Introduction
Gastroparesis is a functional disorder that is characterized by delayed gastric emptying
in the absence of mechanical gastric outlet obstruction leading to symptoms including
nausea, vomiting, pain, early satiety, and bloating [1]. The underlying etiologies can be multifactorial; however, the most observed are
diabetes, postsurgical, or idiopathic [2]. Ye et al. demonstrated that the overall prevalence of gastroparesis was 267.7 per
100,000 US adults [3]. The management of gastroparesis entails lifestyle modification and/or use of prokinetic
agents as first line therapy. In many patients, these agents tend to lose efficacy
over time and patients require additional modalities for persistent or refractory
symptoms. These include use of gastric stimulators, Roux-en-Y gastric bypass, pyloroplasty/pyloromyotomy,
gastrojejunostomy, and/or subtotal gastrectomy [4].
An endoscopic approach called gastric per-oral endoscopic myotomy (G-POEM) has been
evaluated in patients with refractory gastroparesis with notable clinical success
[5]. During G-POEM, submucosal gastric tunneling is performed to approach, identify,
and incise the musculature of the pylorus, similar to technique employed for POEM
in the esophagus to treat patients with achalasia [6]. Recent meta-analyses have demonstrated varying clinical success and efficacy for
G-POEM in managing refractory gastroparesis. Meyboodi et al. demonstrated that G-POEM
was associated with reduction of mean Gastroparesis Cardinal Symptom Index (GCSI)
value by –1.57 (confidence interval (CI): –2.2 to –0.9) and mean gastric emptying
by 22.3 % (CI: –32.9 %–11.6 %) after 5 days [7]. Similarly, Kamal et al. showed reduction of GCSI by –1.4 (CI: –1.9–0.9) at 1-year
follow-up [8]. Neither of these meta-analyses compared G-POEM to surgical approach.
Given the lack of comparative data, we sought to perform a systematic review and meta-analysis
of studies comparing the endoscopic approach i. e. G-POEM, to the surgical pyloromyotomy
for managing refractory gastroparesis.
Patients and methods
Search strategy
This meta-analysis was conducted in accordance to guidelines provided by Preferred
Reporting items for Systematic Review and Meta-Analysis (PRISMA) [9]. A detailed search of the following databases was undertaken from inception through
July 29, 2022: MEDLINE (PubMed interface, National Center for Biotechnology Information),
Embase (Elsevier), Web of Science Core Collection (Clarivate), KCI – Koran Journal
index, Global Index Medicus, and Cochrane Central Register of Controlled Trials (Cochrane/Wiley).
The initial search strategy was suggested by lead investigator (M.A.) that was further
refined and executed by expert medical librarian (W.L.-S.) using appropriate medical
subject headings and related vocabulary and syntax. The following keywords were utilized:
“Pyloromyotomy,” “POEM,” “Endoscopy,” “Surgery,” and “Gastroparesis.” A sample search
strategy using PubMed is highlighted in Supplementary Table 1. Our search was not
restricted to language. Citations were imported and deduplicated using EndNote X9
bibliographic management software (Clarivate, Philadelphia, Pennsylvania, United States).
We did not prepare a review protocol prior to screening/data extraction process.
Study definitions
Symptoms related to gastroparesis were quantified using Gastroparesis Cardinal Symptom
Index (GCSI) score, a previously validated survey that scores on post-prandial fullness,
nausea/vomiting, and bloating [10]. Clinical success was defined as improved gastric emptying study (GES) and/or improvement
in GCSI score based on individual study defined criteria on follow-up. The surgical
group included patients that underwent either pyloromyotomy and/or pyloroplasty using
any approach (open, laparoscopic, robotic).
