Introduction
Endoscopic submucosal dissection (ESD) has become the standard endoscopic resection
modality for superficial esophageal squamous cell carcinoma (SCC) and selected cases
of superficial neoplasia in Barrett’s esophagus (BE) [11 ]. Although ESD allows en bloc resection of superficial lesions without size limit,
the main concern with esophageal resection is the risk of secondary stricture and
its clinical consequences.
Without preventive measures, extensive esophageal ESD is associated with high secondary
stricture rates, particularly as circumferential resection extent is critical in terms
of risk stratification. Previous studies have reported post-ESD esophageal stricture
rates, without preventive measures, of up to 30%–50%, 50%–80%, and 90%–100% for resections
involving more than 50%, 75%, and 90% of the esophageal circumference, respectively
[22 ]
[33 ]
[44 ].
Steroid-based preventive therapy (either by local injection or oral regimen) has been
demonstrated to reduce post-ESD stricture rates to 10%–20%, 30%–40%, and 60%–80% for
circumferential resections of more than 50%, 75%, and 90%, respectively [55 ]
[66 ]
[77 ]
[88 ]. Oral corticosteroids are often administered in Western countries, with tapering
doses for 7–8 weeks. Systemic steroid treatment requires patient compliance and is
associated with well-established systemic side effects in cases of prolonged use,
and cases of fungal infection associated with death have been described in this specific
indication [99 ]
[1010 ]. Moreover, systemic steroid treatment can impair wound healing and therefore compromise
surgical suture healing for possible future complementary surgery [1111 ].
In Eastern countries, a single round of local triamcinolone acetonide injections at
the end of the procedure has been demonstrated to prevent post-ESD esophageal stricture
[1111 ]
[1212 ]
[1313 ]
[1414 ], except in cases of >90% circumferential resection [1515 ]. Oral corticosteroids are therefore used in addition to triamcinolone acetonide
injection for circumferential resections of 90%–100% [1616 ]. The advantages of the local injection protocol include its quick administration,
lack of reliance on patient compliance, and avoidance of systemic side effects. Clinical
adoption of this local injection protocol in Western countries is limited by the availability
of study data that demonstrate its effectiveness in Western patients.
The aim of this study was to assess the efficacy of a simple systematic corticosteroid
stricture prevention protocol after esophageal ESD, with a stratified approach based
on the extent of circumferential resection, with a combination of topical corticosteroid
injections and oral administration for high risk patients.
Methods
Patients
Prospectively collected data from all consecutive patients who underwent esophageal
ESD at H.U.B. Erasme Hospital (Brussels, Belgium) from January 2016 to October 2022
were retrospectively reviewed. From January 2016, a systematic stricture prevention
protocol was applied at the end of esophageal ESD procedures. Follow-up data were
collected from the patients’ medical files and by contacting the referring centers.
Clinical data, plus the endoscopy report and pictures were reviewed to assess the
stricture. Data from the end of clinical and endoscopic follow-up were collected in
April 2023.
The study was approved by the Ethics Committee of H.U.B. Erasme Hospital (P2022/153).
ESD procedure
The ESD procedures were performed with the patients under general anesthesia and with
endotracheal intubation. Each procedure was performed using a gastroscope (GIF-HQ190
or GIF-1500Z, Olympus, Japan) with a transparent cap (Olympus) attached to the endoscope
tip. All procedures were performed under carbon dioxide insufflation. The neoplastic
lesion was lifted using a 25G needle with a commercially available solution (glycerol
10%, fructose 5% in saline) mixed with drops of indigo blue dye. An electrosurgical
ESD knife (1.5-mm tip; DualKnife or DualKnife J, Olympus, Japan in 97% of cases) was
connected to an electrosurgical unit (VIO 300 D/VIO 3, ERBE Elektromedizin GmbH, Germany).
Mucosal incision was performed using Dry Cut E2 30 W (VIO 300 D) or E3.5 mode (VIO
3) and submucosal dissection using Swift Coag E4 30 W (VIO 300 D) or E3.5 mode (VIO
3). Periprocedural bleeding was treated with the closed tip of the knife or with hemostatic
forceps (Coagrasper, Olympus, using soft coagulation E5 50 W [VIO 300 D] or E5.5 [VIO
3]; or bipolar Hemostat WideCup, Pentax, Japan, using bipolar soft coagulation E2.6)
when more severe. The plane of dissection was close to the muscle layer in all cases.
