Background and aims
The current practice of most thoracic surgeons who discover that a patient has an
esophageal perforation involves a number of options. These include primary surgical
repair (when possible) reinforced with a vascularized, pedicle-type muscle flap, nutritional
support, and drainage. Alternative strategies include endoscopic therapies such as
clipping, stenting or a combination thereof, T tube drainage and, in certain cases,
esophageal diversion, with or without esophageal resection as appropriate. Determining
the optimal therapy for our patients requires examining all available options, including
surgery alone, endoscopic therapy alone or a hybrid combination of the two.
Mortality from esophageal perforation is high and attempts to reduce this are important.
One proposed method is to use endoluminal approaches for the treatment of these perforations,
in combination with adequate nutrition and appropriate drainage. Recent reviews and
meta-analyses examining treatment of esophageal perforations have demonstrated that
mortality rates may be lower when stents are used, although there is likely bias in
the selection of these patients [1]
[2]. The use of endoluminal therapies includes endoluminal suturing (OverStitch, Apollo
Endosurgery Inc., Austin, TX, United States), through-the-scope (TTS) clips, and over-the-scope
clips (OTSC, Ovesco Inc., Tubingen, Germany). Success with regard to the use of these
endoscopic devices has been reported in the context of treating gastrointestinal perforations
[3]
[4]
[5]
[6]
[7]. Endoscopic suturing of a stent in the esophagus when there is no stricture present,
such as in the context of a spontaneous perforation, has been reported to successfully
anchor the stent and minimize the risk of migration [7]. With the availability and use of endoluminal therapies [8], it is important to assess their utility in comparison to surgical intervention
for this acute condition. Any benefit of a less invasive endoscopic approach in terms
of leak healing, need for “salvage” intervention, and mortality would be important
for appropriate patient selection.
The aim of this study was to evaluate outcomes for patients with Boerhaave esophageal
perforation who were treated surgically and with endoscopic therapy. Our aim was to
determine the utility of endoscopic therapy in the management of these patients in
order to guide future treatment algorithms.
Patients and methods
We identified all adult patients (> 18 years) who were diagnosed with a spontaneous
esophageal perforation over a 10-year period (1 October 2004 – 1 October 2014). We
excluded all patients who had been diagnosed and treated elsewhere before transfer
to our institution. Those patients who did not consent to their data being used in
the study were also excluded. Spontaneous perforations (Boerhaave perforation) were
defined as perforation of a normal esophagus without previous instrumentation, surgery,
or trauma. Data were collected from prospectively maintained databases. These data
included: patient demographics, comorbidities, the intended treatment (conservative,
endoscopic, surgical or a combination thereof), need for “salvage” intervention (defined
as intervention intended to heal the perforation or treat initial sepsis), persistence
of leak at discharge, discharge location, alimental status at discharge, discharge
on antibiotics or drains in situ, readmission to our institution within 30 days, re-intervention
within 30 days, survival at 30 days, and long-term follow-up.
For those patients who underwent endoscopic therapy, the modalities used were through-the-scope
and over-the-scope clips, endoscopic suturing (using the Apollo device), and esophageal
stenting. To enable endoscopic suturing, an overtube was placed to avoid esophageal
injury. A dual-channel endoscope was inserted through this to the site of the perforation.
Using the suturing device, suitable defects were closed in either an interrupted or
continuous fashion with 2 – 0 polypropylene or 2 – 0 polydioxanone sutures. A tissue
helix was used at times to grip the esophageal mucosa and facilitate suturing.
Results
A total of 17 patients with spontaneous esophageal perforation were identified. Sixteen
patients consented to their data being used in the study. The median age at perforation
was 68 years (range 27 – 88 years). Thirteen patients (81 %) underwent intervention
for their perforation within 24 hours of their diagnosis. The most common investigation
used to diagnose perforation was CT (63 % patients), followed by upper gastrointestinal
endoscopy (25 %) and upper gastrointestinal swallow study (12 %). All patients ultimately
underwent upper gastrointestinal endoscopy to confirm the diagnosis even if the diagnosis
was made with another modality.
Endoscopic therapy was the primary intervention in eight patients (50 %) whereas surgery
was the primary treatment modality in eight patients (50 %). Patients were offered
primary treatment with endoluminal therapy from July 2010 onwards.
Of the eight patients who underwent endoluminal intervention, the median size of perforation
was 1 cm (range 0.8 – 3 cm). Two patients were stented with palliative intent and
were septic and shocked at the time of presentation. The remaining six patients had
no signs of sepsis. Two patients with small perforations were treated successfully
with through-the-scope or over-the-scope clips alone (Ovesco: 0.8 cm perforation,
Resolution clip: 1 cm perforation). Six patients had stent insertion to treat their
esophageal perforation, median size 1 cm (range 1 – 3 cm). Of these six patients,
two underwent attempted closure of the perforation using the Apollo Overstitch device
(one defect completely closed, the other failed). One patient had the defect partially
closed with an Ovesco clip, then required endosuturing to complete the closure before
the stent was finally deployed. Another patient had the defect partially closed with
through-the-scope clips before stent deployment. All of the stented patients underwent
endoluminal stent fixation: 3/6 were fixed with through-the-scope clips and 3/6 with
endosutures. The clinical progress of patients is charted in [Fig.1].
