Development of a predictive model to identify patients most likely to benefit from surgery in metastatic breast cancer

Abstract

Abstract Primary tumor resection for metastatic breast cancer (MBC) has demonstrated a survival advantage, however, not all patients with MBC benefit from surgery. The purpose of this study was to develop a predictive model to select patients with MBC who are most likely to benefit from surgery at the primary site. Data from patients with MBC were obtained from the Surveillance, Epidemiology and End Results (SEER) cohort and patients treated at the Yunnan Cancer Hospital. The patients from the SEER database were divided into surgery and non-surgery groups and a 1:1 propensity score matching (PSM) was used to balance baseline characteristics. We hypothesized that patients who underwent local resection of primary tumors had improved overall survival (OS) compared to those who did not undergo surgery. Based on the median OS time of the non-surgery group, patients from the surgery group were further categorized into beneficial and non-beneficial groups. Logistic regression analysis was performed to identify independent factors associated with improved survival in the surgery group and a nomogram was established using the most significant predictive factors. Finally, internal and external validation of the prognostic nomogram was also evaluated by concordance index (C-index) and using a calibration curve. A total of 7759 eligible patients with MBC were identified in the SEER cohort and 92 with MBC patients who underwent surgery at the Yunnan Cancer Hospital. Amongst the SEER cohort, 3199 (41.23%) patients received surgery of the primary tumor. After PSM, the OS between the surgery and non-surgery group was significantly different based on Kaplan–Meier survival analysis (46 vs. 31 months, P < 0.001), In the surgery group, 562 (55.20%) patients survived for longer than 31 months and were classified in the beneficial group. Significant differences were observed in patient characteristics between the beneficial and non-beneficial groups including age, grade, tumor size, liver metastasis, breast cancer subtype and marital status. These factors were used as independent predictors to create a nomogram. The internally and externally validated C-indices of the nomogram were 0.703 and 0.733, respectively, indicating strong consistency between the actual and predicted survival. A nomogram was developed and used to identify MBC patients who are most likely to benefit from primary tumor resection. This predictive model has the potential to improve clinical decision-making and should be considered routine clinical practice.