Introduction
Esophageal perforation, defined as an acute rupture of the esophageal wall, is a severe
clinical condition with a high mortality rate of 13.3 % [1]. To date there is no consensus about the optimal management of esophageal perforation.
Treatment options range from surgery to endoscopic modalities such as though-the-scope
clips (TTS), over-the-scope clips (OTSC), self-expandable metal stents (SEMS), and
endoscopic vacuum therapy (EVT). Although surgery is supposed to be the gold standard
for managing esophageal perforation, about half of the cases are treated initially
endoscopically (51.3 %), mostly with placement of a metal stent [1]
[2].
Recent studies have shown very promising results in management of upper gastrointestinal
leaks with EVT, including esophageal perforations and postoperative leakage of esophagogastrostomies.
EVT has many theoretical advantages, which are already known based on its use for
vacuum-assisted closure therapy of external wounds, including continuous drainage
that prevents accumulation of secretion, reduction in local edema, increased local
perfusion, and promotion of granulation tissue ingrowth [3]
[4].
An EVT sponge can be manually prepared by assembling a piece of foam in the required
size with a nasogastric tube [5] or a ready-made, commercially available sponge kit can be used (Eso-SPONGE). The
aim of this study was to demonstrate the advantages of EVT in management of acute
esophageal perforation measured by the closure rate of perforations. In addition,
we aimed to underscore the great importance to patient clinical outcomes of early
detection of and therapy for perforation.
Patients and methods
Study design and inclusion criteria
We performed a retrospective, single-center analysis of all patients who presented
with acute esophageal perforation at the Department of Interdisciplinary Endoscopy
of the University Hospital in Marburg, Germany from May 2018 to January 2021, by treating
them with low negative-pressure EVT. This retrospective analysis was approved by the
local ethical review board (Philipps-Universität Marburg, file reference “Studie_ek_mr_210721_denzer-3”).
Each patient provided written informed consent for the procedure. For each patient
there was interdisciplinary consent before starting treatment, considering EVT as
the best possible therapeutic option.
The primary outcome was the rate of closure of perforation with EVT, which was defined
as the endoscopic finding of restoration of continuity of the esophagus, confirmed
by clinical improvement and decrease in inflammatory parameters, even after the start
of oral feeding. Secondary outcomes included: 1) length of the treatment; 2) number
of endoscopic procedures required before closure of the perforation was achieve; and
3) rate of procedure-related complications.
All patients with an acute esophageal perforation were treated with EVT and were included
in the analysis. Perforation was defined as a full-thickness esophageal wall defect.
Patients with a diagnosis of another defect of the esophagus, such as fistulas or
postoperative leaks, were excluded from this analysis. For all patients, the following
parameters were recorded: basic characteristics (age, gender), laboratory data (white
blood cell count and C-reactive protein), and etiology of the perforation. The following
procedure characteristics were also recorded: size of the perforation, its anatomical
location, the time it took to diagnose the perforation, the position of the sponge
(intraluminal or inside the cavity), the number of exchanges of the sponge needed,
the length of treatment and the rate of defect closure. Follow-up was performed by
endoscopy, telephone interviews, and hospital visits.
Procedure
The procedures were performed under conscious sedation by means of intravenous propofol
administration unless the patients were in septic shock with hemodynamic instability,
or at higher-than-normal risk for sedation (according to the American Society of Anesthesiologists
physical status classification), and they underwent general endotracheal anesthesia.
Two experienced endoscopists performed all the procedures. Patients were kept nil
per mouth. Before initial placement of the Eso-SPONGE, a nasojejunal tube was inserted
to maintain enteral nutrition. Broad-spectrum antibiotics and proton pump inhibitors
were administered intravenously.
