Introduction
Gastric outlet obstruction (GOO) is a common complication of malignant tumors arising
from the pancreas and gastric antrum. Less commonly, GOO may also arise due to malignant
infiltration or external compression from tumors arising from bile ducts, gallbladder,
duodenum, ampulla, retroperitoneum, or metastases. This may lead to recurrent vomiting,
dehydration, malnourishment, and inability to tolerate chemotherapy, which severely
impair quality of life [1 ]. Only a minority of patients with malignant GOO present with a resectable tumor;
most often GOO is indicative of locally advanced disease and requires palliative treatment
[2 ]. Traditionally, treatment options for GOO consisted of surgical (open/laparoscopic)
gastroenterostomy or endoscopic duodenal stenting. Three underpowered randomized controlled
trials comparing surgical management with endoscopic duodenal stenting yielded inconsistent
results [3 ]
[4 ]
[5 ]. This has led various societies including the American Gastroenterology Association
and American Society for Gastrointestinal Endoscopy (ASGE) to advise laparoscopic
gastroenterostomy in clinically fit patients with an expected survival of more than
2–6 months because of long-term patency of the surgical anastomosis, whereas duodenal
stenting is reserved for patients with an expected survival of less than 2–6 months
[6 ]
[7 ]. Duodenal stenting leads to rapid relief of symptoms and shows low morbidity compared
with surgical gastroenterostomy but has been associated with a high rate of stent
dysfunction due to tumor ingrowth, which requires reintervention [8 ]. However, in daily clinical practice, survival can be notoriously difficult to predict
and surgeons are often reluctant to subject patients with malignant GOO to surgical
interventions. This, together with patient preference, likely contributes to duodenal
stenting still being used in the majority of patients with malignant GOO.
Endoscopic ultrasound-guided gastroenterostomy (EUS-GE) is a relatively new, minimally
invasive technique that provides rapid relief of symptoms associated with low morbidity
and long-term patency of the anastomosis. Recently, multiple retrospective series
comparing EUS-GE with surgical gastroenterostomy have indeed shown similarly high
technical and clinical success rates, with significantly faster relief of symptoms,
shorter hospital stay, and lower morbidity with EUS-GE [9 ]
[10 ]
[11 ]
[12 ]. Data comparing EUS-GE and duodenal stenting are limited, with most studies not
controlling for confounders, making interpretation of outcomes data difficult [13 ]
[14 ]. More data on this matter are needed to define the most optimal treatment strategy
in these frail patients. The aim of the current study was to compare efficacy, safety,
and stent dysfunction rate of EUS-GE vs. duodenal stenting in patients with malignant
GOO with propensity score matching to correct for confounders.
Methods
Patients and study design
A multicenter retrospective analysis was performed of all consecutive procedures involving
either EUS-GE or duodenal stenting for GOO between January 2015 and May 2021 at the
Amsterdam University Medical Center (Amsterdam UMC) in the Netherlands, IRCSS San
Raffaele Scientific Institute in Italy, and University Hospitals Leuven in Belgium.
Patients were identified by searching local endoscopic electronic databases. The EUS-GE
group included patients from our previous multicenter analysis in which EUS-GE was
compared with laparoscopic gastroenterostomy [9 ]. The study protocol was approved by the Institutional Review Board of Amsterdam
UMC as well as in each participating center.
Inclusion criteria were: 1) symptomatic malignant GOO; 2) endoscopic and/or radiological
confirmation of an obstruction at the gastric antrum or duodenum; and 3) primary intended
intervention of either EUS-GE or duodenal stenting. Exclusion criteria were: 1) previous
duodenal stenting, EUS-GE, or surgical gastroenterostomy; and 2) follow-up data for
less than 30 days post-procedure.
Data collection
Data collection was conducted by manually extracting data from the electronic patient
charts. Data storage, statistical analysis, and the matching procedure were performed
using IBM SPSS statistics for windows version 26.0 (IBM, Armonk, NY, USA).
