Keywords endoscopic gastroplasty - obesity treatment - lifestyle modification - randomized
controlled trials - weight loss - OverStitch - Endomina - POSE
Introduction
Obesity is a growing global health crisis associated with substantial clinical and
socioeconomic burdens.[1 ]
[2 ] Although lifestyle modification (LM)—comprising dietary changes and increased physical
activity—remains the cornerstone of treatment, many patients are unable to achieve
or maintain meaningful weight loss.[3 ] Bariatric surgery offers superior efficacy but is limited by invasiveness, cost,
and potential complications, making it unsuitable for a significant subset of patients.[1 ]
[2 ]
[4 ]
Endoscopic gastroplasty (EG) has emerged as a minimally invasive bariatric option
that bridges the therapeutic gap between conservative and surgical interventions.[5 ]
[6 ]
[7 ] Various EG techniques, such as OverStitch (formerly Apollo Endosurgery, now Boston
Scientific),[8 ]
[9 ] Primary Obesity Surgery Endoluminal (POSE),[10 ]
[11 ] and Endomina,[12 ] have been developed to reduce gastric volume endoscopically, with the goal of inducing
sustained weight loss. Several randomized controlled trials (RCTs) have compared EG
with LM, but individual studies are often underpowered, and reported outcomes have
varied.[8 ]
[9 ]
[10 ]
[11 ]
[12 ] A prior meta-analysis by Chan et al included both RCTs and observational studies,
reporting favorable results for EG.[13 ] However, that analysis pooled heterogeneous data and did not account for differences
in EG technique, limiting its clinical applicability. Furthermore, it lacked robust
sensitivity analyses and did not evaluate the conclusiveness of findings using tools
such as trial sequential analysis (TSA).[14 ]
Given these limitations, we conducted a meta-analysis restricted to RCTs to provide
a more rigorous and unbiased estimate of the efficacy and safety of EG compared to
LM. We performed predefined subgroup analyses based on procedural technique (POSE,
Endomina, OverStitch ESG), applied leave-one-out sensitivity testing, and used TSA
to determine whether current evidence is conclusive.
The primary objective of this study was to evaluate the impact of EG versus LM on
total body weight loss (TBWL) at 12 months. Secondary objectives included TBWL and
excess weight loss (EWL) at 6 months, and EWL at 12 months. Adverse events and publication
bias were also assessed.
Materials and Methods
Protocol
This meta-analysis was conducted in accordance with the Preferred Reporting Items
for Systematic Reviews and Meta-Analyses guidelines ([Fig. 1 ]). A comprehensive literature search was performed across PubMed, Embase, and the
Cochrane Central Register of Controlled Trials (CENTRAL) from inception to March 2025.
Search terms included: “endoscopic sleeve gastroplasty,” “endoscopic gastroplasty,”
“POSE,” “Endomina,” “Apollo OverStitch,” “bariatric endoscopy,” “obesity,” and “randomized
controlled trial.” Reference lists of included studies and relevant reviews were also
screened to identify additional eligible trials.
Fig. 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow
diagram.
Eligibility Criteria
We included RCTs comparing EG with LM alone in adults with obesity (body mass index
[BMI] ≥ 30 kg/m2 ). Trials were required to report at least one of the following outcomes: TBWL at
6 or 12 months, or EWL at 6 or 12 months. Studies involving pharmacological interventions,
surgical procedures, or nonrandomized designs were excluded.
Data Extraction and Quality Assessment
Two independent reviewers (N.J. and A.G.) screened titles, abstracts, and full texts
and extracted data using a standardized form. Discrepancies were resolved by consensus
or by consulting a third reviewer (S.J.). Extracted data included study characteristics,
sample size, interventions, baseline demographics, weight loss outcomes, and adverse
events. The Jadad scale was used to assess study quality.
For studies with incomplete outcome reporting at predefined time points, outcome values
were estimated using linear regression based on reported trends. Specifically, in
the study by Huberty et al, participants in the control arm crossed over to the intervention
arm after 6 months; only pre-crossover data were used, and regression-based extrapolation
was applied to estimate the 12-month control outcomes. For Abad et al, which reported
only 18-month data, approximate values at 6 and 12 months were derived using regression
modeling of the weight trajectory.