Initiation of EVT was performed as soon as possible after the diagnosis of perforation
was made, mostly directly after iatrogenic perforation occurred during a therapeutic
intervention. The site of the perforation was carefully evaluated, and the perforation
cavity was irrigated and debrided, if necessary. After placing an overtube transorally
into the cavity or intraluminally using the endoscope as a guide rail, the sponge
was pushed forward through the overtube to the desired location with a pusher ([Fig. 1]). Alternatively, if the overtube did not fit through an esophageal stenosis or through
a perforation with narrow opening, an air knot was created at the proximal end of
the sponge after stitching through the tube and the knot was captured by a grasping
forceps, which had previously been inserted into the scope, and finally the sponge
was inserted together with the scope in the esophagus. The drain, which was connected
to the sponge, was transnasally channeled and then connected to an external suction
pump; thus, negative pressure could be applied. The external pump was set at a continuous
negative pressure of 125 to 150 mm Hg and high intensity. The sponge was changed twice
a week to prevent adhesion to the esophagus or to the wound cavity and to control
the healing process, until closure of the perforation was endoscopically documented.
After that, oral feeding was started and patient clinical course and laboratory data
were assessed for 2 days more before discharge.
Fig. 1 Technical preparation of EVT. a Eso-SPONGE kit. b Endoscope serves as a guide rail for insertion of the overtube. c, d Sponge placement via overtube with a pusher. e Sponge insertion without overtube by creating an air knot at the proximal end of
the sponge after stitching through the tube and capturing the knot by a grasping forceps.
Statistical analysis
Datasets were compiled and a descriptive analysis for patient and procedural characteristics
was performed by using Microsoft Excel. Continuous variables are presented as medians
with interquantile range. Categorical variables are presented as absolute values and
percentages.
Results
Patient and procedure characteristics
From May 2018 to January 2021, all 10 patients with acute esophageal perforations
were treated with EVT and were included in our retrospective analysis. Baseline patient
and procedure characteristics are presented in [Table 1] and in detail in [Table 2].
Table 1
Baseline patient and procedure characteristics.
|
No. patients (n = 10)
|
Age, years, median (IQR)
|
68.5 (39.5–76.8)
|
Sex, n (%)
|
|
4 (40)
|
|
6 (60)
|
Etiology of perforation, n (%)
|
|
4 (40)
|
|
2 (20)
|
|
1 (10)
|
|
1 (10)
|
|
1 (10)
|
|
1 (10)
|
Anatomical location, n (%)
|
|
4 (40)
|
|
2 (20)
|
|
4 (40)
|
Size of the perforation, mm, median (IQR)
|
17.5 (15–20)
|
Time until start of EVT after perforation, days, median (range)
|
0 (0–2)
|
Position of the sponge, n (%)
|
|
8 (80)
|
|
2 (20)
|
No of Eso-SPONGE change, mean (IQR)
|
2.5 (2–3)
|
Treatment duration, days
|
7.5 (7–11.5)
|
Complete closure of the perforation, n (%)
|
10 (100)
|
Clinical follow-up, months, median (IQR)
|
8 (1.3–18.3)
|
IQR, interquartile range; EGD, esophagogastroduodenoscopy; EVT, endoscopic vacuum
therapy.
Table 2
Detailed patient and procedure characteristics.