Study definitions and end points
The two primary end points were clinical success and stent dysfunction. Both primary
end points were based on the gastric outlet obstruction scoring system (GOOSS). The
GOOSS is an ordinal scoring system ranging from 0 to 3, which is based on the highest
intake tolerability without vomiting (0 = no intake; 1 = liquid only; 2 = soft solids;
3 = full diet) [15 ].
Clinical success was defined as a GOOSS score of at least 2 after the initial intervention
(EUS-GE or duodenal stenting). Stent dysfunction was defined as recurrence of obstructive
symptoms (GOOSS ≤ 1) after initial clinical success.
Secondary end points included technical success, length of hospitalization after initial
EUS-GE or duodenal stenting, adverse events (AEs), and overall survival. Technical
success was defined as successful placement of a duodenal stent crossing the obstruction
site or successful creation of a gastroenteric anastomosis by means of a lumen-apposing
metal stent (LAMS). Intervention-related AEs such as perforation, stent migration,
clinically relevant bleeding requiring intervention and/or blood transfusion, cholangitis,
sepsis, pneumonia, post-procedural fever, and post-procedural pain that occurred within
30 days following the procedure were scored through the ASGE lexicon as mild, moderate,
severe, or fatal [16 ].
Study procedures
EUS-GE
All procedures were performed under deep propofol sedation or general anesthesia.
Prophylactic broad-spectrum antibiotic therapy was routinely administered. EUS-GE
procedures were performed using the wireless EUS-gastroenterostomy simplified technique
(WEST), as described previously [17 ]. In short, a 7-Fr nasobiliary drain or enteral feeding tube was advanced beyond
the stenosis into the proximal jejunum ([Fig. 1a ]). The nasobiliary drain inside the jejunal loop was identified by EUS. Infusion
of saline solution (with or without blue dye) through the drain or feeding tube resulted
in dilation of the jejunal bowel loop ([Fig. 1b ]). The electrocautery-enhanced LAMS (Hot Axios; Boston Scientific Corp., Marlborough,
Massachusetts, USA) was advanced through the gastric wall into the enteric loop using
pure cutting current (100–150 W) ([Fig. 1c ]). After successful entry into the jejunum, the distal flange was deployed under
EUS guidance and carefully retracted against the intestinal wall, creating direct
contact between the gastric and enteral walls ([Fig. 1d ]). The proximal flange was subsequently deployed inside the endoscope working channel
and was then gently pushed out of the endoscope while simultaneously rotating away
the endoscope from the gastric wall. Successful creation of the gastroenteric anastomosis
was confirmed either by (blue dyed) water entering the gastric lumen, or by direct
endoscopic visualization of small intestinal mucosa and/or fluoroscopy ([Fig. 1e,f ]). Either a 15-mm or 20-mm LAMS was used at the discretion of the endoscopist and
according to availability.
Fig. 1 Stepwise approach to endoscopic ultrasound (EUS)-guided gastroenterostomy. a Fluoroscopic image: placement of a 7-Fr nasobiliary catheter or enteral feeding tube
through the gastric or duodenal stenosis at Treitz ligament. The EUS scope was positioned
close to the catheter. b EUS image: the flow of fluid could be seen immediately after saline infusion, which
confirmed visualization of a proximal jejunal bowel loop. c EUS image: dilated jejunal bowel loop after infusion of saline. The electrocautery-enhanced
lumen-apposing metal stent (LAMS) was positioned in line with the small bowel loop. Using
pure-cutting current the stent was brought into the jejunal lumen. d EUS image: the distal flange was released under EUS guidance. The device was then
retracted onto the gastric wall. The proximal flange was deployed inside the working
channel of the endoscope, and slowly pushed out of the endoscope while at the same
time gently pulling back the endoscope from the gastric wall. e Endoscopic image: saline infused through the nasobiliary catheter was, in this case,
dyed with methylene blue and appeared in the gastric lumen after successful LAMS placement.
f Endoscopic view from the stomach showing a patent gastroenteric anastomosis 3 months
after LAMS placement.
Endoscopic duodenal stenting
Endoscopic duodenal stenting was performed under propofol sedation or general anesthesia.
