OUP user menu

★ Editor's choice ★

Ventricular remodelling post-bariatric surgery: is the type of surgery relevant? A prospective study with 3D speckle tracking

Thomas E. Kaier, Douglas Morgan, Julia Grapsa, Ozan M. Demir, Stavroula A. Paschou, Shweta Sundar, Sherif Hakky, Sanjay Purkayastha, Susan Connolly, Kevin F. Fox, Ahmed Ahmed, Jonathan Cousins, Petros Nihoyannopoulos
DOI: http://dx.doi.org/10.1093/ehjci/jeu116 1256-1262 First published online: 25 June 2014


Aims The aim of the study was to examine ventricular remodelling in patients free of cardiac risk factors, before, and 6 months post-bariatric surgery with the new imaging modality of three-dimensional (3D) strain and the comparison of two surgical techniques: sleeve gastrectomy vs. gastric bypass.

Methods and results Fifty-two consecutive patients referred to the Bariatric Services of Imperial College NHS Trust were examined with conventional 2D and 3D strain echocardiography, prior to and 6 months after bariatric surgery. They were all free from cardiac disease. The study cohort's mean age was 44.2 ± 8.7 years and body mass index of 42.4 ± 4.6 g/m2 prior to surgery. Eighteen patients (34.6%) underwent laparoscopic sleeve gastrectomy, and 34 laparoscopic gastric bypass. On 3D speckle tracking, there was significant reverse remodelling post-bariatric surgery [left ventricular (LV) ejection fraction (EF): pre-surgery: 59 ± 8% vs. post-surgery: 67 ± 7%, P < 0.001 and right ventricular (RV) EF: pre-surgery: 60 ± 9% vs. post-surgery: 68 ± 8.2%, P = 0.0001]. Furthermore, there was significant regression of mass (LV mass: pre-surgery: 111 ± 23.5 g vs. post-surgery: 92.8 ± 15.5 g and RV mass: pre-surgery: 95.2 ± 19.8 vs. post-surgery: 67.3 ± 16.3, P < 0.001). RV and LV global strain improved 6 months post-bariatric surgery: global RV strain: pre-surgery −11.7 ± 4 vs. post-surgery −17.52 ± 3.7, P < 0.001; global LV strain: pre-surgery: −20.2 ± 1.7 vs. post-surgery: −26.5 ± 1.86, P < 0.001. Sleeve gastrectomy and gastric bypass had comparable effects.

Conclusion Bariatric surgery has an important effect in reverse LV and RV remodelling and it substantially improves RV longitudinal strain.

  • Remodelling
  • Bariatric surgery
  • Three-dimensional
  • Speckle tracking
  • Obesity


Obesity, the epidemic of the 21st century, has significant impact on morbidity and mortality through a variety of consequences such as cardiovascular complications.1,2 Obesity adversely affects the circulatory system36 with resultant endothelial dysfunction promoting systemic hypertension, coronary artery disease, and vascular calcification. In addition, obesity causes intrinsic changes in the heart including an increase in left ventricular (LV) mass, LV hypertrophy, LV and left atrial (LA) dilatation, and diastolic as well as systolic dysfunction in some cases. Also in obesity, the Frank–Starling curve is shifted to the left because of the increase in LV filling pressure and volume resulting in LV dilatation and, eventually, systolic dysfunction.5 The surgical treatment of obesity is defined as bariatric surgery. The two main types of bariatric surgical procedures are (i) restrictive—a gastric band and (ii) combined restrictive and/or altered absorption—such as Roux-en-Y gastric bypass and sleeve gastrectomy. Currently, most obesity clinics and bariatric centres13 favour the adjustable gastric banding procedure and the proximal Roux-en-Y gastric bypass. Sleeve gastrectomy is rapidly becoming a frequently used alternative to gastric bypass.

Many studies have discussed the ventricular changes after bariatric surgery as assessed with conventional echocardiography.713 Most of the studies comprised of a population with confounding factors, such as hypertension or hypercholesterolaemia. The aim of the study was to examine ventricular remodelling in a group of patients free of cardiac risk factors, before, and 6 months post-bariatric surgery using three-dimensional (3D) strain and to compare two surgical techniques: sleeve gastrectomy vs. gastric bypass. A trend towards a difference in the remodelling effect of either technique would suggest a hormonal or nutritional impact on remodelling.


