Insulin resistance influences the impact of hypertension on left ventricular diastolic dysfunction in a community sample

Background Although obesity‐associated metabolic abnormalities (insulin resistance‐IR) may not play as marked a role in determining left ventricular (LV) diastolic dysfunction (DD) as hypertension, the impact of combinations of these risk factors on DD is unknown. Hypothesis We hypothesized that IR influences the impact of hypertension on DD. Methods In 704 randomly selected participants from a community sample with a high prevalence of hypertension (50.6%) and obesity (46.5%), we determined adiposity indices, IR from the homeostasis model (HOMA‐IR) and LV diastolic function using standard echocardiographic techniques. Results HOMA‐IR was independently associated with lateral wall e' and E/e' (P < 0.05 to P < 0.005) as well as a diagnosis of DD (P < 0.02). Importantly, however, an enhanced relationship between HOMA‐IR and E/e' in hypertensives (n = 356, partial r = 0.15, P < 0.005) as compared to normotensives (n = 348, partial r = 0.02 P = 0.75) was noted. Consequently, as compared to normotensives, with adjustments for confounders, hypertension was independently associated with DD only in those with the highest tertile of HOMA‐IR (odds ratio = 2.65, 95% confidence interval = 1.29‐5.42, P < 0.01), while in those with the lowest tertile of HOMA‐IR, hypertension failed to show a higher prevalence of DD (P = 0.22). Conclusions Insulin resistance enhances the impact of hypertension on LV DD. Thus, DD is more likely to occur with the combination of hypertension and IR.

participants receiving therapy, report on an equal or greater impact of obesity as compared to hypertension on left ventricular (LV) diastolic function. [13][14][15] However, in studies conducted across the full adult age range in a community sample with a high proportion of obesity and hypertension, but a low proportion of participants who were receiving antihypertensive therapy, blood pressure (BP) was noted to be the main determinant of DD. 16 Nevertheless, more recent evidence suggests that a more concentrically remodeled LV determines whether obesity-related insulin resistance (IR) is associated with LV diastolic function. 17 As hypertension is a strong determinant of concentric LV remodeling, the possibility exists that even if obesity or associated IR have only a modest impact on LV diastolic function, that obesity or its metabolic consequences may determine the extent to which DD occurs in hypertension. Consequently, in the present study, we aimed to determine whether adiposity indices or IR influence the extent to which DD occurs in hypertension in a community sample with a high prevalence of hypertension and obesity.
Participants gave informed, written consent. The study design has previously been described. 18

| Demographic and clinical information
A standardized questionnaire was administered to obtain demographic and clinical data. 18 Nurse-derived conventional BP was measured according to guidelines using a mercury sphygmomanometer after 5 minutes of rest in the seated position as previously described. 20 Five consecutive BP readings were obtained using an appropriately sized cuff, 30 to 60 seconds apart. The average of the five readings was taken as the BP. None of the visits had fewer than the planned BP recordings.
Hypertension was defined as the use of antihypertensive medication or if the mean of the five conventional BP measurements was >140 (systolic BP) or 90 (diastolic BP) mm Hg in those not receiving medication.

| Echocardiography
Echocardiographic measurements were performed as previously

| Data analysis
Database management and statistical analyses were performed with SAS software, version 9.4 (SAS Institute Inc., Cary, North Carolina).
Data from individuals were averaged and expressed as mean ± SD or the SE of the mean (SEM). To improve on the distribution of data, HOMA-IR, lateral e', septal e', E/e', and LAV index were logarithmically transformed. To determine independent relations, multivariate adjusted linear (continuous data) or logistic (discrete data) regression analysis was performed. Indexes of diastolic LV function were adjusted for several confounders associated with diastolic function noted in bivariate analysis. Relationships (partial r values) were compared with z-statistics. Table 1 gives the demographic and clinical characteristics of the normotensive and hypertensive participants. More women than men participated in the study and a high proportion of participants, particularly, the hypertensives, were overweight, obese or morbidly obese and had central obesity. As compared to participants recruited prior to TDI becoming available, participants in whom echocardiography was performed once routine TDI became available, were modestly older with more abdominal obesity, but a lower HOMA-IR and LVMI and more were receiving treatment for hypertension (Table S1, Supporting Information). A 5.5% of the normotensives and 25.0% of the hypertensives had DD and this was largely determined by a combination of either reductions in lateral or septal e' and increases in E/e' (75%). No participants had an ejection fraction <40% and 4.4% had an ejection fraction <50%. Of the sample 39.3% had LV hypertrophy (LVH) (LVMI > 80 g/m 1.7 for men and >60 g/m 1.7 for women) and 18.1% had concentric LV remodeling (relative wall thickness >0.42). A greater proportion of hypertensives than normotensives had LVH and concentric LV remodeling.

| Factors associated with LV diastolic function
With adjustments for confounders, systolic BP, and either WC, HOMA-IR, or BMI were independently associated with lateral and septal wall e' and E/e' (Table S2). While WC and systolic BP were independently associated with LAV index, HOMA-IR, and BMI were not (Table S2). However, with adjustments for confounders systolic BP and HOMA-IR, but not WC or BMI were independently associated with the presence of LV DD ( Table 2).

