The safety and usefulness of endoscopy for evaluation of the graft and anastomosis early after esophagectomy and reconstruction.

Abstract

Although rare, graft ischemia and necrosis after esophagectomy is a devastating complication. The aim of this study was to review our experience with early endoscopy for evaluation of the graft and anastomosis after esophagectomy and reconstruction.
From a population of 479 patients who underwent esophagectomy during the years 1996-2003, we identified 102 patients who had endoscopy within 21 days of operation.
Endoscopy was performed a median of 9 days after operation. Graft ischemia, anastomotic leak, or both were found in 63 of the 102 patients. Reoperation was necessary in 27% of these patients, including graft removal in nine patients. In 39 patients, endoscopy demonstrated a healthy graft; only one of these patients (2.6%) required reoperation. No patient with ischemia judged insufficient to warrant graft removal on initial endoscopy subsequently lost their graft. There were no complications or anastomotic injuries associated with early endoscopy.
Endoscopy early after esophagectomy is safe and provides accurate and reliable identification of graft ischemia that can be used to guide the treatment of these patients.

Full text

Page 1

The safety and usefulness of endoscopy for evaluation of the graft
and anastomosis early after esophagectomy and reconstruction
M. S. Maish,1S. R. DeMeester,2E. Choustoulakis,1J. W. Briel,1J. A. Hagen,1J. H. Peters,1J. C. Lipham,1
C. G. Bremner,1T. R. DeMeester1
1Department of Surgery, University of Southern California Keck School of Medicine, 1510 San Pablo Street, Los Angeles, CA 90033, USA
2Department of Cardiothoracic Surgery, University of Southern California Keck School of Medicine, 1510 San Pablo Street, Los Angeles,
CA 90033, USA
Received: 6 May 2004/Accepted: 9 December 2005/Online publication: 28 July 2005
Abstract
Background: Although rare, graft ischemia and necrosis
after esophagectomy is a devastating complication. The
aim of this study was to review our experience with early
endoscopy for evaluation of the graft and anastomosis
after esophagectomy and reconstruction.
Methods: From a population of 479 patients who
underwent esophagectomy during the years 1996–2003,
we identified 102 patients who had endoscopy within 21
days of operation.
Results: Endoscopy was performed a median of 9 days
after operation. Graft ischemia, anastomotic leak, or
both were found in 63 of the 102 patients. Reoperation
was necessary in 27% of these patients, including graft
removal in nine patients. In 39 patients, endoscopy
demonstrated a healthy graft; only one of these patients
(2.6%) required reoperation. No patient with ischemia
judged insufficient to warrant graft removal on initial
endoscopy subsequently lost their graft. There were no
complications or anastomotic injuries associated with
early endoscopy.
Conclusion: Endoscopy early after esophagectomy is
safe and provides accurate and reliable identification of
graft ischemia that can be used to guide the treatment of
these patients.
Key words: Esophagus — Esophagectomy — Endos-
copy — Esophageal reconstruction — Graft ischemia
Few operations rival esophagectomy in complexity or
the potential for perioperative morbidity and mortality.
Many patients are elderly with medical comorbidities,
and respiratory, cardiovascular, and infectious compli-
cations occur commonly. Much of the morbidity of an
esophagectomy is associated with the reconstruction,
because replacing the esophagus requires transposition
of a normally intraabdominal portion of the gastroin-
testinal tract into the chest or neck. The relative ische-
mia of the proximal portion of the graft predisposes
these patients to a high incidence of anastomotic com-
plications after esophagectomy. Less commonly, severe
graft ischemia can lead to transmural necrosis—a life-
threatening complication that may require takedown of
the graft.
Little has been published about the time course and
risk factors for graft ischemia and necrosis after
esophagectomy, and this complication is rare enough
that surgeons who do only an occasional esophagecto-
my may have never encountered it. However, patient
survival in this situation often depends on prompt rec-
ognition and definitive management of the problem.
There is a general reluctance by many surgeons and
gastroenterologists to pass an endoscope through a re-
cent anastomosis. Consequently, a contrast swallow is
typically used to assess the graft and anastomosis after
esophagectomy. Although it is helpful in the elective
setting, use of a contrast swallow to evaluate a patient
with acute clinical deterioration is problematic for sev-
eral reasons. First and foremost, viability of the graft
cannot be evaluated directly but can only be inferred by
the absence of a leak. Another problem is that changes
in mental status often accompany sepsis and clinical
deterioration, and in this setting aspiration of the con-
trast material is a significant concern and could worsen
the clinical situation even further. Lastly, swallow
studies are unreliable in intubated patients, and some
patients may be too ill to be transported safely to radi-
ology for this type of study.
Given these concerns, it has been our strategy to
perform bedside flexible upper endoscopy to evaluate
Presented at the annual meeting of the Society of American Gastro-
intestinal Endoscopic Surgeions (SAGES), Denver, CO, USA,
31 March–4 April 2004
Correspondence to: S. R. DeMeester
Surg Endosc (2005) 19: 1093–1102
DOI: 10.1007/s00464-004-8816-y
? Springer Science+Business Media, Inc. 2005

