Introduction
Even though colonoscopy is considered to be a safe procedure, a number of serious
complications such as acute iatrogenic perforations (AIPs), however small, may carry
risks which are of great concern to endoscopists [1]. Indeed, having to face a new complication during treatment of the original case
can be overwhelming. AIPs have been reported in 0.03 % to 0.08 % of diagnostic colonoscopies
[2]
[3]
[4]
[5]
[6]
[7]. Therapeutic colonoscopy carries a greater risk of AIPs, particularly following
advanced procedures [8]
[9]. In particular, two meta-analyses have reported AIPs rates of 0.9 % and 1.1 % following
endoscopic mucosal resection (EMR) in the colon [10]
[11].
Historically, in most cases, colonoscopic perforation has been treated surgically
with laparotomy or, as is usually the case today, laparoscopically. With advanced
endoscopic clipping techniques being developed during the last decade (through-the-
scope [TTS] clips, with jaw distance reaching 16 mm, a combination of TTS clips with
endoloops and over-the-scope-clip [OTSC system; Ovesco Endoscopy AG, Tubingen, Germany),
endoscopic treatment of AIPs in both diagnostic and therapeutic procedures has been
effective, particularly in perforations diagnosed during the procedure.
In the largest observational case series in the United Kingdom describing post-perforation
outcomes, TTS clips were successfully used to close defects in 83.3 % of cases where
the perforation was visualized by the endoscopist [5]. In the first prospective trial evaluating the efficacy of endoscopic closure of
AIPs of the gastrointestinal tract using the OTSC system [12], authors suggested that the OTSC system might lead to a paradigm shift in the management
of these patients, probably serving as a first-line endoscopic treatment. A case series
evaluating AIPs enforced this assumption [13]. However, data concerning advanced clipping techniques in terms of success rate,
complications, and location of perforation remain limited.
In this retrospective study, the primary objective was to compare the management of
AIPs in the colon (surgery vs. endoscopic treatment) before and after the introduction
of advanced clipping techniques. The secondary objectives included the evaluation
of mortality, hospital stay period, and the success of clip deployment both from a
technical and clinical perspective. To the best of our knowledge, this is one of the
few studies evaluating the management of AIPs in the colon in a large public hospital
over a 24-year period.
Methods
We conducted a retrospective study from July 1996 to February 2020. Data from all
colonoscopies performed at the Benizelion General Hopspital were analyzed using the
electronic database of the GI endoscopic unit which was well maintained and continuously
updated (Appendix 1). The period was divided in two sub-periods; Period 1: July 1996 to December 2012
and Period 2: January 2013 to March 2020. The two time periods were chosen taking
into account the introduction of advanced clipping techniques in our department and
therefore, 2013 was selected as the cut-off point year. This study was approved by
the scientific committee of our hospital. A signed written consent form for this retrospective
study was not required.
All AIPs occurring during a colonoscopy and detected during or immediately (< 4 hous)
after the procedure, were included in the study. Delayed perforations (> 4 hours after
the procedure) [1]
[14] were excluded. Such cases were excluded per protocol since most of them usually
require immediate surgical treatment. Perforation was defined as a complete disruption
of the colonic wall documented by an endoscopic view of either extraintestinal structures
or the presence of gas and colonic contents outside the gastrointestinal tract detected
in a abdominal computed tomography (CT) scan performed after the procedure. Cases
of gas outside the gastrointestinal tract without the presence of colonic contents
or definite instrumentation outside the large bowel lumen were not included. In both
periods, all patients were managed by a multi-disciplinary team including endoscopists,
radiologists, and surgeons.
Perforations during diagnostic colonoscopy were defined as those appearing during
a colonoscopy performed for diagnostic reasons where no tissue was removed from the
colon for therapeutic reasons. Perforations during therapeutic colonoscopy were defined
as those occurring during a polypectomy, EMR or dilatation. Morbidity was defined
as any complication associated to the treatment (surgical or endoscopic) of the perforation.
