Introduction

Reported rates of delayed bleeding (DB) after endoscopic resection with each direct oral anticoagulant (DOAC) are high and heterogeneous. This large-scale multicenter study analyzed lesions resected by colorectal endoscopic submucosal dissection (ESD) under DOACs to compare the DB of each DOAC with warfarin.

Aims & Methods

We retrospectively reviewed 1,019 lesions with DOAC and 459 lesions with warfarin among 34,455 cases of ESD in 47 Japanese institutions from 2012 to 2021. The rate of DB in each DOAC was compared with warfarin. Risk factors of DB for DOAC and warfarin were also investigated.

Results

In the DOAC and warfarin groups, the mean age was 75.1±13.7 and 73.6±10.4 years, and the male rate was 70.2% and 71.4%, respectively. The mean tumor sizes (mm) were 29.5±14.0 and 30.3±16. The en bloc resection rates were 97.5% and 96.5%, respectively. Among 31,968 ESD cases from 46 institutions, the rate of DB was significantly higher in cases with anticoagulation (10.08%, 149/1478) than in cases without anticoagulation (2.21%, 677/30526) (p<0.001). Among the total 34,455 ESD cases, the rates of DOAC and warfarin were 9.32% (95/1019) and 11.76% (54/459), respectively (p=0.149).
The rate of DB with dabigatran (18.26%, 21/115) was significantly higher than with apixaban (10.08%, 25/248, p=0.029), edoxaban (7.73%, 25/323, p=0.001), and rivaroxaban (7.21%, 24/333, p=0.001). The rate of dabigatran was slightly higher than that of warfarin (11.76%, 54/459, p=0.055) and only the rate of rivaroxaban was significantly lower than that of warfarin (p=0.033).
Multivariate analysis showed that the significant risk factors of DB for DOAC were heparin exchange (odds ratio (OR), 95% confidence interval (CI), 2.185, 1.277-3.739, p=0.005), rectal site (2.013, 1.282-3.161, p=0.002), and longer procedure time ≥55 min (2.434, 1.498-3.956, p<0.001). The significant risk factors of DB for warfarin were antiplatelet combination (2.286, 1.274-4.100, p=0.006), rectal location (2.261, 1.248-4.095, p=0.007), and tumor size ≥35mm (2.019, 1.163-3.652, p=0.008).
The number of significant risk factors and the risk of DB (OR, 95%CI) for DOAC were 2.136 (1.301-3.507) and 4.53 (2.525-8.155) for one and two risk factors, respectively. Those for warfarin were 1.646 (0.742-3.650), 3.624 (1.593-8.243), and 13.037 (3.929-43.266) for one, two, and three risk factors, respectively.

Conclusion

The DB rate of dabigatran was higher than that of other DOACs, and only rivaroxaban was significantly lower than that of warfarin. The risk factors for DB with DOACs have been elucidated.

Introduction

Delayed perforation (DP) is a serious complication after colorectal endoscopic submucosal dissection (CESD), often requiring emergent surgery. Traditionally, DP has been managed conservatively or surgically. Recently, successful endoscopic closures have been reported. Since 2021, we have attempted endoscopic closure in patients meeting the following: (1) localized symptoms with stable condition, (2) no or minimal oral intake, and (3) endoscopic accessibility and feasibility of closure.

Aims & Methods

To compare clinical outcomes before and after the introduction of endoscopic closure for DP and to examine closure techniques. Among Among 4,848 CESD procedures (April 2008–December 2024), 16 cases (0.3%) of DP were identified. In Analysis 1, patients were divided into Group A (before closure: 2008–2020) and Group B (after closure: 2021–). In Analysis 2, closure techniques were reviewed. DP was defined as a perforation undetected during ESD but confirmed postoperatively by imaging or endoscopy.

Results

Analysis 1: Group A: 9 patients; Group B: 7. Median age: 72 vs. 70 years; male/female: 5/4 vs. 4/3. Lesion locations (cecum/ascending/transverse/sigmoid/rectum): Group A, 3/2/1/1/2; Group B, 1/3/2/1/0. Median specimen size: 48 mm vs. 46 mm.
Ulcer base closure (complete/partial/none): Group A, 2/2/5; Group B, 2/4/1. Histology (adenoma/serrated lesion/carcinoma in situ/adenocarcinoma/others): Group A, 3/4/1/1; Group B, 4/1/1/1. Median time to DP diagnosis: 63 vs. 55 hours.
Treatment in Group A was determined case-by-case; Group B applied defined criteria. Initial treatments (conservative/surgical/endoscopic): Group A, 5/4/0; Group B, 1/0/6. One case in each group initially managed conservatively required surgery later. Of 7 surgical cases, all underwent partial colectomy or drainage without stoma creation. All 6 endoscopic closure cases avoided surgery. Surgical avoidance rate: 33% vs. 86% (p=0.06).
Median time to oral intake: 8 vs. 7 days (p=0.67). Median hospitalization: 16 vs. 11 days (p<0.05).
Analysis2: All 6 closures were performed without bowel preparation. Two patients had resumed oral intake, but endoscopy was performed early.
Perforation diameter: median 10 mm (5–15). Closure methods: over-the-scope clip (OTSC; n=1), hemostatic clips (n=4), clips with polyglycolic acid (PGA) sheet (n=1).