Case
|
Age, years
|
Sex
|
Etiology of perforation
|
Anatomical location
|
Size, mm
|
Days after perforation to sponge placement
|
Position of the sponge
|
Eso-SPONGE changes, n
|
Treatment duration, days
|
Complete closure
|
1
|
66
|
F
|
Diagnostic EGD; pemphigus vulgaris
|
Upper esophagus
|
10
|
1
|
Intraluminal
|
2
|
7
|
+
|
2
|
25
|
M
|
Foreign body (glass piece)
|
Upper esophagus
|
20
|
0
|
Intraluminal
|
3
|
12
|
+
|
3
|
33
|
F
|
Foreign body (pen)
|
Upper esophagus
|
10
|
0
|
Intraluminal
|
2
|
8
|
+
|
4
|
79
|
F
|
Endoscopic-assisted-diverticulotomy
|
Upper esophagus
|
15
|
0
|
Intraluminal
|
1
|
3
|
+
|
5
|
59
|
M
|
Removal of food bolus
|
Middle esophagus
|
15
|
0
|
Intraluminal
|
2
|
7
|
+
|
6
|
3
|
F
|
Caustic stricture
|
Middle esophagus
|
20
|
0
|
Intraluminal
|
3
|
10
|
+
|
7
|
73
|
F
|
Pneumatic dilation, achalasia
|
Lower esophagus
|
50
|
0
|
Intracavitary and intraluminal
|
5
|
21
|
+
|
8
|
78
|
M
|
Pneumatic dilation, achalasia
|
Lower esophagus
|
20
|
0
|
Intraluminal
|
1
|
4
|
+
|
9
|
79
|
M
|
Pneumatic dilation, achalasia
|
Lower esophagus
|
30
|
0
|
Intraluminal
|
3
|
7
|
+
|
10
|
71
|
F
|
Pneumatic dilation, achalasia
|
Lower esophagus
|
20
|
2
|
Intracavitary and intraluminal
|
5
|
20
|
+
|
EGD, esophagogastroduodenoscopy; F, female; M, male.
We report four cases with a perforation of the upper esophagus, two of which were
associated with ingestion of foreign bodies (a pen and a piece of glass), one of which
occurred accidentally during a diagnostic esophagoscopy in a patient with pemphigus
vulgaris; and the last case occurred as a complication of a flexible endoscopy-assisted
diverticulotomy ([Fig. 2] and [Video 1]).
Fig. 2 Perforation of the upper esophagus after ingestion of a pen. a The pen was captured with a snare. b Perforation after removal of the pen. c Intraluminal placement of the sponge just below the upper esophageal sphincter. d Perforation closure endoscopically confirmed.
Video 1 Management with EVT of an acute perforation of the upper esophagus that occurred
after ingestion of a pen.
Furthermore, we treated two patients with perforation of the middle esophagus. In
one case, the perforation occurred during removal of a food bolus. The other case
involved a 3-year-old child with a caustic stricture of the esophagus, who unfortunately
experienced perforation of the middle esophagus during a gastroscopy 3 weeks after
the caustic ingestion. In that case, we had to cut the sponge to the appropriate size
to fit the small diameter of the child’s esophagus with the caustic stricture. Because
the overtube would not fit in the child’s esophagus, we had to insert the sponge into
the esophagus together with the scope, by creating a knot at the proximal end of the
foam after stitching through the tubing and by grasping the knot with a rat tooth
grasper, which had previously been inserted into the scope ([Fig. 3] and [Video 2]).
Fig. 3 Perforation of the middle esophagus in a child with caustic stricture. a CT scan topogram of a tension pneumothorax. b Intraluminal placement of the sponge. C Site of perforation on Day 3. d Perforation closure on Day 7. e End of EVT on Day 10. f Complete closure documented after injection of contrast through the scope in the
esophagus.
Video 2 Management with EVT of an acute iatrogenic perforation of the middle esophagus that
occurred during diagnostic gastroscopy in a child with caustic stricture.
Moreover, we report four cases with perforation of the lower esophagus after pneumatic
balloon dilation (30 mm) of the lower esophegeal sphincter for achalasia ([Fig. 4] and [Video 3]).
Fig. 4 Perforation of the lower esophagus after pneumatic balloon dilation for achalasia.
a Large perforation with a mediastinal cavity at an early stage (Day 0). b Intracavitary placement of the sponge. c Condition after 3 days of EVT. d Smaller perforation defect after 7 days of EVT. e Perforation closure endoscopically confirmed.
Video 3 Management with EVT of an acute iatrogenic perforation of the lower esophagus that
occurred after pneumatic balloon dilation for achalasia.
