Abstract
PURPOSE: Malignant pleural effusion (MPE) is a severe complication in patients with advanced cancer that is associated with a poor prognosis. Breast cancer is the second leading cause of MPE after lung cancer. We therefore aim to describe clinical characteristics of the patients with MPE combined with breast cancer and construct a machine learning-based model for predicting the prognosis of such patients. METHODS: This study is a retrospective and observational study. Least absolute shrinkage and selection operator (LASSO) and univariate Cox regression analyses were applied to identify eight key clinical variables, and a nomogram model was established. Model performance was evaluated by receiver operating characteristic (ROC) curve, calibration curve, and decision curve analyses. RESULTS: 196 patients with both MPE and breast cancer (143 in the training group and 53 in the ex-ternal validation group) were analyzed in this study. The median overall survival in two cohorts was 16.20 months and 11.37 months. Based on the ROC curves for 3-, 6-, and 12-month survival, the areas under the curves were 0.824, 0.824, and 0.818 in the training set and 0.777, 0.790, and 0.715 in the validation set, respectively. In the follow-up analysis, both systemic and intrapleural chemotherapy significantly increased survival in the high-risk group compared to the low-risk group. CONCLUSION: Collectively, MPE confers a poor prognosis in breast cancer patients. We have developed a first-ever survival prediction model for breast cancer patients with newly diagnosed MPE and validated the model using an independent cohort.