META-ANALYSIS

Laparoscopic Sleeve Gastrectomy Versus Laparoscopic Roux-en-Y

Gastric Bypass
A Systematic Review and Meta-analysis of Weight Loss, Comorbidities, and
Biochemical Outcomes From Randomized Controlled Trials
Yung Lee, BHSc,  y Aristithes G. Doumouras, MD, MPH,  y James Yu, BHSc,  y Ishan Aditya, BHSc,z
Downloaded from https://journals.lww.com/annalsofsurgery by 01UGrXh3ipqzR4DKqW7bOJtSxsRVOheLV9OzOeHq2POX2t1GrKUD6m1aBQdlm0lX7bZxxCFBpuxAM3exuxdXARVaPXROhrEIELxLuiE5dAoxEXL1EbBXOmtN5qOe2K0czeIvdGFOHR5ia7TbUFTzsQ== on 03/16/2021

Scott Gmora, MD,  y Mehran Anvari, MBBS, PhD,  y and Dennis Hong, MD, MSc  yY

(LRYGB) has traditionally been considered the gold standard in

Objective: The aim of this study was to compare weight loss, obesity-related
bariatric surgery because it provides substantial, long-term weight
comorbidities, and biochemical outcomes of LSG versus LRYGB through a
loss and remission of comorbidities.3 However, in recent years,
meta-analysis of randomized controlled trials (RCTs).
laparoscopic sleeve gastrectomy (LSG) has experienced a rapid surge
Summary of Background Data: LSG and LRYGB are the 2 most commonly
in popularity due to its effectiveness in achieving weight loss and
performed bariatric surgeries for the treatment of obesity. The comparative
remission of comorbidities, and its less technically intensive procedure
outcomes of the 2 surgeries is a topic of ongoing debate and medium-term
and the belief that it causes fewer complications.4 In 2016, LSG was the
outcomes remain uncertain.
most commonly performed bariatric surgery in the US and worldwide,
Methods: A search for RCTs comparing LRYGB versus LSG was conducted.
making up 53.6% of operations compared to the 30.1% of RYGB.1–6
Pooled outcomes between 2 procedures were compared using pairwise
Despite the recent change in trend from LRYGB to LSG,
random-effects meta-analysis at 1, 3, and 5-year follow-up time points.
evidence comparing LRYGB to LSG has lagged behind, especially in
Grading of recommendations, assessment, development, and evaluation
medium- and long-term outcomes. Although short-term clinical and
was used to assess certainty of evidence.
metabolic data were promising for LSG, only a few randomized
Results: Thirty-three studies involving 2475 patients were included. LRYGB
controlled trials (RCTs) directly compared LSG to LRYGB and those
resulted in greater loss of body mass index compared to LSG at 1 year [mean
that did typically had small sample sizes, or short-term follow-ups,
difference 1.25 kg/m2, 95% confidence interval (CI) 2.01 to 0.49, P ¼
sparking debate on the appropriateness of LSG in replacing
0.001; moderate certainty of evidence] which persisted at 3 years, but there was
LRYGB.7 In response to the existing uncertainty on weight loss
insufficient evidence at 5 years. Resolution of dyslipidemia was higher for
achieved by LSG compared to LRYGB, the American Society for
LRYGB than LSG at 1 year (risk ratio 0.58, 95% CI 0.46–0.73, P < 0.001;
Metabolic and Bariatric Surgery (ASMBS) produced an updated
moderate certainty of evidence) and 5 years (risk ratio 0.68, 95%CI 0.46–0.99,
position statement in 2017.3 The statement concluded that while
P ¼ 0.04; low certainty of evidence). There was no difference between LRYGB
there was no reliable conclusion regarding which bariatric operation
and LSG for remission of type 2 diabetes, hypertension, and hemoglobin A1c,
produces the greatest weight loss early after surgery, evidence
fasting insulin, homeostatic model assessment of insulin resistance, high-
appeared to support LRYGB as producing greater percent excess
density lipoprotein, and the rate of 30-day major and minor complications.
weight loss (%EWL) compared to LSG after the first year.3
Conclusions: There are insufficient data from RCTs to draw any conclusions
Beyond the uncertainty between LSG and LRYGB for weight
regarding the long-term comparative effectiveness beyond 3 years between
loss, meta-analyses comparing LSG to LRYGB for improvements in
LRYGB and LSG.
co-morbidities have also differed in results, reporting both remission
Keywords: bariatric surgery, gastric bypass, meta-analysis, sleeve and no change in T2DM and hypertension.8,9 To address the vigorous
gastrectomy, systematic review and ongoing debate comparing LSG and LRYGB, 2 major RCTs,
The Swiss Multicenter Bypass or Sleeve Study (SM-BOSS) and
(Ann Surg 2021;273:66–74)
Finnish Sleeve vs. Bypass (SLEEVEPASS) recently provided new
evidence comparing 5-years outcomes after LSG and LRYGB.10,11
O besity has become a growing health issue worldwide and is
associated with comorbidities including type 2 diabetes mellitus
(T2DM), cardiovascular disease, and overall rates of cancer.1 Bariatric
Nonetheless, while SM-BOSS found no significant differences in
percent body mass index (BMI) loss, SLEEVEPASS did find greater
%EWL in the LRYGB group than the LSG group, although this
surgery is the most effective and long-lasting treatment for patients difference was not clinically significant. The 2 trials also came to
with obesity, capable of resolving comorbidities, decreasing mortality, different conclusions on hypertension remission, and metabolic
and improving quality of life.2 Laparoscopic Roux-en-Y gastric bypass differences in LSG compared to LRYGB.
Given the considerable number of high-quality studies with
From the Division of General Surgery, Department of Surgery, McMaster medium-term follow-ups that have accumulated in recent years, and
University, Hamilton, Ontario, Canada; yCentre for Minimal Access Surgery the differing conclusions proposed by RCTs, this systematic review
(CMAS), St. Joseph’s Healthcare, McMaster University, Hamilton, Ontario, and meta-analysis aim to compare LSG and LRYGB on weight loss,
Canada; and zFaculty of Medicine, University of Toronto, Toronto, Ontario,
Canada. comorbidities, and complications.
Y dennishong70@gmail.com.
The authors report no conflicts of interest.
Supplemental digital content is available for this article. Direct URL citations METHODS
appear in the printed text and are provided in the HTML and PDF versions of
this article on the journal’s Web site (www.annalsofsurgery.com). Eligibility Criteria
Copyright ß 2019 Wolters Kluwer Health, Inc. All rights reserved.
ISSN: 0003-4932/19/27301-0066 We included RCTs that compared LRYGB versus LSG in
DOI: 10.1097/SLA.0000000000003671 patients with severe obesity (BMI >40 kg/m2 or >35 kg/m2 with