In eight of 10 cases, EVT was started immediately after the perforation was seen on
endoscopy. In one patient with perforation of the upper esophagus, the initial defect
closure with TTS clips was not sufficient, as shown the next day on a computed tomography
(CT) scan with oral contrast; thus, EVT was started 1 day after perforation. In another
case, the perforation was suspected 2 days after pneumatic dilation based on the increase
of C-reactive protein and confirmed endoscopically, leading to delayed initiation
of EVT. In all but two cases, the sponge was placed intraluminally; in the remaining
two cases, the patients were treated initially with intracavitary placement due to
the large size of the perforation. In seven of 10 cases, a CT scan was performed after
diagnosis of perforation was made to access mediastinitis and extraluminal fluid collections.
Primary and secondary outcomes
The median duration of EVT was 7.5 days (IQR 7–11.5) with a median number of 2.5 sponge
changes (IQR 2–3), leading to successful closure of the perforation in all cases (n = 10/10),
confirmed endoscopically. No procedure-related complication has been documented. Clinical
follow-up data are available for all 10 patients (median 8 months, IQR 1.3–18.3).
Sufficient oral intake confirmed the good clinical course in all of them.
Discussion
Esophageal perforation remains a challenging clinical condition with no treatment
consensus so far. Current guidelines from the European Society of Gastrointestinal
Endoscopy recommend a first-step endoscopic approach over surgery [2]. A high mortality rate is the major concern about surgery. Open surgery for esophageal
perforation has a reported overall mortality rate of around 20 % [6]. A systematic review of non-comparative studies also showed a mortality rate of
17 % with surgical management, whereas SEMS placement had a mortality rate of 7.5 %
[7].
While TTS and OTSC are supposed to be a good option for rather smaller defects [8], for many years, SEMS placement has been the main alternative to surgery, especially
for larger defects [2]. In recent years, EVT has emerged as a definitive treatment of esophageal perforation
and many retrospective studies and case reports have reported very encouraging success
rates, as shown in [Table 3]
[3]
[4]
[5]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]. According to a meta-analysis by Jung et al., EVT has a significantly higher closure
rate and a lower mortality rate than SEMS placement [17]. However, their data were extracted from retrospective studies with almost exclusively
anastomotic leaks. To the best of our knowledge, there is only one comparative study,
which included only two patients with perforations in the EVT arm and eight patients
with perforations in the SEMS arm [18]. That study demonstrated promising results in favor of EVT compared with SEMS, regarding
closure rate (84.4 % vs 53.8 %) and stricture rate (9.4 % vs 28.2 %). Thus, to date,
there are very limited comparative data for EVT and SEMS or other endoscopic treatments
in the setting of esophageal perforation. Besides the well-known complications of
SEMS, including stent migration and tissue overgrowth, the main clinical concern about
SEMS, from our perspective, remains the inability to perform endoscopic lavage and
to drain the wound cavity, as well as to control the healing process, as can be done
during every sponge change.
Table 3
EVT for esophageal perforations.