Page 2

anastomotic integrity as well as the viability of the graft
in patients with signs of clinical deterioration. There-
fore, the aim of this study was to review our experience
with graft ischemia, and to determine the safety and
usefulness of endoscopy for assessing the graft and
anastomosis in patients early in their postoperative
course after esophagectomy and reconstruction.
Methods
Patient population
A retrospective review was conducted from the charts of 479 patients
who underwent esophagectomy with immediate or delayed esophageal
reconstruction at the University of Southern California during the
years 1996–2003. Patients who underwent upper endoscopy within the
first 21 days postoperatively to evaluate their graft or anastomosis
were identified and represent the early endoscopy group. The time
frame of 21 days was selected to maximize the likelihood of capturing
all endoscopies that showed an ischemic graft while minimizing the
inclusion of endoscopies that were done to evaluate an anastomotic
stricture. This study was approved by the institutional review board of
the University of Southern California Keck School of Medicine.
Endoscopy
Upper endoscopy was performed by the attending surgeon at the
bedside in the intensive care unit or, less commonly, in the endoscopy
suite using a standard Olympus 9-mm adult endoscope. Air insuffla-
tion was used as needed to visualize the graft. No special techniques or
precautions were taken during insertion of the endoscope through the
mouth. Because most patients had a nasogastric tube in place, it was
often easiest to follow the nasogastric tube and thereby enter the
esophagus under direct vision. Because most anastomoses were at 21–
24 cm from the incisors, in nearly all cases the endoscope was in the
residual cervical esophagus after insertion, and the anastomosis was
inspected prior to passing the endoscope through it. Particular atten-
tion was paid to the integrity of the anastomosis, the appearance of the
mucosa both around the anastomosis and further distally in the graft,
and the overall appearance of the graft. Typically, when a nasogastric
tube was in place, it was clamped and not removed unless it signifi-
cantly impaired the visualization of an area in question. If the patient?s
respiratory status was marginal, in some circumstances the patient was
electively intubated prior to the endoscopy; otherwise the procedure
was done with light intravenous sedation.
Definition of endoscopic findings
Graft ischemia was defined endoscopically as a gray, blue, or black
appearance to the mucosa, a green or silvery mucous coating that
could not be washed off the mucosa, or a clearly demarcated mucosa
with ulceration (Fig. 1).
Graft necrosis was defined as a gray or yellow-brown, almost
liquefied appearance to the graft (Fig. 2).
Anastomotic leak was defined as endoscopic visualization of a
breakdown of the anastomosis (Fig. 3), evidence during the endoscopy
of air leak into the closed suction drains in the neck, or palpation of
subcutaneous emphysema in the neck or upper chest at the completion
of the procedure.
Video esophagram
Once bowel function returned, patients were sent for a video esopha-
gram to evaluate the anastomosis and graft emptying prior to insti-
tuting oral intake. Typically, the swallow was done 7–10 days after
surgery if the patient was recovering in routine fashion. An irregularity
at the anastomosis consistent with extravasated contrast was consid-
ered to be evidence of a leak.
Clinical parameters and laboratory findings
To assess the relationship of temperature, pulse, cardiac rhythm, and
laboratory values to graft ischemia, we reviewed the hospital records
for the day of and ‡7 days prior to the upper endoscopy to correlate
clinical and laboratory changes with the endoscopic findings.
Statistics
Data are presented as medians. Continuous variables were compared
with the Mann-Whitney test. Categorical variables were compared
using Fisher?s test. Significance was accepted at p < 0.05.
Results
From the 479 patients that underwent esophagectomy
and reconstruction during the years 1996–2003, we
identified 102 patients who had an upper endoscopy
within the first 21 days after surgery (early endoscopy
group). The median age of the early endoscopy group
was 62 years; there were 77 men and 25 women. The
most common indications for the esophagectomy in the
early endoscopy group were esophageal cancer (89%),
Fig. 1. A Mild ischemia evidenced by some mucosal loss and silvery/
gray mucus that will not wash off. B More severe ischemia with areas
of gray and black tissue suggestive of transmural ischemia.
1094

