Triglycerides are related to left ventricular mass in hypertensive patients independently of other cardiometabolic risk factors: the effect of gender

Given the inconsistent results on the prognostic significance of triglycerides (TGs), the purpose of the present study was to investigate the association of plasma TGs with left ventricular mass (LVM) in hypertensive patients. We studied 760 never treated, non diabetic, hypertensive patients. Τransthoracic echocardiography was performed and LVMI was calculated according to the Devereux formula, adjusted to body surface area. Triglycerides were associated with LVMI after adjustment for age, gender, systolic blood pressure (SBP), smoking and fasting glucose (b = 0.08, p = 0.009). This relationship remained significant even after adjustment for BMI, LDL-C and ApoB/ApoA1 ratio (b = 0.07, p = 0.04). Gender-stratified analysis indicated that TGs were related to LVMI in men (p = 0.001) but not in women (p = NS). In addition, TGs were related with LV hypertrophy (LVH) in men, increasing the odds by 7% to present LVMI over 115 g/m2 (OR = 1.07 per 10 mg/dl increase in TGs, p = 0.01). In conclusion, TGs are associated with LVMI in hypertensive patients, independently of other risk factors, including LDL-C. Given the prognostic significance of LVH, it might be suggested that TGs may serve as a useful marker for indentifying hypertensive patients at high risk. The gender discrepancy may suggest a possible gender-specific modulatory effect of TGs on LV structure.

www.nature.com/scientificreports/ Given the prognostic significance of LVH and the limited data on the role of plasma TGs in LV structure of hypertensive patients, we sought to investigate the association of LVM with plasma TGs in never treated hypertensive male and female patients.
In the whole population, abnormal glucose metabolism was demonstrated in 37% of patients, defined either as impaired fasting glucose (IFG) or impaired glucose tolerance (IGT). Patients with IFG/IGT had increased LVMI compared to hypertensive patients with normal glucose metabolism (124 vs 118 g/m 2 , p < 0.001).
In multiple regression analysis, in the whole population, LVMI was related to plasma TGs after controlling for age, gender, smoking, SBP, and plasma fasting glucose (b = 0.08, p = 0.009). This relationship remained significant even after adjustment for BMI, LDL-C and ApoB/ApoA1 ratio ( Table 2). An interaction term of plasma TGs with gender was significant in the final multivariable model, both in continuous and dichotomous format (p = 0.009 and p = 0.03 for TGs as a continuous variable*gender and hypertriglyceridemia*gender, respectively). Stratified analysis by gender indicated that TGs were differentially associated with LVMI. In particular, TGs were related to LVMI in men (b = 0.15, p = 0.001) but not in women (p = NS) after adjustment for age, smoking, SBP and plasma fasting glucose (Fig. 1). This association was remained significant even after adjusting for additional confounders, including BMI, plasma LDL-C and ApoB/ApoA1 ratio.
In the final step of our analysis, we identified male subjects that presented LVH and tested if plasma TGs were independently associated with LVH in this subgroup. Importantly, plasma TGs were associated with LVH (OR = 1.07 per 10 mg/dl increase in plasma TGs, 95% CIs 1.02-1.14, p = 0.01) (Fig. 2) and increased the odds by 7% to present LVH over 115 g/m 2 after controlling for age, smoking, SBP and plasma fasting glucose. This association was not attenuated when other cardiometabolic risk factors, including BMI, plasma LDL-C and ApoB/ApoA1 ratio, were taken into account. Plasma TGs reclassified male subjects into correct categories for the presence or absence of LVH over the core model (age, smoking, SBP and fasting glucose) (continuous net reclassification index, NRI = 38.2%, p = 0.001). The incremental value of TGs over established cardiometabolic risk factors was evident even after additional adjustment for BMI, LDL-C and ApoB/ApoA1 ratio (continuous NRI = 24.1%, p = 0.04). On the contrary, plasma TGs were not associated with LVH in females (Fig. 3).  Figure 1. Differential association of LVMI with plasma TGs in men (blue dots and line) and women (green dots and line). Beta coefficients and p-values are derived from multiple regression analysis of (log) LVMI on TGs after controlling for age, SBP, fasting glucose and smoking.