Inclusion/exclusion criteria
The studies were included based on following parameters: (1) Patients: Adult patients
(≥ 18 years old with refractory gastroparesis of any etiology; (2) Intervention: G-POEM;
(3) Control: Surgical pyloromyotomy (laparoscopic, robotic, and/or open) and (4) Outcomes:
Procedure duration, length of stay (LOS), complications, clinical success, post operative
GCSI score. We included comparative studies i. e. case-control, retrospective/prospective
cohort and/or randomized controlled trials (RCTs). We included abstracts if they met
all study criteria. We excluded case reports, case series < 10 patients, single arm
studies, guidelines, and reviews. If more than one publication was notable for overlapping
data, we included the most recent and updated study to capture the most comprehensive
data.
Screening and data collection
Study screening and data extraction was performed by two independent reviewers (M.A.)
and (M.G). The initial screening was performed using titles and abstracts. Pertinent
studies were then further screened using full texts (where applicable). Data was extracted
using Microsoft Excel (Microsoft, Redmond, Washington, United States). Data regarding
demographics (age, gender), etiology of gastroparesis (diabetic, postsurgical, and
unknown/idiopathic), type of surgery (robotic, laparoscopic, open), and outcomes (LOS,
procedure duration, complications, clinical success, post operative GCSI score) were
collected. Any discrepancy during screening and data extraction was resolved through
mutual discussion.
Data synthesis and analysis
Given the presumed heterogeneity of different surgical approaches and etiology of
gastroparesis, Random effects model using DerSimonian-Laird approach was used a priori
for pooling and comparing outcomes. A correction factor of “0.5” was added when outcomes
were 0 for a given study. Relative risk (RR) for dichotomous outcomes and mean difference
(MD) for continuous outcomes were calculated along with 95 % confidence interval (CI)
and P value. P < 0.05 was considered significant for all assessed outcomes. Study heterogeneity
was assessed using the I2 statistical and value > 50 % was considered substantial heterogeneity [11]. Statistical analysis was performed using Open Meta Analyst (CEBM, University of
Oxford, Oxford, United Kingdom).
The GCSI score was calculated for each group preoperatively and postoperatively. Comparison
was made between postoperative GCSI score directly between the two groups as well
as between the reduction/mean difference (MD) of preoperative and postoperative GCSI
between the two groups.
A subgroup analysis was performed based on individual surgical techniques i. e. pyloroplasty
or pyloromyotomy if two or more studies were available reporting outcomes.
Bias assessment
The risk of bias for individual studies was performed using Newcastle Ottawa scale
(NCOS) for observational/cohort studies and Cochrane risk of bias tool for RCTs [12]
[13]. Publication bias was undertaken via funnel plot for qualitative assessment and
Egger’s regression analysis for quantitative assessment using P value. If publication bias was noted, we attempted the “trim-and-fill” method to
assess the changes in effect size for respective outcome.
Results
A total of 516 studies were identified, of which 324 remained after machine deduplication
was undertaken. After rigorous screening, a total of four studies (1 abstract and
3 full texts) were selected for inclusion in the meta-analysis [14]
[15]
[16]
[17]. The details of study selection are highlighted in the PRISM flow diagram ([Fig. 1]). Of the included studies, all were observational. G-POEM was used as an intervention
in all studies. For surgical techniques, two studies used pyloromyotomy [14]
[16] and two studies used pyloroplasty [15]
[17]. The demographic details and patient characteristics of included studies are summarized
in [Table 1].
Fig. 1 PRISMA flow diagram (G-POEM: Gastric per-oral endoscopic myotomy). From: Page MJ, McKenzie JE, Bossuyt PM et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews.
BMJ 2021; 372: n71.
