Abstract
PURPOSE: To investigate survival-related factors in patients with breast cancer spinal metastases (BCSM) within the context of multidisciplinary treatment. METHODS: A retrospective cohort of 78 cases from July 2010 to December 2021 was recruited. These patients underwent surgery-based multidisciplinary treatment. Collected data included demographics, pathologies, symptoms, surgery-related data, adjuvant therapies, postoperative events, and survival data. The primary outcome was overall survival (OS). Kaplan-Meier survival curves were plotted. Univariate analysis employed the log-rank test, and post-hoc multivariate analysis utilized the Cox regression model. RESULTS: The mean age was 50.9 years. 72 cases (92.3%) reported locoregional pain, and 30 cases (38.5%) presented with neurological dysfunction. The primary pathological subtype was invasive ductal carcinoma (83.3%). SURGICAL PROCEDURES: total en-bloc spondylectomy (6.4%), debulking surgery (61.5%), palliative surgery (32.1%). Postoperatively, both pain and neurological function significantly improved (P < 0.05). Radiotherapy, endocrine therapy, chemotherapy/targeted therapy were given to 56.4%, 60.3%, 61.5% patients, respectively. The estimated OS was 50.0 months. Tomita's scores (P = 0.355) and Tokuhashi's scores (P = 0.461) showed no significant OS association. Univariate analysis indicated that preoperative neurological dysfunction (P = 0.003), postoperative neurological dysfunction (P = 0.051), adjuvant endocrine therapy (P = 0.025), and hormone receptor expression status (P = 0.009) were associated with patient survival. Multivariate analysis identified endocrine therapy as an independent protective factor for prognosis (aHR = 0.070, 95% CI 0.007-0.727, P = 0.026). CONCLUSIONS: Patients with BCSM have experienced prolonged survival. Neurological status, adjuvant anti-drugs, and expression of hormone receptors played crucial roles in predicting survival. Conventional prognostic systems may require modification to incorporate these factors. However, this study has limitations inherent to its retrospective design, single-center cohort, and relatively small sample size, which may affect generalizability.