OUP user menu

Alterations of left ventricular myocardial strain in obese children

Fabien Labombarda , Eva Zangl , Audrey Emmanuelle Dugue , Dominique Bougle , Arnaud Pellissier , Virginie Ribault , Pascale Maragnes , Paul Milliez , Eric Saloux
DOI: http://dx.doi.org/10.1093/ehjci/jes238 668-676 First published online: 16 November 2012


Aims Obesity may have implications in the myocardial structural change, which may contribute to mechanical consequences. Using 2D speckle echocardiography, we looked for myocardial changes and investigated their relation to obesity, inflammation, insulin resistance and physical capacity in children with isolated obesity.

Methods and results Standard echocardiography and 2D strain were prospectively performed in obese children and compared them with age- and sex-matched controls. Z-score body mass index (BMI Z-score), ultra-sensitive C reactive protein, indices of insulin resistance (HOMA-IR) and metabolic stress test were assessed in obese children. Thirty-two consecutive obese patients [age: 12.8 (8–17) years; 15 males; BMI Z-score: 5.8 [2.05–8.6)] were compared with 32 controls. Longitudinal strain and circumferential strain were significantly lower in the obese group (respectively −18.0 ± 2.4% vs. −20.6 ± 2.5%; P = 0.0001 and −18.2 ± 3.5% vs. −20.1 ± 2.3%; P = 0.013), while radial strain did not differ. Longitudinal strain was correlated with HOMA-IR (Pearson's rho = −0.39) and with the exercise capacity (Pearson's rho = 0.62). In the multivariate analysis, after adjusting for age, the mean arterial pressure and left ventricular (LV) mass, the BMI Z-score remained independently related to the longitudinal and circumferential strain.

Conclusion Childhood obesity may be associated with an early alteration of the longitudinal and circumferential LV strain. These findings have potentially significant clinical implications for the outcomes and follow-up of obese children meriting further studies.

  • Obesity
  • Children
  • 2D strain


The prevalence of excess weight and child obesity has been increasing rapidly for several decades.1 The somatic consequences of child obesity are rarely expressed clinically, except in patients with very severe obesity. In the adult, however, there are serious consequences. Longitudinal studies, with a 55- and 57-year follow-up have shown that excess weight and obesity during childhood were associated with increased cardiovascular morbidity and mortality during adulthood.2,3 This excess cardiovascular mortality is mainly attributed to coronary artery disease4,5 and congestive heart failure.6 Studies have demonstrated that child obesity generates lesions which are not expressed during childhood but which lead to life-threatening cardiovascular complications in adulthood.7 In view of the recent increase in the prevalence of child obesity, we can expect to see a recrudescence of cardiovascular morbidity and mortality over the next decades.8 In the adult, echocardiographic studies have demonstrated that obesity led to cardiac remodelling and alterations in the myocardial function characterised by ventricular and auricular dilatation, eccentric left ventricular (LV) hypertrophy, diastolic and systolic dysfunction.9,10 The relation between obesity and cardiac structure and function in children is less well documented. New echocardiographic techniques, such as 2D strain imaging, can be used to detect early alterations in the myocardial function.11,12 We used 2D strain imaging to investigate whether severely overweight children show early abnormalities in myocardial function. In addition we investigated the relations between these myocardial features and the severity of the obesity, the insulin resistance, the inflammation and the physical capacity.


Study population

We prospectively recruited obese patients between 5 and 17 years old followed up at the Paediatric department of the Caen teaching hospital. The exclusion criteria were syndromic obesity, history or clinical signs of cardiac disease, significant chronic illness, sleep apnoea syndrome, diabetes, dyslipidaemia, smoking, and hypertension. A 24-h blood pressure monitoring was systematically performed to exclude masked hypertension. We compared the obese patients with sex- and age-matched healthy controls undergoing echocardiography for heart murmur in whom the echocardiography was normal. To be included in the control group, children had to have no personal antecedent or family history of high blood pressure or hypercholesterolaemia. The children of the control group were enrolled prospectively.