Table 1
Study details and demographics of included patients.
|
Study, year
|
|
Landreneau, 2019
|
Pioppo, 2021
|
AbiMansour, 2020
|
Clapp, 2022
|
|
Type of study
|
|
Case control study
|
Retrospective
|
Retrospective
|
Prospective
|
|
Interventions
|
|
Group 1: G-POEM
Group 2: Laparoscopic pyloroplasty
|
Group 1: G-POEM
Group 2: Laparoscopic pyloromyotomy
|
Group 1: G-POEM
Group 2: pyloromyotomy
|
Group 1: G-POEM
Group 2: Robotic pyloromyotomy
|
|
Total patients
|
|
Group 1: 30
Group 2: 30
|
Group 1: 39
Group 2: 63
|
Group 1: 84
Group 2: 28
|
Group 1: 63
Group 2: 48
|
|
Mean age, (SD)
|
|
Group 1: 44.1 (13.5)
Group 2: 45.4 (14.5)
|
Group 1: 49 (16.5)
Group 2: 45.8 (10.3)
|
Group 1: 50.6 (16.9)
Group 2: 46.2 (17.4)
|
Group 1: 43.9 (14.1)
Group 2: 47.4 (12.4)
|
|
Female (%)
|
|
Group 1: 23 (76.7 %)
Group 2: (76.7 %)
|
Group 1: 33 (84.6 %)
Group 2: 36 (57.1 %)
|
Group 1: 63 (75.0 %)
Group 2: 22 (78.6 %)
|
Group 1: 53 (84.1 %)
Group 2: 44 (91.7 %)
|
|
Mean BMI, (SD)
|
|
Group 1: 24.9 (7.1)
Group 2: 26.1 (6.7)
|
Group 1: 27.7 (7.7)
Group 2: 27.6 (7.5)
|
Group 1: 6.09 (25.1)
Group 2: 25.1 (5.2)
|
Group 1: 28.7 (8.2)
Group 2: 27.3 (5.5)
|
|
Etiology of Gastroparesis
|
Diabetic
|
Group 1: 5
Group 2: 5
|
Group 1: 13
Group 2: 14
|
Group 1: 23
Group 2: 3
|
Group 1: NR
Group 2: NR
|
|
Postsurgical
|
Group 1: 6
Group 2: 6
|
Group 1: 4
Group 2: 16
|
Group 1: 23
Group 2: 12
|
Group 1: NR
Group 2: NR
|
|
Idiopathic/ unknown
|
Group 1: 19
Group 2: 19
|
Group 1: 22
Group 2: 33
|
Group 1: 38
Group 2: 13
|
Group 1: NR
Group 2: NR
|
|
Length of follow-up
|
|
90 days
|
Postoperative
|
NR
|
90 days
|
BMI, body mass index; G-POEM, gastric per-oral endoscopic myotomy; NR, not reported;
SD, standard deviation.
A total of 385 patients were included (216 in the G-POEM group and 169 in the surgical
group). The mean age was 46.9 (± 3.41) and 46.2 (± 0.86) for G-POEM and surgery, respectively.
The female proportion was 79.6 % and 74.0 % for G-POEM and surgery, respectively.
The outcomes for individual studies are shown in [Table 2].
Table 2
Outcomes for individual studies.