A therapeutic gastroscope or pediatric colonoscope was advanced to the site of gastric
or duodenal obstruction ([Fig. 2a ]). A double-lumen catheter and guidewire were then advanced through the stricture
([Fig. 2b ]). Contrast injection under fluoroscopy was used to determine the length of the stricture,
relation to the papilla, and preferable size of the stent ([Fig. 2c ]). An uncovered self-expandable metal stent (SEMS) was then advanced over the wire
and deployed under endoscopic and fluoroscopic guidance ([Fig. 2d ]). Wallflex duodenal stents (Boston Scientific) and Cook Evolution duodenal stents
(Cook Medical, Bloomington, Indiana, USA) 6, 9, and 12 cm in length were used. The
diameter of the stent was 22 mm in all cases. Adequate positioning of the stent was
confirmed by means of fluoroscopy and endoscopy ([Fig. 2e,f ]).
Fig. 2 Stepwise approach to endoscopic duodenal stenting. a Endoscopic image: malignant duodenal stenosis. b Endoscopic image: traversing the stricture with a double-lumen catheter and a guidewire.
c Fluoroscopic image: contrast injection through the catheter to determine the length
of the stricture, relation to the papilla, and preferable size of the stent. d Fluoroscopic image: advancing the self-expandable metal stent over the guidewire
across the stricture. e Fluoroscopic image: the stent was been deployed with the waist located in the middle
of the stent confirming adequate position. f Endoscopic image: proximal side of the uncovered stent deployed proximally to the
stricture.
Statistical analysis.
Categorical and binary variables were reported as frequencies (%) and were compared
through either Fisher’s exact test or Pearson’s chi-squared test. Continuous variables
were reported as means or median with SD or interquartile range (IQR) and were analyzed
through unpaired t tests or Mann–Whitney U test. The reported analyses were performed in the intention-to-treat population (intended
procedure EUS-GE or duodenal stenting regardless of technical success), unless explicitly
mentioned as per protocol (only patients with technical success). Outcomes were reported
as odds ratios (OR) with a 95 %CI. P values were considered statistically significant if < 0.05.
To minimize selection bias of the observed data, a propensity score matching analysis
was conducted. Propensity score was based on age, sex, GOO etiology, and disease stage,
presence of ascites, and peritoneal carcinomatosis. Variables were selected based
on demographic discrepancies of the main cohort and expected factors of influence
based on recent studies [13 ]
[18 ]
[19 ]
[20 ]. A stringent maximum propensity score difference of 0.05 was used for matching.
For the time to event data analysis, the Kaplan–Meier curve and log-rank test were
used.
Results
Main cohort
A total of 246 eligible patients were identified. After excluding 32 patients in whom
follow-up was less than 30 days, the EUS-GE group consisted of 107 patients (50 %)
and the duodenal stenting group included 107 patients (50 %) (see Fig. 1 s and Table 1 s in the online-only Supplementary Material).
Table 1
Main and matched cohort: baseline characteristics.
Main cohort
Matched cohort
EUS-GE (n = 107)
Duodenal stenting (n = 107)
P value
EUS-GE (n = 88)
Duodenal stenting (n = 88)
P value
Age, years
66 (11.8)
67 (11.2)
0.54
66 (12.1)
66 (10.4)
0.98
Female sex, n (%)
54 (50.5)
58 (54.2)
0.68
44 (50.0)
48 (54.5)
0.65
Follow-up duration, median (IQR), days
90.5 (44–177)
50 (27–126)
0.01
103 (43–184)
51 (30–126)
0.01
Primary disease, n (%)
50 (46.7)
70 (65.4)
0.009
50 (56.8)
56 (63.6)
0.44
15 (14.0)
7 (6.5)
0.11
11 (12.5)
5 (5.7)
0.19
12 (11.2)
8 (7.5)
0.48
8 (9.1)
7 (7.9)
> 0.99
10 (9.3)
10 (9.3)
> 0.99
8 (9.1)
10 (11.4)
0.80
20 (18.7)
12 (11.2)
0.18
11 (12.5)
10 (11.4)
> 0.99
Disease stage, n (%)
n = 104
n = 106
34 (32.7)
45 (42.5)
0.16
32 (36.4)
35 (39.8)
0.76
20 (19.2)
17 (16.0)
0.72
20 (22.7)
15 (17.0)
0.45
14 (13.5)
8 (7.5)
0.26
9 (10.2)
8 (9.1)
> 0.99
36 (34.6)
36 (34.0)
> 0.99
27 (30.7)
30 (34.1)
0.78
Disease manifestations, n (%)
31 (29.0)
24 (22.4)
0.28
23 (26.1)
22 (25.0)
0.86
44 (41.1)
27 (25.2)
0.02
28 (31.8)
25 (28.4)
0.74
EUS-GE, endoscopic ultrasound-guided gastroenterostomy; IQR, interquartile range.
