Underwater endoscopic mucosal resection of adenomas and colorectal serrated lesions: a prospective clinical study

Introduction

Detection and endoscopic resection of colorectal lesions such as adenomas and serrated polyps are well-established strategies for the prevention of colorectal carcinoma (CRC) [1,2]. The European Society of Gastrointestinal Endoscopy strongly recommends, with a high level of evidence, that all colorectal polyps should be resected, except for diminutive rectosigmoid lesions that are reliably considered hyperplastic [3]. Endoscopic removal of adenomas, even from tiny lesions, has been widely recommended worldwide to reduce the incidence and mortality associated with CRC [4]. More recently, serrated lesions, particularly from the right colon, formerly considered hyperplastic lesions, have been demonstrated to present a new oncogenic pathway and should also be treated similarly to adenomas [5].

Endoscopic mucosal resection (EMR) using flexible endoscopes was first described in the 1970s and the procedure remains essentially unchanged [6-8]. During EMR, a solution is injected into the submucosal space through a needle catheter to create a cushion between the muscularis propria and the lesion, which allows for a safe and effective resection using electrosurgical energy. This technique is widely used and allows the removal of colorectal adenomas of varying sizes [6]. Underwater EMR (UEMR) is a technique first described by Binmoeller et al [9] and subsequently reported in other studies, which confirmed it was easy to learn, safe and effective [6,9-18]. In contrast to EMR, the intestinal lumen is filled with water instead of air/CO₂, and no submucosal injection is added. Conceivably, submucosal injection reduces the risk of iatrogenic perforation and thermal injury to the deeper tissue layers. However, on some occasions it can paradoxically make capture of flat lesions difficult, increase tension in the intestinal wall secondary to air insufflation and decrease the space within the organ lumen, limiting the working area. Furthermore, in lesions with fibrosis submucosal injection is followed by a non-lifting sign, which prevents resection by this technique [9]. In addition, there is concern about the risk of neoplastic cell implantation to deeper layers when injecting through the polyp [19], although this issue is debatable and still unproven.

In clinical practice, UEMR is not yet disseminated, particularly in Latin American countries, where the literature related to the topic is still scarce. Thus, this prospective study aimed to assess the feasibility, safety and efficacy of the UEMR technique for the resection of adenomas and colorectal serrated lesions in a Brazilian setting.

Patients and methods

UEMR procedure

Colonoscopies were performed by 2 interventional endoscopists with experience in EMR, each having performed more than 1000 EMRs. All patients were evaluated by high-definition colonoscopes (530, 590 and 600 series; Fujifilm Corporation, Saitama-Shi, Saitama, Japan). CO₂ insufflation was used for all procedures (GW-100; Fujifilm Corporation - Saitama-Shi, Saitama, Japan).

Target-lesion macroscopic features were studied with 0.4% indigo-carmine chromoendoscopy and digital chromoendoscopy with Fuji Intelligent Color Enhancement. The lesion’s size was estimated by visual comparison with the diameter of the opened snare before water instillation. Captivator^Ò snares (Boston Scientific Corporation, Marlborough, USA) were used in association with an electrocautery microprocessor (Endocut, 25W, effect 4, ERBE VIO 300S; ERBE Elektromedizim, Tübingen, Germany). Paris and Kudo pit pattern endoscopic classifications were used to characterize the lesions [20,21].

After macroscopic assessment of a lesion, intraluminal air was evacuated and approximately 500-1000 mL of sterile water at room temperature was infused until the lumen was completely filled. The resection started at the lesion’s distal margin, opening the snare and positioning it in order to include a rim of normal mucosa inside, and torque was applied to the colonoscope to maximize its capture. An en bloc tumor resection was always attempted, and when the lesion size was considered unsuitable for en bloc resection the tumor was resected in a piecemeal fashion. The mucosal defect and the resection margins were carefully examined post-procedure and any residual lesion was eradicated by either a snare or a biopsy forceps. Immediate bleedings were controlled with available endoscopic methods. Biopsies of the mucosal defect margins were obtained to evaluate free pathologic margins, in a number that depended on the size of the lesions. All collected material was properly stored in bottles with 10% formaldehyde, labeled and sent for histopathological analysis. The resection area was closed with endoscopic clips (Olympus Medical System Corporation, Tokyo, Japan), based on the Sydney classification [22] and at the discretion of the endoscopist. Fig. 1 is illustrative of the UEMR procedure.

