The main finding of the present study is that IR is associated with LVDD independent of overt diabetes. These finding persist after adjustment for CAD, hypertension, age, sex, history of previous myocardial infarction, history of previous coronary angioplasty, EF and glycaemic control. To our knowledge, this is the first study that demonstrated an association in a population of patients without a history of diabetes focusing on the published current guidelines for the diagnosis of LVDD.
Furthermore, our date confirm the observations that T2DM is associated with LVDD , which is considered a precursor of diabetic cardiomyopathy. In addition, we were able to extend the findings in previous studies showing the association between LVDD and IGT.
These findings are in line with a limited number of studies that assessed the relationship between prediabetes and LVDD mainly in population based studies . Nevertheless, there are some concerns about methodological issues involving the identification of LVDD patients in the previous studies. The criteria used to define LVDD were highly variable and predominantly did not consider the diagnostic guidelines .
Furthermore, most of these studies did not screen for the presence of CAD using coronary angiography. Since CAD has been shown to be associated with LVDD, the lack of information on coronary morphology is a potential source of bias towards an increased prevalence and severity of CAD in patients with LVDD. In contrast, the present study focuses on patient recruitment according to the published current guidelines for the diagnostic workup of LVDD in subjects well characterized for the degree of abnormality in plasma glucose levels and coronary morphology.
We found that IR, as estimated by the HOMA-IR , was strongly associated with LVDD in patient without a history of overt diabetes. In a logistic regression model with LVDD as the dependent variable, this association was found independently of established classic risk factors for LVDD such as female sex, age, CAD, hypertension and diabetic state. In addition, high sensitive c-reactive protein (hsCRP) levels did not act significantly to the model. Given the fact that previous studies suggested suggest a link between low-grade inflammation and the presence of LVDD [23, 24], it is not likely that low grade inflammation was a significant source of bias in our cohort.
As expected, the presence and degree of a glucose metabolism disorder was another strong predictor for the presents of LVDD. These findings suggest that IR linked to glucose metabolism disorders can contribute to the development of LVDD and that both are important determinants of LVDD. It is well established that diabetes is associated with chronic HF [25, 26]. Previous studies have shown that IR, independent of hyperglycemia, predisposes the development of systolic HF [14, 27]. In a study by Ingelsson et al. , IR was the strongest glucometabolic predictor of chronic HF, even in a subsample without diabetes and independently of other established risk factors for HF. Our study showed, for the first time, that IR is a predictor for the prevalence and severity of LVDD in subjects without overt diabetes. Since, in our study, IR was associated with LVDD, MetS and obesity and LVDD was associated with increased BMI and waist circumference, previous described association between obesity and diastolic HF may be mediated, at least in part, by IR.
This is of clinical relevance, since it has been demonstrated that subtle and subclinical signs of LVDD are associated with an increased cardiovascular morbidity and mortality [3, 9, 28]. In a study by Wang et al. , including subjects with cardiac diseases and controls, a reduced E' velocity was an independent predictor of cardiac death, and From et al.  demonstrated an association of increasing E/E' ratio with all-cause mortality in patients with diabetes. Mogelvang et al.  showed that, in the general population, LVDD diagnosed by TDI was a powerful and independent predictor of death. Recently, the predictive value of asymptomatic early diastolic dysfunction for HF has been appreciated , and asymptomatic early diastolic dysfunction is the most prominent characteristic of diabetic cardiomyopathy [8, 31]. Our findings may indicate that the risk for LVDD is already increased in the subclinical phase of glucose metabolism disturbances in subjects with IR, which may precede the development of diabetes. This information would be of clinical importance, because it might strongly justify and encourage the use of therapeutic interventions, including drugs capable of improving insulin sensitivity, with the aim of reducing the risk for diabetic cardiomyopathy.
Although establishing a pathophysiological model linking IR to LVDD is beyond the scope of the present study, several mechanisms for a conditional relationship between IR, glucose metabolism and LVDD should be considered. These mechanism most prominent includes altered insulin signaling, deposition of advanced nonenzymatic glycation end products (AGE) into the ECM , increased myocardial collagen deposition with down- regulation of matrix metalloproteinases (MMPs) and upregulation of tissue inhibitors of metalloproteinases (TIMPs) , and substrate shifts from glucose to free fatty acids  as well as endothelial dysfunction .
