Introduction
Removal of adenomatous polyps during colonoscopy decreases the risk of death from
colorectal cancer [1]. Endoscopic mucosal resection (EMR) is an effective modality for removal of large
and sessile/flat polyps. Endoscopic submucosal dissection (ESD) and surgery are potential
alternatives to EMR. En bloc resection rate is better with ESD compared to EMR but
ESD is associated with increased risk of perforation and requires a higher level of
expertise [2]. Surgery for colorectal polyps is associated with higher risks of morbidity and
mortality and increased costs compared to EMR [3]
[4].
Conventional endoscopic mucosal resection (EMR) with submucosal injection and after
insufflating the colon with air or carbon dioxide is currently the most commonly performed
EMR technique. Underwater endoscopic mucosal resection (UEMR) is a newer alternative
to EMR for removal of large sessile colorectal polyps [5]. Initially reported in 2012, it has since shown promising results. UEMR is performed
by suctioning out gas from the colonic lumen and instilling water immediately before
polyp resection [6]. Filling the colon lumen with water instead of gas decreases wall tension and has
a buoyancy effect on the mucosa and submucosa raising them above the muscularis layer
without the need for submucosal injection [7]. This changes the borders and shape of a lesion, potentially making it easier to
resect completely with snaring. A recent meta-analysis evaluating the efficacy of
UEMR reported a complete resection rate of 96 % and an en bloc resection rate of 57 %
[8]; however, this analysis was confined to single-arm non-comparative studies. UEMR
has also shown better results compared to EMR in the management of recurrent polyps
after prior piecemeal EMR [9]. These polyps are often difficult to resect en bloc due to fibrosis from prior attempts
at polyp resection. Studies comparing EMR and UEMR have reported conflicting results.
Therefore, we conducted a systematic review and meta-analysis to compare UEMR and
EMR for resection of flat and sessile colorectal polyps.
Methods
Data sources and search strategy
We followed the Preferred Reporting items for Systematic Review and Meta-Analysis
(PRISMA) guidelines to conduct this systematic review and meta-analysis [10]. We performed a comprehensive search of several databases including MEDLINE using
the PubMed interface (1950–present), Embase (1947–present), Web of Science Core Collection
(1965–present), and Cochrane Register of Controlled Trails (1966-present) from inception
through 11/12/2019. The search strategy was designed by study investigators (F.K.
and Z.K.) and was refined and conducted by an experience medical librarian (W.L.-S.)
using keywords and controlled vocabularies (MeSH and Emtree headings) for the topics
of endoscopic mucosal resection, underwater, polyps, and colon. A sample search is
included from Embase in Appendix A. There was no limitation of language in conducting the search. Publication types
of editorial, guidelines, case reports, and reviews were excluded when possible. The
keywords used in search included “Endoscopic-mucosal-resection” OR “EMR” OR “endoscopic-aspiration-mucosectomy”
OR “endoscopic-mucosa-resection” OR endoscopic-mucosectomy” OR “endoscopic mucosal
resection” AND “water” OR “underwater” OR “UEMR” OR “U-EMR” OR ”water” AND “polyp”
OR “polyps” OR “polypoid-lesion” AND “colon” OR “colorectal” OR “colonic” OR “rectum”
OR “rectal” OR “colonic-neoplasm” OR ‘intestine polyp’/exp.
Two authors (F.K. and M.A.K.) independently reviewed the titles and abstracts of studies
retrieved from the search to identify those that fulfilled inclusion criteria. To
maximize the yield of the search, the reference lists of identified studies were also
reviewed to identify any additional relevant studies. Search strategy is illustrated
in [Fig. 1].
Fig. 1 PRISMA flowchart. From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group
(2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA
Statement. PLoS Med 6 (7): e1000097. doi:10.1371/journal.pmed1000097
Two authors (F.K. and M.A.K.) searched for observational studies and randomized controlled
trials (RCTs) comparing the efficacy and/or safety of UEMR and EMR for resection of
sessile colorectal polyps. Non-comparative studies and studies with patients < 18
years of age or with animal data were excluded. All articles were downloaded into
Endnote 7.0, a bibliographic database manager. Duplicate citations were removed. We
included both fully published studies and abstracts.
