Introduction
Colorectal cancer (CRC) is the third most common cancer in the US with 135 – 420 new
cases per year [1 ]. Screening colonoscopy with polypectomy substantially reduces CRC incidence and
mortality [2 ]
[3 ]. However, an unexpected high rate of incomplete resection of colon polyps has been
reported in some recent studies, and this has in turn been related to a higher risk
of post-colonoscopy interval CRC [4 ].
Diminutive (< 5 mm) and small (6 – 9 mm) polyps represent the vast majority of polyps
removed at screening colonoscopy [5 ]. In theory, two major techniques are available for these lesions, namely biopsy-forceps
polypectomy and snare polypectomy. The former, however, has been associated with a
much higher rate of incomplete resection, especially for small polyps, and its use
is not generally recommended [6 ]. The latter may be further classified into hot (HSP) and cold (CSP) snare polypectomy.
The basic difference is the use of a high frequency generator for HSP. Such use may,
on the one hand, minimize immediate post-polypectomy bleeding by coagulation, but,
on the other, it may also damage deeper vessels with increased risk of delayed bleeding
[7 ]
[8 ] or even perforation. For this reason, CSP is usually considered safer, while resulting
in equivalent rates of complete resection, and its use has dramatically increased
in recent years [9 ]
[10 ]
[11 ]
[12 ], due also to the development of specific CSP-snares. CSP- and HSP-incomplete resection
rates (IRR) for ≤ 10 mm polyps range widely, being 0.5 – 6.4 % and 1.2 – 7.4 %, respectively
[12 ]
[13 ]
[14 ]
[15 ], and these estimates are much lower compared with forceps-polypectomy [6 ]
[16 ]. However, any advantage of one technique (HSP vs. CSP) over the other remains unclear
[17 ].
The current European Society of Gastrointestinal Endoscopy (ESGE) guidelines recommend
CSP as the preferred technique for removal of diminutive polyps (size ≤ 5 mm) due
to high rates of complete resection, adequate tissue sampling for histology, and low
complication rates [18 ]. They also suggest CSP for sessile polyps 6 – 9 mm in size because of its superior
safety profile, although evidence comparing efficacy with HSP is lacking [18 ].
Recently, randomized controlled studies (RCTs) have compared the efficacy of HSP with
CSP for colorectal polyps ranging between 4 and 10 mm with a disparity in results
[19 ]
[20 ]
[21 ]. Given this discrepancy and the lack of supporting evidence for current guidelines
for polypectomy, we performed a systemic review and meta-analysis of available evidence
with the objective to compare the IRR and adverse events between CSP and HSP when
removing polyps between 4 and 10 mm in size.
Materials and methods
A comprehensive search was performed using PubMed, Embase, and abstracts of conferences
presented at Digestive Diseases Week and the United European Gastroenterology Week.
These sources were searched using the keywords “polypectomy”, “cold polypectomy”,
“cold snare polypectomy”, “hot snare polypectomy”, “remnant adenoma”, and “endoscopic
mucosal resection” to extract studies up to December 2017. Only English language and
human studies were searched. Related data suggested by PubMed were also searched.
Two authors (RJ and MA) individually searched the databases for article screening.
The abstracts were not blinded for authors, institutions or journals during review.
Only data from randomized controlled trials comparing the efficacy of and adverse
events related to the use of CSP with those of HSP (with or without submucosal injection)
for removal of 4 – 10 mm colorectal polyps were included in our study. Such ‘4 – 10
mm’ definition was preferred over the generally reported ‘6 – 9 mm’ for small polyps,
in order not to exclude relevant studies, and also include 4-, 5- or 10-mm polyps.
In addition, we limited our inclusion to only those studies where biopsy specimens
were obtained from the post-polypectomy resection margin to confirm the absence of
any residual polypoid tissue (i. e. endoscopic and histologic radical resection).
If multiple articles were found from the same institution, then the most recent article
was used for analysis. Letters to editors, case reports, case series, case-control,
and cohort studies were excluded.
Data collection and bias assessment
Baseline demographic data (age, gender), number of study participants, number of polyps
examined, morphology of polyps, histology of polyps, location of polyps, IRR, type
of snare used, procedure time, and complication rates were extracted from each study.
Paris classification [22 ] and Vienna classification [23 ] systems were used to assess the morphology and histology of polyps, respectively.
