The Dialogue Between Diabetes and Diastole

This editorial refers to ‘A randomized trial of the impact of strict glycaemic control on myocardial diastolic function and perfusion reserve: a report from the DADD (Diabetes mellitus And Diastolic Dysfunction) study’†† doi:10.1093/eurjhf/hfn018 by Christina Jarnert et al. on page 39 Even in early reports, diabetic cardiomyopathy was related to hyperglycaemia-induced myocardial microvascular damage because of its association with glomerulosclerosis and retinopathy.1 Diabetic cardiomyopathy was initially classified as a dilated cardiomyopathy as it featured prominent left ventricular (LV) enlargement and depressed LV systolic function. Over the last decade, however, diastolic LV dysfunction was identified as an earlier manifestation of diabetic cardiomyopathy.2,3 This restrictive phenotype of diabetic cardiomyopathy was also related to hyperglycaemia. Impaired diastolic LV distensibility was thought to result from advanced glycation end products (AGEs) causing strong collagen cross-links in the myocardial extracellular matrix.4 Because of the presumed importance of hyperglycaemia in all clinical phenotypes of diabetic cardiomyopathy, it was speculated that strict glycaemic control, preferably with insulin, could retard or even reverse LV dysfunction in diabetic cardiomyopathy. The hypothesis of strict glycaemic control retarding or reversing diastolic LV dysfunction in patients with Type 2 diabetes mellitus was elegantly tested in a small observational study, which revealed that insulin-induced amelioration of glycaemic control to parallel improvements of diastolic LV function and myocardial perfusion reserve.5 Inspired by these encouraging results, Jarnert et al. from the Karolinska Institute in Stockholm engaged in a larger randomized trial (diabetes mellitus and diastolic dysfunction; DADD), that evaluated the impact of strict glycaemic control on myocardial diastolic function and perfusion in patients with Type 2 diabetes mellitus. The result of this trial is reported in this issue of the European Journal of Heart Failure.6 The DADD trial unfortunately failed to confirm the results of the earlier observational study. In patients with Type 2 diabetes mellitus, strict glycaemic control induced no improvement in either diastolic LV function or myocardial perfusion reserve. Although this failure could relate to flaws in study design causing a statistical Type II error, the negative outcome could also suggest that mechanisms other than hyperglycaemia account for diastolic LV dysfunction in diabetes mellitus. The DADD investigators intentionally selected Type 2 diabetes patients with evidence of diastolic LV dysfunction but without confounding comorbidities that could also induce diastolic LV dysfunction. They therefore rigorously excluded Type 2 diabetes patients with ischaemic heart disease, heart failure, atrial fibrillation, poorly controlled hypertension, peripheral vascular disease, renal insufficiency, and LV hypertrophy. Probably because of the exclusion of patients with heart failure, atrial fibrillation, arterial hypertension, and LV hypertrophy, the DADD investigators ended up with a study population that only consisted of patients with subtle evidence of diastolic LV dysfunction. This was especially evident from the E/E' ratio of the entire patient cohort (=7.4 ± 2.2), the E/A ratio of the entire patient cohort (=1.0 ± 0.3) and the left atrial size of the insulin treated group [Left Atrial Volume Index (LAVI) = 31 ± 6 mL/m2]. An E/E' < 8 implies normal pulmonary capillary wedge pressures, and a LAVI of 31 mL/m2 is only 10% above the upper limit of the normal LAVI value. Based on this set of diastolic LV dysfunction indices, the study group was essentially composed of patients with slow LV relaxation or filling but with preserved diastolic LV distensibility. LV relaxation or filling kinetics depend mainly on LV recoil and early diastolic LV load in contrast to LV diastolic distensibility, which more closely reflects myocardial material properties.7 In a study group whose diastolic LV dysfunction consists of slow LV relaxation or filling but preserved LV diastolic distensibility, it is difficult to demonstrate a beneficial effect of strict glycaemic control on diastolic LV dysfunction because reduced AGEs formation and less collagen cross-links in the myocardial extracellular matrix will mainly affect diastolic LV distensibility and much less LV relaxation or filling kinetics. Recruitment of a patient population with diastolic LV dysfunction consisting only of slow LV relaxation or filling probably contributed to the negative outcome of the DADD trial. The exclusive enrolment of patients in whom diastolic LV dysfunction could be solely attributed to diabetes mellitus was possibly another reason for the negative outcome of the DADD trial. In the MONICA population-based survey, 58% of the patients with diastolic LV dysfunction were hypertensive and 29% had LV hypertrophy.8 The New York heart failure consortium completed a prospective multicentre registry of patients hospitalized for diastolic heart failure and reported similar findings.9 Pre-existing arterial hypertension was observed in 75% of patients and atrial fibrillation in 25% of patients. From these epidemiological data, it seems that it would be difficult to identify patients with diastolic LV dysfunction who do not also have either arterial hypertension, LV hypertrophy, or atrial fibrillation. Nevertheless, the DADD trial specifically insisted on absence of these confounding comorbidities. Moreover, Type 2 diabetes mellitus is also closely intertwined with arterial hypertension and LV hypertrophy. Arterial hypertension is an intrinsic part of the metabolic syndrome that frequently leads to Type 2 diabetes mellitus;10 and LV hypertrophy is associated with insulin resistance as hyperinsulinaemia stimulates prohypertrophic signalling in insulin-responsive tissues such as the myocardium.11 Because of the close associations between Type 2 diabetes mellitus and arterial hypertension or LV hypertrophy, identifying patients in whom diastolic LV dysfunction can be solely attributed to Type 2 diabetes mellitus, is probably an elusive goal. The pursuit of this goal in the DADD trial ended up being counterproductive as it yielded a study