Advanced clipping techniques were defined as the application of the OTSC system, multiple
clipping with TTS clips, or the combined technique using TTS clips plus endoloop [15]
[16]. The choice of clipping technique was at the discretion of the endoscopist.
Technical success of clip application was the adequate endoscopic closure of the perforation
followed by the absence of leakage as confirmed by a CT scan performed after the procedure.
The absence of complications related to clip application were defined as clinical
success of clip application.
The presence of colonic contents detected in an abdominal CT performed after the procedure,
the failure of endoscopic treatment, the inadequate colon preparation and the deterioration
of patient's clinical condition were the parameters taken into account by the multidisciplinary
team for making the decision to treat the patient surgically [1]
[17]. The protocol briefly included surgical or endoscopic treatment in those cases where
either extraintestinal structures were observed, or the presence of gas and colonic
contents outside the gastrointestinal tract were detected in an abdominal CT scan,
performed after the procedure. In each case the committed team decided at its discretion.
All colonoscopies were performed by trained endoscopists (no trainees were involved).
In colonoscopies performed after 2012 the Boston Bowel Preparation Scale [18] was recorded. All colonoscopies were performed with the technique of one operator,
with conscious sedation. After 2012, all endoscopic procedures in our department are
performed using CO2. Position changes of the patient were performed when it was needed.
In cases where the perforation was treated surgically, the surgeon was asked to measure
the length of perforation. In cases where the perforation was treated endoscopically,
perforation length was measured by visual comparison using open biopsy forceps of
predetermined width or/and the diameter of the colonoscope.
Statistical analysis
Continuous data were compared with unpaired t or Mann Whitney tests as appropriate.
Categorical variables were tested using corrected χ2 or two-sided Fisher exact tests for univariate comparisons, as appropriate. Criterion
for statistical significance was P < 0.05.
Results
Period 1
The total number of colonoscopies performed at our hospital during Period 1 was 33055.
Among them, 21375 were performed for diagnostic and 11680 for therapeutic purposes.
We observed 15 perforations: 10 occurred during diagnostic and 5 during therapeutic
colonoscopies ( [Table 1]). Thus, the overall iatrogenic colonic perforation rate was 0.04 % (0.04 % for diagnostic
and 0.04 % for therapeutic colonoscopies). Most perforations were recorded in the
sigmoid colon (9/15, 60 %) and 13/15 (86.6 %) of patients were females. In terms of
treatment, 14 patients were treated surgically and 1 endoscopically ( [Table 2]). The mean size of the perforation in patients having either diagnostic or therapeutic
colonoscopy was 23.2 ± 3.3 and 11.2 ± 1.7 mm, respectively.
Table 1
Clinical parameters of patients at study entry in the 2 studied periods.
|
Variables
|
Period 1
|
Period 2
|
|
Number of patients
|
15
|
11
|
|
Sex (male)
|
2
|
1
|
|
Age (yr)
|
67.6 ± 12.7
|
69.1 ± 9.5
|
|
Location of perforation
|
|
Rectum
|
1
|
1
|
|
Sigmoid
|
9
|
9
|
|
Descending
|
–
|
–
|
|
Transverse
|
1
|
–
|
|
Cecum
|
4
|
1
|
|
Cause
|
|
Diagnostic endoscopy
|
10
|
7
|
|
Therapeutic endoscopy
|
5
|
4
|
|
EMR
|
5
|
3
|
|
Dilation
|
–
|
1
|
EMR, endoscopic mucosal resection.