The percentage circumferential mucosal defect and the length of mucosal defect were
evaluated by the endoscopist at the end of the resection based on a visual appreciation.
The specimen size was measured after being pinned onto cork before histologic evaluation
by expert pathologists.
Systematic protocol for stricture prevention
The extent of the mucosal defect was classified into three groups: C1, circumferential
resection of <50%; C2, resection between 50% and 89%; C3, circumferential resection
of ≥90%. Triamcinolone acetonide (Kenacort, Bristol-Myers Squibb, Japan) was systematically
injected for mucosal defects ≥50% of the esophageal circumference. For mucosal defects
of ≥90% of the circumference, triamcinolone acetonide was injected but this was also
followed with an oral methylprednisolone regimen, with a tapering dose for 7 weeks.
The injection of triamcinolone acetonide (50 mg) was performed in a single session
at the end of the procedure using a 25G needle with multiple injections into the mucosal
edges and into approximately every square centimeter along the small residual submucosal
layer of the entire resection bed, without any injection through the muscle layer
([Fig. 1Fig. 1 ]; [Video 1Video 1 ]). For defects involving ≥90% of the esophageal circumference, an oral methylprednisolone
regimen (an oral prednisolone-equivalent treatment that is available in Belgium) was
systematically prescribed as follows (according to the equivalent prednisolone doses):
24 mg for 15 days, 20 mg for 15 days, 16 mg for 7 days, 12 mg for 7 days, and 8 mg
for 7 days.
Fig. 1
Fig. 1 Endoscopic images showing: a a mucosal defect after large esophageal endoscopic submucosal dissection (ESD); b triamcinolone injection being performed after large esophageal ESD; c the resected sample; d mucosal healing post-ESD at follow-up endoscopy.
Local triamcinolone injection (along with oral steroids in cases with ≥90% circumferential
defect) is given to prevent post-endoscopic submucosal dissection esophageal stricture.Video
1Video 1
Post-ESD management
All patients were kept fasted after the ESD procedure until clinical evaluation by
the operator. A clear liquid diet was permitted from 6 hours after ESD in the absence
of chest/abdominal pain, dysphagia, abdominal distension, fever, or any other concerning
clinical symptoms or signs. Soft diet was introduced gradually from the following
day. Proton pump inhibitors were administered intravenously (80 mg twice daily [bid])
for 24 hours following the ESD and orally (40 mg bid) at discharge for 1 month.
Patients underwent systematic follow-up with an initial appointment within 4–6 weeks
after endoscopic resection to exclude the presence of symptoms of esophageal stricture.
If the patient had symptomatic esophageal stricture, an esophagogastroduodenoscopy
(EGD) was performed to prepare for an endoscopic balloon dilation (EBD) session. EBD
was performed under fluoroscopic guidance using a CRE Balloon Dilation Catheter (10–18
mm, Boston Scientific Corp., USA), with a systematic 2-week interval before reaching
18 mm, and then adapted to clinical response. In asymptomatic patients, systematic
endoscopic follow-up was scheduled according to the recommendations of the most recent
guidelines, with an initial EGD at 3–6 months after the ESD [11 ].
Patients treated for Barrett’s neoplasia were scheduled for a restaging EGD at 3 months
for ulcer healing assessment and new biopsy sampling before eventual further radiofrequency
ablation (RFA) treatment.
Outcomes and definitions
The primary outcome of this study was the incidence of symptomatic stricture at 3
months from the ESD. Stricture was defined as any narrowing of the esophageal lumen
at the level of the ESD scar, in combination with any degree of patient dysphagia
which therefore required EBD.
The secondary outcomes were the cumulative stricture rate assessed by endoscopy within
6 months of ESD, the mean number of EBD sessions, the rate of refractory stenosis,
the safety of steroid administration, and the end of stricture follow-up rate.