Outcomes of interest and inclusion criteria
The outcomes of interest included rates of en bloc resection, complete macroscopic
resection (defined by complete endoscopic resection of polyp with absence of any macroscopically
visible polyp fragments on endoscopic views at the end of resection), recurrent or
residual polyps on follow-up colonoscopy, complete resection confirmed by histology,
and adverse events (AEs). Regarding the outcome of “complete resection confirmed by
histology”, two studies reported data as R0 resection defined as en bloc resection
with histologically confirmed negative margins [11]
[12]. One study reported data on incomplete resection rate based on pathologic assessment
of biopsies from the resection margin of polyps [13]. Incomplete resection was defined as the presence of any adenomatous or serrated
pathology in the resection margin. We assumed that the rest of the polyps had been
completely resected and confirmed by histology.
Our analysis only included sessile and flat polyps ≥ 10 mm that were resected using
UEMR. We excluded pedunculated polyps that were resected underwater. Sessile or pedunculated
polyps < 10 mm resected underwater were also excluded.
Data extraction
Two authors (F.K. and S.S.) assessed the eligibility of selected studies and extracted
data using data extraction forms specifically designed for this study. Any disagreement
between the authors was discussed with a third author (K.F.B.) and agreement was reached
by consensus. Extracted data included study design, year and country of publication,
patient demographics, inclusion criteria, exclusion criteria and – for each group
– rates of en bloc resection, piecemeal resection, complete macroscopic resection,
recurrent/residual polyps on follow-up colonoscopy, as well as resection times of
polyps and AEs ([Table 1] and [Table 2]).
Table 1
Characteristics of included studies.
|
Study, year
|
Country, Single vs multicenter
|
Type of study
|
Total number of patients
|
Males
|
Number of operators
|
Inclusion criteria
|
Exclusion criteria
|
|
Schenck et al, 2017
|
USA, Single center
|
Retrospective, cohort
|
99
|
62
|
NR
|
Patients who had EMR or UEMR of a polyp ≥ 15 mm in the colon or rectum, which had
not undergone prior attempted endoscopic resection.
|
NR
|
|
Cadoni et al, 2017
|
Italy, Taiwan, UK, Multicenter
|
Retrospective, cohort
|
287
|
190
|
4
|
Patients > 18 years who underwent colonoscopy from January 2015 – December 2016 with
polypectomy. Sessile, flat and pedunculated polyps ≥ 6 mm removed by hot snare either
using gas insufflation polypectomy (GIP) and submucosal injection or underwater.
|
NR
|
|
Yamashina et al, 2019
|
Japan, Multicenter
|
RCT
|
210
|
139
|
28
|
Patients ≥ 20 years undergoing endoscopic resection for colorectal mucosal lesions
that were 10–20 mm in diameter
|
Pedunculated lesions; residual lesions after endoscopic resection; and lesions in
patients with inflammatory bowel disease, familial polyposis, electrolyte abnormality,
coagulopathy, or severe organ failure
|
|
Kim et al, 2014
|
USA, Single center
|
Retrospective cross-sectional
|
80
|
50
|
1
|
Patients referred to an interventional colonoscopy clinic for recurrence after piecemeal
EMR of colorectal laterally spreading tumors (LSTs)
|
Lesions smaller than 8 mm were excluded
|
|
Yen et al, 2019
|
USA, Multicenter
|
RCT
|
255
|
248
|
1
|
All ≥ 18 years old patients scheduled for outpatient colonoscopy, Polyps > 5 mm in
size
|
Hospitalized patients, uninterrupted antithrombotic therapy at the time of colonoscopy,
uncorrected coagulopathy (INR > 1.5) or thrombocytopenia (platelet count < 50,000),
significant comorbidities ASA classification ≥ 4, diminutive (< 5 mm) polyps, pedunculated
lesions, lesions with endoscopic evidence of deep submucosal invasion.
|
|
Rodriguez-Sanchez et al, 2019
|
Spain, Multicenter
|
Prospective cross-sectional
|
137
|
NR
|
2
|
All colorectal endoscopic resections of lesions > 15 mm performed with both techniques
(UEMR and CEMR) were included.
