NHLBI WORKING GROUP ON
CELLULAR AND MOLECULAR MECHANISMS
OF DIABETIC CARDIOMYOPATHY

MEETING SITE: BETHESDA, MARYLAND

MINUTES OF AND CONSENSUS RECOMMENDATIONS:

This document is a review of the current state of knowledge and research in diabetic cardiomyopathy and recommendations for future research initiatives. It summarizes presentations and discussions held on the above date by a panel of scientific experts and NIH staff.

STATEMENT OF RATIONALE FOR THE WORKING GROUP

The development of heart failure in patients with diabetes mellitus is a problem of major clinical and epidemiologic importance. Heart failure can occur in patients with diabetes mellitus in the absence of coexistent hypertension and/or hemodynamically significant epicardial coronary artery stenoses (subsequently referred to as coronary artery disease). Clinical reports in the 1970s first described such patients, who were considered to have a diabetic cardiomyopathy, usually of a dilated cardiomyopathy with low ejection fraction. Of far greater epidemiologic importance, however, is the risk when diabetes mellitus is combined with coronary artery disease and/or hypertension. Diabetes mellitus patients with acute myocardial infarction have approximately twice the incidence of heart failure and death compared to non-diabetic patients. Diabetes mellitus and hypertension is also a potent combined risk factor for heart failure. Increased risks for heart failure and death in patients with diabetes mellitus are especially pronounced in women, African-Americans and certain Native Americans. Many diabetes mellitus patients without overt heart disease have evidence of subclinical myocardial abnormalities, including echocardiographic-Doppler patterns suggestive of slowed relaxation and/or decreased compliance, abnormal ultrasonic tissue characterization consistent with altered connective tissue, and abnormal left ventricular functional responses during exercise. It is not known whether these subclinical abnormalities are the substrate for the subsequent development of heart failure, nor what factor(s) are responsible for its pathogenesis.

Despite the obvious importance of diabetic cardiomyopathy, a modern understanding of this complex, and very likely multi-factorial, problem at the cellular and molecular level is lacking, especially in patients. In addition, there has been little emphasis on developing novel approaches to prevention and treatment.

The purpose of the Working Group was to discuss our current understanding of diabetic cardiomyopathy, especially at the cellular/molecular level, and identify the most promising and critical areas for future research efforts in the field.

SUMMARY OF PREVIOUS RESEARCH

Early clinical studies of patients with apparent diabetic cardiomyopathy were accompanied by very limited histologic examination of tissue obtained by endomyocardial biopsy. Findings included increased PAS+ connective tissue (mainly perivascular), basement membrane thickening (also with PAS+ material), and thickening of the walls of small arteries and arterioles. These abnormalities were not striking in relation to the severity of clinical disease. Replacement fibrosis in particular was not a prominent feature, arguing against myocardial necrosis as a major causative factor. At about the same time, the epidemiologic importance of diabetes mellitus as a risk factor for heart failure was first recognized, initially through observations made in the Framingham Study. Subsequent clinical research, most of which was accomplished in the 1980s through the early 1990s, included the following major contributions:

1. Careful pathologic examination of post-mortem tissue from patients with diabetes mellitus and advanced heart disease, typically in combination with coronary artery disease and/or hypertension. These patients had substantial cardiac hypertrophy and enlargement with major increases in connective tissue and replacement fibrosis. Although it was impossible to sort out their relative contributions, the concept that evolved from these studies was that diabetes mellitus and coronary artery disease/ hypertension were synergistic in increasing the risk of heart failure and death. This concept continues to the present day.

2. Epidemiologic studies of the natural history of acute myocardial infarction in patients with coexistent diabetes mellitus. These revealed remarkably high mortality and risk for development of heart failure. Women, African-Americans, and some Native Americans appear to be at especially high risk. Despite modern treatment, more recent studies continue to reveal excess morbidity and mortality from acute myocardial infarction in patients with diabetes mellitus.

