Introduction
Benign esophageal strictures are a challenging clinical problem and may occur secondary
to gastroesophageal reflux disease, external beam radiation, caustic ingestion, or
surgical anastomosis following esophagectomy [1 ]
[2 ]. A common cause of benign strictures of the esophagus is development of anastomotic
strictures following resection of the distal esophagus and proximal stomach for treatment
of esophageal cancer [3 ]
[4 ]. The mainstay of management of benign esophageal strictures is esophageal dilation.
Several techniques for dilation exist and the most commonly used ones are Savary-Gilliard
bougie dilation (BD) and through-the-scope (TTS) balloon dilation [5 ].
While 80 % to 90 % of patients initially respond to dilation, most patients require
more than one session to reach adequate dilation, and nearly 10 % will experience
a refractory or recurrent stricture despite repeated dilation [6 ]. Perforation is the most serious complication of esophageal dilation, and carries
a mortality rate of approximately 20 % [7 ]. The overall perforation rate has been reported to be 0.1 % to 0.4 %, but rates
as high as 1.9 % have been reported with BD [8 ]. The rate of significant bleeding after dilation, likely the result of mucosal disruption
caused by dilation, has been reported to be 0.4 %. Furthermore, retrosternal pain
is reported in 0 % to 5 % of cases after dilation [9 ].
Adjuncts to management of difficult benign esophageal strictures include steroid injection,
cautery, and topical application of the antiproliferative agent mitomycin C. Most
recently, esophageal stenting techniques have been used in patients with esophageal
strictures. Options for stenting include nonmetal (plastic) expandable stents, biodegradable
stents, and self-expanding metal stents (SEMS) [6 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ]
[15 ]. Over the past 20 years, SEMS have become available in a wide variety of materials
and configurations. Stainless steel and nitinol (an alloy of nickel and titanium)
are the most common materials used in SEMS design, and the wire mesh construction
of the stent can be braided or woven [16 ]. SEMS are also available in uncovered (bare) versions as well as versions that are
partially covered (PC) or fully covered (FC) with silicone membranes. PC and FCSEMS
are designed to reduce tissue ingrowth through the mesh of the stent. As a result,
the FCSEMS appear to allow for long-term removability, thus permitting their use in
both malignant and benign disease [10 ]
[12 ]
[17 ].
Use of FCSEMS for management of postoperative complications following esophagectomy
has not been well studied. However, several studies have included small numbers of
patients with refractory anastomotic esophageal strictures who were treated with FCSEMS
[18 ]
[19 ]
[20 ]
[21 ]. Results from these studies have indicated that while FCSEMS are in place, generally
over a period of a few months, the patients experience a reduction in dysphagia symptoms
associated with anastomotic strictures, but symptoms return shortly after the stent
is removed in all but 20 % to 30 % of patients. In addition, these studies report
frequent complications associated with esophageal stenting, particularly stent migration
and erosion of tissue surrounding the stent. Our study adds to the literature a randomized
comparison of anastomotic esophageal stricture treatment using temporary placement
of a FCSEMS to repeated BD.
Patients and methods
Study design
This was a prospective, multicenter, randomized, controlled, open-label clinical trial.
The study was conducted in the Netherlands and Brazil and was approved by the ethics
committee of the University Medical Center Utrecht, the Netherlands, and thereafter
by the ethics committees of all participating centers. In Brazil, the study was also
approved by the Brazilian regulatory Competent Authority prior to enrollment. All
patients provided written informed consent at the time of enrollment. The study was
conducted in accordance with the latest version of the Declaration of Helsinki. The
trial was sponsored by Boston Scientific Corporation and is registered at www.ClinicalTrials.gov , number NCT01699542. All procedures were performed by endoscopists with extensive
experience with treatment of esophageal diseases, including benign esophageal strictures.