Author
|
Journal
|
Patients
|
Defect size, mm
|
Success rate, n (%)
|
Mortality rate, n (%)
|
Complication rate, n (%)
|
Days after perforation to sponge placement
|
Sponge changes, n, median
|
Duration of therapy, days, median
|
Position of the sponge
|
Mastoridis et al.[9]
|
Minim Invasive Ther Allied Technol. 2020
|
3
|
NA
|
3 (100)
|
0 (0)
|
0/3 (0)
|
1 (0–8)
|
3
|
13
|
IL 3/IC 0
|
Jung et al.[4]
|
Digestion 2020
|
7
|
NA
|
5 (71.4)/4 (57.1)[1]
|
1 (14.3)
|
NA
|
NA
|
6.4
|
25.5
|
IL 4/IC 3
|
Leeds et al.[5]
|
J. Gastrointest. Surg. 2019
|
17
|
NA
|
16 (94)
|
NA
|
NA
|
12.8
|
5.2
|
23.8
|
NA
|
Bludau et al.[3]
|
Surg. Endosc. 2018
|
18
|
NA
|
13 (72.2)
|
3 (16.7)
|
NA
|
NA
|
2.8[2]/4.5[3]
|
8.8[2]/15.6[3]
|
IL 14/IC 5[4]
|
Pournaras et al.[10]
|
World J. Surg. 2018
|
14
|
NA
|
13 (92.9)
|
2 (14.9)
|
NA
|
NA
|
NA
|
NA
|
NA
|
Ooi et al.[11]
|
ANZ J. Surg. 2018
|
5
|
20
|
3 (60)
|
1 (20)
|
NA
|
5
|
6
|
20
|
NA
|
Laukoetter et al.[12]
|
Surg. Endosc. 2017
|
13
|
NA
|
49 (94.2)/42 (80.8)[1]
[5]
|
0
|
0
|
5[2]/5.5[3]
|
6[2]/5[3]
|
23.5[2]/25[3]
|
NA
|
Kuehn et al.[13]
|
J. Gastrointest. Surg. 2016
|
10
|
NA
|
10 (100)
|
NA
|
0
|
NA
|
5
|
15
|
NA
|
Moschler et al.[14]
|
Endosc. Int. Open 2015
|
5
|
NA
|
5 (100)
|
NA
|
NA
|
NA
|
0 (range 0–6)
|
5
|
IL 3/IC 2
|
Loske et al.[15]
|
Endosc. Int. Open 2015
|
10
|
12,5
|
10 (100)
|
0
|
0
|
0
|
1
|
5
|
IL 8/IC 2
|
Heits et al.[16]
|
Ann. Thorac. Surg. 2014
|
10
|
NA
|
9 (90)/7 (70)[1]
|
1 (10)
|
7 (70)
|
NA
|
5.4
|
NA
|
NA
|
EVT, endoscopic vacuum therapy; NA, not available; IL, intraluminal; IC, intracavitary.
1 EVT alone.
2 spontaneous perforations.
3 Iatrogenic perforations.
4 Combination possible.
5 Overall success, not separately for perforations.
Our retrospective analysis showed complete perforation closure in all patients (n = 10/10,
100 %) using only EVT. This very promising outcome is comparable to the results of
the other, mostly retrospective studies (57.1 %–100 %) [3]
[4]
[5]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]. The results of these retrospective studies are difficult to compare due to the
different clinical settings and therapeutic algorithms used in each center. In the
study of Ooi et al., EVT was initiated only due to failure of other treatments, which
included surgery and placement of external drains, endoscopic closure, and percutaneous
external drainage [11]. This might have led to the lower success rate of 60 %.
Moreover, another well-known, important parameter for successful treatment is the
time from perforation to initiation of EVT. Heits et al. reported that the length
of EVT and hospital stay was significantly longer for patients with an interval of
more than 24 hours before initiating EVT [16]. In our cohort, EVT was initiated immediately after perforation in eight of 10 cases,
in one case the next day and in the last case after 2 days. This might explain the
shorter duration and higher success rate of EVT in our study. Loske et al. also reported
having started EVT immediately after diagnosis, which may contribute to complete closure
in all patients and to short treatment duration of median 5 days [15]. In our cohort, therapy duration was slightly longer (median 7.5 days). Notably,
our case in which EVT was started 2 days after perforation had the longest duration
of therapy (21 days), which led to an increase in the median value for treatment duration
in our series.
Furthermore, in our case series, intraluminal placement of the sponge, for the most
part, was sufficient for defect closure, except in two patients in whom the sponge
was placed initially intracavitary. In keeping with previous series, we suggest intraluminal
placement of the sponge at an early stage after perforation and intracavitary placement
for larger perforations or when perforation is diagnosed late, which often I sassociated
with mediastinal fluid collection [13]
[15].
Conclusions
In conclusion, our data confirm promising results with EVT for acute esophageal perforation
published so far. EVT seems to be an effective and safe definitive treatment option,
provided that perforation is early detected and a patient’s healing process is evaluated
every 3 to 4 days.