Introduction
Iatrogenic esophageal fistula after esophagectomy is a serious complication for esophageal
and gastric cancers. It can lead to longer hospital stay and high postoperative morbidity
and mortality rates. As esophageal fistula is refractory, management of it is often
challenging.
Interventional endoscopic treatments including the application of glue and endoscopic
clips and stent insertion are increasingly used for treatment of esophageal fistulas
[1]. A main limitation of covered stent is the risk of migration, with several studies
identifying rates more than 50 % [2]
[3]. On the other hand, application of glue has become a promising method in clinical
practice for avoiding reoperation after gastrointestinal leakage [1]
[4]. The use of cyanoacrylate glue was first reported in 1983 as a treatment for tracheoesophageal
fistula [5]. Since then, there have been many case reports of successful fistula closures by
using glue [1]
[6]
[7]
[8]
[9]
[10]. In many of these reports, n-butyl cyanoacrylate polymer or 2-octyl cyanoacrylate
polymer was used via endoscopic injection [5]
[6]
[7]. We propose that alpha-cyanoacrylate (α-CA) monomer is ideal for glue embolization
against intractable esophageal fistulas because α-CA monomer glue spreads faster and
thus creates a bond more rapidly than the other polymer glues [1]
[10]
[11]. We previously demonstrated 4 case series with esophageal fistulas that were successfully
treated with endoscopic injection of α-CA monomer [12]. Excluding our preliminary report, only 1 case report has so far examined the usefulness
of α-CA monomer for the treatment of esophageal fistulas [8].
This prospective phase II trial aims to test the hypothesis that endoscopic injection
of α-CA monomer is effective and safe in patients with intractable esophageal fistulas.
We planned this study to apply the endoscopic treatment in a clinical setting.
Patients and methods
Eligibility criteria
This study was conducted according to a protocol reviewed and approved by the Wakayama
Medical University Hospital Ethical Committee on Human Research. All patients gave
written informed consent before enrolling in the study. The study protocol was registered
at the University Hospital Medical Information Network (UMIN000018486).
Participants were enrolled in the trial between January 2015 and March 2017 and regarded
as eligible if they met the following criteria: (i) aged between 20 and 85 years;
(ii) invalid cases with more than 1 wk of conservative medical treatment for intractable
esophageal fistulas; (iii) had no severe disturbance of liver or renal functions,
as indicated by aspartate aminotransferase and alanine aminotransferase levels of
< 300 IU/L, and creatinine level of < 3.0 mg/dL. Exclusion criteria were as follows:
(i) patients with esophagotracheal fistulas; (ii) patients with pulmonary fistulas;
(iii) patients with shock vital.
Surgical procedure (video)
Details of the endoscopic treatment methods performed at our institute have been previously
described [12]. In brief, all patients were treated with endoscopic injection of α-CA monomer glue.
Endoscopic treatments were performed while patients were under sedation induced by
intravenous diazepam. Patients were placed in the left decubitus position. We used
a standard endoscope with a single accessory channel (GIF H260; Olympus, Tokyo, Japan).
In the image-guided therapy suite, a mixture of α-CA monomer (Aron Alpha A; Sankyo,
Tokyo, Japan) and oily contrast agent (Lipiodol; Guerbet, Tokyo, Japan) in a ratio
of 0.3 to 1.7 mL was endoscopically injected through the fistula using a dispersion
tube and 2.5-mL glue syringe ([Fig. 1]). We used 50 % glucose to push the solution from the syringe into the dispersion
tube ([Fig. 1]). We repeated this procedure every 1 or 2 wk until the fistula was closed.
Fig. 1 Schematic drawing of the endoscopic injection of cyanoacrylate glue. a Glue syringe: a mixture of cyanoacrylate and oily contrast agent (lipiodol) in a
ratio of 0.3 to 1.7 mL in a 2.5-mL syringe. b Pushing syringe: 5 mL of 50 % glucose in a 10-mL syringe.
Study design
We previously confirmed safety of this endoscopic treatment [12]. In this study, we adopted a phase II study design to evaluate the response rate.
The primary endpoint of this study was to assess the objective response rate (fistulas
closure rate). According to endoscopic findings 4 wk after the endoscopic treatment,
we considered the patients with reduction of the esophageal fistula orifice to be
successful cases.
Secondary endpoints were to evaluate treatment-related toxicity and the late complications
rate. Postoperative complications were analyzed according to the Clavien-Dindo classification
[13]; complications higher than grade II were regarded as clinically significant. The
planned sample size of the phase II study was 25 patients, a number that was required
to confirm the null hypothesis that the 95 % confidence interval of the expected overall
response rate (85 %) would be less than 60 % under the condition of α-error of 0.05
and β-error of 0.2.