Patients
Patients with a refractory symptomatic esophagogastric anastomotic stricture post-esophagectomy
were enrolled at four centers, three in the Netherlands and one in Brazil. Eligible
patients were 18 years or older and had esophageal anastomotic benign stricture post-esophagectomy
(esophagogastric strictures), which was performed at the same institution where patient
enrollment and follow-up was planned; baseline Ogilvie dysphagia scores of Grade 2
(ability to swallow semi-solid foods), Grade 3 (ability to swallow liquids only),
or Grade 4 (unable to swallow liquids) [22 ]; between two and five prior dilations to at least 16 mm in diameter (first dilation
no more than 6 months and last dilation no more than 24 months post-esophagectomy);
a stricture unable to pass with a standard endoscope of approximately 9.8 mm diameter,
and were willing and able to comply with study procedures and provide written informed
consents. Major exclusion criteria included strictures within 2 cm of the upper esophageal
sphincter, strictures more than 5 cm in length, dysphagia related to a motility disorder,
non-anastomotic esophageal strictures, esophagocolonic strictures, prior esophageal
stent placement, and active erosive esophagitis.
Procedures
Patients were randomized 1:1 to endoscopic BD or temporary placement of a FCSEMS,
the WallFlex Esophageal FC stent (Boston Scientific, Marlboro, Massachusetts, United
States) which is made of nitinol wires braided to form a cylindrical mesh. The stent
contains a flare at both ends to minimize risk of migration. A suture is threaded
just on the proximal end and may be used to remove or reposition the stent. Stent
sizes were 18-mm body diameter and 103-, 123-, or 153-mm length; or 23-mm body diameter
with 105-, 125-, or 155-mm length.
The WallFlex stent was chosen for this study because it is an easy-to-place stent,
even in inexperienced hands. Because it is available in a fully-covered version, it
is amenable for use for benign indications. The stent has a good safety record with
a low risk of adverse events (AEs). Finally, the combination of radial and axial force
and the flexibility of the stent theoretically reduce the risk of stent migration
[23 ].
Patients in the SEMS group were treated with a single FCSEMS at time 0, with endoscopic
imaging confirmation after placement. FCSEMS were removed per protocol at 8 weeks
(± 7 days) post-placement or sooner if complications occurred, at the discretion of
the treating physician and patient. After stent removal, dilation for recurrence of
dysphagia was repeated as needed, and these dilation procedures were recorded until
end of study at 12 months after enrollment. The BD group received dilations up to
at least 16 mm, in one to four dilation sessions, per standard treatment; the date
of the last dilation session was considered time 0, completion of the initial study
treatment. In cases of recurrent dysphagia in the BD group, dilation was repeated
and recorded until 12 months after enrollment.
Follow-up
After initial study treatment, all patients completed an in-person or by-phone visit
on Day 2, Week 2, and Months 1, 2, 3, 6, 9, and 12. In the SEMS group, the Month 2
visit coincided with endoscopic removal of the FCSEMS. At each follow-up visit, information
on AEs was collected, Quality of life (QoL) was evaluated per the EQ-5D-5 L questionnaire,
a validated questionnaire for describing and valuing general health based on five
dimensions including mobility, self-care, usual activities, pain/discomfort, and anxiety/depression
[24 ]. A patient’s reported pain and satisfaction with therapy was assessed using a 10-point
visual analog scale (VAS) and episodes of dysphagia (at baseline, daily for the first
month and weekly thereafter up to 12 months) were evaluated as recorded in the patient’s
diary, and endoscopic and/or fluoroscopic imaging was conducted in the FCSEMS group
at the discretion of the investigator. Each subject was followed for 12 months post-initial
study treatment.
Endpoints
The primary endpoint was the number of dilation procedures for management of dysphagia
within 12 months following initial study treatment. Secondary endpoints included AEs
related to the device and/or procedure; technical stent placement success, defined
as the ability to deploy the stent in satisfactory position across the stricture with
a standard 9.-mm endoscope; technical stent removal success, defined as the ability
to remove the stent without complications; time to recurrence of dysphagia, defined
as inability to pass the stricture; stent migration with or without symptoms; number
of reinterventions within 12 months following the initial study treatment; dysphagia
scores throughout follow-up; patient report of pain and satisfaction with the therapy
(by VAS); and change in QoL.
