Risk Factors for Mediastinal Lymph-Node Metastasis in Siewert Type I/II Esophagogastric Junction Cancers

Background/Objectives: Currently, a definitive diagnostic strategy for predicting mediastinal lymph-node (MLN) metastasis remains poorly established in Siewert Type I and II esophagogastric junction adenocarcinoma (EGJAC). This study aimed to evaluate long-term survival and identify objective, reliable preoperative risk factors for MLN metastasis to optimize perioperative treatment and surgical approach. Methods: We retrospectively reviewed a well-defined cohort of 133 patients with Siewert Type I (33.1%) and II (66.9%) EGJAC who underwent R0 resection and complete follow-up at our institution between 2005 and 2025. Preoperative risk factors for histological MLN metastasis were evaluated using receiver operating characteristic (ROC) curve analysis and univariate/multivariate logistic regression models. Results: Thorough upper, middle, and lower MLND were completed in 57.9%, 72.9%, and 94.7% of patients, respectively. Pathological MLN metastasis was confirmed in 31.6% of patients and was significantly associated with poor overall and recurrence-free survival. ROC analysis demonstrated that an MLN short-axis diameter of ≥5 mm on CT, esophageal involvement length (EIL) in upper GI series ≥ 30 mm, and EIL in endoscopy ≥ 60 mm predict histological MLN metastasis. In univariate analysis, Siewert Type I, EIL in upper GI series ≥30 mm, EIL in endoscopy ≥ 60 mm, clinical tumor depth ≥ cT3, and CT node size ≥ 5 mm were correlated with MLN metastasis. Multivariate logistic regression analysis revealed that a clinical tumor depth of ≥cT3, EIL in upper GI series ≥ 30 mm, and a CT node size of ≥5 mm were independent predictors of MLN metastasis. Conclusions: Clinical tumor depth, EIL, and size of MLN are powerful independent preoperative risk factors for MLN metastasis in Siewert Type I/II EGJAC. Utilizing these objective indices may help identify high-risk patients requiring intensive perioperative multidisciplinary strategies.