Background
Diastolic dysfunction of the left ventricle (LV) is characterized by delayed active relaxation and increased chamber stiffness [
1]. The condition is most commonly associated with ischemic heart disease and/or hypertension with subsequent concentric remodeling or hypertrophy of LV (LVH) [
2]. LV diastolic dysfunction and LVH are powerful independent predictors of future cardiovascular morbidity and mortality [
3,
4], and identification of other hemodynamic and non-hemodynamic factors associated with the development of these often subclinical cardiac conditions may unveil novel targets for prevention.
Both LV diastolic dysfunction and LVH are common findings among patients with diabetes mellitus (DM) [
5‐
7]. The associations are independent of concomitant hypertension and ischemic heart disease, which has led to the term
diabetic cardiomyopathy, defined as ventricular dysfunction in a patient with DM, occurring independently of an otherwise recognized cause [
8]. However, the pathogenic mechanisms for development of this condition are poorly understood. Insulin resistance and the accompanying hyperinsulinemia may constitute an important pathophysiological link in subjects with obesity, glucose intolerance, or overt DM, but the majority of studies so far have yielded inconsistent results, particularly due to inadequate adjustment for key confounders, notably body size, blood pressure, and glucose levels [
9‐
11].
Additionally, most previous studies addressing this subject have been of cross-sectional nature, which, besides the inability to establish causality, are further limited by the fact that there is evidence to suggest that the changes in LV structure and function may be consequences of long-term, rather than short-term, metabolic abnormalities [
12].
Therefore, we conducted this study in subjects free from DM and overt cardiovascular disease at baseline aiming to examine whether lower insulin sensitivity as determined by homeostatic model assessment (HOMA-%S) was associated with increased left ventricular mass (LVM) and presence of LV diastolic dysfunction at long-term follow-up, independently of body mass index (BMI).
Discussion
In this prospective population-based cohort study comprising middle-aged male, apparently healthy subjects, we found that greater BMI, but not lower insulin sensitivity defined as the lowest HOMA-%S quartile was associated with later detection of increased LVM and grade 2 or 3 LV diastolic dysfunction. Low insulin sensitivity (HOMA-%S Q1) was only associated later recognition of grade 2 or 3 LV diastolic dysfunction in univariable analysis, but the association was lost after adjusting for BMI. As expected, BMI and HOMA-%S were significantly correlated.
Conflicting results have been reported regarding the relationship between insulin sensitivity and both indexed and non-indexed LVM, and to our knowledge, none of these were prospective. In a small study comprising 29 non-obese, glucose-tolerant subjects with borderline hypertension,
Phillips et al. found a significant independent association between LVMI and insulin sensitivity [
21]. Subjects were age-wise comparable to our cohort; however, insulin sensitivity was derived by frequent sampling during OGTT. Likewise,
Sundström et al. [
22] found a borderline significant association between insulin resistance according to HOMA (HOMA-IR) and LVMI in normotensive subjects, but they did not adjust for glucometabolic status. On the contrary, in one of the largest studies to date, based on the
Framingham Heart Study [
10] cohort, HOMA-IR was associated with increased LVM among women only, although the relationship was largely accounted for by obesity. Further supporting our results, neither
Galvan et al. [
11] nor
Ebinc et al. [
23] were able to detect an association between LVM and insulin resistance, independently of BMI, in subjects without DM. Although a number of other studies exist, direct comparison is challenging, especially due to limited sample sizes and heterogeneous study populations, with the majority having assessed subjects already at increased cardiovascular risk, i.e. subjects who were obese, had DM, or were hypertensive. Such individuals have greater insulin resistance and LVM than the general population [
24,
25], providing limited pathophysiological understanding regarding subjects in whom non-hemodynamic LVH-inducing mechanisms are likely to be more important. Adjustment for these important confounders has also been quite variable [
11]; for instance, some of the described relations of insulin resistance to LVM may have been mainly due to the effects of blood pressure. A true association between insulin LV size and function is therefore more likely be revealed in a general population-based study.