Sample size calculation
Based on prior experience of the senior investigator (PS), after initial treatment
with a SEMS, approximately half the patients do not require dilation procedures for
management of dysphagia, and approximately half require one to three dilations in
the subsequent 12 months [2 ]. After initial treatment with BD, approximately 30 % of patients need at most one
dilation and approximately 70 % of patients need two to four dilations in the 12-month
period [5 ]. This yields an expected number of dilations of 1.0 with standard deviation (SD)
of 1.15 in the SEMS group, and 2.3 with SD of 1.39 in the BD group. The hypothesis
of the study was superiority of the mean number of dilation procedures in the SEMS
group compared to the BD group, with a margin of 1.30, 80 % power and two-sided alpha
of 0.05. Testing this hypothesis required 20 patients per group. Adding 10 % of patients
to compensate for potential attrition during follow-up yielded the initial 44-patient
enrollment goal. Due to slow enrollment, the study was discontinued after 33 months
when 18 patients had been enrolled and completed the 12-month follow-up.
Randomization
Patients were randomized to a 1:1 ratio between the Stent Arm (Group A) and BD Arm
(Group B). Randomization was stratified by clinical site using random block sizes
of four. Randomization sequence was created using SAS version 9.3 and executed at
the time of initial treatment by study staff at each participating center. After a
patient was enrolled and passed screening, site staff referred to the randomization
sequence to determine treatment assignment. An electronic database was the primary
method to implement randomization with envelopes as a back-up method. At the end of
trial enrollment, a total of 18 patients were randomized, with nine assigned to the
Stent Group and nine to the BD Group.
Statistical analyses
Baseline data (patient demographics and medical history) and post-procedure information
was summarized using descriptive statistics (e. g., mean, standard deviation, number
of data points used) for continuous variables with normal distribution or count variables,
interquartile range for continuous variable with non-normal distribution, and frequency
tables for discrete variables. A negative binomial model was used to compare the primary
endpoint and the number of dilation procedures for management of dysphagia during
12-month follow-up between the BD and SEMS group. Univariate and multivariate analysis
were performed per protocol to assess possible predictors of the primary endpoint.
Factors from the univariate model with P ≤ 0.20 were modeled multivariately using a stepwise procedure in a negative binomial
model. The significance thresholds for entry and exit into the model was set to P < 0.10. Change in overall QoL health score was analyzed using repeated measures analysis.
Results
Participant flow
A total of 22 patients were assessed for eligibility, of whom four were excluded for
not meeting inclusion criteria (3 patients) and investigative site error during enrollment
(1 patient). Eighteen patients were deemed eligible and were subsequently randomized,
as shown in [Fig. 1 ].
Fig. 1 CONSORT diagram of study flow.
Baseline
Baseline (BL) characteristics were similar between the two study groups ([Table 1 ]). Mean age was 67 ± 7 years and 67 % of patients were male. Esophagectomy was transthoracic
in 78 % (14/18) and transhiatal in 22 % (4/18). Mean time from esophagectomy to first
dilation was 95 ± 36 days. Mean stricture length was 0.7 cm (range 0.3 – 1.5 cm),
located at 21.6 ± 2.9 cm from the incisors. The SEMS group had lower baseline QoL
scores (median 65 vs. 80, P = 0.021), but similar dysphagia scores (median 2 vs. 2, P = 0.206) and number of BDs up to 12 months prior to baseline treatment (mean 2 vs.
2.4, P = 0.566).
Table 1
Demographics and baseline information.
BD
SEMS
P value[1 ]
Age
66.6 ± 7.7
66.6 ± 6.3
1.0
Male
66.7 % (6/9)
66.7 % (6/9)
1.0
Type of esophagectomy
0.577
Transthoracic
88.9 % (8/9)
66.7 % (6/9)
Transhiatal
11.1 % (1/9)
33.3 % (3/9)
Max dilation (second dilation) (mm)
15.9 ± 2.0 (9)
16.4 ± 0.9 (9)
0.450
Dysphagia score[2 ]
2.0 (2.0, 2.0)
2.0 (2.0, 3.0)
0.206
Dysphagia score = 2
100 % (9/9)
66.7 % (6/9)
0.206
Dysphagia score > 2
0.0 % (0/9)
33.3 % (3/9)
0.206
Number of pretreatment dilations
2.4 ± 1.0
2.0 ± 0.0
0.566
Quality of life overall health score
80.0 (70.0, 80.0)
65.0 (50.0, 70.0)
0.021
BD, bougie dilation; SEMS, self-expanding metal stents. Treatment group is per randomization.