The cumulative stricture rate assessed by endoscopy within 6 months of ESD was defined
as any endoscopic feature of stricture that occurred in the patient between the time
of the ESD procedure up until 6 months later at endoscopic follow-up. Given the lack
of a universally used standard definition for “refractory esophageal stricture” in
the literature, we proposed a simplified definition as follows: refractory stenosis
was defined as the failure to achieve 18 mm in three EBD sessions [1717 ]. In these specific cases, patients were evaluated for steroid injection, stricturoplasty,
and/or stenting. The end of stricture follow-up rate was defined as any symptomatic
stricture treated with EBD between ESD and the end of the patient’s follow-up.
To avoid confusing factors in stricture rate assessment, patients were censored at
the time of further RFA therapy for BE, further endoscopic resection, or further esophageal
surgery.
Any adverse event (AE) after local injection of triamcinolone was noted, with specific
attention paid to findings of healing delay, ulcer, mediastinal abscess, delayed perforation,
fever, systemic complications, and worsening of diabetes.
Potential factors associated with stricture development were also evaluated with univariate
and multivariate analysis.
Histologic definitions
R0 resection was defined as en bloc tumor resection with free vertical and horizontal
margins, described as follows:
free horizontal margin (HM0), the absence of high grade dysplasia (HGD)/carcinoma
within the lateral margin
free vertical margin (VM0), the absence of neoplasia within the deep margin.
The risk stratification for the resection was defined, according to the latest ESGE
guidelines [11 ], as follows:
very low risk resection: en bloc R0 resection, pT1a (no more than m2 for SCC), differentiated
lesion, and no lymphovascular invasion
low risk resection: en bloc R0 resection, no lymphovascular invasion, pT1a m3 or pT1b
sm1 (invasion <200 µm from the muscularis mucosae) for SCCs or pT1b sm1 (<500 µm from
the muscularis mucosae) for BE-associated lesions, well-differentiated lesions, and
VM0
local risk resection: piecemeal resection or tumor-positive horizontal margin of a
lesion otherwise meeting very low risk criteria
high risk resection: any lesion with positive vertical margin, lymphovascular invasion,
deep submucosal invasion (>200 µm from the muscularis mucosae for SCCs or >500 µm
from the muscularis mucosae for BE-associated lesions), or poorly differentiated lesions.
Statistical analysis
Statistical analysis was performed using R v1.4.1717.
Discrete variables are shown as counts (percentage) and continuous variables as mean
(SD). Non-normally distributed variables are summarized by the median (interquartile
range [IQR]).
A Cox regression model was used to determine the association between stricture and
patient age and sex, circumferential extent of the mucosal defect, length of the mucosal
defect, esophageal location of the lesion, type of histologic lesion, and history
of chemoradiotherapy. Based on the results of univariate analysis, a Cox regression
model was used to determine the association between circumferential extent of the
mucosal defect, length of the mucosal defect, and post-ESD stricture. Harrell c-index
analysis was used to determine the cutoff value for resection length that was most
discriminating in terms of whether or not a stenosis occurred. The Variance Index
Factor for a multivariate model was used to detect multicollinearity.
The symptomatic stricture rate incidence at 3 months and the cumulative stricture
rate assessed by endoscopy within 6 months of ESD were depicted using the Kaplan–Meier
approach.
Results
During the study period, a total of 111 patients (median age 69 years [range 36–98
years]; 70% men) were treated by a total of 130 esophageal ESD procedures. The indications
comprised 41% SCC, 54% esophageal adenocarcinoma or Barrett’s dysplastic lesions,
and 5% other types of lesions (e.g. submucosal tumor, neuroendocrine tumor).
The median size of the resected specimen was 40 mm (range 10–130 mm), with an en bloc
resection rate of 98% and a complete endoscopic resection rate of 97% ([Table 1Table 1 ]). Specimen histologic analysis showed an R0 rate of 83%, an HM0 rate of 87%, and
a VM0 rate of 96%. Oncologic risk stratification of the ESD specimen revealed a “very
low risk resection” rate of 50%, a “low risk resection” rate of 10%, a “local risk
resection” rate of 6%, and a “high risk resection” rate of 33% [11 ]. The median circumferential extent of the post-ESD mucosal defects was 50% (range
20%–100%). The extent of the mucosal defect was classified into three groups; C1,
circumferential resection <50% (63 ESDs); C2, 50%–89% (59 ESDs); and C3, ≥90% (8 ESDs).