Page 3

benign disease or caustic ingestion (7%), and esophageal
perforation (4%). A diagnosis of cancer was significantly
more common among patients in the early endoscopy
group than in the entire population of patients who
underwent esophagectomy (p = 0.002) (cancer = 359/
479 [75%], benign disease = 120/479 [25%]). The
esophagectomy was done as a transthoracic procedure
in 49 patients, a transhiatal resection in 51 patients, and
a thoracoscopic/laparoscopic esophagectomy in two
patients. In 99 patients the reconstruction was done
concomitantly with the esophagectomy, while three pa-
tients underwent reconstruction as a delayed procedure.
The graft was placed in the posterior mediastinum in 90
patients and substemally in 12 patients. The esophageal
replacement was a gastric pull-up in 75 patients and a
colon interposition in 27 patients. In the entire group,
there were 295 gastric pull-ups (62%) and 184 colon
interpositions (38%). Although a gastric pull-up was the
most common means of reconstruction, patients with
benign disease and those under 70 years of age were
significantly more likely to have a colon interposition
(Table 1).
The median time to the first endoscopy in all 102
patients was 9 days, and the earliest endoscopy was
performed only 2 days after esophagectomy and gastric
pull-up. The most common indication for endoscopy
(60% of patients) was clinical deterioration, defined as
some combination of fever, tachycardia, rising white
blood cell (WBC) count, dyspnea, hypoxia, or acid–base
disturbance on arterial blood gas (Table 2). In the 61
patients for whom clinical deterioration was the indi-
cation for endoscopy, ischemia of the graft or an anas-
tomotic disruption significant enough to explain the
clinical deterioration was present in 36 cases (59%).
However, in 25 patients with clinical deterioration, the
graft was found to be healthy and the anastomosis in-
Fig. 2. The characteristic yellow-brown appearance of transmural
ischemia and developing liquification necrosis of a graft.
Fig. 3. A A small anastomotic leak in an otherwise healthy graft. B A
large anastomotic disruption with adjacent areas of graft ischemia.
Table 1. Patient and operative details in 102 patients who had early
endoscopy after esophageal reconstruction
Stomach
(n = 75)
Colon
(n = 27)p value
Age
< 70 yr
‡ 70 yr
Sex
Female
Male
Indication
Cancer
Benign disease
Type
Transhiatal
Transthoracica
Graft location
Posterior mediastinum
Substernal
Neoadjuvant therapy
Yes
No
Medical comorbiditiesb
Present
Absent
47
29
25
2
0.003
20
55
50.5
22
71
4
20
7
0.007
40
35
11
16
0.4
67
8
23
4
0.7
25
50
80.8
19
44
31
10
17
0.07
aAlso includes two patients who had thoracoscopic/laparoscopic
esophagectomy
bPresence or absence of significant medical comorbidities, including
hypertension, diabetes, coronary artery disease or prior myocardial
infarction, renal failure, or emphysema
1095

