Body composition changes at 12 months following different surgical weight loss interventions in adults with obesity: A systematic review and meta‐analysis of randomized control trials

Summary To determine relative lean mass and fat mass changes in adults with obesity following surgical weight loss interventions, a systematic review and meta‐analysis was conducted. The Cochrane Central Register of Controlled Trials, PubMed, Web of Science, EMBASE, and Scopus were screened for eligible studies. Inclusion criteria included randomized controlled trials (RCTs) performed in populations with obesity (body mass index ≥30 kg/m2) aged over 18 years, who underwent any type of bariatric surgery and reported body composition measures via dual‐energy X‐ray absorptiometry or bio‐electrical impedance analysis. Authors conducted full text screening and determined that there were six RCTs eligible for inclusion, with data extracted at 12 months post‐surgery. Meta‐analysis revealed that, relative to gastric banding, Roux‐en‐Y gastric bypass (RYGB) led to greater total body mass loss (mean difference [MD]: −9.33 kg [95% CI: −12.10, −6.56]) and greater fat mass loss (MD: −8.86 kg [95% CI: −11.80, −5.93], but similar lean mass loss (MD: −0.55 kg [95% CI: −3.82, 2.71]. RYGB also led to similar changes in total body mass, fat mass, and lean mass compared with sleeve gastrectomy. RYGB results in greater 12‐month weight and fat loss, but similar changes in lean mass, compared with gastric banding. Further RCTs comparing body composition changes following different bariatric surgery procedures are required.

There are several types of bariatric surgery, and the most commonly performed are Roux-en-Y gastric bypass (RYGB), sleeve gastrectomy (SG), and gastric banding. 5 RYGB involves creating a small pouch of stomach that is attached directly to the jejunum, which reroutes ingested food directly from the esophagus to the jejunum. 6 In SG, the normal intestinal pathway remains intact, but the size of the stomach is reduced by about 80% via surgical removal; and in gastric banding, a band is placed around the proximal stomach so that a small pouch is created proximal to the band, restricting the amount of food entering the region of the stomach distal to the band. 6 The mechanism by which weight loss occurs varies between procedures, but typically involves restriction of food intake, as well as a change in hormonal and signaling pathways that affect appetite and metabolism. 6 A disadvantage of successful weight loss interventions is that they often result in significant bone and muscle loss, 7,8 and this may lead to negative outcomes including increased fracture risk. 9,10 Maintenance of lean muscle mass is beneficial for metabolic health and decreases fracture risk. [11][12][13] Evidence also suggests greater lean mass is protective against T2DM, osteoporosis, and obesity. 14 In comparison, loss of lean mass is associated with declines in muscle strength 15 and function 16 and an increased risk for all-cause mortality. 17 In individuals undergoing weight loss, losses in lean mass also contributes towards subsequent weight regain. 18 Therefore, interventions that maximize fat mass losses while retaining lean mass are likely to result in optimal outcomes for people afflicted with obesity.
A recent meta-analysis 19 that investigated body composition changes following bariatric surgery in non-randomized studies found an average lean mass loss one-year post-RYGB of 10 kg (P < 0.001), but despite this loss, lean mass to body mass ratio increased by almost 12%. This meta-analysis also reported smaller absolute lean mass losses post gastric banding (7 kg) and SG (9.5 kg), with smaller increases in lean mass to body mass ratios (8.1% and 5.7%, respectively). 19 Therefore, RYGB might lead to greater fat mass losses while preserving lean mass relative to other surgical weight loss interventions. To date, no meta-analyses have directly compared fat and lean mass changes following different types of bariatric surgeries in randomized controlled trials (RCTs) to establish whether some lead to a more favorable body composition relative to others. Thus, the aim of this systematic review and meta-analysis was to compare changes in body composition (in particular, fat mass and lean mass changes) in adults afflicted with obesity undergoing different surgical weight loss interventions commonly utilized worldwide.