Clinical assessment

The demographic details of age, gender, weight, height, and heart rate were recorded. The body surface area (BSA) was calculated according to the Dubois formula and expressed in square metre. The body mass index (BMI) was calculated according to the formula: BMI = weight (kg)/height (m)2. Obesity was defined when BMI adjusted to the sex and age, exceeded the 97th percentile according to the French reference values.13 The BMI was expressed in Z-score. The systolic blood pressure (SBP) and the diastolic blood pressure (DBP) were measured at rest after 10 min in the decubitus position with a calibrated automatic blood pressure monitor (Datascope® DUOTM).

Bioelectrical impedance analysis

In obese children, the fat mass and the fat-free mass were assessed by bioelectrical impedance analyzer (Bodystat® 1500 MDD bio-impedance analyser).


Fasting blood samples were taken in the obese children to analyse the lipid balance (total cholesterol, HDL cholesterol, triglycerides, and LDL cholesterol calculated according to the Friedwald formula; Beckman Coulter DXC 800, method: cholesterol oxydase, cholesterol esterase), fasting glycaemia, insulinaemia (Bi-insulin-IRMA (CIS bio International)) and ultra-sensitive C-reactive protein (us-CRP; Beckman Coulter DXC 800, method: turbidimetric, normal value <3 mg/l). The homeostasis model assessment–insulin resistance (HOMA-IR) was used as the insulin resistance index and calculated according to the formula: [fasting glycaemia (mmol/l) × insulinaemia (µmol/l)/22.5). An index of >2 defined the insulin resistance.

Metabolic stress test

A stress test with respiratory gas exchange measurements was conducted for each obese patient according to the Bruce protocol adapted to the patient's age. The functional exercise capacity was assessed using the maximal oxygen consumption (VO2 max) expressed as a percentage of the theoretical value according to the Wassermann standards (Oxycon Pro Eric Jaeger GmbH, Hoechberg, Germany).

Echocardiographic image: acquisition and analysis

Images were gathered with a standard ultrasound machine (iE33- Philips®, Best, The Netherlands) equipped with a 5-MHz phased-array transducer. Each acquisition was made in the left lateral decubitus position and apnoea. The electrocardiogram was recorded continuously. The echocardiographic studies were digitally stored and all the measurements were performed off-line. We chose the limit of 5 years old because, in our experience, most of the children younger than 5 years old are not able to cooperate for the apnoea during echocardiography.

Conventional echocardiography

The LV ejection fraction (LVEF) was assessed using the biplane Simpson's method in the apical view. The right ventricular end-diastolic dimension (RV-EDD), the LV end-diastolic dimension (LV-EDD), the interventricular septum end-diastolic thickness (IVS-EDT) and the LV posterior wall end-diastolic thickness (LVPW-EDT) were measured in a time motion (TM) mode in parasternal long-axis view. The LV mass (LVM) was calculated by Devereux's formula14 and indexed to the height raised to the power 2.7.15 The mitral Doppler signal was recorded in the apical four-chamber view, with the Doppler sample volume placed at the tip of the mitral valve. The peak velocities of early (E) and late (A) filling waves, early/late filling ratio of peak velocities (E/A) and the mitral deceleration time (MDT) were measured on the basis of transmitral flow velocities. Systolic pulmonary artery pressure (SPAP) was estimated from continuous-wave Doppler of tricuspid regurgitation using the Bernoulli equation. The tricuspid regurgitation velocity was recorded from the apical view and the parasternal short-axis view. The right atrial pressure was estimated by assessing the diameter of the inferior vena cava and the percentage decrease in its diameter during inspiration.

Tissue Doppler imaging

Acquisitions in the pulsed-wave tissue Doppler imaging (TDI) mode were made in a apical four-chamber view. The sample volume (4 mm thick) was placed at the basal level of the right ventricular and LV free walls to measure the tricuspid peak systolic velocity (St), early tricuspid peak diastolic velocity (Et), mitral peak systolic velocity (Sa), and early mitral peak diastolic velocity (Ea). For each TDI parameter, the average of three cycles was used.