3. Non-invasive studies utilizing echocardiography-Doppler, radionuclide ventriculography, and stress testing in patients with diabetes mellitus without overt heart disease. These revealed abnormalities of relaxation and/or compliance, ultrasonic tissue characterization, and abnormally small increases in contractile functional parameters during exercise in, surprisingly, large numbers of patients. The incidence of these abnormalities appeared to be independent of diabetes type (insulin dependent vs non-insulin dependent), but was positively correlated with the presence of other complications of diabetes mellitus such as neuropathy, nephropathy, and retinal vasculopathy.

These clinical observations stimulated work in experimental models of diabetes mellitus. By far the most extensively employed model was, and continues to be, streptozotocin (STZ)-induced diabetes mellitus in the rat. Rats given STZ for weeks to months develop hypoinsulinemic diabetes mellitus, consistent depression of contraction, and, most especially, relaxation. Effects on "passive" diastolic compliance have been variable. These contractile abnormalities were correlated with several alterations in excitation-contraction coupling, ion transport and exchange, including prolongation of the action potential, and depressed sarcoplasmic reticular calcium pumping and sarcolemmal Na/Ca exchange. Abnormalities of the contractile machinery have also been detected in mechanical studies in skinned strips, in addition to a V1 to V3 isomyosin switch.

STZ-diabetes mellitus is a reliable, well-characterized model. However, results are confounded by the fact that STZ is a general toxin which causes several other abnormalities in addition to diabetes mellitus, for example, hypothyroidism. Several other models of mild diabetes mellitus in rats, sometimes in combination with hypertension, have been studied, but in much more limited fashion. Alloxan is another toxin which has been used to produce diabetes mellitus, mainly in large mammals and more long-term preparations. Interestingly, these animals seem to have abnormalities of diastolic compliance in association with changes in the connective tissue matrix that are more prominent than defects in contractile performance.

Taken together, the clinical and experimental work have led to only limited insights into the mechanism(s) of diabetic cardiomyopathy. The direct effects of abnormal carbohydrate metabolism and excessive fatty acid oxidation have been implicated as an important proximate cause of diabetic cardiomyopathy in STZ treated rats. Free radical production with altered lipid content of membranes is one postulated mechanism. However, several other consequences of abnormal carbohydrate metabolism, for example, depressed pyruvate dehydrogenase activity and inadequate energy availability, have also been proposed to account for diabetic cardiomyopathy. Possible molecular/genetic mechanisms involved in diabetic cardiomyopathy in STZ-diabetes mellitus have in general not been well characterized. Microvascular disease and protein glycation/glycosylation, important mechanisms of complications in patients with diabetes mellitus, do not have particularly good counterparts in experimental models. The direct effects of hyperglycemia are also difficult to characterize in intact animals. Thus, the roles of these major components of diabetes mellitus in causation of diabetic cardiomyopathy have been particularly difficult to evaluate.

There has also been virtually no progress in characterizing the cellular and molecular features of diabetic cardiomyopathy in patients, much less their mechanisms. Defining cardiac changes in diabetic humans is obviously extremely important in light of the fact that diabetes is a chronic disease, not well represented by animal models. Moreover, it remains unclear whether there are differences between insulin dependent and non-insulin dependent forms of diabetes mellitus and what role insulin resistance plays in its development.

Finally, there remains some skepticism, particularly amongst clinicians, as to whether a distinct diabetic cardiomyopathy actually exists. The "conventional wisdom" is that if diabetic cardiomyopathy does exist its cause is diabetic vasculopathy, possibly of the microvasculature, despite the fact that there is no more support for this mechanism than for any of the other known complications of diabetes mellitus. The Working Group came to a rapid consensus that there is a wealth of evidence supporting the existence of diabetic cardiomyopathy as a major and distinct complication of diabetes mellitus, although there remain more questions than answers with regard to the underlying pathogenesis and pathophysiology.