Fifty-two consecutive patients referred to the Bariatric Services at Imperial College NHS Trust were examined with conventional 2D and 3D strain echocardiography, prior to and 6 months after bariatric surgery. Patients were eligible for the study if they underwent laparoscopic sleeve gastrectomy or gastric bypass surgery and were free from cardiac disease or associated risk factors such as hypertension, hypercholesterolaemia, diabetes, and family history of heart disease. Patients with obstructive sleep apnoea were excluded from the study, as it may influence pulmonary pressures and right ventricular (RV) systolic function.

Initially, a comprehensive baseline 2D echocardiography (2DE) examination was performed for the detailed anatomical description and estimation of ventricular dimensions and function. All patients underwent a detailed assessment of diastolic dysfunction, including tissue Doppler imaging and the measurement of LA volume index. All the 52 patients were assessed for biventricular volumes, ejection fraction (EF), and mass. RV longitudinal strain was assessed with 3D strain.

The local ethics committee has approved the study and the subjects gave written informed consent. The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.

Three-dimensional echocardiography


The same operator performed all examinations (J.G.—8 years of experience on 2DE and 3DE with European Accreditation). A second operator (S.S.) with 3 years' experience on echocardiography was employed for the assessment of inter-observer reproducibility of 3D RV strain. 3DE acquisitions were obtained using the Philips IE 33 equipped with a central X3 transducer (frequency of 3–4 MHz, volumetric frame rate 16–24 frame/s, imaging depth 6–16 cm, rotation speed 6 Hz, and pulse length 2.5 cycles). Images were acquired from apical four-chamber views with the patient in the left decubitus position during a breath hold of 7 s. Care was taken for the views to be acquired laterally while on the apical four-chamber view, for RV optimization. In addition, the depth and width were adjusted to include the RV apex. Furthermore, the sector width was reduced to include solely the RV cavity and by including four beats of acquisition, we achieved a frame rate of >30–35 fps during full-volume acquisition.


Images were then transferred to an offline workstation (4D analysis, TomTec, Munich, Germany). Serial short-axis reconstructions of biventricular volumetric data sets were then obtained and endocardial contour was traced at 7 mm intervals with cross-reference to long-axis images for identification of the tricuspid annulus. End-diastolic phase was defined as the peak of the R-wave of the QRS complex and end-systolic as the first frame before opening of the tricuspid and mitral valve. Papillary muscles and trabeculations were excluded from endocardial mapping. End-diastolic and end-systolic biventricular volumes and EF were calculated offline using the method of summation of discs and semi-automated border detection. Stroke volume (SV) was calculated by the subtraction of the end-systolic volume (ESV) from end-diastolic volume (EDV), while EF was calculated as EDV − ESV/EDV. Valve opening and closure were identified for the accurate volume determination.

Using the same full-volume 3DE data set, epicardial boundaries of the RV wall were identified and traced to calculate an epicardial cast of the RV at end-diastole. The volume of this cast was then subtracted from the endocardial cast and the volume of the RV myocardium was derived. By multiplying myocardial volume by the density of myocardial muscle (1.05 g/mL), RV mass was calculated. Care was taken to distinguish the endocardium from the ventricular septum.

Three-dimensional speckle analysis was performed using the Tomtec 4D speckle tracking software on a Philips IE33 ultrasound machine. As the software is designed for the LV, a modified methodology was devised and employed for the RV (Figure 1). We therefore assessed both LV and RV indices.

Figure 1:

Tomtec 4D speckle tracking software with modified methodology for the right ventricle.

The long-axis A-plane was adjusted so that it would cut directly through the centre of the RV, dividing the septum and RV free wall in half, and through the distal apical portion. The long-axis B-plane was aligned in order to achieve maximum proximity (but without intruding) to the inter-ventricular septum and through the distal apical portion, so that the anterior segments could be reliably differentiated as RV free wall. We included the sum of the six segments for the RV and, separately, the three segments of the RV free wall.

Statistical analysis

Results are presented as mean ± SD for continuous variables. Results are presented as absolute numbers or percentages for categorical variables. Differences in continuous variables between groups were tested using independent t-test, Mann–Whitney U-test or paired t-test, as appropriate. Differences in categorical variables between groups were tested using the χ2 test with Yates Correction. All statistical analyses were performed using the Statistical Package for Social Sciences (SPSS 16.0, Inc., IL, USA). A P-value of <0.05 was considered statistically significant.