66 | www.annalsofsurgery.com Annals of Surgery  Volume 273, Number 1, January 2021

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Annals of Surgery  Volume 273, Number 1, January 2021 LSG Versus LRYGB

obesity-related comorbidities). Exclusion criteria were (1) non-RCTs Risk of Bias Assessment and Certainty of Evidence
(cohort studies, case series and reports, reviews, letters, and edito- Risk of bias for individual RCTs was assessed using the
rials) (2) studies which did not compare LRYGB versus LSG (3) Cochrane Collaboration’s tool for assessing risk of bias in RCTs.14
studies with no relevant primary or secondary outcomes of interest Certainty of evidence for estimates derived from each meta-analysed
(4) non-human studies (5) studies with less than 10 eligible patients. outcome was assessed by grading of recommendations, assessment,
When the results of a single trial were reported for multiple time development, and evaluation (GRADE).15
points across more than 1 publication, we collected the most com-
plete data for every follow-up time point reported across Statistical Analysis
all publications. All statistical analysis and meta-analysis were performed on
STATA, version 14 (StataCorp, College, TX) and Cochrane Review
Outcomes Assessed Manager 5.3 (London, United Kingdom) with a level of significance
Primary outcome was changed in BMI at 1, 3, and 5 years set at P of <0.05. We performed pairwise meta-analyses using a
after bariatric surgery. Secondary outcomes were: (1) remission of DerSimonian and Laird random-effects model for continuous and
comorbidities including T2DM, hypertension, and dyslipidemia at 1, dichotomous outcomes. Pooled effect estimates were obtained by
3, and 5 years after surgery (2) changes in biochemical calculating the mean difference (MD) for continuous outcomes and
outcomes including hemoglobin A1c (HbA1c) (%), fasting glucose risk ratio (RR) for dichotomous outcomes along with their respective
(mg/dL), fasting insulin (uIU/mL), homeostatic model assessment 95% confidence intervals (CI) to confirm the effect estimation. In
of insulin resistance (HOMA-IR), total cholesterol (mg/dL), high- addition, mean and standard deviation were estimated for studies that
density lipoprotein (HDL) (mg/dL), low-density lipoprotein (LDL) only reported median and interquartile range using the estimation
(mg/dL), triglycerides (mg/dL) at 1, 3, and 5 years (3) operating method proposed by Wan et al. Assessment of heterogeneity was
time (minutes) (4) 30-days minor complications (Clavien-Dindo completed using the inconsistency (I2) statistic. Studies reporting
Grade 1 and 2) and 30-days major complications (Clavien-Dindo outcomes at follow-up time points that were not exactly at 1, 3, and
Grade 3 and 4).12 5 years were pooled into time point closest to 1, 3, or 5 years.
Sensitivity analysis by excluding these studies was conducted to
Search Strategy ensure that estimate of effect is not driven by these studies. Subgroup
We searched the following databases covering the period from analysis based on preoperative BMI >40 kg/m2 and BMI 40 kg/m2
database inception through January 2019: MEDLINE, EMBASE, was conducted. We considered I2 higher than 50% to represent
Cochrane Central Register of Controlled Trials, PubMed, and the considerable heterogeneity. Publication bias was assessed using a
major clinical trial registries (ClinicalTrials.gov: http://clinicaltrials.- funnel plot for outcomes that contained more than 10 RCTs as having
gov/; International Clinical Trials Registry Platform Search Portal: less trials than 10 RCTs can lead to bias in interpretation of the funnel
http://apps.who.int/trialsearch/) were searched for ongoing trials. plot.16 We performed meta-analyses of outcomes based on follow-up
The search was designed and conducted by a medical librarian with time points of 1, 3, and 5 years after surgery. These time points were
input from study investigators. The search strategy included key- selected after data extraction as these were the most common time
words such as ‘‘gastric bypass’’ and ‘‘sleeve gastrectomy’’ (complete points reported across all trials.
search strategy shown in Supplementary Table 1, http://link-
s.lww.com/SLA/B813). We did not include the term ‘‘randomized RESULTS
trial’’ to ensure that all randomized studies were captured manually
through the screening process. We also searched the references of Study Characteristics
published studies and grey literature manually to ensure that relevant From 5783 potentially relevant citations from the search, 33
articles were not missed. We did not discriminate full texts by studies met the inclusion criteria.10,11,17– 47 Fig. 1 depicts a preferred
language. This systematic review and meta-analysis are reported reporting items for systematic reviews and meta-analyses flow
in accordance with the preferred reporting items for systematic diagram of study selection process. Studies were conducted between
reviews and meta-analyses.13The protocol of this study was regis- 2006 and 2018 in 13 countries including Brazil, China, Finland,
tered in the Prospective Register of Systematic Reviews (PROS- France, Greece, Israel, Netherlands, New Zealand, Poland, Spain,
PERO). Sweden, Switzerland, and United States. All studies were RCTs
comparing LRYGB versus LSG in patients with class 2 obesity or
Data Abstraction greater (BMI 35 kg/m2). Peterli et al published 5 studies of the
At least 2 reviewers independently screened the searched same trial at follow-up time points of 3 months, and 1, 2, 3, and 5
titles, abstracts, and full texts after the inclusion and exclusion years. Salminen et al published 3 studies of the same trial at follow-
criteria. Reviewers were not blinded to authors, institution, or the up time points of 1 and 6 months, and 1, 3, and 5 years. Vix et al
journal where the manuscript was published. Discrepancies that published 2 studies of the same trial at follow-up time points of
occurred at the title and abstract screening stages were resolved 6 months and 1 year. In total, 2475 patients were included, with 1223
by automatic inclusion to ensure that all relevant papers were not randomized to LRYGB and 1252 randomized to LSG. Of these
missed. Discrepancies at the full-text or data abstraction stage were patients, 70% were female with a weighted mean age of 43.4 (range,
resolved by consensus between 2 reviewers and if disagreement 29.3–51.5) years, and a weighted mean preoperative BMI of 43.47
persisted, a third reviewer was consulted. Two reviewers indepen- (4.29) kg/m2. The detailed characteristics of included trials are
dently conducted data abstraction onto a standardized spreadsheet reported in Supplementary Table 2, http://links.lww.com/SLA/
designed a priori. The following data were abstracted from included B813. Raw values for all outcomes are reported in Supplementary
studies: study characteristics (author, country, year of publication, Tables 3 and 4, http://links.lww.com/SLA/B813.
single or multi-center design, funding source, inclusion and exclu-
sion criteria), patient demographics (mean age at time of surgery, % Body Mass Index
female, number of patients included, mean BMI before and Supplementary Table 5, http://links.lww.com/SLA/B813
after surgery), follow-up time points, type of bariatric surgery, presents the meta-analysis of all outcomes with its certainty of
and outcomes. evidence according to GRADE. In total, 16 RCTs (n ¼ 1673)