Figure 2.
Difference in plasma TGs between hypertensive males with and without LV hypertrophy (LVMI below or above 115 g/m 2 ). p-value was derived from logistic regression analysis after adjustment for age, SBP, plasma fasting glucose and smoking.  10 . Limited data in hypertensive population come from small studies which have shown a relationship of low plasma HDL-C with LVM and diastolic function [7][8][9] . Plasma triglycerides have been associated with LVH 11 , although with inconsistent results 8 . However, in the above studies, plasma LDL-C was not included as a confounder in the linear regression analysis and the incremental value of plasma TGs was not investigated.
Our present findings support a strong, independent association of plasma TGs with LVMI in hypertensive patients, over and above plasma LDL-C. Given the prognostic significance of LVH, this finding may, at least in part, explain the residual cardiovascular risk observed in patients with optimal plasma LDL-C levels.
The pathophysiological mechanisms underlying this relationship are not fully elucidated. However, it might be argued that insulin resistance may mediate part of the association of LVMI with plasma TGs given the data on the unfavorable role of insulin resistance on LVstructure and function [12][13][14][15] . To further support this notion, the strong independent relationship of LVMI with fasting glucose in our population may imply a possible modulatory role of abnormal glucose metabolism, through the mechanism of insulin resistance. Although we recruited nondiabetic patients, a part of our population exhibited abnormal glucose metabolism defined either as IFG or IGT. This sub-population had increased LVMI compared to hypertensive patients with normal glucose metabolism. Data have shown that non-diabetic individuals with IFG and IGT have a 3 to tenfold greater probability of LVH compared to subjects with normal glucose tolerance 16 . Aortic stiffness, a major determinant of LVH may also serve as a potential mediator to the relationship of plasma TGs with LVMI given previous results demonstrating an association of plasma TGs and impaired glucose metabolism with aortic stiffness [17][18][19][20] .
From a mechanistic perspective, a more prominent underlying mechanism explaining the adverse effect of atherogenic dyslipidemia, mainly plasma TGs, on LVstructure may be related to cardiac steatosis, a recently recognized cardiometabolic condition that associates hypertriglyceridemia with LVH. In physiological conditions, most of the energy used by the myocardium is derived from beta oxidation of fatty acids. Most of the fatty acids inserting myocardium are used for energy production, whereas only a small amount is stored in the intracellular myocardial lipid pool 21,22 . When there is an imbalance between lipid storage and lipolysis in cardiomyocytes, cardiac steatosis and myocardial hypertrophy are observed 23 . Human studies have demonstrated an increased myocardial triglyceride content in patients with diabetes mellitus II 24 and generalized lipodystrophy 25 , which was independently associated with concentric LV remodeling in these group of patients. Use of imaging modalities, such as the magnetic resonance spectroscopy, for the measurement of myocardial lipid content in hypertensive patients, may elucidate the potential role of cardiac steatosis as an underlying mechanism for the association of LVMI with plasma TGs.
The intriguing finding of the disassociation of LVMI with plasma TGs between men and women, also merits attention. A possible explanation for the lack of an independent association between LVMI and plasma TGs in women may be attributed to the difference in tissue metabolic profile and myocardial energy use between the two genders. Recently, an experimental study showed that females, compared to males, may be protected from cardiomyopathy through mechanisms related to genetically-determined normal cardiac glucose uptake and preserved cytochrome c oxidase activity 26 , a mitochondrial enzyme regulating tissue metabolism and energy production. Future, translational research studies need to confirm the above, novel, experimental data and shed new light into the pathophysiology of LVH. www.nature.com/scientificreports/ Finally, the absence of measured insulin levels and the subsequent estimation of insulin resistance that could explain, at least in part, the observations of the present study, may be considered as a possible limitation of our study. Moreover, the lack of studying a non-hypertensive group should also be taken into consideration given that the demonstration of a strong association of plasma TGs with LVMI in a non-hypertensive, high risk population would strengthen the results of the present study and would extend the clinical implications in other populations.