|
NR
|
|
Chien et al, 2019
|
USA, Japan, Taiwan, Multicenter
|
Retrospective cohort
|
223
|
NR
|
1
|
Age ≥ 20 years, medium to large lesions (≥ 10 mm) with Paris classification 0-Is,
IIa, or IIc morphology
|
NR
|
RCT, randomized controlled trial; NR, not reported; UEMR, underwater endoscopic mucosal
resection; ASA, American Society of Anesthesiologist; CEMR; conventional endoscopic
mucosal resection
Table 2
Patient characteristics and outcomes of interest.
|
Studies
|
Groups
|
No. patients in each group
|
No. polyps in each group
|
Polyp size in each group
|
En bloc resection rate
|
Piecemeal resection
|
Complete macroscopic resection
|
Recurrent/residual polyps
|
Polyp resection time
|
IPPB
|
DPPB
|
Perforation
|
|
Schenck et al, 2017
|
UEMR
EMR
|
46
53
|
73
62
|
20 (15–40)
20 (15–70)
|
NR
|
52
40
|
72
54
|
4/55
13/46
|
NR
|
0
0
|
3
0
|
0
0
|
|
Cadoni et al, 2017
|
UEMR
EMR
|
NR
|
81
77
|
10 (9.25–15)
10 (8–15)
|
58
51
|
23
26
|
NR
NR
|
NR
NR
|
2 (0.8–5)
3.3 (2.5–6)
|
10
11
|
1
0
|
0
0
|
|
Yamashina et al, 2019
|
UEMR
EMR
|
108
102
|
108
102
|
14 (7–25)
13.5 (7–25)
|
96
76
|
NR
|
NR
|
NR
|
2.75 (1.9–4.5)
2.9 (2.16–4.43)
|
0
0
|
3
2
|
0
0
|
|
Kim et al, 2014
|
UEMR
EMR
|
36
44
|
36
44
|
18.5 ± 14.3
16.9 ± 8.9
|
17
7
|
NR
|
NR
|
2/20
13/33
|
NR
|
0
0
|
0
2
|
0
0
|
|
Yen et al, 2019
|
UEMR
EMR
|
128
127
|
68
50
|
9.9 ± 5.8
9.9 ± 6.4
|
48
32
|
NR
|
NR
|
NR
|
3.8 (0.34)
5.4 (0.35)
|
5
3
|
0
0
|
0
0
|
|
Rodriguez-Sanchez et al, 2019
|
UEMR
EMR
|
137
|
50
112
|
20.78 (15–50)
30.38 (15–70)
|
31
55
|
19
45
|
50
100
|
1/27
16/111
|
NR
|
1
6
|
0
4
|
0
1
|
|
Chien et al, 2019
|
UEMR
EMR
|
115
108
|
121
121
|
17 ± 7.2
16.6 ± 6.5
|
106
100
|
15
21
|
|
|
8.6 ± 6.4
10.8 ± 8.3
|
7
19
|
1
1
|
1
0
|
NR, not reported; UEMR, underwater endoscopic mucosal resection; CEMR, conventional
endoscopic mucosal resection; IPPB: immediate post-polypectomy bleeding; DPPB, delayed
post-polypectomy bleeding
Statistical analysis
Data were analyzed using a fixed effect model and summarized as pooled risk ratio
(RR) with 95 % confidence interval (CI). P < 0.05 was considered statistically significant. P < 0.1 for Cochran Q test or an I2 > 50 % indicated significant heterogeneity. Publication bias was assessed using funnel
plots and Egger’s test. When publication bias was detected, Duval and Tweedie’s trim
and fill test was used to report adjusted effect size. All statistical analyses were
performed using Review Manager (RevMan, version 5.3 for Windows; The Cochrane Collaboration,
The Nordic Cochrane Centre, Copenhagen, Denmark, 2014).
