Introduction
Colorectal cancer (CRC) is the third most common cancer worldwide [11]. The majority of tumors arise from colorectal polyps and colonoscopy exerts its
beneficial effect by polypectomy [22]. Ideally polypectomy should be safe, efficient, cost-effective and curative. Recurrence
occurs due to incomplete resection which compromises clinical outcomes and increases
cost.
The overwhelming majority of colonic polyps are less than 10mm and are safely and
effectively managed with en bloc or oligo-piecemeal cold snare polypectomy. The absence
of electrocautery all but eliminates the risks of perforation and post-polypectomy
bleeding [33]. This safety and simplicity holds appeal for resection of larger lesions. While
these have traditionally been removed by hot snare endoscopic mucosal resection (EMR),
the use of diathermy carries with it a small but significant risk of perforation and
delayed bleeding [44]
[55].
The upper limit of lesion size where piecemeal cold snare polypectomy (p-CSP) remains
clinically effective, without excessive and burdensome recurrence rates, is currently
unknown. We undertook a modelling study to address this important question.
Methods
Incomplete resection rates for hot snare polypectomy (HSP) and CSP were determined
after a review of relevant literature. A formal systematic review with a comprehensive
electronic search strategy was not employed. Rather, targeted literature searches
were conducted for publications available up to 1st June 2022. The PubMed database
was searched using the following terms: “cold snare” AND polypectomy AND incomplete
(n = 36); “cold snare” AND polypectomy AND residual (n = 37).”
There were no restrictions on the type of publication and no restriction on study
size. Language was limited to English. Data extracted included: author, journal, year
of publication, polyp size, injectate used, study size, and rate of incomplete resection
(Table 1s, available online-only in Supplementary material).
There is a paucity of data for incomplete resection rates, particularly for piecemeal
p-CSP. The data does include a recent large multicenter trial (n = 1393) which reported
an incomplete resection rate of 1.5% for polyps ≤10mm [66]. However, there is little data reporting the incomplete resection rate using p-CSP
for polyps >20mm. Given the scarcity of relevant studies, a reasonable surrogate measure
was taken from the CARE study [77]; this reported an incomplete resection rate of 17.3% for polyps 10–20mm (n = 110).
This may underestimate the actual incomplete resection rate for p-CSP given that this
study used HSP [77].
Snare size and shape
A 10-mm snare was assumed for CSP modelling. Resection pieces vary between 6 and 8mm
in diameter [88]. A 15-mm snare was assumed for HSP, with a resection diameter of 15mm. The snare
shape has been assumed to be circular, as a surrogate for the true slightly elliptical
shape of a snare.
Resection error risk
Resection error risk (RER) is defined as the likelihood of incomplete removal of adenomatous
tissue per snare resection. This constant risk applies to each individual snare resection
during piecemeal EMR and thus the overall risk accumulates with each sequential resection.
Based on a 20-mm polyp area with a 3-mm margin of normal tissue (530.93mm2), the number of p-CSP resections assuming 8-mm resection pieces (area 50.26mm2) is 10.56. This is supported by the “disc covering problem” which estimates the smallest
number of smaller discs required to cover a larger disc [99]. Clinically, this equates to 10 or 11 sequential snare resections. Dividing 17.3%,
the incomplete resection rate from the literature, by the number of sequential snares
(10 or 11) the p-CSP RER is estimated at 1.57%–1.73%.
HSP has evolved with advances such as thermal ablation of the defect margin which
have greatly reduced recurrence rates [1010]
[1111]. The incomplete resection rate with HSP is also likely to be lower due to the tissue
destruction from the thermal penumbra of the hot snare. However, conservatively, based
on the study by Pohl et al., it is reasonable to assume an incomplete resection rate
of 17.3% for 20-mm polyps [77]. A 20-mm polyp resected with a 15-mm hot snare, with 15-mm resection pieces would
require 10 resections. RER for HSP is estimated at 1.73%.
Incomplete resection risk equation
The overall risk of incomplete resection (RIR) = 1 – (1 – RER)n, where n is the total number of resections required. This dynamic equation estimates
the cumulative risk of missed adenoma with each piecemeal snare resection.
