Introduction
The quality of endoscopic resections has become a focal point in endoscopy research
where the target is techniques with higher rates of complete resection for small polyps
(cold snare for lesions < 10 mm in diameter) or large lesions (endoscopic submucosal
dissection (ESD), for large lesions > 30 mm in size) [1 ]
[2 ]. Endoscopic mucosal resection (EMR) with snare is the technique recommended to resect
non-invasive colorectal neoplastic lesions (10 – 30 mm) effectively and with low morbidity
[1 ]. Nevertheless, such resections are only curative when the lesion is resected in
a single specimen (en bloc) with free margins (R0 resection). In the case of piecemeal
resection (more than one piece) or in the case of incomplete en bloc resection (undetermined – Rx,
or invaded margins – R1), the risk of local recurrence increases and a second colonoscopy
is recommended within 6 months to detect and resect potential local recurrence. Thus,
improving the quality of EMR to increase the rate of R0 resection and to reduce the
need for follow-up procedures should become a target of research. It has been reported
that conventional EMR is associated with 62 – 65 % en bloc resection for lesions < 20 mm
in diameter but R0 resection has never been evaluated precisely [1 ]
[3 ]
[4 ].
We previously described the anchoring EMR (A-EMR) technique [5 ] which consists of creating a small hole in the submucosa with the tip of a conventional
EMR snare to anchor the snare tip in the surrounding margin to reduce sliding and
to enlarge the snare opening. This technique is currently used in four centers in
France and in this study, we retrospectively assess the current effectiveness and
safety of A-EMR.
Methods
Design
This was a retrospective multicenter pilot study in four French endoscopy tertiary
care centers with experience in EMR and ESD, and where A-EMR was routinely practiced.
In total, 10 operators participated in the study. They had each performed >200 EMR
procedures before beginning A-EMR.
Inclusion criteria
The study included all consecutive A-EMR procedures attempted between May 2017 and
December 2017 with en bloc intent for sessile or flat lesions between 10 and 30 mm
in size. Evaluation of lesion size was done endoscopically; however, size is known
to be imprecisely measured this way and may be influenced by the operator’s experience.
The classic landmark used to help in size evaluation is the comparison with an open
biopsy forceps (measuring about 7 mm in length). Once the specimen was resected, it
was systematically stretched on cork using needles and accurately measured.
Exclusion criteria
The study excluded lesions with pedunculated shape (Ip of Paris classification) or
with invasive shape (ulcerated type III of Paris classification), recurrent or residual
lesions after previous resection as well as other causes of severe fibrosis (ulcerative
colitis), and lesions with a high risk of superficial adenocarcinoma requiring ESD.
Anchoring-EMR procedure
In all cases, 10 or 25 mm conventional snares from Olympus were used (SD-210U, Olympus,
Tokyo,
Japan) according to lesion size. Snare size was chosen by the operator during the
procedure, and
use of a distal cap fitted to the colonoscope tip was left to the operator’s discretion.
After
injection of colored fluid (indigo carmine blue, Carmine, SERB, Paris, France) into
the
submucosal layer, anchoring of the snare tip was performed by creating a small hole
in the
mucosa a small distance from the lesion edges using electric cutting current (Endocut
Q, Erbe,
Tübingen, Germany). This incision aimed to reach the submucosa and securely anchor
the snare tip
there ([Fig. 1 ], [Video 1 ]).
Fig. 1 Anchoring procedure. First an incision is made into the mucosa using the tip of the
snare (a ) followed by anchoring of the tip and snare opening (oval shape of snare) (b ). Then, pressure on the snare leads to a larger round shape (c ) and then resection is performed after snare closure (d ).
Video 1 Snare shape depending on pressure and an example of the procedure.