Page 4

tact. A defined cause for the clinical deterioration was
subsequently determined in 20 of these patients (80%)
and included pneumonia (45%), wound infection or
abscess (30%), line sepsis (10%), pulmonary embolism
(10%), and liver failure (5%).
Graft ischemia
Upper endoscopy demonstrated some degree of graft
ischemia in 49 patients (48%). Assuming that patients
who did not undergo endoscopy had no significant graft
ischemia, 10.2% of the 479 total grafts were ischemic.
The median time interval between esophageal recon-
struction and the endoscopy showing graft ischemia in
these 49 patients was 8 days. In 37 patients, the anas-
tomosis was intact by endoscopic examination, but some
degree of graft ischemia was present; in 12 patients,
there was both graft ischemia and an anastomotic dis-
ruption. Most commonly, graft ischemia was limited to
the area around the anastomosis (59%), but it extended
distally to involve more of the graft in 41% of patients.
Ischemia was present on endoscopy in 55% of colon
interpositions and 45% of gastric pull-ups (p = 0.4 for
patients with early endoscopy), corresponding to 8.1%
of colon interpositions and 11.5% of gastric pull-ups in
the entire group of 479 patients (p = 0.3 for the entire
group). There was no difference in the prevalence of
ischemia based on the indication for the esophagectomy,
type of esophagectomy, location of the graft, use of
neoadjuvant therapy, or presence of significant medical
comorbidities in the group who had early endoscopy
(Table 3). Of the 49 patients with graft ischemia, 13
(27%) required reoperation. The remaining 36 patients
demonstrated resolution of the graft ischemia on follow-
up endoscopy and did not require reoperation.
Graft loss
A total of nine patients required reoperation and take-
down of their graft for significant ischemia with necro-
sis. This represents a graft loss rate of 8.8% for the early
endoscopy group and 1.9% for the entire population of
479 patients. There was no difference in the loss rate
between the two graft types ([7/75 [9.3%] for stomach
and 2/27 [(7.4%]) for colon, p = 1.0). Likewise, al-
though the numbers are small, there was no difference in
Table 2. Indication for endoscopy and endoscopic findings in 102 patients with early endoscopy after esophageal reconstruction
EGD findings (n = 102)
EGD indicationGraft ischemia (%)Anastomotic leak (%) Both (%)Neither (%)
Clinical deterioration (n = 61)
Neck wound drainage (n = 14)
Delayed graft function (n = 9)
Surgeon concern about graft intraoperatively (n = 2)
Bleeding (n = 3)
Dysphagia (n = 5)
Abnormal swallow findings (n = 8)
28 (46)
1 (7)
1 (11)
1 (50)
3 (100)
3 (60)
0
4 (6.5)
5 (36)
2 (22)
0
0
0
3 (38)
4 (6.5)
5 (36)
0
0
0
0
3 (38)
25 (41)
3 (21)
6 (67)
1 (50)
0
2 (40)
2 (25)
EGD, endscopy
Table 3. Incidence of ischemia in 102 patients who underwent early endoscopy after esophageal reconstruction.
Factorn
Ischemia present
(n = 49)
Ischemia absent
(n = 53)p value
Indication
Cancer
Benign disease
Type
Transhiatal
Transthoracica
Graft location
Posterior mediastinum
Substernal
Neoadjuvant therapy
Yes
No
Medical comorbiditiesb
Present
Absent
91
11
44
5
47
6
1.0
51
51
28
21
23
30
0.2
90
12
44
5
46
7
0.8
33
69
16
33
17
36
1.0
54
48
29
20
25
20
0.24
aAlso includes two patients who had thoracoscopic/laparoscopic esophagectomy
bPresence or absence of significant medical comorbidities, including hypertension, diabetes, coronary artery disease or prior myocardial infarction,
renal failure, or emphysema
1096