Figure 1 Adenoma with laterally spreading component along the ascending colon (20 mm in size). (A) Before underwater endoscopic mucosal resection (UEMR). (B) Chromoendoscopy with indigo-carmine. (C, D) Piecemeal UEMR. (E) Mucosal defect after piecemeal UEMR. (F) Closure of the mucosal defect and hemostasis with endoscopic clips

Results

From January to July 2019, 39 patients with 54 colorectal lesions were referred to endoscopic resection. Three patients were excluded because of endoscopic signs of deep submucosal invasion (n=2) and refusal to sign the consent form (n=1). Thus, 36 patients (58.5% female) with 51 colorectal lesions underwent UEMR and entered the study (Fig. 2).

Figure 2 Enrolment of patients UEMR, underwater endoscopic mucosal resection

Table 1 presents the clinico-pathological characteristics of the population. The average age was 63 years, ranging from 33-86 years, and the mean/median size of the lesions was 16.24/13 mm (range: 7-60 mm). Regarding the location, 38 (74.5%) lesions were found in the right colon, 10 (19.6%) in the left colon, and 3 (5.8%) in the rectum. Histopathological evaluation revealed 43.1% tubular adenomas, 13.7% tubulovillous adenomas, 41.2% serrated lesions (of these 25.5% were hyperplastic polyps, 11.7% were sessile serrated polyp and 4% were traditional serrated adenomas), and 2% intramucosal adenocarcinoma.

Table 1 Characteristics of patients and endoscopic findings (N = 36 patients)

Complete resection was achieved in 44 of 51 lesions (86.3%), with 27 lesions removed en bloc (52.9%) and 24 resected in a piecemeal fashion (47.1%). When we considered only lesions ≥20 mm, we obtained an en bloc resection rate of 23.5%, which reached up to 47% for lesions ≥15 mm. During the study 27 lesions <15 mm were resected, 15 (55.5%) in the first half of the study, and 12 (44.4%) in the second half. All 15 lesions from the first half were removed en bloc. Of the 12 lesions removed during the second half of the study, 5 were resected in a piecemeal manner.

UEMR failed in 2 lesions (3.9%), requiring conversion to standard EMR. Endoscopic resection was accomplished successfully in 96.1% of cases. The mean/median resection time was 16.97/9.19 min, respectively (range: 2.57-105 min). Only 1 patient presented delayed bleeding (2%), which occurred on the 6^th day after resection and was treated with endoscopic clips. Minor intraoperative bleeding occurred in 5 patients (9.8%) and was also controlled endoscopically (4 with endoscopic clips and 1 using the snare tip associated with electrocautery soft-coagulation). Blood transfusions were not necessary in any of the cases. There was no intestinal perforation or post-polypectomy syndrome.

Of the 49 lesions treated exclusively by UEMR, 5 (10.2%) presented residual lesions disclosed on histology of a biopsy of the resection margins; therefore, a surveillance colonoscopy was proposed. Follow-up colonoscopy ranged from 3-6 months and was undertaken in only 2 cases, one presenting scar without residual lesion and the other presenting a polypoid lesion upon the scar that required rescue UEMR, with histopathology confirming a hyperplastic polyp.

Discussion

The role of standard EMR with submucosal injection is already well established for removal of colorectal lesions. UEMR without injection of solution into the submucosa has been used as an alternative method to standard EMR, and some series have demonstrated its safety and clinical effectiveness [6,9-18]. Water submersion allows a greater area of mucous surface to be captured inside the open snare, as there is less distension of the intestinal lumen as well as an apparent contraction and reduction in the diameter of the lesion upon the mucosa, and as a consequence offers greater potential for en bloc resection [24]. Furthermore, water submersion permits the following: a better visualization of margins via the optical effect of a natural “zoom” provided by the refraction of water [25]; reduction of wall damage induced by a diathermic snare secondary to the heat-sink effect of water [26]; better identification of specific bleeding points [27]; cost reduction, since it is not necessary to use a solution and needle for injection; and according to some authors, a reduction in the procedure time [13,18]. In addition, UEMR may be useful in challenging scenarios, such as recurrent lesions [15], and lesions difficult to access [28].