Table 2
Treatment characteristics of patients in the two studied periods.
|
Variables
|
Period 1 (n = 15)
|
Period 2 (n = 11)
|
|
Treatment[1]
|
|
|
|
Surgery
|
14
|
3
|
|
Endoscopic treatment
|
1
|
8
|
|
Surgery
|
|
Open surgery-primary closure
|
14
|
–
|
|
Laparoscopic surgery-primary closure
|
–
|
2
|
|
Open surgery-stoma formation
|
–
|
1
|
|
Endoscopic treatment
|
1
|
8
|
|
TTS clips
|
1
|
2
|
|
TTS clips plus endoloop (king closure)
|
–
|
1
|
|
OTSC system
|
–
|
5
|
|
Failed endoscopic treatment
|
1
|
2
|
|
Failed endoscopic treatment with TTS clips
|
1
|
2
|
|
Failed endoscopic treatment with OTSC system
|
–
|
–
|
|
Hospital stay (days)[2]
|
6.9
|
4.0
|
|
Mortality
|
1
|
–
|
OTSC, over-the-scope-clip; TTS, through-the- scope.
1
P < 0.01
2
P < 0.01
Surgery
All cases were diagnosed during the procedure or immediately (< 4 hours) following
it. More specifically, in nine diagnostic colonoscopies, extraintestinal structures
were detected during the procedure and the patients were referred for surgical treatment
without any attempt for endoscopic treatment. In one therapeutic colonoscopy, a complete
disruption of the colonic wall was observed during an EMR procedure. The defect was
closed using six TTS clips. However, a CT scan performed immediately after the procedure
revealed the presence of extraintestinal gas and colonic contents in the peritoneal
cavity, thus deeming it necessary for the patient to be treated surgically. In four
therapeutic cases, a CT scan was performed due to clinical signs of perforation after
the completion of the procedure, showing the presence of extraintestinal gas and colonic
contents, leading us to the decision of treating all of them surgically. In all surgically
treated cases, a primary closure of the defect was performed with open surgery ( [Table 2] and [Table 3]). No colonic resection was needed.
Table 3
Perforations during therapeutic procedures and lesion characteristics.
|
Cases
|
Period
|
Lesion
|
Location
|
Therapeutic procedure
|
Lesion size (mm)
|
Treatment
|
|
1
|
1
|
Sessile polyp
|
Transverse
|
EMR one piece
|
15
|
Open surgery-primary closure
|
|
2
|
1
|
Sessile polyp
|
Cecum
|
EMR piecemeal
|
25
|
Open surgery-primary closure
|
|
3
|
1
|
Sessile polyp
|
Cecum
|
EMR piecemeal
|
20
|
Open surgery-primary closure
|
|
4
|
1
|
Sessile polyp
|
Cecum
|
EMR piecemeal
|
30
|
Open surgery-primary closure
|
|
5
|
1
|
Sessile polyp
|
Cecum
|
EMR piecemeal
|
30
|
Open surgery-primary closure
|
|
6
|
2
|
Sessile polyp
|
Cecum
|
EMR one piece
|
15
|
Laparoscopic surgery- primary closure
|
|
7
|
2
|
Sessile polyp
|
Sigmoid
|
EMR piecemeal
|
30
|
Endoscopic treatment, TTS clips
|
|
8
|
2
|
Benign stenosis
|
Colorectal anastomosis
|
Endoscopic balloon dilation
|
20
|
Open surgery-Stoma formation
|
|
9
|
2
|
Sessile polyp
|
Sigmoid
|
EMR one piece
|
20
|
Endoscopic treatment-OTSC system
|
OTSC, over-the-scope-clip; TTS, through-the-scope; EMR, endoscopic mucosal resection.
TTS clips
In one diagnostic colonoscopy, a complete disruption of the sigmoid wall was observed.
The defect was closed using eight TTS clips. A CT scan performed after the procedure
revealed only the presence of extraintestinal gas which, in turn, was successfully
treated conservatively.
Outcome
Fourteen patients were discharged within a reasonable timeframe ( [Table 2]) as no complication was observed associated to the surgical or endoscopic treatment
of the perforation. One 81-year-old woman with colon cancer and leukemia, who had
been treated surgically for a perforation in the sigmoid colon during a diagnostic
colonoscopy, died 2 days after surgery due to severe complications.
The technical and clinical success rates for TTS clips application were 50 % (1/2)
and 100 % (1/1) respectively.
