Introduction
Endoscopic polypectomy and endoscopic mucosal resection (EMR) have been the standard of care for many years; however, these methods should be avoided when challenging lesions are encountered. Lesions >20 mm, with severe submucosal fibrosis, or residual/recurrent lesions on scars after previous resection require advanced endoscopic resection techniques to achieve en bloc resection. In this regard, endoscopic submucosal dissection (ESD) has been introduced to overcome these challenges and improve resection results. Notably, according to Japanese guidelines and the Japanese medical health insurance system, superficial colonic neoplastic lesions >20 mm that are expected to be intramucosal or slightly invasive submucosal cancer should be considered for ESD [11 ]. However, colorectal ESD is still considered a challenging technique [22 ]
[33 ]
[44 ].
In order to manage lesions with severe submucosal fibrosis, we have developed the pocket-creation method (PCM) [55 ]
[66 ]. PCM is one of the techniques recommended for colorectal ESD in the 2023 technical review from the European Society of Gastrointestinal Endoscopy [77 ]. However, during opening of the pocket, the surrounding mucosa gradually loses traction, and the lesion becomes very unstable, making PCM-assisted ESD technically challenging. To overcome this, we developed a novel technique named the pocket-creation method with single-clip traction (PCM-CT) [88 ]
[99 ]. No prospective study has compared the efficacy of PCM-CT with that of conventional PCM; therefore, we prospectively compared the two techniques in this multicenter, randomized, controlled trial.
Methods
Study design
This randomized controlled clinical trial was carried out at four Japanese institutions. The trial followed the Declaration of Helsinki, and the trial protocol received the approval of the ethics committee at each of the four participating institutions: Jichi Medical University Hospital (B20–153), Fukushima Medical University Aizu Medical Center (RK2021–001), Kansai Medical University Medical Center (2021247), and Jyoban Hospital of Tokiwa Foundation (JHTF-2021–003). The manuscript was prepared according to the checklist of the Consolidated Standards of Reporting Trials (CONSORT) 2010 Statement (see the online-only supplementary material) [1010 ].
Patient inclusion and exclusion criteria
Patients were enrolled if they had superficial colonic neoplastic lesions >20 mm that were expected to be intramucosal or slightly invasive submucosal cancer according to Japanese guidelines for colorectal cancer [11 ]. Before study enrollment, all patients underwent preoperative colonoscopy. The lesion size was estimated with an endoscopic measuring device (M2–3U; Olympus, Tokyo, Japan). The tumor invasion depth was evaluated based on nonmagnifying endoscopic findings and magnifying endoscopic findings using Japan Narrow Band Imaging Expert Team (JNET) classification [1111 ] and/or Kudo’s pit pattern [1212 ]. Additionally, endoscopic ultrasonography was performed when submucosal invasive cancer was suspected. Patients were excluded if they had: lesions extending to the appendiceal orifice, colonic diverticula, or ileocecal valve; patients with diagnosed ulcerative colitis, Crohn’s disease, hematological abnormalities, or severe organ failure; and patients with lesions >100 mm in diameter. We excluded these very challenging lesions to ensure patient safety, as trainees participated in the trial. Patients with rectal lesions were also excluded. When a patient had multiple lesions, only the lesion located most orally was registered in the trial to prevent operator selection bias for a lesion treated with PCM or PCM-CT. Patients under the age of 20 years or over 91 years, and patients unable to consent to the procedure were excluded from the study. All patients provided written informed consent after receiving a detailed explanation of endoscopic procedures and study participation by researchers.
Operators and assistants
Seven endoscopists performed all procedures. Endoscopists with experience of 20–100 colonoscopies were selected as operators to eliminate bias based on ESD skills; these endoscopists were not able to perform colorectal ESD independently, but did have significant ESD experience in the upper gastrointestinal tract, where they could operate without supervision. In institutions with multiple operators, an equal number of trial cases was assigned to each operator. If an operator exceeded the predetermined number of 100 cases of colorectal ESDs during the study period, the additional cases were also included in the study. The participating ESD operators were independent endoscopists with experience of total colonoscopy, polypectomy, EMR, more than 20 gastric ESDs, and certification from the Japanese Society of Gastrointestinal Endoscopy, or equivalent skills. In addition to the above, participating endoscopists had experience of at least 20 colorectal ESDs to ensure patient safety and research quality. Participating endoscopists were assisted by expert endoscopists who had experience of more than 100 colorectal ESDs and of supervising ESD procedures. The assistants were randomly assigned.
