Andrew Nett, MD and Kenneth F. Binmoeller, MD
Conventional endoscopic mucosal resection (EMR) involves the use of submucosal fluid injection to lift a lesion away from the muscularis propria with the intention of making resection safer and easier. This practice has disseminated despite the lack of studies proving the clinical benefit of submucosal injection prior to EMR. Submucosal injection prior to EMR can at times make EMR more difficult and riskier if the injection is misguided. A limitation of conventional EMR is a high recurrence rate, particularly with larger lesions requiring piecemeal resection. Conventional EMR may also be technically difficult if scarring is present or if lesions are located in difficult locations such as at the ileocecal valve, at the appendiceal orifice (AO), or across colonic folds. Underwater EMR (UEMR) was developed and reported by Binmoeller et al1 in 2012 as an alternative resection method that eliminates the need for submucosal injection prior to EMR. The method is based upon the observation that complete filling of the gastrointestinal (GI) lumen with water causes the submucosa and mucosa to float away from the wall’s muscular layer and involute into the lumen, thereby obviating the need for submucosal injection. UEMR has been found to be easy to learn by endoscopists experienced in conventional EMR techniques. Large series have shown UEMR to be associated with a higher complete macroscopic resection rate, lower recurrence rate, and reduced need for repeat procedures to achieve curative resection with no difference in adverse event rates compared to historical controls that underwent conventional EMR. Compared to endoscopic submucosal dissection (ESD), UEMR may achieve comparable complete resection rates without recurrence while resulting in less complications and consuming substantially less procedure time. If the proposed advantages of UEMR are confirmed in controlled trials, this technique could be considered as a potential, and possibly superior, alternative to conventional EMR.
EMR has provided an alternative to surgical resection for the treatment of large and/or sessile mucosal neoplasia. Conventionally, submucosal injection of saline or other fluids is performed prior to resection based on the rationale that fluid accumulation within the submucosa acts to separate superficial lesions from the underlying muscular layer, protecting against perforation through the colonic wall when snare resection is performed. Fluid injection is also thought to lift sessile or flat lesions, thereby facilitating entrapment within a snare. The submucosal cushion that is created is also thought to limit inadvertent transmural thermal injury by increasing the distance over which heat can dissipate between the electrocautery probe and extramural tissue.2,3 Another purported benefit of submucosal injection is the potential identification of the presence of deeply invading neoplastic tissue, which will prevent tissue lifting despite injection (“nonlifting sign,” NLS); although the absence of lift may also occur if submucosal scarring is present.
Submucosal injection prior to EMR has been widely embraced as the standard of care, despite the lack of studies proving its clinical benefit. Noteworthy is a remarkably high incidence of residual or recurrent neoplasia reported after conventional EMR, ranging from 15% to 55%.3–6 While recurrent lesions may often be diminutive when surveillance colonoscopy is performed at a short interval following initial resection, their presence necessitates repeat procedures or eventual surgical intervention. The primary predictors of adenomatous lesion recurrence are the performance of piecemeal resection and large polyp lesion size (which often necessitates piecemeal resection). Piecemeal resection is typically required for lesions > 2 cm because the ability to perform en bloc resection on larger lesions is limited by snare size and the risk of capturing the muscle layer during snaring.7,8 Use of argon plasma coagulation (APC) during initial resection for the purpose of neoplasia ablation and the occurrence of intraprocedural bleeding have also been associated with recurrent/residual adenoma on multivariate analysis.8
ESD is an alternative therapeutic modality for treatment of large or flat GI neoplasia. Per the Japanese guidelines, ESD is indicated for resection of lesions > 2 cm with Paris IIc or IIa + IIc morphology. Laterally spreading tumors (LSTs), nongranular type, are also an indication for ESD given that they have a high rate of multifocal submucosal invasion.9 Deeper invasion will limit lift from submucosal injection, frequently necessitating piecemeal resection, which in turn complicates histological assessment of invasion depth and the adequacy of the resection margins.3 The Japanese guidelines also suggest that any lesion > 3 cm in size should be considered for ESD.9 In the aforementioned scenarios, guideline advisement for ESD is based on the benefits of en bloc removal. While ESD may facilitate enhanced en bloc and complete resection vs EMR, it also can be a cumbersome, lengthy, and expensive process with increased risk of complication. Thus, both conventional EMR and ESD are imperfect as resection methodologies.