2D strain analysis

For the 2D strain analysis, the standard two-dimensional acquisitions in a grey scale were made in the apical four-chamber incidence and in short-axis parasternal view at the mitral papillary muscles. All images were recorded with a high image rate of >50 Hz (a mean image rate in the short-axis parasternal view: 83 ± 13 Hz, mean image rate in the apical four-chamber view: 74 ± 11 Hz). Each incidence was recorded during a short apnoea in the form of three consecutive loops. The echocardiographic records were stored on the machine hard disk then transferred to a computer for post-processing analysis with QLAB advanced Quantification software (version 6.0 Philips®). In a short axis, the software performed automatic circular contouring that the operator adjusted and positioned on the endocardium-chamber interface. In the apical view, fast semi-automatic contouring of the endocardium was carried out by placing three points on the image (basal septum, basal lateral wall and apex) at the endocardium–chamber interface. The software then suggested a region of interest of adjustable thickness that could be repositioned by the operator and which must correspond to the thickness of the wall to be analysed. The operator ensured contouring and optimal tracking of the movements of each wall segment by the software. When the myocardial tracking was qualified as being optimal by the operator, the software was started to analyse the global and segmental strain represented as coloured curves. The strain peak corresponded to the mean of the strain peaks of each sub-region (six sub-region short-axis view, seven sub-regions in the apical view). The longitudinal strain was quantified from an apical incidence (Figure 1A). The circumferential strain and the radial strain were obtained from a short-axis parasternal incidence (Figure 1B).

Figure 1

2D strain imaging: multidirectional assessment of the left ventricular strain. Longitudinal strain is calculated at the apical four-chamber view; global longitudinal strain value is obtained from the average of seven segments (A). Radial and circumferential strains were calculated at the mitral papillary muscles short-axis view; radial strain and circumferential strain values are obtained from the average of obtaining segments (B).

Statistical analysis

Qualitative covariates were described using frequencies and percents, whereas quantitative ones, using mean and standard error, or median and IQR (inter-quartile range), depending on the normality assumption. Comparisons of quantitative variables between the obese and controls groups were done with a Student's t-test or a Wilcoxon's rank sum test (depending on the normality assumption) even though the controls were individually matched to obese. In fact, matching was used in order to have two groups globally comparable but pairs are weakly linked. As described by the STROBE (Strengthening The Reporting Of Observational Studies in Epidemiology), for case–control studies, if taking the matching into account does have a little effect on the estimates, authors may choose to present an unmatched analysis.16 Several correlations between strains and clinical measurements were also measured and tested by the Pearson's product-moment coefficient (estimated by r) and test. Multivariate analyses were also conducted in order to estimate and test the link of the BMI Z-score on each strain, after adjusting for independent possible risk factors. We added in the model all covariates [among age, indexed LVM, and mean blood pressure (MAP)] linked to the studied strain with a P-value of <20% in the univariate analysis. To take the intra-pair correlation into account, linear mixed models were used. The interactions between the independent factors and the strains were also tested in the multivariate models. Intra-observer and inter-observer variability for strain measurements was assessed using a coefficient of variation for a randomly chosen sample of 10 study children. The type I error was fixed at 5% for each test and the whole statistical analysis was computed using the R software version 2.12.1.

The Caen University Hospital ethical review board approved the study. The participants and at least one of their parents provided informed consent.



Thirty-two consecutive Caucasian obese patients (age: 8–17, 15 boys) were enrolled in this prospective study. Thirty-two healthy children, matched in sex and age with the obese patients, were extracted from a control group of 68 children enrolled prospectively during the same period.

The clinical and biological characteristics of our population are shown in Table 1. Although none of the obese children had a prior diagnosis of hypertension, the SBP was significantly higher in this group. The two groups were comparable in terms of DBP, mean arterial pressure (MAP) and heart rate. In the obese group, 12 children (37.5%) presented insulin resistance and 8 (25%) had an abnormal CRP value.