ESD procedures
A colonoscope (EC-580RD/M [Fujifilm, Tokyo, Japan] or PCF-H290TI [Olympus]), a carbon dioxide insufflation regulation unit (GW-1 [Fujifilm] or UCR, [Olympus]), and a mechanical water pump (JW-2 [Fujifilm] or OFP-2 [Olympus]) were used for all procedures. VIO-300D (ERBE Elektromedizin GmbH, Tübingen, Germany) or ESG-400 (Olympus) were used for diathermy. The settings of these diathermy devices were mucosal incisions (Endo-Cut I, effect 1, duration 1–4, interval 1), submucosal dissection (swift coagulation, effect 4, 25–30 W), and coagulation (soft coagulation, effect 4, 80 W) for VIO-300D, and mucosal incisions (Pulse cut fast, effect 1, 50 W), submucosal dissection (forced coagulation, effect 3, 30 W), and coagulation (soft coagulation, effect 3, 50 W) for ESG-400. Either the DualKnife (KD-650Q; Olympus) or Flushknife (DK2620J-B15S-; Fujifilm) was used as an electrosurgical knife. In both groups, a conical cap (ST hood, DH-33GR; Fujifilm) and 0.4% sodium hyaluronate solution (MucoUp; Seikagaku Corp., Tokyo, Japan) for submucosal injection were used. When performing PCM-CT, a general purpose, reopenable clip with 16 mm opening width (SureClip Plus, ROCC-F-26–235-C; Micro-Tech Co., Ltd. Nanjing, China) was used as the traction device.
The assistant (expert endoscopist) was allowed to take over from the operator for patient safety in the event of intraoperative perforation, intraoperative uncontrollable bleeding, or when the operator was unable to continue the procedure appropriately. In addition, when the procedure time exceeded 2 hours, the assistant was allowed to take over to avoid operator fatigue and diminished concentration; these decisions were made by the assistant of the procedure.
Pocket-creation method with/without single-clip traction
The PCM was performed as previously reported [55 ]
[66 ]
[1313 ]
[1414 ]
[1515 ]. PCM-CT was also performed as previously reported [88 ]
[99 ] ([Fig. 1Fig. 1 ], [Fig. 2Fig. 2 ]).
Fig. 1
Fig. 1 Pocket-creation method with single-clip traction (PCM-CT). First, 0.4% sodium hyaluronate solution is injected into the submucosa. a Mucosal incision. b Creation of a submucosal pocket under the tumor in the same manner as for the pocket-creation method (PCM). c Circumferential mucosal incision around the lesion. d Normal mucosa from the anal side of the partially resected tumor is grasped with a reopenable clip, without deploying it. e The entrapped mucosa is pulled toward the opposite wall and the clip further captures the opposing mucosa. Capture of both the tumor and opposing mucosa is visually confirmed and the clip is then deployed. f PCM-CT stretches the submucosa, exposing the appropriate submucosal dissection plane.
Fig. 2
Fig. 2 Sequential pictures of the pocket-creation method with single-clip traction. a A 20-mm laterally spreading tumor, nongranular, pseudo-depressed type in the sigmoid colon; 0.4% hyaluronic acid was injected. b An initial mucosal incision was made 1 cm distant from the tumor. c Creating the submucosal pocket in the same manner as for the pocket-creation method. d A circumferential incision was made after creation of the submucosal pocket under the tumor. e Grasping the anal edge of the partially dissected specimen with a reopenable clip. f Attaching the specimen to the mucosa of the opposite intestinal wall. g The remaining submucosa is stretched to facilitate resection. h The connecting reopenable clip is finally removed with grasping forceps after completion of the submucosal dissection. i Pinned resected specimen. Pathology reported a T1 cancer of well differentiated adenocarcinoma with 900 µm submucosal invasion, positive lymphovascular invasion, and negative margins.
