From January 2009 to December 2015, patients who underwent ESD for gastric neoplasms at Chung-Ang University Hospital (Seoul, Korea) were initially enrolled. Gastric adenoma and carcinoma, especially EGC, were categorized as gastric neoplasms in this study. Gastric neoplasm was confirmed with forceps biopsy under esophagogastroduodenoscopy (EGD) guidance (GIF-H260 or GIF-Q260; Olympus, Tokyo, Japan). During EGD, if dysplasia or carcinoma was suspected, two to four biopsy samples were obtained with forceps (FB-25 K-1 [Olympus] or BX 420 [PriMed Instruments Inc., Ontario, Canada]). ESD was indicated when the following criteria were met: any lesions with low- to high-grade dysplasia, regardless of size, or well- to moderately differentiated adenocarcinoma confined to the mucosa without evidence of lymph node or distant metastases on abdominal computed tomography or endoscopic ultrasonography. However, in some patients, the pathological results after ESD were revealed to be outside of the criteria. In such cases, whether to perform additional surgery or ESD was decided according to the physician’s discretion and the patient’s wish. The following patients were excluded from the study: those who had no endoscopy at the 3-month follow-up visit and those who underwent surgical resection after endoscopic treatment owing to adverse events or lymphovascular or margin invasion in the pathological specimen. The medical records of patients were reviewed with regard to age, sex, medication, Helicobacter pylori (H. pylori) infection, and endoscopic characteristics including tumor size, location, macroscopic type, and gross morphology. The study protocol was approved by the Institutional Review Board (IRB) of Chung-Ang University Hospital, Korea (IRB no. C2015155).

Each endoscopic finding and report at 3 months follow-up were reviewed to determine the size, location, macroscopic type, and gross morphology of the lesion. The location of lesions was defined according to seven sections of the stomach: antrum, angle, lower body, mid body, high body, fundus, and cardia. The location of the lesion was classified following the Japanese Classification of Gastric Carcinoma [12]. According to the Paris endoscopic classification, the macroscopic type of the lesion was divided into the superficial elevated, flat, or depressed types [13]. In addition, gross morphology including erythema, ulcer, erosion, or nodularity, H. pylori infection, intestinal metaplasia, and chronic atrophic gastritis were also examined.

In the present study, the key endoscopic findings at the resection scar were defined on the basis of gross morphology, as follows: irregular erythema, mucosal defect (erosion or ulcer), and nodularity (Fig. 1). We defined irregular erythema at the resection scar as an irregular reddish mucosal change, and we also included findings of engorged capillary vessels at the scar into irregular erythema. Ulcer was defined as a ≥5-mm mucosal break with a well-defined crater, whereas erosion was defined as a superficial mucosal defect without an ulcer crater [14]. Nodularity was defined as the presence of an irregularly raised or nodular surface without a dominant mass [15]. In this study, recurrent tumor at the resection site after 12 months was defined as local recurrence.

All lesions were removed through ESD by three experienced gastrointestinal endoscopists (B.J.K., J.G.K., and C.H.C) by using a single-channel endoscope (GIF-Q260J, GIF-H260Z, or GIF-H290; Olympus). The patients were sedated by using a course of intravenous midazolam (0.05 mg/kg) or propofol and remifentanil, with a target-controlled infusion system under bispectral index (BIS) monitoring by an experienced anesthesiologist. The initial target effect-site concentration of propofol and remifentanil was 1.5 μg/mL and 1.5 ng/mL, respectively. The plasma concentration was increased or decreased by 0.2 μg/mL or 0.2 ng/mL, targeting a 60~80 BIS score.

In all patients, dot markings for ESD were first placed 5 mm beyond the tumor margins with argon plasma coagulation. A saline solution (100 mL 0.9% saline mixed with 1 mg epinephrine and a small amount of indigo carmine) was then injected into the submucosal layer, and the submucosal layer around the lesion was dissected by using a needle knife (Nadel-Papillotom, 99020121; MTW Endoskopie, Wesel, Germany), insulated-tip knife (KD-611L; Olympus), and/or flex knife (KD-630L; Olympus). If necessary, saline injection was repeated and endoscopic hemostasis was achieved during the procedure. A high-frequency electrosurgical current generator (VIO 300D; ErbeElektromedizin, Tübingen, Germany) was used during the processes of marking, mucosal incision, submucosal dissection, and hemostasis. A complete resection was defined as a resection in which the resected tumor had tumor-free horizontal and vertical margins. Complete en bloc resection was defined as en bloc resection achieved endoscopically along with histological confirmation of tumor-free margins.