|
Study, year
|
Mean procedure duration, mins (SD)
|
Mean LOS, days (SD)
|
Mean Postoperative GCSI score, (SD)
|
Mean reduction in GCSI pre/post, (SD)
|
Clinical success
|
Normal GES
|
Complication rates
|
Reintervention rates
|
Readmisson rates
|
|
Landreneau, 2019
|
G-POEM: 33.9 (18.8)
Surgery: 99.9 (41.8)
|
G-POEM: 1.4 (1.0)
Surgery: 4.6 (5.6)
|
G-POEM: 2.4 (1.5)
Surgery: 2.3 (1.5)
|
G-POEM: 1.6 (0.3)
Surgery: 1.7 (0.3)
|
G-POEM: 6
Surgery: 5
|
G-POEM: 8
Surgery: 7
|
G-POEM: 1
Surgery: 5
|
G-POEM: 1
Surgery: 1
|
G-POEM: 2
Surgery: 5
|
|
Pioppo, 2021
|
G-POEM: 58.0 (27.6)
Surgery: 78.4 (13.1)
|
G-POEM: 1.3 (1.0)
Surgery: 4.2 (0.7)
|
G-POEM: 1 (0.8)
Surgery: 1.7 (0.6)
|
G-POEM: 2.8 (0.8)
Surgery: 1.5 (0.7)
|
G-POEM: 36
Surgery: 52
|
G-POEM: NR
Surgery: NR
|
G-POEM: 5
Surgery: 21
|
G-POEM: NR
Surgery: NR
|
G-POEM: NR
Surgery: NR
|
|
AbiMansour, 2020
|
G-POEM: 60.6 (29.6)
Surgery: 151.8 (72.0)
|
G-POEM: 1.5 (2.4)
Surgery: 10.9 (20.8)
|
G-POEM: 1.1 (1.3)
Surgery: 2.3 (1.1)
|
G-POEM: 1.8 (1.6)
Surgery: 1 (0.9)
|
G-POEM: 24
Surgery: 9
|
G-POEM: 12
Surgery: 6
|
G-POEM: 9
Surgery: 11
|
G-POEM: 8
Surgery: 2
|
G-POEM: NR
Surgery: NR
|
|
Clapp, 2022
|
G-POEM: 25.3 (6.1)
Surgery: 91.5 (10.4)
|
G-POEM: NR
Surgery: NR
|
G-POEM: 2.5 (1.2)
Surgery: 2.0 (1.1)
|
G-POEM: 0.9 (1.2)
Surgery: 1.9 (1.2)
|
G-POEM: NR
Surgery: NR
|
G-POEM: NR
Surgery: NR
|
G-POEM: 12
Surgery: 6
|
G-POEM: NR
Surgery: NR
|
G-POEM: 12
Surgery: 3
|
GCSI, Gastroparesis Cardinal Symptom Index; GES, gastric emptying study; G-POEM, gastric
per-oral endoscopic myotomy; LOS, length of stay; lNR, not reported; SD, standard
deviation.
Procedure duration
The mean procedure time was significantly lower for G-POEM compared to surgery (MD:
–59.47 min, CI: –87.57 to –31.37 min, P < 0.001, I2 = 96.6 %) ([Fig. 2a]). Consistent results were obtained on subgroup analysis for patients that underwent
surgical pyloroplasty (MD: –66.24 min, CI: –69.84 to –62.64 min, P < 0.001, I2 = 0 %). The subgroup analysis although showed shorter duration for G-POEM compared
to surgical pyloromyotomy however this result was not statistically significant (MD:
–54.59 min, CI: –123.95–14.78 min, P = 0.12, I2 = 95.65 %).
Fig. 2 Mean difference for a procedure duration, b LOS, c postoperative GCSI, d preoperative/postoperative difference in GCSI between G-POEM and surgical group. CI,
confidence interval; LOS, length of stay.
Length of stay
The LOS was significantly shorter for the G-POEM group compared to the surgical group
(MD: –3.10 days, CI: –4.21 to –1.98 days, P < 0.001, I2 = 27.4 %) ([Fig. 2b]). The subgroup analysis was possible for surgical pyloromyotomy only. Although lower
LOS was noted for G-POEM, the result did not achieve statistical significance (MD:
–4.93, CI: –10.80–0.94, P = 0.10, I2 = 62.7 %).
Post-GCSI score
The post-procedure GCSI was evaluated by all four studies and was not significantly
different (MD: –0.33, CI: –1.09–0.43, P = 0.39, I2 = 89.6 %) ([Fig. 2c]). The subgroup analysis demonstrated higher post operative GCSI score for G-POEM
group compared to surgical pyloroplasty group (MD: 0.46, CI: 0.02–0.90, P = 0.04, I2 = 0 %). However, G-POEM showed lower post operative GCSI score compared to surgical
pyloromyotomy group (MD: – 0.91, CI: –1.39 to –0.43, P < 0.001, I2 = 65.82 %).
