Introduction
Colorectal polyps are potential precursor lesions of colorectal cancer (CRC), and
endoscopic resection reduces CRC-related mortality [1]. The recommended method of endoscopic removal depends, among others, on polyp morphology
and size. According to guidelines of the European Society of Gastrointestinal Endoscopy
(ESGE) and the US Multi-Society Task Force on Colorectal Cancer (USMSTF), endoscopic
mucosal resection (EMR) or endoscopic submucosal dissection (ESD) are advised for
the removal of nonpedunculated colorectal polyps (NPCPs) sized ≥20 mm [2]
[3]. However, recommendations for polyps sized 10–20 mm are less clear [4]
[5]. Nevertheless, a recent meta-analysis found that conventional hot-snare EMR with
submucosal injection is the most commonly performed resection technique for lesions
sized 10–20 mm [6].
Despite the effectiveness of EMR, there remains a risk of recurrent or residual lesions,
particularly with piecemeal EMR (pEMR), which has an increased risk of recurrence
compared with en bloc resections [2]
[7]
[8]. While high-quality studies investigating the incidence, predictors, and preventative
techniques for NPCPs sized ≥20 mm are increasingly being performed, studies specifically
addressing recurrence rates and potential predictors for resecting NPCPs sized 10–20
mm remain scarce [8]
[9]
[10]
[11]. Previous studies reported recurrence or incomplete resection rates of up to 20.4%
for EMRs of NPCPs sized 10–20 mm [4]
[6]
[7]. However, these studies frequently pooled data of various polyp sizes and resection
techniques, obscuring a clear understanding of the risk of recurrence after pEMR in
this specific size category.
Therefore, the aim of this retrospective, multicenter cohort study was to evaluate
the recurrence rate, associated factors, and post-EMR scar identification rate in
patients undergoing early surveillance colonoscopy (ESC) after pEMR of NPCPs sized
10–20 mm.
Methods
Study design
This retrospective, multicenter cohort study was conducted at five hospitals in the
Netherlands: two teaching hospitals and three non-teaching hospitals. The Medical
Ethical Committee Oost-Nederland waived the requirement for ethical approval according
to Dutch law (Dutch Medical Research Involving Human Subjects Act), reference 2021–13058,
issued on July 15, 2021, due to the retrospective design of the study and negligible
risk for patients. Consequently, the need for informed consent procedures was waived.
Approval from each participating study site was obtained, ensuring adherence to local
standards and protocols. The study was reported according to the Strengthening the
Reporting of Observational Studies in Epidemiology (STROBE) statement.
Patients
Historic electronic medical record data from patients at each study site underwent
review by a study team member following the application of specific search terms,
including “piecemeal,” “endoscopic mucosal resection,” “EMR,” or “pEMR,” to identify
cases from the endoscopy patient cohorts. Adult patients were included in the cohort
if they underwent hot-snare EMR with submucosal injection in a piecemeal fashion (minimum
of two pieces per resection), of an NPCP sized 10–20 mm, and were reported as complete
resections, between January 1, 2014, and December 31, 2021, and subsequently underwent
surveillance colonoscopy within 3–9 months after primary resection. This timeframe
aligns with the national 2013 Dutch post-polypectomy guidelines and 2013 ESGE guidelines,
which recommended early repeat colonoscopy 6 months after pEMR [12]
[13]. Although NPCPs sized 20 mm are generally considered large NPCPs in polypectomy
guidelines [2]
[3], NPCPs reported as 20 mm were included in the study cohort to address potential
terminal digit bias, which frequently leads to an overrepresentation of reported lesion
sizes ending in 0 or 5 [14]. In addition, patient records were re-reviewed through August 31, 2024, to collect
additional information from surveillance colonoscopies conducted after ESC (SC2).
Patients with polyposis syndrome or known/suspected inflammatory bowel disease were
excluded.
Study variables collected
Data collection was performed at each study site using a standardized electronic data
collection form in CastorEDC (Ciwit B.V., Amsterdam, The Netherlands) with prespecified
criteria for coding. Study variables included patient demographics, study site, primary
pEMR characteristics (including year of resection, level of training [nurse endoscopist,
junior or senior endoscopist]), and whether the endoscopist was considered an EMR
expert, defined as performing an average of ≥30 EMRs or ESDs per year based on historical
data. Lesion location was categorized per colonic segment and dichotomized to proximal
colon (defined as proximal to the splenic flexure) or distal colon (defined as distal
to the splenic flexure). Additional variables included lesion size, Boston Bowel Preparation
Scale (BBPS) score, use of adjuvant modalities, presence of intraprocedural bleeding,
placement of a tattoo, and histopathological diagnosis based on the Vienna criteria
[15].
