A Prognostic Model for Breast Cancer With Liver Metastasis

Background: Breast cancer with liver metastasis consists of a group of heterogeneous diseases, and survival time may be significantly different, ranging from a few months to several years. The present study aimed to develop and externally validate a prognostic model for breast cancer with liver metastasis (BCLM). Methods: In total, 1022 eligible patients from January 2007 to December 2018 were selected from Fudan University Shanghai Cancer Center (FUSCC) and were temporally in the training (n = 715) and validation (n = 307) set. According to regression coefficients found in the multivariate Cox regression analysis, the final results were transformed into the prognostic scores. On the basis of these scores, patients were finally classified into three risk groups, including low-, intermediate-, and high-risk groups. Bootstrapping was used for internal validation. Then, time-dependent receiver operating characteristic (ROC) curves and calibration plots were used to assess discrimination and calibration of this prognostic model in the validation set. Results: Molecular subtypes, metastatic-free interval (MFI), extrahepatic metastasis, and liver function tests were identified as independent prognostic factors in the multivariate analysis. According to risk stratification, intermediate-risk (hazard ratio (HR) 2.12, 95% confidence interval (CI) 1.74–2.58, P < 0.001) and high-risk groups (HR 6.94, 95% CI 5.25–9.16, P < 0.001) had significantly worse prognoses in comparison with the low-risk group regarding overall survival (OS) from the time of metastasis. The median OS in these three groups were 39.97, 21.03, and 8.80 months, respectively. These results were confirmed in the internal and external validation cohorts. Conclusions: Based on molecular classification of tumors, routine laboratory tests, and other clinical information easily accessible in daily clinical practice, we developed a clinical tool for BCLM patients to predict their prognosis. Moreover, it may be useful for identifying the subgroup with unfavorable prognosis and individualization of treatment.