Conclusion

Endoscopic closure for DP was associated with higher surgical avoidance and shorter hospitalization, without worsening outcomes. It may offer an effective treatment option for selected patients.​

Introduction

Endoscopic submucosal dissection (ESD) is emerging as the standard treatment for suspected superficial submucosal invasion lesions. Considering the global population ageing, understating the feasibility of this approach in elderly patients is of the utmost importance. Studies in this field have shown a tendency towards reduced resection rates and higher rates of complications.

Aims & Methods

We sought to assess the effectiveness and safety of colorectal ESD for patients aged ≥ 80 yo compared to 65-79 yo patients. We prospectively enrolled consecutive patients aged >64 years undergoing colorectal ESD between January 2019 and December 2023 from ten Italian Tertiary Referral Centres. Patients were classified as Elderly Group (E-Group, age >80 yo) and non-elderly group (NE-Group, 65-79 yo). Procedure outcomes and adverse events were compared between E-Group and NE-Group and between patients with comorbidities and healthy (1-CM-Group and H-Group).

Results

A total of 980 patients were included [269 (27.5%) in the E-Group and 711 (72.5%) in the NE-Group]. Table 1 summarises the principal data and outcomes.

En-bloc resection and R0 rate did not statistically differ (p=0.30; p=0.05) between E and NE-Groups. Oncological curative resection was achieved in the two groups without significant differences (p=0.41). Histological outcomes were comparable between the two groups, except for lymphovascular invasion [E:14 (5.2%) vs NE:14 (2.0%); p=0.007 – OR:2.9 (CI:1.3-6-6)]. Regarding clinical course, intraprocedural and 14 days-adverse events (AEs), no statistically significant differences were found between the two groups, except for a higher rate of post-procedure delirium in the E-Group [6 (2.2%) vs 1 (0.1%); p=0.001–OR:16.2 (CI:1.9-135.2)]. 1-CM-Group showed a higher rate of intraprocedural bleeding (p=0.001), fever onset (p<0.001), C-reactive protein (p=0.006) and procalcitonin increase (p=0.02), delayed perforation (p=0.03) and systemic infections (p=0.02) compared to the H-Group. The presence of one or more comorbidity, but not the age, was associated with a poorer clinical course [p<0.001; OR:2.3 (CI:1.5-3.6)] and poor overall course (poor clinical course combined with 14 days AEs) [p=<0.001; OR:2.1 (CI:1.4-3.0)]. ASA classification (OR: 1.36; CI:1.01-1.83), the presence of one comorbidity (OR: 1.73; CI:1.14-2.63), and procedure time (OR: 1.01; CI:1.00-1.01) were poor outcome predictors. After excluding patients with at least one comorbidity, still no differences in terms of poor overall course (p=0.21) were registered between E and NE-Group.

Conclusion

Colorectal ESD is an effective and safe procedure in elderly patients. Physicians should ackowledge delirium as a possible AE in patients aged over 80. The presence of comorbidities should drive the execution of a comprehensive pre-procedural assessment and post-procedural care, to optimize outcomes and minimize complications.

Introduction

Clinically significant delayed bleeding (CSDB) is the most frequent complication after colorectal endoscopic submucosal dissection (ESD). The Limoges Bleeding Score was recently developed to identify patients at risk of CSDB after colorectal ESD. The efficacy of prophylactic clipping to prevent bleeding has been a topic of ongoing debate.