Table 1
Table 1 Baseline characteristics of the 111 patients included in the study and the 130 ESD
procedures they underwent.
Characteristic
n (%), unless otherwise specified
ESD, endoscopic submucosal dissection; SCC, squamous cells carcinoma.
Sex, male
78 (70)
Age, median (range), years
69 (36–98)
Histology
54 (41)
71 (55)
5 (4)
Resection specimen size, median (range), mm
40 (10–130)
En bloc resection rate
127 (98)
Complete endoscopic resection rate
126 (97)
Histologic R0 rate
108 (83)
Circumferential extent of mucosal defect, median (range), %
50 (20–100)
Length of mucosal defect, median (range), mm
40 (15–130)
Extent of mucosal defect
63 (49)
59 (45)
8 (6)
Clinical follow up, median (range), days
337 (31–1912)
Endoscopic follow up, median (range), days
184 (0–1707)
Triamcinolone acetonide local injection alone was performed for 59 ESDs and combination
therapy of triamcinolone acetonide local injection and oral corticosteroids was used
for eight ESD procedures ([Fig. 2Fig. 2 ]).
Fig. 2
Fig. 2 Protocol flowchart.
Among the study population, 11 patients subsequently underwent esophagectomy and were
censored from esophageal stenosis assessment (three patients in group C1, and eight
patients in group C2): none of them underwent extensive ESD before the additional
surgery and none developed stricture symptoms between the ESD procedure and the additional
surgery. A further three patients underwent BE RFA and were also censored from esophageal
stenosis assessment at 3 months.
From ESD to the first endoscopic assessment at 3–6 months, four more patients were
censored from stricture assessment owing to esophageal surgery, seven more for further
RFA ablation treatment, and two more for complementary endoscopic resection. From
the first endoscopic assessment to the end of follow-up outcome assessment, 16 patients
were censored from stricture assessment because of esophageal surgery, RFA ablation
treatment, or complementary endoscopic resection (Fig. 1s , see online-only Supplementary material)
Clinical end-of-study follow-up was available in 110 patients after a median of 337
days (range 31–1912 days). Endoscopic follow-up at 3–6 months was available for 58%
of patients (75/130).
Primary outcome
The symptomatic stricture rate at 3 months was 8.4% (95%CI 3.2%–13.2%) ([Fig. 3Fig. 3 ]). In total, by 3 months from ESD, we had observed post-ESD esophageal strictures
in 12/116 ESDs (10.3%): 0/57 in group C1; 7/51 (13.7%) in group C2; and 5/8 (62.5%)
in group C3 ([Table 2Table 2 ]). Altogether, the stricture rate for >75% circumference resections was 25.8% (8/31).
Fig. 3
Fig. 3 Kaplan–Meier curve of the symptomatic clinical stricture rate at the 3-month assessment.
Table 2
Table 2 Stricture and refractory stricture rates at 3 months after endoscopic submucosal dissection
(ESD).1
Circumferential extent
<50%
50%–89%
≥90%
1 11 patients underwent esophagectomy and three underwent Barrett’s esophagus RFA and
were therefore censored from esophageal stenosis assessment at 3 months.
Number of procedures
57
51
8
Post-ESD esophageal strictures, n (%)
0 (0)
7 (13.7)
5 (62.5)
Refractory strictures, n (%)
0 (0)
0 (0)
3 (37.5)
Secondary outcomes
Symptomatic esophageal strictures were treated by EBD. The median time before the
first dilation session was 43.5 days (range 29–98 days). The mean number of EBD sessions
was 7 (range 2–17), with a mean dilation caliber of 18 mm.
Three patients failed to achieve 18 mm in three EBD sessions and were considered to
have refractory stenosis. They were managed by esophageal stent placement ([Table 2Table 2 ]). One patient needed two successive esophageal stents to be placed and also then
a gastrostomy for recurrent stenosis. For the second patient, three stents were placed
and then removed, with recurrence of dysphagia. He unfortunately died of acute respiratory
failure a few months later. The third patient underwent placement of a stent for 3
months, with resolution of dysphagia after its removal.