| METHODS
This study was conducted in accordance with the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement, 20 and a PRISMA checklist can be found in Table S1. The study protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (registration no: CRD 42020173485). A protocol was not prepared for this systematic review and meta-analysis.

| Search strategy and study selection
We systematically searched Embase (1980 to Present), PubMed, Web of Science, Scopus, and Cochrane Central Register of Controlled Trials for relevant articles from inception until September 2021. We also searched bibliographies of included articles and our personal reference libraries. Our complete search strategy can be found in Table S2. Title, abstract, and full-text screening was performed by two independent reviewers (AS and PJ) using Covidence software (Veritas Health Innovation, Melbourne, Australia).
We included parallel-group RCTs with populations aged 18 years and over afflicted with obesity (body mass index [BMI] > 30 kg/m 2 ) and underwent any type of bariatric surgery procedure and reported body composition measures in the form of fat or lean mass via dual-energy X-ray absorptiometry (DXA) or bio-electrical impedance analysis (BIA). Studies were excluded if they were not published in English.

| Data extraction and quality assessment
Data were independently extracted by two reviewers (AS and PJ).
We extracted health status, age, sample size, age, post-surgery fat mass, fat-free mass, body mass, and lean mass data from eligible studies at baseline and at 12 months post-surgery. For data extraction and analysis, we combined adjustable gastric banding with nonadjustable gastric banding into one group, as these two procedures are comparable in terms of weight loss mechanism and overall efficacy. 21 Fat-free mass and lean mass are composed of different body composition compartments, with the key difference being the addition of bone mass in fat-free mass measurements. However, despite fat-free mass and lean mass being different absolute values, the pre-and post-surgery change in these body composition compartments is comparable. Thus, change in fat-free mass was considered the same as change in lean mass for the analysis.
We performed an additional sensitivity analysis to evaluate whether combining fat-free mass and lean mass would yield similar results.
All calculations were performed using Review Manager (RevMan). 22 Quality assessment (risk of bias) of included studies was performed by two reviewers (AS and PJ) using the Cochrane Risk of Bias 2 tool, and discrepancies were adjudicated by a third reviewer (JM). 23 Studies were classified as "low," "some concerns," or "high risk of bias" based on the following five bias domains: (1) randomization process, (2) deviation from intended intervention, (3) missing outcome data, (4) measurement of the outcome, and (5) selection of the reported results.
A Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) assessment was performed to evaluate the quality of the outcomes used in the meta-analysis. 24 The use of five different domains (risk of bias, inconsistency, indirectness, imprecision, and publication bias) is assessed to determine the degree of confidence in the estimate of effect derived from the meta-analysis.
The quality of evidence is defined as high, moderate, low, or very low.

| Statistical analysis
Mean differences in body composition changes between the two intervention groups were calculated as the mean change in control group minus the mean change in the intervention group, or as mean follow-up values in the control group minus mean follow-up values in the intervention group. For studies missing standard deviations of mean changes, [25][26][27] we imputed a correlation co-efficient of 0.7 and performed sensitivity analyses with co-efficient of 0.5 and 0.9. Body composition data were pooled using random-effects meta-analyses, based on the DerSimonian and Laird model. 28 Tau 2 , Q-statistic, and I 2 values were used to measure heterogeneity between studies. Meta-analyses were performed using RevMan. 22 A P value < 0.05 was considered statistically significant.