Randomization and blinding
Tumor morphology and the institution were used for randomization. Tumor morphology was used to assign protruded lesions (0-I) and laterally spreading tumors of granular type as sessile types, and nongranular laterally spreading tumors as flat types. Randomization was performed using Research Electronic Data Capture (Vanderbilt University, Nashville, Tennessee, USA) as the electronic data capture system, and a stratified permuted block randomization method. Blinding to study participants and pathologists was conducted. Histopathological diagnosis followed the World Health Organization classification [1616 ]: low grade adenoma, high grade dysplasia, and T1 cancer. Colonic ESD operators and assistants were determined before the start of ESD, and the operator and assistant were not informed of the PCM or PCM-CT group assignment based on randomization until the start of ESD.
Definition of recorded data
ESD completion was defined as completion of ESD with an en bloc resection using the assigned ESD method without changing to the other method or changing operator during the procedure.
The procedure time was defined as the time from the start of the first mucosal incision to the end of the lesion excision. The dissection speed was defined as “short diameter of excised specimen × long diameter of excised specimen × 0.25 × 3.14/procedure time” [1717 ].
En bloc resection was defined as the resection of the lesion in one single piece. R0 resection was defined as the histopathological affirmation of an en bloc resection with negative vertical and horizontal margins.
Postoperative bleeding and perforation were documented as adverse events. Delayed bleeding was defined as overt bleeding occurring within 14 days after ESD, that resulted in a hemoglobin level drop of at least 2 g/dL and required endoscopic hemostasis or blood transfusion [1818 ]. Perforation was defined as a deep mucosal injury that resulted in direct exposure between intraperitoneal and intestinal lumen during or after the procedure; the former was defined as intraoperative perforation, and the latter as delayed perforation. Intraoperative perforation was diagnosed when the intraperitoneal cavity was confirmed by endoscopic visualization. Delayed perforation was diagnosed when evidence of free air was seen at computed tomography scan or X-ray.
In addition, the following information specific to the current study was defined (Fig. 1s ). The time from the first mucosal incision to completion of the submucosal pocket was defined as the pocket-creation time. The remaining time until the complete excision of the lesion was defined as the pocket-opening time. The procedure time was the total of the pocket-creation time and pocket-opening time.
The following information was additionally defined in the PCM-CT group. The time from the appearance of the clip on the endoscopic view to the completion of clip deployment prior to PCM-CT was defined as the clip deployment time. If a clip fell off during the procedure and had to be redeployed, the additional clip deployment time was added.
Outcomes
The primary outcome of this study was the dissection speed, which is commonly used to estimate the efficiency of ESD. The procedure time, pocket-opening time, rate of en bloc resection, rate of R0 resection, and rate of adverse events were analyzed as secondary outcomes. Subgroup analyses were conducted for tumor location, morphology, and size (<30 mm or ≥30 mm).
Sample size
PCM had a dissection speed of 16.0 mm2 /min in a previous multicenter trial [1414 ], and PCM-CT had a dissection speed of 20 mm2 /min in 30 consecutive cases in a recent retrospective study [99 ]. Therefore, we expected that PCM-CT could increase the efficiency of pocket opening by about 20% compared with PCM. To detect a significant difference between the groups with a significance level of 0.05 (two sided) and a power of 80%, 45 patients in each group were required and 90 patients in total; considering 10% dropouts, the study population was set at 100 patients. EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [1919 ], a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria), was used for the calculation.
Statistical analysis
The primary outcomes were analyzed based on the intention-to-treat analysis. Categorical data were analyzed using the chi-squared test or Fisher’s exact test. Continuous variables, which are quantitative data, were compared using the Mann–Whitney U test or t test. P < 0.05 (two sided) was considered significant. All statistical analyses were performed using EZR.
Results
Participant flow and recruitment
From May 2021 to May 2023, 100 patients with 100 colonic tumors were included in the trial and randomly assigned to the PCM-CT group (n = 49) or PCM group (n = 51) ([Fig. 3Fig. 3 ]).
Fig. 3
Fig. 3 Flow chart of the study. ITT, intention to treat; PCM-CT, pocket-creation method with single-clip traction; PCM, pocket-creation method.
Baseline data
Baseline characteristics of patients, lesions, and experience of each operator are presented in [Table 1Table 1 ]. A total of 70 men and 30 women, with a mean age of 67.8 (SD 11.3) years, were included.
Table 1
Table 1 Baseline characteristics of patients and procedures.
Characteristics
PCM-CT
(n = 49)
PCM
(n = 51)
PCM-CT, pocket-creation method with single-clip traction; PCM, pocket-creation method.1 ERBE Elektromedizin GmbH, Tübingen, Germany.2 Olympus, Tokyo, Japan.3 Fujifilm, Tokyo, Japan.4 Flat type: laterally spreading tumor (LST), nongranular type; elevated type: protruded lesion (0-I) and LST, granular type.
