Integrating CD45 + CD3+ and CD3 + CD8+ T-cell markers into nomograms for predicting survival in gastroesophageal junction carcinoma.

To develop and validate prognostic nomograms for overall (OS) and progression-free survival (PFS) in gastroesophageal junction (GEJ) carcinoma to enable risk-stratified clinical decision-making. In this retrospective study of 618 patients, prognostic factors were selected using LASSO regression. Nomograms were constructed and validated by assessing discrimination (C-index, time-dependent ROC curves), calibration, and clinical utility (decision curve analysis). Patients were stratified into high- and low-risk groups based on nomogram-derived scores. The OS nomogram incorporated serum CD3 + CD8+ cells, mismatch repair status, metastatic lymph node count, and maximum tumor diameter. It demonstrated superior discriminative performance compared with the AJCC 9th edition staging system (C-index 0.716, 95% CI 0.679-0.752 vs. 0.683, 95% CI 0.646-0.721). The PFS nomogram included CD45 + CD3+CD19 + cells, CD3 + CD8+ cells, carcinoembryonic antigen, metastatic lymph nodes, and tumor diameter, also outperforming AJCC staging (C-index 0.709, 95% CI 0.676-0.743 vs. 0.647, 95% CI 0.610-0.684). Both models showed excellent calibration. Decision curve analysis confirmed their clinical utility across risk thresholds, with significant survival differences between risk groups (log-rank test, p < 0.001). Restricted cubic spline analysis revealed nonlinear associations between key immune cell levels (CD45 + CD3+ and CD3 + CD8+) and survival outcomes (all p < 0.05). This study establishes the first clinically validated risk-stratification model for GEJ carcinoma. The nomograms precisely identify low-risk patients who may avoid unnecessary adjuvant therapy and high-risk patients likely to benefit from treatment intensification, thereby enhancing personalized management.