AREAS OF CURRENT RESEARCH REVIEWED BY THE WORKING GROUP

Transformation of Normal to Abnormal Myocardium by Diabetes Mellitus: Role of Hyperglycemia and Protein Kinase C Activation
There are multiple manifestations of diabetes mellitus that could potentially cause diabetic cardiomyopathy, including hyperglycemia per se, altered carbohydrate metabolism (increased fatty acid oxidation, altered energy metabolism), extracellular and intracellular protein glycosylation, and microvascular disease. A central, unresolved issue in the pathogenesis of diabetic cardiomyopathy is the relative contributions of specific manifestations of diabetes mellitus (e.g., hypo- or hyperinsulinemia, hyperglycemia, increased fatty acid oxidation, dyslipidemia) vs secondary effects of these manifestations on cell signaling (e.g., protein kinase C activation) and/or gene expression. In rodent models of diabetes mellitus (STZ, BB/Wor), concentric remodeling and hypertrophy of the left ventricle may be observed in association with "diastolic dysfunction". Increased expression of atrial natriuretic peptide and upregulation of angiotensin converting enzyme activity suggest that this process shares a number of features with other causes of pathologic hypertrophy.

One potential mechanism of diabetic cardiomyopathy is activation of protein kinase C as a response to hyperglycemia. Protein kinase C very likely has a critical signaling role in cardiac hypertrophy by virtue of its effects on angiotensin converting enzyme activity, nitric oxide synthase, the MAP kinase/early gene response system, and contractile proteins. In rodent models of diabetic cardiomyopathy, total protein kinase C activity is modestly increased. Preliminary data with respect to isoform specificity suggest that the -isoform may be selectively upregulated in STZ-diabetes mellitus, but other isoforms are selectively elevated in other forms of heart failure. Upregulation occurs before rather than after the appearance of left ventricular dysfunction. A mechanism that may serve as another source of diacylglycerol and amplify protein kinase C upregulation is activation of tyrosine kinase secondary to protein kinase C activation of angiotensin converting enzyme and increased angiotensin II production. Tyrosine kinase also activates the MAP kinase system. Finally, although phospholipase D activity is found to be decreased in diabetic myocardium, phosphotidate phosphohydrolase activity is increased, providing another source of diacylglycerol.

Low-dose ethanol is known to inhibit diacylglycerol and tyrosine kinase while upregulating protein kinase C. Treatment of STZ-diabetes mellitus rats with low-dose ethanol appears to prevent the structural changes and protein kinase C upregulation induced by STZ. Thus, complex signaling relationships between protein kinase C, tyrosine kinase , angiotensin converting enzyme, and MAP kinase may be involved in both structural and biochemical changes during STZ-diabetes mellitus.

Exposure of isolated cardiac myocytes in a cell culture system to hyperglycemia is one approach to this aspect of diabetic cardiomyopathy, which allows stringent control of conditions and testing of discrete hypotheses. Maintaining cardiac myocytes in a hyperglycemic environment for only 1-2 days results in alterations in cellular electrophysiology and excitation-contraction coupling that are remarkably similar to those observed with short-term experimental diabetes mellitus. Specifically, the cellular action potential is prolonged, and cytosolic calcium clearing and relaxation are impaired. Suggested mechanisms include up-regulation of Ca²⁺ currents, down-regulation of K⁺ currents and Na/Ca exchange, and/or depressed sarcoplasmic reticulum calcium ATPase (SERCA-2) activation. The signaling pathway appears to involve N-linked glycosylation of certain proteins as well as elevated protein kinase C activity. Interfering with the processing of glucosamine or the inhibition of protein kinase C prevents these hyperglycemia-induced abnormalities. The effects of hyperglycemia on the myocyte do not seem to be caused by increased osmolarity, and they can be attenuated by certain antidiabetic agents (e.g., metformin and troglitazone) but not by others (e.g., glyburide). Here again, preliminary evidence implicates a role for hyperglycemia-induced up-regulation of protein kinase C activity, in addition to intracellular glycosylation and changes in intracellular calcium concentration. One or more of these factors may then trigger changes in gene expression and potentially a host of other alterations in protein function.

