Study design and patients
This was a single-center, retrospective study conducted at Tonan Hospital, Hokkaido, Japan. The study protocol was approved by the institutional review board of Tonan Hospital, and the study was conducted in compliance with the principle of the Declaration of Helsinki of 1964 and later versions.
We analyzed the medical records, endoscopic reports, and pathological findings of consecutive patients with EGCs who underwent ESD or gastrectomy with lymphadenectomy at our hospital between January 2010 and October of 2014. The inclusion criteria were differentiated-type EGC, age ≥20 years, availability of detailed pathological diagnosis, and follow-up more than 5 years or any-cause death within 5 years after the procedures. The exclusion criteria were multiple synchronous gastric cancers, locally recurrent cancers, remnant stomach or gastric tube cancers, and cancers with a neuroendocrine carcinoma component. All resected EGCs were classified by patient age, sex, tumor location, and macroscopic morphological type. The tumor location was divided into the upper, middle, and lower part of the stomach, and the macroscopic type was classified according to the JCGC guidelines .
The resected specimens obtained via ESD or gastrectomy were evaluated pathologically according to the JCGC guidelines . When lymphatic invasion was suspected, immunohistochemistry using D2-40 was performed. Vascular invasion was determined using Elastica van Gieson staining. Histological types were classified following the Japanese Gastric Cancer Association (JGCA) guidelines . Well differentiated, moderately differentiated, and papillary adenocarcinoma were classified as differentiated-type carcinoma, while poorly differentiated adenocarcinoma and signet ring cell carcinoma were classified as undifferentiated-type carcinoma. All resected lesions were categorized into either PD or MD types (differentiated-type-predominant carcinoma including undifferentiated-type component) according to the proportions of components at histopathology.
The clinicopathological features analyzed were maximum tumor size, tumor depth, histological type, presence of lymphatic and vascular invasion, ulcerative findings, horizontal margin (for cases treated with ESD), vertical margin (for cases treated with ESD), the rate of endoscopic curative resection (for cases treated with ESD), and the rate of LNM. Additionally, we analyzed the LNM rate of both surgical and ESD cases meeting curative endoscopic resection criteria. The criteria for endoscopic curative resection stipulated in the JGCA guidelines were as follows : differentiated mucosal cancer regardless of size in the absence of ulceration, tumors < 30 mm in diameter in the presence of ulceration, and tumors < 30 mm with invasion into the superficial layer of the submucosa (SM <500 µm). With respect to MD carcinoma, the lesions with areas of undifferentiated type carcinoma exceeding 2 cm or undifferentiated type component in the part that had invaded the submucosa were defined as non-curative resection. Although the patients with endoscopically non-curative resection underwent additional gastrectomy with lymphadenectomy, some high-risk patients, such as those with advanced age or severe comorbidities, selected non-surgical observation. In curative ESD cases and non-curative ESD cases without additional gastrectomy, LNM negativity was defined as the absence of LNM on follow-up computed tomography (CT) for at least 5 years after ESD.
Analysis of long-term prognosis
The long-term prognosis was analyzed with respect to the recurrence of gastric cancer, overall survival, and EGC-specific survival at 5 years. In addition, the long-term prognosis of both surgical and ESD cases meeting curative endoscopic resection criteria was analyzed. Survival and recurrence were determined using medical records. The patients who were referred to another hospital after treatment were surveyed by letters to the referred hospitals. The prognosis of patients who completed follow-up was investigated using telephone calls to the patients or their family members.
Quantitative variables were expressed as the mean, while categorical variables were presented as total numbers and percentages. Pearson’s chi-squared test and Mann-Whitney U-test were applied as appropriate. Survival rates from the date of treatment were calculated by using the Kaplan-Meier method and compared using the log-rank test. A p-value < 0.05 was considered statistically significant. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R 2.13.0 (R Foundation for Statistical Computing, Vienna, Austria) .