Baseline characteristics are shown in [Table 1 ]. Pancreatic cancer-induced GOO was more frequent in the duodenal stenting group
than in the EUS-GE group (65.4 % vs. 46.7 %; P = 0.009). With regard to disease manifestations, peritoneal carcinomatosis was present
more frequently in the EUS-GE group (41.1 % vs. 25.2 %; P = 0.020), while the presence of ascites did not significantly differ between the
two groups (29.0 % vs. 22.4 %).
Technical success of EUS-GE and duodenal stenting was 94 % (95 %CI 90 %–99 %) vs.
98 % (95 %CI 96 %–100 %), respectively. Clinical success was 90 % (95 %CI 84 %–96 %)
with EUS-GE and 77 % (95 %CI 68 %–85 %) with duodenal stenting. Stent dysfunction
occurred less frequently after EUS-GE (3 %; 95 %CI 0–6 %) than after duodenal stenting
(30 %; 95 %CI 14 %–46 %).
Propensity score matching analysis
Propensity score matching allocated 88 patients in each group (1:1), resulting in
a total of 176 patients. No significant differences in baseline characteristics were
found between the two groups ([Table 1 ]). The overall characteristics of the propensity score-matched cohort included a
mean age of 66 years (SD 11.5), the majority being female (52.3 %), and the underlying
disease being pancreatic cancer (56.8 %), while peritoneal metastasis and ascites
were present in up to a third of patients.
Technical success was achieved in 83/88 (94 %; 95 %CI 89 %–99 %) EUS-GE patients and
in 86/88 (98 %; 95 %CI 95 %–100 %) duodenal stenting patients, with no significant
difference between the two groups (P = 0.44) ([Table 2 ]). Clinical success rate was higher after EUS-GE (80/88; 91 %; 95 %CI 85 %–97 %)
than after duodenal stenting (66/88; 75 %; 95 %CI 66 %–84 %; P = 0.008). Per-protocol clinical success was 96 % (95 %CI 92 %–100 %) after EUS-GE
and 77 % (95 %CI 68 %–86 %) after duodenal stenting, which was a significant difference
(P < 0.001). Median time to clinical success was shorter after EUS-GE (1 day [IQR 1–2])
than after duodenal stenting (2 days [IQR 2–3]; P < 0.001). Median length of hospitalization was similar between the two groups, at
4 days (IQR 2–10.8) after EUS-GE vs. 4 days (IQR 1–9.5) after duodenal stenting.
Table 2
Matched cohort: outcome comparisons.
EUS-GE (n = 88)
Duodenal stenting (n = 88)
OR (95 %CI)
Efficacy
Primary outcomes
83 (94) [89–99]
86 (98) [95–100]
0.39 (0.07–2.04)
80 (91) [85–97]
66 (75) [66–84]
3.33 (1.39–8.00)
80 (96) [92–100]
66 (77) [68–86]
8.06 (2.30–28.57)
1 (1–2)
2 (2–3)
1 (0–1)
1 (0–1)
1 (1) [0–4]
17 (26) [15–37]
0.04 (0.01–0.28)
243 (N/A)
57 (27–169.5)
Secondary outcomes
4 (2–10.8)
4 (1–9.5)
85 (43–157)
57 (18.5–130.5)
Safety
Overall adverse events, n (%) [95 %CI]
9 (10.2) [3.8–16.7]
18 (20.5) [11.9–29.0]
0.44 (0.19–1.05)
ASGE AE severity grading system, n (%) [95 %CI]
2 (2.3) [0–5.5]
6 (6.8) [1.5–12.2]
0.32 (0.06–1.62)
3 (3.4) [0–7.3]
9 (10.2) [3.8–16.7]
0.31 (0.08–1.19)
4 (4.5) [0.1–9.0]
2 (2.3) [0–5.5]
2.05 (0.37–11.49)
0 (0)
1 (1.1) [0–3.4]
0.99 (0.97–1.01)
EUS-GE, endoscopic ultrasound-guided gastroenterostomy; OR, odds ratio; IQR, interquartile
range; N/A, not applicable; ASGE, American Society for Gastrointestinal Endoscopy;
AE, adverse event.