Then the snare was deployed progressively and adjusted around the lesion trying to
respect free margins between lesion edges and snare closure. Anchoring allowed the
snare to be enlarged after application of pressure ([Fig. 1 ] and [Fig. 2 ]). Once closed, resection was performed as usual with the Endocut Q. Immediately
after resection, the resected area was assessed to detect both muscular damage according
to Sydney’s classification (partial damage with target sign or complete transmural
perforation) [6 ] and residual neoplastic tissue. In the case of perforation, endoscopic closure was
attempted. If a single snare EMR resected the whole lesion without residual tissue,
resection was considered to be en bloc endoscopically. In the case of residual tissue
detected using white light, virtual chromoendoscopy and magnification if needed, additional
snare resection(s) (with or without anchoring) was performed leading to a piecemeal
resection automatically considered to be R1. After resection, the specimen was stretched
on cork with needles and fixed in buffered formalin for pathological assessment. Lesions
were then sliced into 2 mm sections followed by analysis of resection margins to assess
R0 status. Pathologists used their conventional technique to analyze the margins and
were not aware of the future retrospective evaluation.
Fig. 2 Snare shape depending on amount of pressure on the anchoring point. The snare is
oval without pressure on the tip (a ) and becomes round when a pressure is applied to the tip (b ).
Outcomes
Primary outcome was the proportion of R0 resections defined histologically by the
presence of lateral and deep free margins around the lesion after A-EMR.
Secondary outcomes assessed after A-EMR were: the characteristics of R0 resection
(size, operator, distal cap use); the proportion of successful A-EMR defined by the
ability to anchor the snare in the submucosa without slipping of the snare during
closure (tip of the snare fixed in the submucosa throughout snare closure); the proportion
of en bloc resection defined endoscopically; the proportion of immediate and delayed
adverse events (within the first month) including perforation (complete or partial,
i. e. with target sign) and bleeding, as well as the proportion of adverse events
leading to further surgery.
Data collection
All data were retrieved from the endoscopy, pathology, and hospitalization reports
and were collected, anonymized, and collated in a spreadsheet (Excel; Microsoft, Redmond,
WA, United States). Data were verified by an independent research assistant (MMG).
Statistical analysis
Baseline characteristics and outcome variables were described by the mean, standard
deviation (SD) and range for continuous variables, and by frequencies and percentages
for categorical variables. Comparisons were performed using Fisher’s exact test. Multivariate
analysis was performed by penalized logistic regression, including all factors significant
in univariate analysis. A P value < 0.05 was considered to be statistically significant. All analyses were performed
using R software version 3.4.2. (R Core Team. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/ ).
Ethics
This study was in compliance with the Declaration of Helsinki and received approval
from the ethics committee of the Hospices Civils de Lyon (March 7, 2018). All of the
patients in the study consented to participate.
Declaration
This study was declared on the database of the United States National Library of Medicine
(clinicaltrials.gov) under the number NCT03467451.
Results
From May 2017 to December 2017, 141 consecutive lesions in 125 patients (mean age:
65.4 years; range 30 – 89 years) were resected by 10 operators in four French centers
using the A-EMR technique. Among them, 112 patients had 1 lesion (89.6 %), 10 had
2 (8.0 %), and 3 patients had 3 lesions (2.4 %). The size of 6 lesions was not determined
pathologically as these were piecemeal resections; the mean (pathologically determined)
size of the 135 lesions was 19.8 mm (range 8 – 40 mm; SD: 7.1). Lesions were adenomas
or intramucosal adenocarcinomas in 89/141 cases (63.1 %), sessile serrated lesions
in 51 cases (36.2 %), and invasive submucosal adenocarcinoma in one case (0.7 %).
All lesions included in the study were endoscopically determined to be between ≥ 10
and ≤ 30 mm; according to pathological assessment, there were 87 lesions between 8
and 20 mm, 38 lesions between 21 and 30 mm, and 10 lesions > 30 mm; the characteristics
of the resected lesions are presented in [Table 1 ].
Table 1
Characteristics of lesions resected by A-EMR.