ß 2019 Wolters Kluwer Health, Inc. All rights reserved. www.annalsofsurgery.com | 67

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Lee et al Annals of Surgery  Volume 273, Number 1, January 2021

FIGURE 1. PRISMA diagram –

transparent reporting of system-
atic reviews and meta-analysis
flow diagram outlining the search
strategy results from initial search
to included studies. PRISMA indi-
cates preferred reporting items for
systematic reviews and meta-anal-
yses.

reported changes in BMI at 1 year after bariatric surgery. Patients 1.04, P ¼ 0.12; I2 ¼ 40%; 508 patients; 9 trials; high certainty of
who received LRYGB had significantly greater decrease in BMI than evidence) (Fig. 3A), 3 years (RR 0.88, 95% CI 0.72–1.07, P ¼ 0.19;
LSG by 1.25 kg/m2 at 1 year (95% CI 2.01 to 0.49, P ¼ 0.001; I2 I2 ¼ 0%; 208 patients; 4 trials; moderate certainty of evidence)
¼ 88%; 1673 patients; 16 trials; moderate certainty of evidence) (Fig. 3B), or 5 years after surgery (RR 0.79, 95% CI 0.57–1.10, P ¼
(Fig. 2A). At 3 years after surgery, the LRYGB group had signifi- 0.17; I2 ¼ 57%; 351 patients; 4 trials; low certainty of evidence)
cantly greater decrease in BMI by 1.71 kg/m2 than the LSG group (Fig. 3C).
(95% CI 2.68 to 0.74, P < 0.001; I2 ¼ 47%; 595 patients; 5 trials;
moderate certainty of evidence) (Fig. 2B). However, at 5 years after Cardiovascular Risk Profile
surgery, the data were insufficient to estimate the difference in BMI The remission of dyslipidemia was reported by 4 trials (n ¼
change between 2 procedures with precision (MD 1.46, 95% CI 364) at 1 year, and 6 for hypertension (n ¼ 630) at 1 year. Compared
3.15 to 0.23, P ¼ 0.09; I2 ¼ 91%, 719 patients; 4 trials; low to LSG, LRYGB had significantly greater remission of dyslipidemia
certainty of evidence) (Fig. 2C). However, much of the heterogeneity at at 1 year (RR 0.58, 95% CI 0.46–0.73, P < 0.001; I2 ¼ 0%; 364
5 years was introduced by a trial by Ruiz-Tovar et al40 and after patients; 4 trials; moderate certainty of evidence) and 5 years (RR
performing a sensitivity analysis excluding this study, weight loss 0.68, 95% CI 0.46–0.99, P ¼ 0.04, I2 ¼ 77%; 351 patients; 4 trials;
favored LRYGB at 5 years with no heterogeneity (MD 2.20, 95% CI low certainty of evidence) (Supplementary Fig. 1, http://link-
2.36 to 2.04, P < 0.001; I2 ¼ 0%, 353 patients; 3 trials). There were s.lww.com/SLA/B813). However, there was no difference in the
2 studies with preoperative BMI less than 40.37,46 Subgroup analysis of remission of hypertension between the 2 surgeries at 1 year (RR
patients with preoperative BMI greater than 40 did not change the 0.91, 95% CI 0.81–1.01, P ¼ 0.08; I2 ¼ 0%; 478 patients; 5 trials;
effect estimate that is present at 1 and 3 years after surgery. high certainty of evidence) or 5 years post-surgery (RR 0.86, 95% CI
0.68–1.10, P ¼ 0.24; I2 ¼ 47%; 446 patients; 4 trials; high certainty of
Type 2 Diabetes evidence) (Supplementary Fig. 2, http://links.lww.com/SLA/B813).
The rate of remission for T2DM was reported by 9 trials (n ¼
508) at 1 year, 4 trials (n ¼ 208) at 3 years, and 4 trials (n ¼ 351) at 5 Lipid Biochemical Outcomes
years. There was no significant difference in the rate of remission for Ten trials reported levels of triglycerides (n ¼ 753) at 12
T2DM between LRYGB and LSG at 1 year (RR 0.86, 95% CI 0.71– (range, 3–60) months, 6 for total cholesterol (n ¼ 429) at 1 year, 6 for

68 | www.annalsofsurgery.com ß 2019 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Annals of Surgery  Volume 273, Number 1, January 2021 LSG Versus LRYGB

FIGURE 2. Pairwise random-effects meta-analysis forest plot comparing laparoscopic sleeve gastrectomy versus laparoscopic Roux-
en-Y gastric bypass on (A) Change in BMI 1 yr after surgery. (B) Change in BMI 3 yr after surgery. (C) Change in BMI 5 yr after
surgery. BMI indicates body mass index.