Underwater Endoscopic Mucosal Resection
Colonoscopy using the water submersion, or underwater technique has been shown to enhance the diagnostic yield of colonoscopy. Multiple randomized controlled trials comparing the underwater exchange and conventional gas-insufflation techniques have demonstrated improved patient comfort, decreased sedation needs, and enhanced adenoma detection rates.10–16 Visualization underwater has a magnification and focus effect that enhances lesion resolution, helping to demarcate lesion margins with or without use of Narrow Band Imaging (Olympus).17 The effect of water also causes mucosal lesions to float into view, potentially revealing those that may be concealed behind folds during gas insufflation.
This floating effect can be clearly visualized under endoscopic ultrasound (EUS). The endosonographic observation that the mucosa and submucosa float away from the muscularis layer when the colon is filled with water inspired the UEMR technique.1 With water submersion, the mucosa and submucosa will form multiple involutions resembling rugal folds of the stomach while the deeper muscularis layer remains circular (Figure 15-1). By removing intraluminal air, colonic wall tension is decreased and the wall reassumes its native thickness. Once underwater, tissue buoyancy—due to the fat density of the submucosa and its antigravity effect—will lift overlying adenomatous mucosal lesions away from the muscularis propria, thereby eliminating the need for submucosal injection. Natural wall layer separation reduces the chance of unintentional snare entrapment of the muscularis. Intraluminal water also serves as a heat sink, protecting the deeper colonic wall from thermal injury. In combination, layer separation and thermal dissipation are thought to decrease the risk of early and delayed perforation as well as postpolypectomy electrocautery syndrome.18 By not performing submucosal injection, one can also avoid potential, albeit rare, complications due to inadvertent fluid injection through the entire thickness of the colonic wall. In the case of tattoo injection, perforation, peritonitis, abscess formation, unintentional marking of the small intestine, adhesion ileus, and development of inflammatory pseudotumor have been reported.19 Some of these complications, however, may be induced by an inflammatory response to the tattoo ink; risks may not exist for other types of submucosal fluid injection. Skipping unnecessary submucosal injection does reduce procedure time.
Unacceptably high reported recurrent rates following conventional EMR beg the question of other potential detriments to submucosal injection. While recurrence may be due to inadvertent incomplete resection, the act of submucosal injection itself may potentially be a root cause of recurrence. Needle-tract seeding to the GI wall during EUS-guided fine-needle aspiration (FNA) has been reported among various tumor types.20 A systematic review found a 2.7% overall incidence of needle-tract tumor seeding following hepatocellular carcinoma biopsy.21 Transperitoneal biopsy of hilar cholangiocarcinoma has also been associated with increased risk of peritoneal metastases.22 In cases of hilar cholangiocarcinoma, concern over tumor seeding following FNA can preclude patients from transplant listing if EUS-FNA has been performed. Why then, should one ignore that injection may spread neoplastic cells along the submucosa and into deeper tissue layers of the colonic wall despite the known phenomenon of needle-tract seeding? During EMR, avoiding needless submucosal injection eliminates this risk.
Binmoeller et al1 published the first series establishing the novel UEMR technique in 2012. Sixty-two large sessile benign colorectal lesions (>2 cm in size, mean size 3.4 cm) were removed using UEMR in 60 consecutive patients over 11 months. Surveillance colonoscopy was performed 3 months thereafter and biopsies were taken of the resection scar site as well as any tissue suspicious for residual or recurrent adenoma. Clip closure was feasible and performed in 27% of resections. Among the 90% of patients who adhered to follow-up colonoscopy, possible recurrent adenoma was seen in 1.9% though this consisted of a 5-mm adenomatous nodule along a tethered fold adjacent to the prior resection site, which may have represented a missed residual satellite lesion rather than true recurrence. Delayed bleeding occurred in 5% of patients following resection. No perforations occurred.