View this table:
Table 1

Physical and biological characteristics in obese and control children

Obese children (n = 32)Controls (n = 32)P
 Gender (M/F)15/17 (47%)15/17 (47%)
 Age (years)12.8 ± 2.112.8 ± 2.1
 Weight (kg)73.7 ± 2140.9 ± 7<0.0001
 Height (m)154.8 ± 13149.2 ± 11.10.071
 BSA (m²)1.83 ± 0.231.34 ± 0.24<0.0001
 BMI (kg/m²)30.2 (20.8–42.6)18.1 (16.5–19.6)<0.0001
 BMI Z-score6.25 (3.93–7.46)0 (−0.24–0.38)<0.0001
 Heart rate (bpm)82 (73–90)80.5 (74–93)0.62
 SBP (mmHg)114.8 ± 11.4106 ± 8.40.0015
 DBP (mmHg)65.6 ± 965.5 ± 60.95
 MBP (mmHg)82 ± 8.379.1 ± 6.20.13
 Fat mass (%)20.8 (13.7–32.1)
 Fat free mass (%)50 (43.6–61.15)
 Fat free mass/fat mass0.4 (0.3–0.6)
 Glucose (mmol/l)4.8 (4.6–5.3)
 Insulin (mU/l)9.6 (6.2–12.3)
 HOMA-IR2 (1.3–2.9)
 Total cholesterol (g/l)1.7 (1.4–2)
 LDL cholesterol (g/l)1.04 ± (0.4–1.64)
 HDL cholesterol (g/l)0.5 ± (0.34–0.8)
 Triglycerides (g/l)0.9 (0.7–1.2)
 Us CRP (mg/l)2.1 (0.8–5.3)
  • BSA, body surface area; BMI, body mass index; bpm, beat per minute; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; HOMA-IR, homeostasis model assessment of insulin resistance; US CRP, ultra-sensitive C reactive protein (qualitative covariates were described using frequencies and percents, whereas quantitative ones, using mean and standard error, or median and IQR (inter-quartile range), depending on the normality assumption).

Standard echocardiography and tissue Doppler

The echocardiographic data are shown in Table 2.

View this table:
Table 2

Echocardiographic characteristics in obese and control children

Obese children (n = 32)Controls (n = 32)P
2D standard parameters
 LV-EDD (mm)45.7 ± 5.943.3 ± 4.10.07
 IVS-EDT (mm)7.3 ± 1.46.4 ± 10.0049
 LVPW-EDT(mm)7 (6–7)7 (6–8)0.23
 RV-EDD (mm)22.9 ± 5.617.5 ± 3.1<0.0001
 LVM (g)116.7 ± 31.679.1 ± 19.2<0.0001
 LVM/height2.7(g/m2.7)0.14 ± 0.040.11 ± 0.02<0.0001
 LVEF (%)64 ± 6.563.2 ± 5.10.63
E (cm/s)108.6 ± 14.6106.9 ± 20.20.70
A (cm/s)63 ± 11.658.5 ± 13.70.17
E/A1.7 ± 0.481.8 ± 0.450.30
 MDT (ms)188.4 ± 29.3186.6 ± 24.20.80
 Sa (cm/s)11 ± 1.811.5 ± 1.90.23
 Ea (cm/s)16.9 ± 3.118.1 ± 2.50.08
 E/Ea6.8 ± 1.96 ± 1.50.06
 St (cm/s)14 (13–15)13 (12–14.2)0.20
 Et (cm/s)13.9 ± 1.814.4 ± 2.10.87
 SPAP(mmHg)25 (20–26.5)25 (22.5–25)0.60
2D strain
 Radial strain (%)28.8 ± 6.230.6 ± 4.40.21
 Circumferential strain (%)−18.2 ± 3.5−20.1 ± 2.30.013
 Longitudinal strain (%)−18 ± 2.4−20.6 ± 2.50.0001
  • LV EDD, left ventricular end diastolic dimension; IVS EDT, inter ventricular septum end diastolic thickness; LVPW EDT, left ventricular posterior wall end diastolic thickness; RV EDD, right ventricular end diastolic diameter; LVM, left ventricular mass; LVEF, left ventricular ejection fraction; E, mitral early peak velocity; A, mitral late peak velocity; MDT, mitral deceleration time; Sa, mitral annulus systolic peak velocity; Ea, mitral annulus early peak diastolic velocity; St, tricuspid peak systolic velocity; Et, tricuspid annulus early peak diastolic velocity, SPAP, systolic pulmonary artery pressure.

  • Bold values indicate significant difference at P < 0.05 level.

LV morphology and function

The IVS-EDD and indexed LV mass were significantly larger in the obese group. There was no significant differences between the two study groups with regard to LVEF, LV-EDD, LVPW-EDD, standard diastolic function parameters (E, A, E/A, and MDT) and TDI parameters. The results are shown in Table 2.