LDL (n ¼ 429) at 1 year, and 9 for HDL (n ¼ 549) at 12 (range, 3–60) B813). There was no significant difference between HDL levels
months. The LRYGB group had significantly greater reduction in between the 2 surgeries (P ¼ 0.06) (Supplementary Fig. 6, http://
triglycerides levels than the LSG group (MD 12.60 mg/dL, 95% CI links.lww.com/SLA/B813).
24.78 to 0.42, P ¼ 0.04; I2 ¼ 91%; 753 patients, 10 trials;
moderate certainty of evidence) (Supplementary Fig. 3, http://link- Diabetes Related Biochemical Outcomes
s.lww.com/SLA/B813). Moreover, LRYGB had significantly lower Overall, 10 trials reported outcomes for fasting glucose (n ¼
total cholesterol levels compared to LSG at 1 year after surgery (MD 563) at 1 year, 11 for %HbA1c (n ¼ 761) at 12 (range, 3–60) months,
15.55 mg/dL, 95% CI 21.98 to 9.11, P < 0.001; I2 ¼ 73%; 429 7 for fasting insulin (n ¼ 340) at 12 (range, 3–12) months, and 7 for
patients; 6 trials; moderate certainty of evidence) but no difference HOMA-IR (n ¼ 340) at 12 (range, 3–60) months. Although LRYGB
was found at 3 years (P ¼ 0.22) (Supplementary Fig. 4, http:// did not significantly reduce fasting glucose compared to LSG at
links.lww.com/SLA/B813). Similarly, patients with LRYGB had 1 year, LRYGB did reduce fasting glucose compared to LSG at
significantly lower LDL levels compared to patients with LSG at 5 years (MD 21.04 mg/dL, 95% CI 33.94 to 8.14, P ¼ 0.001; I2
1 year (MD 19.04 mg/dL, 95% CI 28.66 to 9.42, P ¼ < 0.001; ¼ 87%; 231 patients; 3 trials; low certainty of evidence) (Supple-
I2 ¼ 93%; 429 patients; 6 trials; moderate certainty of evidence), mentary Fig. 7, http://links.lww.com/SLA/B813). Conversely, there
3 years (MD 14.50 mg/dL, 95% CI 16.56 to 12.45, P < 0.001; was no difference between LRYGB and LSG for changes in %
I2 ¼ 71%; 407 patients; 3 trials; low certainty of evidence), but not HbA1c (P ¼ 0.45), fasting insulin (P ¼ 0.41), and HOMA-IR (P ¼
5 years (P ¼ 0.65) (Supplementary Fig. 5, http://links.lww.com/SLA/ 0.44) (Supplementary Fig. 8, http://links.lww.com/SLA/B813).

ß 2019 Wolters Kluwer Health, Inc. All rights reserved. www.annalsofsurgery.com | 69

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Lee et al Annals of Surgery  Volume 273, Number 1, January 2021

FIGURE 3. Pairwise random-effects meta-analysis forest plot comparing laparoscopic sleeve gastrectomy versus laparoscopic Roux-
en-Y gastric bypass on (A) Remission of type 2 diabetes 1 yr after surgery. (B) Remission of type 2 diabetes 3 yr after surgery. (C)
Remission of type 2 diabetes 5 yr after surgery.

Perioperative Outcomes generation. Allocation concealment was present in 97% (32/33) of

Operating time for bariatric surgery was reported by 4 trials (n studies, and 88% of studies (29/33) adequately explained incomplete
¼ 462). LSG had a significantly shorter operating time compared to outcome data or loss to follow-up. However, blinding of participants
LRYGB (MD 50.58, 95% CI 76.29 to 24.86, P ¼ 0.0001; I2 ¼ occurred in only 9% (3/33) of studies, blinding of healthcare
95%; 4 trials; 462 patients) (Supplementary Fig. 9, http://link- providers in 9% (3/33) of studies, and blinding of outcome assess-
s.lww.com/SLA/B813). Due to the inadequate reporting of late ment in 15% (5/33) of studies. No study had selective reporting of
(greater than 30 days) complications across majority of the trials, outcomes or other biases. Therefore, following the Cochrane Risk of
meta-analysis was only possible for early (within 30 days) minor and Bias tool, the majority of the studies had low selection bias, high-
major complications. There was no difference in minor early com- performance bias, high detection bias, and low reporting bias.
plications between LRYGB and LSG (RR 0.82, 95% CI 0.52–1.27, P GRADE certainty of evidence is summarized in Supplemen-
¼ 0.37; I2 ¼ 25%; 1128 patients; 10 trials; moderate certainty of tary Table 5, http://links.lww.com/SLA/B813. No outcomes were
evidence) (Fig. 4A). Similarly, from 10 RCTs (n ¼ 1518), there was rated down for risk of bias because most studies had adequate
no difference in major early complications between LRYGB and randomization, allocation concealment, low attrition bias, and low
LSG (RR 0.82, 95% CI 0.58–1.16, P ¼ 0.25; I2 ¼ 0%; moderate reporting bias. However, the majority of studies did not blind
certainty of evidence) (Fig. 4B). Specific complications reported by participants, healthcare providers, and outcome assessors due to
each trial is presented in Supplementary Table 6, http://links. the nature of the comparison including surgery. Nonetheless, blind-
lww.com/SLA/B813. ing has less impact on the objective outcomes analyzed in the present
meta-analysis such as BMI, comorbidities, and biochemistry profiles
Quality Assessment of Studies of patients. Several outcomes including BMI, biochemical outcomes,
A summary of the risk of bias across all studies is provided in and dyslipidemia, and T2DM at 5 years were rated down for
Fig. 5. In brief, all included trials had appropriate random sequence inconsistency because of either high heterogeneity (I2 >50%), or