Results
Patients
Twenty-five invalid patients with more than 1 wk of conservative medical treatment
for intractable esophageal fistulas were enrolled in this study between January 2015
and March 2017. [Table 1] shows the clinical characteristics and surgical outcomes of the 25 patients (male:
22, female: 3), with a median age of 69 years (range: 47 – 83 years). The primary
disease was esophageal cancer in 16 patients (64 %), and the cause of the esophageal
fistula of these 16 patients was anastomotic leakage of cervical esophagus after the
subtotal esophagectomy with esophagogastrostomy. The remaining 9 patients (36 %) had
gastric cancer as primary disease. The cause of esophageal fistula was anastomotic
leakage of abdominal esophagus after total gastrectomy with esophagojejunostomy. All
patients received conservative medical treatment of over a week, such as fasting and
antibiotic administration. In all 16 patients with anastomotic leakage of cervical
esophagus after subtotal esophagectomy, we opened the cervical wound and noted the
presence of fistula. In 7 of the 9 patients with an anastomotic leakage of abdominal
esophagus after total gastrectomy, we performed percutaneous drainage of intra-abdominal abscess.
The size of fistula orifice in the esophagus according to endoscopic findings was
8 mm in a median major axis (range: 3 – 20 mm).
Table 1
Demographic and clinical data of the patients.
|
Case
|
Gender/age
|
Primary disease
|
Primary operation
|
Site of fistula/major axis of fistula orifice (mm)
|
Primary fistula treatment
|
Time from diagnosis of fistula to injection of CA glue (wk)
|
Total number of injections of CA glue/duration of endoscopic therapy
|
Clinical outcomes of endoscopic therapy
|
Reoperation procedure after endoscopic treatment
|
Follow-up (mo)/clinical course
|
|
Failure cases
|
|
1
|
M/70
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/10
|
Open drainage, antibiotic treatment
|
2
|
3 T/9 W
|
Failure
|
Fistulectomy plus muscle flap
|
24/full recovery
|
|
2
|
M/68
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/8
|
Open drainage, antibiotic treatment
|
3
|
6 T/6 W
|
Failure
|
Fistulectomy plus muscle flap
|
9/full recovery
|
|
3
|
M/65
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/12
|
Open drainage, antibiotic treatment
|
3
|
11 T/11 W
|
Failure
|
Fistulectomy plus muscle flap
|
6/full recovery
|
|
Success cases
|
|
4
|
M/64
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/3
|
Open drainage, antibiotic treatment
|
2
|
2 T/5 W
|
Completely sealed
|
None
|
27/full recovery
|
|
5
|
M/47
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/8
|
Open drainage, antibiotic treatment
|
1
|
2 T/6 W
|
Completely sealed
|
None
|
26/full recovery
|
|
6
|
F/67
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/5
|
Open drainage, antibiotic treatment
|
5
|
2 T/6 W
|
Completely sealed
|
None
|
23/full recovery
|
|
7
|
M/76
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/8
|
Open drainage, antibiotic treatment
|
3
|
1 T/4 W
|
Completely sealed
|
None
|
30/full recovery
|
|
8
|
M/54
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/10
|
Open drainage, antibiotic treatment
|
2
|
4 T/5 W
|
Completely sealed
|
None
|
11/cancer death
|
|
9
|
M/53
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/3
|
Open drainage, antibiotic treatment
|
1
|
1 T/5 W
|
Completely sealed
|
None
|
19/full recovery
|
|
10
|
M/75
|
Esophageal cancer
|
Subtota esophagectomy
|
Anastomotic site with cervical esophagus/5
|
Open drainage, antibiotic treatment
|
4
|
14 T/14 W
|
Completely sealed
|
None
|
12/full recovery
|
|
11
|
M/68
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/7
|
Open drainage, antibiotic treatment
|
2
|
2 T/7 W
|
Completely sealed
|
None
|
7/full recovery
|
|
12
|
M/62
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/5
|
Open drainage, antibiotic treatment
|
2
|
3 T/4 W
|
Completely sealed
|
None
|
7/full recovery
|
|
13
|
M/65
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/6
|
Open drainage, antibiotic treatment
|
2
|
12 T/13 W
|
Completely sealed
|
None
|
9/full recovery
|
|
14
|
M/63
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/10
|
Open drainage, antibiotic treatment
