Obesity and bariatric surgery have been associated with changes in ventricular function and structure. The aim of the present study was to assess the long-term changes in left ventricular (LV) and right ventricular (RV) function and structure in patients with morbid obesity—body mass index ≥40 kg/m 2 or ≥35 kg/m 2 with co-morbidities—who had lost weight after bariatric surgery compared to nonsurgical controls. We reviewed 57 patients with morbid obesity who had undergone gastric bypass surgery and who had undergone echocardiography before and after surgery. A reference group (n = 57) was frequency matched for body mass index (±2 kg/m 2 ), gender, age (±2 years), and follow-up duration (±6 months). After a mean follow-up of 3.6 years, the LV mass and LV mass indexed by height had decreased in the patients who had undergone bariatric surgery and had lost weight. In contrast, these measurements had increased in the patients who had not undergone bariatric surgery. The difference between these 2 groups remained significant after adjusting for potential confounders. At follow-up, neither the patients nor controls showed a significant change in ejection fraction, LV myocardial performance index, or RV myocardial performance index. In the study population as a whole, multivariate analysis showed a positive correlation between the change in body weight and ventricular septum thickness (R = 0.33), posterior wall thickness (R = 0.31), LV mass (R = 0.38), RV end-diastolic area (R = 0.22), and estimated RV systolic pressure (R = 0.39), all with p values <0.05. In conclusion, body weight changes in patients with morbid obesity were associated with changes in LV structure independent of improvement in obesity-related co-morbidities, including obstructive sleep apnea. Weight loss improved the RV end-diastolic area and might prevent progression to RV dysfunction.
We conducted a longitudinal study with retrospective assessment of clinical and anthropometric variables and echocardiographic off-line measurements to test the hypothesis that major weight loss is associated with the long-term decrease in left ventricular (LV) mass and improvement in LV and right ventricular (RV) function, independent of concomitant improvement of systemic blood pressure and other obesity-related co-morbidities, including obstructive sleep apnea. We assessed long-term changes in LV and RV structure and function in patients with morbid obesity—body mass index (BMI) ≥40 kg/m 2 or ≥35 kg/m 2 with co-morbidities—after surgically induced weight loss from gastric bypass surgery. We used obese patients matched for clinical characteristics who had not undergone any bariatric procedure as the comparison group.
Methods
We included all the patients who had undergone Roux-en- Y gastric bypass for morbid obesity from 1995 to 2005 at Mayo Clinic (Rochester, Minnesota) and who had undergone ≥2 transthoracic echocardiographic examinations, one within 2 years before and one at ≥6 months after bariatric surgery. For the patients with multiple echocardiograms before surgery, we selected the one performed closest in time to the date of surgery, and for the patients with multiple echocardiograms after bariatric surgery, we selected the study furthest from the surgery date. We selected a control group by frequency matching for gender, age (±2 years), BMI (±2 kg/m 2 ) at the first echocardiogram, and follow-up duration (±6 months). The controls were selected from patients with morbid obesity who had undergone ≥2 echocardiographic evaluations at Mayo Clinic (Rochester, Minnesota) from 1995 to 2005 but who had not undergone bariatric surgery. We excluded patients with a heart transplant, a medical history of cardiomyopathy secondary to a specific disease, pericardial disease, and severe valve disease, as well as those with technically poor echocardiograms. The institutional review board approved the present study.
