Anthropometric and Biochemical Markers as Possible Indicators of Left Ventricular Abnormal Geometric Pattern and Function Impairment in Obese Normotensive Children

Εmerging data indicate that various effects of obesity on the cardiovascular system can be evident during childhood. The aim of this study was to detect early changes in left ventricular structure and function in obese normotensive children and explore possible associations of these changes with anthropometric and biochemical parameters. Normotensive 8–11-year-old obese and normal weight children were included in the study. They all underwent anthropometric measurements, laboratory tests, and echocardiography study by conventional and tissue Doppler to assess geometric pattern and function of left ventricle. Statistically significant differences in most anthropometric and metabolic parameters were noticed between groups. Obese children showed higher left ventricular mass index (LVMI) (40.05 ± 9.44 vs. 28.31 ± 6.22), lower E/A ratio (1.76 ± 0.33 vs. 2.08 ± 0.56), and higher E/e’ (6.04 ± 1.13 vs. 5.43 ± 0.96) compared to lean peers. Waist-to-height ratio and hs-CRP correlated significantly with E/A in the obese group. Left ventricular hypertrophy was present in 47.2% of obese children and eccentric was the prominent type. Waist-to-height ratio and serum cortisol levels in plasma increased the odds of having any type of abnormal ventricular geometric pattern. Echocardiographic evaluation of left ventricle and diastolic function could be considered for obese normotensive children based on waist-to-height ratio, hs-CRP, and serum cortisol.


Introduction
Childhood obesity, defined as abnormal or excessive fat accumulation that presents a risk to health, is one of the most challenging health problems around the world, with increasing rates in developed but also in developing countries during the last decades [1]. It is estimated that 40 million children under the age of 5 years and more than 330 million children and adolescents aged 5-19 years were overweight or obese in 2016 [2]. It seems that a high percentage of obese children tend to remain obese through adolescence and adulthood [3] with higher cardiovascular risk [4]. Overweight or obesity in adolescence may account for as much as 20% of cardiovascular deaths and 25% of deaths from coronary heart disease in midlife [5]. Incidence of comorbidities such as hypertension, insulin resistance, type 2 diabetes, non-alcoholic liver disease, and dyslipidemia seems to be higher in obese teens compared to normal weight peers [6].

Materials and Methods
Our sample was prospectively selected from children visiting the outpatient clinic of our Pediatric Cardiology Department for routine checkup and issue of health certificate, from September 2014 to October 2017. Inclusion criteria were age 7.5-11.5 years old, puberty stage Tanner 1 and BMI above 95th percentile according to IOTF cut off levels [25] for the case group or below 85th percentile for the control group. Exclusion criteria were cardiac disorders, hormonal disorders or any other chronic condition. Children and their guardians were fully informed about the protocol, approved by the Bioethics Committee of the Medical School of Aristotle University of Thessaloniki and written informed consent was given by parents (3850, 14-06-2012).
All subjects underwent full physical examination and complete medical history was recorded. Anthropometric measurements were taken twice from the right side of the body, with children standing still in erect position, by the same trained examiner. Height (in meters) was measured to the nearest 0.1 cm with a standard stadiometer (Seca 213, 22,089 Hamburg, Germany), weight (in kg) to the nearest 0.1 kg with a standard scale (Seca 813, 22,089 Hamburg, Germany) and various body circumferences to the nearest 0.1 cm with a standard measuring tape (Seca 201, 22,089 Hamburg, Germany), after removal of shoes and heavy clothes. Neck, waist, and hip circumferences were obtained by placing the tape just below the thyroid cartilage, on the midway between the lowest ribs and the iliac crest and at the level of great trochanters, respectively [26,27]. BMI was assessed from weight and height using Quetelet's equation (weight [kg]/height[m 2 ]). Additionally, skinfold thickness at three sites were measured with a standard caliper (Slimguide Skinfold): Triceps fold (back of the mid-upper arm), subscapular fold (below the shoulder blade) and suprailiac fold (on the middle of the distance between the lower rib and the top of iliac crest). Final value was the average of 3 consecutive cycles of measurements. Puberty status was assessed according to Tanner staging method by the same trained physician [28]. Waist-to-height ratio (WHtR) and waist-to-hip ratio (WHR) were also calculated [29,30].