Period 2
Similarly, 36831 colonoscopies were performed in Period 2. Of these, 22239 were diagnostic
and 14592 were therapeutic. We observed 11 perforations: seven occurred during diagnostic
and four during therapeutic colonoscopies ( [Table 1]). Thus, the overall iatrogenic colonic perforation rate was 0.02 % (0.03 % for diagnostic
and 0.02 % for therapeutic colonoscopies). Most of the perforations were observed
in the sigmoid colon 9/11 (81.8 %), while 10/11 patients (90.9 %) were women. Regarding
treatment, three patients were treated surgically and eight endoscopically ( [Table 2]). The mean size of the perforation in patients having either a diagnostic or therapeutic
colonoscopy was 24.4 ± 2.2 and 18.7 ± 4.7 mm, respectively.
OTSC system
In four diagnostic colonoscopies, extraintestinal structures (all in the sigmoid colon)
were identified during the procedure. In all patients, an OTSC system was applied
and the defect was effectively closed ( [Table 2]). A CT scan performed in all cases after the application of OTSC system revealed
the presence of only extraintestinal gas. In one therapeutic colonoscopy ( [Table 3]) a complete disruption of the sigmoid wall was observed during an EMR (en block
resection of a 2-cm sessile polyp in the sigmoid colon). The defect was successfully
closed with an OTSC and was then checked via a CT scan which revealed only the presence
of extraintestinal gas. All five patients uneventfully received further conservative
treatment. We did not face any complications related to the OTSC system application.
TTS clips
In two colonoscopies (one diagnostic and one therapeutic; EMR in the sigmoid colon)
a complete rupture of the sigmoid wall was recognized ( [Table 2] and [Table 3]). The defect was closed by applying seven and six TTS clips in each case respectively,
and the CT scans performed after the procedures revealed the presence of only extraintestinal
gas. Thereafter, the patients were treated conservatively without complications. Finally,
in another diagnostic colonoscopy a complete disruption of the rectal wall was observed
after a retroflexion manoeuvre. Perforation was effectively treated with eight TTS
clips and endoloop (king closure). CT scan after the procedure showed the presence
of only extraintestinal gas. Once again, the patient received further conservative
treatment without complications
Surgery
In one therapeutic colonoscopy, a complete rupture of the cecum wall was identified
during an EMR procedure. The defect was closed using eight TTS clips. However, the
closure was not considered efficient and the patient was treated surgically. A primary
closure of the defect was immediately performed laparoscopically. In another diagnostic
colonoscopy, a complete disruption of the sigmoid wall was observed. The defect was
closed using seven TTS clips. However, the closure was not considered adequate and
the patient was referred for surgery and a primary closure of the defect was performed
laparoscopically ( [Table 2] and [Table 3]). In both cases TTS clips closure was considered incomplete since defect in the
colonic wall was still observed, being not amenable to further endoscopic treatment
and the decision to treat the patients surgically was made. The OTSC was not available
at that time. In a third patient, a perforation occurred during endoscopic balloon
dilation of a benign colorectal anastomosis stricture after surgery for adenocarcinoma
in the sigmoid colon. The patient was immediately treated with open surgery with a
formation of a temporal stoma.
Outcome
Eleven patients were discharged within a reasonable timeframe ( [Table 2]). Both laparoscopically treated patients were discharged 3 days after the operation.
No complications associated with the surgical or endoscopic treatment of the perforation
were observed in these patients during hospitalization.
The technical success rate for TTS clip application was 60 % (3/5) including the case
with the endoloop combination. The technical success rate for OTSC system application
was 100 % (5/5). The clinical success rate for all clipping devices was 100 %. The
percentage of patients undergoing laparoscopic surgery was 66.6 % (2/3).
