Colorectal endoscopic mucosal resection with submucosal injection of epinephrine versus hypertonic saline in patients taking antithrombotic agents: propensity-score-matching analysis

This retrospective chart review included 204 patients taking antithrombotic agents among 995 consecutive patients who underwent colonic EMR at the National Hospital Organization Ureshino Medical Center from April 2012 to March 2018. Patients >20 years of age who fulfilled the following criteria were candidates for the study: i) polyp diameter < 20 mm; ii) use of antithrombotic agents, including antiplatelet agents (low-dose aspirin, clopidogrel, ticlopidine, and cilostazol) and anticoagulants (warfarin and direct oral anticoagulants); and iii) normal coagulogram (platelet count: 140,000–379,000/μL, prothrombin time international ratio: 1.5–2.6). Patients with polyp diameter > 20 mm, abnormal coagulogram, and/or impaired normal blood clotting were excluded. Informed consent for the procedures was obtained from all patients who underwent colorectal EMR. The present study was conducted according to the Ethical Guidelines for Medical and Health Research Involving Human Subjects. The study protocol and the consent procedure were approved by the Ethics Review Committee of the National Hospital Organization Ureshino Medical Center (approval number 18–17).

During the study period, 10 endoscopists with more than 3 years of experience in gastrointestinal colonoscopy performed colorectal EMR procedures using submucosal injection of either saline–epinephrine (0.01%) or hypertonic saline (10% NaCl), according to the judgment of the endoscopist. Patients were divided into two groups: the saline–epinephrine group (Group A) and the hypertonic saline group (Group B) (Fig. 1). The EMR procedure was performed with a colonoscope (PCF-Q260AZI; Olympus, Tokyo, Japan), snare (SnareMaster; Olympus), and electrosurgical unit (VIO 300, ERBE; Elektronedizin, Tübingen, Germany), with appropriate use of butylscopolamine or glucagon. Sedation was not used, except in patients with procedure-related pain or sedation request; in these patients, diazepam (5–10 mg), midazolam (0.05–0.1 mg/kg), pentazocine (15 mg), or pethidine hydrochloride (35 mg) was administered with monitoring of cardiorespiratory function [40].

After injection of saline–epinephrine (Fig. 2) or hypertonic saline (Fig. 3) into the submucosa with needle forceps to create an adequate mucosal bulge, the colorectal polyp was snared and resected [17, 22)]. Prophylactic clipping was performed in all patients. When bleeding was clinically suspected after EMR, emergency colonoscopy was performed to achieve hemostasis with hemoclips and/or electrocoagulation [19)].

Information on antithrombotic agents was recorded, including type, number, and management (continuation, cessation, or replacement) and comorbidity including Charlson comorbidity score was detected. Information on polyp lesions (size, location, and histological and macroscopic classifications), the endoscopist (specialist or trainee), and treatment outcomes (procedure time and en bloc resection rate) were reviewed after EMR. Specialist endoscopists were defined as those who had performed more than 40 EMR procedures over a period of at least 3 years after attaining the required fundamental skills and knowledge [41, 42)]. Immediate EMR bleeding was defined as hemorrhage during the procedure; post-EMR bleeding was defined as bleeding that occurred at least 1 h after the procedure. Adverse events of immediate EMR bleeding, post-EMR bleeding, perforation, and mortality within 30 days were recorded. Results are expressed as means ± standard deviations (SD).

The χ² test was used to identify differences in the effectiveness rate between the two groups. Student’s t test was used for unpaired data to determine differences in means between the two groups. As indicated in Fig. 1, the two groups were compared after propensity-score matching. Propensity-score-matching analysis was used to control factors that might influence EMR treatment outcomes and adverse events. The two groups were matched in a 1:1 ratio (Group A, n = 80; Group B, n = 80) with propensity-score matching adjusted for five covariates (age, sex, anticoagulant agents, antiplatelet agents, and endoscopist) to minimize inherent bias (Table 5). This model yielded a c statistic of 0.67, indicating the ability to differentiate between Groups A and B. The caliper width of propensity-score matching was 0.05. Differences were considered statistically significant at P < 0.05. Univariate and multivariate logistic regression was performed to assess risk factors for immediate and post-EMR bleeding, with the explanatory variables of age, sex, anticoagulant agents, antiplatelet agents, multiple agents, heparin bridge therapy, polyp size, number of polyps, histological classification, endoscopist, and type of injection. All statistical analyses were performed with JMP version 13.0.0 (SAS Institute, Tokyo, Japan).