Percutaneous endoscopic gastrostomy (PEG) is a relatively safe procedure; however, acute and chronic complications of PEG have been reported. We aimed to determine risk factors associated with complications and 30-day mortality after PEG, based on 11 years of experience at a single tertiary hospital. In total, 401 patients who underwent first PEG insertion at the Asan Medical Center, Seoul, Korea, between January 2005 and December 2015 were eligible. Medical records were retrospectively reviewed to determine clinical characteristics and outcomes of 139 and 262 patients who underwent pull-type and introducer-type PEG, respectively. The median age of the overall population was 68 years, and the median body mass index was 19.5 kg/m2. Acute and chronic complications developed in 96 (23.9%) and 105 (26.2%) patients. Acute ileus and chronic tube obstruction were significantly more frequent in the introducer-type PEG group (p = 0.033 and 0.001, respectively). The 30-day mortality rate was 5.0% (median survival: 10.5 days). Multivariate analysis revealed that underlying malignancy was a predictor of acute complications; age ≥ 70 years and diabetes mellitus were predictors of chronic complications. The median follow-up was 354 days. Neurologic disease and malignancy were the most common indications for PEG. Neurologic diseases were classified into two groups: stroke and the other neurologic disease group (including dementia, Parkinson’s disease, neuromuscular disease, and hypoxic brain damage). Multivariate analysis showed that 30-day mortality was significantly lower in the other neurologic disease group and higher in patients with platelet count < 100,000/μL, and C-reactive protein (CRP) ≥ 5 mg/dL. PEG is a relatively safe and feasible procedure, but it was associated with significantly higher early mortality rate in patients with platelet count < 100,000/μL or CPR≥5mg/dL, and lower early mortality rate in neurologic disease group including dementia, Parkinson's disase, neuromuscular disease, and hypoxic brain damage. In addition, acute complications in patients with underlying malignancy, and chronic complications in patients aged ≥70 and those with diabetes mellitus should be considered during and after PEG.

Proton pump inhibitor (PPI)-based triple therapy consisting of PPI, amoxicillin, and clarithromycin, is the recommended first-line treatment for Helicobacter pylori infection. However, the eradication rate of triple therapy has declined over the past few decades. We analyzed the eradication rate and adverse events of triple therapy to evaluate current practices in Korea. A comprehensive literature search was performed up to August 2013 of 104 relevant studies comprising 42,124 patients. The overall eradication rate was 74.6% (95% confidence interval [CI], 72.1%-77.2%) by intention-to-treat analysis and 82.0% (95% CI, 80.8%-83.2%) by per-protocol analysis. The eradication rate decreased significantly from 1998 to 2013 (P < 0.001 for both intention-to-treat and per-protocol analyses). Adverse events were reported in 41 studies with 8,018 subjects with an overall incidence rate of 20.4% (95% CI, 19.6%-21.3%). The available data suggest that the effectiveness of standard triple therapy for H. pylori eradication has decreased to an unacceptable level. A novel therapeutic strategy is warranted to improve the effectiveness of first-line treatment for H. pylori infection in Korea.

Recently, significant improvements have been made in artificial intelligence. The artificial neural network was introduced in the 1950s. However, because of the low computing power and insufficient datasets available at that time, artificial neural networks suffered from overfitting and vanishing gradient problems for training deep networks. This concept has become more promising owing to the enhanced big data processing capability, improvement in computing power with parallel processing units, and new algorithms for deep neural networks, which are becoming increasingly successful and attracting interest in many domains, including computer vision, speech recognition, and natural language processing. Recent studies in this technology augur well for medical and healthcare applications, especially in endoscopic imaging. This paper provides perspectives on the history, development, applications, and challenges of deep-learning technology.

Joon Sung Kim, Byung-Wook Kim, Do Hoon Kim, Chan Hyuk Park, Hyuk Lee, Moon Kyung Joo, Da Hyun Jung, Jun- Won Chung, Hyuk Soon Choi, Gwang Ho Baik, Jeong Hoon Lee, Kyo Young Song, and Saebeom Hur, The Korean Society of Gastroenterology, Korean College of Helicobacter and Upper Gastrointestinal Research, Korean Society of Gastrointestinal Endoscopy, The Korean Gastric Cancer Association, Korean Society of Interventional Radiology. Gut and Liver 2020;14:560-70. https://doi.org/10.5009/gnl20154

Subepithelial lesions occasionally found in the stomach of patients undergoing endoscopy may be either benign lesions or tumors with malignant potential. They may also appear due to extrinsic compression. Discrimination of gastric subepithelial lesions begins with meticulous endoscopic examination for size, shape, color, mobility, consistency, and appearance of the overlying mucosa. Accurate diagnosis can be achieved with endoscopic ultrasonography, which provides useful information on the exact size, layer-of-origin, and characteristic morphologic features to support a definitive diagnosis. Endoscopic ultrasonography also aids in the prediction of malignant potential, especially in gastrointestinal stromal tumors. Features of subepithelial lesions identified on endoscopic ultrasonography can be used to determine whether further diagnostic procedures such as endoscopic resection, fine needle aspiration, or core biopsy are required. Endoscopic ultrasonography is a valuable tool for diagnosis and clinical decision making during follow-up of gastric subepithelial lesions.