Trial Sequential Analysis
To assess the conclusiveness of cumulative evidence for the primary outcome, TSA was
performed using TSA software version 0.9.5.10 Beta (Copenhagen Trial Unit, Denmark).
The cumulative Z-statistic was plotted against O'Brien-Fleming monitoring boundaries.
Assumptions included a two-sided alpha of 0.05, 90% statistical power, an anticipated
mean difference (MD) of 8% in TBWL at 12 months, and a pooled standard deviation derived
from included studies. The required information size (RIS) was calculated, and cumulative
evidence was assessed using an inverse variance-weighted Z-curve.
Outcomes
The primary outcome was the MD in percentage TBWL at 12 months between EG and LM.
Secondary outcomes included MD in TBWL and EWL at 6 months, and EWL at 12 months.
Adverse events, including serious and total events, were also evaluated.
Data Synthesis and Statistical Analysis
Meta-analyses were conducted using a random-effects model (DerSimonian and Laird method)
to account for between-study variability. For continuous outcomes, pooled MDs with
95% confidence intervals (CIs) were calculated. Heterogeneity was assessed using the
I
2 statistic. Sensitivity analysis was performed via leave-one-out testing. Subgroup
analyses were conducted by EG technique (POSE, Endomina, OverStitch ESG). Publication
bias was evaluated through visual inspection of funnel plots and Egger's test. The
certainty of evidence for each outcome was assessed using the Grading of Recommendations
Assessment, Development and Evaluation (GRADE) approach, considering study limitations,
inconsistency, indirectness, imprecision, and publication bias.
Results
A total of five RCTs comprising 696 participants were included in the final meta-analysis
([Table 1 ]). All studies compared EG with LM in adults with obesity (BMI ≥ 30 kg/m2 ) and reported weight loss outcomes at 6 and/or 12 months. The trials evaluated three
EG techniques: the POSE system (evaluated by Miller et al and Sullivan et al),[10 ]
[11 ] the Endomina platform (Huberty et al),[12 ] and the OverStitch system (Abu Dayyeh et al and Abad et al).[8 ]
[9 ] All trials were conducted in Western populations and demonstrated comparable baseline
characteristics across intervention and control groups.
Table 1
Characteristics of included randomized controlled trials
Miller et al, 2017
Sullivan et al, 2017
Huberty et al, 2020
Abu Dayyeh et al,
2022
Abad et al, 2024
Design
Multicenter RCT
Multicenter RCT
Multicenter RCT
Multicenter RCT
Multicenter RCT
Site
Three centers, Europe
Eleven centers, USA
Two centers, Europe
Nine centers, USA
Four centers, Spain
Inclusion criteria
Age 20–60 years, obesity class I and III, failure of conservative weight loss measures,
no significant weight
change (± 5.0% TBWL) in the last 6 months, had an American
Society of Anesthesiologists score of ≤ 2, not taken any
weight-loss medications for ≥ 6 months, agreed not to have
additional weight-loss interventions or liposuction for
≥ 30 months after study enrollment, and been willing to cooperate
with postoperative dietary recommendations and
assessments
Age 22–60 years, obesity class 1 with at least one nonsevere comorbid obesity-related
condition, or BMI ≥ 35 and < 40 kg/m2 without any condition, cannot opt other weight loss measures for the next 24 months
Age 18–65 years, class I or II obesity, must be able to comply with study protocol,
must live within 75 km of treatment site, following the bariatric multidisciplinary
workup
Age 21–65 years, class I or II obesity, failed nonsurgical weight loss methods interventions,
willing to comply with study protocol