Fig. 1 Flowchart detailing the treatment options and outcomes for patients with esophageal
perforations observed in our cohort.
“Salvage” intervention was required for 2/8 endotherapy and 1/8 surgical patients
([Fig. 1]). Salvage intervention involved endoluminal therapy in all three patients ([Table 1]). Median lengths of stay (LOS) were 26.5 days (range 7 – 80) and 9 days (range 5 – 42)
for the surgery and endotherapy groups, respectively (P = 0.04). We excluded from this analysis the two patients in the endotherapy group
who died in hospital. Five patients in the surgery group (63 %) and three patients
in the endotherapy group (50 %) were discharged home. One patient in each group was
discharged systemically well but with a persistent leak. Three patients in the surgery
group (3/8; 38 %) and two patients in the endotherapy group (2/6; 33 %) were discharged
home with drains. One patient in the surgery group (13 %) and three patients in the
endotherapy group (50 %) were discharged on antibiotics. Six patients in the surgery
group (75 %) and four patients in the endotherapy group (67 %) were discharged with
enteral nutrition. More patients in the endotherapy group were discharged NPO (nothing
by mouth) (3/8 surgery, 38 % v 4/6 endotherapy, 67 %). Two patients in the stented
group died in hospital. These patients were considered unfit for surgery before any
intervention. Readmission within 30 days occurred in 3/6 and in 0/8 patients in the
endotherapy and surgical groups, respectively. Reasons for readmission included blocked
feeding tube, dysphagia after resuming oral intake, and dislodged chest tube and feeding
jejunostomy tube in association with sepsis. One diabetic patient in the surgery group
with Boerhaave syndrome related to hyperemesis gravidarum required multiple stricture
dilatations for dysphagia. She subsequently developed a fistula and paraesophageal
abscess. Her initial surgery was a left thoracotomy, repair of esophageal perforation
and pleural flap.
Table 1
Salvage endotherapy for failed earlier intervention.
|
Patient
|
Initial intervention
|
Salvage intervention
|
Outcome
|
|
71-year-old man
|
Esophageal perforation closed with both OTSC and suturing with Apollo endosuturing
device.
A 12.5-cm × 23 mm FCSEMS was also placed.
|
Day 12: Left VATS decortication and chest drain insertion for empyema.
Day 28: Overlapping stents (12.5 cm × 23 mm).
Day 65: Stents removed. Persistent leak. A 15.5 cm × 23 mm PCSEMS was placed.
|
Day 91: Stent in stent removal of stents. No leak.
|
|
66-year-old woman
|
Esophageal perforation closed with TTS clips. A 12 cm × 23 mm PCSEMS was also placed
and TTS clips used to anchor the stent.
|
Day 78: Stent removed – persistent fistula.
Suture closure of the fistula was performed and a 10.3 cm × 23 mm PCSEMS placed.
|
Day 111: Stent within a stent (FCSEMS).
Day 129: Stents removed.
Persistent fistula-thermal preparation (APC) with suture closure.
Subsequent stricture – dilatations required.
|
|
75-year-old woman
|
Left thoracotomy, repair of esophageal perforation and intercostal muscle flap.
Day 13: Feeding jejunostomy.
|
Day 20: Persistent leak.
Endoscopic assessment of the leak and closure using TTS clips and a 12.3 cm × 18 mm
FCSEMS. The stent was sutured to anchor it in place.
|
Day 76: Stent removed, no leak.
|
APC, argon plasma coagulation; OTSC, over-the-scope clip; TTS, through-the-scope;
VATS, video assisted thoracoscopic surgery; FCSEMS, fully covered self-expanding metal
stent; PCSEMS, partially covered self-expanding metal stent.
Discussion
Spontaneous esophageal perforation is a rare but life threatening pathology. We have
identified 17 patients with this clinical problem treated at a tertiary care center
over a 10-year period which correlates with other series in comparable centers [10]
[11]
[12]. Minimally invasive surgical intervention is increasing, as is endotherapy for the
treatment of many gastrointestinal pathologies, including esophageal perforation and
esophageal anastomotic leak. Endotherapy for spontaneous esophageal perforation and
subsequent leaks includes management with through-the-scope and over-the-scope clips,
sutures or stents [13]
[14]
[15] ([Fig. 2 a], [Fig. 2 b], [Fig. 2 c], [Fig. 2 d], [Fig. 2 e]). In this study, size of perforation and the absence of sepsis determined the success
of clips or stents (with endoluminal closure of the perforation) as primary therapy.
Fig. 2 a Contrast esophagram demonstrating a Boerhaave esophageal perforation.
Fig. 2 b Combined radiological and endoscopic assessment of the perforation.
Fig. 2 c Endoscopic and radiological guidance to position the Ovesco over-the-scope clip to
close the perforation.
