Comparison of efficiency and safety between dual-clip and rubber band-assisted ESD and conventional ESD for colonic lateral spreading tumors (LSTs) with different levels of technical difficulty: a retrospective case–control study

This was a retrospective cohort study of patients who underwent colonic ESD for the treatment of LSTs at Binzhou Medical University Hospital, a regional medical center and general teaching hospital in Shandong Province of China. This study was approved by the ethical committee of Binzhou Medical University Hospital (Registration code 2022-LW-11). Informed consent was obtained from all patients before the procedure. All the data used in this study were collected through an endoscopic information system deployed in our medical center (Qingdao Medicon Co., Ltd, Qingdao, China). The STROBE checklist was followed to ensure the paper quality.

All cases of colonic lesions referred for ESD treatment were included between Jan 1st, 2019 and Jan 1st, 2022, when the following inclusion criteria were met: (1) laterally spreading tumor (LST) ≥ 10 mm in size or adenomatous polyp with a broad base ≥ 10 mm and (2) no evidence of lymph node metastasis based on abdominal computed tomography (CT). Exclusion criteria included the following: (1) all rectal cases; (2) cases treated with hybrid ESD technique characterized by partial submucosal dissection followed by snare-assisted resection, due to its potential impact on the outcome of operation duration; (3) cases that were referred to C-ESD procedure but the trimming process was jammed for a long time, leading to unplanned application of DCRB-ESD (unplanned DCRB-ESD cases), due to the impact on the outcome of operation duration and the introduced heterogeneity.

Yun-Shi Zhong et al. proposed a clinical score model for grading the technical difficulty of colorectal ESD (CS-CRESD), which predicts the probability of accomplishing colorectal ESD within 60 min and can be applied to grade the technical difficulty before the procedure [22]. This prediction model is based on the following four independent risk factors: tumor size (1 point when size 30–50 mm, 2 points when size ≥ 50 mm), the circumference of the lesion (2 points when circumference ≥ 2/3 and 0 points when less), lesion location (1 point for cecum, 2 points for flexure, 1 point for dentate line), and LST-NG lesion (1 point). The total score classifies the technical difficulty into easy (score = 0), intermediate (score = 1), difficult (score = 2–3), and very difficult (score ≥ 4).

Based on personally experience, the endoscopist (Q. Niu) evaluated the difficulty of the ESD procedure. The decision whether to perform C-ESD or DCRB-ESD was made by the endoscopist personally right before the markers were proposed.

ESD cases were all performed by an endoscopist (Q. Niu) who performed more than 100 gastrointestinal tract ESDs annually. The complication rate of ESDs performed by this endoscopist last year (2021) is 0.85% (two post-ESD gastral bleedings out of 236 gastrointestinal ESD cases, without any perforation).

The C-ESD procedure was conducted with a colonoscope (CF-HQ290I/CF-Q260JI, Olympus Co., Tokyo, Japan) and a high-frequency surgical device (VIO 200D, ERBE, Tübingen, Germany). A transparent plastic cap (D-201-11804; Olympus, Japan) was initially attached to the tip of the endoscope during this procedure to obtain better vision and maintain tension. Markers were pointed at least 5 mm to the edge of the LSTs using DualKnife (KD-650Q; Olympus, Tokyo, Japan). A mixed solution of indigo carmine, epinephrine, and saline was injected to fully lift the colonic mucosa layer with an injection needle (NM-200U-0523; Olympus, Japan). Then, electric trimming was conducted with a margin of 5 mm from the markers. Then, the trimming process was performed from the distal edge with the plastic cap introduced into the space of the submucosal layer. After the LSTs was fully resected, hemostatic forceps (HBF-16/1800; Micro-Tech Nanjing Co., Ltd, China) were used to perform coagulation to reduce bleeding risk. During this procedure, the patient’s position was adjusted when needed to control the direction of traction.

The DCRB-ESD procedure was basically the same as C-ESD, with the extra help of a dual-clip and rubber band to obtain better vision and tension. Rubber bait bands (SeaKnight, Shanghai, China) were adopted. The band was approximately 3 mm in outer diameter, 1.5 mm in inner diameter, and 1 mm in thickness and could be directly deployed through the instrument channel of the endoscope (Fig. 1). When traction was needed during the ESD procedure, a clip (ROCC-D-26-195, Micro-Tech Nanjing Co., Ltd, China) holding the band edge was delivered through the instrument channel and was fixed on the margin of LSTs. Then, another clip was delivered to the colon. With one branch of the clip across the band, the clip was then dragged and fixed to the proper position of the colonic wall to gain traction. After resection, the clip fixed onto the colonic wall was removed by foreign body forceps (Fig. 2).