Activation and upregulation of the signal transduction pathway involving diacylglycerol and protein kinase C has also been observed in human cardiomyopathy unrelated to diabetes mellitus. In diabetes mellitus, in addition to hyperglycemia, protein glycosylation and oxidative stress may also activate this pathway. As indicated, protein kinase C influences a multiplicity of cellular functions, including production of extracellular matrix, cytokines, contractile protein function, calcium handling and contractility, and growth responses (early response genes, MAP kinase). In heart failure, there is evidence that the isoform, and possibly the isoform, of protein kinase C are preferentially increased. To assess the consequences of increased protein kinase C- activity, both a specific inhibitor and a transgenic mouse that overexpresses protein kinase C- have been studied. The inhibitor may be capable of preventing some of the complications of experimental diabetes mellitus. Overexpression of protein kinase C- results in development of a cardiomyopathy characterized by hypertrophy, loss of myocytes, fibrosis and markedly depressed contractile function. mRNA for collagen, TGF-, -myosin heavy chain, ANF and c-fos are all increased. Thus, activation of protein kinase C is of considerable interest as a potential mechanism of diabetic cardiomyopathy.

Transgenic Models of Altered Glucose Metabolism: Role of the GLUT4 Transporter
The glucose transporter protein, GLUT4, is normally the major glucose transport system. In insulin resistant diabetes mellitus, especially in association with obesity, reduced insulin transport into skeletal muscle and fat cells is the major metabolic defect. To better understand factors that regulate whole body insulin sensitivity and glucose transport, gene knockout technology has been employed to produce mice with one null allele of GLUT4 (+/-) and with a complete knockout of the GLUT4 gene (-/-).

GLUT4 +/- mice develop a pattern closely resembling human type 2 diabetes mellitus. As expected, these animals have reduced GLUT4 expression and glucose transport in muscle and adipose tissue, in addition to reduced glycolysis and glycogen synthesis. In an outbred genetic background, severe insulin resistance (quantified by euglycemic/hyperinsulinemic clamps) and a number of other pathologic changes noted in human type 2 diabetes mellitus are present, including hypertension, myocardial tissue changes consistent with a cardiomyopathy, and renal glomerular lesions. These changes occur in males only. Females are able to normalize glucose transport through as yet uncertain mechanisms. The phenotype in these animals is strongly dependent on age, progressing from normoglycemia/insulinemia, to normoglycemia/hyperinsulinemia, to hyperglycemia/insulinemia. Unlike human type 2 diabetes mellitus, GLUT4 +/- mice do not become obese, although their adipocytes are enlarged. Thus, the GLUT4 +/- mouse has considerable potential for understanding the development of diabetic cardiomyopathy in insulin-resistant diabetes mellitus independent of coincident obesity.

Interestingly, GLUT4 -/- mice, while exhibiting relatively subtle abnormalities of glucose and lipid metabolism including abnormal glucose clearance from the blood, do not become diabetic and do not display the sorts of diabetic complications seen in +/- mice. This is accounted for by upregulation of both the GLUT1 transporter and most likely another, as yet poorly characterized transport mechanism. Surprisingly, the -/- mice develop severe cardiac hypertrophy in the absence of hypertension, and yet have improved recovery of function following global ischemia-reperfusion when compared to the heterzygous animals. GLUT4 -/- animals, while not diabetic, may provide clues to interactions between abnormal transport mechanisms and stimuli for cardiac hypertrophy, as well as protection from ischemia.

Apoptosis and Ventricular Remodeling in Diabetes Mellitus
Multiple abnormalities of both excitation-contraction coupling and the contractile machinery have been documented in STZ-diabetes mellitus, but a role for apoptosis in this model of diabetes mellitus has not previously been sought. This is pertinent in view of increasing evidence of apoptosis in other forms of cardiomyopathy. In addition, the relatively modest histopathologic changes in STZ-diabetes mellitus suggest the possibility of cell dropout due to apoptosis.

In preliminary studies designed to assess whether significant amounts of apoptosis occur after one month of STZ-diabetes mellitus in the rat, heart to body weight ratios were increased, but absolute heart weight was decreased because body weight fell faster than heart weight. Morphologic analysis revealed a 10% decrease in myocyte mass despite a 13% increase in myocyte cell volume. This apparent inconsistency was due to a remarkable, 28% decrease in total myocyte number in the left ventricle. As in previous studies of STZ-diabetes mellitus, histologic changes were modest. Studies using both TdT and Taq analyses of apoptosis suggest that both calcium and pH dependent DNAases are activated, and that necrosis may also occur despite the absence of histologic sequelae. Parallel in vitro studies in cultured myocytes suggest that hyperglycemia per se may be a stimulus for increased apoptosis. These preliminary studies suggest a previously unappreciated role for apoptosis and cell necrosis in STZ-induced diabetic cardiomyopathy.

