Patients
Between January 2000 and December 2013, the data of 1822 patients who underwent radical surgery at the Department of Gastric Surgery, Sun Yat-sen University Cancer Center were retrieved. Eligibility criteria for patient inclusion comprised of: (1) histologically confirmed diagnosis of gastric adenocarcinoma and R0 gastrectomy; (2) no other synchronous malignancy; (3) no preoperative chemotherapy or perioperative radiotherapy; (4) gastrectomy and lymphadenectomy based on the Japanese Gastric Cancer treatment guidelines12; (5) more than 15 post-operative pathologically reported lymph nodes, and; (6) a postoperative survival time of more than 1 month; considered as the non-surgical relative death. Patients with carcinoma of the gastric stump after gastric resection for benign disease were excluded from this study. Retrieved data included the patient gender, tumor size, histological grade, status of vascular invasion, nerve invasion, adipose connective tissue invasion, GBA invasion, depth of invasion (pT), nodal status (pN), distant metastasis (pM), and the number of retrieved lymph nodes. Pathological staging was performed according to the 8th edition of the AJCC cancer staging manual.
Follow-up protocol
Follow-ups were performed by telephone, email, or outpatient department visits. The last follow-up date was December 2018. The postoperative follow-ups included clinical and laboratory examinations every 3 months for the first 2 years at our outpatient department, every 6 months from the third to fifth postoperative years, and then annually thereafter or until the patient died. Overall survival (OS) was defined as the time from the operation to death or the last follow-up.
Range of GBAI
Three important ligaments compose the boundary of the GBA, namely the medial gastrodiaphragmatic ligament (MGDL), lateral gastrodiaphragmatic ligament (LGDL) and gastropancreatic ligament (GPL), as shown in Figure 1. The hepatogastric ligament contains two layers, the anterior layer and posterior layer. The rightward extension of anterior layer becomes the peritoneum of the anterior gastric wall at the lesser curvature. The posterior layer turns posteriorly and right laterally at the lesser curvature, connecting with the right diaphragmatic crura. This peritoneum reflex from the lesser curvature to the right diaphragmatic crura is the so called MGDL which is the right boundary of the GBA. The right posterior extension of the peritoneum of the posterior gastric wall from the fundus and the great curvature to the left diaphragmatic crura becomes the LGDL, which is the left boundary of the GBA. The inferior extensions of MGDL and LGDL becomes the GPL at the superior border of the pancreas which is the inferior boundary of GBA. The left gastric artery and vein pass through the GPL. Based on the 3rd English edition of Japanese classification of gastric carcinoma, the definition of the station number 1, 2, 7, and 9 LNs stations are: the right paracardial LNs including those along the first branch of the ascending limb of the left gastric artery; the left paracardial LNs include those along the esophagocardiac branch of the left subphrenic artery; LNs along the trunk of the left gastric artery between its root and the origin of its ascending branch; and the LNs along the celiac artery, respectively13. Thence, we consider the adipose tissues invasion in the area of LNs station number 1, 2, 7, and 9 or attached to the primary tumor involving the proximal of stomach as GBAI.
Histologic Evaluation of GBAI
For each patient, all postoperative pathological slides were reviewed to evaluate the presence of GBAI. Existence of GBAI was examined by reviewing the adipose tissues attached to the primary tumor involving the proximal stomach, and the LNs station number 1, 2, 7, and 9. For pathologic examination of these adipose tissues, tumor deposits (TDs) located >5mm from the leading edge of primary tumors and LNs were considered as the GBAI, irrespective of their shape, contour, and size.
In this study, adipose tissue invasion was defined as TDs residing in the adipose tissues neighboring the stomach. These TDs were identified using these three primary characteristics: (1). they were located in the adipose tissues and were separated from the primary tumor or lymph nodes; (2). there were no structures of blood or lymphatic vessel, lymphatic nodes, nerves around them; (3). they were enveloped by a proper fascia and could be clearly discriminated from peritoneal seeding.
Imaging assessment
One radiologist performed all imaging reassessment. She was blinded to the clinical and survival data of the patients. CT images were evaluated independently at the workstation using transverse CT (images were reconstructed with a 5-mm section thickness). Metastasis were categorized into five types, namely as retroperitoneal infiltration, peritoneal, hematogenous, lymphatic, and others. The radiological manifestations including posterior-pancreas infiltration, para-aortic infiltration, mesenterium roots thickening, and thickening of the rectal lining were considered as retroperitoneal infiltration, as shown in Figure 2; peritoneal nodules/thickening, Douglas pouch nodules and abdominal mass were considered as peritoneal metastasis; liver, lung, and bone were considered as hematogenous metastasis; para-aortic, left supraclavicular, mediastinum, porta hepatis LNs were considered as lymphatic metastasis; umbilical region, ovary mass, anastomotic astium were considered as others. It was difficult to classify ascites as retroperitoneal infiltration or peritoneal metastasis, so ascites was considered as others.
Statistical analysis
The 5-year overall survival rate was estimated using the Kaplan-Meier method and univariate comparisons between groups was performed using the log-rank test. In the multivariate analyses, the Cox’s proportional hazards model was carried to estimate the relative risks and to identify corresponding prognostic factors. All data analyses were performed using the SSPS software (version 22.0, Stata Corporation, Texas, USA). A P value less than 0.05 (two-sided) was considered as statistically significant.