Predetermined subgroup analyses were conducted to evaluate randomized controlled trials
(RCTs) separately from observational studies. Sensitivity analyses were performed
based on inclusion and exclusion of outlier studies where the study population differed
substantially from other studies to explain heterogeneity. One study [9] was considered to be an outlier. It compared UEMR and EMR in the management of recurrent
polyps after prior attempts at resection. Additionally, some polyps in that study
might have been < 10 mm, although most were > 10 mm; median polyp sizes were 12 mm
and 14 mm in the UEMR and EMR groups, respectively.
Results
Search strategy yield and quality assessment
The search strategy yielded 139 articles, of which 61 were removed as duplicates.
Of the remaining 78 studies, 43 were excluded after title and abstract review. Bibliographic
review of the remaining 35 revealed three additional studies. Therefore, 38 articles
underwent full text review. Finally, seven studies [9]
[11]
[12]
[13]
[14]
[15]
[16] with 1291 patients met our inclusion criteria and were included in analysis. 537
polyps were removed by UEMR and 568 by EMR. Two studies [11]
[13] were RCTs; the remaining five were observational [9]
[12]
[14]
[15]
[16]. [Table 1] summarizes the characteristics of included studies. [Table 2] summarizes data of outcomes of interest. Cadoni et al [12] had also presented data on pedunculated polyps which we did not include in our analysis.
Quality assessment
We used the Newcastle-Ottawa scale (NOS) for assessment of methodological quality
of observational studies, and the Cochrane tool for assessing risk of bias for RCTs.
The Cochrane tool assesses the presence of selection bias by evaluating the methods
of randomization and allocation concealment; performance and detection of biases by
checking for blinding of personnel and outcome assessment, respectively; attrition
and reporting bias by evaluating for incomplete and selective reporting of data, respectively.
In the RCTs included in our meta-analysis, the blinding of endoscopists to treatment
allocation was not possible. Therefore, all RCTs had high risk of performance bias.
However, all RCTs had low risk of selection, detection, attrition and reporting biases.
The risk of bias assessment of RCTs is summarized in Supplementary Table 1.
NOS measures quality in the three parameters of selection, comparability and outcome,
and allocates a maximum of 4, 2, and 3 points to each, respectively. High-quality
studies score more than 7, moderate-quality studies score between 5 and 7 and low-quality
studies score less than 5 on this scale. Four observational studies included in our
analysis were of high quality and one study was of moderate quality. The quality assessment
of observational studies is summarized in Supplementary Table 2. Two authors (F.K. and M.A.K.) independently performed the quality assessment and
any disagreement was discussed with a third author (C.W.H.).
Meta-analysis
En bloc resection
Six studies [9]
[11]
[12]
[13]
[15]
[16] reported data on en bloc resection for 464 polyps removed by UEMR and 506 by EMR.
Rates with UEMR and EMR were 77 % and 63 %, respectively. We found that the rate of
en bloc resection was significantly higher (better) with UEMR compared to EMR; Pooled
RR (95 % CI) 1.16 (1.08, 1.26), Cochran Q test P = 0.08, I2 = 49 % ([Fig. 2a]). No publication bias was detected based on funnel plot (Supplementary Fig. A). Subgroup analysis confined to RCTs still showed statistically significant higher
en bloc resection rate with UEMR; pooled RR (95 % CI) 1.16 (1.03, 1.31), I2 = 0 %. Subgroup analysis including observational studies showed similar results;
pooled RR (95 % CI) 1.16 (1.05, 1.29), I2 = 69 %.
Fig. 2 Forest plot to compare en bloc resection rate and recurrent/residual polyps between
UEMR and EMR.
Sensitivity analysis excluding the study by Kim et al again showed significantly higher
en bloc resection rate with UEMR; pooled RR (95 % CI) 1.13 (1.04, 1.22) and heterogeneity
decreased substantially (I2 = 0 %). Sensitivity analysis by excluding one study [15] where mean polyp size in the EMR group (30 mm) was larger than UEMR group (20 mm)
also showed consistent results, Pooled RR (95 % CI) 1.15 (1.06, 1.25).