| Literature search and study characteristics
Database searches identified 2887 unique citations. The full text of 310 articles were screened; data extraction was sought for 17 studies, but 11 were further excluded at this stage. [29][30][31][32][33][34][35][36][37][38][39] Overall, six were included in the meta-analysis ( Figure 1). Descriptive characteristics of included studies are presented in Table 1. Four studies were conducted in individuals with T2DM, 7,25,26,40 and the other two studies included adults with obesity without specified health conditions. 27,41 Age varied across all studies (range: 18-65 years) and all had combined cohorts of men and women. All six studies included F I G U R E 1 PRISMA diagram of information flow in this systematic review and metaanalysis T A B L E 1 Descriptive characteristics of randomized controlled trials comparing changes in body composition in adults with obesity undergoing weight loss while undertaking bariatric surgery RYGB. 7,[25][26][27]40,41 In two studies, RYGB was compared with gastric banding (either adjustable or vertical), 7,41 and in four studies, RYGB was compared with SG. [25][26][27]40 All studies provided data at exactly 12 months post-surgery, however one study 7 continued for a total of 36 months. Five studies included sufficient data to compare total body, lean, and fat mass changes. 7,[25][26][27]40 Of these five studies, two reported fat-free mass instead of lean mass changes. 26,40 For one study, 41 the total body mass changes were extracted from a graph using OriginLab. 42 None of the eligible studies included protein supplementation. Body composition at baseline was similar in different bariatric surgery groups. In those who underwent RYBG, mean total body, fat, and lean mass at baseline were 114 kg, 49 kg, and 60 kg, respectively. Mean total body, fat, and lean mass at baseline were 112 kg, 50 kg, and 55 kg, respectively, in those who underwent gastric banding. In those who underwent SG, mean total body, fat, and lean mass at baseline were 115 kg, 47 kg, and 62 kg, respectively.

| Risk of bias assessment
Study quality assessments performed using the Cochrane Risk of Bias 2 tool are presented in Figure 2. Most studies had low selection bias.
One study 26 was classified as having "some concerns" in domain 2 due to lack of concealment for treatment allocations. Two studies 40,41 were identified as having unclear risk of bias due to missing F I G U R E 2 Study quality assessment using Cochrane Risk of Bias 2 tool outcome data (domain 3). All other domains (1, 4, and 5) were considered low risk in all studies.

| GRADE assessment
The GRADE assessment findings are summarized in Tables 2 and 3.
For the RYGB versus gastric banding outcomes (Table 2), the quality of evidence was evaluated as either moderate or high. We downgraded the quality of evidence for lean mass outcomes due to substantial heterogeneity. Publication bias was undetectable for all outcomes due to the limited number of studies included. Otherwise, the outcomes were of high quality of evidence due to no serious risk of bias, no serious indirectness or imprecision for all outcomes, and a large effect size for the body mass and lean mass change outcomes.
For the RYGB versus SG outcomes (  Of the studies, one 41 was identified as having some concerns of risk of bias. Thus, there was only one study 7 remaining for the sensitivity analysis. Alone, this study demonstrated that both fat mass and lean mass losses were significant at 12 months (Figure 3).

| RYGB versus SG
Four studies were eligible for inclusion in the RYGB versus SG metaanalysis, with a total of 170 participants (Figure 4). 25 studies 26,40 which had some risk of bias were also the same two studies which reported fat-free mass instead of lean mass results.
We performed additional sensitivity analyses where we substituted our imputed correlation coefficient of 0.7 with values of 0.5 and 0.9 and neither coefficient changed our overall findings.

| RYGB versus gastric banding
Unfortunately, due to one of the two included studies having a mod- Changes in food preference might also explain why fat mass losses were greater in those who underwent RYGB relative to gastric banding. RYGB is associated with increased consumption of "healthy" foods (such as fruits, fish, and less processed foods), 47,48 with energy intake from fat consumption decreasing from 37.0% immediately before surgery to 25.3% 2 years post-surgery (P < 0.0001). 48 In com- gastric banding continued to be 36% total fats. 49 In both studies, patients were given similar dietary advice post-surgery, but no specific diet was prescribed. Additionally, another study used the validated "Swedish Obese Subjects" questionnaire to calculate percentage of total energy intake for patients 1-year post-surgery and found that those who underwent gastric banding consumed higher amounts of sweets and fats compared with those who underwent RYGB. 41 Conversely, energy from fruits and vegetables following RYGB was over 10% of daily total intake as compared with around 2.5% for those who underwent gastric banding (P < 0.0001). A recent systematic review 50 analyzed the difference in food preferences post-surgery. It determined that poorer diets post-gastric banding compared with RYGB may be due to the purely restrictive nature of the procedure and minimal influence on metabolism of food, with gastric banding patients more likely to feel hungry and adopt a high-calorie diet to compensate. 50 Further research is required to understand changes in eating behavior and its influence on body composition changes in those who undergo RYGB compared with gastric banding.