Comparison of the two periods
The rate of surgery was 93.3 %% (14/15) for Period 1, and 27.2 % (3/11) for Period
2 (P < 0.01) ( [Table 2]) ( [Fig.1]). The subsequent rate of endoscopic treatment increased from 6.6 % to 72.7 %. In
the subgroup of cases where AIPs occurred during a diagnostic colonoscopy, the respective
rate of endoscopic treatment increased from 0 % for Period 1 to 85.7 % for Period
2. The mortality rate was 8.3 % (1/15) for Period 1 and 0 % (0/11) for Period 2 of
the study, and this difference was not significant (P = 1). The mean hospital stay was 6.9 days, and 4 days for Period 1 and 2 respectively
(P < 0.01) ( [Table 2]).
Fig. 1 Comparison of treatment during the two time periods.
The Boston Bowel Preparation Scale score was > 6 in the cases which were amenable
to be calculated.
Discussion
In this retrospective study, we primarily evaluated the management of colonic AIPs
(surgery vs endoscopic treatment) between Period 1 and Period 2. A definite shift
to endoscopic treatment (from 6.7 % to 72.7 %) was demonstrated. Secondarily, there
was a significant reduction in length of hospitalization during Period 2. It would
be noteworthy to mention that the technical and clinical success rate of the OTSC
system application remained 100 % even in large perforations observed during a diagnostic
colonoscopy ~2 cm.
AIPs related to colonoscopy is defined as the recognition of gas or colonic contents
outside the colon [1]
[19] or any clearly visible sign of perforation identified endoscopically during or in
time related to endoscopy. In the present study, cases with gas outside the GI tract
but without the presence of colonic contents or definite instrumentation outside the
large bowel lumen were not included. In other words, cases with colonic perforation
were included where endoscopic or surgical treatment was mandatory because otherwise,
serious complications would appear. This could explain the similar perforation rates
between diagnostic and therapeutic colonoscopies.
It is well documented that the timing of diagnosis is important for the management
of the AIPs of the colon and the patient outcome [20]
[21]. Morbidity and mortality have significantly been related to the delay of AIPs diagnosis
> 24 hours [22]. In this study, all AIPs occurring during a colonoscopy that were detected during
or immediately (< 4 hours) after the procedure were included in the study. Therefore,
the associated mortality and morbidity rates were rather low. No cases with delayed
diagnosed perforations were included. After searching rigorously across our electronic
database, we did not find any case of delayed colonic perforation requiring surgical
treatment. As recommended, all patients in our unit are thoroughly evaluated after
colonoscopy for clinical symptoms suggestive of perforation and are not allowed for
discharge until they are completely free of such symptoms [1]. In cases of doubt, these patients may undergo an abdominal CT scan with the intention
of detecting radiologic findings compatible with perforation. We acknowledge that
we have a low threshold for performing abdominal CT scan in patients with post-colonoscopy
symptoms suggesting perforation, and, by adopting this policy, the number of cases
with delayed colon AIPs has practically been eliminated.
There is enough evidence that perforations occurring during diagnostic colonoscopies
are larger than those occurring during therapeutic colonoscopies [17]
[23]
[24] and this was confirmed in our data. The OTSC system is considered more effective
in defects larger than 1 cm, performed mainly during diagnostic colonoscopy due to
its ability to capture and sustain a larger volume of tissue with a higher compression
force compared to conventional TTS clips. Furthermore, relatively large defects can
also be sutured by using dedicated grasping forceps. It is of the utmost importance
to avoid suctioning the extraluminal structures into the cap of the OTSC system and
therefore, caution is strongly recommended in the phase of pulling the wall defect
to the cap via suction. Suction of extraluminal structures into the cap of the OTSC® system have sparsely been reported [25]
[26]. To minimize such a risk and to succeed in the closure of even larger defects, the
use of a twin grasper has been recommended. In this study, we did not use the twin
grasper device. However, we did not face any complications coming from the OTSC system
application. The technical and clinical success rate of OTSC was 100 %, while it was
mostly used in large defects performed during diagnostic colonoscopy. We could make
a speculation that the extended experience in the use of OTSC system and Full Thickness
Resection Device (FTRD, Ovesco Endoscopy AG, Tubingen, Germany) devices in our department
[27] has resulted in the uneventful application of the OTSC system. Nevertheless, the
number of presented cases treated with the OTSC system is rather small and a misjudgement
could not be excluded. In other studies, a few failures concerning the technical success
of the OTSC system application have been reported [12]
[13]
[28]
[29].