Aims & Methods

This study aimed to evaluate the real effectiveness of complete clip closure in preventing CSDB after colorectal ESD.
We performed a retrospective analysis of data prospectively collected from ten-centers between 2017 and 2023. Specifically, pre- and periprocedural data were collected prospectively, while post-procedural data, including the evaluation of CSDB, was retrospectively assessed. All epithelial colorectal lesions treated by ESD were included. The patients were allocated into the closure and non-closure groups which were compared before and after propensity-score matching based on risk factors of the Limoges Bleeding Score (age > 75 yrs, size > 50 mm, ASA III-IV, location in rectum and anticoagulant/antiplatelet). We further analyzed post-ESD CSDB in the subgroups of patients on anticoagulants and with high risk of bleeding (5 to 8 points according to Limoges score).

Results

A total of 3431 patients were included, 1236 in the closure group and 1985 in non-closure group. The propensity score matching based on risk factors of the Limoges Bleeding Score created 1064 matches pairs. The rate of CSDB was not significantly different between the two groups (closure group 7.04% vs non-closure group 6.8%; p = 0.85), nor after propensity score matching (closure group 7.47% vs non-closure group 6.11%; p = 0.38). The subgroup analysis including patients on anticoagulants (165 patients in closure group vs 251 in non-closure group) and those identified at high-risk by Limoges score (181 in each group), found no significant difference of CSDB between the groups (p= 0.31 and p =0.16, respectively).

Conclusion

In our population, prophylactic clipping after colorectal ESD did not significantly reduce the incidence of CSDB. These findings were confirmed even when the analysis was restricted to high-risk patients. Considering the technical challenges of complete clip closure, it could be reserved for selected clinical scenarios.

References

[1] Saito H, Igarashi K, Hirasawa D et al. The risks and characteristics of the delayed bleeding after endoscopic submucosal dissection for early gastric carcinoma in cases with anticoagulants. Scand J Gastroenterol 2020; 55: 1253–1260
[2] Boda K, Oka S, Tanaka S et al. Clinical outcomes of endoscopic submucosal dissection for colorectal tumors: a large multicenter retrospective study from the Hiroshima GI Endoscopy Research Group. Gastrointest Endosc 2018; 87: 714–722
[3] Bordillon P, Pioche M, Wallenhorst T et al. Double-clip traction for colonic endoscopic submucosal dissection: a multicenter study of 599 consecutive cases (with video). Gastrointest Endosc 2021; 94: 333–343
[4] Probst A, Ebigbo A, Märkl B et al. Endoscopic submucosal dissection for early rectal neoplasia: experience from a European center. Endoscopy 2017; 49: 222–232
[5] Albouys J, Montori Pina S, Boukechiche S et al. Risk of delayed bleeding after colorectal endoscopic submucosal dissection: the Limoges Bleeding Score. Endoscopy 2024; 56(2):110-118

Introduction

Piecemeal endoscopic mucosal resection (EMR) vs endoscopic submucosal dissection (ESD) has been a long standing debate in the endoscopic community worldwide with multiple studies done to evaluate the preferred method to manage large colorectal neoplasms. Piecemeal EMR with thermal ablation of the margin has been strongly advocated and reported by certain groups to be a simple, inexpensive and effective technique and this has been preferred widely due to its perceived ease of performance and shorter learning curve.¹ However, close, repeated surveillance colonoscopy is recommended even when the lesion is thought to have been completely resected² as residual tumors can cause interval cancers.³ Some groups, mainly from east Asian countries, have recommended en bloc resection by ESD due to significantly lower risk of local recurrence and precise pathological diagnosis.⁴ But its usage is limited due to concerns about risk of serious adverse events. With improvement in training and devices, colorectal ESD is now being performed by various centers globally. This study aims to report the outcomes of an international, multicenter study on colorectal ESD.

Aims & Methods

Patients who underwent colorectal ESD for lesions greater than 20mm between January 2022 to April 2024 from centers in USA, Japan, and India respectively were included in this study. Recurrent lesions after endoscopic therapy, neoplasms identified in patients with ulcerative colitis, and non-epithelial lesions were included. Primary outcomes were en bloc and R0 resection rate, delayed adverse events (bleeding/perforation), and local recurrence rate. Secondary outcomes were procedure time, lesion size, and pathology. Univariate and multivariate analysis were done to identify the risk factors for primary outcomes.

Results

309 patients (median size, 40(27.5-60)mm; location rectum 52.4%, left colon 26.2%, right colon 21.4%) were involved in this study. The en bloc resection rate was 96.1% and R0 resection rate was 90.3%. 3 patients experienced delayed bleeding which was managed endoscopically. 2 patients underwent surgery for delayed perforation. 6 patients had a local recurrence. 67(23%) patients had intramucosal carcinoma on pathology. Univariate analysis showed lesser procedure time (median(IQR) 145(80-236.25) vs 270(180-300) minutes, P= 0.005) and smaller lesion size (median(IQR) 40(26-56.25) vs 60(40-90) mm, P= 0.004) were significant factors for successful en bloc resection. However, lesser procedure time (median(IQR) 74(29.5-97.5) vs 150(82.5-240) minutes, P= 0.034) led to an increased delayed adverse event rate. Multivariate analysis using Cox regression showed en bloc resection (odds ratio 0.038; 95% confidence interval 0.003-0.421; P= 0.008) was an independent factor preventing local recurrence.

