Abstract
BACKGROUND: Triple-negative breast cancer (TNBC) lacks targeted therapies and precise prognostic tools. This study developed a prognostic nomogram integrating clinicopathological factors and treatment response dynamics to improve survival prediction. METHOD: Data from 2,978 TNBC patients (SEER database, 2000–2020) were analyzed. Independent prognostic factors were identified via Cox regression. A nomogram incorporating race, AJCC N/M stage, tumor size, surgery type, and pathological response (pCR/pPR/pNR) was constructed. Performance was evaluated using C-index, ROC-AUC, calibration, decision curve analysis (DCA), and compared to AJCC-TNM staging. RESULT: Multivariate analysis identified N3 stage (HR = 4.13), M1 stage (HR = 1.77), tumor size ≥ 90 mm (HR = 1.84), mastectomy (HR = 1.28), and pathological non-response (pNR, HR = 6.87) as independent risk factors (all P < 0.05). The nomogram achieved superior discrimination (C-index: 0.780 [training], 0.773 [validation] vs. TNM’s 0.715–0.720). AUCs for 1-/3-/5-year survival were 0.858/0.823/0.820 (training) and 0.0.864/0.802/0.799 (validation). Calibration errors were < 5% for 1–3-year predictions. DCA demonstrated a 7–10% net benefit increase over TNM staging, with 3.9 additional correct decisions per 100 patients at the 40% risk threshold. CONCLUSION: This nomogram dynamically integrates pathological treatment response, significantly outperforming TNM staging (ΔC-index = + 0.066). It enables personalized risk stratification and clinical decision-making, particularly for guiding therapy intensification in high-risk subgroups (e.g., N3/pNR). Future models should incorporate molecular biomarkers (e.g., PD-L1, BRCA) and socioeconomic variables to enhance precision. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s12672-025-04251-y.