Development and Validation of a Competing-Risk Nomogram for Predicting Breast Cancer–Specific Survival in Young Women with Early-Stage Breast Cancer: A SEER-Based Study

Abstract

Background & Objective: Breast cancer in women aged ≤40 years is biologically aggressive and carries a disproportionate cancer-specific mortality risk. Existing prognostic tools predominantly predict overall survival, rarely account for competing risks of non-cancer death, and have not evaluated whether primary treatment adds prognostic value beyond tumor biology. This study aimed to develop and validate a competing-risk nomogram for predicting 3- and 5-year breast cancer–specific survival (BCSS) in this population. Methodology: Data were extracted from the SEER 17-registry database (2010–2015). Women aged 18–40 years with histologically confirmed T1–2N0–1M0 invasive breast cancer were included (n = 8,221) and randomized into training (n = 5,755) and validation (n = 2,466) cohorts (7:3). Univariate and multivariate Cox regression, scaled Schoenfeld residuals, and Fine–Gray subdistribution hazard models were applied. A nomogram was constructed from independent prognostic factors and validated using the concordance index, bootstrap resampling, time-dependent AUC, Brier score, and calibration curves. Results: Over a median follow-up exceeding 167 months, 823 breast cancer–specific deaths and 119 other-cause deaths occurred. Multivariate analysis identified race, differentiation grade, lymph node status, molecular subtype, and T stage as independent predictors of BCSS. Primary treatment variables (surgery type, radiotherapy, chemotherapy) were associated with worse univariate survival but lost significance after adjustment for biological covariates and were excluded from the final model. Fine–Gray subdistribution hazard ratios closely paralleled Cox hazard ratios, indicating negligible informative censoring bias. The nomogram demonstrated a C-index of 0.704 (optimism-corrected 0.698) in the training cohort and time-dependent AUCs of 0.794 (3-year) and 0.757 (5-year) in the validation cohort, with good calibration. Conclusion: This validated competing-risk nomogram provides individualized 3- and 5-year BCSS probability estimates for young women with early-stage breast cancer based on readily available clinicopathological variables, offering a practical tool to refine risk stratification and inform shared decision-making in this high-risk population. Key Words: Competing Risk Nomogram, Breast Cancer, Survival.