Age, mean (SD), years
67.9 (12.8)
67.7 (9.8)
Sex
9 (18)
21 (41)
40 (82)
30 (59)
Antithrombotic medication
8 (16)
6 (12)
41 (84)
45 (88)
Diathermy
43 (88)
46 (90)
6 (12)
5 (10)
Electrosurgical knife
40 (82)
39 (77)
9 (18)
12 (24)
Tumor location
5 (10)
8 (16)
21 (43)
16 (31)
10 (20)
10 (20)
2 (4)
5 (10)
11 (22)
12 (24)
Morphology4
23 (47)
23 (45)
26 (53)
28 (55)
Operators [previous colorectal ESDs]
11 (22)
11 (22)
6 (12)
5 (10)
13 (27)
11 (22)
8 (16)
11 (22)
6 (12)
9 (18)
4 (8)
4 (8)
1 (2)
0 (0)
ESD completion rate
One of the 49 patients in the PCM-CT group had their ESD discontinued due to abdominal pain associated with intraoperative perforation and was excluded from any further analysis ([Fig. 3Fig. 3 ]). In the PCM-CT group, four patients had a change of operator during the procedure, including one case of overtime and three cases of procedural difficulties. In the PCM group, seven patients had a change of operator during the procedure, including two cases of overtime and five cases of procedural difficulties. In addition, one patient who had an operator switch due to procedural difficulties also had a change of assigned procedure, from PCM to PCM-CT.
Treatment results and outcomes
ESD data were collected from 99 patients and are shown in [Table 2Table 2 ]. The mean (SD) dissection speeds were 27.0 (14.5) and 21.4 (10.8) mm2 /min in PCM-CT and PCM groups, respectively, which were significantly different (95%CI 0.5 to 10.7, P = 0.03). In the PCM-CT group, both the pocket-opening time and procedure time were lower compared with the PCM group, but the differences were not statistically significant. The rate of en bloc resection, R0 resection, and adverse events were not significantly different between the groups. Clinicopathological features of the resected lesions are shown in [Table 2Table 2 ].
Table 2
Table 2 Endoscopic submucosal dissection-related data (intention-to-treat analysis).
PCM-CT
(n = 48)
PCM
(n = 51)
95%CI
P value
PCM-CT, pocket-creation method with single-clip traction; PCM, pocket-creation method; N/A; not applicable.
Data are mean (SD) unless otherwise stated.
Hyaluronic acid, mL
57.8 (29.8)
62.3 (37.8)
–18.2 to 9.1
0.51
Procedure time, minutes
64.8 (47.6)
81.8 (57.9)
–38.2 to 4.3
0.12
Dissection speed, mm2 /min
27.0 (14.5)
21.4 (10.8)
0.5 to 10.7
0.03
25.6 (12.5)
16.3 (9.2)
–0.5 to 19.0
0.06
18.9 (6.6)
21.9 (10.0)
–14.5 to 8.3
0.56
41.0 (14.7)
32.9 (13.4)
–3.9 to 20.1
0.17
21.1 (12.0)
18.7 (6.6)
–6.9 to 11.6
0.60
18.6 (8.6)
17.2 (4.7)
–6.0 to 8.8
0.69
25.6 (6.9)
19.8 (7.1)
–6.4 to 17.9
0.29
5.6
N/A
N/A
N/A
22.6 (11.1)
18.3 (8.3)
–1.2 to 9.7
0.13
31.3 (16.3)
24.8 (12.3)
–1.8 to 14.9
0.12
Pocket-creation time, minutes
34.9 (22.1)
44.0 (29.0)
–19.4 to 1.3
0.08
Pocket-opening time, minutes
30.0 (28.9)
37.8 (33.0)
–20.2 to 4.6
0.22
En bloc resection, n (%)
48 (100)
51 (100)
0
>0.99
Delayed perforation, n (%)
0 (0)
0 (0)
0
>0.99
Delayed bleeding, n (%)
2 (4)
5 (10)
0.04 to 2.6
0.44