Polyps were classified as either right-sided (if found in the cecum, ascending colon
or transverse colon) or left-sided (if found in the descending colon, sigmoid colon,
or rectum). IRR was defined as the presence of any residual polypoid tissue in biopsied
specimen post-polypectomy with either technique. Procedural time was defined as the
time required from identification of polyp to complete resection of the polyp using
either technique [19 ]
[21 ].
Complications included immediate bleeding during the procedure, delayed bleeding after
the procedure, and perforations of the intestinal wall. Immediate bleeding was defined
as continuous hemorrhage usually for ≥ 30 seconds immediately after polypectomy [19 ]. Delayed bleeding was defined as hemorrhage after colonoscopy requiring endoscopic
hemostasis. All three studies have excluded patients who were treated with antithrombotic
agents.
Outcome measures
The primary outcome of our study was IRR and secondary outcomes were complication
rates, polyp retrieval rates, and procedure time.
Risk of bias assessment
The risk of bias was evaluated based on guidelines from the Cochrane Handbook for
Systematic Reviews of Interventions [24 ] using the following: adequacy of random sequence generation, allocation concealment,
blinding of the participant, blind outcome assessment, incomplete outcome data, and
selective outcome reporting.
Statistical analysis
Pooled odds ratio (OR) and 95 % confidence interval (CI) were calculated for primary
and secondary outcomes. For pooled analysis of rare events, risk difference was used
as an estimate to compare any possible detectable difference, when applicable. Heterogeneity
of the study was assessed using the I
2 statistic [25 ]. Percentages of 25 % (I
2 = 25), 50 % (I
2 = 50), and 75 % (I
2 = 75) were considered to be a low, moderate, and high degree of heterogeneity, respectively.
In the presence of substantial heterogeneity (I
2 > 50 %), a random effect model was used as the pooling method; otherwise, a fixed
effect model was adopted as the pooling method. Statistical software used was Review
Manager (RevMan) v.5.3 (The Cochrane Collaboration, Oxford, Oxfordshire, UK).
Results
A total of 244 studies were retrieved using the databases ([Fig. 1 ]). After screening through the title, abstracts, and full texts, a total of three
RCTs were selected [19 ]
[20 ]
[21 ]. All three studies were published between April 2017 and August 2017 ([Table 1 ]). A total of 1051 patients were found to have 1485 small polyps (4 – 10 mm in size)
which were randomized to either the HSP group (n = 741 polyps) or the CSP group (n = 744
polyps). Of 1485 colorectal polyps, 1266 polyps were included in the final analysis
with 630 polyps in the HSP group and 636 polyps in the CSP group. Two RCTs [19 ]
[21 ] included only neoplastic polyps for the final analysis while one RCT [20 ] included all colorectal polyps which were excised. The mean age of patients randomized
to the CSP group was 64.1 years old, while those randomized to the HSP group was 65.3
years old, as described in two studies [20 ]
[21 ]. The mean percentages of males were 55.5 % in the CSP group and 56.8 % in the HSP
group.
Fig. 1 PRISMA flow diagram for selection of studies.
Table 1
Details of the three studies included.
Reference
Patient age, mean ± SD, years
Gender male, %
No. of polyps, n
Polyp size, mean ± SD, mm
Macroscopic appearance, n
Histological findings, n
Location of the polyp, n
No. of all adverse events, n
Incomplete resection rate, %
Complete retrieval rate, %
0 – I
0 – II
TA
AA
HP
SSP
OP
Right
Left
Cold snare polypectomy
Kawamura et al., 2017 [19 ]
NA
NA
341
NA
234
107
333
10
21
11
12
207
134
34
1.8
98.2
Zhang et al., 2017 [21 ]
64.5 ± 7.7
96 (53.6 %)
267
7.4 ± 1.4
164
81
189
45
46
25
7
141
126
5
8.5
100
Papastergiou et al., 2018 [20 ]
63.1 ± 10.3
46 (59.7 %)
83
8.2 ± 1.6
38
45
59
0
7
17
0
40
43
3
7.2
92.8
Hot snare polypectomy
Kawamura et al., 2017 [19 ]
NA
NA
346
NA
234
112
336
15
32
9
7
193
153
16
2.6
99.3
Zhang et al., 2017 [21 ]
65.8 ± 9.4
101 (56.4 %)
258
7.7 ± 1.5
164
74
175
41
49
26
15
140
118
3
1.5
100
Papastergiou et al., 2018 [20 ]
64.1 ± 10.9
45 (57.7 %)
81
8.3 ± 1.4
34
47
60
0
8
13
0
36
45
1
3.7
95.1
NA, not available; TA, tubular adenoma; AA, advanced adenoma; HP, hyperplastic polyps;
SSP, sessile serrated polyps; OP, other polyps.