Recurrent/residual polyps
Three observational studies [9]
[14]
[15] reported data on recurrent/residual polyps on follow-up colonoscopy. One hundred
two polyps had been removed by UEMR and 190 by EMR. At surveillance colonoscopy, rates
of recurrent/residual polyps with UEMR and EMR were 7 % and 22 %, respectively. Pooled
analysis showed that recurrent/residual polyps were less likely to be seen after UEMR;
pooled RR (95 % CI) 0.26 (0.12, 0.56), I2 = 0 % ([Fig. 2b]) and this analysis reached statistical significance. Sensitivity analysis by excluding
one study [15] where mean polyp size in the EMR group (30 mm) was larger than in the UEMR group
(20 mm) also showed consistent results; pooled RR (95 % CI) 0.26 (0.11, 0.59), I2 = 0 %
Complete macroscopic resection
Three observational studies [9]
[14]
[15] reported data on complete macroscopic resection with 159 polyps resected by UEMR
and 218 by EMR. UEMR was associated with a statistically significant higher rate of
complete macroscopic resection compared to EMR; pooled RR (95 % CI) 1.28 (1.18, 1.39),
I2 = 94 % ([Fig. 3a]). Sensitivity analysis by excluding the study by Kim et al also showed a significantly
higher rate of complete macroscopic resection with UEMR compared to EMR, pooled RR
(95 % CI) 1.12 (1.06, 1.19) and heterogeneity decreased substantially (I2 = 0 %).
Fig. 3 Forest plot to compare complete macroscopic resection and histologic resection rate
between UEMR and EMR.
Complete resection confirmed by histology
Three studies (2 RCTs and 1 observational study) [11]
[12]
[13] reported data on this outcome and showed superiority of UEMR which reached statistical
significance; Pooled RR (95 % CI) 0.75 [0.57, 0.98], Cochran Q test P = 0.24, I2 = 29 % ([Fig. 3b]). Analysis of RCTs only showed similar results; pooled RR (95 % CI) 0.66 (0.47,
0.93), Cochran Q test P = 0.25, I2 = 25 %
Adverse events
Seven studies (2 RCTs and 5 observational) reported data on AEs. Immediate post-polypectomy
bleeding (IPPB) requiring endoscopic therapy and delayed post-polypectomy bleeding
(DPPB) were the most common AEs occurring with 5.5 % and 1.5 % of polypectomies, respectively.
There were no statistically significant differences in the rates of AEs between the
two techniques; pooled RR (95 % CI) was 0.68 (0.44, 1.05), I2 = 17 % ([Fig. 4]). Subgroup analysis including RCTs only was consistent; pooled RR (95 % CI) 1.30
(0.44, 3.86), I2 = 0 %. Sensitivity analysis by excluding the study by Kim et al showed similar results;
pooled RR (95 % CI) 0.70 (0.46, 1.09), I2 = 26 %.
Fig. 4 Forest plot to compare adverse events between UEMR and EMR.
Subgroup analysis based on type of adverse events showed a significantly lower rate
of IPPB following UEMR; pooled RR (95 % CI) was 0.60 (0.36, 0.99), I2 = 11 %. There was no significant difference in the rate of DPPB between the two techniques;
pooled RR (95 % CI) was 1.07 (0.43, 2.70), I2 = 0 %. There was one case of perforation with each technique.
Discussion
During the last several decades, EMR has been the treatment modality of choice for
managing sessile colorectal polyps over 10 mm in diameter. However, the rate of en
bloc resection decreases with increase in polyp size over 10 mm [17] and practice guidelines recommend that en bloc resection be limited to Iesions ≤ 20 mm
in the colon. UEMR, a newer technique first described in 2012 [7], is gaining popularity for the management of large sessile colorectal polyps. UEMR
eliminates submucosal injection prior to EMR based on the rationale that the anti-gravity
“floating” effect of water immersion on the mucosa and submucosa relative to the muscularis
layer, in addition to preservation of wall thickness in a non-distended lumen, makes
this step unnecessary. Advantages of eliminating submucosal injection are avoidance
of displacement or distortion of the polyp anatomy that may make resection more difficult,
avoidance of bleeding from needle puncture, and avoidance of the risk of dysplastic
or neoplastic cell seeding. Thermal injury to deeper wall layers may be decreased
due to the heat-sink effect of water submersion. Successful resection of polyps refractory
to EMR due to scarring after prior attempted resection or instrumentation has been
reported with UEMR [7]. Comparative studies had reported conflicting results thereby justifying this systematic
review and meta-analysis.