Cardiac examination was conducted by the same experienced pediatric cardiologist. Firstly, blood pressure was measured after sitting quiet for 5 min, twice in both arms and in a leg, with an appropriately sized cuff, using a Dinamap monitor. BP values >90th percentile for age, height and gender [32] were considered elevated and patients were excluded, according to the study protocol. Echocardiography was performed afterwards by using a General Electric Vivid 3 medical system, with an appropriate for patient size probe, with the subject lying in the left lateral decubitus position.
M-mode parameters captured in the longitudinal axis included: Left atrial diameter (LAD), intraventricular septal thickness at diastole (IVSd), left ventricular end diastolic dimension (LVIDd), posterior wall dimension at diastole (LVPWd), left ventricular mass (LVM), intraventricular septal thickness at systole (IVSs), left ventricular end systolic dimension (LVIDs), posterior wall dimension at systole (LVPWs), and ejection fraction (EF). LVM was indexed (LVMI) to height in meters raised to the 2.7 power, as it is considered more reliable for pediatric patients [33]. Left ventricular hypertrophy (LVH) was defined by values above 40 gr/m 2.7 and 45 gr/m 2.7 , for girls and boys respectively [33]. Relative wall thickness (RWT) was calculated according to formula: (IVSd+LVPWd)/LVIDd and values above 0.41 were considered abnormal, as it represents the 95th percentile [34]. LVH was categorized into four patterns, based on LVMI and RWT. Normal when both LVMI and RWT were within normal limits, eccentric hypertrophy when only LVMI was elevated, concentric hypertrophy when both LVMI and RWT were abnormal and concentric remodeling when RWT was high in the presence of normal LVMI [35]. Pulsed wave Doppler measurements of the mitral inflow obtained in the apical four-chamber view with the sample volume placed between the mitral valve leaflet tips included: early diastolic mitral velocity (E), late diastolic mitral velocity (A) and E/A ratio (indicator of LV diastolic function).
Tissue Doppler imaging (TDI) was also performed to assess diastolic function. The sample volume was placed at the medial (septal) and lateral (wall) mitral annulus and e' and a' velocities were measured. The average of velocities from both sites was taken into account. The ratio of early mitral flow velocity (E) to early diastolic velocity of the mitral annulus (e') was used as a surrogate of LV filling pressure in early diastole, with high E/e' ratio reflecting impaired relaxation. Measurements were obtained with minimal angle of incidence (angle between the direction of wall motion and the Doppler beam).
Continuous variables were presented as mean ± standard deviation (SD) and categorical variables as frequencies and percentages. Normality was tested using the Kolmogorov-Smirnov test. Differences in continuous variables between the study groups were assessed using independent sample t-tests and Mann-Whitney U-tests for normally and abnormally distributed data, respectively. Spearman correlation co-efficient (rho) was used to describe the correlation between continuous variables. Variables found to be correlated by any of the above tests were then entered in a multivariate regression model. Stepwise multiple regression analysis was used to determine which independent predictor variables explained a significant fraction of the variance of the dependent variables. Linear or logistic regression was used accordingly to the nature of the dependent variable. Not normally distributed variables were log transformed in order to be included in regression analysis. Statistical significance was set at p < 0.05. Statistical analyses were performed using the Statistical Package for Social Sciences 25.0 program.

Results
A total of 62 children were enrolled in the study and the anthropometric and biochemical characteristics of children are shown in Tables 1 and 2. There were no differences in age and sex distribution between the 2 groups. However, statistically significant differences were revealed in all anthropometric parameters and biochemical markers except for total cholesterol, HbA1c and cortisol levels.  TSH was within normal limits across the sample. HOMA-IR was ≥3.16 in 38.9% of obese children. No differences in anthropometric, metabolic, and cardiac parameters between insulin resistant and the rest of obese children were noticed.
Systolic blood pressure was by protocol < 90th percentile, but was significantly higher in obese children. Diastolic blood pressure did not differ between groups.