Right ventricular morphology and function

The right ventricle was significantly larger in the obese group. The right ventricular function and the SPAP were comparable between the two groups (Table 2).

2D strain analysis

6.3% of segments were excluded from the strain calculation because of inadequate tracking. The longitudinal strain and the circumferential strain were significantly lowered in the group of obese children, while radial strain did not differ (Table 2, Figure 2).

Figure 2

Comparison of radial strain (A), circumferential strain (B), and longitudinal strain (C) between obese children and controls.

Relationship between strain and the clinical, biological and echocardiographic parameters

The longitudinal strain and the circumferential strain were significantly correlated with the BMI Z-score (respectively: r = −0.5; P < 0.0001 (Figure 3) and r = −0.31; P < 0.01), while there was no correlation with the radial strain. The longitudinal strain was significantly correlated with the HOMA-IR (r = −0.39; P = 0.03), while radial strain and circumferential strain did not correlate with HOMA-IR (radial strain: r = −0.34; P = 0.06; circumferential strain: r = 0.072; P = 0.7). The us-CRP and the fat-free mass/fat mass ratio were not correlated with the various strains. We did not find a relationship between strain parameters and indexed LVM (radial strain: r = −0.016; P = 0.9; circumferential strain: r = 0.2; P = 0.11; longitudinal strain: r = 0.24; P = 0.058). In the multivariate analysis, after adjusting for age, MAP, and indexed LVM, the BMI Z-score remained independently related to the longitudinal strain and to the circumferential strain (Table 3).

View this table:
Table 3

Multivariate stepwise linear regression models with strain parameters

 Longitudinal strainCircumferential strain
BMI Z-score0.500.00010.310.003
Age (years)
LVM/height2.7 (+0.1)−1.290.25
  • BMI, body mass index; MBP, mean blood pressure; LVM, left ventricular mass; β, estimation of slope.

  • Bold values indicate significant differences at P < 0.05 level.

Figure 3

Relationship between BMI Z-score and longitudinal strain.

Relationship between strain and exercise capacity

Of those undergoing metabolic exercise testing (n = 27, 84%), the longitudinal strain was correlated with the reduced exercise capacity evaluated by VO2 max (r = 0.62; P < 0.05) (Figure 4)

Figure 4

Relationship between VO2 max (expressed in % of theoretical value) and longitudinal strain.


The intra-observer reproducibility was 7% for the longitudinal strain, 7% for the circumferential strain and 11% for the radial strain. The inter-observer reproducibility was 8% for the longitudinal strain, 9% for the circumferential strain and 13% for the radial strain.


Our study demonstrated that childhood obesity is associated with an alteration in the longitudinal and circumferential LV function. These modifications of the longitudinal and circumferential function appear to be related to the severity of the obesity and may contribute to reduced exercise capacity. The BMI is independently related to the longitudinal and circumferential strains, even after adjusting for age, mean arterial pressure and indexed LV mass.

LV morphology and function

The LV morphological modifications observed in our study are comparable with previous reports.17,18 This LV remodelling is matched with functional modifications. A certain degree of the diastolic dysfunction in the obese group may be suggested regarding the decrease in the circumferential strain. The circumferential strain is strictly connected with torsion, an essential promoter of the diastolic function. These data are consistent with the results of several studies conducted in obese patients, whether children or young adults, with no other cardiovascular risk factors.19,20 In our study, 2D strain revealed decreased LV longitudinal and circumferential strain in obese children, while the LVEF was preserved. The conclusions of the previous echocardiographic studies disagree concerning the LV systolic function in the obese child. Based on the LVEF study, most echocardiographic studies conducted on obese children have been concluded in the absence of LV systolic dysfunction.18,19 Our data are consistent with a study conducted by Di Salvo et al.21 which analysed the LV systolic function using TDI and reported a reduction in the longitudinal strain assessed in obese children without high blood pressure. The same observations were made in young adults with isolated obesity in whom only TDI or 2D strain were able to identify subclinical abnormalities of the LV function.20,22 To our knowledge, we performed for the first time an evaluation of the LV function in the obese child with the analysis of the radial, longitudinal and circumferential components of the contraction by 2D strain, currently the reference technique for the assessment of myocardial strain. 2D strain has demonstrated its superiority over TDI by eliminating the constraints generated by analysis of velocity, in particular of the Doppler angle, and by offering better reproducibility.23 In addition, unlike LVEF, 2D strain is relatively independent of the load conditions, making it an ideal tool with obese patients who exhibit a high preload. We did not find a significant difference between obese children and control for radial strain. Several potential hypotheses may be evoked to explain this result. The difficulty to assess the radial strain in parasternal short-axis view may be a first explanation. In this view radial strain is calculated perpendicularly at the longitudinal motion and is very influenced by the longitudinal out-of-plane motion, much more than the field of circumferential motion. Secondly, the small population of the study may explain that the radial strain did not differ significantly between the two groups. Nevertheless, a trend towards lower radial strain values in the obese group, without reaching a level of significance, has to be noted.

