Background
Triple-negative breast cancer (TNBC) is an aggressive subtype with poor outcomes, particularly in elderly patients. Chemotherapy remains the primary systemic treatment, but its effectiveness in TNBC patients aged 70 years and older remains controversial due to comorbidities and poor treatment tolerance. This study evaluates the survival impact of adjuvant chemotherapy in this population using a competing risk analysis.
Methods
A retrospective cohort of 4,855 elderly TNBC patients (≥70 years) was extracted from the SEER database (1995–2016). Propensity score matching (PSM) was applied to balance baseline characteristics between chemotherapy and non-chemotherapy groups. Overall survival (OS), cancer-specific survival (CSS), and competing risks of cancer-specific death (CSD) and other-cause death (OCD) were analyzed using Kaplan-Meier and Fine-Gray subdistribution proportional hazards models. A competing risk-based nomogram was developed to predict individualized survival outcomes.
Results
After PSM, the chemotherapy group showed significantly improved OS (p < 0.05), but no significant difference in 5-year CSD (16.88% vs. 19.01%, p = 0.1664). However, chemotherapy reduced the 5-year cumulative incidence of OCD (8.93% vs. 18.55%, p < 0.001). Multivariate competing risk analysis identified marital status, tumor grade, receipt of radiation therapy, T stage, and N stage as independent predictors of CSD. The nomogram incorporating these factors demonstrated high predictive accuracy (AUC: 1-year = 0.788, 3-year = 0.768, 5-year = 0.747) and excellent calibration.
Conclusion
Adjuvant chemotherapy provides limited direct benefit in reducing CSD but significantly decreases OCD in elderly TNBC patients, improving OS. The competing risk analysis offers a comprehensive understanding of mortality risks and underscores the value of individualized treatment. The validated nomogram supports precision medicine approaches for this high-risk population. Future studies should address underlying biological mechanisms and validate findings in multi-regional cohorts.