R0 resection, n (%)
48 (100)
49 (96)
0.2 to Inf
0.50
Tumor size, mm
32.4 (10.8)
34.4 (15.1)
–7.2 to 3.3
0.45
Specimen size, mm
44.7 (11.7)
46.3 (16.4)
–7.4 to 4.1
0.57
Operators switched, n (%)
4 (8)
7(14)
0.1 to 2.5
0.53
Histology, n (%)
21 (44)
16 (31)
0.46
21 (44)
28 (55)
6 (13)
5 (83)
1 (17)
7 (14)
5 (71)
2 (29)
1 (2)
2 (4)
0.08 to 96.4
>0.99
2 (4)
0 (0)
0 to 4.1
0.19
0 (0)
1 (2)
0.02 to Inf
>0.99
0 (0)
1 (2)
0.02 to Inf
>0.99
We also performed a per protocol analysis ([Table 3Table 3 ], [Fig. 3Fig. 3 ]) in 88 patients with completed ESD. This analysis of PCM-CT vs. PCM showed mean (SD) dissection speeds of 28.3 (14.3) vs. 22.8 (10.6) mm2 /min (95%CI 0.1 to 10.8, P = 0.045), procedure times of 58.3 (39.8) vs. 70.3 (42.7) minutes (95%CI –29.5 to 5.4, P = 0.17), and pocket-opening times of 27.1 (25.1) vs. 31.8 (25.5) minutes (95%CI –15.4 to 6.0, P = 0.39). In this subgroup analysis, the difference in dissection speed between PCM-CT and PCM group was statistically significant (P = 0.045).
Table 3
Table 3 Endoscopic submucosal dissection-related data (per protocol analysis).
PCM-CT
(n = 44)
PCM
(n = 44)
95%CI
P value
PCM-CT, pocket-creation method with single-clip traction; PCM, pocket-creation method; N/A, not applicable; Inf, infinity.
Data are mean (SD) unless otherwise stated.
Hyaluronic acid, mL
56.8 (28.3)
58.4 (36.2)
–15.5 to 12.1
0.81
Procedure time, minutes
58.3 (39.7)
70.3 (42.7)
–29.5 to 5.4
0.17
Dissection speed, mm2/min
28.3 (14.3)
22.8 (10.6)
0.1 to 10.8
0.045
25.9 (13.1)
18.8 (8.1)
–3.7 to 17.8
0.18
20.6 (5.6)
21.9 (10.0)
–13.2 to 10.5
0.80
41.0 (14.7)
32.9 (13.4)
–3.9 to 20.1
0.17
22.6 (12.4)
21.6 (6.4)
–11.7 to 13.7
0.87
18.6 (8.6)
17.2 (4.7)
–6.0 to 8.8
0.69
25.6 (6.9)
19.8 (7.1)
–6.4 to 17.9
0.29
N/A
N/A
N/A
N/A
23.7 (11.3)
20.4 (7.9)
–2.9 to 9.4
0.29
32.5 (15.7)
24.8 (12.3)
–0.6 to 15.9
0.07
Pocket-creation time, minutes
31.2 (17.5)
38.5 (21.2)
–15.6 to 0.9
0.08
Pocket-opening time, minutes
27.1 (25.1)
31.8 (25.5)
–15.4 to 6.0
0.39
En bloc resection, n (%)
44 (100)
44 (100)
0
>0.99
Delayed perforation, n (%)
0 (0)
0 (0)
0
>0.99
Delayed bleeding, n (%)
2 (5)
4 (9)
0.3 to 24.2
0.68
R0 resection, n (%)
44 (100)
42 (95)
0.2 to Inf
0.49
Tumor size, mm
32.1 (10.4)
33.5 (13.7)
–6.6 to 3.7
0.58
Specimen size, mm
44.3 (11.2)
45.6 (14.9)
–6.9 to 4.2
0.64
Histology, n (%)
20 (46)
14 (32)
0.36
18 (41)
24 (55)
6 (14)
5 (83)
1 (17)
6 (14)
4 (67)
2 (33)
1 (2)
2 (5)
0.08 to 101
>0.99
2 (5)
0 (0)
0 to 4.2
0.19
0 (0)
1 (2)
0.03 to Inf
>0.99
0 (0)
1 (2)
0.03 to Inf
>0.99
Subgroup analysis
Subgroup analyses were conducted for tumor location, morphology, and size (<30 mm or ≥30 mm) in relation to the dissection speed. There was a significant difference in mean (SD) dissection speed in favor of the PCM-CT group for lesions in the right colon (28.5 [15.7] vs. 21.4 [9.5] mm2 /min [95%CI 0.8 to 13.3], P = 0.03)] and especially in the ascending colon (31.4 [17.8] vs. 21.3 [9.9] mm2 /min [95%CI 0.1 to 20.2], P = 0.048)].
