BackgroundThere is a significant survival difference and lack of effective treatment among breast cancer patients with liver metastasis. This present study aimed to construct a novel prognostic score for predicting the prognosis and locoregional treatment benefit of de novo metastatic breast cancer with liver metastasis (BCLM).MethodsIn total, 2,398 eligible patients between 2010 and 2016 were selected from the Surveillance, Epidemiology, and End Results (SEER) database. They were assigned to the training set including 1,662 patients (2010–2014) and validation set comprising 736 patients (2015–2016) depending on the time of diagnosis. The prognostic score was based on regression coefficients in the multivariate Cox regression analysis. And then, patients were stratified into low-, intermediate-, and high-risk groups by the prognostic score. The discrimination and calibration of prognostic score were evaluated using time-dependent receiver operating characteristic (ROC) curves analysis and calibration curves, respectively. Subgroup analysis was performed to evaluate locoregional surgery and chemotherapy benefit in different risk groups.ResultsAge, race, insurance and marital status, T stage, pathological grade, molecular subtypes, and extrahepatic metastasis were identified as independent prognostic variables in the prognostic score. The prognostic score showed high discrimination power with an area under the curve (AUC) of 0.77 and 0.72 and excellent agreement suggested by calibration plots in the training and validation sets, respectively. Intermediate-risk [hazard ratio (HR) 2.39, 95% confidence interval (CI) 2.09–2.73, P<0.001] and high-risk groups (HR 4.88; 95% CI 4.13–5.76; P<0.001) had significantly worse prognosis in comparison with the low-risk group. The median overall survival (OS) in three prognostic groups were 44, 18, and 7 months, with a 3-year survival rate of 56, 23, and 7%, respectively. Apart from the high-risk group (HR 0.79; 95% CI 0.56–1.10; P=0.157), the low-risk (HR 0.64; 95% CI 0.49–0.84; P=0.001) and intermediate-risk groups (HR 0.68; 95% CI 0.55–0.85; P=0.001) could benefit from the surgery of primary site, while chemotherapy improved prognosis in all risk groups.ConclusionsA prognostic score was developed to accurately predict the prognosis of de novo BCLM patients. Moreover, it may be useful for further subdividing them into different risk groups and helping guide clinicians in treatment decisions.