|
3
|
4 T/8 W
|
Completely sealed
|
None
|
24/full recovery
|
|
15
|
M/70
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/3
|
Open drainage, antibiotic treatment
|
1
|
7 T/8 W
|
Completely sealed
|
None
|
21/full recovery
|
|
16
|
M/75
|
Esophageal cancer
|
Subtotal esophagectomy
|
Anastomotic site with cervical esophagus/5
|
Open drainage, antibiotic treatment
|
8
|
13 T/12 W
|
Completely sealed
|
None
|
13/cancer death
|
|
17
|
M/74
|
Gastric cancer
|
Totalgastrectomy
|
Anastomotic site with abdominal esophagus/20
|
Drainage, antibiotic treatment
|
3
|
4 T/4 W
|
Completely sealed
|
None
|
30/full recovery
|
|
18
|
M/69
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/15
|
Drainage, antibiotic treatment
|
2
|
3 T/4 W
|
Completely sealed
|
None
|
29/full recovery
|
|
19
|
M/81
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/10
|
Drainage, antibiotic treatment
|
2
|
1 T/4 W
|
Completely sealed
|
None
|
24/full recovery
|
|
20
|
M/71
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/12
|
Drainage, antibiotic treatment
|
3
|
2 T/8 W
|
Completely sealed
|
None
|
19/cancer death
|
|
21
|
M/64
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/14
|
Antibiotic treatment
|
2
|
3 T/8 W
|
Completely sealed
|
None
|
14/full recovery
|
|
22
|
M/83
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/8
|
Drainage, antibiotic treatment
|
3
|
6 T/6 W
|
Completely sealed
|
None
|
13/full recovery
|
|
23
|
F/79
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/4
|
Antibiotic treatment
|
2
|
3 T/4 W
|
Completely sealed
|
None
|
11/full recovery
|
|
24
|
F/77
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/3
|
Drainage, antibiotic treatment
|
2
|
2 T/4 W
|
Completely sealed
|
None
|
10/full recovery
|
|
25
|
M/79
|
Gastric cancer
|
Total gastrectomy
|
Anastomotic site with abdominal esophagus/6
|
Drainage, antibiotic treatment
|
2
|
3 T/6 W
|
Completely sealed
|
None
|
6/full recovery
|
CA: cyanoacrylate; M: male; F: female; T: times; W: weeks.
Endoscopic treatments
All patients were treated with endoscopic injection of α-CA monomer glue. [Fig. 2] shows the pre- and posttreatment endoscopic findings of the fistula with the complete
healing (case 19). The median duration from diagnosis of esophageal fistula to initiation
of the endoscopic treatment was 2 wk (range: 1 – 8 wk). The median number of injections
of α-CA monomer glue was 3 times (range: 1 – 14 times), and the median duration of
endoscopic treatment was 6 wk (range: 4 – 16 wk).
Fig. 2 Endoscopic findings. a Endoscopic appearance of fistula opening. b A view of the fistula opening after cyanoacrylate glue injection. c Fistula closure. Yellow arrows show the fistula site.
Primary and secondary endpoints
After treatment, complete closure of the fistula was verified in 22 patients by endoscope.
Endoscopic injection of α-CA monomer glue was ineffective in 3 patients. The overall
success rate therefore was 88 % (22/25). The 3 patients in which the treatment was
ineffective were all esophageal cancer patients with cervical esophageal fistulas.
They required reoperation of the fistulectomy plus muscle flap under general anesthesia.
During the median follow-up period of 14 mo (range: 6 – 30 mo), 3 patients died of
cancer, but the remaining patients completely recovered without recurrence of esophageal
fistula.
In both early and late phases, there were no complications higher than Clavien-Dindo
grade II in patients who received endoscopic treatments with injection of α-CA monomer
glue.
Discussion
This is the first prospective study of endoscopic treatment of an esophageal fistula
using α-CA monomer injection. In our study of 25 patients, complete closure of the
esophageal fistula was possible in 22 patients after endoscopic injection of α-CA
monomer; the overall success rate was 88 %. In addition, there were no complications
associated with this endoscopic treatment and fistula recurrence. This phase II study
met its primary endpoint of high fistula closure rate and its secondary endpoint of
low complication rate. Therefore, we propose that endoscopic injection of α-CA monomer
is a feasible and safe procedure for the treatment of nonhealing esophageal fistulas.