1
P values calculated from t -test for continuous variable with normal distribution, Wilcoxon test for continuous
variable with non-normal distribution, negative binomial models for count variable,
and Fisher’s exact for binary variable.
2 Median (IQR) presented for continuous variables with non-normal distribution.
Initial treatment
Nine FCSEMS were placed, seven patients received an 18-mm FCSEMS with a length of
103 mm and two patients received a 23-mm FCSEMS with a length of 105 mm. Technical
placement and removal success in the SEMS group was 100 %. SEMS were removed by retrieval
of the suture at the proximal end of the SEMS in 67 % of patients (6/9) and from the
stomach in 33 % of patients (3/9). Two patients required early removal due to symptomatic
stent migration.
Among the nine BD patients, six had a single dilation for initial treatment (16 mm
for 3 patients, 17 mm for 1 patient and 18 mm for 2 patients), two patients had two
dilation sessions (14 mm followed by 16 mm after 15 days in 1 patient and 15 mm followed
by 17 mm after 16 days in 1 patient), and one patient had three dilation sessions
(14 mm followed by 16 mm after 7 days and 18 mm after 8 days).
Treatment of recurrent dysphagia
Post-treatment dilations and reinterventions are listed in [Table 2 ]. Dysphagia was the cause of reintervention in 89 % (23/26) of the BD group and 98 %
(49/50) of the SEMS group. In the BD group, other causes for reintervention are as
followings: one patient had a gastroscopy to check for recurrent stenosis, one patient
had odynophagia and a third patient reported complaints with passing food but a slight
stricture could be passed with a diagnostic scope (approx. 9.8-mm diameter) and hence
this was not qualified as dysphagia caused by the stricture. In the SEMS group, one
patient had a reintervention due to pleuritic carcinomatosa of the right lung and
had a chest tube into the right pleura and pleurodesis. The mean number of post-treatment
dilations for dysphagia was not different between the groups (2.4 for the BD group;
5.4 for the SEMS group; P = 0.159). A Kaplan-Meier analysis of time to first recurrence after initial treatment
for the BD group and after removal for the SEMS group demonstrated no difference between
groups (P = 0.576) ([Fig. 2 ]). Median time to recurrence of dysphagia after treatment was 33 days for the BD
group and 36 days for the SEMS group. The most common reason for recurrent dysphagia
was stricture recurrence. The most common reintervention was BD, 92 % of the interventions
in the BD group and 94 % in the SEMS group. Both groups experienced similar reductions
in frequency of dilation (calculated as number of dilations divided by duration from
first to last dilation) after the initial study treatment (P = 0.931). In the BD group, median frequency of dilation decreased from 2.7 before
study treatment to 0.2 interventions per month after study treatment. In the SEMS
group, the median frequency decreased from 2.9 to 0.3.
Table 2
Post-initial treatment dilations and reinterventions.
BD
SEMS
P value[1 ]
Number of post-dilations due to dysphagia through 12 months
2.4 ± 2.5
5.4 ± 5.4
0.159
Number of post-dilations through 12 months
2.7 ± 2.6
5.4 ± 5.4
0.183
Frequency of dilation per month post-treatment[2 ]
,
[3 ]
0.2 (0.1, 0.3)
0.3 (0.2, 1.1)
0.283
Total number of reinterventions
2.9 ± 2.7
5.6 ± 5.3
0.168
Time to first recurrence of dysphagia[3 ]
33 (21, 33)
36 (24, 71)
0.576
Reintervention due to dysphagia
88.5 % (23/26)
98.0 % (49/50)
0.113
Reason for dysphagia recurrence
Stent migration
0.0 % (0/23)
6.1 % (2/33)
0.507
New stricture formation
4.3 % (1/23)
0.0 % (0/33)
0.411
Stricture recurrence
95.7 % (22/23)
87.9 % (29/33)
0.639
Other
0.0 % (0/23)
6.1 % (2/33)[4 ]
0.507
Type of Reintervention
Study stent removed
0.0 % (0/26)
4.0 % (2/50)
0.544
Bougie dilation
96.2 % (24/26)
94.0 % (47/50)
1.0
Balloon dilation
0.0 % (0/26)
4.0 % (2/50)
0.544
Other
7.7 % (2/26)
20.0 % (10/50)[5 ]
0.202
BD, bougie dilation; SEMS, self-expanding metal stents
1
P values calculated from Wilcoxon test for continuous variable with non-normal distribution,
negative binomial models for count variable, fisher’s exact for binary variable, and
log-rank test for time to event variable.