| RYGB versus SG
Our finding that RYGB resulted in similar body composition changes compared with SG is supported by multiple cohort studies comparing adults with obesity undergoing RYGB or SG. [51][52][53] One study followed up participants at 6 months 51 and another after 1 year, 52 and the F I G U R E 3 Mean differences in body composition outcomes after Roux-en-Y gastric bypass and gastric banding F I G U R E 4 Mean differences in body composition outcomes after Roux-en-Y gastric bypass and sleeve gastrectomy most recent was a long-term review at 6.7 years post-bariatric procedure. 53 At each of these time-points, there were no significant differences in body composition outcomes. [51][52][53] Additionally, a recent meta-analysis 54 found body mass changes post-surgery were sustained over 5 years for both RYGB and SG. Although our limited data meant that we could not explore long-term body composition changes in our meta-analysis, our findings suggest that the lean and fat mass changes post RYGB and SG are comparable at 12 months.
There is evidence that different initial body weight and body composition influences the partitioning of weight loss as either fat or lean mass, and effects of initial weight on weight loss and body composition changes were not explored in this study. However, given that only RCTs were included in the analysis and initial total body, fat and lean mass were similar between groups reduces the influence on body weight and body composition outcomes in our study. The Forbes theory equation is used to predict changes in fat-free mass during weight change as a function of the initial body fat mass. 55 However, this equation may only be valid for measuring moderate amounts of weight loss; for more extreme weight change as seen post-bariatric surgery, this equation underestimates the proportion of fat-free mass loss. 55 Hall's theory may be a better predictor of relative loss of fatfree mass versus fat mass following bariatric surgery, 56 but there is no universally accepted equation for predicting change in fat-free versus fat mass for people undergoing dramatic weight loss.
As seen in amended Table 1, baseline body composition was very similar between groups that were compared. We have also presented the mean total body, fat and lean mass at baseline in each surgical group in our results text, which were also very similar between different surgical groups. interventions. We were therefore unable to determine whether observed differences are wholly attributable to procedure type or in part related to postoperative management. We were unable to assess publication bias due to the limited studies available for the analysis.

| Limitations
Due to the limited number of RCT available assessing body composition changes after different bariatric procedures, we also were unable to compare all types of surgery in current practice (e.g. anastomosis gastric bypass and biliopancreatic diversion with duodenal switch).
Therefore, our conclusions are limited to the procedures presented in our analyses rather than an overview of all procedures in current practice.
Regarding measurement tools used to assess body composition, we were unable to perform an additional analyses to compare effects of use of DXA versus BIA to quantify body composition due to the low number of included studies. BIA uses electrical currents to determine conductivity of different tissue and equations to calculate the proportion of body fat based on the assumption of a constantly hydrated body. 57 In contrast, DXA uses X-ray attenuation by different tissues to estimate body composition and its accuracy can be influence by body size. 58 Therefore, in populations with obesity, differences in total body water levels and body composition (e.g., greater tissue depth) may decrease the accuracy of body fat and fat-free mass estimates. 57 There were two types of DXA machines (GE/Lunar and Hologic) that were utilized by our included studies, and while DXA is not considered a gold-standard measurement for body composition, it is widely recommend due to its high precision, low cost, and wide availability. 59 Each type has different degrees of precision due to pro- This meta-analysis demonstrated that RYGB leads to greater total body and fat mass losses, but similar lean mass losses, compared with gastric banding at 12 months. Body composition changes following RYGB relative to SG were similar. The limited studies available for this analysis highlight the necessity for further RCTs comparing body composition changes following different bariatric surgery procedures. In particular, additional studies of durations beyond 12 months are