The study cohort constituted of 52 bariatric patients with satisfactory acoustic windows and a mean age of 44.2 ± 8.7 years and a body mass index (BMI) of 42.4 ± 4.6 g/m2. Eighteen patients (34.6%) underwent laparoscopic sleeve gastrectomy, while 34 patients had a laparoscopic gastric bypass. Twelve patients (23%) were excluded from the 3D strain assessment due to poor acoustic windows for 3D. All demographics are demonstrated on Table 1.

View this table:
Table 1

Demographics of 52 patients undergoing bariatric surgery

Echocardiographic valueMean/median (SD)
Age (years)44.2 (±8.7)
Body mass index (kg/m2)42.4 (±4.6)
Body surface area (m2)2.9 (±0.8)
Systolic blood pressure (mmHg)125 (±19)
Diastolic blood pressure (mmHg)68 (±11)
Heart rate (bpm)78 (±12)
Haemoglobin (g/dL)13.5 (±0.8)
Total cholesterol4.9 (4–5.5)
High density lipoprotein (mmol/L)1.3 (1.1–1.5)
Glucose (mmol/L)5.3 (4.9–6)
HbA1C (%)5.5 (5–6.2)
Creatinine (μmol/L)78 (70–95)
Alanine aminotransferase (U/L)32 (25–40)
  • Continuous data are presented as mean (SD) or median (25–75th percentiles).

Comparison between pre- and post-bariatric surgery

Six months post-bariatric surgery, all patients managed to achieve significant reduction of BMI: pre-surgery: 42.4 ± 4.6 vs. post-surgery: 31.5 ± 2.6, P < 0.001.

Conventional echocardiographic indices

All echocardiographic indices were initially indexed to body surface area (Table 2, for full dataset see Supplementary data online). There was a significant reduction in wall thickness post-bariatric surgery (inter-ventricular septum thickness: pre-surgery: 10.17 ± 1.17 mm vs. post-surgery: 8.15 ± 0.9 mm, P < 0.001) and in LV dilatation (end-diastolic diameter: pre-surgery: 47.9 ± 4.2 mm vs. post-surgery: 43.6 ± 3.5 mm, P < 0.001). While the patients demonstrated impaired diastolic filling (E/A ratio: pre-surgery: 0.98 ± 0.3, E/E′ = 5.2 ± 0.4) pre-surgery, this improved significantly post-bariatric surgery: E/A ratio post-surgery: 1.3 ± 0.26, P < 0.001. E/E′ ratio remained similar (4.9 ± 0.6, P = 0.64). LA volume index was similar after surgery (pre-surgery: 30.9 ± 7 mm/m2 vs. post-surgery: 28.9 ± 6.7 mm/m2, P = 0.8).