	Factors	Variables		P value
En bloc resection rate		En bloc resection (n=297)	Piecemeal resection (n=12)
	Procedure Time (median (IQR) minutes)	145 (80 – 236.25)	270 (180 – 300)	0.005
	Lesion Size (median (IQR) mm)	40 (26 – 56.25)	60 (40 – 90)	0.004
Recurrence rate		Recurrence (n=6)	No recurrence (n=257)
	En Bloc resection n (%)	4 (66.7)	246 (98)	0.009
Adverse events		Delayed adverse events (n=5)	No delayed adverse events (n=304)
	Procedure Time (median (IQR) minutes)	74 (29.5 – 97.5)	150 (82.5 – 240)	0.034

Conclusion

This international, multicenter study with data from three countries shows favorable outcomes for colorectal ESD for large lesions. The high en bloc and R0 resection rate, and low adverse event rate show increased global adaptation to this procedure. Reduced local recurrence is an added benefit. Data from a more centers globally will be valuable to establish ESD as the standard for large colorectal neoplasms.

References

1. Sidhu M, Shahidi N, Gupta S et al. Outcomes of thermal ablation of the mucosal defect margin after endoscopic mucosal resection: a prospective, international, multicenter trial of 1000 large nonpedunculated colorectal polyps. Gastroenterology. 2021 Jul 1;161(1):163-70.
2. Gupta S, Lieberman D, Anderson JC et al. Recommendations for Follow-Up After Colonoscopy and Polypectomy: A Consensus Update by the US Multi-Society Task Force on Colorectal Cancer. Gastrointest Endosc 2020;91:463-485.e5.
3. Rutter MD, Beintaris I, Valori R et al. World Endoscopy Organization Consensus Statements on Post-Colonoscopy and Post-Imaging Colorectal Cancer. Gastroenterology 2018;155:909-925.e3.
4. Ohata K, Kobayashi N, Sakai E et al. Long-term outcomes after endoscopic submucosal dissection for large colorectal epithelial neoplasms: a prospective, multicenter, cohort trial from Japan. Gastroenterology. 2022 Nov 1;163(5):1423-34.

Introduction

Colorectal endoscopic submucosal dissection (ESD) is an effective treatment for large colorectal tumors, but the technique is more difficult and is associated with more complications such as bleeding and perforation compared to EMR. Therefore, accurate identification of blood vessels and muscle layers is essential for safe colorectal ESD. However, it is often difficult to perform the treatment while recognizing all of the image information on the monitor. Therefore, in this study, we conducted a proof of concept study to determine whether deep learning can be used to identify the submucosa and muscular layer in images taken during ESD procedures.

Aims & Methods

We used 11 procedure videos randomly selected from cases of colorectal ESD performed at our center between January 2020 and July 2022. The videos were cropped every second as still images, and these still images were labeled with semantic segmentation (hereafter referred to as segmentation). This is a method of labeling each pixel in an image, and Artificial intelligence (AI) trained with this method can segment regions of the target image according to the tissues and devices depicted in the image.
In this study, 16 different labels were performed for each pixel of each image, including mucosa, submucosa, muscular layer, and blood vessels. Of the 11 labeled case images, 9 cases were used for training and 1 case with no duplicate training was used for validation. The training images were machine learned with a representative method of medical image segmentation called U-NET and evaluated for accuracy.

Results

Characteristics of patients and tumors of the 11 cases in this study were male to female ratio of 7:4, mean age of 71.9 years (36-83), mean tumor diameter 25.2 mm (3-54 mm), colon/rectum ratio of 8:2, adenoma to T1 cancer ratio 5:6. One thousand and five hundreds sixty-one images were used for training, and 90 images were used for evaluation as validation. The accuracy (corresponding to the number of pixels correctly answered out of the labeled pixels) of the trained AI to the validation images was 83.0%. The submucosa, muscular layers, and vessels, which have a large labeled area, were relatively discriminated.

Conclusion

The use of segmentation technology demonstrated the feasibility of real-time ESD navigation by AI. In the near future, we aim to further improve the system by increasing the number of training images.