| Impact of IR on LV diastolic function in hypertensives and normotensives
In both normotensives and hypertensives, WC (and BMI) as well as HOMA-IR were independently associated with lateral wall e' (Table 3). However, in normotensives, but not in hypertensives, WC or BMI were independently associated with E/e', while in hypertensives, but not normotensives, HOMA-IR was independently associated with E/e' (Table 3). Importantly, as compared to normotensives, this translated into an independent effect of hypertension on E/e' (and lateral wall e') only in those hypertensives with a HOMA-IR in the upper two tertiles (Figure 1). Although   Figure S1), these effects were largely attributed to age differences.
Indeed, beyond age and other confounders hypertension was only independently associated with an increased odds of DD in those with the higher tertiles of HOMA-IR, WC, or BMI ( Figure 2). Importantly, a stepwise increase in the odds of DD occurred across tertiles of HOMA-IR, but inconsistent effects were noted for WC and BMI ( Figure 2).

| DISCUSSION
The main findings of the present study are as follows: In a community sample with a high prevalence of obesity, indices of excess adiposity and IR were independently associated with indices of LV diastolic function (e' and E/e'). However, while HOMA-IR was independently associated with lateral wall e' in both hypertensives and normotensives, HOMA-IR was independently associated with E/e' in hypertensives, but not in normotensives. Consequently, HOMA-IR determined whether hypertensives developed LV DD as compared to normotensives. In contrast, adiposity indices were associated with indices of diastolic function less well in hypertensives as compared to normotensives and adiposity indices were consequently not independently associated with DD.
Although several large studies have demonstrated that indices of excess adiposity are strongly and independently associated with indices of LV DD, [13][14][15] these studies have been confounded by the use of predominantly elderly populations (where age is the principle determinant of DD), in select patients referred for echocardiography, or in samples with a high proportion of participants receiving antihypertensive therapy. [13][14][15] In contrast, in an alternative study conducted in a much smaller cohort of the present community, but across the adult age range and in a sample with a high prevalence of obesity and hypertension, where antihypertensive therapy was employed in only half the hypertensives, 16 BP was noted to contribute far more to DD than adiposity indices. In keeping with this prior study 16 in the present study   conducted in a much larger study sample of the same community we show that while BP and HOMA-IR translated into DD, adiposity indices failed to do so. The ability of HOMA-IR to associate with DD while adiposity indices did not, we attribute to an impact of HOMA-IR, but not adiposity indices on E/e in the hypertensive, but not in the normotensive BP range. These data therefore suggest that IR is an important contributor to DD, but mainly in those with hypertension.
An important caveat of the present study is that the results do   Table S3. *P < 0.02, **P < 0.001, ***P < 0.0001 vs normotensives. †P < 0.05 vs hypertensives HOMA-IR tertile 1 or vs hypertensives waist circumference tertile 1. e', myocardial tissue lengthening in early diastole at the mitral annulus; E/e', transmitral early blood flow velocity/velocity of the mean value of lateral and septal wall myocardial tissue lengthening in early diastole at the mitral annulus; LV, left ventricle HOMA-IR or indices of excess adiposity and DD, as described in the present study, are more likely to reflect relations between an excess adiposity and actual DD than those previously described. [13][14][15] An explanation for the impact of HOMA-IR on E/e' (an index of LV filling pressures) and hence DD in hypertensives, but not normotensives in the present study, is unclear. It is possible that because hypertension is associated with concentric LV remodeling, and as recently demonstrated, a more concentrically remodeled LV determines the impact of IR on E/e', 17 that IR only contributes to LV filling pressures when the LV is more concentrically remodeled. Importantly, however, the presence of hypertension in the present study did not influence the association between HOMA-IR and lateral wall e', an index of LV relaxation, while previous work does show an impact of relative wall thickness on relations between HOMA-IR and lateral wall e'. 17 Hence, the impact of hypertension on relations between HOMA-IR and E/e' may not be accounted for just by an effect of the extent of concentric LV remodeling, but by an alternative as yet unidentified factor.
There are several potential implications of the findings of the present study. First, in contrast to adiposity indices, which had less of an ability to determine the impact of hypertension on DD, the assessment of IR may better characterize hypertensives at risk of developing DD and hence heart failure with a preserved ejection fraction. In this regard, longitudinal studies are required to assess this question. Second, targeting IR as opposed to an excess adiposity per se with behavioral modification in hypertensives may have marked benefits to preventing the development of DD and hence heart failure with a preserved ejection fraction. In this regard, intervention studies are required to assess these hypotheses.
The limitations of the present study are as follows: First, this is a cross-sectional study and hence, we cannot draw conclusions FIGURE 2 Impact of insulin resistance (homeostasis model-homeostasis model of insulin resistance [HOMA-IR]), waist circumference (WC) or body mass index (BMI) on the odds of independent associations between hypertension and left ventricular diastolic dysfunction. Adjustments in the left panels are for age, sex, pulse rate, regular smoking, regular alcohol consumption, and diabetes mellitus. Tertiles of HOMA-IR, WC, and BMI are defined in Table S3 regarding causality. Whether the development of IR influences that of LV DD in those with hypertension rather than normotension therefore requires further study. Second, we are not statistically powered to perform sex-specific analysis and hence, it is possible that as more women than men volunteered for the study, the results may relate mainly to women.
In conclusion, in a relatively large community-based sample with a high prevalence of obesity and hypertension, we show that independent of confounders, the extent of IR influences whether hypertension translates into diastolic dysfunction. These data suggest that from a clinical perspective, hypertensives with IR may be particularly prone to the development of LV DD and thus possibly the progression to heart failure with a preserved ejection fraction. Consequently, targeting IR in hypertension may have marked benefits for preventing the development of heart failure.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.