Fig. 2 d Contrast esophagram demonstrating no further esophageal leak after placement of the
Ovesco clip.
Fig. 2 e Endoscopic appearance of the sealed esophageal perforation using an Ovesco clip.
The principles of management of the esophageal perforation, however, have not changed.
The need for sepsis control, repair of the perforation, feeding via a jejunostomy
tube, and venting of gastric contents remains. When endotherapy is used to seal the
perforation, stenting across the gastroesophageal junction does increase the risk
of a life threatening aspiration. The availability of stents with one-way anti-reflux
valves (EndoMAXX EVT, Merit, Salt Lake City, Utah, United States) has the potential
to eliminate this specific adverse event. Until these are in widespread clinical use,
jejunostomy feeding and gastric decompression are recommended to avoid this.
Patients with spontaneous esophageal perforation exhibit a spectrum of severity. Patients
with significant pleural and mediastinal contamination are at high risk for sepsis
and systemic inflammatory response syndrome (SIRS). Accordingly, based on this experience,
we suggest the following endoscopic approach: for patients presenting early (< 24
hours) with a defect of ≤ 1 cm, closure with a through-the-scope or over-the-scope
clip or suture. In the presence of extravasation of contrast from the esophagus or
pleural effusion, empyema or mediastinal contamination, radiological or surgical (video-assisted
thoracoscopic surgery (VATS) or thoracotomy) adjunctive drainage is performed. For
those patients with a large defect ≥ 1 cm, primary endoluminal closure can be attempted
combined with stent placement, with pleural drainage as needed. Our experience identified
patients with large perforations as having a need for eventual decortication. Surgery,
thoracoscopic or open, should include drainage of sepsis, repair of the defect, and
a muscle flap to cover this repair, in addition to a feeding tube (jejunostomy) and
venting gastrostomy. For patients who have lost more than 50 % of the esophageal diameter
due to a more severe tear, delay in treatment or inadequate drainage, esophageal diversion
or exclusion become major management considerations.
Salvage intervention for patients in this study always involved endoluminal therapy.
One patient with a persistent leak after surgery required stenting and two stented
patients required re-stenting. Stenting is preferable to clipping or suturing in this
situation as the tissue is suboptimal for these closure methods. In the absence of
sepsis in those treated primarily with early endoluminal therapy, further endoluminal
therapy is desirable for leak control and minimization of sepsis risk. In post-surgical
patients, re-intervention can be challenging, which makes an endoluminal option appealing
as long as sepsis is well controlled.
CT was the most common diagnostic tool used to detect a leak. This is favored over
a contrast swallow because of the additional information that can be provided, e. g.
extent of any collections and presence of mediastinal contamination. All patients
underwent subsequent upper gastrointestinal endoscopy after diagnosis (if this was
not the diagnostic modality). This is essential in order to define the anatomy of
the esophageal perforation and guide treatment.
Median length of stay was lower in the endotherapy group, while readmission was higher
at 50 %. Patients who had the least contamination and physiological insult were those
selected for primary endotherapy. Patients undergoing endotherapy also spent a greater
proportion of their recovery out of hospital which may account for the increased readmission
rates in this group. Of those patients who were readmitted after endotherapy, only
one patient stayed in hospital for more than 2 days. Our surgical in-hospital and
30-day mortality rate in patients treated with curative intent was 0 %. In the endotherapy
group, two patients died, but those patients were treated with palliative intent.
The median duration that stents were in place was 59.5 days (range 23 – 76). During
this time, two patients required multiple stents to be removed and replaced. Suturing
for stent fixation has been supported in the literature and applied to our patients.
Use of the current suturing device for this application in order to reduce stent migration
is not formally approved in the FDA clearance. Although our small series showed no
esophago-aortic fistulas, after review of the published literature [16]
[17]
[18]
[19]
[20]
[21], there is a well-recognized risk for this complication. Improved stent management
may be able to reduce this risk and will require more formal study which may examine
stent sizes, length of dwell time, stent free periods, and stent exchange.
With a paucity of Level 1 evidence, the expert consensus opinion from both our thoracic
surgery division and the gastroenterology division favors a shorter duration of stenting
and, if the leak is still present at the time of removal, use of a new stent that
is of different length, size, or landing zone is encouraged. Aortic erosions do occur,
and most appear to occur at the flare of the stent or at the location of perforation
where the stent is in direct contact with the wall of the vessel. It is for this reason
that we also recommend a muscle flap to interpose when a leak directly abuts the aorta
or tracheobronchial tree.
In conclusion, our 10-year experience demonstrates that endoscopic repair of Boerhaave
perforations can be useful in carefully selected patients without evidence of systemic
sepsis. We have also found that endoscopic therapy such as stenting is particularly
valuable as a “salvage” intervention regardless of whether the primary intervention
was endoscopic or surgical. The benefits of endoscopic therapy and esophageal preservation
may be offset against an increased risk of readmission in those patients primarily
treated endoscopically. Our experience has shown that it is possible to manage this
life-threatening pathology with surgery and endotherapy with low mortality.