2 Defined as time 0 to last follow-up or end of study for BD and stent removal to last
follow-up or end of study for SEMS
3 Median (IQR) presented for continuous variables with non-normal distribution and
4 One patient experienced two incidences of recurrent dysphagia; cause unknown
5 Other types of reintervention are: gastroscopy to check for recurrent stenosis (1)
and gastroscopy performed but no stenosis (1) for BD and needle knife (9) and Chest
tube into right pleura and pleurodesis (1) for SEMS.
Fig. 2 Kaplan-Meier analysis of time to first recurrence. The start time of the 12-month
follow-up period for the bougie dilation (BD) group was the same as time 0, which
is defined as the end of the initial treatment period. However, for the stent (SEMS)
group, the start time of follow-up was the time of stent removal rather than stent
placement.
Multivariate analysis
A multivariate analysis identified location of anastomotic stricture, age and treatment
modality – SEMS or BD – as predictors of the number of reinterventions due to dysphagia
([Table 3 ]). Age was not a significant predictor at the 0.05 level when adjusting for other
covariates. Multivariate analysis demonstrated that with each centimeter increase
of distance from the incisors to the stricture, the number of reinterventions for
dysphagia during 12 months after the initial treatment decreased by 32 % (P = 0.006).
Table 3
Multivariate analysis of number of reinterventions due to dysphagia.
Predictor
Rate ratio
P value[1 ]
SEMS vs. BD
2.4
0.053
Location of anastomotic stricture
0.7
0.006
BD, bougie dilation; SEMS, self-expanding metal stent
1 Negative binomial model is used.
Patient-reported outcomes
Pain by VAS in the SEMS group was generally low, with median ranging from 0 at baseline
to a maximum of 3 at Day 2, and a return to low levels (median 0) at Day 14 which
remained low during follow-up. In the BD group, pain levels were consistently low
(median 0) during follow-up. Patient satisfaction with treatment was similar between
the groups (SEMS median ranging from 8 to 9.5; BD median ranging from 8 to 10). The
repeated measures analysis on QoL overall health score change suggests a significant
difference between SEMS group and the BD group (P = 0.043). As seen in [Fig. 3 ], patients in the SEMS group seemed to experience a greater QoL improvement following
initial study treatment compared with the BD group. It is possible that the significant
difference in baseline scores between the two groups may have contributed to the greater
improvement in the SEMS group. In other words, the SEMS group improved more because
there was more room for improvement in this group. Additionally, when interpreting
the QoL improvement, one should consider the under-enrollment and the small sample
size and the similar satisfaction scores between the two groups. It is possible that
patients were equally satisfied with the treatment but SEMS patients experienced higher
improvement in overall QoL due to fewer procedures to manage dysphagia while the stent
was in place. Details of patient-reported outcomes, including VAS pain score and patient
satisfaction score, are shown in [Fig. 4 ] and [Fig. 5 ].
Fig. 3 Change from baseline in quality of life (QoL) scores using the EQ-5D-5 L questionnaire.
The repeated measures analysis showed a significant effect of treatment on overall
health score change (P = 0.043) as well as a significant effect of time (P = 0.028). No significant interaction effect between treatment and time was found.
Fig. 4 Visual analog scale (VAS) pain scores.
Fig. 5 Patient satisfaction with treatment.
Adverse events
A total of eight AEss were related to either the procedure or the device, of which
seven were non-serious and one serious. In the SEMS group, one patient experienced
aspiration during FCSEMS removal (serious), one experienced recurrent dysphagia due
to stent migration, one had stent-related gastroesophageal reflux, one had foreign
body sensation and retching due to foreign body sensation (both related to stent),
one experienced thoracic pain and cervical pain during FCSEMS placement procedure,
and one patient had epigastric pain after FCSEMS removal.