Multiple subsequent prospective series and one cohort study describe successful and efficacious application of the UEMR technique for colonic lesions. In 2014, Wang and colleagues5 published a series of 43 lesions averaging 20 mm in size in 21 patients. Complete resection was successful on the first attempt in 98% of lesions, though this included necessary use of hot-biopsy forceps stripping of islands of residual lesion in 11.6%, corresponding with larger lesion size (P = .001). Whereas Binmoeller et al1 used no thermal ablation for therapy, this study relied on APC application for residual neoplastic islands in 16.3% of cases. Incomplete resection occurred in one case, resulting from deliberate abandonment of endoscopic resection after noticing involvement of the AO. Delayed bleeding followed simultaneous resection of a 6 cm tubulovillous and 1 cm cecal tubular adenoma. There were no incidences of postpolypectomy syndrome or perforation. Uedo and colleagues23 published a subsequent prospective series showing good results in 11 consecutive patients confirming technique efficacy and safety for large lesions. A larger prospective series described a total of 81 polyps in 72 consecutive patients removed by UEMR with a 100% complete resection rate and 0% recurrence rate at follow-up colonoscopy at 3 months.24 En bloc resection was achieved in 68% of cases, albeit on relatively small lesions (42% of polyps were 10 to 15 mm in size, 58% were 15 mm or larger). These favorable outcomes can be maintained, however, while treating larger lesions.
When submersed underwater, the colon mucosa and submucosa retain a contracted and involuted state. By folding into the lumen when underwater, large colonic lesions, while having unchanged total surface area and length, will occupy a smaller plot at the colon wall compared to their flattened, more spread-out state, under air insufflation. Despite having a fixed maximum size, a resection snare will thus be able to capture underwater lesions much larger than the snare opening. In a prospective series examining larger lesions, 50 patients with 53 LSTs with a median size of 30 mm (20 to 40 mm range) were treated with UEMR.25 En bloc resection was possible in a high number of patients, achieved in 55% using a 33 mm snare. The delayed bleeding rate remained low at 4% and no perforations were observed. Residual adenoma was found at follow-up colonoscopy with resection site biopsy in 5% of cases (median follow-up at 31 weeks, range 7 to 71 weeks).
Overall, > 300 cases of UEMR have been reported (Table 15-1). In these prospective series and cohort publications, delayed bleeding rates have occurred at 0.5% to 6.7% with 1 case requiring blood transfusions and 2 cases reporting pursuit of colonoscopy for hemostatic intervention.2,3 Perforations have not occurred in these prospective series, though a case report by Ponugoti and Rex26 describes an incident of perforation during UEMR performed in a retroflex position. The authors speculate that retroflexion reduced the protective water immersion effect of reduced wall stretching by causing stress on the colonic wall and consequent wall thinning.
Underwater Endoscopic Mucosal Resection Technique
UEMR is typically performed with an adult, high-definition, single-channel colonoscope with an auxiliary water jet (CF-H180AL; Olympus). Use of the adult colonoscope over a pediatric colonoscope carries the advantage of near focus capability, sharpening underwater images. A cap is mounted on the colonoscope tip (Model D-201-15004; Olympus), which aids in fold flattening and scope tip stabilization. Glucagon may be administered to reduce peristalsis when necessary, but is often deliberately avoided because colon wall contractions allow large lesions to prolapse into the resection snare. We routinely perform water-exchange intubation, but if the lesion is reached and identified under gas insufflation, the first step is removal of gas from the colonic segment of interest followed by infusion of room temperature sterile water to achieve complete submersion of the lesion (usually between 100 and 300 mL) using a flushing pump. High-definition white-light and narrow-band imaging examination of the lesion pit pattern is performed underwater to evaluate the pit and vascular pattern. If stigmata of malignancy are present (eg, Paris type IIc morphology, Kudo Type V pit pattern), EUS is performed to evaluate for an invasive malignancy. If the lesion is an appropriate candidate for resection, we make diathermic markings within 1 to 2 mm of the perimeter of the lesion margins under high-definition imaging. We prefer to use APC by applying the 7 Fr probe tip directly to the mucosal surface and delivering a short burst of coagulation flow (0.8 L/minute, 30 W), which produces a round coagulum dot that is easily recognized during resection (Figure 15-2). We routinely mark the perimeter because snare resection creates cautery artifact that may be difficult to differentiate from adenomatous change.