Asymptomatic LV diastolic dysfunction is the most prominent characteristic of diabetic cardiomyopathy [
8,
26]. However, since LV diastolic function may already be impaired in the pre-diabetic or even preclinical phase of glucometabolic disturbances, i.e. before the onset of sustained hyperglycemia, an independent mechanistic role of insulin resistance may exist [
27‐
29]. Furthermore, LV diastolic dysfunction may be evident at less severe insulin resistance when compared to the values associated with measureable LV structural changes [
30]. In a population-based study including 1063 subjects,
Fontes-Carvalho et al. found a significant association between higher insulin resistance according to HOMA-IR and worse LV diastolic function, including lower lateral é velocity and higher E/é ratio [
31]; however, without adjusting for systolic blood pressure and BMI. Further supporting these results, both
Dinh et al. [
32] and
Hwang et al. [
30] found an independent association between insulin resistance and both presence and severity of LV diastolic dysfunction, independently of overt DM; however, neither studies adjusted for body size. In addition, all the above mentioned studies were cross-sectional. In the present study, we showed that BMI itself, but not insulin sensitivity, was significantly associated with E/é and grade 2 or 3 diastolic dysfunction, in multivariable analysis. Nevertheless, direct comparison between different studies is complicated by use of variable methods for assessment of both LV diastolic function and insulin sensitivity as well as inadequate adjustment for relevant confounders.
The highly prevalent co-existence of insulin resistance, hyperinsulinemia, obesity, hypertension, and DM makes it difficult to dissect the separate role of each of these conditions for development of subclinical cardiac damage [
11]. However, at present time, there is insufficient convincing evidence to conclude that LVM and prevalence of LV diastolic dysfunction are greater among subjects with low insulin sensitivity or insulin resistance, when adequate care is taken to adjust for DM, blood pressure, and body size. Therefore, uncertainty remains, as to why subjects with increased body size have greater LV size and worse diastolic function [
33‐
35]. Several other risk factors, e.g. elevated blood pressure, glucose and cholesterol levels, are associated with obesity, and the progressive addition of metabolic risk factors seems to be associated with greater LVM [
36]. Just as proposed for diabetic cardiomyopathy, the myocardial alterations associated with obesity are likely to be a result of several similar synergistically acting mechanisms [
8], and it is possible that low insulin sensitivity, as previously suggested for fasting plasma glucose [
20], primarily acts as an effect modifier of these other risk factors. This hypothesis is supported by the fact that most studies reporting a positive association between insulin resistance and subclinical cardiac damage have involved subjects with other risk factors as well, and our chances of finding positive associations were weakened by the strict selection criteria employed, aiming to exclude subjects with prevalent DM or cardiovascular disease. Although we were not directly able to find synergistic interactions between insulin sensitivity and the traditional risk factors in the present study, this could have resulted from the relatively small sample size. The exact reason why baseline BMI, but not insulin sensitivity, was associated with later cardiac damage in our study, is uncertain. However, HOMA-indices display pronounced biological variation, which decreases the chances of finding such significant associations. Furthermore, the HOMA-indices show considerable temporal changes, which may further reduce the predictive value at long-term follow-up. Although BMI changes over time as well, the fluctuations may more often be unidirectional [
37,
38].
Regardless of whether or not insulin resistance is the main mediator of subclinical cardiac damage in obesity, pre-diabetes, and DM, the strong association between higher baseline BMI and later detection of structural and functional LV changes in our study suggests that early weight loss in overweight or obese subjects may halt the progression of adverse cardiac alterations, specifically reduce the risk of LVH and diastolic dysfunction. This is supported by the results from the
Coronary Artery Risk Development in Young Adults study, in which increasing BMI over a 5- to 10-year follow-up in generally healthy adults was associated with increasing LVM [
33,
34]. Further supporting this early preventive strategy is the weak ability of therapeutic interventions, e.g. intensive glycemic control [
39,
40] and antihypertensive drugs [
41,
42], to lower cardiovascular risk, when DM is overt, because the myocardial damage may have become partly to completely irreversible at this point. Lastly, early weight loss is associated with other clinically relevant benefits, including a lower risk of hypertension, dyslipidemia, and DM itself, which may further reduce the risk of cardiovascular complications [
43]. The lack of our detection of a beneficial effect of BMI reduction in the present study was most likely related to the small number (
n = 32) of subjects who actually lost weight during the study period and the possible BMI fluctuations over time that we were unable to account for. Even in those having lost weight, the loss was very subtle in most subjects. In addition, most individuals were normo- or overweight, but not obese, making detection of substantial benefits difficult.