Age 18–70 years, histological evidence of MASH, NAS score > 3, BMI > 30 kg/m2
Exclusion criteria
History of bariatric,
gastric, or esophageal surgery, stricture, or other anatomy or
condition that could preclude passage of endoluminal instruments,
gastroesophageal reflux disease (L.A. classification of
grade B, C, or D), known hiatal hernia > 3 cm, pancreatic
insufficiency/disease; active peptic ulcer; pregnancy or plans of pregnancy within
12 months; present corticosteroid use;
inflammatory gastrointestinal disease; coagulation disorders;
hepatic insufficiency or cirrhosis; >2 years type 2 diabetes
mellitus (HbA1C > 6.5) or uncontrolled type 2 diabetes
(HbA1C > 7%); diabetes treatment with insulin; quit smoking
in last 6 months; immunosuppression; portal hypertension or
varices; or active gastric ulcer disease, outlet obstruction, or
stenosis
History of bariatric, gastric or esophageal surgery, stricture or other esophageal
anatomical defect, severe GERD, hiatal hernia > 3 cm, inflammatory gastrointestinal
diseases, type II DM > 10 years, HbA1c > 7, known hormonal or gastric cause for obesity
Achalasia or any other motility disorders, severe esophagitis, gastro-duodenal ulcer,
heart disease, uncontrolled diabetes mellitus or hypertension, TBWL > 5% over last
6 months, severe comorbidity, GI stenosis or obstruction, previous bariatric therapy,
impending gastric surgery 60 days postintervention
History of foregut, GI surgery or bariatric surgery, Inflammatory GI disease, hiatal
hernia > 4 cm, achalasia or any other motility disorder, severe coagulopathy, any
major illness such as cardiac, pulmonary, etc.
History of prior bariatric surgery, acute cardiac event, heart failure, liver cirrhosis, > 5%
TBWL in 6 months, esophagogastric varices, hepatocellular carcinoma, retroviral disease,
any unstable disease or medical condition that could reduce life expectancy to less
than 2 years
Sample size
Active arm
Control arm
34 (POSE)
10 (lifestyle)
221 (POSE)
111 (sham)
49 (EndoMina)
22 (Lifestyle)
85 (ESG)
124 (control)
20 (ESG)
20 (sham)
Age, y, mean (SD)
Active arm
Control arm
38.3 (10.3)
38.5 (12.5)
44.2 (8.6)
45.3 (9.1)
37.6 (9.9)
45.3 (11.7)
47.3 (9.3)
45.7 (10.0)
55.15 (10.9)
53.05 (11.8)
Male, n (%)
Active arm
Control arm
9 (26.5)
1 (10.0)
26 (11.8)
10 (9.0)
3 (6.0)
2 (9.0)
9 (12.0)
18 (16.0)
11 (55.0)
11 (55.0)
Diabetes mellitus, n (%)
Active arm
Control arm
1 (2.9)
1 (10.0)
4 (7.0)
2 (7.4%)
18 (23%)
36 (33%)
9 (45.0)
11 (55.0)
Weight, kg, mean (SD)
Active arm
Control arm
99.9 (11.1)
96.8 (12.1)
99.7 (12.2)
98.7 (11.6)
93.3 (8.8)
94.7 (9.5)
N = 77, 110
98.4 (12.3)
99.1 (12.8)
106.15 (21.85)
106.50 (18.15)
BMI, kg/m2 , mean (SD)
Active arm
Control arm
36.2 (3.3)
37.2 (3.7)
36.0 (2.4)
36.2 (2.2)
34.8 (2.7)
34.2 (2.5)
35.5 (2.6)
35.7 (2.6)
37.54 (4.81)
38.17 (4.76)
Abbreviations: BMI, body mass index; DM, diabetes mellitus; GERD, gastroesophageal
reflux disease; GI, gastrointestinal; HbA1C, glycated hemoglobin; MASH, metabolic
dysfunction-associated steatohepatitis; NAS, nonalcoholic fatty liver disease activity
score; RCT, randomized controlled trial; SD, standard deviation; TBWL, total body
weight loss.
Total Body Weight Loss at 12 Months
Total Body Weight Loss at 12 Months
The primary outcome of interest was the MD in TBWL at 12 months between the EG and
LM groups. Pooled analysis revealed that EG was associated with a statistically significant
improvement in TBWL at 12 months, with a MD of 7.67% (95% CI: 4.38–10.96). However,
statistical heterogeneity was considerable (I
2 = 92.5%), reflecting substantial variability in the magnitude of benefit across studies
([Fig. 2 ]).
