Survival Estimation, Prognostic Factors Evaluation, and Prognostic Prediction Nomogram Construction of Breast Cancer Patients with Bone Metastasis in the Department of Bone and Soft Tissue Tumor: A Single Center Experience of 8 Years in Tianjin, China

Abstract

Purpose. Bone metastasis in breast cancer remains globally concerned. Accurate survival estimation would be beneficial for clinical decision-making, especially for the patients with potential indications of surgery. Based on a retrospective cohort from China, the study aimed to construct a prognostic prediction nomogram for breast cancer patients with bone metastasis. Methods. Breast cancer patients with bone metastasis diagnosed between 2009 and 2017 in our department were retrospectively selected. The total cohort was divided into construction and validation cohorts (ratio 7 : 3). A nomogram was constructed to predict the probability of survival, and the performance of model was validated. Results. A total of 343 patients were enrolled with 243 and 100 patients in construction and validation cohorts, respectively. The median overall survival for the total cohort was 63.2 (95% CI: 52.4–74.0) months. Elevated ALP (HR = 1.71, 95% CI: 1.16–2.51; $P=0.006$ ), no surgery for breast cancer (HR = 2.19, 95% CI: 1.30–3.70; $P=0.003$ ), synchronous bone metastasis (HR = 1.98, 95% CI: 1.22–3.22; $P=0.006$ ), and liver metastasis (HR = 1.68, 95% CI: 1.20–2.37; $P=0.003$ ) were independent prognostic factors for worse survival. The independent predictors and other five factors (including age at diagnosis, ER status, PR status, Her-2 status, and the performance of bisphosphonate) were incorporated to construct the nomogram. The C-index was 0.714 (95% CI: 0.636–0.792) and 0.705 (95% CI: 0.705) in the construction cohort and validation cohort, respectively. All the calibration curves were close to the 45-degree line, which indicated satisfactory calibration. Conclusion. A retrospective study aiming at prognostic estimation of breast cancer patients with bone metastasis was designed. Four independent prognostic factors were identified and a prognostic nomogram was constructed with satisfactory discrimination and calibration. The model could be used in survival estimation and individualized treatment planning.