The reduction in the GCSI score preoperatively and postoperatively between the G-POEM
and surgical groups was compared and no difference was observed (MD: 0.27, CI: –0.62
– 1.16, P = 0.55, I2 = 96.45 %) ([Fig. 2d]). Consistent result was obtained when G-POEM was compared to surgical pyloroplasty
only (MD: –0.49, CI: –1.32–0.34, P = 0.25, I2 = 89.7 %). G-POEM showed higher reduction of GCSI postoperatively compared to surgical
pyloromyotomy group (MD: 1.09, CI: 0.60–1.57, P < 0.001, I2 = 66.32 %).
Clinical success/normal GES
Three studies assessed clinical success on follow-up. No significant difference in
clinical success was noted for either group (OR: 0.98, CI: 0.32–3.00, P = 0.98, I2 = 49.34 %) ([Fig. 3a]). Subgroup analysis between surgical pyloromyotomy group and G-POEM was also consistent
(OR: 0.95, CI: 0.14–6.69, P = 0.96, I2 = 74.7 %).
Fig. 3 Odds ratio for a clinical success, b normal GES, c reintervention rate, d readmission rate, and e any complication rates between G-POEM and surgical group. CI, confidence interval.
The postoperative rate of normal GES was assessed by two studies and was not significantly
different between the G-POEM and surgical groups (OR: 0.56, CI: 0.20–1.60, P = 0.28, I2 = 25.84 %) ([Fig. 3b]).
Readmission/reintervention rate
Only two studies assessed the reintervention rates, and no significant difference
was noted between the G-POEM and surgical group (OR: 1.27, CI: 0.31–5.13, P = 0.74, I2 = 0 %). A subgroup analysis was not applicable ([Fig. 3c]).
Only two studies assessed readmission rate and no significant difference was noted
between the G-POEM and surgical groups (OR: 1.20, CI: 0.13–11.31, P = 0.87, I2 = 76.5 %) ([Fig. 3 d]). Both studies used pyloroplasty and hence subgroup analysis for pyloromyotomy was
not applicable.
Adverse events
The overall rates of adverse events (AEs) were assessed by all studies and no significant
difference was noted between G-POEM and surgery (OR: 39, CI: 0.13–1.18, P = 0.10, I2 = 69.5 %) ([Fig. 3e]). Subgroup analysis was consistent for G-POEM and surgical pyloroplasty (OR: 0.66,
CI: 0.08–5.80, P = 0.71, I2 = 69.2 %). Lower overall rates of AEs were noted for G-POEM compared to pyloromyotomy
(OR: 023, CI: 0.11–0.49, P < 0.001, I2 = 0 %).
Risk of bias
The risk of bias using the NCOS is highlighted in Supplementary Table 2. The risk
of bias was applicable for three studies with full text, and each had a score of 6
or more signifying moderate to low risk [15]
[16]
[17]. One study was an abstract and hence risk of bias assessment was not possible [14].
Discussion
Gastroparesis is a debilitating disease that severely impairs the quality of life
of affected patients. Unfortunately, there are limited interventions for the management
of gastroparesis. The alteration of pyloric musculature via mechanical disruption,
to reduce barriers to gastric outflow, using surgical and endoscopic techniques have
been explored.
The approaches that have been performed to achieve pyloric muscle disruption include
pyloroplasty or pyloromyotomy. The difference between the two techniques is that in
pyloromyotomy, a longitudinal incision is made in the avascular plane after the duodenum
is grasped just distal to the level of pylorus. The muscle fibers are spread with
a spreader and pyloric edges are mobilized thus confirming successful pyloromyotomy
[16]. In pyloroplasty, a longitudinal incision is performed through the pylorus and the
incision is then closed using Heineke-Mikulicz technique [15]. A previous retrospective study comparing these two techniques in children with
delayed gastric emptying and gastroesophageal reflux did not find an advantage of
one over the other [18]. Toro et al. assessed the impact of laparoscopic pyloroplasty on patients with gastroparesis
and noted improvement of symptoms in 82 % of patients postoperatively [19]. Similarly, Shada et al. described the largest single center experience for pyloroplasty
and noted a 90 % improvement in GES among affected patients following the procedure
[20]. G-POEM has shown promise with similar efficacy compared to surgical techniques.