In all patients, digital off-line analyses of the myocardial performance index (MPI) and fractional area change of the left ventricle and right ventricle and pulmonary vascular resistance were obtained from the original videotaped signals. The off-line analysis was performed by a single observer (CAG) who was unaware of the clinical history of the patients. We measured the Doppler intervals for LV MPI from the velocity intervals of mitral inflow and LV outflow. The interval from the cessation to onset of mitral inflow (time “a”) was equal to the sum of isovolumetric contraction time, ejection time, and isovolumetric relaxation time. The time for LV ejection (time “b”) was the duration of the LV outflow velocity profile. The MPI was calculated as (a − b)/b. Similar intervals for the RV were measured from the recordings of tricuspid inflow and RV outflow in the parasternal short-axis view. The RV MPI was calculated using the same formula. The fractional area change of the RV and LV was measured from the apical 4-chamber view. End-diastole was identified by the onset of the R wave, and end-systole was identified as the smallest LV and RV cavity size immediately before the opening of the mitral and tricuspid valves. The LV and RV areas were calculated from the average of 3 measurements. The fractional area change was calculated using the following formula: (end-diastolic area − end-systolic area)/end-diastolic area. In the patients in whom Doppler assessment of pulmonary flow was obtained, we measured: (1) pre-ejection period, the interval between the onset of tricuspid regurgitation and the onset of pulmonary systolic flow; (2) acceleration time, the interval between the onset of ejection to peak flow velocity; and (3) ejection time. We calculated the pulmonary vascular resistance using the following equation: pulmonary vascular resistance = −0.156 + 1.154 × [(pre-ejection period/acceleration time)/total systolic time]. The interventricular septum, posterior wall thickness, LV end-systolic diameter, and end-diastolic diameters were measured from the parasternal long axis. The ejection fraction (EF) was calculated using the modified Quinones formula and LV mass using the Devereux formula. Additionally, we determined the LV mass indexed by height. Indexing the LV mass by body surface area was considered inappropriate for the present study, because the body surface area incorporates the weight into the calculation. Therefore, hypothetically, the weight loss associated with a decreased LV mass might not show any change in the LV mass when indexed by body surface area because both the numerator and denominator would change in the same direction. The RV systolic pressure was estimated by adding the right atrial pressure to the transtricuspid gradient calculated using the modified Bernoulli equation (transtricuspid pressure gradient = 4 × tricuspid regurgitation velocity 2 ).
The clinical and demographic data were obtained from the medical records of the clinical encounter nearest to the date of the baseline echocardiogram. We abstracted information regarding the presence of hypertension, diabetes mellitus, dyslipidemia, pulmonary disease, atrial fibrillation, obstructive sleep apnea, idiopathic cardiomyopathy, and coronary artery disease. The definition of coronary artery disease required a diagnostic angiogram with findings of coronary stenosis of ≥50% luminal diameter or a documented history of myocardial infarction, angina, coronary artery bypass grafting, or percutaneous coronary revascularization. Weight, height, heart rate, and blood pressure were obtained from the echocardiogram reports or the clinical visit findings closest in time to the echocardiogram.
Because no consensus exists about the best measure of weight change, we decided to use 2 different approaches. First, the rate of weight loss was calculated by dividing the change in body weight between the surgery date and the last echocardiogram by the interval between the surgery and the last echocardiogram in months. For the control group, this measurement was calculated by dividing the change in body weight between the first and last echocardiograms by the interval between the 2 studies in months. This parameter took into account the duration of follow-up. Second, we calculated the percentage of weight loss by dividing the change in body weight between the 2 echocardiograms by the weight at the first electrocardiogram. This reflected the percentage of weight lost in relation to the patient’s total weight.
We divided the statistical analyses into 2 sets. The first set compared the measurements in patients who had undergone gastric bypass surgery with the measurements in the control group. For the second set of analyses, we pooled the data from the entire study population (patients with and without bariatric surgery) to analyze a wide spectrum of body weight changes over time against the changes in echocardiographic measurements. All values for quantitative measures are expressed as the mean ± SD.
At baseline, we assessed the differences between the groups with and without bariatric surgery using an unpaired t test for continuous variables and a chi-square test for categorical variables. Differences between baseline and follow-up within the groups were assessed using a paired t test. We compared the changes between cases and controls using an unpaired t test, instead of a paired t test, because we used a frequency match strategy, not an individual match technique. Multivariate analysis was performed to assess bariatric surgery as a predictor of change in LV and RV structure and function. We also created different models by adjusting for potential confounders of this association. We constructed a base model for LV structure and function that included patient age at the first study, gender, and a history of systemic hypertension, diabetes mellitus, and coronary artery disease. Next, we added the change in blood pressure and obstructive sleep apnea. The base model for RV structure and function included patient age, gender, chronic obstructive pulmonary disease, and obstructive sleep apnea. Next, we added diabetes mellitus, coronary artery disease, and systemic hypertension.