The risks of bias assessment are reported in [Fig. 2 ].
Fig. 2 Risk of bias per Cochrane risk of bias assessment among studies included.
Type of snare
The snares used in these studies were: Captivator II 10 mm (Boston Scientific, Natick,
MA, United States) [19 ]
[20 ], the Acusnare (Cook Medical, Bloomington, IN, USA) [20 ], Snare-Master 10 mm (Olympus Medical, Tokyo, Japan) [20 ]
[21 ] and Exacto/9 mm (US Endoscopy, Mentor, OH, United States) [19 ]. For HSP, electro-cautery was performed with the ERBE ICC 200 electrosurgical generator
(ERBE Electromedizin, Tübingen, Germany) [20 ]
[21 ]. Only one study mentioned the setting for cautery as fractionated cutting mode ENDO
CUT Q (effect 3; cut duration 1; cut interval 3) [20 ]. Sub-mucosal injection before polypectomy was performed for both groups in one study
[19 ], however, it was only performed for the HSP group in the other two studies [20 ]
[21 ].
Polyp characteristics
Based on the morphology of polyps, 436 out of 691 (63.1 %) and 432 out of 685 (63.1 %)
were classified as type 0 – I polyps, while 255 out of 691 (36.9 %) and 263 out 685
(38.4 %) were classified as type 0 – II in the CSP and HSP group, respectively. On
a histological basis, the following polyps were removed: 1152 (78.1 %) tubular adenomas,
111 (7.5 %) advanced adenomas (prominent villous component and high grade dysplasia),
86 (5.8 %) sessile serrated polyps/adenoma, 83 (5.6 %) hyperplastic polyps and 43
(2.9 %) polyps without histological diagnosis. The locations of polyps were reported
in all studies with a total of 757 out of 1485 (51 %) found to be right-sided. All
three studies reported the average polyp sizes which were not significantly different
on comparison. The size averaged between 5.4 and 8.3 mm as provided in [Table 1 ].
Endoscopic mucosal resection
Submucosal injection before polypectomy for endoscopic mucosal resection (EMR) was
performed for both groups in one study [19 ]; however, it was only performed for selected polyps in the HSP group in the other
two studies based on endoscopist preference [20 ]
[21 ]. The injected submucosal solution included 1:10 000 epinephrine in one of the studies
that allowed submucosal injection for HSP only [21 ] while the other two studies did not use any epinephrine in the injected submucosal
solution.
Incomplete resection rate
IRR was assessed via four-quadrant biopsy along with biopsy of the base of the post-polypectomy
site in two studies [19 ]
[20 ] and biopsy of the left and right lateral margins in one study [21 ]. IRR was collectively found to be 2.4 % (15/630) and 4.7 % (30/636) for HSP and
CSP group, respectively. The study heterogeneity was substantial (I
2 = 73 %), hence a random effects model was used to assess the difference in incomplete
resection. The pooled odds ratio was 0.51 (95 % CI 0.13 – 1.99, P = 0.33). The difference in IRR between HSP and CSP was not statistically significant
([Fig. 3 ]).
Fig. 3 Forest plot of all RCTs assessing incomplete resection rate (IRR) between HSP/EMR
and CSP group.
Adverse events
All three studies reported adverse events (immediate procedural bleeding, delayed
bleeding, and perforations). A total of 741 polyps were removed using HSP and 744
polyps were removed employing CSP. Clinically and statistically, a lower rate of overall
adverse events was observed with HSP compared to CSP (20/546 [3.7 %] vs. 36/545 [6.6 %],
pooled OR 0.53, 95 % CI 0.3 – 0.94, P = 0.03; I
2 = 0 %) ([Fig. 4a ]).