Fig. 2 Forest plot – total body weight loss (TBWL) at 12 months.
To explore this heterogeneity, a leave-one-out sensitivity analysis was performed.
Notably, exclusion of the Sullivan et al study reduced the heterogeneity to 86.53%
and increased the pooled effect size to a MD of 8.795%. While exclusion of the Abu
Dayyeh et al study reduced the heterogeneity to 79.92% with pooled effect size reduced
to 6.198%, suggesting that these trials introduced significant between-study variability.
Despite this heterogeneity, the overall direction of effect consistently favored EG
over LM across all trials ([Supplementary Table S1 ]).
To assess the conclusiveness of the accumulated evidence, a TSA was conducted using
a two-sided alpha of 0.05 and 90% power ([Fig. 3 ]). The anticipated effect size was set at 8% TBWL, based on clinical relevance and
prior literature. The cumulative Z-curve surpassed both the O'Brien-Fleming boundary
and the RIS (67 participants), indicating that the available evidence is sufficient
to establish the superiority of EG over LM for 12-month TBWL. These results mitigate
concerns related to sparse data and repeated significance testing, thereby strengthening
the confidence in the observed effect.
Fig. 3 Trial sequential analysis (TSA).
Subgroup Analysis by Procedural Technique
Subgroup Analysis by Procedural Technique
To determine whether the efficacy of EG varied based on the endoscopic technique used,
subgroup analyses were conducted for the three procedural approaches ([Table 2 ]). The OverStitch system demonstrated the greatest magnitude of effect, with a pooled
MD of 10.82% (95% CI: 6.17–15.46; I
2 = 73.3%). The Endomina system, evaluated in a single RCT, showed a mean TBWL difference
of 6.50%, although pooled heterogeneity could not be calculated due to the single-study
data. The POSE system yielded the most modest benefit, with a pooled MD of 5.61% (95%
CI: 1.56–9.66; I
2 = 80.1%). The test for subgroup differences was statistically significant (Q_between = 37.01;
p < 0.001), indicating that the type of endoscopic platform plays a critical role in
the degree of weight loss achieved ([Fig. 4 ]).
Table 2
Outcomes by EG technique overall and subgroup analysis
Technique
Overall
Apollo OverStitch
Endomina
POSE
p -Value
MD TBWL 12 months (95% CI); I
2 (%)
7.67% (4.38–10.96); 92.5%
10.82% (6.17–15.46); 73.3%
6.50% (4.57–8.43); 0%
5.61% (1.56–9.66); 92.2%
0.001
MD EWL 12 months (95% CI); I
2 (%)
26.24% (11.19–41.29); 98.3%
41.75% (29.10–54.39); 56.6%
15.90% (11.59–20.11),)%
19.08% (4.31–33.86); 95.5%
0.001
MD TBWL 6 months (95% CI); I
2 (%)
6.98% (3.76–10.20); 95.3%
8.42% (4.26–12.59); 63.6%
8.30% (6.97–9.63); 0%
5.33% (0.18–10.49); 97.0%
0.001
MD EWL 6 months (95% CI); I
2 (%)
24.85% (13.44–36.27); 96.4%
32.77% (19.38–46.16); 48.8%
25.20% (22.26–28.14); 0%
19.64% (−2.21 to 41.49); 97.3%
0.001
Abbreviations: CI, confidence interval; EG, endoscopic gastroplasty; EWL, excess weight
loss; MD, mean difference; POSE, Primary Obesity Surgery Endoluminal; TBWL, total
body weight loss.
Fig. 4 Bubble plot – total body weight loss (TBWL) at 12 months.
Secondary Outcomes: TBWL and EWL at 6 and 12 Months
Secondary Outcomes: TBWL and EWL at 6 and 12 Months
Secondary outcomes further supported the efficacy of EG across earlier time points
and additional weight metrics. At 6 months, EG resulted in a pooled MD in TBWL of
6.98% (95% CI: 3.76–10.20; I
2 = 95.3%). Sensitivity analysis excluding the Sullivan et al study reduced heterogeneity
to 38.3% and increased the pooled MD to 8.54%, again highlighting this study's impact
on between-study variability ([Supplementary Fig. S1 ]).
