Abstract
Background: In adenocarcinoma of esophagogastric junction (AEG), the relationship between tumor size (TS) and lymph node metastasis (LNM) is unclear. This study aimed to explore the relationship between TS and LNM, and to construct a prediction model for LNM. Materials and Methods: Data from 4649 Siewert type II AEG patients were retrospectively acquired from the Surveillance, Epidemiology, and End Result (SEER) database. TS data was analyzed as a continuous variable, but also divided into 1-cm-interval categorical groups for further analysis. The logistic regression model and restricted cubic spline (RCS) model was used to explore the relationship between TS and LNM, after adjusting for covariates. Internal validations as well as external validation (Single-Center data) were used to check our LNM prediction model. Results: TS and LNM showed a significant relationship in the logistic regression analysis, regardless of the TS data being entered as a continuous or a categorical variable, after adjusting for covariates. The logistic regression model and RCS consistently showed that larger TS resulted in larger Odds Ratio (OR) values. When tumors were larger than 4 cm, the OR value remained relatively constant. The receiver operator characteristic curve evaluated the nomogram by the area under the curve (AUC) (AUC=0.737, in internal validation; AUC=0.626, in external validation), and the calibration curve of the nomogram showed an improved prediction system. Conclusions: In Siewert type II T1-T3 stage AEG patients, we reported that LNM increased with TS up to 4-cm, and our nomogram provided a simple tool to predict LNM.