Fig. 4 Forest plot of all RCTs assessing: a total complication rate; b immediate bleeding rate; c delayed bleeding rate between HSP/EMR and CSP group.
When assessing adverse events separately, immediate bleeding rate was lower in the
HSP group (18/546 [3.3 %]) compared to the CSP group (36/545 [6.6 %]), yielding a
pooled OR 0.48 (95 % CI 0.27 – 0.86, P = 0.01; I
2 = 0 %) ([Fig. 4b ]). All patients with immediate post-polypectomy bleeding required endoscopic hemostasis
to control bleeding and was clinically successful in all three studies. Two studies
used hemostatic clip placement for endoscopic hemostasis [20 ]
[21 ] while one study did not mention the technique used for endoscopic hemostasis [19 ].
The delayed bleeding rates were not different between the two groups (2/546 [0.4 %]
vs. 0/545 [0 %]) ([Fig. 4c ]). Delayed bleeding occurred in only two patients who underwent HSP in one of the
studies [19 ] while the other two studies did not have any occurrence of delayed bleeding in either
group. Since delayed bleeding was considered a “rare event” among the studies included,
we used pooled risk difference as pooled estimate to detect if any difference existed
between the groups. Pooled risk difference was zero and results were not statistically
significant (P = 0.45).
No case of perforation was reported in either group in any of the three studies. Preventive
hemostasis, defined as prophylactic coagulation of vessels or red spots in the ulcer
or clipping of a non-bleeding post-polypectomy mucosal defect, was allowed in the
HSP group, but not in the CSP group in one study [19 ].
Polyp retrieval rate
The polyp retrieval rate was not clinically or statistically significantly different
between the two groups. Collective retrieval rates were 99 % for the HSP group and
98.1 % for the CSP group yielding a pooled OR of 1.9 (95 % CI 0.74 – 4.83, P = 0.18; I
2 = 0 %) as shown in [Fig. 5 ].
Fig. 5 Forest plot of all RCTs assessing polyp retrieval rate between HSP/EMR and CSP group.
Procedure time
Procedural time was defined as the time required from identification of polyp to complete
resection of the polyp using either technique [19 ]
[21 ]. Only two studies reported procedural time for polypectomies. Kawamura et al. reported
median times of 83 seconds and 60 seconds for the HSP and CSP group, respectively
[19 ]. Zhang et al. had comparatively longer times for their polypectomies with mean procedural
times of 330 seconds and 282 seconds, respectively for the HSP and CSP group [21 ].
Discussion
The results of this meta-analysis suggest that there is no significant difference
between cold and hot snare in achieving a complete resection of 4 – 10 mm colorectal
polyps, but there is a higher rate of adverse events with cold snare, albeit this
is limited to immediate post-polypectomy bleeding that was amenable to endoscopic
treatment in all of the cases.
The main result of our analysis is the equivalence between the two techniques in achieving
a complete endoscopic resection. This is clinically relevant as completeness of resection
represents by far the dominant end point when approaching the risk/benefit ratio of
endoscopic resection for sub-centimetric lesions. Our meta-analysis, however, showed
that there is residual uncertainty on the equivalence between the two techniques,
as shown by the very high heterogeneity level in this estimate. In detail, in two
of the three studies, the rate of incomplete resection was 2- to 6-fold higher with
cold snare than hot snare [20 ]
[21 ], while in the remaining study a very low rate of incomplete resection was shown
in both of the arms [19 ]. The risk of incomplete resection with cold snare, was however < 10 % in all three
trials, marginalizing the clinical impact of a possible superiority of the hot snare
technique for such an end point. For instance, the overall 4.7 % rate of incomplete
resection with cold-snaring is much lower than the nearly 20 % estimated for forceps
biopsy when removing diminutive polyps [16 ]
[26 ].
Adverse events related to both CSP and HSP include risk of bleeding (immediate bleeding
during the procedure or delayed bleeding after colonoscopy) and intestinal perforation.