The high mortality after acute myocardial infarction in diabetic patients is generally considered to be caused by an increased risk of heart failure. One possible explanation is that myocardial necrosis is superimposed on a pre-existing cardiomyopathic process, which reduces cardiac reserve. Another is that the remodeling process after acute myocardial infarction differs in patients with diabetes mellitus. An interaction of post-acute myocardial infarction remodeling and apoptosis is of interest because factors known to be operative in diabetes mellitus such as altered insulin levels, oxidative stress and growth factors, also influence apoptosis.

This interaction has been studied in vitro using a tissue culture model in which neonatal myocytes are first exposed to high glucose concentrations for three days followed by a hypoxic period as a surrogate for ischemia. High glucose treatment results in markedly reduced apoptosis following hypoxia. This is at least in part a generic response to osmotic stress. An alternative mechanism of protection may be related to hyperglycemia-induced changes in expression and/or phosphorylation (by tyrosine kinase) of BCL-2, which is known to be pro-apoptotic. It is unclear at present whether glucose protection against hypoxic apoptosis in vitro is relevant to in vivo remodeling in either a positive or negative fashion.

Role of Insulin-Like Growth Factor-1 in Diabetic Cardiomyopathy
Insulin-like growth factor-1 (IGF-1) promotes cardiac growth, exerts positive effects on contractility and relaxation, promotes relaxation of vascular smooth muscle, and is a potentially important paracrine/autocrine factor in the myocardium. IGF-1 is produced by both vascular smooth muscle cells and cardiac myocytes and is subject to multiple regulatory influences. In vascular smooth muscle, its effects appear to be mediated by increased nitric oxide production. In contrast, hyperglycemia decreases nitric oxide production. In rat papillary muscle preparations and isolated ventricular myocytes, IGF-1 causes dose-dependent increases in tension development/shortening and calcium transients, but does not alter relaxation parameters. These effects are attenuated by nitric oxide synthase inhibition. In STZ-diabetes mellitus rats (5-7 days), the above-mentioned effects of IGF-1 are not observed, suggesting an intrinsic defect at the level of the myocyte whose mechanism is unknown at present. Similarly, short-term STZ-diabetes mellitus results in loss of effects of IGF-1 on vascular muscle, although normal responses to norepinephrine and KCl are retained. Since IGF receptors are up-regulated in this model, the signaling pathway must be disrupted downstream of the receptors. The possibility that paracrine/autocrine factors (e.g., IGF-1) contribute to the development of diabetic cardiomyopathy is an area that has received little previous attention.

Characterization of Diabetic Cardiomyopathy in Human Myocardium
Hemodynamic and endomyocardial biopsy assessment of patients with heart failure presumed due to diabetic cardiomyopathy has been mentioned previously, as have noninvasive studies in patients without overt evidence of heart disease. There has been no assessment of abnormalities of myocardial function in patients with combined diabetes mellitus and coronary artery disease, who are at high risk for heart failure and death if they suffer acute myocardial infarction. Advances in tissue preservation now allow assessment of the in vitro mechanical performance of strips of left ventricular myocardium obtained via epicardial biopsy during surgery or from explanted hearts. Studies performed in myocardial tissue from patients with mitral regurgitation and dilated cardiomyopathy (without diabetes mellitus) reveal a key functional abnormality, depression of the force-frequency relation (FFR). The positive FFR probably results from increased calcium cycling as contraction frequency is increased and is a key mechanism whereby ventricular function is augmented during exercise. Depression of the FFR has been strongly correlated with depressed sarcoplasmic reticular calcium pumping and decreased expression of SERCA-2. Recently, this approach has been applied to patients with combined diabetes mellitus and coronary artery disease who requires coronary artery bypass surgery. Patients selected for study had normal left ventricular contraction pattern, and no history of prior acute myocardial infarction or hypertension. They were compared to matched bypass surgery patients without diabetes mellitus. Left ventricular myocardial strips were stimulated to contract isometrically at varying contraction frequencies. In patients with combined diabetes mellitus and coronary artery disease, contractile performance is normal at rest, but the FFR is depressed compared to patients with coronary artery disease alone. The extent of depression is intermediate between coronary artery disease patients and mitral regurgitation/diabetes mellitus. Preliminary results of simultaneous calcium transient measurements indicate that the amplitude of the transient decreases in parallel with force measurements, implicating abnormal calcium handling as a mechanism of FFR depression. These results represent the first in vitro documentation of an underlying cardiomyopathy in patients with combined diabetes mellitus and coronary artery disease, and may provide a mechanism whereby contractile performance is impaired, especially during stress, resulting in a high incidence of heart failure after acute myocardial infarction.