We calculated Pearson’s correlation coefficients to test the univariate relations between the change in body weight (including the 2 measurements of weight change) and the change in LV and RV function and structure. The entire group of participants had a wide range of weight changes, including patients in whom no change in weight occurred. We performed multiple linear regression analyses using changes in the measurements that were significant in the univariate analyses as the dependent variables and changes in body weight expressed as the rate of weight loss as an independent variable, as well as other covariates that might be implicated in the change in LV and RV function. Furthermore, we stratified the pooled data according to the presence or absence of systemic hypertension and obstructive sleep apnea, and we tested the correlation between the change in body weight and the change in echocardiographic measurements. We compared the correlation coefficients between the patients with and without each of these diseases.
A subgroup analysis for systolic LV function included patients with a baseline EF of ≤50%. We compared the patients who had and had not undergone gastric bypass surgery and also determined the correlation coefficients between the change in body weight and change in EF, adjusting for obesity-related co-morbidities.
To assess the reliability of the measurements, 2 independent observers measured a randomized sample of 10% of all echocardiograms. The interobserver concordance was evaluated using Pearson’s correlation coefficients and a t test of the differences between the 2 measurements.
Results
A total of 64 patients who had undergone bariatric surgery met the inclusion criteria. Of these 64 patients, 7 were excluded because of a history of heart transplant (2 patients), severe valvular heart disease (4 patients), and cardiac amyloidosis (1 patient). Thus, we included 57 patients and 57 frequency-matched controls. Of the 57 patients who underwent gastric bypass, 22 were men (39%), their age at the first echocardiogram was 52 ± 9 years, and their BMI was 49 ± 9 kg/m 2 . The mean age, gender, follow-up duration, and BMI were similar in the surgical and control subjects. The baseline clinical characteristics of the patients and controls are listed in Table 1 .
| Variable | Patients (n = 57) | Controls (n = 57) | p Value |
|---|---|---|---|
| Men | 22 (39) | 22 (39) | 0.99 |
| Age (years) | 51 ± 9 | 52 ± 10 | 0.57 |
| Follow-up (months) | 45 ± 25 | 41 ± 21 | 0.33 |
| Body mass index (kg/m 2 ) | 49 ± 9 | 48 ± 8 | 0.33 |
| Height (cm) | 170 ± 12 | 170 ± 11 | 0.90 |
| Weight (kg) | 143 ± 33.2 | 141 ± 34 | 0.60 |
| Systemic hypertension ⁎ | 50 (89%) | 40 (70%) | 0.02 † |
| Diabetes mellitus | 34 (61%) | 35 (61%) | 0.97 |
| Alcohol use | 8 (14%) | 4 (7%) | 0.23 |
| Current smoker | 1 (2%) | 9 (16%) | 0.005 † |
| Previous smoker | 21 (38%) | 13 (23%) | 0.05 |
| Dyslipidemia ‡ | 38 (68%) | 34 (60%) | 0.53 |
| Obstructive sleep apnea | 42 (82%) | 39 (68%) | 0.08 |
| Chronic obstructive pulmonary disease | 7 (13%) | 9 (16%) | 0.48 |
| Heart failure | 18 (32%) | 22 (39%) | 0.80 |
| Coronary artery disease § | 19 (34%) | 15 (26%) | 0.46 |
| Pulmonary embolism | 2 (4%) | 5 (9%) | 0.26 |
| Pulmonary hypertension | 8 (14%) | 13 (23%) | 0.39 |
| Idiopathic cardiomyopathy | 11 (20%) | 14 (25%) | 0.56 |
| Permanent pacemaker | 5 (9%) | 4 (7%) | 0.96 |
| Atrial fibrillation | 10 (18%) | 19 (33%) | 0.07 |
⁎ Systemic hypertension required documented diagnosis, antihypertensive treatment, or recorded blood pressure of ≥125/85 mm Hg.
‡ Dyslipidemia required documented diagnosis or mean total cholesterol level of ≥240 mg/dl, low-density lipoprotein of ≥160 mg/dl, or use of lipid-lowering therapy.
§ Coronary artery disease required diagnostic angiogram with findings of coronary stenosis of ≥50% luminal diameter or documented history of myocardial infarction, angina, coronary artery bypass grafting, or percutaneous coronary revascularization.
