Development and Validation a Survival Prediction Model and a Risk Stratification for Elderly Locally Advanced Breast Cancer.

Purpose: we aimed to develop an individualized survival prediction model for elderly locally advanced breast cancer (LABC) and stratify its risk to assist in the treatment and follow-up of patients.

Methods: Elderly LABC data were extracted from the Surveillance, Epidemiology, and End Results (SEER) database. The best model was screened using Cox, least absolute shrinkage and selection operator (LASSO) and best subset regression to construct the nomogram. After internal and external validation of this model, risk stratification was established, and differences between risk groups were assessed using Kaplan-Meier method.

Results: A total of 10,697 elderly LABC patients were divided into a training group (n = 7131) and a validation group (n = 3566) with a 5-year overall survival rate of 57.6% [confidence interval (CI): 56.4%-58.7%]. A nomogram was developed using age, marital status, histological grading, estrogen and progesterone receptors, surgery, radiation therapy, and chemotherapy as predictors. This model was evaluated and validated to perform well, with a discrimination index of 0.744 (95% CI: 0.734-0.753). Patients were divided into low, medium and high groups based on risk scores, and there was a significant difference in survival between the 3 groups.

Conclusion: The prognosis of elderly LABC was poor. The nomogram constructed based on prognostic factors could accurately predict the prognosis, which would provide a reference for treatment and follow-up.