The sigmoid colon and rectosigmoid junction are the most common sites of AIPs during
a diagnostic procedure due to direct mechanical injury applied by the colonoscope
during insertion [5]
[6]
[23]
[30]. In agreement with the existing literature, most perforations in this study occurred
in the sigmoid colon, particularly those during diagnostic colonoscopy, and it is
therefore difficult to make comparisons based on these data, with regard to the effectiveness
of advanced clipping technique in the different segments of the colon. Indeed, there
is lack of data to adequately evaluate this issue. It is worth noting that the use
of the OTSC system in the right colon is related to certain technical obstacles due
to the rather difficult advancement of the colonoscope mounted with the OTSC system.
However, the experience with FTRD, a rather similar device, has assisted in overcoming
obstacles concerning the advancement of the mounted colonoscope [27]
[31].
In the first prospective trial evaluating the efficacy of endoscopic closure of AIPs
of the GI tract using the OTSC system, Voermans et al [12] revealed a 92 % success rate in the colon among the 13 cases included. Additionally,
in this study the assumption that the OTSC system might lead to a paradigm shift in
the management of these patients was done closing the defects with the OTSC system
as a first-line treatment. Retrospective studies and reviews in GI perforations have
enforced these assumptions [13]
[28]
[29]
[32]. Our study comes some years later to confirm the paradigm shift, specifically in
the colon, by analyzing a large colonoscopic historical cohort in both a disgnostic
and therapeutic setting. Our data revealed a significant reduction regarding length
of hospitalization in patients treated endoscopically or laparoscopically, leading
to the speculation that this might be associated with lower morbidity, although it
was not clearly proven by these data. After a multidisciplinary approach, all patients
treated endoscopically for AIPs of the colon remained in hospital to receive intravenous
antibiotic treatment despite not having developed any complications. We may assume
that if this strategy had not been adopted, the difference in the length of hospitalization
between the two periods would be greater.
In this study the technical and clinical success rate of TTS clipping is 50 %. In
the largest retrospective observational case series in Europe describing post-perforation
outcomes, TTS clips were successfully used to close defects in 83.3 % of cases where
the perforation was visualized by the endoscopist [5]. This discrepancy might be explained by the exclusion of EMR cases with type 3 and
4 mural injury (Sydney Classification of Deep Mural Injury (DMI) following EMR [33]) which are considered perfectly amenable to the TTS clipping. Indeed, only cases
with defects within a white cautery ring and observed contamination were included
(type 5) (Sydney Classification of DMI following EMR). Indeed, this was also applied
to the Period 1 cases, where the entry criterion was defined as the presence of colonic
contents outside the colon, i. e. open contamination that is not only limited to the
presence of extraluminal gas which does not necessarily presuppose open perforation.
There are some limitations in our study. First, this study is a retrospective one
and it was performed in the setting of one center, thus the results cannot be easily
generalized. Second, the number of patients included is small. However, this limitation
is unavoidable since perforation is not a frequent complication of colonoscopy. Finally
and most importantly, it is difficult to make accurate comparisons with previous studies,
and in particular comparisons concerning the perforation rates, since confounding
factors are created by variations in methodology as well as by differences in the
definition of the perforation. The strength of this study is that the database in
our unit was well maintained, continuously updated and resulted in minimizing the
possibility of underestimating the cases of perforation and the subsequent required
treatment.
Conclusions
In conclusion, data from this historical cohort study have clearly shown a decrease
in the surgery rate of AIPs occurring in both diagnostic and therapeutic colonoscopies
and detected during or immediately after the procedure. In addition, after the endoscopic
or laparoscopic treatment of AIPs, a significant reduction in the required days of
hospitalization was observed.