After ESD, all patients received 40 mg pantoprazole intravenously or took 30 mg lansoprazole orally once daily on the day they underwent ESD and throughout the length of their hospital stay, which was usually 4~5 days. If no ESD-related adverse events occurred, the patients were allowed a liquid diet on the first or second day after ESD. The patients were discharged the day after starting diet. After discharge, the patients were instructed to take 30 mg lansoprazole once a day for 4 weeks. The patients were examined with conventional endoscopy at 3, 6, and 12 months after ER and annually thereafter.

At our center, pathologic diagnosis was made on the basis of the third edition of the Japanese Classification of Gastric Carcinoma [11]. The pathologic type, depth of invasion, lymphatic and venous involvement, and the presence of tumor at the resection margin (horizontal and vertical) were evaluated.

Statistical analysis was performed with SPSS ver. 21.0 (Statistical Package for Social Sciences; SPSS Inc., Chicago, IL, USA). For univariate analysis, categorical variables were analyzed by using a chi-square test or Fisher exact test. Those variables with P<0.20 in the univariate analyses were examined in multivariate analysis by using binary logistic regression models. A P value of <0.05 was considered significant.

The baseline clinicopathologic characteristics of the study population are summarized in Table 1. The mean age of the patients was 64.4 years, and 66.3% of them were men. Comorbidities including hypertension (n=106), diabetes mellitus (n=46), cardiovascular disease (n=19), chronic kidney disease (n=11), and liver cirrhosis (n=5) were found. Nine patients had a history of peptic ulcer. Eighty-three patients were current smokers, whereas 124 patients were alcohol drinkers. The number of patients who were taking aspirin was 37, and 15 patients were taking an antiplatelet agent other than aspirin, seven were taking nonsteroidal anti-inflammatory drugs, and one was taking an anticoagulant and steroid.

Among the total lesions resected by using ESD, 201 were adenomas and 73 were EGCs. Almost lesions were located at the antrum or angle, and the macroscopic shape of most lesions was superficially elevated. The mean size of the tumor was 14.6 mm, and the mean area of the tumor was 212.4 mm². The H. pylori infection was observed in 82 patients.

Therapeutic outcomes are summarized in Table 2. En bloc resection rate was 98.9% (271/274). Complete resection rate and curative resection rate were 97.0% (266/274) and 97.0% (266/274), respectively. With regard to the gross morphology of the scar, 101 lesions (36.9%) were classified as irregular erythema, 85 (31.0%) as nodular lesions, and 104 (38.0%) as unhealed mucosal defects. The mean follow-up duration was 24.5 months. There were four recurrence cases (Table 3). Concerning additional treatment methods, two patients with recurrence underwent surgical resection, whereas one underwent argon plasma coagulation treatment. The remaining one patient had low-grade dysplasia and refused to undergo further treatment, although endoscopic follow-up had been done.

Atypical scar patterns were defined as irregular erythema, nodularity, or mucosal defect. In univariate analysis for the irregular erythema scar pattern, the following factors showed marginal significance (P<0.100): male sex and specimen size >40 mm (Table 4). Likewise, the nodularity scar pattern was related to male sex, age younger than 65 years, alcohol drinking, smoking, and a procedural duration >1 hour. Mucosal defect was related to alcohol drinking, smoking (inverse relationship), history of peptic ulcer disease, infra-angle location (angle and antrum), cancer, lesion size >20 mm, and specimen size >40 mm. The irregular erythema with nodularity scar pattern was related to male sex, smoking, infra-angle location, and procedural duration >1 hour. Furthermore, the irregular erythema with nodularity and mucosal defect scar pattern was related to male sex, liver disease, chronic kidney disease, and hemostasis at second-look EGD in univariate analysis. These related factors were examined with multivariate analysis. Table 5 showed that male sex was an independent risk factor of the irregular erythema scar pattern (OR, 2.25; 95% CI, 1.15-4.39; P=0.018) and smoking was an independent risk factor of the nodularity scar pattern (OR, 2.51; 95% CI, 1.30-4.85; P=0.006). In addition, infra-angle location (OR, 8.20; 95% CI, 1.74-38.4; P=0.008) and cancer were related to the mucosal defect scar pattern as independent risk factors (OR, 2.81; 95% CI, 1.54-5.14; P=0.001). Male sex was demonstrated as an independent risk factor of the irregular erythema with nodularity scar pattern (OR, 3.46; 95% CI, 1.13-10.64; P=0.030). Liver disease and chronic kidney disease were demonstrated as independent risk factors of the irregular erythema with nodularity and mucosal defect scar pattern (OR, 16.99; 95% CI, 1.72-167.88; P=0.015 and OR, 12.71; 95% CI, 1.79-90.27; P=0.011, respectively).