Development And Validation Of A Novel Clinical Prognostic Model For Breast Cancer Patients With Malignant Pleural Effusion

Background: MPE(Malignant pleural effusion)is one of the common complications in patients with advanced malignant tumors,usually with poor prognosis and high mortality.In recent years,it has received a lot of attention from clinicians because of its increasing incidence.The mechanism of occurrence and development of MPE is very complex,and there are no clear clinical treatment guidelines at present.It is of great significance to evaluate the severity and prognosis of such patients as soon as possible.Most previous studies of malignant pleural effusions have focused on the field of lung cancer.Breast cancer is the second leading cause of malignant pleural effusion,and it has unique molecular characteristics and a high degree of heterogeneity.Therefore,it is necessary to construct a reliable prognostic model in a cohort of patients with breast cancer combined with MPE.Objectives: Herein,patients with breast cancer combined with malignant pleural effusion were investigated for relevant factors affecting prognosis,including clinical features,pathological features and laboratory indicators.The ultimate goal is to construct a novel machine learning-based model for predicting the prognosis of patients with MPE combined with breast cancer to facilitate risk stratification of patients and guide follow-up treatment.Methods: This study is a retrospective and observational study.After collecting baseline clinical information of patients,LASSO(least absolute shrinkage and selection operator)analysis and univariate Cox regression analysis were used to identify prognostic risk factors and construct a nomogram model.The validation includes internal and external verification.Model performance was evaluated by ROC(receiver operating characteristic curve),calibration curve,and decision curve analyses.Finally,Kaplan-Meier survival curve and Log-rank test were used to evaluate the risk stratification ability of the model.Results: A total of 196 patients with breast cancer combined with MPE(143 in the training set and 53 in the validation set)were included in this study.The pleural effusion in all patients was confirmed as MPE by cytology.The median age of patients in the training set was 57 years,and 76% of breast cancers were estrogen or progesterone receptor positive.Univariate Cox regression analysis showed that 13 variables had predictive effects on the survival of patients.In the application of LASSO regression,the model with the lowest mean error was selected through cross-validation.The nomogram model contained key clinical variables,among which the variable with the highest contribution to the model was the level of glucose in pleural effusion,followed by the disease-free survival of the patients and the ratio of blood neutrophils to lymphocytes.The model was able to classify patients into two different subgroups,high and low risk,with significant differences in survival curves between the different risk groups(Logrank P <0.0001 for the training set and Log-rank P =0.0212 for the validation set).The ROC for 3-,6-,and 12-month survival were plotted with an area under the ROC curve of 0.824,0.824,and 0.818 for the training set and 0.777,0.790,and 0.715 for the validation set,respectively.In addition,using this model,patients can be divided into high and low risk groups,and there are significant differences in survival curves between different risk groups(training set Log-rank P < 0.0001,validation set Log-rank P =0.0212).Follow-up exploratory analysis found that systemic intravenous chemotherapy and intrapleural infusion chemotherapy could significantly improve the survival of patients in the high-risk subgroup,while the survival benefits of chemotherapy in the low-risk subgroup were not significant.Conclusions: This study confirmed that the prognosis of breast cancer patients with MPE is poor and there is significant difference.Based on the clinical characteristics and laboratory indexes of patients,a prognostic prediction model was developed and verified in this study.This model is a first-ever survival prediction model developed in a cohort of breast cancer patients combined with MPE.It shows a good performance and can accurately predict the prognosis of patients and provide guidance for subsequent individualized treatment.