Novel Therapeutic Approaches to Diabetic Cardiomyopathy
0000Until recently, the only therapy envisioned for prevention and treatment of the complications of diabetes mellitus has been tight control of blood glucose with insulin and/or oral hypoglycemic regimens. Results in general have been inconclusive. The recent introduction of agents that directly increase sensitivity to insulin (e.g., metformin and troglitazone) offer one new approach. Looking toward the future, application of gene therapy is of course of great interest. The ultimate efficacy of gene therapy will be linked to better understanding of the underlying pathophysiology of diabetic cardiomyopathy. Some recent developments with implications for gene therapy are indicative of the sorts of novel approaches that might be employed.

One alteration in calcium handling proteins identified in rodent models of diabetes mellitus is decreased activity of SERCA-2. SERCA-2 is also depressed in non-diabetic failing myocardium. A number of possibilities exist to explain this and other alterations in calcium handling, including reduced insulin signaling at the myocyte level, hyperglycemia per se, increased fatty acid oxidation and free radical production with damage to membrane lipids, and cytokine-mediated effects (e.g., interleukin-6 decreases SERCA-2 expression). Increasing SERCA-2 activity is therefore a potential novel therapeutic approach to diabetic cardiomyopathy. To test this hypothesis, the effects of STZ-induced diabetes on contractile function in isolated hearts and papillary muscle strips have been tested in transgenic mice that overexpress SERCA-2. SERCA-2 overexpression resulted in partial restoration of function compared to wild type mice treated with STZ, supporting this as a therapeutic modality. Advances in in vivo viral transfection techniques offer the possibility of transfecting large numbers of myocytes with transgenes that would be particularly useful in treating the dysfunction (e.g., by interfering with the inhibitory action of phospholamban on SERCA-2).

Recently, transgenic mice with modification of genes encoding proteins of oxidative metabolism have been developed. These animals may shed light on the relation between altered energy metabolism and depressed contractile function in diabetic cardiomyopathy and conceivably suggest targets for treatment. New genetic screening approaches to identify small molecules with potential use as drugs to protect the heart from metabolic injury are also of considerable interest. For example, a group of peptides selected for high affinity binding to DnaK, the bacterial orthologue of mammalian hsp70, demonstrate marked cytoprotection during substrate deprivation. Based on their binding affinity, this may be related to enhanced chaperone activity.

Since skeletal muscle insulin-resistance is a major risk factor contributing to metabolic disturbances of diabetes mellitus, insight into the cellular mechanisms associated with these changes and the effects of drug interventions and exercise may greatly enhance the ability to treat and prevent diabetic cardiomyopathy. An example of this is the identification of intracellular signaling links between the firing of motor nerves controlling the activity of skeletal muscles and transcription of genes encoding proteins that are differentially expressed in Type I vs Type IIb muscle fibers. These discoveries could lead to novel approaches to increasing insulin sensitivity and beneficial effects with respect to diabetic cardiomyopathy.

