would result in or explain the observed diastolic dysfunction in this study. It should be considered that cardiac lipotoxicity could simply be a byproduct of other pathologic processes that coexist in the diabetic heart, and processes such as altered myocardial calcium homeostasis or myocardial fibrosis may play as important a role as cardiac lipotoxicity in the development of DCM (3).
The authors should be commended for their attempt to exclude individuals with myocardial ischemia by screening with dobutamine stress echocardiography (DSE). However, the negative predictive value of DSE is around 60% to 86% (11). Therefore, it is conceivable that some subjects may have had obstructive epicardial CAD. Also DSE does not reliably evaluate microvascular coronary disease that has been associated with diastolic dysfunction (12).
Due to the observational study design, the authors are correct in asserting that no conclusions about causality should be drawn from this study. The observation that myocardial TG content is associated with diastolic dysfunction, although important, does not imply cause and effect. Future study should focus on individuals with the clinical syndrome of diabetic cardiomyopathy to better determine the implications of the current findings.
The widespread clinical use of MRS is unlikely given the technical challenges involved in data acquisition, and the precision required in measuring myocardial steatosis is not insignificant given the observed <1% myocardial TG content reported by Rijzewijk et al. (9) and McGavock et al. (6). However, this study confirms that MRS will be a valuable research tool in understanding the cardiac pathology involved in obesity, insulin resistance, metabolic syndrome, and DM.
Is myocardial steatosis a modifiable condition? Administration of pioglitazone, a peroxisome proliferator-activated receptor-gamma agonist, to diabetic subjects resulted in a reduction of both myocardial and hepatic TG content (13). Due to the observed relationship between central adiposity and myocardial TG content, it can be postulated that lifestyle modification through exercise and weight reduction strategies could result in a reduction of myocardial TG content. This is a hypothesis that can and should be tested in future studies evaluating myocardial steatosis.
Despite these limitations, Rijzewijk et al. (9) have provided important translational research that supports the findings of animal model data and gives us further insight into the underlying pathophysiology of the diabetic heart. Although this study does not firmly establish a causative role for myocardial TG accumulation in diabetic cardiomyopathy, it adds additional supportive evidence, is consistent with preliminary evidence from McGavock et al. (6), and leads us to believe that there is an important role for additional studies with interventions aimed at reduction of myocardial TG content (thiazolidinediones or weight loss/exercise) and subsequent reassessment of myocardial TG content and parameters of diastolic dysfunction.
↵⁎ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
- American College of Cardiology Foundation
- Boudina S.,
- Abel E.D.
- Nichols G.A.,
- Hillier T.A.,
- Erbey J.R.,
- Brown J.B.
- Hunt S.A.
- McGavock J.M.,
- Lingvay I.,
- Zib I.,
- et al.
- Chiu H.C.,
- Kovacs A.,
- Blanton R.M.,
- et al.
- Sharma S.,
- Adrogue J.V.,
- Golfman L.,
- et al.
- Rijzewijk L.J.,
- van der Meer R.W.,
- Smit J.W.A.,
- et al.
- Torriani M.,
- Thomas B.J.,
- Barlow R.B.,
- Librizzi J.,
- Dolan S.,
- Grinspoon S.
- Galderisi M.