Endoscopic Mucosal Resection for Colonic Mucosal Neoplasia and Evaluation of Long-Term Recurrence

A Single-Center Experience of 500 Cases

Saeed Ali, MD; Neelam Khetpal, MD; Evgeny Idrisov, MD; Asad Ur Rahman, MD; Sameen Khalid, MD; Yuan Du, MS; Udayakumar Navaneethan, MD; Shyam Varadarajulu, MD; Robert Hawes, MD; Muhammad Khalid Hasan, MD


South Med J. 2021;114(4):199-206. 

In This Article


Study Design and Patients

We performed a longitudinal retrospective study at our advanced endoscopy unit at AdventHealth Orlando Hospital from January 1, 2013 to April 26, 2017. A total of 480 patients with 500 polyps (LSLs) were included. All of the LSLs were previously identified on colonoscopy by the patients' primary gastroenterologists and were referred to our center for endoscopic resection. Inclusion criterion was based on age 18 years and older and polyp size ≥20 mm. The study was approved by the hospital's institutional review board.

Lesion Classification

Lesions were carefully assessed using white light and narrow-band imaging (NBI) endoscopy and were classified using three criteria: the Paris classification,[28] surface morphology (granular, nongranular, or mixed),[29] and Kudo pit patterns I to V.[30] NBI (Olympus America, Center Valley, PA) and high-definition imaging were used to define pit pattern classification. The lesion size was measured using snares of known dimensions.

Procedural Techniques

Colonoscopy with EMR was performed by one of four advanced endoscopists. The split-dose bowel regimen was used. Monitored anesthesia care was the preferred mode of anesthesia. Colonic insufflation was achieved using carbon dioxide. Olympus colonoscopes CF-HQ190 or PCF-H190 were used according to endoscopist preference.

A mixture of injection solution containing Voluven solution (6% hydroxyethyl starch 130/0.4 in 0.9% sodium chloride) dyed with methylene blue or indigo carmine was injected into the submucosa beneath the lesion for the lift before resection. Epinephrine 1:100,000 was added to the submucosal injection fluid. All of the components of the solution were mixed before the injection. Carr-Locke needles (US Endoscopy, Mentor, OH) were used for submucosal injection. Lesions were resected using the sequential inject and resect EMR technique (Figure 1).[11,31,32] Resection included a small margin of normal tissue (1–3 mm) around the lesion. Snares were used according to endoscopist preference (spiral SnareMaster [Olympus America], oval AcuSnare [Cook Medical, Bloomington, IN], Captivator snare [Boston Scientific, Marlborough, MA], and Histolock snare [US Endoscopy]). The cautery settings for snare resection were endocut 2-1-4 or coagulation with an effect of 2 and 18 W (VIO 300, ERBE, Tübingen, Germany). Complete snare excision was the goal in all of the patients. If this was not possible, then cold forceps avulsion followed by snare tip soft coagulation to the avulsion bed or argon plasma coagulation was used (soft coagulation for ablation was used at the setting of effect of 4 and 80 W). Normal-appearing resection margins were not routinely ablated. Visible blood vessels not actively bleeding were not ablated prophylactically (Figure 2A and 2B; Supplemental Digital Content Video 1, http://links.lww.com/SMJ/A216; sequential steps of inspection using white light and NBI, followed by injection and resection of colon polyp; following resection, the area is carefully examined to confirm the completeness of polyp resection]), and the resection site was not routinely clipped. After the procedure, the tissue was retrieved and sent for histopathology.

Figure 1.

Sequential steps of inspect, inject, resect, and inspect. A, Polyp in cecum as examined with NBI; B, after submucosal injection; C, after resection of polyp, margins showing no retained abnormal tissue; and D, after complete resection of polyp. NBI, narrow-band imaging.

Figure 2.

A, Polyp in the ascending colon with white light examination; B, postresection of the polyp; C and D, SC1 examination showing a typical healthy scar with white light and NBI without recurrence (puckering of normal folds into the central scarred area with normal overlying well-healed mucosa). NBI, narrow-band imaging.

Patients were observed for 2 to 3 hours postprocedure for any immediate complications. In the absence of any obvious complications, patients were discharged on a clear liquid diet for 24 hours. Preprocedural anticoagulation and antiplatelet therapy were managed according to American Society for Gastrointestinal Endoscopy guidelines.[33]

Patients with successful endoscopic resection of their LSLs, with no cancer on histopathology, subsequently underwent SC at 4 to 6 months (SC1) and 16 to 18 months (SC2). If recurrence was noted at SC1, then subsequent SC2 was performed at 6 months from SC1.

During SC, the EMR scar was evaluated with white light and NBI. A photograph of the scar was obtained (Figures 2C and 2D). A normal-appearing post-EMR scar was not routinely biopsied. If RRA was suspected, then a biopsy was obtained to determine the presence or absence of histologically determined recurrence, and the abnormal tissue was treated endoscopically. Treatment included resection by a snare or avulsion by cold biopsy forceps, followed by snare tip soft coagulation. Dedicated expert gastrointestinal histopathologists reviewed all of the pathology specimens.


Patients were contacted via telephone within 1 week after endoscopy to document complications. Patients who had cancer on histology on initial EMR were referred for surgery.


Early recurrence was defined as the presence of RRA at SC1 with histological confirmation as an adenoma. Late recurrence was defined as presence of RRA at the EMR site at SC2. Postprocedural complications included delayed bleeding, immediate perforation, delayed perforation, and abdominal pain. Delayed bleeding was defined as bleeding resulting in emergency department visit, hospitalization, or repeat colonoscopy. Immediate perforation was defined as a full-thickness defect in the colonic wall. Delayed perforation was defined as patients presenting to the hospital with signs and/or symptoms suggestive of peritonitis.

Statistical Analysis

The results for continuous variables were summarized by mean, median, and range. Frequencies and percentages were used to summarize categorical variables. Student t or Mann-Whitney tests were used to compare the distribution of continuous variables by outcome. Pearson's or Monte Carlo estimate χ 2 or Fisher exact tests were used for the association between categorical variables and outcome. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to quantify the level of association. Two-tailed tests with a significance level of 5% were used throughout. Univariate analysis was only done for patients with follow-up at SC1.

Candidate variables for inclusion in a multiple logistic regression model were any variable with P ≤ 0.1 on univariate analysis where there was no category with "0" as the data point. As such, lesion location and histology were not included in multiple logistic regression models, although both had P ≤ 0.1. The continuous variable of lesion size was grouped into categories. Statistical analysis was performed using SAS version 9.4 (SAS Institute, Cary, NC).