Occurrence of esophageal fistula following esophagectomy is a severe complication
because gastrointestinal contents drain into the thoracic or peritoneal cavities,
leading to intrapleural abscess or peritonitis. The sealing of the fistulas orifice
by endoscopic injection of superglue, which is exemplified by α-CA monomer, therefore,
is an important step. In addition, α-CA monomer has the advantages of having both
stronger adhesive properties and stronger antibacterial properties compared with other
polymer glues, such as n-butyl 2-cyanoacrylate polymer and 2-octyl cyanoacrylate polymer
[1]
[14]. Therefore, α-CA monomer glue is suitable for application in a wet and infected
cavity. In this study, the treatment success rate of 88 % is one of the best compared
with previous retrospective studies [1]
[10]
[11]
[15]. In 3 case series with 22 [10], 10 [11], and 9 patients [15], respectively, the concomitant presence of fistula with this technique has proven
to be associated with a lower success rate. Notably, n-butyl 2-cyanoacrylate polymer
was used for glue embolization in the studies. We affirm that using α-CA monomer would
be more successful for this technique.
The injection of α-CA monomer often causes endoscopic troubles because of the strong
fast-acting adhesive characteristics associated with the superglue. In general, α-CA
monomer is an acrylic resin that rapidly polymerizes in the presence of a little water.
Because the presence of moisture causes the glue to set, exposure to normal levels
of humidity in the air causes polymerization to start within several seconds. Therefore,
we mixed the α-CA monomer and oily the contrast agent (lipiodol) at a ratio of 0.3
to 1.7 mL in a 2.5-mL syringe. We developed this mixture ratio in terms of efficiency
of the infusion [12]. Lipiodol has the added benefit of allowing radiological confirmation of the injection
and identification of embolization to be performed. If the glue syringe containing
the α-CA monomer and lipiodol is made air-tight, we could endoscopically inject the
solution safely in the image-guided therapy suite. On the other hand, it is generally
accepted that the polymerization of α-CA monomer is delayed in the presence of a large
quantity of water. Therefore, we used a 50 % glucose solution to push the solution
from the syringe into the endoscopic dispersion tube. In our experiments, 50 % glucose
solution was suitable for the pushing of the embolic solution [12]. We succeeded in performing the endoscopic injections of α-CA monomer in all cases
using the above-mentioned devices and methods.
In our study, the endoscopic treatment for esophageal fistulas closure failed in 3
patients (12 %). In these patients, the size of the fistula orifice in the esophagus
was larger than 8 mm of the median (8, 10, and 12 mm, respectively). In addition,
the fistula sites were all located in the flexion of the cervical esophagus. The hold
of endoscopic position during this treatment was therefore difficult, and the infusion
volume of α-CA monomer glue during endoscopic treatment may be insufficient. The strong
pressure exerted on the flexion of the cervical esophagus during swallowing may have
caused the nonhealing fistulas. Furthermore, the systemic inflammatory reactions in
failed patients were high. We postulate that stable patients who have fistulas in
the abdominal esophagus after total gastrectomy with esophagojejunostomy may be managed
by this endoscopic treatment. While the majority of patients with cervical esophageal
fistulas after subtotal esophagectomy with esophagogastrostomy are candidates for
this endoscopic treatment, it has its limitations in certain patients – for example,
patients with fistula larger than 8 mm, patients with fistula that render it difficult
to hold an endoscopic position, and unstable patients with evidence of sepsis. In
these patients, other endoscopic treatments, such as endoscopic clips and stent insertion,
are also inadequate adaptations [1]
[16]. In our series, all 3 patients in which this endoscopic treatment was refractory
were cured by surgical reinterventions with fistulectomy plus muscle flap. However,
a more aggressive surgery, such as esophagectomy with proximal and distal stoma, is
advisable before this endoscopic treatment is planned in unstable patients with a
septic condition.
This phase II study had several limitations. It was conducted in a single institution
without randomized controlled study. Due to the small sample size, findings from this
do not allow established clinical application. Indeed, many patients with esophageal
fistulas can be treated conservatively, and their fistulas may close during follow-up
even when endoscopic intervention was not performed. Therefore, in our 22 success
cases who received endoscopic injection of α-CA monomer, the patients who were available
for conservative treatment and the patients who really needed endoscopic treatment
to make the fistula closure might be mixed. In order to prove the clinical application
of this endoscopic treatment with injection of α-CA monomer, we are going to plan
a large multicenter, randomized, double-blind, placebo-controlled phase III clinical
trial in the future.
In conclusion, endoscopic treatment with injection of α-CA monomer facilitated healing
of post-esophagectomy fistula in 88 % of patients without complications. This suggests
that the treatment is effective and safe for patients with esophageal fistulas.
Video 1 Endoscopic treatment with injection of α-CA monomer for esophageal fistulas.