Data analyses
Polyp area, snare area, and defect area were estimated using area =πr2 , where r was the radius. A 3-mm edge of normal tissue was added to the resection size, irrespective
of polyp size. RER has been assumed as constant with each resection. Data were computed
and graphically illustrated using Mathematica 12.3.1 (Wolfram Research) by a mathematician
with expertise in mathematical modeling.
Results
Polyp area
Polyp area increases quadratically with increasing polyp radius ([Fig. 1Fig. 1]). For example, the areas of a 5-mm, 10-mm, and 20-mm polyp are 19.6mm2, 78.5mm2, and 314.2mm2, respectively ([Table 1Table 1], [Fig. 2Fig. 2]
,
[Fig. 3Fig. 3]). This difference is further emphasized when comparing a 30-mm and 40-mm polyp which
have areas of 706.9mm2 and 1256.6mm2 respectively. With the addition of a 3-mm margin of normal tissue, the area difference
between a 10-mm and 20-mm polyp is 329.9mm2 (2.6 times larger).
Fig. 1
Fig. 1 Risk of incomplete resection (RIR). A mathematical model demonstrating the RIR with
cold snare polypectomy and hot snare polypectomy for increasing polyp size.
Table 1
Table 1 Mathematical modelling of risk of incomplete resection (RIR) for hot and cold and
hot snare polypectomy. Polyp area, based on polyp and defect area, resection error
risk (RER), and number and size of resections.
|
Polyp diameter,mm
|
|
5
|
10
|
15
|
20
|
25
|
30
|
35
|
40
|
|
Polyp area,mm2
|
19.63
|
78.54
|
176.71
|
314.16
|
490.87
|
706.86
|
962.11
|
1256.64
|
|
Defect area,mm2
|
95.03
|
201.06
|
346.36
|
530.93
|
754.77
|
1017.88
|
1320.25
|
1661.90
|
|
6-mm resections
|
|
Number of resections
|
3.36
|
7.11
|
12.25
|
18.78
|
26.70
|
36.00
|
46.70
|
58.79
|
|
Risk of incomplete resection (RER 1.57%)
|
5.20%
|
10.60%
|
17.60%
|
25.70%
|
34.50%
|
43.40%
|
52.20%
|
60.60%
|
|
8-mm resections
|
|
Number of resections
|
1.89
|
4.00
|
6.89
|
10.56
|
15.01
|
20.25
|
26.26
|
33.06
|
|
Risk of incomplete resection (RER 1.73%)
|
2.95%
|
6.13%
|
10.33%
|
15.39%
|
21.14%
|
27.42%
|
34.00%
|
40.74%
|
|
Hot snare resections (15-mm)
|
|
Number of resections
|
0.54
|
1.14
|
1.96
|
3.00
|
4.27
|
5.76
|
7.47
|
9.40
|
|
Risk of incomplete resection (RER 1.73%)
|
0.94%
|
1.97%
|
3.36%
|
5.10%
|
7.18%
|
9.56%
|
12.22%
|
15.13%
|
Fig. 2
Fig. 2 Cold snare polypectomy for Paris 0-IIa polyp <10mm. a 3-mm Paris 0-IIa colorectal adenoma. b En bloc R0 excision with thin-wire cold snare. Note wide margin of normal mucosa with
snare placement. c The halo effect with blanching of the surrounding mucosa and hyperemia of the adenoma.
d Post-excision halo effect. e Water expansion of the defect. f Careful inspection of margin.
Fig. 3
Fig. 3 Piecemeal cold snare polypectomy for Paris 0-IIa lesion >20mm. a–c Laterally spreading lesion, >20mm, examined with high definition white-light and narrow-band
imaging. d Lesion after submucosal injection. e Piecemeal cold snare excision. f Careful inspection of the post-resection defect margin.
Resection number
Resection number is influenced by polyp area and sequential resection area ([Table 1Table 1]). A 20-mm polyp (with a 3-mm margin of normal tissue) removed by p-CSP, would requires
18.8 or 10.6 resections, using piecemeal resection sizes of 6-mm or 8-mm diameter,
respectively.