For EWL, EG outperformed LM at both 6 and 12 months ([Supplementary Figs. S2 ] and [S3 ]). The pooled MD in EWL at 6 months was 24.85% (95% CI: 13.44–36.27; I
2 = 96.4%), and at 12 months was 26.24% (95% CI: 11.19–41.29; I
2 = 98.3%). Leave-one-out analyses showed influence of an individual study on the overall
effect size ([Supplementary Table S1 ]).
Subgroup Analysis of Secondary Outcomes
Subgroup Analysis of Secondary Outcomes
Subgroup analyses of 6- and 12-month EWL by procedural technique revealed substantial
differences ([Table 2 ]). At 6 months, the OverStitch ESG subgroup achieved the greatest EWL (32.77%; 95%
CI: 19.38–46.16; I
2 = 48.8%), followed by Endomina (25.20%; 95% CI: 22.26–28.14; I
2 = 0%) and POSE (19.64%; 95% CI: −2.21 to 41.49; I
2 = 97.3%). The differences between subgroups were statistically significant (Q_between = 73.12;
p < 0.001). Similarly, at 12 months, the OverStitch system demonstrated superior performance
with a pooled EWL of 41.75% (95% CI: 29.10–54.39; I
2 = 56.6%), compared with Endomina (15.90%; 95% CI: 11.59–20.21; I2 = 0%) and POSE (19.08%; 95% CI: 4.31–33.86; I
2 = 95.5%). The subgroup effect remained significant (Q_between = 217.54; p < 0.001), reinforcing the clinical relevance of procedural selection in EG.
Adverse Events
Safety profiles across the studies were favorable. The pooled incidence of any adverse
event in the EG group was 10.8% (95% CI: 6.2–17.8%; I
2 = 48.6%). Serious adverse events were infrequent, with a pooled incidence of 2.9%
(95% CI: 1.3–6.1%; I
2 = 0%). Most reported events were mild to moderate in severity and included gastrointestinal
bleeding, transient nausea, dehydration, and abdominal discomfort. No serious device-related
complications were reported in the POSE or Endomina trials. In the OverStitch ESG
subgroup, isolated cases of perigastric fluid collections and prolonged pain were
noted, but all were self-limiting or managed conservatively. No serious adverse events
were observed in the LM control groups.
Publication Bias
Publication bias was assessed using funnel plots and Egger's test. Visual inspection
of funnel plots demonstrated overall symmetry, and Egger's test did not detect statistically
significant small-study effects for any of the primary or secondary outcomes ([Supplementary Fig. S4 ]; p > 0.05). These findings suggest a low risk of publication bias in the included studies.
GRADE assessment showed moderate certainty of evidence for TBWL at 6 and 12 months,
and low certainty for EWL outcomes due to heterogeneity and imprecision. These ratings
support the overall robustness of the primary outcome while highlighting the need
for further studies on secondary endpoints ([Supplementary Table S2 ]).
Discussion
In this meta-analysis of five RCTs involving a total of 696 participants, EG was found
to be significantly more effective than LM alone in achieving weight loss at both
6 and 12 months. EG led to a pooled MD of 7.67% in TBWL at 12 months, and similarly
favorable outcomes were observed for EWL at both 6- and 12-month follow-up intervals.
The robustness of these findings were confirmed through leave-one-out sensitivity
analyses, and TSA demonstrated that the cumulative evidence was both adequately powered
and conclusive, crossing the O'Brien-Fleming monitoring boundary and surpassing the
RIS.[14 ] LM forms the cornerstone of all bariatric and nonbariatric weight management strategies.
Structured LM programs incorporating dietary optimization, physical activity, behavioral
support, and psychological counseling have consistently demonstrated modest but clinically
meaningful weight loss. In fact, LM remains an essential cointervention even in trials
evaluating endoscopic or surgical therapies, reinforcing its foundational role in
obesity care.