70 | www.annalsofsurgery.com ß 2019 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Annals of Surgery  Volume 273, Number 1, January 2021 LSG Versus LRYGB

FIGURE 4. Pairwise random-effects meta-analysis forest plot comparing laparoscopic sleeve gastrectomy versus laparoscopic Roux-
en-Y gastric bypass on (A) minor early (less than 30 d) complications. (B) Major early (less than 30 d) complications.

varied point estimates with little overlap of CIs. Outcomes with large cholesterol at 1 year, LDL at 1 and 3 years, and fasting glucose at 5
CIs overlapping no effect, or fewer than 400 patients were also years. In addition, though not significant, diabetes remission rates at
downgraded for imprecision. Other outcomes with 10 or more RCTs every interval favored LRYGB. LSG has long been thought to result
had low publication bias as their funnel plots were symmetrical in fewer complications than LRYGB; however, no differences in
(Supplementary Fig. 10, http://links.lww.com/SLA/B813). Overall, major or minor early complications were found in this systematic
there was low certainty of evidence for change in BMI at 5 years, review. Despite the inclusion of only RCTs, the overall certainty of
remission of T2DM at 5 years, dyslipidemia, and biochemical out- evidence ranged from low to high across all outcomes according
comes at 5 years after surgery. There was a high certainty of evidence to GRADE.
for remission of T2DM at 1 year after surgery, and remission of Previous reviews have explored the effects of LRYGB com-
hypertension at all time points. There was a moderate certainty of pared to LSG. A meta-analysis by Yang et al included 15 RCTs and
evidence for change in BMI at 1 and 3 years, minor and major 1381 patients.48 The study concluded that LRYGB results in greater
complications, and all other outcomes. weight loss than LSG at 3 and 5 years but not at time points less than
3 years. In contrast, the present study included 1094 more patients by
conducting a comprehensive search of the literature and concluded
DISCUSSION that LRYGB results in a greater reduction in BMI at 1 and 3 years, but
This is the most comprehensive systematic review and meta- results were inconclusive at 5 years. Furthermore, although Yang et al
analysis to date comparing LRYGB with LSG. In this review with concluded that LSG results in fewer complications than LRYGB, the
2475 patients, LRYGB results in a significantly greater decrease in definition of a complication was unclear.48Our review stratified
BMI than LSG at 1 and 3 years after surgery. However, certainty of complications by severity as major or minor and by time point as
evidence was low for BMI loss at 5 years because of wide CIs that early and late to reduce ambiguity in accordance with ASMBS
include a large effect favoring LRYGB. Therefore, additional studies reporting standards, concluding that there were no differences in
are needed to improve the precision of the point estimate. Comparing either major or minor early complications.49 The findings of the
the impact of the 2 surgeries on comorbidities, LRYGB results in a current study also agree with 2 previous meta-analyses by Shoar et al
greater remission of dyslipidemia than LSG at both 1- and 5-years and Li et al that found statistically improved long-term weight loss
post-surgery. LRYGB also provides greater improvements in bio- for LRYGB compared to LSG.50,51 However, Shoar et al and Li et al
chemical outcomes than LSG, including greater reductions in total provided conflicting reports on the remission of comorbidities and

ß 2019 Wolters Kluwer Health, Inc. All rights reserved. www.annalsofsurgery.com | 71

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Lee et al Annals of Surgery  Volume 273, Number 1, January 2021

included 3 and 11 RCTs respectively, with both reviews lacking

several recently published high-quality trials.10,11,41 The present
review also examines important biochemical outcomes including
HbA1c, fasting glucose, total cholesterol, LDL, HDL, and triglyc-
erides. Finally, our study differs from previous reviews in its rigorous
assessment of included studies, both on the individual study level for
risk of bias using the Cochrane Risk of Bias Tool, and on the body of
evidence level using GRADE.
The outcomes of the present review are supported by previous
high-quality studies. Although this meta-analysis reports that
LRYGB results in a significantly greater reduction in BMI compared
to LSG at 1 year, a 1.25 kg/m2 reduction in BMI is unlikely to be
clinically significant. Typically, a 5%–10% reduction in weight is
considered to be clinically important with some studies calling for
even greater weight reductions.52Two recent high-quality RCTs have
also reported a marginal difference in weight loss between LRYGB
and LSG. The SLEEVEPASS trial found a trend towards higher
%EWL at 5 years for LRYGB compared to LSG (57% vs 49%);
however, this difference did not meet prespecified equivalence
margins for clinical significance.11 The SM-BOSS trial also found
a trend towards improved excess BMI loss at 5 years for LRYGB
compared to LSG (68% vs 61%); however, this difference was not
significant after adjustment for multiple comparisons.10 Conversely,
a retrospective cohort study of 46,510 patients using the National
Patient-Centered Clinical Research Network did find a significantly
greater percentage total weight loss at 5 years with RYGB compared
to LSG (25.5% vs 18.8%), albeit with a higher rate of complications
for RYGB.53 However, the study’s retrospective, nonrandomized
design limits its conclusions and outcomes for comorbidities were
not reported. In contrast, both the SLEEVEPASS and SM-BOSS
trials found similar outcomes for comorbidities to the present study,
concluding that there were no differences between LRYGB and LSG
for the remission of type 2 diabetes.10,11 Additionally, in the SLEE-
VEPASS trial, a higher percentage of patients receiving LRYGB
achieved remission in dyslipidemia and in the SM-BOSS trial, a trend
towards remission of dyslipidemia favoring the LRYGB group was
seen (P ¼ 0.03 unadjusted, P ¼ 0.09 adjusted for multiple compar-
isons).10,11 These results support the benefit for dyslipidemia in the
LRYGB group found in the present meta-analysis. In the analyses of
BMI change at 5 years after surgery, the trial by Ruiz-Tovar et al
appeared to be an outlier, with a much larger sample size than other
trials but with a mean that favored greater BMI loss after LSG.40
However, Ruiz-Tovar et al’s trial presented methodological issues,
including limited data regarding patient selection, baseline character-
istics, and biological outcomes, and did not report missing data
including for their primary outcome of excess BMI loss.40 Impor-
tantly, the sensitivity analysis excluding this trial demonstrated a
significantly greater loss of BMI in the LRYGB group compared to
LSG at 5 years with no heterogeneity (MD 2.20, 95% CI 2.36 to
2.04, P < 0.0001; I2 ¼ 0%, 353 patients; 3 trials). It is difficult to
assess potential biases without knowing precisely how the trial by
Ruiz-Tovar was conducted, but it is likely that these biases are a
major source of heterogeneity and imprecision in this review. How-
ever, another source of heterogeneity could be the lack of generaliz-
ability of results from the remaining 3 studies due to a strict inclusion
criterion. Although the sensitivity analysis likely represents a true
signal for the 3 remaining trials, it is overall difficult to conclude that
the results are completely representative of the true difference
without future trials in other cohorts with long-term (5 years)
FIGURE 5. Cochrane risk of bias assessment of included ran- follow-ups.
domized controlled trials. The findings of the present study at 1-year and 3-year follow-
up support the 2017 ASMBS guidelines for LSG, suggesting that
RYGB may provide greater weight loss compared to LSG.3 For
context, using the average height, BMI difference of 1.25 kg/m2