Characteristics
Lesions, n = 141
Location
24 (17.0)
53 (37.6)
9 (6.4)
16 (11.3)
13 (9.2)
15 (10.6)
11 (7.8)
0
Pathologically-determined size
87 (64.4)
38 (28.1)
10 (7.4)
6
Paris classification
52 (36.9)
2 (1.4)
67 (47.5)
5 (3.5)
1 (0.7)
3 (2.1)
11 (7.8)
0
Histology subtype
89 (63.1)
51 (36.2)
1 (0.7)
0
Technical results
The injection before EMR was performed with saline serum in 113 cases (80.1 %), with
hyaluronic acid (0.4 % solution; Sigmavisc, Life Partners Europe, Paris, France) in
24 (17.0 %), and with a glycerol mixture in 4 cases (2.8 %). Snare anchoring through
the mucosa was feasible in 136 cases (96.5 %). In five cases, anchoring was not feasible
since the snare slipped away from the anchoring point before complete snare closure.
A cap was fitted to the tip of the colonoscope in 54 cases (38.3 %), no cap was used
in 54 cases (38.3 %), and data were missing in 33 cases (23.4 %). A 25 mm snare was
used in 127 cases (90.1 %) and a 10 mm snare in 14 cases (9.9 %). Endoscopically,
en bloc resection was performed in 115/141 cases (81.6 %).
Histology results
R0 resections were obtained in 99/141 cases (70.2 %). The proportion of R0 resections
obtained was significantly different according to lesion large diameter: 82.8 % (72/87)
for lesions <20 mm in size, 55.3 % (21/38) for lesions between 20 and 30mm in size,
and 50.0 % (5/10) for those > 30 mm in size (P = 0.002). When anchoring was attempted but not achieved (5 cases), the proportion
of R0 resections was 0 % although the proportion obtained when anchoring was feasible
was 72.8 % and the difference was statistically significant (P = 0.002). There was no statistically significant difference in the proportion of
R0 resections according to the other factors investigated, including operator or type
of solution injected into the submucosa ([Table 2 ]). There were three operators with a proportion of R0 below the overall value (70.2 %)
and two who attained 100 % R0. In multivariate analysis adjusted for the size of the
lesion, the effect of the success of anchoring was still significant (P value = 0.025).
Table 2
En bloc and R0 resection according to lesion and procedure characteristics.
Characteristics
n
En bloc resection (%)
R0 resection (%)
P value
Success of anchoring
0.002
136
115 (84.6)
99 (72.8)
5
0 (0.0)
0 (0.0)
0
0
0
Pathology-determined size
0.002
87
80 (92.0)
72 (82.8)
38
26 (68.4)
21 (55.3)
10
8 (80.0)
5 (50.0)
6
6
6
Center
0.737
41
33 (80.5)
27 (65.9)
55
47 (85.5)
38 (69.1)
21
16 (76.2)
15 (71.4)
24
19 (79.2)
19 (79.2)
0
0
0
Operator
0.399
5
5 (100)
5 (100)
7
5 (71.4)
5 (71.4)
13
9 (69.2)
6 (46.2)
12
10 (83.3)
6 (50.0)
12
10 (83.3)
9 (75.0)
21
16 (76.2)
15 (71.4)
21
19 (90.5)
16 (76.2)
25
21 (84.0)
17 (68.0)
24
19 (79.2)
19 (79.2)
1
1 (100)
1 (100.0)
0
0
0
Histology subtype
0.339
90
70 (77.8)
66 (73.3)
51
45 (88.2)
33 (64.7)
0
0
0
Cap assisted
0.283
54
45 (83.3)
42 (77.8)
54
44 (81.5)
36 (66.7)
33
33
33
Injection medium
0.671
113
93 (82.3)
77 (68.1)
4
3 (75.0)
3 (75.0)
24
19 (79.2)
19 (79.2)
0
0
0
Adverse events
Perforations occurred in 5/141 resections (3.5 %) in 125 patients (5/125, 4.0 %) including
complete transmural perforation in three cases (2.1 %) and partial perforation with
target sign in two cases (1.4 %). Complete perforations never occurred for lesions
< 20 mm in size (0/87, 0 %), in one case for lesions between 20 and 30 mm in size
(1/38, 2.6 %), and in two cases >30 mm in size (2/10, 20 %). All were treated conservatively
with endoscopic closure using hemoclips, and none led to salvage surgery. One delayed
bleeding (melena) occurred after 24 hours (0.7 %) and which stopped spontaneously
without a new colonoscopy; blood transfusion was not necessary.