Limitations
Although the participation rates of 71 % in MPP and 72 % in MPP-RES, respectively, are considered high, one may still argue that the study subjects did not represent a truly random population sample since people who agree to take part may be healthier than the general population. All subjects in the present study were male, limiting the applicability of the results in females. Moreover, our exclusion of a vast amount of the original study population in order to get a cohort of apparently healthy subjects, who were alive and underwent echocardiography, however with variable follow-up periods, may introduce further selection bias, including survival bias, and lack of adequate power.
Insulin sensitivity was not assessed according to the gold standard method, i.e. the hyperinsulinemic euglycemic clamp technique [
44]. However, the use of fasting BG and insulin was justified by the fact that HOMA-derived parameters are strongly related to clamp-measured insulin sensitivity and insulin resistance in subjects both with and without DM [
27,
45]. Sensitivity analyses with respect to the prediction of LV size and LV diastolic dysfunction were performed, before settling on the use of Q1 as cut-off in the present study. Moreover, data on glucose tolerance would have been desirable; however, in the present study, inclusion of these data would have reduced sample size even further.
Similarly, the use of BMI as a surrogate marker for obesity has some limitations [
46‐
48], and the inclusion of other measures, e.g. waist circumference and waist-to-hip ratio would have been preferred; however, these measurements were not available. The accuracy of BMI for diagnosing obesity is especially limited for individuals with BMI between 25.0 and 29.9 kg/m
2, in men and in the elderly. However, BMI or plain body weight may still be the best way to evaluate changes in body fat content over time, because increments in body weight or BMI most likely represent fat gain [
49].
Linear LV measurements have prognostic value and are feasible, especially when studying large populations. However, the method is based solely on basal dimensions, unable to accommodate for LV shape and size changes that might occur along the long axis of the chamber, and the formula for calculating LVM assumes normal LV geometry and cubes the linear measurements. Therefore, even small errors may significantly influence the calculated mass [
17]. Although echocardiography does not directly measure the same parameters for diastolic function that are measured invasively, it is still the most practical and recommended routine clinical approach. However, some limitations deserve mentioning. Minimal angulation is essential for reliable spectral Doppler measurements. Additionally, the usefulness of é velocity in normal subjects may be limited, as preload increases é in these subjects. The correlation between E/é between 8 and 15 using septal é (9–13 in the present study) and mean LV diastolic pressure displays wide variability [
50]. Therefore, although increased E/é is indicative of an elevated LV filling pressure, it should not be used as stand-alone parameter when drawing conclusions about LV diastolic dysfunction. Furthermore, our grading of LV diastolic dysfunction could have been more robust, had we also been able to incorporate the left atrial volume index [
19]. Lastly, the lack of an echocardiography at baseline prevented us from directly assessing LV structural and functional changes over time.
Competing interest
All authors have completed the Unified Competing Interest form at
www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organization for the submitted work; no financial relationship with any organizations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.
Authors’ contribution
MN contributed substantially to the conception of the hypothesis, the design of the work, analysis of the work, interpretation of data for the work, drafting the work, critical revision for important intellectual content, final approval of the version to be published, and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. MP contributed substantially to the conception of the hypothesis, the design of the work, analysis of the work, interpretation of data for the work, drafting the work, critical revision for important intellectual content, final approval of the version to be published, and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. OG contributed substantially to the analysis of the work, interpretation of data for the work, critical revision for important intellectual content, the final approval of the version to be published, and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. ML contributed substantially to the acquisition of data for the work, critical revision for important intellectual content, the final approval of the version to be published, and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. PN contributed substantially to the acquisition of data for the work, the interpretation of data for the work, critical revision for important intellectual content, final approval of the version to be published, and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. MO contributed substantially to the conception of the hypothesis, the design of the work, analysis of the work, interpretation of data for the work, drafting the work, critical revision for important intellectual content, final approval of the version to be published, and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All authors read and approved the final manuscript.