* Per-protocol cohorts.
Median follow-up was 85 days (IQR 43–157) in the EUS-GE group and 57 days (IQR 18.5–130.5)
in the duodenal stenting group. Recurrent GOO occurred in 1/80 (1 %; 95 %CI 0–4 %)
EUS-GE patient due to stent migration after 243 days, and in 17/66 duodenal stenting
patients (26 %; 95 %CI 15 %–37 %; P < 0.001). Median time to stent dysfunction was 243 days after EUS-GE and 57 days
(IQR 27–169.5) after duodenal stenting. Kaplan–Meier analysis showed higher probability
of dysfunction-free survival for EUS-GE (hazard ratio 27.4, 95 %CI 4.2–28.2; P < 0.001) with a 6-month probability of remaining recurrence free of 100 % compared
with 65.0 % with duodenal stenting ([Fig. 3 ]). Both recurrent GOO rates and stent dysfunction-free survival rate by Kaplan–Meier
analyses revealed a significantly higher stent dysfunction rate after duodenal stenting
compared with EUS-GE.
Fig. 3 Kaplan–Meier curve with time to event (stent dysfunction) analysis (log-rank test
P < 0.001). EUS-GE, endoscopic ultrasound-guided gastroenterostomy.
AEs occurred in 9/88 patients (10.2 %; 95 %CI 3.8 %–16.7 %) after EUS-GE and in 18/88
patients (20.5 %; 95 %CI 11.9 %–29.0 %) after duodenal stenting, and were similar
between the two groups (P = 0.09). The AEs in the EUS-GE group consisted of infectious complications (aspiration
pneumonia [n = 1; 1.1 %] or cholangitis [n = 3; 3.4 %]), bleeding (n = 1; 1.1 %),
and post-procedural pain (n = 1; 1.1 %); in three cases (3.4 %) intraperitoneal LAMS
maldeployment resulted in emergency salvage surgery. AEs after duodenal stenting comprised
infectious complications, including aspiration pneumonia (n = 4; 4.5 %), cholangitis
(n = 4; 4.5 %), post-procedural pain (n = 4; 4.5 %), bleeding (n = 3; 3.4 %), atrial
fibrillation (n = 1; 1.1 %), and stent migration (n = 1; 1.1 %). Aspiration pneumonia
occurred in four patients who were under general anesthesia with tracheal intubation,
and in one patient under deep sedation. The severity of AEs was not significantly
different between EUS-GE and duodenal stenting ([Table 2 ]). Median survival after EUS-GE was 85 days (43–157) vs. 57 days (18.5–130.5) after
duodenal stenting.
Discussion
This study reports the first propensity score-matched comparison between EUS-GE and
endoscopic duodenal stenting in malignant GOO, and contains, to the best of our knowledge,
the largest comparative cohort published to date. Our data indicate that, while technical
success was similar between the two groups, EUS-GE showed higher clinical success,
superior long-term stent patency, and similar AE rates compared with duodenal stenting.
To date, two smaller unmatched retrospective studies compared EUS-GE with duodenal
stenting in malignant GOO [13 ]
[14 ]. A recent meta-analysis included an additional three abstracts resulting in a total
of 659 patients published in the literature [21 ]. None of these studies used propensity score matching in an attempt to correct for
confounders such as the type of underlying malignancy, peritoneal carcinomatosis,
or the presence of ascites.
