Prognosis comparison and prognostic model construction in patients with Siewert type II adenocarcinoma of the esophagogastric junction after neoadjuvant therapy between proximal gastrectomy and total gastrectomy.

The incidence of adenocarcinoma of the esophagogastric junction (AEG) is rising globally. As the most common subtype, Siewert type II AEG still lacks consensus on surgical strategy. For patients treated with neoadjuvant therapy, long-term survival differences between proximal gastrectomy (PG) and total gastrectomy (TG) remain unclear, and surgical-stratified prognostic models are currently unavailable. This study aimed to compare long-term outcomes of PG versus TG in Siewert type II AEG patients after neoadjuvant therapy, and to construct stratified prognostic models for individualized clinical decision-making and prognostic assessment. A total of 1,364 patients with Siewert type II AEG who received neoadjuvant therapy followed by PG or TG between 2004 and 2015 were enrolled from the Surveillance, Epidemiology, and End Results (SEER) database. Propensity score matching (PSM) was performed at a 2:1 ratio with a caliper width of 0.01 to balance confounding factors between groups. Survival analyses for overall survival (OS) and cancer-specific survival (CSS) were conducted both before and after PSM using the Kaplan-Meier analysis, followed by multivariate Cox proportional hazards regression to verify the independent association between surgical approach and long-term prognosis. Prognostic models for OS and CSS were separately constructed for the PG and TG groups. Variables were first screened by univariate Cox regression, and those with P<0.05 were entered into multivariate analysis; independent factors with P<0.05 were retained to establish nomograms. Model performance was validated using the concordance index (C-index), receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). Survival analyses before and after PSM consistently demonstrated that patients undergoing PG exhibited significantly superior OS and CSS compared with those receiving TG, with all comparisons achieving statistical significance (P<0.05). Multivariate Cox proportional hazards regression further identified TG as an independent risk factor associated with adverse long-term prognosis. To enable individualized risk stratification, we constructed four prognostic models stratified by surgical approach: the OS model for PG patients incorporated gender, age, tumor (T) stage, node (N) stage, and metastasis (M) stage; the OS model for TG patients comprised age, N stage, M stage, and the number of examined lymph nodes (ELNs); the CSS model for PG patients included gender, T stage, N stage, and M stage; and the CSS model for TG patients consisted of N stage and M stage. All models prioritized tumor-node-metastasis (TNM) stage-related indices as core prognostic predictors and exhibited robust discriminatory capacity, satisfactory calibration performance, and favorable clinical utility. In Siewert type II AEG patients receiving neoadjuvant therapy, PG confers better long-term survival. For patients with marked tumor regression, PG achieves an optimal balance between oncological safety and survival benefit. Prognostic models stratified by surgical approach aid precise clinical risk stratification, individualized surgical decision-making and postoperative management.