Discussion
The current study confirmed that FCSEMS placement is a technically feasible and a
clinically effective treatment for benign refractory anastomotic strictures after
esophagectomy during FCSEMS indwell, including greater QoL improvement during FCSEMS
treatment compared with continued BD. However, the number of dilations required in
the 12 months after treatment were comparable between the two groups. Patients’ reports
of pain and satisfaction were similar between the groups.
The study failed to confirm our initial hypothesis, namely that placement of a FCSEMS
would reduce the number of dilations over a follow-up time of 12 months after initial
treatment. Several factors may have contributed to this outcome. First, the sample
size was smaller than intended, thus complicating detection of small differences between
the two methods. Based on the 18 subjects randomized, the power to detect the predefined
relevant difference in the primary outcome was low and consequently the chance of
detecting a true effect may be reduced. Similarly, multivariate analysis results and
QoL results should be interpreted with caution. Due to slow enrollment over 33 months,
it was decided to discontinue enrollment before the predefined required number of
patients was included. The main reason for slow enrollment is that patients tend to
be more willing to receive standard of care than to try out a new treatment, therefore,
one inclusion criterion was relaxed, namely the number of dilations to reach 16 mm
in diameter since esophagectomy. Originally this was two dilations, but was expanded
to allowing two to five dilations. Another limitation was that time 0 started at the
last dilation session for the BD group and at stent placement for the SEMS group,
because that time point reflects the time at which the stricture is first dilated
to the largest diameter. That, however, led to the patients in the SEMS group having
a 2-month shorter window to document potential recurrence of dysphagia during 12 months
of follow-up after time 0. To adjust for this issue, the Kaplan Meier analysis ([Fig. 2 ]) was conducted starting at the time of stent removal for the SEMS group and at the
time of the last dilation for the BD group.
Other limitations of the study include lack of data on details on the anastomotic
technique and that the caliber of stricture was not evaluated by barium study. Gastroesophageal
reflux disease (GERD) is a common occurrence in esophagectomy patients and may contribute
to esophageal pathology. In future studies, the potential role of possible associated
GERD should be considered.
In addition, there were no clear criteria for dilation at the time of recurrence of
dysphagia in the study. Some patients received reintervention to manage dysphagia
without an observed increase in dysphagia score.
In the current study, the stent was removed after 8 weeks to avoid tissue overgrowth
with new stricture formation as a result. During the time that the stent was in place,
dilations for dysphagia were avoided; after removal of the stent, the median number
of episodes of dysphagia requiring intervention increased. These findings suggest
that stent placement for benign strictures is effective, but only as long as the stent
is in place. Its effect is not sustained once the stent has been removed. Some anastomotic
strictures are highly refractory due to ongoing ischemia leading to recurrent fibrosis.
The number of related AEs was higher in the SEMS group. Based on literature and our
own experience, we see no reason to suspect a direct causal relationship in the finding
of increased number of related events, and instead believe this to be an artifact
of the small sample size.
Despite not meeting the primary endpoint, patients in the SEMS group experienced a
higher QoL improvement from baseline compared with the BD group, and the differences
between two groups were significant at Months 1 and 2 as well as toward the end of
the follow-up (at Months 9 and 12). Note that QoL score at baseline in the SEMS group
was significantly lower than in the BD group. To account for baseline scores, QoL
score change from baseline in percentage was calculated instead of the raw score.
The peak improvement of the QoL in the SEMS group occurred at the time of stent removal
(8 weeks). Both groups had decreased frequency of interventions for dysphagia during
the study. Larger studies are needed to evaluate the recurrent dysphagia rate for
FCSEMS vs. standard dilation in patients with refractory anastomotic strictures. Longer-term
utilization of FCSEMS and outcome measures comparing interventions during the time
of stenting only versus a standard dilation group would be relevant outcome measures
for subsequent studies.
Conclusion
It can be concluded that in the current study, no significant difference was found
between SEMS placement and repeat BD with regard to the number of BDs thereafter.
Nevertheless, temporary FCSEMS indwell may offer reduced treatment burden as long
as the stent remains in place and offers similar (if not greater) QoL improvement
from baseline sustained for 12 months compared to repeat BD for patients with refractory
anastomotic esophageal strictures. In addition, biodegradable stents could be considered
as a patient-friendly alternative for FCSEMS placement, as stent removal is avoided.