In the current study, clinical success rates, using an intention-to-treat analysis,
were significantly higher after EUS-GE than after duodenal stenting. Similar clinical
success rates of EUS-GE were recently reported in a retrospective analysis of 19 centers
including 267 patients ([Table 3 ]) [22 ]. Clinical success rates of duodenal stenting were slightly lower compared with previous
large studies on this topic, which may be related to the relatively high incidence
of peritoneal carcinomatosis in our study ([Table 3 ]) [23 ]
[24 ]
[25 ]. When comparing only technically successful procedures (per-protocol analysis) in
our study, the benefit of EUS-GE with regard to clinical success was even more striking.
EUS-GE has several features that may explain this superior clinical benefit. First,
the short length of the LAMS (1.5 cm) is likely to facilitate food passage into the
small intestine better than the longer duodenal stent (6–12 cm). Second, the LAMS
in EUS-GE is placed some distance away from the tumor and expands to its full diameter,
whereas compression by the tumor may prevent the SEMS from reaching its full diameter,
which may impair food passage. The positive effect of a larger stent diameter was
confirmed by our previous and other studies showing a superior clinical effect of
the 20-mm over the 15-mm LAMS when compared with surgical gastroenterostomy [9 ]
[22 ]. In addition, the time to clinical success, considered one of the most beneficial
features of duodenal stenting, was even further reduced in the EUS-GE group.
Table 3
Overview of largest series on endoscopic ultrasound-guided gastroenterostomy and endoscopic
enteral stenting.[* ]
First author, year [ref.]
Design, geographic area
Patients, study dates
Treatment and technical details
Efficacy, %
Safety, AEs, %
Long-term outcomes
EUS-GE
Bejjani, 2021 [22 ]
Retrospective multicenter
Europe (7), North America (12)
n = 267
2018–2020
EUS-GE
Freehand EC-LAMS, or balloon assisted EUS-GE
Tech. success: 95.5
Clin. success: 87.0
12.4
Follow-up: 72 days (IQR 23–160)
Recurrence: 6.4 %
Current study
Retrospective, multicenter
Europe (3)
n = 107
2015–2021
EUS-GE
Freehand EC-LAMS
Tech. success: 94.4
Clin. success: 89.7
12.1
Follow-up: 91 days (IQR 44–177)
Recurrence: 3.1 %
Endoscopic enteral stent
Lee, 2009 [24 ]
Prospective, single center
Asia
n = 154
1998–2007
UCSEMS, PCSEMS
Tech. success: 100
Clin. success: 97.4
3.2
Follow-up: 108 days (95 %CI 60–151)
Recurrence: 17.5 %
Costamagna, 2012 [23 ]
Prospective, multicenter
Australia (1), Europe (9), North America (2)
n = 202
2006–2008
UCSEMS
Tech. success: 98.0
Clin. success: 91.0
10.9
Follow-up: 94 days
(95%CI 79–112)
Recurrence: 14.4%
Tringali, 2014 [25 ]
Prospective, multicenter
Africa (1), Australia (1), Europe (3), North America (1), South America (1)
n = 106
2009–2011
UCSEMS
Tech. success: 99.1
Clin. success: 84.5
25.0
Follow-up: 47 days (range 0–195)
Recurrence: 17.6 %
Current study
Retrospective, multicenter
Europe (3)
n = 107
2015–2021
UCSEMS
Tech. success: 98.1
Clin. success: 76.6
17.8
Follow-up: 50 days (IQR 27–126)
Recurrence: 29.3 %
AE, adverse event; EUS-GE, endoscopic ultrasound-guided gastroenterostomy; EC-LAMS,
electrocautery-enhanced lumen-apposing metal stent; Tech., technical; Clin., clinical;
IQR, interquartile range; UCSEMS, uncovered self-expandable metal stent; PCSEMS, partially
covered self-expandable metal stent.
* Studies describing treatment of malignant gastric outlet obstruction with either
EUS-GE or endoscopic enteral stent comprising > 100 cases were selected. Prospective
studies were preferred over retrospective studies when available. AEs are reported
using the definition of the current study.