In our study, the majority of subjects with IR and IGT had a mild form of LVDD (grade I, relaxation abnormalities). Since relaxation is an active, dynamic and energy-consuming myocardial process, impaired relaxation may be due to a reduction in the energy supply. The above mentioned abnormalities in the free acid metabolism may be important contributors to the abnormal myocardial relaxation in subjects with IR. High levels of free acids lead to an inhibition of glucose oxidation, resulting in reduced myocardial ATP availability .
In addition, IR can lead to sympathetic nervous system activation , which is related to an increased response to angiotensin II  and increases the stimulating effects of angiotensin II on collagen production , leading to fibrosis and likely subsequent the development of LVDD. Alterations in myocardial structure are usually minimal in the early stages of diabetes and may be partially reversible. As the disease progresses, accumulation of collagen becomes obvious and may play a major role in the development of LVDD . Furthermore, insulin resistance independently influences arterial stiffness , and MacIsaac et al  demonstrated a link between arterial resistance and diastolic dysfunction in type 2 diabetes, indicating that vascular and LVDD in glucose metabolism disturbances are manifestations of common pathophysiological mechanisms.
Interestingly, even in subjects without a history of diabetes before inclusion into the study, the HbA1c was significantly correlated with the E/E'ratio, a parameter indicative for LVDD with elevated filling pressures. In addition, HbA1c correlated with the LAi, a parameter that indicates long standing LVDD. As the HbA1c incorporates metabolic disturbances over a longer period of time, the LAi reflects a cumulative effect of different contributors to LVDD of longer duration and is less vulnerable to acute changes in preload and afterload, which might have an acute impact on diastolic function. Therefore, the LAi could be labeled as the "HbA1c" of diastolic dysfunction abnormalities.
Two recent studies, the "ADVANCE" trial  and the "ACCORD"- trial , reported no significant benefit from intensive HbA1c lowering in terms of cardiovascular outcomes in subjects with long standing diabetes. Similarly, outcomes of recent trials in subjects with HFnEF were frequently disappointing [45–48]. The average duration of diabetes at the start of the "ACCORD" and "ADVANCED" studies ranged from 8 to 11,5 years. Subgroup analysis in "ACCORD" showed that intensive glycaemic control led to fewer cardiovascular complications in diabetic subjects with shorter disease duration and with no antecedent cardiovascular events at baseline. The cardiovascular benefit of intensive glycaemic control in subjects with shorter diabetes duration and no pre-existing cardiovascular disease was also supported by the follow up of the United Kingdom Prospective Diabetes Study (UKPDS) patients . Analogous, the neutral outcome in HFnEF trials might be attributed to the recruitment of patients with advanced diastolic HF and concomitant reduced systolic function, indicating long standing myocardial disease. Likely, therapeutic interventions have failed because the myocardial damage might have become partly irreversible.
Therefore, we speculate that an early intervention is necessary to avoid or reverse LVDD as the first stage in the development of diabetic cardiomyopathy [8, 50]. Early treatment strategies should address functional myocardial abnormalities characteristically observed in subjects with diabetes, IR and MetS such as a shift in the myocardial metabolism from glucose to free fatty acids or changes in the ECM turnover. Thiazolidinediones, which are capable to restore glucose utilization, have recently been shown to favorably modify diastolic function as evident from improvement the in E'septal velocity . In this content, one should recognize that physical activity, which can improve insulin sensitivity, was shown to prevent the development of cardiovascular diseases in type 2 diabetes and can improve diastolic function and exercise capacity in subjects with diastolic heart failure .
In our study, we did not use the gold standard in the assessment of insulin sensitivity, i.e. glucose clamp . However, previous studies have shown that HOMA-IR is strongly related to clamp-measured insulin resistance in both diabetic and non diabetic subjects [19, 54]. Therefore, the HOMA-IR seems to be a reliable diagnostic tool and practicable alternative in the clinical setting in the assessment of IR. Furthermore, the rates of CAD and cardiovascular risk factors were high in this study population. Therefore, the present results may not be readily represent the general population. Nevertheless, association between LVDD, IR and glucose metabolism remains significant after adjustment for CAD and hypertension as covariates into multivariate regression models. Although we based the diagnosis of LVDD on current guidelines which have recently been published , their clinical value has yet to be prospectively validated. Lastly, our cross sectional study design does not permit any conclusions on causality.