To our knowledge, this is one of the first studies so far to report the use of UEMR in Brazil, since Chaves et al demonstrated its feasibility for resection of sessile serrated adenomas [29] and for difficult colorectal polyps [30] in previous case series. Our data show that UEMR was endoscopically successful in 96.1% of cases, and on only 2 occasions was conversion to EMR using submucosal injection necessary. In the first case, the lesion was in the sigmoid colon and the water did not remain in the lumen even after the patient’s position was changed. This failure may have been related to the learning curve effect. In addition, Yamashina et al [31] reported that air remaining inside part of the lumen before water infusion can create a pressure gradient and push water away from the lesion; we think this may also have occurred in this particular case. The second case was a 35 mm lesion located in the hepatic flexure, over and behind a fold, which prevented good access for adequate snaring and ultimately required submucosal injection to improve the exposure. Siau et al [16] also reported that UEMR may not be feasible in special circumstances, such as recurrent polyps, depressed (0-IIc) lesions and larger lesions with suboptimal access.

In the present case series, we obtained a complete resection rate of 86.3% and an en bloc resection rate of 52.9%, accomplished in an average time of 16.97 min. These findings corroborate the results of a recent systematic review, which included 10 prospective and retrospective studies [32] and demonstrated a complete resection rate of 96.9% and en bloc resection rate of 57.1% using UEMR. In the only randomized controlled study comparing UEMR with EMR [31] for lesions of intermediate size (10-20 mm), with 108 lesions in the UEMR group and 102 lesions in the EMR group, the R0 resection rate was 69% for UEMR versus 50% for EMR, while en bloc resection was achieved in 89% with UEMR and 75% with EMR, showing that UEMR was superior as regards both en bloc and R0 resection rates for intermediate-sized lesions and hence establishing the clinical benefit of the technique in this subgroup.

It is important to note that lesion size affects the en bloc resection rate, especially for lesions over 20 mm [16]. Considering only lesions ≥20 mm, we obtained an en bloc resection rate of 23.5%, while for lesions ≥15 mm this value rose to 47%. We did not observe an improvement in the en bloc resection rate for lesions <15 mm in diameter in the second half of the study as expected, perhaps because the endoscopists were already experienced and had already passed the learning curve. Some authors advocate that UEMR has a greater potential for en bloc resection as compared to EMR, particularly due to the “effect of tumor size reduction” induced by water immersion [15,24,31], and that could lead to lower rates of local recurrence, which typically range from 14-55% after piecemeal EMR [15]. In the review by Spadaccini et al [32], a recurrence rate of 8.8% was identified using UEMR, while 13.8% was reported by Hassan et al [33] using conventional techniques, although the median polyp size in those 2 studies were 33 mm and 23 mm respectively, which may affect this analysis. In our study, we had 5 patients (10.2%) with residual lesions and only 2 of them underwent a surveillance colonoscopy; thus, we do not have sufficient data to adequately analyze the recurrence or residual lesion rate. Control examination is still one of the best parameters to say whether the resection was complete or not, and to establish the presence of recurrence and/or residual lesions.

We can also highlight a good security profile using UEMR. Our findings show that bleeding occurred in 11.8% of cases. It is noteworthy that, of these, 5 (9.8%) were minor and immediately controlled during the procedure; they should therefore not be viewed as an adverse event. We observed a similar rate of delayed bleeding (2%) to Spadaccini et al (2.8%) [32]. There were no cases of intestinal perforation in the study, a feared complication that has seldom been reported so far [17,18,34]. In contrast, some studies involving EMR describe bleeding values ranging from 2-11% and perforation rates around 1.5% [33,35-39]. Sánchez et al [40] believe that the safety profile offered by UEMR can be explained by the low probability of deep damage to the intestinal wall, due to the heat-sink effect of water. We particularly think that in the absence of the edema effect caused by injected solution, the post-resection mucosal defect is smaller and less deep using UEMR, which can facilitate the placement of endoscopic clips when needed.

This study had some limitations that should be highlighted. First, it was performed in a single tertiary center by 2 endoscopists experienced in EMR. Therefore, it is not known whether it can be reproduced in other institutions, despite the fact that the endoscopists who performed the procedures had not received any specific training in UEMR before the beginning of the study and just followed the technical details described in the literature. Secondly, the study sample was relatively small. Finally, patient follow up was inadequate, and thus long-term results such as recurrence rate could not be precisely assessed.

In conclusion, this study indicated that UEMR for the removal of adenomas and colorectal serrated lesions is feasible, safe and effective in clinical practice in Brazil. The technique may be included in the therapeutic arsenal of institutions that manage colorectal lesions, although further well-designed randomized controlled trials must be encouraged to support its generalized implementation, as well as to assess differences in lesion recurrence compared to EMR.