Endoscopic duodenal stenting is prone to recurrent GOO due to tumor ingrowth through
the meshes of the stent [23 ]
[24 ]
[25 ]. In line with these findings, almost one-third of patients with duodenal stents
in the current study experienced stent dysfunction. The use of a covered SEMS instead
of an uncovered SEMS has been shown to prevent tumor ingrowth and reduce stent dysfunction
[26 ]. Covered SEMSs are, however, associated with a higher risk of stent migration and
post-procedural pancreatitis due to obstruction of the papilla, and are seldom used
given these concerns [27 ]
[28 ]. In EUS-GE, the stent is placed at a distance from the tumor, so that stent obstruction
due to tumor ingrowth is seldom a concern. In the current analysis, stent dysfunction
occurred in only 1.3 % of cases, which is in line with previous published data showing
that stent obstruction in EUS-GE is uncommon and generally caused by either food impaction
or buried LAMS [29 ].
EUS-GE is a more invasive procedure than duodenal stenting, and may lead to more severe
AEs such as peritonitis and perforation, even requiring salvage surgery. Yet our study
showed that, in the hands of endoscopists who have received adequate training in expert
high-volume settings, EUS-GE is a safe procedure [30 ]. Indeed, in our series, AEs associated with EUS-GE were not more frequent nor more
severe than after duodenal stenting.
The choice of therapy for frail patients with malignant GOO remains a matter of debate
and will only finally be resolved once the results from adequately powered randomized
controlled trials become available. Currently, based on the outcomes of previously
underpowered randomized studies comparing duodenal stenting with surgical gastroenterostomy,
duodenal stenting is advocated for use in patients with a prognosis of less than 2–6
months or a poor performance status (World Health Organization [WHO] performance score
≥ 3), whereas surgical gastroenterostomy has been advised in fitter patients with
a better prognosis [6 ]
[7 ]. The findings of our study suggest that EUS-GE may offer a valuable alternative
to the currently employed options: EUS-GE is similar to duodenal stenting as a minimally
invasive option that provides rapid symptom relief and is associated with low morbidity
and early hospital discharge, yet provides an anastomosis that achieves surgical-range
efficacy. Together with the significantly lower need for re-interventions due to less
stent dysfunction, EUS-GE thus seems preferable over duodenal stenting. The latest
guideline of the European Society of Gastrointestinal Endoscopy indeed recommends
EUS-GE, when performed in an expert setting, as an alternative to duodenal stenting
or surgical gastroenterostomy [31 ]. Duodenal stenting may potentially be reserved for patients with an ultra-limited
short-term prognosis in whom EUS-GE is not feasible, for instance when the proximal
jejunal loop cannot be visualized from the stomach, or when a large volume of intervening
ascites obscures localization of a jejunal loop preventing safe placement. The presence
of ascites also warrants attention toward underlying diffuse peritoneal metastatic
disease, which increases the risk of gastrointestinal dysmotility and downstream enteral
obstruction. Both may severely affect the clinical success rate of EUS-GE as well
as duodenal stenting, and should be considered as a relative contraindication.
Our study has certain limitations. The retrospective design introduces confounders.
However, by using propensity score matching analysis, we have tried to overcome these
limitations, allowing a fair comparison between the two techniques. Although data
regarding most important confounders such as etiology, presence of ascites, and (peritoneal)
metastases could be reliably collected retrospectively, others, such as WHO performance
scale, could not. However, overall post-procedural survival was not different between
the two groups after matching, reducing the concern of different baseline frailty.
Finally, despite the promising outcomes of the EUS-GE group, generalizability of these
results outside tertiary academic centers with specific expertise in interventional
EUS remains difficult. Moreover, to date, the procedure includes an off-label use
of the LAMS.
In conclusion, this study showed that in patients with malignant GOO, EUS-GE and duodenal
stenting displayed similar technical success and AEs rates. The higher clinical success,
shorter time to clinical success, and lower GOO recurrence suggest that EUS-GE should
be preferred over duodenal stenting when adequate expertise is available.