A key strength of this study lies in its technique-specific subgroup analysis, which
revealed important differences in efficacy among the three EG platforms. The OverStitch
system showed the greatest magnitude of weight loss benefit, followed by Endomina
and POSE. These findings highlight the clinical relevance of device selection in endoscopic
bariatric interventions and suggest that OverStitch ESG may offer superior outcomes
where available. These differences may partly reflect the technical and design variations
among the platforms.
The OverStitch system is the most widely adopted platform and enables full-thickness
suturing via a dedicated double-channel endoscope, allowing robust gastric plication.[8 ]
[9 ] POSE uses a proprietary endoscope-mounted system, but the earlier version (POSE
1.0) had limited depth of tissue capture and plication strength. Since then, POSE
2.0 has been developed with improved anchors, a shorter procedure time, and better
intragastric remodeling, potentially enhancing clinical efficacy.[15 ]
[16 ] Endomina differs by using an external triangulation platform, which allows use with
a standard single-channel endoscope and offers internal suturing, potentially increasing
its accessibility in centers without advanced equipment.[12 ]
[17 ] These anatomical, technological, and logistical differences may influence both procedural
adoption and clinical outcomes, and should be carefully considered when interpreting
the comparative effectiveness of EG techniques.
Our findings are broadly consistent with prior meta-analyses that support the superiority
of EG over LM for weight loss. However, most earlier studies included heterogeneous
designs and observational data, which can introduce bias and dilute technique-specific
insights. For example, the meta-analysis by Chan et al demonstrated overall efficacy
of EG but combined randomized and nonrandomized studies and did not account for differences
in EG platforms. In contrast, our meta-analysis exclusively included RCTs and corrected
for crossover effects and incomplete data reporting using validated regression-based
estimation. This methodologically rigorous approach provides a more precise and clinically
applicable evaluation of EG efficacy.
In addition to efficacy, the safety profile of EG was favorable across trials. The
pooled rate of serious adverse events was low (2.9%), and most complications were
mild and self-limiting, including abdominal discomfort, transient nausea, and perigastric
fluid collections. No serious device-related events were reported in studies using
the POSE or Endomina platforms. While minor complications were more frequent with
OverStitch ESG, they were manageable with conservative treatment. No adverse events
occurred in the LM arms. These findings support the overall safety and tolerability
of EG, especially in patients who are not suitable candidates for bariatric surgery.[18 ]
This study has several notable strengths. It represents the most current and comprehensive
meta-analysis limited to RCTs evaluating EG versus LM. Subgroup analyses based on
procedural technique offer novel insight into platform-specific efficacy, and the
application of TSA strengthens the internal validity of the primary outcome. Risk
of bias and certainty of evidence were formally evaluated using the GRADE framework,
and publication bias was not evident.
There are few limitations in the current study. The number of included RCTs was relatively
small, and moderate to high heterogeneity was observed for several outcomes. Although
this heterogeneity was partially explained by subgroup and sensitivity analyses, residual
variation likely reflects differences in operator experience, procedural technique,
and population characteristics. We could not perform subgroup analysis according to
baseline BMI due to nonavailability of data. Two of the included trials required extrapolation
of data at 6 or 12 months using regression modeling, which, although conservative
and sensitivity-tested, may introduce some uncertainty. Additionally, all included
trials were conducted in Western populations, limiting the generalizability of findings
to other geographic and ethnic groups, including Asian and Indian populations.[6 ]
[7 ]
[19 ]
[20 ]
In conclusion, EG is significantly more effective than LM alone for short- and intermediate-term
weight loss, with an acceptable safety profile. Among currently available techniques,
OverStitch ESG appears to provide the greatest benefit, although platform-specific
characteristics should be considered when choosing an intervention. These findings
support the integration of EG into obesity treatment pathways, particularly for patients
who are ineligible or unwilling to undergo bariatric surgery. Future research should
focus on evaluating long-term durability of weight loss, metabolic improvements, cost-effectiveness,
and broader applicability across global populations.
Supplementary Table S1 Leave-one-out sensitivity analysis
Supplementary Table S2 GRADE summary of findings: EG vs. LM for obesity management