View this table:
Table 2

Comparison of all patients before and after bariatric surgery

Echocardiographic indexPre-bariatric surgeryPost-bariatric surgeryP-value
BMI (g/m2)42.4 ± 4.631.5 ± 2.6<0.001
 Sleeve gastrectomy vs. gastric bypass43.6 ± 540.4 ± 4.231.4 ± 2.830 ± 1.90.416
LVFS (%)37.91 ± 5.742.6 ± 6<0.001
 Sleeve gastrectomy vs. gastric bypass47.3 ± 447.8 ± 4.142.2 ± 7.443.1 ± 3.40.87
LA diameter (mm)40.8 ± 2.934.9 ± 4<0.001
 Sleeve gastrectomy vs. gastric bypass41.2 ± 3.742 ± 235.4 ± 4.636.2 ± 3.60.663
A-wave0.83 ± 0.230.75 ± 0.160.0625
E-wave0.79 ± 0.220.98 ± 0.16<0.001
E/A ratio0.98 ± 0.31.3 ± 0.26<0.001
DT (ms)173.7 ± 44.5190.1 ± 36.90.0246
E/E′ ratio5.2 ± 0.44.9 ± 0.60.64
LA volume index (mm/m2)30.9 ± 728.9 ± 6.70.8
RVSP (mmHg)30.4 ± 11.922.8 ± 5.70.0046
RVMPI0.37 ± 0.120.24 ± 0.06<0.001
IVRT (ms)109.6 ± 11.357.1 ± 25.4<0.001
RVOT-AT (ms)91.4 ± 17.5107.2 ± 12.35<0.001
RV-S wave (cm/s)11.92 ± 3.513.08 ± 2.9<0.001
TAPSE (mm)26.3 ± 5.825.6 ± 5.50.56
LVEDV (mL)169.4 ± 43.2120.36 ± 19.6<0.001
 Sleeve gastrectomy vs. gastric bypass169.6 ± 65.4142.4 ± 29.5123.3 ± 25.6120.7 ± 15.80.23
LVESV (mL)68.87 ± 2339.2 ± 10<0.001
 Sleeve gastrectomy vs. gastric bypass61.9 ± 24.662.8 ± 26.538.6 ± 10.343.1 ± 130.528
LVSV (mL)100.5 ± 29.881.1 ± 18.20.001
 Sleeve gastrectomy vs. gastric bypass61.9 ± 24.662.8 ± 26.584.7 ± 22.277.6 ± 160.023
LVEF (%)59 ± 867 ± 7<0.001
 Sleeve gastrectomy vs. gastric bypass63.3 ± 6.657.2 ± 11.568.2 ± 764.2 ± 9.50.032
LV mass (g)111 ± 23.592.8 ± 15.5<0.001
 Sleeve gastrectomy vs. gastric bypass115.5 ± 25.5106.4 ± 16.499.3 ± 19.693.8 ± 17.40.969
RVEDV (mL)143.2 ± 33.2109.6 ± 15.7<0.001
 Sleeve gastrectomy vs. gastric bypass146.5 ± 29.6128 ± 27114 ± 10.5113.9 ± 160.907
RVESV (mL)57.2 ± 20.835.1 ± 10.4<0.001
 Sleeve gastrectomy vs. gastric bypass57.3 ± 22.451.5 ± 18.236 ± 1237.2 ± 12.2
RVSV (mL)86 ± 20.474.5 ± 14.30.0048
 Sleeve gastrectomy vs. gastric bypass89.2 ± 13.776.4 ± 17.878 ± 1176.7 ± 19.80.742
RVEF (%)60 ± 968 ± 8.20.0001
 Sleeve gastrectomy vs. gastric bypass61.9 ± 959.9 ± 9.568.6 ± 966.7 ± 110.702
RV mass (g)95.2 ± 19.867.3 ± 16.3<0.001
 Sleeve gastrectomy vs. gastric bypass94.87 ± 16.789.6 ± 19.371.7 ± 1874.2 ± 19.3
LV-global strain (%)−20.9 ± 2.3−26.5 ± 1.86<0.001
 Sleeve gastrectomy vs. gastric bypass−20.2 ± 1.719.4 ± 1.1−26.7 ± 1.3−27.09 ± 2.4
RV-global strain (%)−11.7 ± 4−17.52 ± 3.7<0.001
 Sleeve gastrectomy vs. gastric bypass−10.1 ± 4−11.8 ± 3.2−19.1 ± 3.7−17 ± 4.1
RV-free wall strain (%)−11.8 ± 6−18.3 ± 4.4<0.001
 Sleeve gastrectomy vs. gastric bypass−12.6 ± 5.111.3 ± 2.9−19.2 ± 3.7−16.9 ± 3.6
  • BMI, body mass index; LV, left ventricular; RV, right ventricular; LVEDV, LV end-diastolic volume; LVESV, LV end-systolic volume; LVSV, LV stroke volume; LVEF, LV ejection fraction; LVmass: LV mass; RVEDV, RV end-diastolic volume; RVESV, RV end-systolic volume; RVSV, RV stroke volume; RVEF, RV ejection fraction; RVmass: RV mass; LVFS, left ventricular fractional shortening; DT, deceleration time; RVSP, right ventricular systolic pressure; RVMPI, RV myocardial performance index; IVRT, isovolumic relaxation time; RVOT-AT, RV outflow tract acceleration time; RV-S, velocity of the tricuspid anular systolic motion; TAPSE, Tricuspid annular plane systolic exertion.