72 | www.annalsofsurgery.com ß 2019 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Annals of Surgery  Volume 273, Number 1, January 2021 LSG Versus LRYGB

represents about 10 pounds for average female patients and 11.5 3. Ali M, Chaar E, Ghiassi S, et al. Surgery for obesity and related diseases
ASMBS guidelines/statements American Society for Metabolic and Bariatric
pounds for average male patients, which could represent up to 15% of Surgery updated position statement on sleeve gastrectomy as a bariatric
weight loss. Nonetheless, no current trials have compared LSG and procedure. Surg Obes Relat Dis. 2017;13:1652–1657.
LRYGB with long-term (>10 years) follow-up. An obesity-related 4. Major P, Wysocki M, Dworak J, et al. Analysis of laparoscopic sleeve
comorbidity that often plays a significant role in decision-making is gastrectomy learning curve and its influence on procedure safety and periop-
GERD. Unfortunately, limited number of trials have so far compared erative complications. Obes Surg. 2018;28:1672–1680.
GERD status after LRYGB and LSG.10,27 Given the growing body of 5. Mechanick JI, Youdim A, Jones DB, et al. Clinical practice guidelines for the
perioperative nutritional, metabolic, and nonsurgical support of the bariatric
evidence demonstrating that LSG may lead to the development of surgery patient-2013 update: cosponsored by american association of clinical
GERD or worsen existing GERD, future trials should aim to compare endocrinologists, The obesity society, and american society for metabolic &
the long-term GERD outcome difference between LRYGB and LSG. bariatric surgery. Obesity. 2013;21:S1–S27.
This study has several limitations. First, only a small number 6. Angrisani L, Santonicola A, Iovino P, et al. IFSO worldwide survey 2016:
of studies reported outcomes at a follow-up time of 5 years, limiting primary, endoluminal, and revisional procedures. Obes Surg. 2018;28:3783–
3794.
the certainty of evidence on medium-term outcomes comparing
7. Benaiges D, Más-Lorenzo A, Goday A, et al. Laparoscopic sleeve gastrec-
LRYGB versus LSG. Nonetheless, RCTs that reported outcomes tomy: more than a restrictive bariatric surgery procedure? World J Gastro-
at 5 years were typically higher quality than trials that reported only enterol. 2015;21:11804–11814.
short-term outcomes.10,11 Second, low numbers of patients were 8. Yip S, Plank LD, Murphy R. Gastric bypass and sleeve gastrectomy for type 2
studied for some outcomes including fasting glucose at 5 years and diabetes: a systematic review and meta-analysis of outcomes. Obes Surg.
HOMA-IR, and results were inconclusive for BMI loss at 5 years, 2013;23:1994–2003.
supporting a need for future studies. Third, substantial heterogeneity 9. Li J-F, Lai D-D, Lin Z-H, et al. Comparison of the long-term results of roux-
en-y gastric bypass and sleeve gastrectomy for morbid obesity. Surg Laparosc
existed across outcomes. This may be because of differences in Endosc Percutan Tech. 2014;24:1–11.
postoperative management, surgeons’ skill level, and patient pop- 10. Peterli R, Wölnerhanssen BK, Peters T, et al. Effect of laparoscopic sleeve
ulations across the 13 countries studied. Nevertheless, the effects of gastrectomy vs laparoscopic Roux-en-Y gastric bypass on weight loss in
key outcomes including BMI and comorbidities including dyslipi- patients with morbid obesity: the SM-BOSS randomized clinical trial. JAMA.
demia were similar across different time points, making it less likely 2018;319:255–265.
for heterogeneity to explain the results found in the present review. 11. Salminen P, Helmiö M, Ovaska J, et al. Effect of laparoscopic sleeve
gastrectomy vs laparoscopic Roux-en-Y gastric bypass on weight loss at
Fourth, the present review analyzed changes in BMI rather than 5 years among patients with morbid obesity: the SLEEVEPASS randomized
%EWL, which is now a more common outcome.11 This was because clinical trial. JAMA. 2018;319:241–254.
many included trials did not report %EWL, precluding a uniform 12. Dindo D, Demartines N, Clavien P-A. Classification of surgical complica-
weight loss outcome data collection. Nonetheless, change in BMI is tions: a new proposal with evaluation in a cohort of 6336 patients and results of
still a commonly used outcome for weight loss and is a recommended a survey. Ann Surg. 2004;240:205–213.