Risk of incomplete resection (RIR)
RIR is directly proportional to the number of resections, influenced by polyp size
and snare area. The predicted RIRs for a 20-mm versus 40-mm polyp removed by p-CSP
with a 6-mm resection piece size are 25.7% and 60.6% respectively. The predicted risks
of incomplete resection for a 20-mm versus 40-mm polyp removed by p-CSP with an 8-mm
resection piece size are 15.4% and 40.7%, respectively ([Table 1Table 1], [Fig. 1Fig. 1]).
Discussion
It has been shown that p-CSP is safer than HSP. However, anecdotally for adenomas
recurrence seems to be excessive. We undertook a mathematical modeling study to better
understand the size at which recurrence from p-CSP might become particularly burdensome.
Data are scarce for the rates of incomplete resection for piecemeal CSP and HSP. After
extensive review of the literature, we used contemporary data from the highest quality
publications to inform our model. The RIR using p-CSP for polyps >20mm in size and
6-mm resection pieces was 25.7%. The corresponding value with 8-mm resection pieces
was 15.4% and comparable to the pre-thermal ablation recurrence rates prior to thermal
ablation of the margin. However, as size increased RIR for p-CSP became excessive
at >40% for lesions ≥40mm.
The risk profile of CSP is clearly advantageous. Delayed bleeding and perforation
are extremely rare events. The mitigation of bleeding risk is also very appealing
in an era where anticoagulant and antiplatelet agents are common. The safety profile
of CSP has been demonstrated in a large retrospective study (n = 562) that compared
conventional EMR to p-CSP for resection of large ≥20-mm sessile serrated lesions [1212]. There were no adverse events reported in the p-CSP group compared to delayed bleeding
and significant deep mural injury (Sydney classification type 3–5) [44] in the EMR group of 5.1% (n = 18) and 3.4% (n = 12) respectively. In both groups
technical success was 100% and recurrence at 18 months was similarly low at <2% [1212].
Paradoxically, despite the excellent safety profile of p-CSP, there is a hesitancy
to resect wide margins of normal tissue and this potentially leads to higher recurrence
rates. Without meticulous attention to ensure overlapping of sequential snares, there
is a risk of incomplete resection with each snare resection. Each pass with the snare
wire carries a small but cumulative error risk. The risk becomes unacceptably high
when removing ever-larger lesions. In this study, the RIR for a 40-mm polyp was >40%
with an 8-mm resection piece size and >60% with a 6-mm resection size.
Using modeling we have demonstrated that as lesion size increases, the risk of incomplete
resection disproportionately increases and becomes potentially unacceptably high at
>40% for lesions 40mm. This indicates that current techniques or devices are inadequate
to manage larger lesions in this way.
There is a lack of literature that assesses the natural history of incompletely resected
adenomas. Approximately one-third of post-colonoscopy CRCs are found within the same
colorectal segment where the adenoma was resected suggesting that these could be a
result of incomplete resection [1313]. Pohl et al.’s 2021 study (n = 233) demonstrated a greater risk for neoplastic polyps
in colon segments after incomplete polyp resection compared to segments with a prior
complete resection (odds ratio 3.0, 95%CI 1.12–8.17] [1414]. There is also no guarantee that these patients will comply with a surveillance
program or that the polypectomy site will be found during surveillance procedures.
Moreover the financial burden of surveillance is significant.
In contrast to HSP, p-CSP relies purely on a wide resection margin and sequential
snare resection overlap to prevent recurrence. Key factors in reducing recurrence
rates include resecting a 2–3-mm margin of normal tissue and meticulous post-resection
examination of the p-CSP defect [1212]
[1515]. Although thin-wire snares (0.30mm) have been developed for CSP, in an international
multicenter randomized controlled trial (n = 1393) the rate of incomplete polyp resection
was similar to that with use of a thick-wire snare (0.47mm). There was only a single
case of post-endoscopy bleeding; this was seen in the thin-wire snare group [66]. Although high quality technique is critical for optimizing outcomes, further reductions
in RER are likely to be driven by advances in device design. Intraprocedural bleeding
has also been identified as an independent risk factor for recurrence [1616]
[1717]. In comparison to HSP, intraprocedural bleeding is more frequent when using CSP.