conclusion
Although no etiological relationships can be established from the present study, our intriguing finding of the independent relationship of LVMI with plasma TGs in hypertensive patients may have important clinical implications. Considering the adverse prognostic role of LVH, it might be suggested that the plasma TGs, may explain part of the residual cardiovascular risk observed in patients with optimal plasma LDL-C levels. Moreover, TGs may serve as a useful marker for identifying hypertensive patients at high risk of LVH. Future studies need to clarify whether treatment of increased plasma TGs may promote LVH reduction in hypertensive patients. Finally, the gender discrepancy regarding the association of LVMI with plasma TGs provides new insights into the pathophysiology of LVH in both genders and gives the impetus for further research.

Methods
Study population. This is a cross-sectional study. We studied 760 non-diabetic, never treated, hypertensive patients recruited from Hypertension Unit of 1st Cardiology Department from 2007 to 2015. Arterial hypertension was defined according to the European Society of Hypertension Guidelines 27,28 . Briefly, office blood pressure (BP) was measured by an oscillometric sphygmomanometer, taking at least three measurements spaced by 1 min, allowing the patients to rest for 10 min before examination. Measurement of brachial SBP ≥ 140 mmHg and DBP ≥ 90 mmHg were considered as systolic and diastolic arterial hypertension, respectively. Mean arterial pressure was calculated as DBP + 1/3 (SBP-DBP).
Plasma total cholesterol, HDL-C, LDL-C, TGs and glucose were measured with standard techniques by autoanalyzers after fasting for 12 h. Patients with fasting glucose levels ≥ 100 mg/dl were subject to 2 h oral glucose tolerance test. Plasma levels of ApoA1 and ApoB were measured by nephelometry (QUED, OSAN for ApoA1 and ApoB, respectively, Behring Diagnostics, Marburg, Germany).
Patients with heart failure, significant valvular disease, cerebrovascular disease and chronic, systemic diseases were excluded from the study. Lipid lowering treatment was also an exclusion criterion. Height and weight were recorded and body mass index (BMI) was calculated. The study was conducted according to the principles of Helsinki Declaration. All patients gave their informed consent to participate in the study which was approved by the Institutional Research and Ethics Committee (Athens Medical School, University of Athens).

Statistical analysis.
Continuous variables are presented as mean value ± standard deviation or median value. Categorical variables are presented as absolute frequencies and percentages. Normal distribution of continuous variables was evaluated by the Kolmogorov-Smirnov test and graphically by histograms.
Difference in baseline characteristics between males and females were assessed by independent samples Student's t-test or non-parametric Mann-Whitney test for continuous variables and chi-squared test for nominal ones. Multiple regression analysis was used to test the association of plasma TGs with LVMI after adjustment for age, gender, smoking, BMI, SBP, plasma fasting glucose, LDL-C and ApoB/ApoA1 ratio.
In order to assess the differential effect of gender on LVMI, we forced included relevant interaction terms (i.e. gender*parameter of interest) in the final model. Nested regression models, with and without interaction, were compared by Log-Likelihood ratio test. In case of significant interaction-term, stratified analysis by gender was performed. Finally, exploratory analysis implemented dichotomous LVMI (LVH yes/no) and a. evaluated the independent association of plasmaTGs with the outcome (LVH) in the multiple regression analysis and b. assessed the prognostic significance of plasmaTGs beyond the model that included age, gender, SBP and LDL-C by calculating continuous NRI 29 .
Exact p values < 0.05 were considered as statistically significant. Data analysis was performed with SPSS software, version 17.0 (Chicago, IL, USA).