In a large, prospective, non-randomized study, Repici et al. reported the rate of
immediate bleeding for CSP to be 1.8 % [10 ]
[27 ]. The rate for immediate bleeding has been reported to be 0 – 1.4 % for HSP [13 ]
[15 ]. The immediate bleeding rates in our meta-analysis were 6.6 % and 3.3 % for CSP
and HSP groups, respectively. There were two major limitations in the included studies
which could have affected immediate bleeding rate between the two techniques. One
of the three studies included allowed submucosal injection which also included epinephrine
in the HSP group only [21 ], and which could have decreased immediate post-polypectomy bleeding in the HSP group. Also,
only one study [19 ] defined immediate bleeding to be continuous bleeding for ≥ 30 seconds while the
other two studies did not provide any specific definition. Immediate oozing is destined
to occur after cold snare polypectomy due to lack of coagulation current. In most
cases, it is a slow, capillary bleeding that spontaneously stops and does not need
intervention [13 ]
[15 ].
The incidences of delayed post-polypectomy bleeding observed in previous prospective
studies were 0.6 – 2 % and 0 % for the HSP and CSP groups, respectively [10 ]
[15 ]
[28 ]
[29 ]. The delayed bleeding rates in our study were 0.3 % and 0 % for the HSP and CSP
groups, respectively, which are consistent with previous studies. As this was a rare
event in our included studies, we used risk difference as pooled estimate; pooled
risk difference was zero and results were not statistically significant (P = 0.45) suggesting there was no difference between either method. Colonic perforation
has been reported to be below 1 % using HSP and 0 % with CSP [30 ]
[31 ]. The overall adverse events for our meta-analysis were lower for the HSP group compared
to CSP, with sub-analysis revealing immediate post-procedural bleeding to be lower
for the HSP group. Clinically, this does not have any major significance as immediate
bleeding is usually self-limiting and managed with observation on the majority of
occasions or with post-procedural clipping immediately after polyp removal [10 ]
[32 ]. Since all three studies excluded patients treated with antithrombotic agents, the
use of antithrombotic agents did not have any influence on post-polypectomy bleeding.
Several factors pose limitations to our meta-analysis. Firstly, despite a wide literature
search on the main electronic databases, only three RCTs were available in the literature
which were evaluated and compared in this systematic review and meta-analysis. Secondly,
there was no follow-up colonoscopy for these patients to confirm “true” complete resection
as polyps may recur in previously biopsied negative margin sites. Further studies
are needed that not only biopsy the resection margins to confirm complete removal
but also assess the patients at a follow-up surveillance colonoscopy and biopsy the
scar area to determine the IRR. The post-polypectomy sites can also be evaluated by
performing endoscopic mucosal resection of the 1 – 3 mm margin around the resection
site to determine IRR as suggested by Matsuura et al. [33 ]. Third, the rates of incidence of colon cancer and mortality from colon cancer were
not assessed in these studies which undermines the true efficacy of removing small
colorectal polyps. Large, multicenter long-term studies are needed to assess the incidence
of colon cancer when comparing the two techniques in removing small colorectal polyps.
Moreover, two studies reported four-quadrant biopsy along with biopsy of base of polypectomy
site compared to one study that biopsied only left and right resected margins to display
any residual polypoid tissue. This can imply potential bias and possibly false IRR.
There is potential bias involved with different physicians as each endoscopist has
a different skill set with different procedural times. There was also substantial
heterogeneity noted between the three included studies which could be partially explained
by different types of snare used between studies for polypectomy and the difference
in the utility of submucosal injection between CSP and HSP groups in the included
studies. Also, immediate bleeding was only defined in one study, while the other studies
used the term “intra-procedural bleeding” that was summed together as “immediate bleeding”
for our systematic review. Finally, we only searched articles published in the English
language and may have missed relevant studies published in other languages.
Despite the above limitations, the main strength of this meta-analysis is the fact
that all three studies being compared were randomized controlled trials (RCTs) with
a low risk of bias based on the Cochrane risk of bias assessment scale that demonstrated
incomplete resection rate (IRR) using post-polypectomy biopsy to confirm residual
tissue. Furthermore, all RCTs assessed polyp size between 4 and 10 mm which is in
accordance with our study objective and further decreased the bias among polyp sizes.
In conclusion, our study findings show that there is no statistically significant
difference between the two techniques, although the rate of overall complications
is higher with CSP; however, this is self-limiting. Based on currently available RCTs,
we suggest that, clinically, either CSP or HSP can be safely used as one of the standard
techniques for the resection of 4 – 10 mm colorectal polyps. More studies are needed
in future to assess recurrence of polyps and incidence of cancer to further gauge
the efficacy of these two techniques.