CONSENSUS RECOMMENDATIONS OF THE WORKING GROUP

As indicated in the introductory comments above, the Working Group agreed that diabetic cardiomyopathy is an extremely important and complex problem and that very little progress has been made in understanding its underlying pathophysiology, specific manifestations and natural history in patients, and treatment. The Group especially wished to dispel any questions as to the existence of diabetic cardiomyopathy as a distinct entity and the notion that it is not worth detailed investigation because it is simply a consequence of diabetes mellitus-related vasculopathy. Vascular disease may certainly be a cause of diabetic cardiomyopathy, but this in no way diminishes its importance and/or the need for more knowledge in this area. The Group was particularly struck by the fact that poor outcomes due to cardiovascular disease in patients with diabetes mellitus, especially those due to heart failure, are associated with angioplasty as compared to coronary bypass surgery. In consideration of the above, the Working Group strongly recommends that diabetic cardiomyopathy be targeted as an area for special emphasis in research funding and supports the timely development of a Request for Applications (RFA) to support future investigations. The Working Group also felt that this is an area in which collaborative, multidisciplinary research is especially appropriate. The Working Group agreed that the following constitute the most promising areas for funding:

1. Research designed to identify basic mechanisms of disease and novel therapeutic interventions using innovative molecular and cellular approaches. Emphases should be on distinguishing and elucidating the contributions of direct "toxic" effects of diabetes mellitus from those due to altered gene expression. Correspondingly, the roles of altered organelle and protein function vs cell death (either by necrosis or apoptosis) in the pathophysiology of diabetic cardiomyopathy should also be established. A great deal of effort should be focussed on the fundamental reasons for gender, racial, and ethnic variations in diabetes mellitus and associated cardiomyopathy.

2. The development and utilization of experimental models with high relevance to human disease should be encouraged and used to answer specific, mechanistic questions. Examples are: animals with insulin resistance as a single risk factor and in combination with other risk factors (e.g., hypertension and/or obesity), models that are particularly amenable to testing novel therapeutic interventions, and longer-term models that better simulate human disease.

3. Delineation of the connections between altered intracellular signaling, including those related to glucose transport and protein kinase C activation, and altered contractile function and cell death. These mechanisms of disease have been largely unappreciated by investigators working in the field until relatively recently.

4. A better understanding of the relation between altered energy metabolism and myocardial function in diabetes mellitus in particular and heart failure in general. Is there the possibility of a unifying hypothesis related to abnormal fatty acid oxidation and what are the key metabolic factors that contribute to diabetic cardiomyopathy? What is the relation between "diastolic dysfunction" and energy metabolism?

5. Elucidation of the mechanisms contributing to the modification of cardiac responses to ischemia -reperfusion and oxidative stress in diabetes mellitus. Is this the key factor in the high mortality of acute myocardial infarction in patients with diabetes mellitus? Does substrate utilization in the diabetic heart influence responses to ischemia and reperfusion?

6. Research designed to better characterize the epidemiology and pathophysiology of diabetic cardiomyopathy in patients. Recognizing the limitations of short-term animal models in relation to a chronic and extraordinarily complex human disease, it is critical to better characterize organ level and myocardial cellular/molecular alterations in patients with diabetes mellitus. In vitro studies employing tissue from patients and modern, non-invasive assessments of myocardial function and metabolism (magnetic resonance spectroscopy/imaging, PET scanning, echocardiography-Doppler and ultrasonic tissue characterization) are likely to be most fruitful in delineating human natural history and pathophysiology. Non-cardiac markers of diabetic cardiomyopathy (e.g., skeletal muscle alterations, retinal vasculopathy) should also be sought. Clues gained from these types of studies could be used to test mechanistic hypotheses relevant to human disease. Other key areas are whether there are meaningful differences between type 1 and type 2 diabetes mellitus and quantification of gender, racial and ethnic differences. Establishment of a registry of patients with diabetic cardiomyopathy based on noninvasive markers might be useful in delineating genetic markers of diabetic cardiomyopathy and identifying high risk patients, tracking disease incidence and natural history and providing a framework for intervention trials.

7. Clinical trials to test specific interventions. These could be designed to track clinical endpoints (death, development of heart failure) and/or non-invasive markers as delineated above. The list of potential interventions is large, for example, "tight control" of blood glucose, insulin sensitizers, protein kinase C inhibitors and ACE inhibitors, and is likely to increase in the future. A useful strategy might be to establish a mechanism to perform relatively short-term, feasibility studies followed by larger trials with longer term clinical endpoints when indicated.