We found that UEMR was associated with a significantly higher en bloc resection rate
than EMR. Results were similar in subgroup analyses including RCTs only and on sensitivity
analyses. This is clinically important since the incomplete resection of sessile colorectal
polyps is associated with > 15 % local recurrence rates [18] necessitating more frequent surveillance colonoscopies and increasing the risk of
interval cancer development. This analysis, however, was limited by moderate heterogeneity
(I2 = 49 %) which may be due to difference in polyp sizes and possibly different locations.
We addressed this issue by performing a sensitivity analysis which led to substantial
decrease in heterogeneity (I2 = 0 %). Another limitation of this analysis is the small number of RCTs (only two)
in the subgroup analysis. This number is too small to make any firm conclusions. These
limitations limit the validity of our results and more RCTs are warranted to further
evaluate the possible superiority of UEMR in achieving a higher en bloc resection
rate compared to EMR.
Consistent with the higher rate of en bloc resection, UEMR was also associated with
significantly reduced rates of residual or recurrent polyps seen at surveillance colonoscopy.
Reported polyp recurrence rates after EMR can be as high as 15 % to 50 % [19]
[20]. A subgroup analysis based on fully published articles in peer-reviewed journals
confirmed that UEMR led to lower rates of residual or recurrent polyps. However, the
analysis of “recurrent or residual polyps” included only three studies and is underpowered
to make any firm conclusions.
UEMR was associated with significantly less IPPB that required endoscopic therapy.
Otherwise, we found no significant differences in the occurrence of adverse events
between the two techniques. It is not entirely clear why UEMR should be associated
with less IPPB. One possible explanation is that UEMR produces a smaller mucosal defect
than EMR. In EMR, the submucosal injection causes the borders of polyps to expand,
thereby producing a larger defect after snare resection. In UEMR, floating of mucosa
and submucosa over muscularis layer leads to a change in shape of lesions and some
flat and sessile lesions become smaller and more polypoid in configuration [11]. Resection of such lesions leads to smaller defects. Some studies have shown that
fewer clips are required after UEMR than EMR [13], supporting this hypothesis.
This is the first systematic review and meta-analysis to compare the efficacy and
safety of UEMR with EMR. Our comprehensive literature search identified a large number
of relevant studies. Our analyses may have been weakened by the inherent limitations
of meta-analyses and of the individual included studies, most of which were observational.
Due to differences in methodology, combining RCTs and observational studies in a meta-analysis
may raise the issue of validity of results. Lack of randomization in observational
studies, which is necessary to control measured and unmeasured confounding, can affect
the validity of study results [21]. To address this issue, we performed predetermined subgroup analyses based on types
of studies (i. e., RCTs vs observational) and noticed no difference in results. In
addition, we also performed sensitivity analyses by excluding some observational studies
that were considered to be outliers (had substantial differences from other studies)
with no change in results.
There was significant heterogeneity in the analysis for some outcomes. However, we
were able to address this by performing sensitivity analyses. Most of the studies
included had one or two operators. A formal propensity score matching was not done
in most of the studies, which can possibly affect the outcomes due to differences
between groups. However, some important parameters such as polyp sizes were comparable
between the two groups, as is evident from [Table 2], except for one study [15] where mean polyp size in the EMR group (30 mm) was larger than in the UEMR group
(20 mm). We performed sensitivity analysis by excluding that study with no change
in results.
We could not evaluate the difference in time required by both procedures as studies
had not presented adequate data for such analysis. Finally, analyses of some of the
outcomes such as “recurrent or residual polyps”, “complete macroscopic resection”
and “complete resection confirmed by histology” included only three studies each so
these analyses are not sufficiently powered to make definite conclusions. Consequently,
more studies are required to evaluate these outcomes.
Conclusion
In summary, UEMR was associated with significantly higher en bloc resection rates
than EMR and with lower rates of recurrence and IPPB. If confirmed in randomized controlled
trials, UEMR is likely to disseminate as the approach of choice for the resection
of large colorectal polyps.