outcome according to the ASMBS reporting standards.49 Fifth, 13. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535.
patients and surgeons were unblinded to the type of surgery per-
14. Higgins JPT, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration’s
formed. In particular, both LRYGB and LSG possess unique com- tool for assessing risk of bias in randomised trials. Br Med J. 2011;343. doi:
plications, such as internal hernia being exclusive to LRYGB. 10.1136/bmj.d5928.
Therefore, blinding would make the management of complications 15. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating
challenging, and would be difficult to justify ethically.10 Only quality of evidence and strength of recommendations. BMJ. 2008;336:924–926.
Kehagias et al reported the blinding of patients and medical staff.27 16. Lau J. The case of the misleading funnel plot. BMJ. 2006;333:597–600.
In conclusion, LRYGB resulted in greater BMI loss at 1 and 3 17. Murphy R, Clarke MG, Evennett NJ, et al. Laparoscopic sleeve gastrectomy
years; however, there was insufficient randomized evidence to draw versus banded Roux-en-Y gastric bypass for diabetes and obesity: a prospec-
tive randomised double-blind trial. Obes Surg. 2018;28:293–302.
any conclusions regarding weight loss between the 2 procedures at 5
18. Nemati R, Lu J, Dokpuang D, et al. Increased bile acids and FGF19 after
years. No differences between the 2 procedures were found in sleeve gastrectomy and Roux-en-Y gastric bypass correlate with improvement
remission of type 2 diabetes, despite a trend at every time interval in type 2 diabetes in a randomized trial. Obes Surg. 2018;28:2672–2686.
favoring LRYGB, hypertension, and rates of major and minor 19. Olbers T, Björkman S, Lindroos A, et al. Body composition, dietary intake,
complications. Compared to LSG, LRYGB provides a higher remis- and energy expenditure after laparoscopic Roux-en-Y gastric bypass and
sion of dyslipidemia and lower LDL and total cholesterol levels. laparoscopic vertical banded gastroplasty: a randomized clinical trial. Ann
Surg. 2006;244:715–722.
Large RCTs with low risk of bias and long-term (>5 year) follow up
20. Paluszkiewicz R, Kalinowski P, Wróblewski T, et al. Prospective randomized
are necessary to provide valid data on the relative effectiveness of clinical trial of laparoscopic sleeve gastrectomy versus open Roux-en-Y
LRYGB and LSG for long-term weight loss. gastric bypass for the management of patients with morbid obesity. Video-
surgery Other Miniinvasive Tech. 2012;4:225–232.
ACKNOWLEDGMENTS 21. Peterli R, Wölnerhanssen B, Peters T, et al. Improvement in glucose metabo-
lism after bariatric surgery: comparison of laparoscopic Roux-en-Y
The authors thank Dr. Lawrence Mbuagbaw for the biostatis- gastric bypass and laparoscopic sleeve gastrectomy. Ann Surg. 2009;
tics and research methods review of the manuscript. Moreover, 250:234–241.
authors thank Dr. Stefan Schandelmeier for his insights on research 22. Peterli R, Steinert RE, Woelnerhanssen B, et al. Metabolic and hormonal
methods and utilizing grading of recommendations, assessment, changes after laparoscopic Roux-en-Y gastric bypass and sleeve gastrectomy:
development, and evaluation (GRADE) for this review. a randomized, prospective trial. Obes Surg. 2012;22:740–748.
23. Peterli R, Borbély Y, Kern B, et al. Early results of the swiss multicentre
bypass or sleeve study (SM-BOSS). Ann Surg. 2013;258:690–695.
REFERENCES 24. Ramón JM, Salvans S, Crous X, et al. Effect of Roux-en-Y gastric bypass vs
1. Welbourn R, Hollyman M, Kinsman R, et al. Bariatric surgery worldwide: sleeve gastrectomy on glucose and gut hormones: a prospective randomised
baseline demographic description and one-year outcomes from the fourth trial. J Gastrointest Surg. 2012;16:1116–1122.
IFSO global registry report 2018. Obes Surg. 2018;29:782–795. 25. Peterli R, Wölnerhanssen BK, Vetter D, et al. Laparoscopic sleeve gastrectomy
2. Sjöström L, Lindroos A-K, Peltonen M, et al. Lifestyle, diabetes, and versus Roux-Y-gastric bypass for morbid obesity—3-year outcomes of the
cardiovascular risk factors 10 years after bariatric surgery. N Engl J Med. prospective randomized swiss multicenter bypass or sleeve study (SM-BOSS).
2004;351:2683–2693. Ann Surg. 2017;265:466–473.

ß 2019 Wolters Kluwer Health, Inc. All rights reserved. www.annalsofsurgery.com | 73

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.