Adverse events
One patient in the PCM-CT group had their ESD discontinued due to intraoperative perforation with uncontrollable abdominal pain. The patient was excluded from any further analysis as no ESD data were collected. He was managed by endoscopic clip closure of the muscle defects and conservative treatment with fasting and prophylactic intravenous antibiotics. There was no delayed perforation or intraprocedural uncontrolled bleeding in either group. Delayed bleeding was documented in two patients in the PCM-CT group and in five patients in the PCM group. There were no fatal adverse events during the study period.
Discussion
ESD has been accepted worldwide as minimally invasive treatment of superficial colorectal cancer [44 ]. Although it is still a clinically challenging procedure, it is gradually becoming safer and easier due to the development of dedicated attachments [2121 ]
[2222 ], traction methods [2323 ]
[2424 ]
[2525 ]
[2626 ]
[2727 ], and resection strategies such as PCM [55 ]
[66 ]
[1313 ]
[2828 ]. Overall, effective cancer management requires complete local resection and assessment of metastasis risk. In particular, the risk factors are associated with the submucosa of the resected specimen. Therefore, the goal of endoscopic treatment is to achieve complete local excision of the tumor with submucosa of sufficient depth to ensure negative margins.
The PCM procedure secures an adequate depth of submucosa in the resected specimen, with minimal thermal injury, to allow the risk of lymph node metastasis to be determined [2828 ]
[2929 ]. PCM-CT and PCM both achieved high R0 resection rates in the present study ([Table 2Table 2 ]). Given these advantages, PCM has proven to be effective in colorectal ESD regardless of tumor size, morphology [1313 ]
[1515 ], or location [3030 ], and is a useful strategy among other techniques available in ESD [1414 ]
[3131 ]. In addition, the effectiveness of PCM has also been reported for gastric [3232 ]
[3333 ] and duodenal [3434 ] ESD. However, during pocket opening in PCM procedures, the submucosal pocket can be challenging, as the more the pocket is opened, the less stable the endoscope tip becomes.
PCM-CT was developed to overcome the technical difficulty associated with opening the submucosal pocket. Our results revealed that PCM-CT had significantly better performance than conventional PCM in dissection speed (27.0 [14.5] vs. 21.4 [10.8] mm2 /min [95%CI 0.5 to 10.7], P = 0.03). The mean clip deployment time in PCM-CT was 2.3 minutes, and the overall procedure time was much shorter than that with PCM (64.8 vs. 81.8 minutes). As a side note, we cannot accurately calculate the dissection speed for opening the pocket because we cannot measure only the area around the pocket on a resected specimen.
This study shows that the addition of clip traction reproducibly improves the conventional PCM procedure. Most conventional traction techniques are dedicated devices or hand-made combinations of clip and a connecting part. Most PCM-CTs were completed using only a single general purpose reopenable clip as a traction device for each procedure, indicating that PCM-CT was also cost-effective. The SureClip Plus that was used, costs 3500 Japanese Yen (JPY), and is cheaper than other dedicated traction devices; for example, the S-O clip (TC1H05; Zeon Medical Inc. Tokyo, Japan) is 5000 JPY, SureClip Traction Band (ETD00005, ETD00006; Micro-Tech Co., Ltd.) is 5200 JPY, ProdiGI Traction Wire (ERD-TW20, ERD-TW35; Medtronic, Minneapolis, USA) is 20 500 JPY, and FlexLifter (LA-400; Olympus) is 35 000 JPY. Furthermore, the single-clip traction could connect the specimen to the contralateral wall by decreasing the amount of luminal gas, and adjust the traction force by increasing or decreasing the gas. The application of a traction clip from the early stages of the ESD could be useful. However, based on our experience, achieving sufficient submucosal elevation through local injection of sodium hyaluronate, followed by mucosal incision and several superficial submucosal dissections while gently lifting the mucosa with the sheath portion of the knife, facilitates easy insertion of the tip of the ST hood into the submucosal layer in most cases. In contrast, if traction clips are placed immediately after incision this could interfere with the subsequent procedure, and could also damage the specimen owing to unintentional traction force. In PCM, traction is usually necessary only when the pocket is opened, as the endoscopic manipulation becomes unstable. Recently, double-clip traction ESD reported by Bordillon et al. [2020 ] achieved a significantly faster dissection speed (39.4 mm2 /min) than PCM-CT (27.0 mm2 /min). These excellent outcomes might be due to participating endoscopists having more ESD experience compared with our nonindependent operators. Meanwhile, a Japanese multicenter randomized controlled trial of traction-assisted ESD (CONNECT-C trial) [2626 ] showed a slower dissection speed (16 mm2 /min) than in our study. The CONNECT-C trial concluded that the traction method did not significantly shorten ESD time but could be useful for large tumors or nonexpert operators.
