Abstract
This retrospective cohort study aimed to investigate clinicopathological differences between early and late recurrence in young breast cancer patients aged 40 years or younger, identify independent prognostic factors for post-recurrence overall survival (prOS), and develop and validate a nomogram model for predicting prOS. A total of 515 breast cancer patients initially diagnosed between January 2018 and January 2022 at The First Affiliated Hospital of Xi'an Jiaotong University were consecutively enrolled. All patients were aged 40 years or younger and experienced first recurrence during follow-up, which continued until January 2026. Patients diagnosed between January 2018 and December 2020 comprised the training cohort (n=294), while those diagnosed between January 2021 and January 2022 formed the temporal validation cohort (n=221). Early recurrence was defined as first recurrence within 24 months of initial treatment; late recurrence occurred beyond 24 months. The training and validation cohorts showed comparable baseline clinicopathological characteristics (all P>0.05). In the training cohort, patients with early recurrence exhibited more aggressive tumor features and higher rates of distant and visceral metastasis compared to those with late recurrence. Median prOS in the training cohort was 55 months, with significantly shorter prOS in the early recurrence group (P<0.001). Multivariate Cox regression identified the following as independent risk factors for prOS: early recurrence (hazard ratio [HR]=2.578, 95% confidence interval [CI]: 1.711-3.883), tumor size T3-T4 (HR=1.950, 95% CI: 1.321-2.878), lymph node positivity (HR=2.190, 95% CI: 1.512-3.172), human epidermal growth factor receptor 2 (HER2)-overexpressing subtype (HR=2.347, 95% CI: 1.255-4.387), triple-negative breast cancer (TNBC) (HR=2.293, 95% CI: 1.250-4.205), receipt of neoadjuvant chemotherapy (HR=1.782, 95% CI: 1.277-2.488), and age younger than 35 years (HR=1.424, 95% CI: 1.012-2.003). The nomogram was constructed based on these factors. In the training cohort, the area under the curve (AUC) for predicting 1-year, 3-year, and 5-year prOS was 0.845, 0.820, and 0.649, respectively, with a concordance index (C-index) of 0.756. In the validation cohort, AUC values were 0.720, 0.725, and 0.601, respectively, with a C-index of 0.658. Calibration curves and decision curve analysis demonstrated acceptable predictive accuracy and clinical utility. In conclusion, early recurrence serves as an important predictor of poor post-recurrence survival in young breast cancer patients aged 40 years or younger. The nomogram incorporating recurrence timing, tumor burden, and molecular subtype can predict prOS with reasonable accuracy and may guide clinical risk stratification and individualized management.