Surveillance colonoscopy characteristics included interval to surveillance colonoscopy,
level of training, identification of the post-EMR scar, identification of placed tattoos,
BBPS score, and histopathological diagnosis of recurrence or biopsies from the post-EMR
site. Data were collected on the presence of recurrence at the identified scar, or
advanced neoplasia, defined as lesions ≥10 mm, or those containing high grade dysplasia
(HGD) at SC2. The end of follow-up was defined as the detection of recurrence at ESC
or completion of SC2.
Primary and secondary outcomes
The primary outcome was the recurrence rate at ESC after pEMR of 10–20-mm NPCPs. Recurrence
rates were reported both for all colonoscopies and for cases where the scar was identified.
Recurrence was defined as the presence of macroscopic or microscopic tissue at the
post-EMR site, confirmed by histopathological diagnosis ([Fig. 1]). Secondary outcomes included the rate of identified post-EMR scars during ESC,
variables potentially associated with recurrence and scar identification, and the
presence of recurrence at the identified scar or advanced neoplasia within the same
colonic segment as the primary resection (cecum, ascending colon, transverse colon,
descending colon, sigmoid colon, or rectum) during SC2.
Fig. 1 Local recurrence after piecemeal endoscopic mucosal resection (pEMR) of nonpedunculated
colorectal polyps (NPCP). a A 15-mm NPCP located in the cecum, which was resected by pEMR. b Scar identified with recurrence at early surveillance colonoscopy.
Statistical analysis
Patient and procedural characteristics were summarized using descriptive statistics.
Categorical variables were presented as absolute frequencies and percentages, while
continuous variables were reported as means with SD or medians with interquartile
ranges (IQRs), as appropriate. Inferential comparisons were conducted using independent
t tests, Mann–Whitney U tests, or chi-squared tests, as appropriate. Wilson Score Interval was used to calculate
95%CIs where applicable.
Potential associations with recurrence and scar identification were assessed using
mixed-effects binary logistic regression models to account for clustering. Study site
was included as a random effect. Variables with a P value ≤0.20 in the univariable analysis for the outcome of interest were included
in the multivariable model. Backward elimination was used, sequentially removing the
variable with the highest P value until only variables with P ≤ 0.05 remained, while maintaining a maximum of one covariate per 10 events of the
outcome of interest.
Cumulative incidence functions were used to calculate the incidence of recurrence
or advanced neoplasia at SC2 for cases with and without scar identification at ESC.
To minimize potential bias related to the surveillance interval, cases with synchronous
lesions sized ≥20 mm at ESC were excluded. Competing risks analysis was performed
using Gray’s test to compare the incidence of recurrence or advanced neoplasia between
cases with and without scar identification at ESC. Competing events included death,
canceled SC2 due to comorbidities or advanced age, and transfer of care to another
hospital.
The level of statistical significance was set at P < 0.05, unless otherwise specified. No imputation of missing data was performed. Statistical
analysis was performed using Statistical Package for Social Sciences program, version
29 (IBM Corp., Armonk, New York, USA) and R Statistical Software (v.4.1.3; packages
“cmprsk” and “binom”; R Foundation for Statistical Computing, Vienna, Austria).
Results
Patient and procedural characteristics
A total of 389 patients with 426 NPCPs sized 10–20 mm who underwent pEMR and ESC were
included in the cohort. The median age was 68.0 years (IQR 63.0–72.0), and 219 patients
(56.3%) were male. NPCPs were resected by 103 endoscopists, with a median of 3 (IQR
1–6) pEMRs procedures per endoscopist. The median NPCP size was 15.0 mm (IQR 12.8–20.0
mm). Most NPCPs (80.3%; 342/426) were located in the proximal colon, and were resected
by senior endoscopists in 367 cases (86.2%) and by EMR experts in 98 cases (23%).
Adjuvant margin thermal ablation (MTA) with snare-tip soft coagulation was performed
for 67 NPCPs (15.7%) and argon plasma coagulation was used for 25 (5.9%). A tattoo
was placed at the primary pEMR site of 76 NPCPs (17.8%). Additional characteristics
are presented in [Table 1] (see also Table 1s–3s, and Fig. 1s in the online-only Supplementary material).