Discussion
Anchoring-EMR is effective in achieving en bloc and R0 resections of colorectal neoplasia
lesions, in particular, those <20 mm in size. The results presented herein for the
A-EMR technique compare well with the literature as it is reported that only 62 – 65 %
of resections are en bloc with conventional EMR [1 ]
[3 ]
[4 ]
[7 ]; interestingly, it is difficult to compare R0 resection rates as they do not seem
to be reported in studies evaluating EMR.
Current quality indicators for endoscopic resections require en bloc resection for
small polyps (< 10 mm), and on this basis, cold snaring of these is strongly recommended
[1 ]. For larger lesions (> 30 mm), EMR is mostly piecemeal in intent, whereas ESD has
a high R0 resection rate [8 ]
[9 ]
[10 ]
[11 ]. Nevertheless, ESD is time consuming and requires expensive devices and thorough
training. Alternative hybrid strategies have been developed to push the boundaries
of ESD indications, but the resulting R0 resection rate was far from perfect and,
although marginally quicker, these use the same devices [12 ] which does not meaningfully change the cost.
The A-EMR technique associates a simple technical trick with conventional EMR snares
and with a good overall proportion of en bloc and R0 resections in different colon
segments, including more than 60 % of resections in the ascending colon, known to
be a technically challenging location. In this study, when considering only lesions
< 20 mm in size, which are the most suitable for en bloc EMR according to ESGE guidelines
[1 ], A-EMR achieved a high quality level as the proportion of R0 was comparable to that
reported for ESD in Japanese studies [8 ]
[13 ]
[14 ], but without any perforations. This is of interest because, as in the case of R0
resections for non-invasive neoplasia, the risk of local recurrence is theoretically
null and therefore there is probably no need for follow-up colonoscopy to detect local
recurrence [15 ] despite the recent ESGE guidelines [16 ].
For larger lesions, although there was no comparator group, the proportion of R0 may
be better than for conventional EMR, but does not reach that obtained with ESD, and
it is of note that perforations occurred in such cases. Among these, the proportion
of perforations was 6.2 %, which is considerably higher than the 1.2 % reported after
conventional EMR in comparably sized lesions [17 ]. Anchoring could lead to a deeper catching and, as a result, to an increased risk
of perforation, but comparative data are needed to evaluate the morbidity of this
technique compared to conventional EMR. The latter point highlights the main limitation
of this study.
A general point to consider is that, in light of the relatively high perforation rate
in this series, operators using A-EMR should be careful if en bloc resection is attempted
for a lesion > 2 cm in size. Furthermore, the number of procedures for each operator
was low, which precluded analysis of a potential operator effect on the R0 result.
This was further compounded by the non-standardized solution used for the submucosal
cushion and the non-systematic use of a distal cap. Although not significant in this
sample, a larger difference between the frequency of en bloc (88.2 %) and R0 resections
(64.7 %) appeared for sessile serrated lesions than for adenomas (respectively, 77.8
and 73.3 %). This larger difference underlines the difficult delineation of sessile
serrated lesions endoscopically and then the technical challenge to remove those lesions
with free margins as previously reported [18 ]; this could affect the results in a future comparative study. Another point to consider
is that the expertise of the physicians involved is also a limitation since most of
them were also expert in ESD and were used to accessing the submucosa, which may affect
the generalizability of the results.
Our opinion is that the main limitation of conventional EMR is the lack of snare fixation
at the distal point with a risk of snare sliding to the lesion edge leading to incomplete
(R1), uncertain (Rx) or piecemeal resections. This is a postulation that led us to
create and develop this method of A-EMR.
To summarize, the A-EMR technique appears to be promising with a high proportion of
R0 for lesions of 10 to 20 mm diameter without any perforation. It could also offer
an alternative to ESD or to hybrid techniques to reach R0 for lesions between 20 and
30 mm saving time and money by using conventional devices. Randomized comparative
studies are required using standardized procedures to conclude as to the potential
benefits of this technique.