The baseline and follow-up anthropometric, hemodynamic, and echocardiographic features are listed in Table 2 . At baseline, the patients and controls were similar. At follow-up, those who had undergone bariatric surgery had had a significant decrease in body weight (42 ± 25 kg), BMI (15 ± 8 kg/m 2 ), body surface area (0.42 ± 0.21 m 2 ), and heart rate (11 ± 22 beats/min). We also found a decrease in ventricular septum thickness (p <0.0001), posterior wall thickness (p <0.0001), LV mass (p = 0.0006), and indexed LV mass (p = 0.0006) in patients who had undergone gastric bypass surgery. In contrast, these measurements increased in the patients who had not undergone surgery; the difference between these 2 groups was statistically significant. In those without gastric bypass surgery, we found an increased RV end-diastolic area (p = 0.0008) and RV end-systolic area (p = 0.06). The RV systolic pressure decreased in the group of patients who had undergone bariatric surgery but increased in the control group (p = 0.02). At follow-up, neither the patients nor controls showed a significant change in EF, LV MPI, or RV MPI. Also, when we compared the changes between the 2 groups, no significant difference was found in EF (p = 0.56), LV MPI (p = 0.18), or RV MPI (p = 0.42). Multivariate analysis showed that the differences between the surgical patients and controls persisted for ventricular septum thickness (p = 0.002), posterior wall thickness (p <0.0001), LV mass (p = 0.0007), and LV mass index (p = 0.0006). RV systolic pressure had a borderline significance on multivariate analysis (p = 0.07).
| Variable | Patients | Controls | ||||
|---|---|---|---|---|---|---|
| n | Baseline | Follow-Up | n | Baseline | Follow-Up | |
| Weight (kg) | 57 | 143 ± 33 | 100 ± 26 ⁎ | 57 | 141 ± 33 | 138 ± 34 † |
| Body mass index (kg/m 2 ) | 57 | 49 ± 9 | 35 ± 8 ⁎ | 57 | 48.3 ± 8 | 47.1 ± 8.8 † |
| Body surface area (m 2 ) | 57 | 2.51 ± 0.36 | 2.24 ± 0.34 ⁎ | 57 | 2.47 ± 0.34 | 2.40 ± 0.31 † |
| Heart rate (beats/minute) | 57 | 78 ± 16 | 67 ± 17 ‡ | 57 | 77.8 ± 17.7 | 75.7 ± 15.8 § |
| Systolic blood pressure (mm Hg) | 57 | 135 ± 21 | 129 ± 17 | 57 | 134.1 ± 18.9 | 128.8 ± 20 |
| Diastolic blood pressure (mm Hg) | 57 | 77 ± 15 | 72 ± 11.0 ¶ | 57 | 77.7 ± 13.9 | 76.4 ± 13 |
| Two-dimensional echocardiography | ||||||
| Ventricular septum thickness (mm) | 57 | 12.0 ± 2.2 | 10.6 ± 1.9 ⁎ | 57 | 11.9 ± 2.6 | 12 ± 2.8 § |
| Posterior wall thickness (mm) | 57 | 11.7 ± 1.6 | 10.5 ± 1.8 ⁎ | 57 | 11.1 ± 1.9 | 11.5 ± 2 ∥ |
| Left ventricular end-diastolic dimension (mm) | 57 | 50.3 ± 7.6 | 51.4 ± 6.5 | 57 | 53.3 ± 7.9 | 53.4 ± 8.6 |
| Left ventricular end-systolic dimension (mm) | 57 | 32.7 ± 8.2 | 33.9 ± 7.1 | 57 | 35.1 ± 9.1 | 36.5 ± 9.7 |
| Left ventricular myocardial performance index | 44 | 0.37 ± 0.20 | 0.30 ± 0.21 | 53 | 0.41 ± 0.21 | 0.42 ± 0.25 |
| Left ventricular ejection fraction (%) | 57 | 58 ± 12 | 56.1 ± 11 | 57 | 56.3 ± 13.2 | 53.3 ± 13.2 |
| Left ventricular ejection fraction ≤50 (%) | 9 | 44.8 ± 7 | 59.5 ± 10.1 | 10 | 44.9 ± 7.9 | 58.6 ± 14.1 |