There was a significant reduction in pulmonary pressures following bariatric surgery (RV systolic pressure: pre-surgery: 30.4 ± 11.9 mmHg vs. post-surgery: 22.8 ± 5.7 mmHg, P = 0.0046). All RV indices indicated improvement of diastolic function post-surgery, except the tricuspid annular plane systolic excursion measurement which remained comparable: pre-surgery: 26.3 ± 5.8 mm vs. post-surgery: 25.6 ± 5.5 mm, P = 0.56.

Three-dimensional speckle tracking

There was clear reverse remodelling post-bariatric surgery of both ventricles (Table 2):

  1. LV EF: pre-surgery: 59 ± 8% vs. post-surgery: 67 ± 7%, P < 0.001.

  2. RV EF: pre-surgery: 60 ± 9% vs. post-surgery: 68 ± 8.2%, P = 0.0001.

  3. LV mass: pre-surgery: 111 ± 23.5 g vs. post-surgery: 92.8 ± 15.5 g.

  4. RV mass: pre-surgery: 95.2 ± 19.8 g vs. post-surgery: 67.3 ± 16.3 g, P < 0.001.

RV longitudinal strain significantly improved 6 months post-bariatric surgery:

  1. Global RV strain: pre-surgery: −11.7 ± 4 vs. post-surgery: −17.52 ± 3.7, P < 0.001 and strain of RV free wall: pre-surgery: −11.8 ± 6 vs. post-surgery: −18.3 ± 4.4, P < 0.001.

  2. Global LV strain: pre-surgery: −20.9 ± 2.3 vs. post-surgery: −26.5 ± 1.86, P < 0.001.

Comparison of laparoscopic sleeve gastrectomy vs. gastric bypass

Eighteen patients (34.6%) underwent laparoscopic sleeve gastrectomy, while 34 patients had a laparoscopic gastric bypass. The two groups of different surgical techniques were compared before and after surgery (Table 2), and there was no significant difference in the echocardiographic indices measured.

Intra- and inter-observer reproducibility of 3D strain

Thirty-one patients were examined for intra- (J.G.—8 years of echocardiography experience) and inter-observer (J.G. and S.S.—3 years of echocardiography experience) reproducibility for RV EF and RV global strain (Figure 2). There was good intra-observer reproducibility for RV EF (ICC = 0.78, P < 0.001, mean bias = −1.5%, SD of bias = 5.7%, Figure 2A), whereas inter-observer reproducibility was slightly lower (Intraclass correlation coefficient (ICC) = 0.67, P = 0.003, mean bias = 0.7%, SD of bias = 6.3%, Figure 2B). The intra-observer reproducibility was equally good for RV global strain (ICC = 0.76, P < 0.001, mean bias = 0.76%, SD of bias = 2.5%, Figure 2C). Whereas inter-observer reproducibility was lower than intra-observer (ICC = 0.65, P = 0.004, mean bias = −0.36%, SD of bias = 2.3%, Figure 2D).

Figure 2:

(A) Intra-observer reproducibility for RVEF. (B) Inter-observer reproducibility for RVEF. (C) Intra-observer reproducibility of RV global strain. (D) Inter-observer reproducibility for RV global strain.


This is the first study to employ 3D speckle tracking on bariatric patients who were free of cardiac risk factors, before, and 6 months following surgery and with the use of two different techniques of bariatric surgery. We confirmed substantial reverse remodelling of biventricular volumes and mass. Alaud-din et al. in 19999 were the first group to demonstrate a significant decrease in LV dilatation (27.3–9.1%, P < 0.05) and in hypertrophy (45.5–0%, P < 0.05) in only 12 patients following bariatric surgery, after a 54.8 ± 1.9 kg weight loss. Rocha et al.14 reported thinning of the ventricular septum, LV posterior wall, and an increase in LV diastolic dimension at 6 months after weight loss surgery in 23 obese patients and normalization of the LV mass at 3 years post-weight loss. In the large SOS study from Sweden, Kardassis et al.15 identified obese patients who underwent bariatric surgery and they reported that weight loss is associated with lower cavity size, wall thickness, and mass. The surgical treatment of obesity also improved diastolic filling as demonstrated with the E/A ratio returning to normal. Diastolic impairment may be an early composite of diastolic heart failure, and the reversal to normal may indicate recovery of LV function post-bariatric surgery. Leichman et al.16 and Valezi and Machado17 described normalization of LV diastolic function at 3 and 9 months after bariatric surgery—similar to Karason et al.18 who—in the SOS study—found that surgical treatment of obesity led to significant recovery of LV EF and cardiac output post-surgery.18 A similar study19 examined obese patients with transthoracic echocardiography before and 1 year after Roux-en-Y gastric bypass, and there was a significant increase in LV EF32 (70.2 ± 7.2 vs. 72.9 ± 6.4%, P < 0.05).