Landreneau et al. demonstrated that G-POEM and laparoscopic pyloroplasty were similar
in terms of postoperative GCSI score (2.4 ± 1.5 vs 2.3 ± 1.5, P = 0.85) and improved GES (85.7 % vs 83.3 %, P = 0.91) [15]. In this study, we demonstrated that G-POEM resulted in similar efficacy to surgery
with regard to the following parameters: postoperative GCSI score (MD: –0.33, CI:
–1.09–0.43), reduction in GCSI score preoperatively and postoperatively (MD: 0.27,
CI: –0.62–1.16), and similar rates of obtaining a normal GES (OR: 0.56, CI: 0.20–1.60).
The major attractiveness of G-POEM is its cost-effectiveness. A previous study showed
that G-POEM overall had a 26 % lower procedural cost than surgical pyloroplasty [21]. In this study, we further demonstrated that G-POEM also reduced the overall procedure
time compared to surgery (MD: –59.47 min, CI: – 87.57 to –31.37 min) and overall length
of hospital stay (MD: –3.10 days, CI: –4.21 to –1.98), both of which may contribute
to further reducing the cost. The overall lower procedural time reduces the risk of
prolonged anesthesia as well.
We compare our study to a previously published article on similar topic by Mohan et
al. The authors performed systematic review, pooled analysis and comparison between
the two groups i. e. G-POEM and surgical pyloroplasty. The authors demonstrated similar
clinical success between surgical pyloroplasty, and G-POEM based on GES and GCSI score
which is in line with our analysis [22]. However, the authors could not perform the analysis of mean difference between
pre- and post-GCSI score between the two groups due to lack of data. Further, the
comparison of clinical success between two groups was made indirectly as all included
studies except one were single arm. Our study only included comparative studies and
hence we were able to perform comparative meta-analysis on more outcomes.
Interestingly, a lower overall postoperative GCSI was noted for surgical pyloroplasty
compared to G-POEM, however, the reduction in GCSI preoperatively and postoperatively
were comparable between the two groups. One major limitation of this analysis was
that only two studies utilized pyloroplasty. Furthermore, one of the studies explored
the efficacy of pyloroplasty using a laparoscopic approach and other study used a
robotic technique. We were unable to account for this heterogeneity in our outcome.
Future RCTs, directly comparing individual surgical approaches, can further expand
on the relative benefits and efficacy of each technique.
Our study had some notable limitations. The most important limitations were the lack
of RCTs as well as low number of included studies with lower overall patients included
in respective intervention groups. We were unable to account for outcomes based on
specific etiology of gastroparesis, surgical access (open, laparoscopic, robotic),
and prior exposure to other interventions such as medications (metoclopramide, erythromycin,
domperidone) or botulinum toxin injection. Further, the included studies mostly included
patients at highly advanced tertiary care centers, and hence the generalizability
of the results is questionable. Lastly, the follow-up of patients was not consistent
across the studies and hence important outcomes such as duration and timing of improvement
in symptoms, GES and reduction in GCSI were not consistently assessed. All these aspects
can explain the high heterogeneity noted in our outcomes. Despite the limitations,
this study provides the largest comprehensive comparative analysis currently available
between G-POEM and surgery. We performed subgroup analysis based on type of surgery
i. e., pyloroplasty vs pyloromyotomy and although few outcomes were different owing
to the low number of studies, these were not clinically significant.
Conclusions
In conclusion, G-POEM may be favored over surgical approach for refractory gastroparesis
as it was more cost-effective while demonstrating comparable efficacy. Future randomized
controlled trials (RCTs) should be performed to confirm these results.