Potential mechanisms

The haemodynamic and neurohormonal consequences induced by obesity may explain the morphological and functional modifications reported in our study. The excess adipose tissue in obese patients is responsible for insulin resistance whose deleterious effects on the heart have been reported24: insulin may aggravate hypervolaemia due to salt and water retention; insulin may act as a myocardial growth factor25; insulin may activate the sympathetic nervous system with an increased response to angiotensin II which is already abnormally high due to the increase in the adipose tissue.26,27 The adverse effects of angiotensin II which acts as vasoconstrictor, hypertrophic and fibrosing agent are known. Some authors have suggested the existence of a distinct form of obese cardiomyopathy induced by insulin resistance.28,29 A proinflammatory state is detectable in obese children from the age of 3.30 Deleterious effect of the chronic inflammation on LV cardiac function have been reported: inflammation mediators induce extracellular matrix alterations, myocyte metabolism disturbance and hypertrophy, and negative inotropic effects31 and contribute to insulin resistance by inhibiting the action of the insulin.32 While no link has been established between us-CRP and obesity, the inflammation may favour early alteration of myocardial strain. Lastly, the excess of fatty acids released by the intra-abdominal adipocytes could exert direct toxicity on the myocyte, with the increase in myocyte triglyceride content likely to favour apoptosis.33

Clinical implications

Obesity generates lesions which remain silent during childhood but which are expressed in adulthood by excess cardiovascular mortality mainly attributed to coronary artery disease and heart failure. Severe childhood obesity is associated with early endothelial dysfunction and increased stiffness of elastic arteries which are established markers for coronary-artery atherosclerosis.7 Similarly, our results suggested that childhood obesity may be associated with early myocardial dysfunction; this myocardial dysfunction may represent a first step towards the development of heart failure. If confirmed in large-scale prospective studies, 2D strain could identify patients exhibiting a high risk of developing cardiac insufficiency. The presence of a possible early myocardial dysfunction may support the idea that childhood obesity must be actively treated. Weight loss in the child corrects deleterious cardiovascular metabolic abnormalities, such as insulin resistance. 2D strain could help assess the beneficial effects of weight loss by objectivizing the reversibility of the myocardial dysfunction at an early stage, as it has been suggested in the adult with TDI.34 Other studies are required to determine the prognostic value of these subclinical abnormalities of the cardiac function.


We have several limitations in this study. First, although it is a prospective work, we are aware that the small study population represents a limitation to draw a formal conclusion. Secondly, we included exclusively Caucasian patients and so we cannot extend our results to African children and then, to the whole population of obese children. Thirdly, metabolic blood tests and lipid studies were not performed on the healthy controls. At last, we did not correlate myocardial strain parameters with specific markers of renin–angiotensin and sympathetic nervous system, known to be increased in these sub-groups of children.


Our study suggested that childhood obesity may be associated with an early alteration in the longitudinal and circumferential LV strain. Further studies must be conducted to specify the prognostic value of these myocardial strain abnormalities detected by 2D strain in the obese child.


No financial assistance was received to support this study.


Acknowledgment to Dr David Jacobi, CHRU de Tours and Inserm U921, Université François-Rabelais de Tours, Tours, France for helpful discussion.

Conflict of interest: None declared.


View Abstract