Page 5

the loss rate based on the type of esophagectomy, po-
sition of the graft, use of neoadjuvant therapy, or the
presence of significant medical comorbidities. Interest-
ingly, no patient with benign disease as the indication
for their esophagectomy lost his or her graft. The indi-
cation for endoscopy in seven of the nine patients who
required takedown of their graft was clinical deteriora-
tion. The other two patients were not systemically ill but
were noted to have a change in the character of their
drain output. The reoperation occurred on the same day
as the endoscopy in all patients who lost their graft and
was done at a median of 7 days after the initial recon-
struction. The earliest a graft was removed was 2 days
after esophagectomy; the latest was 14 days.
Anastomontic leak
Partial anastomotic disruption was found by endoscopy
in a total of 26 patients (25%) and was associated with
graft ischemia in 12 patients. In the other 14 patients,
there was no endoscopic evidence of ischemia, and only
an isolated anastomotic disruption was found on
endoscopy. The prevalence of an anastomotic leak ten-
ded to be higher in gastric pull-ups (29%) than colon
interpositions (15%), but this difference did not reach
statistical significance (p = 0.2). The majority of pa-
tients (65%) with an anastomotic leak were treated
expectantly without the need for reoperation. Of the 14
patients with a leak but no graft ischemia, nine were
kept nothing per month (NPO) and observed with suc-
cessful resolution of the leak, but five required reoper-
ation for incision and drainage of the neck wound
(n = 3) or drainage of a pleural or mediastinal abscess
(n = 2). None of the patients in this group lost their
graft. In contrast, of the 12 patients with both graft
ischemia and anastomotic leak, two patients required
takedown of the graft and two others required incision
and drainage of the neck wound. The remaining eight
patients were observed, and in each case the ischemia
resolved and the anastomosis healed.
Airway fistulas
Three patients developed fistulas to their airway after
esophageal reconstruction. All three patients had a gas-
tric pull-up placed in the posterior mediastinum that was
ischemic on endoscopy, and each underwent reopera-
tion. In one patient with gastric necrosis the graft was
taken down. The fistula in the second patient was con-
trolled with a tracheal stent. In the third patient, the
fistula was closed with a pedicled latissimus muscle flap.
All three survived and were discharged from the hospital.
Comparison with contrast video esophagram
There were 72 contrast video swallows performed in the
group of 102 patients with early endoscopy at a median
of 10 days after surgery. In 53 patients, the video
swallow and endoscopic findings concurred, whereas in
19 patients the findings differed (Fig. 4). In those where
the findings concurred, there was a similar distribution
of patients with the video preceding the endoscopy and
vicsa versa. In the 19 patients with discordant findings,
11 patients were found to have a leak on video swallow
that had not been seen on endoscopy done a median of 5
days prior to the video swallow. In eight others with
discordant findings, the endoscopy showed a leak that
had not been seen on video swallow, which was done in
five patients at a median of 7 days prior to the endos-
copy and in three patients at a median of 30 days after
the endoscopy.
Clinical and laboratory indicators of graft ischemia
In an effort to identify clinical parameters that could
indicate the presence of an ischemic graft, we assessed
the daily maximum temperature, pulse, cardiac rhythm,
white blood cell (WBC) count, and change in base ex-
cess on arterial blood gas in patients with ischemia
necessitating graft take down, patients with ischemia
that resolved without reoperation, and patients with a
healthy graft and an intact anastomosis (Figs. 5–8).
Although none of the parameters were statistically sig-
nificant between groups, patients with ischemia that led
to graft loss tended to have more tachycardia and atrial
arrhythmias, were more likely to have a fever and ele-
vated WBC count, and were the only ones to demon-
strate a change from a positive to a negative base excess
on arterial blood gas. However, none of these parame-
ters could be used to reliably distinguish one group from
another.
Fig. 4. Findings for of the 72 patients who had a contrast video
swallow. In 53 studies, the findings at endoscopy and on video swallow
concurred, whereas in 19 cases the findings were discordant. Among
the discordant studies, there were 11 patients who had a video study
showing an anastomotic leak a median of 5 days after endoscopy did
not detect an anastomotic leak; however, in more than half of these
patients, ischemia was seen on the endoscopy, and it may have sub-
sequently led to the leak seen on contrast swallow. There were also five
patients who had a normal video swallow, but on endoscopy a median
of 7 days later, four patients were found to have a leak and one patient
had an ischemic graft. The studies are complementary for detecting
anastomotic leaks, but only endoscopy shows graft ischemia.
1097