Table 1 Patient, polyp, and primary and early surveillance colonoscopy characteristics.
|
|
Overall
|
No recurrence at ESC
|
Recurrence at ESC
|
|
Data are presented as n (%) unless otherwise stated.
APC, argon plasma coagulation; BBPS, Boston Bowel Preparation Score; CRC, colorectal
cancer; ESC, early surveillance colonoscopy; ESD, endoscopic submucosal dissection;
FIT, fecal immunochemical test; IQR, interquartile range; MTA, margin thermal ablation;
NPCP, nonpedunculated colorectal polyp; (p)EMR, (piecemeal) endoscopic mucosal resection;
STSC, snare-tip soft coagulation.
¹Family history was not reported in 53 patients (no recurrence group n = 52; recurrence
group n = 1).
2Defined as one first-degree relative diagnosed before the age of 50 years, or one
first-degree relative diagnosed aged 50–70 years and one second-degree relative diagnosed
before the age of 70 years, or two or more first-degree relatives diagnosed aged 50–70
years.
3Proximal colon was defined as all segments proximal to the splenic flexure.
4EMR expert was defined as an endoscopist with a minimum of 30 performed EMRs or ESDs
per year on average, based on historic data.
5Median calculated on the total number of cases (n = 93) in which number of pieces
per resection was reported.
6BBPS score was not reported for 73 patients (no recurrence group n = 63; recurrence
group n = 10).
|
|
Patient information
|
n = 389
|
n = 355
|
n = 34
|
|
Age, median (IQR), years
|
68.0 (63.0–72.0)
|
68.0 (63.0–72.0)
|
68.0 (63.0–73.3)
|
|
Sex
|
|
|
219 (56.3)
|
199 (56.1)
|
20 (58.8)
|
|
|
170 (43.7)
|
156 (43.9)
|
14 (41.2)
|
|
Family history1
|
|
|
30 (8.9)
|
28 (9.2)
|
2 (6.1)
|
|
|
8 (2.4)
|
8 (2.6)
|
0 (0)
|
|
Reason for colonoscopy
|
|
|
7 (1.8)
|
7 (2.0)
|
0 (0)
|
|
|
159 (40.9)
|
143 (40.3)
|
16 (47.1)
|
|
|
120 (30.8)
|
109 (30.7)
|
11 (32.4)
|
|
|
49 (12.6)
|
47 (13.2)
|
2 (5.9)
|
|
|
49 (12.6)
|
44 (12.4)
|
5 (14.7)
|
|
|
5 (1.3)
|
5 (1.4)
|
0 (0)
|
|
NPCP and procedural information
|
n = 426
|
n = 391
|
n = 35
|
|
Lesion size, median (IQR), mm
|
15.0 (12.8–20.0)
|
15.0 (12.0–20.0)
|
15.0 (15.0–18.0)
|
|
Location of primary lesion
|
|
|
342 (80.3)
|
318 (81.3)
|
24 (68.6)
|
|
|
84 (19.7)
|
73 (18.7)
|
11 (31.4)
|
|
pEMR performed by
|
|
|
367 (86.2)
|
338 (86.4)
|
29 (82.9)
|
|
|
49 (11.5)
|
43 (11.0)
|
6 (17.1)
|
|
|
10 (2.3)
|
10 (2.6)
|
0 (0.0)
|
|
|
98 (23.0)
|
93 (23.8)
|
5 (14.3)
|
|
No. of pieces per pEMR, median (IQR)5
|
2.0 (2.0–3.0)
|
2.0 (2.0–3.0)
|
3.0 (2.0–3.3)
|
|
Intraprocedural bleeding, yes
|
55 (12.9)
|
54 (13.8)
|
1 (2.9)
|
|
Post-EMR clip placement, yes
|
116 (27.2)
|
108 (27.6)
|
8 (22.9)
|
|
Use of MTA
|
|
|
67 (15.7)
|
64 (16.4)
|
3 (8.6)
|
|
|
25 (5.9)
|
24 (6.1)
|
1 (2.9)
|
|
Tattoo placed, yes
|
76 (17.8)
|
67 (17.1)
|
9 (25.7)
|
|
ESC information
|
n = 426
|
n = 391
|
n = 35
|
|
Time to ESC, median (IQR), weeks
|
27.0 (23.8–31.0)
|
27.0 (24.0–31.0)
|
27.0 (20.0–34.0)
|
|
ESC performed by
|
|
|
314 (73.7)
|
285 (72.9)
|
29 (82.9)
|
|
|
76 (17.8)
|
70 (17.9)
|
6 (17.1)
|
|
|
36 (8.5)
|
36 (9.2)
|
0 (0)
|
|
|
97 (22.8)
|
88 (22.5)
|
9 (25.7)
|
|
BBPS of segment of post-pEMR site6
|
|
|
248 (70.3)
|
226 (68.9)
|
22 (88.0)
|
|
|
98 (27.8)
|
95 (29.0)
|
3 (12.0)
|
|
|
6 (1.7)
|
6 (1.8)
|
0 (0)
|
|
|
1 (0.3)
|
1 (0.3)
|
0 (0)
|
|
Biopsy of post-pEMR scar
|
23 (8.8)
|
18 (7.9)
|
5 (14.3)
|