Lee et al Annals of Surgery  Volume 273, Number 1, January 2021

26. Helmiö M, Victorzon M, Ovaska J, et al. Comparison of short-term outcome of 40. Ruiz-Tovar J, Carbajo MA, Jimenez JM, et al. Long-term follow-up after
laparoscopic sleeve gastrectomy and gastric bypass in the treatment of morbid sleeve gastrectomy versus Roux-en-Y gastric bypass versus one-anastomosis
obesity: a prospective randomized controlled multicenter SLEEVEPASS gastric bypass: a prospective randomized comparative study of weight loss and
study with 6-month follow-up. Scand J Surg. 2014;103:175–181. remission of comorbidities. Surg Endosc. 2019;33:401–410.
27. Kehagias I, Karamanakos SN, Argentou M, et al. Randomized clinical trial of 41. Biter LU, van Buuren MMA, Mannaerts GHH, et al. Quality of life 1 year after
laparoscopic Roux-en-Y gastric bypass versus laparoscopic sleeve gastrec- laparoscopic sleeve gastrectomy versus laparoscopic Roux-en-Y Gastric
tomy for the management of patients with BMI < 50 kg/m2. Obes Surg. bypass: a randomized controlled trial focusing on gastroesophageal reflux
2011;21:1650–1656. disease. Obes Surg. 2017;27:2557–2565.
28. Helmiö M, Victorzon M, Ovaska J, et al. SLEEVEPASS: a randomized 42. Gounder ST, Wijayanayaka DR, Murphy R, et al. Costs of bariatric surgery in
prospective multicenter study comparing laparoscopic sleeve gastrectomy a randomised control trial (RCT) comparing Roux en Y gastric bypass vs
and gastric bypass in the treatment of morbid obesity: preliminary results. sleeve gastrectomy in morbidly obese diabetic patients. N Z Med J.
Surg Endosc. 2012;26:2521–2526. 2016;129:43–52.
29. Schneider J, Peterli R, Gass M, et al. Laparoscopic sleeve gastrectomy and 43. Ignat M, Vix M, Imad I, et al. Randomized trial of Roux-en-Y gastric bypass
Roux-en-Y gastric bypass lead to equal changes in body composition and versus sleeve gastrectomy in achieving excess weight loss. Br J Surg.
energy metabolism 17 months postoperatively: a prospective randomized trial. 2017;104:248–256.
Surg Obes Relat Dis. 2016;12:563–570. 44. Kalinowski P, Paluszkiewicz R, Wróblewski T, et al. Ghrelin, leptin, and
30. Tang Q, Sun Z, Zhang N, et al. Cost-effectiveness of bariatric surgery for type glycemic control after sleeve gastrectomy versus Roux-en-Y gastric bypass—
2 diabetes mellitus. Medicine (Baltimore). 2016;95:e3522. results of a randomized clinical trial. Surg Obes Relat Dis. 2017;13:181–
31. Viana EC, Araujo-Dasilio KL, Miguel GPS, et al. Gastric bypass and sleeve 188.
gastrectomy: the same impact on IL-6 and TNF-a. Prospective clinical trial. 45. Karamanakos SN, Vagenas K, Kalfarentzos F, et al. Weight loss, appetite
Obes Surg. 2013;23:1252–1261. suppression, and changes in fasting and postprandial ghrelin and peptide-YY
32. Vix M, Diana M, Liu K-H, et al. Evolution of glycolipid profile after sleeve levels after Roux-en-Y gastric bypass and sleeve gastrectomy. Ann Surg.
gastrectomy vs. Roux-en-Y gastric bypass: results of a prospective random- 2008;247:401–407.
ized clinical trial. Obes Surg. 2013;23:613–621. 46. Lee W-J, Chong K, Ser K-H, et al. Gastric bypass vs sleeve gastrectomy for
33. Yang J, Wang C, Cao G, et al. Long-term effects of laparoscopic sleeve type 2 diabetes mellitus. Arch Surg. 2011;146:143–148.
gastrectomy versus roux-en-Y gastric bypass for the treatment of Chinese type 47. Keidar A, Hershkop KJ, Marko L, et al. Roux-en-Y gastric bypass vs sleeve
2 diabetes mellitus patients with body mass index 28-35 kg/m2. BMC Surg. gastrectomy for obese patients with type 2 diabetes: a randomised trial.
2015;15:88. Diabetologia. 2013;56:1914–1918.
34. Woelnerhanssen B, Peterli R, Steinert RE, et al. Effects of postbariatric 48. Yang P, Chen B, Xiang S, et al. Long-term outcomes of laparoscopic sleeve
surgery weight loss on adipokines and metabolic parameters: comparison gastrectomy versus Roux-en-Y gastric bypass for morbid obesity: results from
of laparoscopic Roux-en-Y gastric bypass and laparoscopic sleeve gastrec- a meta-analysis of randomized controlled trials. Surg Obes Relat Dis.
tomy—a prospective randomized trial. Surg Obes Relat Dis. 2011;7:561–568. 2019;15:546–555.
35. Werling M, Fändriks L, Björklund P, et al. Long-term results of a randomized 49. Brethauer SA, Kim J, El Chaar M, et al. Standardized outcomes reporting in
clinical trial comparing Roux-en-Y gastric bypass with vertical banded metabolic and bariatric surgery. Surg Obes Relat Dis. 2015;11:489–506.
gastroplasty. Br J Surg. 2013;100:222–230. 50. Shoar S, Saber AA. Long-term and midterm outcomes of laparoscopic
36. Vix M, Liu K-H, Diana M, et al. Impact of Roux-en-Y gastric bypass versus sleeve gastrectomy versus Roux-en-Y gastric bypass: a systematic review
sleeve gastrectomy on vitamin D metabolism: short-term results from a and meta-analysis of comparative studies. Surg Obes Relat Dis.
prospective randomized clinical trial. Surg Endosc. 2014;28:821–826. 2017;13:170–180.
37. Zhang Y, Zhao H, Cao Z, et al. A randomized clinical trial of laparoscopic 51. Li J, Lai D, Wu D. Laparoscopic Roux-en-Y gastric bypass versus laparo-
Roux-en-Y gastric bypass and sleeve gastrectomy for the treatment of morbid scopic sleeve gastrectomy to treat morbid obesity-related comorbidities: a
obesity in China: a 5-year outcome. Obes Surg. 2014;24:1617–1624. systematic review and meta-analysis. Obes Surg. 2016;26:429–442.
38. de Barros F, Setúbal S, Martinho JM, et al. Early endocrine and metabolic 52. Warkentin LM, Majumdar SR, Johnson JA, et al. Weight loss required by
changes after bariatric surgery in grade III morbidly obese patients: a the severely obese to achieve clinically important differences in health-
randomized clinical trial comparing sleeve gastrectomy and gastric bypass. related quality of life: two-year prospective cohort study. BMC Med.
Metab Syndr Relat Disord. 2015;13:264–271. 2014;12:175.
39. Schauer PR, Bhatt DL, Kirwan JP, et al. Bariatric surgery versus intensive medical 53. Arterburn D, Wellman R, Emiliano A, et al. Comparative effectiveness and
therapy for diabetes — 5-year outcomes. N Engl J Med. 2017;376:641–651. safety of bariatric procedures for weight loss. Ann Intern Med. 2018;169:741–750.

74 | www.annalsofsurgery.com ß 2019 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.