We performed univariate and multivariate regression analysis using a Cox regression
model to better determine factors associated with stricture development in the setting
of our systematic prophylactic protocol, including all potential prognostic factors:
patient age, sex, extent of circumferential mucosal defect, length of mucosal defect,
esophageal location of the lesion, histologic type of lesion, and history of chemoradiotherapy.
Circumferential extent and length of the mucosal defect remained independent prognostic
factors: 1.05 (95%CI 1.01–1.08; P = 0.02) and 1.02 (95%CI 1.00–1.05; P = 0.04), respectively ([Table 3Table 3 ]). A cutoff of 60 mm for mucosal defect length was associated with a more than 15-fold
risk of developing a stricture, according to the Harrel c-index analysis, which measures
the discriminatory capacity of a logistic regression model. The Variance Index Factor
for the multivariate model was 1.47, excluding any multicollinearity between these
variables.
Table 3
Table 3 Factors associated with stricture development according to a Cox regression model.
Variable
Univariate analysis
Multivariate analysis
HR (95%CI)
P value
HR (95%CI)
P value
HR, hazard ratio.
Age
0.99 (0.95–1.05)
0.91
Sex
1.98 (0.63–6.25)
0.24
Circumferential extent of mucosal defect, continuous
1.06 (1.03–1.09)
<0.001
1.05 (1.01–1.08)
0.02
Length of mucosal defect, continuous
1.04 (1.03–1.06)
<0.001
1.02 (1.00–1.05)
0.04
Length of mucosal defect ≥60 mm
5.41 (1.19–24.73)
0.03
We did not find any risk factors for esophageal stricture associated with the type
of
lesion. The distribution of cases within the different esophageal locations (upper,
middle,
or lower esophagus) did not allow for useful analysis. Among patients presenting with
a
post-ESD stricture, there was one case in the “lower esophagus” category compared
with 11 in
the other two categories, which jeopardized adequate statistical analysis. In terms
of
history of chemoradiotherapy, none of our patients had received prior or adjuvant
therapy.
The cumulative stricture rate assessed by endoscopy within 6 months of ESD was 10.4%
(95%CI 4.6%–15.9%) ([Fig. 4Fig. 4 ]).
Fig. 4
Fig. 4 Kaplan–Meier curve of the cumulative stricture rate assessed by endoscopy within 6
months of the endoscopic submucosal dissection.
No AEs were observed in patients receiving an injection of triamcinolone. Neither
were any AEs noted during follow-up of the patients who received both triamcinolone
injection and oral corticosteroids.
Discussion
This study demonstrates the efficacy of a systematic prospective protocol to prevent
post-ESD esophageal strictures in a Western tertiary center. Without preventive measures,
stricture rates can reach 70%–100% [1515 ], especially when the extent of the resection is >75% of the esophageal circumference
[33 ]
[44 ]
[1818 ]. Among the currently used preventive methods, local steroid injection and oral steroids
are standard treatments, as they have demonstrated efficacy for the prevention of
strictures and the requirement for EBD [44 ]
[1212 ]
[1313 ]
[1919 ]
[2020 ]. More recently, the triamcinolone acetonide filling method has been developed and
has been demonstrated to be effective in prospective series [2121 ]
[2222 ]. Steroids have an inhibitory effect on inflammation and fibrosis, with suppression
of proliferation of the subepithelial layer fibrous tissue being the hypothesized
mechanism [2323 ]. Long-term administration of systemic steroids is however associated with AEs, such
as infection, worsening of diabetes mellitus, secondary adrenocortical hypoplasia,
and osteoporosis. Furthermore, systemic oral steroid therapy may compromise proper
wound and anastomosis healing in patients who may in future need an esophagectomy
after ESD [2424 ].
To our knowledge, this is one of the first Western studies to propose a systematic
protocol for the prevention of post-ESD esophageal stricture according to the extent
of the submucosal resection. However, the originality of the present study is in proving
that the systematic implementation of a steroid-based local injection protocol is
effective in Western countries, even though multiple factors (genetics, type of resected
cancer [squamous vs. adenocarcinoma], ESD technique, and triamcinolone acetonide injection)
may differ between Eastern and Western countries.