Scar identification and recurrence rates at ESC
ESC was performed after a median interval of 27.0 weeks (IQR 23.8–31.0) following
the
primary pEMR ([Table 1]). The post-pEMR scar was identified in 262 of 426 colonoscopies performed at ESC
(61.5%; 95%CI 56.8–66.0) ([Table 2]). Scar identification rate was 81.6% (62/769; 95%CI 71.4–88.7) for tattooed
resection sites compared with 57.1% (200/350; 95%CI 51.9–62.2) for non-tattooed resection
sites (P<0.001). Overall, recurrence at the post-pEMR site was
observed in 35 of 426 NPCPs (8.2%; 95%CI 6.0–11.2), with a median recurrence size
of 5 mm
(IQR 1–7). Recurrence rates were 9.0% (26/289; 95%CI 6.2–12.9) for adenomas and 7.0%
(9/128;
95%CI 3.7–12.8) for serrated lesions. Recurrence was observed in 4 of 92 lesions treated
with MTA (4.3%; 95%CI 1.7–10.7), and in 31 of 334 lesions not treated with MTA (9.3%;
95%CI
6.6–12.9). Recurrent lesions contained no HGD and had no HGD in their primary resection.
In
colonoscopies with a post-pEMR scar identified (n = 262), recurrences were observed
in 35
(13.4%; 95%CI 9.8–18.0). Additional results are reported in [Table 2], and Tables 4s–8s.
Table 2 Overall recurrence rates, recurrence rates per adenomatous or serrated lesions, and
scar identification rates at early surveillance colonoscopy.
|
|
Early surveillance colonoscopy
|
|
Data are presented as n/N (%) [95%CI] unless otherwise stated. 95%CI was calculated
with Wilson Score Interval.
HGD, high grade dysplasia; IQR, interquartile range; NPCP, nonpedunculated colorectal
polyp; MTA, margin thermal ablation.
1Serrated lesions included sessile serrated lesions (n = 105), hyperplastic polyps
(n = 17), and traditional serrated adenomas (n = 6).
2Other lesions (n = 9) included NPCPs with indefinite or missing histopathologic diagnosis.
3Size of recurrence was reported in 19 of 30 cases with macroscopic recurrence.
|
|
Overall recurrence
|
35/426 (8.2) [6.0–11.2]
|
|
Recurrence of adenomas
|
26/289 (9.0) [6.2–12.9]
|
|
Recurrence of serrated lesions1
|
9/128 (7.0) [3.7–12.8]
|
|
Recurrence of NPCPS with HGD
|
0/35 (0) [0.0–9.9]
|
|
Recurrence of other NPCPs2
|
0/9 (0) [0.0–29.9]
|
|
Recurrence of NPCPs 16–20 mm
|
10/171 (5.8) [3.2–10.4]
|
|
Recurrence of NPCPs 10–15 mm
|
25/255 (9.8) [6.7–14.1]
|
|
Recurrence of NPCPs treated with MTA
|
4/92 (4.3) [1.7–10.7]
|
|
Recurrence of NPCPs treated without MTA
|
31/334 (9.3) [6.6–12.9]
|
|
Recurrence size,3 median (IQR), mm
|
5.0 (1.0–7.0)
|
|
Identification of post-pEMR scar
|
262/426 (61.5) [56.8–66.0]
|
|
Identification of tattooed resection sites
|
62/76 (81.6) [71.4–88.7]
|
|
Identification of non-tattooed resection sites
|
200/350 (57.1) [51.9–62.2]
|
|
Overall recurrence of cases with scar identification
|
35/262 (13.4) [9.8–18.0]
|
Associations with recurrence and scar identification at ESC
Our mixed-effects binary regression model analysis found no variables significantly
associated with recurrence at ESC in either univariable or multivariable analysis
([Table 3]). The mixed-effects model found that tattoo placement (odds ratio [OR] 3.50; 95%CI
1.81–6.77), a segmental BBPS score of 3 vs. ≤2 at ESC in the same segment as the primary
resection (OR 2.07; 95%CI 1.28–3.37), and post-EMR clip placement (OR 2.13; 95%CI
1.28–3.56) were independently and statistically significantly associated with scar
identification at ESC ([Table 4]).