Subanalysis of the present study revealed that PCM-CT had a significantly faster dissection speed than PCM in the right colon (28.5 [15.7] vs. 21.4 [9.5] mm2 /min [95%CI 0.8 to 13.3], P = 0.03), especially in the ascending colon (31.4 [17.8] vs. 21.3 [9.9] mm2 /min [95%CI 0.1 to 20.2], P = 0.048). This could be due to the stretchability and thickness of the submucosal tissue in the ascending colon, an observation based on our extensive experience with colonic ESD. Therefore, traction force will stretch the submucosal tissue more effectively and secure good visibility of the submucosa, thus facilitating submucosal dissection.
Histopathology of the ESD specimens revealed 37% (37/99) low grade adenoma. According to Japanese guidelines and the medical health insurance system, ESD should be offered to patients with endoscopically suspected early cancer including high grade dysplasia regardless of the final histopathological report. This might account for some cases of overestimation at optical diagnosis when a large lesion was considered too big for a proper en bloc resection with conventional EMR. Additionally, it should be noted that the diagnostic yield of JNET type 2B is still controversial [3535 ].
This study has some limitations. First, the operating endoscopists were not blinded, although patients and pathologists were completely blinded to the assigned group. Nevertheless, we cannot exclude that some operating endoscopists unconsciously preferred one technique over the other. Notwithstanding this lack of blinding, all operators were trainees and as such, they were evaluated on their speed, effectiveness, and performance, and regardless of their preferred technique they were always focused on the best outcome for every lesion. Second, the decision to change the operators was subjective and it was done only after consultation with the assistant endoscopist; however, switching was based on definite criteria. We think that this was an adequate measure to support the fairness of this study bias when comparing ESD completion rates between the two groups. Third, the skill level of the operating endoscopists could not be completely equal, although only endoscopists with experience of between 20 and 100 colorectal ESDs at the beginning of the study were included. Rectal lesions were excluded, as rectal ESD is considered less challenging; the procedure is naturally assisted by the gravity traction of changing a patient’s body position. Fourth, the rate of delayed bleeding (7/99 [7%]) in this study was higher than previously reported [99 ]
[1616 ]. Six of the seven patients with delayed bleeding underwent endoscopic hemostasis despite there being no hemoglobin drop of ≥2 g/dL or the need for blood transfusion. Moreover, in 5/6 patients, the bleeding had already stopped spontaneously at the time of endoscopic intervention. Only one patient in the PCM group had a hemoglobin drop of 3.0 g/dL that required blood transfusion. Fifth, although both “dissection speed” and “procedure time” had been registered as primary outcomes in the University Hospital Medical Network Clinical Trials Registry, we decided to change the “procedure time” to a secondary outcome. Sixth, adjustment factors should have also included lesion size and location; however, when this study was designed, the population was not large enough, and thus these two factors were reserved for a secondary analysis. Finally, one patient in the PCM-CT group was excluded from the analysis after discontinuation of the procedure due to uncontrollable abdominal pain caused by intraoperative perforation during pocket opening. It is important to note that intraoperative perforation with mucosal incision in the pocket opening is not specific to PCM-CT. The patient was discharged after conservative treatment following immediate clip closure of the perforation; laparoscopic surgery was performed subsequently to remove the lesion.
Conclusion
According to this study, PCM-CT appeared to further improve PCM by increasing the resection speed. Additionally, both PCM-CT and PCM achieved high R0 resection rates in the present study, thus reinforcing the pre-existing knowledge that PCM is highly effective for colonic ESD.