Therefore we suggest the use of dilute adrenaline 1:100000 added to the injectate
to minimize bleeding when performing p-CSP for lesions >20mm.
Preventing recurrence is important for several reasons. Due to the fibrosis typically
encountered around a recurrent polyp, resection is more challenging [1818]. This increases the risk of complications and resultant morbidity. Additional healthcare-related
costs are attributed to an often prolonged procedure time to treat the recurrence,
an increased risk of admission post-procedure and the need for additional procedures
for ongoing surveillance. All of this assumes that the patient is adherent to the
management plan and is not lost to follow-up. Hence, techniques to reduce recurrence
risk are essential.
The results of this study have several important clinical implications. Piecemeal-CSP
almost completely eliminates the risk of post-procedural bleeding and perforation;
however recurrence is greatly increased in comparison to conventional EMR. Lesion
location, morphology, and predicted histology, and patient frailty and comorbidities
need to be factored into the process of selecting the appropriate resection technique;
this ideally would minimize complications, achieve successful lesion excision, and
avoid burdensome surveillance. For example, in the right colon, large nonpedunculated
colonic polyps ≥20mm have a 10%–12% risk of post-procedural bleeding in the absence
of defect closure, but many of these lesions are very large and defy current closure
techniques. However the risk of recurrence is also related to lesion size. These are
some of the factors that need to be borne in mind. These considerations usher in a
new approach to colonic tissue resection which may become more individualized and
bespoke.
This study has several limitations. Firstly, a targeted review of the literature rather
than a systematic review was employed to estimate RER for our equation. We acknowledge
this method could have biased our model. However, our modelled RIRs are consistent
with recurrence rates reported in a recent systematic review [1919]. Abdallah et al. reported a recurrence rate of 12.3% (3.4%–35.7%) for polyps ≥20mm resected by p-CSP), derived from 4 studies [1212]
[2020]
[2121]
[2222]. Unfortunately, recurrence rates for adenomas and SSLs were not reported separately.
SSLs are particularly suited to p-CSP due to their thin mucosal profile, and therefore
have a lower recurrence rate, compared to adenomas. In particular, Van Hattem et al.
included only SSLs (n = 562, 4.3%) [1212]. The recurrence rate for adenomas ≥20mm is therefore likely higher than concluded
by Abdallah et al. and more in line with our model.
Secondly, this model assumes that each resection is an independent entity. In clinical
practice however, each resection is related in varying degrees, to the previous resections.
This model also does not account for snare overlap. Based on current literature, accounting
for this is not feasible. Thirdly, it was assumed that polyp and snare shape are circular
to make calculation of area and number of resections more practicable. This assumption
probably underestimates the number of resections required to remove large noncircular
lesions. Fourthly, technical challenges such as difficult access and intraprocedural
bleeding have not been accounted for but are likely to increase the number of resections
for a lesion of any given size. Fifthly, the analysis was based on a very low incomplete
resection rate, in procedures conducted by experts who had received prior training,
and it probably underestimates the true numerical value of RER. Other studies have
reported a rate of incomplete resection of 8%–10%. Finally, during p-CSP re-excision
of suspicious areas may serve to reduce the rate of incomplete resection; to what
extent this occurs and its efficacy is currently unknown and not feasible to model.
Conclusions
It is important for any endoscopist undertaking p-CSP to have a deep understanding
of the fundamentals which underpin endoscopic resection technique. These include sound
optical evaluation skills and a willingness to resect a wide margin with meticulous
defect inspection and sequential snare overlap. This study demonstrates that the benefits
of p-CSP may begin to be outweighed by the risk of incomplete resection for polyps
>30–40mm. Further research is needed to develop better tools to minimize incomplete
resection when using p-CSP, and to understand the upper limit for p-CSP where its
safety advantages become outweighed by the burden of recurrence and its sequelae.