Table 3 Binary logistic mixed-effects model regression results for variables associated with
local recurrence at early surveillance colonoscopy.
|
|
Univariable analysis
|
Multivariable analysis1
|
|
OR (95%CI)
|
P
|
OR (95%CI)
|
P
|
|
APC, argon plasma coagulation; CRC, colorectal cancer; ESC, early surveillance colonoscopy;
ESD, endoscopic submucosal dissection; FIT, fecal immunochemical test; HGD, high grade
dysplasia; MTA, margin thermal ablation; OR, odds ratio; (p)EMR, (piecemeal) endoscopic
mucosal resection; STSC, snare-tip soft coagulation.
1Study site was used as a random effect in this mixed-effects model to account for
site-level clustering. Variables with a P value ≤0.20 in the univariable analysis were included in the multivariable analysis.
Backward elimination was applied, removing variables sequentially by P value until only those with P ≤ 0.05 remained. As no variables met this final inclusion criterion, the first step
of the model is presented. Due to the limited number of recurrence cases, the number
of variables was restricted to a maximum of three in the model.
2Analysis was performed on the 370 cases with data on family history.
3Proximal colon defined as all segments proximal to the splenic flexure.
4EMR expert was defined as an endoscopist with a minimum of 30 performed EMRs or ESDs
per year on average, based on historic data.
5Tertiles of performed pEMRs per endoscopist were categorized based on the 33rd and
66th percentile.
6Analysis was performed on the 93 cases with data on the number of pieces per resection.
|
|
Age at primary resection, years
|
1.02 (0.98–1.08)
|
0.35
|
|
|
|
Interval to ESC
|
1.00 (0.95–1.06)
|
0.89
|
|
|
|
Female sex
|
0.81 (0.40–1.64)
|
0.55
|
|
|
|
Reason for colonoscopy
(ref: FIT+ screening)
|
|
|
|
|
|
|
0.87 (0.39–1.91)
|
0.72
|
|
|
|
|
0.33 (0.08–2.90)
|
0.20
|
|
|
|
|
0.97 (0.33–2.90)
|
0.96
|
|
|
|
Positive family history for CRC2
(ref: no family history)
|
0.63 (0.14–2.80)
|
0.55
|
|
|
|
Primary lesion in proximal colon
(ref: distal colon)3
|
0.54 (0.25–1.16)
|
0.11
|
0.55 (0.26–1.20)
|
0.14
|
|
Primary pEMR performed by junior endoscopist (ref: senior endoscopist)
|
1.28 (0.48–3.46)
|
0.62
|
|
|
|
pEMR performed by EMR expert4
|
0.48 (0.18–1.33)
|
0.16
|
0.50 (0.18–1.38)
|
0.18
|
|
No. of pEMR performed per endoscopist (ref: bottom tertile, 1–5 pEMRs)5
|
|
|
0.78 (0.32–1.89)
|
0.58
|
|
|
|
|
0.73 (0.32–1.68)
|
0.46
|
|
|
|
Primary lesion size, mm
|
1.00 (0.91–1.10)
|
0.96
|
|
|
|
No. of pieces per resection6
|
1.37 (0.55–3.37)
|
0.49
|
|
|
|
Intraprocedural bleeding
|
0.17 (0.23–1.28)
|
0.09
|
|
|
|
Post-pEMR clip placement
|
0.75 (0.31–1.80)
|
0.51
|
|
|
|
Adjuvant use of MTA
|
0.42 (0.14–1.24)
|
0.12
|
0.44 (0.15–1.29)
|
0.13
|
|
|
0.46 (0.13–1.57)
|
0.21
|
|
|
|
|
0.42 (0.05–3.25)
|
0.41
|
|
|
|
Serrated lesion
|
0.81 (0.37–1.80)
|
0.61
|
|
|
|
HGD
|
0.68 (0.09–5.43)
|
0.72
|
|
|
Table 4 Binary logistic mixed-effects model regression results for variables associated with
scar identification at early surveillance colonoscopy.
|
|
Univariable analysis
|
Multivariable analysis1
|
|
OR (95%CI)
|
P
|
OR (95%CI)
|
P
|
|
BBPS, Boston Bowel Preparation Scale; ESC, early surveillance colonoscopy; HGD, high
grade dysplasia; MTA, margin thermal ablation; OR, odds ratio; (p)EMR, (piecemeal)
endoscopic mucosal resection.
