Development and internal validation of a prognostic nomogram for resected non-metastatic Siewert type II–III gastroesophageal junction adenocarcinoma: a single-center retrospective study

Prognostic assessment of resected gastroesophageal junction (GEJ) adenocarcinoma relies primarily on American Joint Committee on Cancer (AJCC) tumor–node–metastasis (TNM) staging, which does not incorporate comorbidity burden, lymph node ratio (LNR), signet-ring cell histology, or resection margin status. Integrated prognostic models specific to resected Siewert type II–III GEJ adenocarcinoma remain lacking. We developed and internally validated a nomogram for this population and compared its discrimination with AJCC staging. We retrospectively analyzed 206 patients with M0 Siewert type II–III GEJ adenocarcinoma who underwent resection at a single center (2010–2019). Independent predictors of overall survival were identified using multivariable Cox regression and assembled into a nomogram. Performance was assessed by internal validation with 1000 bootstrap resamples, calibration, and decision curve analysis, and was benchmarked against AJCC 8th edition staging. Patients were stratified into risk groups by tertiles of the nomogram score. Over a median follow-up of 112 months, 156 of 206 patients (75.7%) died. Five variables were independently prognostic: age-adjusted Charlson Comorbidity Index (hazard ratio [HR] 1.14), pathologic T4 category (HR 3.84), LNR (HR 1.31 per 0.1 increment), positive resection margin (HR 1.70), and signet-ring cell carcinoma (HR 2.12). The nomogram showed good discrimination (concordance index 0.800, bootstrap-corrected 0.793; 95% confidence interval [CI] 0.764–0.834), exceeding AJCC 8th edition staging (0.641; 95% CI 0.601–0.677) with non-overlapping intervals, and demonstrated adequate calibration and positive net benefit on decision curve analysis. Risk stratification separated patients into three groups with 5-year overall survival of 84%, 40%, and 4%. This internally validated nomogram integrates comorbidity, nodal, and histologic factors to provide individualized risk estimates that complement anatomic staging in resected Siewert type II–III GEJ adenocarcinoma. Prospective external validation is needed before routine clinical use.