Regarding echocardiographic findings (Table 3), structural parameters (LAD, IVSd, LVIDd, LVPWd, and LVMI) were significantly higher in obese children. Ejection fraction was within normal range and similar between 2 groups. Mitral inflow velocity, expressed by E/A ratio, was normal (above 1) in all children but significantly lower in the obese group (1.76 ± 0.33 vs. 2.08 ± 0.56).  TDI measurements revealed impaired diastolic relaxation in obese children expressed by a higher E/e' ratio (6.04 ± 1.13 vs. 5.43 ± 0.96).
Correlations by Spearman coefficient (rho) for both groups of LVMI, E/A, and E/e' with various parameters are shown in Table 4. When only obese children were analyzed then the only correlations maintained were those of LVMI with anthropometric parameters (BMI, WHtR, NC) and E/A with WHtR and hs-CRP levels. None of the recorded parameters correlated with E/e' in the obese group. (Table 4).
Among obese children 47.2% (12/36) had abnormal geometric pattern with a predomination of eccentric hypertrophy (Figure 1). WHtR and the log transformed plasma cortisol concentration were statistically different between those with and without remodeling. By means of binary logistic regression in obese children, the estimated odds ratio for the presence of any type of abnormal geometry increased for every one unit  WHtR and the log transformed plasma cortisol concentration were statistically different between those with and without remodeling. By means of binary logistic regression in obese children, the estimated odds ratio for the presence of any type of abnormal geometry increased for every one unit increase of WHtR (OR: 1.296, 95% CI:1.046-1.605), p: 0.018) and log serum cortisol (OR:17.305, 95% CI:1.62-185.08, p: 0.018) ( Table 5). In a multiple linear regression model only hs-CRP was significant predictor of E/A (Beta = −0.727, p = 0.012) with adjusted R 2 of 0.147.

Discussion
This study showed that changes in cardiac structure and diastolic function can be detected early in obese prepubertal children even in the absence of hypertension. Thickness of cardiac walls, left ventricular mass and indices of filling capacity of left ventricle, differed significantly between obese and lean subjects. Nearly 1 out of 2 children had some kind of abnormal pattern of LVH and the risk of remodeling increased in proportion to WHtR and serum cortisol levels. Diastolic function expressed by E/A correlated significantly with WHtR and hs-CRP levels.
Increased intraventricular septal thickness and higher LVMI in obese compared to normal weight children have been noticed previously in studies, varying in regard to sample size, blood pressure status, and puberty stage [16,36,37]. In our sample we tried to eliminate confounders like hypertension and puberty and still dimensions of left ventricle as well as LVMI were significantly associated to several anthropometric indices, confirming that obesity has an independent impact on cardiac structure. Interestingly, LVMI showed stronger relation to WHtR rather than BMI, probably implicating stronger influence of central adiposity [38,39], but this relationship was not verified in some studies [40,41].
Left ventricular hypertrophy has been traditionally considered a form of end organ damage in hypertensive population. Nearly half of the obese group met criteria for LVH, despite the fact that none of the children selected was hypertensive. This finding implies that excess of fat and consequent metabolic dysregulation may play a more important role in changes in left ventricle than hypertension in obese subjects [24]. Our data showed that elevated WHtR ratio and serum cortisol levels increased the odds for having any type of left ventricular remodeling. The possible role of cortisol excess has been investigated in patients with Cushing syndrome, in whom a high prevalence of LVH has been reported, not related to blood pressure levels [42] and reversible to some degree after successful treatment [43]. It has been proposed that possible mediators for the effect of cortisol excess to LVH are hypertension, enhancement of noradrenalin and angiotensin II responsiveness and cardiomyocyte proliferation [44,45]. It must be noted though that concentrations of morning cortisol in all children were within normal limits and did not correlate to indices of adiposity, like previously reported [46]. Perhaps adverse effects occur at lower circulating levels of cortisol in obese children, which has to be clarified by further studies.
The type of pattern of LVH in obesity remains a field of controversy. The Bogalusa Heart Study in young adults showed that adulthood and childhood BMI were significant determinants of eccentric LV hypertrophy, while the presence of diabetes mellitus in adulthood and diastolic blood pressure in childhood predicted the development of concentric LVH [47]. In obese subjects with concurrent hypertension, concentric hypertrophy and concentric remodeling seem to be more prevalent [16,48]. Results in obese children and adolescents are inconclusive with some of them showing higher percentages of eccentric [10,49], while others of concentric type [50,51]. The prominent type of hypertrophy in our sample was eccentric which, in the context of absence of hypertension, can be interpreted as an adaptation to increased preload due to greater metabolic requirements, circulating blood volume and cardiac output, caused by excess adiposity [24].