1Study site was used as a random effect in this mixed-effects model to account for
site-level clustering. Variables with a P value ≤0.20 in the univariable analysis were included in the multivariable analysis.
Backward elimination was applied, removing variables sequentially by P value until only those with P ≤ 0.05 remained.
273 cases with missing values were not included in the analysis.
|
|
Interval to ESC, weeks
|
1.00 (0.97–1.04)
|
0.89
|
|
|
|
Primary lesion size, mm
|
1.09 (1.03–1.15)
|
0.003
|
|
|
|
Tattoo placed, yes
|
3.16 (1.69–5.90)
|
<0.001
|
3.50 (1.81–6.77)
|
<0.001
|
|
BBPS score of 3 at ESC2
(ref: BBPS of ≤2)
|
1.95 (1.22–3.11)
|
0.005
|
2.07 (1.28–3.37)
|
0.003
|
|
ESC performed by
(ref: senior endoscopist)
|
|
|
|
|
|
Junior endoscopist
|
0.69 (0.37–1.29)
|
0.25
|
|
|
|
Nurse endoscopist
|
0.66 (0.32–1.38)
|
0.27
|
|
|
|
ESC performed by EMR expert
|
1.18 (0.73–1.92)
|
0.51
|
|
|
|
ESC performed by the same endoscopist as the primary pEMR
|
0.99 (0.61–1.61)
|
0.96
|
|
|
|
Location of primary lesion
(ref: rectum)
|
|
|
|
|
|
Cecum
|
0.998 (0.42–2.35)
|
0.99
|
|
|
|
Ascending colon
|
0.60 (0.26–1.37)
|
0.22
|
|
|
|
Hepatic flexure
|
0.36 (0.12–1.06)
|
0.06
|
|
|
|
Transverse colon
|
0.47 (0.20–1.12)
|
0.09
|
|
|
|
Splenic flexure
|
1.21 (0.20–7.42)
|
0.84
|
|
|
|
Descending colon
|
0.83 (0.26–2.71)
|
0.76
|
|
|
|
Sigmoid
|
0.88 (0.28–2.75)
|
0.83
|
|
|
|
Post-pEMR clip placement
|
1.60 (1.01–2.54)
|
0.047
|
2.13 (1.28–3.56)
|
0.004
|
|
Adjuvant use of MTA
|
1.52 (0.92–2.51)
|
0.10
|
|
|
|
Serrated lesion
|
1.23 (0.79–1.91)
|
0.36
|
|
|
|
HGD in primary lesion
|
0.94 (0.32–2.72)
|
0.90
|
|
|
Recurrence or advanced neoplasia at surveillance colonoscopy after ESC
After exclusion of cases with recurrence (n = 35) and synchronous lesions sized ≥20
mm at ESC (n = 15), 376 cases remained in the post-ESC surveillance cohort. Of these
cases, 244 (65%) underwent SC2 and 44 cases (12%) had not yet reached SC2 (Fig. 2s). The median interval between ESC and SC2 was 32 months (IQR 14–40). The overall
scar identification rate at SC2 was 86 of 244 cases (35.2%; 95%CI 29.5–41.4). The
scar identification rate was increased in cases with scar identification at ESC (64/148,
43.2%; 95%CI 35.5–51.3) compared with cases without scar identification at ESC (22/96,
22.9%; 95%CI 15.6–32.2; P = 0.001) (Table 9s).
The cumulative incidence of recurrence or same-segment advanced neoplasia at 5 years
after ESC was 7.5% (95%CI 3.6–11.3) for cases with scar identification during ESC
and 8.5% (95%CI 3.0–13.9) for cases without scar identification. The cumulative incidence
for only recurrence during SC2 at 5 years after ESC was 5.3% (95%CI 2.0–8.5) for cases
with scar identification during ESC and 5.3% (95%CI 1.4–9.2) for cases without scar
identification. Our competing risk model indicated that the overall rates of recurrence
and advanced neoplasia were similar between cases with and without scar identification
at ESC (P = 0.97). Competing events included cancelation of SC2 due to advanced age (11 vs.