This is the first study to employ 3D strain for biventricular assessment of obese patients. In our study, we found a significant improvement in diastolic function, normalization of RV myocardial performance index, and an increase in the systolic wave of the RV free wall. Obesity may be associated with early signs of RV dysfunction and likely contributes to increase of pulmonary pressures. The RV is more susceptible than the LV to volume changes.20 As a result, bariatric patients may present with signs of RV dilatation and diastolic dysfunction; in our study, we demonstrated that early findings of impending RV dysfunction normalized after surgical treatment of obesity: we observed a significant reduction of RV volumes, an increase of RV EF, and an RV mass regression. Owan et al.21 from the Utah obesity study demonstrated a significant increase in RV fractional area change with conventional echocardiography and improvement of RV function in 423 patients who underwent gastric bypass surgery.

Three-dimensional speckle tracking

In the group of bariatric patients who underwent 3D speckle tracking, there was a significant improvement in RV longitudinal strain post-surgery, confirming the recovery of RV function. While 3D strain has been widely used22,23 for the assessment of LV performance in ischaemic and non-ischaemic cardiomyopathies, it was rarely used in the assessment of the RV. When 3D speckle tracking echocardiography was initially applied to the left ventricle, it required myocardial motion analysis that aims to overcome the limitations of 2D speckle tracking.2325 It showed good reproducibility similar to cardiac magnetic resonance imaging for LV EF24 and proved predictive for recovery after myocardial infarction.23 It would thus appear that 3D speckle tracking is superior to 2D speckle tracking2326 and is promising for the global and regional assessment of the right ventricle.26 In our study, we employed both global longitudinal strain and RV free wall strain, in order to remove any possible bias from inter-ventricular dependence. The effect of bariatric surgery on the improvement in RV diastolic function translated into significant improvement in RV longitudinal strain, when assessed globally or isolated on the RV free wall.

Comparison of surgical techniques

Bariatric surgery has been the only effective treatment of obesity with substantial and sustained weight loss, and it has a perioperative risk similar to cholecystectomy.2729 Furthermore, laparoscopic bariatric surgery reduces the risk of complications when compared with open abdominal surgery. Our study compared the two techniques with regard to echocardiographic indices, and there was no difference on ventricular remodelling. The study cohort is too small to rule out endocrine causes contributing to myocardial remodelling, but there have been other studies which compared the two surgical techniques: Zhang et al.30 studied a total of 558 patients who underwent either sleeve gastrectomy (35.8%) or gastric bypass (64.2%) for morbid obesity. It was demonstrated that sleeve gastrectomy markedly improves most obesity-related comorbidities. Compared with gastric bypass, sleeve gastrectomy may be equally effective in reducing sleep apnoea, hyperlipidaemia, hypertension, diabetes, and musculoskeletal disease. Laparoscopic sleeve gastrectomy has been gaining acceptance, because it has shown good short- and mid-term results as a single procedure for morbid obesity.31 Even though gastric bypass is performed more commonly, sleeve gastrectomy is yielding promising with comparable results. Our study also showed that both surgical techniques result in similar LV and RV reverse remodelling.


Bariatric surgery has an important effect in reverse LV and RV remodelling, and it improves substantially RV longitudinal strain. Sleeve gastrectomy and gastric bypass are comparable surgical techniques with regard to ventricular remodelling.


Twelve patients had suboptimal acoustic windows and were, therefore, excluded from the analysis of 3D speckle tracking. Furthermore, our follow-up was shorter (at 6 months) compared with other studies (with a follow-up of up to 1 year post-bariatric surgery)—nevertheless, any possible effect on the left and right ventricles are demonstrated early. Most of the patients have endocrine assessment at their local hospital; therefore, we could not obtain their full endocrine profile.


View Abstract