Specimens were fixed onto a foam plate and submerged into formalin after ESD according to the JSCCR guidelines criteria [2].

According to the JSCCR guidelines criteria, en bloc resection was defined as resection without fragmentation. Histologically complete resection (R0) was defined as en bloc resection with negative horizontal margin invasion (HM0) and negative vertical margin invasion (VM0). Curative resection was defined as R0 resection with no risk of lymph node metastasis. Perforation was defined as a complete hole through the colonic muscle during the treatment or clinical evidence upon postoperative radiological findings. Bleeding was defined as clinical evidence of bleeding after ESD.

The ESD procedure duration was defined as the time duration from submucosal injection to specimen retrieval.

Operation duration was included as the primary outcome. The R0 resection rate, curative resection rate, tumor differentiation, tumor infiltration, bleeding, perforation, and additional surgery were included as secondary outcomes.

All patients in this study were referred to a routine endoscopic follow-up procedure in the outpatient department.

All calculations were conducted using the SPSS statistical software package (SPSS 26; Chicago, IL, USA). Continuous variables are presented as the mean and standard deviation. A chi-squared test was used for comparisons between categorical variables, and an independent t test was used for comparisons between continuous variables. Fisher’s exact test was used when necessary. Nonparametric statistics were performed when data did not match the normal distribution, where variables are presented as median and interquartile range. Two-sided P values less than 0.05 were considered statistically significant.

To reduce the heterogeneity of baselines, the propensity score matching (PSM) module of SPSS was employed. Variables employed in the PSM procedure were age, sex, LST size, circumference, LST location, LST classification and LST located in unfavorable location, with a matching ratio of 1:1. The match tolerance was set at 0.02.

Between Jan 1st 2019 and Jan 1st 2022, a total of 127 patients were enrolled, of whom 46 and 81 patients were grouped into the DCRB-ESD and C-ESD groups, respectively (Fig. 3). The baseline characteristics were well balanced between the two groups except for sex and LST classification (Table 1).

The median ESD procedure times were 32.0 and 41.0 min in the DCRB-ESD group and C-ESD group, respectively (p < 0.05). The operation durations of cases with different CS-CRESD scores were different (p < 0.05). In subgroups with a score of 0, DCRB-ESD spent more time than C-ESD (45.0 vs. 22.50, p < 0.05), but in subgroups with a score of 1 ~ 3, DCRB-ESD was faster than C-ESD (22.0 vs. 38.0, 42.0 vs. 52.0, 35.0 vs. 60.0, p < 0.05). In subgroups with a score of 4–5, the between-group operation duration was not significantly different due to limited cases, although DCRB-ESD was faster than C-ESD greatly (55.5 vs. 100.5 and p = 0.33, 146 vs. 172 and p = 1.0, respectively) (Table 2).

The en bloc resection rate in both groups was 100.0%. The R0 resection rates in both groups were not significantly different (92.59% vs. 97.83% in the C-ESD group and DCRB-ESD group, respectively, p = 0.10). The curative resection rates in both groups were not significantly different (91.36% in the C-ESD group and 91.30% in the DCRB-ESD group, p = 0.19). The post-ESD bleeding rates were 1.23% and 2.17%, respectively (p = 0.68). There were no perforation cases in either group. No adverse events, such as a clip dropping or rubber band rupturing, occurred in this study (Table 2).

When referring to lesion differentiation and depth of tumor invasion between the two groups, no significant difference was found (p = 0.68). Two patients in the DCRB-ESD group and three patients in the C-ESD group underwent additional surgery, and no metastasis was found (Table 2).

91.3% of the DCRB-ESDs were performed with one set of DCRB. Two sets of DCRBs were used separately in three lesions due to large lesion area, and three sets were used in one LST-NG lesion due to unclear visualization of the submucosa.

Since LST classification could potentially influence the ESD procedure time, PSM was performed, and the unbalance was corrected. The results were basically the same. Although the statistical significances in some subgroups were influenced, the overall results still follow the same trend after PSM (Fig. 4).