In this study, we found that a single session of local injection of 50 mg triamcinolone
acetonide into the residual submucosal layer and the mucosal edges of the resection
bed directly after an ESD that has created a defect of between 50% and 89% of the
circumference led to a stricture rate of 13.7% (Fig. 2s ). More specifically, for resections creating a circumferential defect of >75%, the
stricture rate was 25.8%, a rate that is similar to those reported in other retrospective
studies [1313 ]
[1515 ]
[2525 ]
[2626 ]. Comparison with these other studies remains a challenge because of differences
in the study protocols (triamcinolone dose, number of injections, timing of injection,
combination or not with oral steroids).
Our protocol was easy to implement in routine practice considering the typically short
in-hospital stay after ESD in Western countries, and without systemic steroid administration.
Furthermore, this protocol has a simple design based on the extent of the resection
field at the end of the ESD procedure ([Fig. 2Fig. 2 ]). Another advantage of this local triamcinolone acetonide injection protocol is
that, for a majority of our cohort, it avoids any misunderstandings with the patient
when prescribing a cumbersome 7 week course of tapered oral steroids. This oral steroid
regimen can sometimes be confusing, mainly when referred patients are managed by multiple
physicians (i.e. referring physician, oncologists, and surgeons), as can be the case
with multidisciplinary management of some complex cases. Moreover, this strategy relies
on a quick technique, which is easy to learn and implement. Further studies with systematic
measurement of the injection time are needed to confirm the additional procedure time.
A major challenge is the substantial risk of stricture after subcircumferential and
circumferential resection, even after a combination of preventive measures [1515 ]. In a retrospective study in 2016, Kadota et al. reported a 71% stricture rate for
extensive resection (14 ESD cases) [2727 ]. In 2018, Iizuka et al. reported a lower stricture rate of 36.5% for such case (11
ESD cases) [2828 ], while the findings of our study fall between these two studies with a stricture
rate of 62.5% (5/8 ESD cases). We know from previous studies that local triamcinolone
acetonide injection is insufficient for extensive resections and this is the reason
why we added the tapering dose of oral steroids in this specific setting [2929 ].
Other methods for post-ESD esophageal stricture prevention have been described, such
as mechanical methods (metal stent, biodegradable stent), antiproliferative methods
(botulinum toxin injection, oral tranilast, mitomycin C injection), regenerative methods
with autologous cell sheet transplantation, and protective methods (carboxymethyl
cellulose sheet, polyglycolic acid sheet alone or in association with triamcinolone)
[1515 ]. However, some of these methods are more technically demanding and should be assessed
in randomized controlled studies before general implementation into routine practice.
EBD is the first-line treatment for post-ESD esophageal stricture. There is no recommended
balloon size or number of dilation sessions for post-ESD stricture [1515 ]. Early preventive balloon dilation is described in several Japanese studies, but
this approach has proven to be inferior to oral steroids for preventing stricture
after esophageal ESD [2020 ]. As previously mentioned, early endoscopy for balloon dilation was not systematically
performed in our study and was reserved only for patients with symptomatic esophageal
stricture. In our study, the mean number of EBD sessions was 6 (range 2–17), with
a mean dilation caliber of 18 mm. We did not observe any complications after EBD sessions.
In subgroup analysis, three patients developed refractory strictures that were treated
with esophageal self-expandable metal stent placement.
We acknowledge several limitations in this study. First, this was a non-comparative
retrospective study in a single center, which therefore limits the level of evidence.
Second, all patients were treated with the same single dose of 50 mg triamcinolone,
regardless of the resection size and length, with the aim of simplifying the protocol
in routine practice; however, there are currently no recommendations for the dose
or number of local injection sessions. This requires further investigation with large-scale
randomized controlled trials.
In conclusion, we validated our systematic stricture prevention protocol using a single
session of triamcinolone acetonide local injections for all >50% circumferential esophageal
ESDs, with added oral steroids for cases with ≥90% circumferential defects, demonstrating
a global low stricture rate, especially for circumferential resection extents <90%.
As a single session treatment without long-term potential systemic side effects, the
triamcinolone injection method appears to be beneficial for patients and could be
considered as a new preventative practice for use in Western countries.