10 cases), comorbidities (6 vs. 5 cases), non-CRC-related death (3 vs. 6 cases), and
transfer of care to another hospital (5 vs. 1 case), for cases with and without scar
identification, respectively ([Fig. 2], Figs. 2s and 3s).
Fig. 2 Cumulative incidence function for advanced neoplasia and recurrence at subsequent
surveillance colonoscopies according to scar identification at early surveillance
colonoscopy.
Discussion
In this multicenter cohort study, we found an overall recurrence rate of approximately
8% at ESC after pEMR of NPCPs sized 10–20 mm. The post-pEMR scar at the primary resection
site was identified in only 62% of cases, with a corresponding recurrence rate of
13% at ESC, suggesting that the true recurrence rate may lie within this range. We
found no NPCP or procedural characteristics significantly associated with recurrence.
Our findings are similar to the results of a recent, single-center, retrospective
study investigating recurrence after pEMR of NPCPs sized 10–19 mm compared with those
sized ≥20 mm [16]. This study reported recurrence rates of 10.5% for NPCPs sized 10–19 mm and approximately
22% for lesions sized ≥20 mm, and identified lesion size as the only independent variable
associated with recurrence. Notably, patients without post-pEMR scars identified were
excluded, and MTA of the resection site was performed in approximately 40%. In comparison,
a 2020 meta-analysis by Djinbachian et al. reported a 20% incomplete resection rate
for lesions sized 10–20 mm resected with submucosal injection [6]. However, the studies in this meta-analysis were predominantly included before 2019
and reported no data about MTA use.
During our study period (2014–2021), the recurrence rate decreased over time (Fig. 1s). This decline may be attributable to the inclusion of relatively fewer cases from
sites with initially higher recurrence rates later in the study period, although a
reduction in recurrence rates was also observed in these high-recurrence centers.
While our mixed-effects regression analysis was underpowered to detect significant
associations, likely due to a low number of recurrences and high variability among
endoscopists (103 in total), certain variables showed trends, including EMR expertise
and MTA use. The latter may be a relevant factor, as the role of MTA in reducing recurrence
is supported by studies from Klein et al. (2019) and Sidhu et al. (2021), which reported
recurrence rates of 5.2% (5.4% for piecemeal resections) and 1.4%, respectively, for
larger NPCPs (≥20 mm) when MTA was applied [17]
[18]. Additionally, a recent meta-analysis investigating recurrence rates after EMR of
NPCPs >15 mm, reported an OR of 0.18 (95%CI 0.13–0.26) for recurrences with adjuvant
MTA after EMR compared with EMR alone [19]. In our study, MTA was used in only 21% of cases, with recurrence rates of approximately
4% for lesions treated with MTA and 9% for lesions without MTA treatment. Although
the use of MTA increased during the study period, possibly contributing to the observed
decline in recurrence rates, it was not statistically significant in the regression
analysis. The ESGE guidelines recommend MTA for larger NPCPs (≥20 mm); however, this
recommendation has not yet been extended to NPCPs of 10–20 mm [2]. Nonetheless, hypothetically, the underlying mechanism of action should remain consistent
across lesion sizes, and the potential benefits of adjuvant margin ablation should
likely be extended to pEMR of NPCPs of 10–20 mm. To our knowledge, no randomized studies
have yet investigated the effect of margin ablation after (piecemeal) EMR of NPCPs
in this size range. Therefore, as MTA is increasingly used in daily clinical practice,
future research should evaluate the efficacy of MTA not only in reducing recurrence
for large (≥20 mm) lesions but also in (piecemeal) resections of smaller lesions.
For our recurrence rate, we considered both microscopic and macroscopic recurrences.
Biopsies were taken in fewer than 10% of cases at ESC, with varying rationale, but
predominantly yielding normal mucosa. This aligns with previous studies that have
demonstrated that thorough inspection is generally sufficient for optical scar diagnosis
[20]
[21]. Remarkably, no recurrences were identified in biopsies taken at non-teaching sites,
suggesting potential challenges with optical scar diagnosis or a more conservative
biopsy approach. However, these findings should be interpreted with caution due to
the low biopsy rate and limited sample size.