| Left ventricular fractional area change (%) | 50 | 39.4 ± 8.3 | 40.3 ± 10 | 56 | 39.8 ± 7.9 | 40.6 ± 8.1 |
| Left ventricular mass (g) | 57 | 239.9 ± 87.5 | 208.2 ± 63.9 ‡ | 57 | 251.1 ± 92.6 | 258.8 ± 95.3 § |
| Left ventricular mass index (g/m 2 ) | 57 | 1.41 ± 0.47 | 1.22 ± 0.34 ‡ | 57 | 1.46 ± 0.49 | 1.51 ± 0.53 § |
| Right ventricular end-diastolic area (m 2 ) | 47 | 13.7 ± 5.2 | 14.6 ± 4.8 | 55 | 18.1 ± 4.5 | 20.2 ± 4.9 ‡ |
| Right ventricular end-systolic area (m 2 ) | 47 | 7.4 ± 3.8 | 7.8 ± 3.7 | 55 | 9.7 ± 3.2 | 10.5 ± 3.9 # |
| Right ventricular fractional area change (%) | 47 | 47.4 ± 9.5 | 47.6 ± 11.6 | 55 | 46.4 ± 7.9 | 48.3 ± 9.1 |
| Right ventricular myocardial performance index | 21 | 0.17 ± 0.11 | 0.19 ± 0.11 | 38 | 0.22 ± 0.14 | 0.27 ± 0.17 |
| Pulmonary vascular resistance | 25 | 1.96 ± 0.93 | 1.67 ± 0.87 | 37 | 2.08 ± 0.96 | 1.96 ± 0.93 |
| Right ventricular systolic pressure | 25 | 40.4 ± 13 | 37 ± 9.6 | 38 | 34.7 ± 12.6 | 38.5 ± 14.9 § |
⁎ p <0.0001 (comparing baseline with follow-up);
† p <0.0001 (comparing patients and controls);
‡ p <0.001 (comparing baseline with follow-up);
§ p <0.05 (comparing patients and controls);
¶ p <0.05 (comparing baseline with follow-up);
∥ p <0.001 (comparing patients and controls);
The analysis of the pooled data showed a positive and significant correlation between the rate of weight loss and ventricular septum thickness (R = 0.33), posterior wall thickness (R = 0.31), LV mass (R = 0.38), LV mass index (R = 0.39), RV end-diastolic area (R = 0.22), and RV systolic pressure (R = 0.39; Figures 1 and 2 ). After adjustment for several variables, these correlations remained significant in all models ( Table 3 ). When the percentage of weight loss was used as the weight change variable, the results were essentially the same, except for the RV end-diastolic area (R = 0.04) and RV systolic pressure (R = 0.22).
| Model 1 | Model 2 | Model 3 | Model 4 | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| E | SE | p Value | E | SE | p Value | E | SE | p Value | E | SE | p Value | |
| Models for left ventricle | ||||||||||||
| Ventricular septum thickness | 0.4167 | 0.1144 | 0.0004 | 0.4082 | 0.1171 | 0.0007 | 0.4076 | 0.1178 | 0.0008 | 0.3601 | 0.1151 | 0.0023 |
| Posterior wall thickness | 0.3198 | 0.0973 | 0.0014 | 0.2808 | 0.0976 | 0.0048 | 0.2823 | 0.0987 | 0.0051 | 0.2719 | 0.0991 | 0.0071 |
| LV mass | 15.2608 | 3.6125 | <0.0001 | 15.7879 | 3.6231 | <0.0001 | 15.8885 | 3.6577 | <0.0001 | 14.9461 | 3.6454 | <0.0001 |
| LV mass index (g/m 2 ) | 0.0910 | 0.0211 | <0.0001 | 0.0941 | 0.0212 | <0.0001 | 0.0949 | 0.0214 | <0.0001 | 0.0892 | 0.0213 | <0.0001 |
| Models for right ventricle | ||||||||||||
| RV end-diastolic area | 0.6941 | 0.3118 | 0.0283 | 0.7620 | 0.3190 | 0.0189 | 0.7706 | 0.3237 | 0.0193 | 0.7568 | 0.3325 | 0.0252 |
| RV systolic pressure | 3.0411 | 0.9149 | 0.0015 | 2.8462 | 0.9292 | 0.0033 | 2.9093 | 0.9334 | 0.0029 | 2.5466 | 0.9925 | 0.0131 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