Significant differences in diastolic function between groups, expressed by lower E/A ratio and higher E/e', emerged in accordance with previous studies [21,22,52]. The ratio of peak early and late transmitral flow velocities E/A, correlated significantly with WHtR ratio and hs-CRP, but not with left ventricular mass like previously shown [53]. Both WHtR and hs-CRP are connected to visceral adiposity and low grade inflammation [54,55]. Waist-to-height ratio has been tested as a surrogate for abdominal obesity in adults and children and seems to be a better predictor of cardiometabolic risk than BMI [56,57]. High sensitivity CRP is a biomarker that quantifies low grade systemic inflammation, in the absence of overt systemic inflammatory or immunologic disorders and has been used as a predictor of cardiovascular events [58], but has been also associated with diastolic dysfunction and heart failure in adults [59][60][61]. In the study of Dahiya et al. in obese adolescents, CRP was an independent determinant of LV diastolic function [21]. Our study yielded significant association between hs-CRP in even younger children with shorter history of obesity. Thus it would be reasonable to suggest that increase in left ventricular filling pressures in obese children without further comorbidities are probably mediated by proinflammatory factors released by adipose tissue, rather than structural changes.
According to latest guidelines, TDI measurements are mandatory for the assessment of diastolic function in adults, since 3 of the 5 criteria are based on this method. Mitral annular diastolic velocity e' is less load dependent than conventional transmitral flow velocities and by using E/e' the effect of LV relaxation impairment on mitral E velocity is corrected [62]. Several pediatric studies incorporated TDI for the evaluation of diastolic function in obese children with conflicting results. Lambobarda et al. found no significant difference in E/e' ratio between obese children and lean controls [17] and the same applies to the study of Barbosa et al. in which only higher velocities a' of mitral annulus in late diastole were recorded [63]. On the contrary, Ghandi et al. reported a statistically significant dampening in E/e' in the case group [52]. In the present study, E/e' ratio was significantly higher in the obese group in accordance to previous studies [22,53,64] and correlated significantly with WHR, skinfold and uric acid levels when both groups were considered together. It must be noted that despite the significant differences between groups none of the children met adult criteria for diastolic dysfunction [62]. Nevertheless, these differences indicate subtle preclinical changes, which may persist and deteriorate into adulthood.
By using 3.16 as a cut off for HOMA-IR, about 40% of obese children showed impaired insulin sensitivity, presumably attributed to obesity as prepubertal children were selected in order to rule out as much as possible the confounding effect of puberty in insulin resistance. Insulin homeostasis model showed significant correlation to anthropometric indices and metabolic parameters. However, relation of HOMA IR to structural and functional cardiac parameters was not detected in our sample like previously reported [16,50,65].
There were certain limitations in our study. Our sample was relatively small but still statistically significant differences between the two groups emerged. Blood pressure levels were measured at a single visit which cannot rule out the masked hypertension and the white coat hypertension phenomena [66]. Assessment of LV by m-mode method is based on geometrical assumptions, but is considered fairly accurate in normally shaped ventricles. Thus m-mode is still recommended in the evaluation of LV mass and categorization into different geometric patterns [35]. It has been shown that reduced sensitivity to insulin can be present even before first signs of puberty and it is more pronounced in children with increased cardiometabolic risk [67], so selecting of prepubertal obese children cannot completely rule out the effect of insulin resistance. The cross sectional nature of the study does not allow conclusions about cause-effect relationships, however it seems that subclinical diastolic dysfunction is partly mediated by factors related to adipose tissue so further studies are needed to shed more light in the underlying pathophysiology and investigate the reversibility of these findings after weight loss.

Conclusions
Our study demonstrated that changes in geometry of left ventricle and diastolic function can be present in obese children even in the absence of hypertension. Moreover, parameters such as WHtR, hs-CRP and cortisol levels were predictors of these changes and could serve as possible indicators of obese children at greater cardiometabolic risk, requiring further echocardiographic evaluation. Future studies in larger samples could define proper cut off values of these easy to use parameters.