In our study, the post-pEMR scar was only identified in approximately 60% of the colonoscopies
at ESC. This moderate identification rate contrasts with scar identification rates
ranging from 93% to 99.7% in studies evaluating optical assessment of post-polypectomy
scars after EMR of lesions sized ≥15 or ≥20 mm, without tattooing [20]
[22]
[23]. These higher identification rates can likely be attributed to studies conducted
in expert centers, with a focus on scar recognition and typically involving larger
lesions. While tattoo placement was associated with higher scar identification rates
in our study, increasing identification to approximately 82% from 57%, additional
studies are needed to evaluate the utility of tattoo placement specifically for 10–20-mm
NPCPs.
Our findings suggest that scar identification at ESC did not influence rates of recurrence
or advanced neoplasia at subsequent (second) surveillance colonoscopy (SC2). This
suggests that when a thorough inspection at ESC does not result in scar identification,
the likelihood of a missed recurrence appears low, as similar rates of recurrence
and advanced neoplasia were observed at SC2 regardless of scar identification at ESC.
Our study design precluded reliable data collection on clinically relevant outcomes
such as post-colonoscopy CRC. However, prior studies indicate that while piecemeal
or incomplete resections are associated with post-colonoscopy CRC, the absolute risk
following incomplete resection remains low [24]
[25]. Given this low absolute risk and the absence of CRC or HGD at SC2 in our cohort,
regardless of scar identification at ESC, the likelihood of missing clinically relevant
recurrences appears minimal. Moreover, recurrences in our study were typically small
and without HGD, allowing for efficient treatment using conventional endoscopic techniques,
such as hot snare polypectomy with snare-tip soft coagulation or cold-forceps avulsion
with adjuvant snare-tip soft coagulation, as demonstrated in a recent study [26], thereby mitigating further potential malignant progression.
While our study reported a substantial recurrence rate, the timing and necessity of
ESC should be carefully balanced against individual patient characteristics and overall
use of endoscopy resources. Given that most recurrences were small and without HGD,
extending the ESC interval beyond the timeframe of our study (9 months) may still
allow for the detection of metachronous lesions. Furthermore, from a sustainability
perspective, careful consideration of appropriateness of endoscopic procedures is
considered one of the most important factors in mitigating the environmental impact
of gastrointestinal endoscopy [27]. Additionally, nearly a quarter of our patients did not undergo SC2 due to factors
such as comorbidities or advanced age, raising questions about the clinical benefit
of performing ESC in these cases. Ultimately, the decision to perform ESC should be
based on the likelihood of clinically relevant recurrence, patient-specific characteristics,
and evolving demands on endoscopy services.
The strengths of this study include the multicenter approach, with both teaching and
non-teaching hospitals participating. To our knowledge, this is the first multicenter
study specifically addressing the recurrence rate after pEMR of NPCPs sized 10–20
mm. Additionally, as our inclusion period aligned with the implementation of the 2013
Dutch national and ESGE post-polypectomy guidelines, our real-world cohort reflects
the yield of performing early repeat ESC to identify the post-pEMR scar and potential
recurrences [12]
[13].
Our study also has some limitations. First, our methodology of patient selection and
differences in the accessibility of electronic medical records across study hospitals
prevented the inclusion of patients who underwent pEMR of NPCPs sized 10–20 mm and
were referred to standard surveillance recommendations for low-risk lesions [28]
[29]. This design may have introduced bias, as endoscopists might have chosen ESC over
a longer interval for patients suspected of being at higher recurrence risk, potentially
increasing the proportion of lesions with an increased recurrence risk in our study.
Second, our regression analysis for recurrence was underpowered due to the low absolute
incidence of recurrence in our cohort, as reflected by relatively wide confidence
intervals, and should be interpreted with caution. Third, our study included NPCPs
ranging from 10 to 20 mm, contrary to current ESGE and USMSTF guidelines, which adopt
a 20-mm cutoff and recommend a 6-month follow-up for piecemeal resections of polyps
sized ≥20 mm [2]
[3]
[28]
[29]. However, we included this size range to address the risk of terminal digit bias
in reporting polyp sizes, particularly considering the retrospective design of our
study.
In conclusion, our study found a substantial recurrence rate after pEMR of NPCPs sized
10–20 mm at ESC. Although this recurrence rate was relatively high, most recurrences
were small and without advanced neoplasia. This suggests that, when deciding to perform
ESC, modestly extending the surveillance interval beyond that of our study may be
appropriate to detect potential metachronous lesions. Additionally, although scar
identification at ESC was moderate, our SC2 findings suggest that the incidence of
late recurrence or clinically relevant advanced neoplasia was comparable regardless
of scar identification at ESC.