population with only subtle derangements of LV relaxation or filling and as such it considerably reduced the power of the trial. As the investigators recruited a patient population with only subtle evidence of diastolic LV dysfunction, the sensitivity and specificity of the criteria used to define diastolic LV dysfunction becomes an important issue. The DADD study used a set of criteria, whereby the patient had to satisfy at least one of the following: E/A ratio < 0.75, E/A ratio decreasing by 0.5 after Valsalva, E' < 8 cm/s, E/E' > 15 or LAVI > 32 mL/m2. It is of interest to note that only E/E' > 15 was retained in a recent consensus document of the heart failure and echocardiography associations of the European Society of Cardiology (HFA-Echo-ESC) as providing stand-alone, diagnostic evidence of diastolic LV dysfunction.12 Furthermore, of the entire DADD patient cohort, just one patient satisfied the E/E'> 15 criterion. The criteria for diastolic LV dysfunction proposed by the HFA-Echo-ESC have also been tested in a practical scenario.13,14 This showed mitral flow velocity Doppler to confer no additional information for the diagnosis of diastolic LV dysfunction in the presence of Tissue Doppler Imaging. Moreover, it also revealed that LAVI had to exceed at least 40 mL/m2 to achieve optimal sensitivity and specificity for left atrial size as a measure of diastolic LV dysfunction. Based on the criteria proposed by the HFA-Echo-ESC, most of the patients enrolled in the DADD trial did not have diagnostic evidence of diastolic LV dysfunction. Sensitivity and specificity of the criteria used to define diastolic LV dysfunction become, especially important when recruiting patients with only subtle evidence of diastolic LV dysfunction. This certainly applies to Type 2 diabetes patients without comorbidities, as evident from the conflicting data provided by two studies on the prevalence of diastolic LV dysfunction in asymptomatic normotensive Type 2 diabetes patients. These two studies, respectively, reported a 0% prevalence and a 75% prevalence of diastolic LV dysfunction in these patients.15,16 Although a statistical type II error related to patient selection could be involved in the negative outcome of the DADD study, the reported failure of strict glycaemic control to improve diastolic LV function in Type 2 diabetes patients could also suggest that hyperglycaemia was not an important contributor to the diastolic LV dysfunction observed in these patients. In Type 2 diabetes, myocardial damage not only results from hyperglycaemia but also from hyperinsulinaemia, inflammation, and oxidative stress.17,18 A recent study observed improvement of diastolic LV function in Type 2 diabetes patients treated with rosiglitazone. This improvement of diastolic LV function was related more closely to the fall in malondialdehyde, a plasma marker of oxidative stress, than to the fall in HbA1c, insulin or interleukin-6.19 Another recent study used LV endomyocardial biopsies and assessed the relative contributions of AGEs deposition, fibrosis, cardiomyocyte hypertrophy, and cardiomyocyte resting tension to the high diastolic LV stiffness of the failing diabetic heart.20 In diabetic patients with diastolic heart failure, cardiomyocyte hypertrophy and high cardiomyocyte resting tension were mainly responsible for the rise in diastolic LV stiffness; whereas in diabetic patients with systolic heart failure, AGEs deposition and fibrosis were the main contributors to diastolic LV dysfunction. As the patients of the DADD trial had abnormal diastolic but normal systolic LV function, they more closely resembled diastolic than systolic heart failure patients. AGEs deposition and fibrosis were therefore of lesser importance for their diastolic LV dysfunction than cardiomyocyte hypertrophy and resting tension. As insulin-based glycaemic control is specifically aimed at correcting AGEs deposition, this treatment obviously failed to correct diastolic LV dysfunction resulting from cardiomyocyte hypertrophy. Furthermore, in the strict glycaemic control group of the DADD trial, the insulin-based treatment raised their already elevated insulin levels. Since insulin is a potent stimulus for cardiomyocyte hypertrophy, worse hyperinsulinaemia could have induced more cardiomyocyte hypertrophy with a concomitant further rise in cardiomyocyte resting tension, the main cause for the diastolic LV dysfunction in these patients. Over the study period, the strict glycaemic control group of the DADD trial also gained weight whereas the other group lost weight. An experimental study observed a relation between excess body weight and diastolic LV dysfunction. When rabbits were fed a high fat diet, they developed obesity after 12 weeks and in the isolated heart preparation, the LV stiffness of the obese animals was significantly higher than in the control animals.21 Hence, the weight gain observed in the strict glycaemic control group could also have contributed to the negative outcome of the DADD trial. The DADD trial observed no benefit of strict glycaemic control on myocardial diastolic function and perfusion reserve in Type 2 diabetes patients. This negative result could be a statistical Type II error caused by the mild diastolic LV dysfunction of the study population, the recruitment of Type 2 diabetes patients without other comorbidities and the use of enrolment criteria for diastolic LV dysfunction with suboptimal sensitivity and specificity. The negative outcome of the DADD trial could, however, also hint at hyperglycaemia not being the only mechanism for diastolic LV dysfunction of the diabetic heart. Hyperinsulinaemia is an equally important mechanism for diastolic LV dysfunction in Type 2 diabetes patients, especially when systolic LV function is preserved. In the DADD trial, diastolic LV dysfunction possibly failed to improve with strict glycaemic control because of a balance between the beneficial action of euglycaemia on myocardial AGEs and the adverse effect of hyperinsulinaemia on cardiomyocyte hypertrophy. Rigorous and strict insulin-based glycaemic control perhaps illustrates Voltaire's quote ‘Le mieux est l'ennemi du bien’ (The perfect is the enemy of the good). This work is supported in part by the Dutch Heart Foundation (Grant 2006B035; Titin in diastolic heart failure).