Prediction for Survival of 1 313 Patients with Breast Cancer Based on Lasso-Cox Regression and Hierarchical Clustering Analysis of Clinicopathological Factors
[Objective]To explore the prediction for survival of breast cancer patients based on Lasso-Cox regression and Hierarchical clustering analysis of clinicopathological features.[Methods]A total of 1 313 breast cancer patients admitted in the Affiliated Cancer Hospital of Xinjiang Medi-cal University from January to December 2016 were included in this study,including 1 139 survival cases and 174 fatal cases.Lasso-Cox regression was used to analyze variables associated with the prognosis of breast cancer patients.Hierarchical clustering based on categorical data was performed to reveal the similarity of clinical and pathological characteristics,and Kaplan-Meier survival anal-ysis curves were plotted to compare the prognosis between different categories of breast cancer pa-tients.[Results]Fourteen clinicopathological features related to the prognosis of breast cancer pa-tients were screened out by Lasso-Cox regression,including T stage,N stage,M stage,clinical stage,neoadjuvant therapy,type of surgery,lymph node dissection,adjuvant therapy,ER positive expression,molecular typing,brain metastasis,lung metastasis,liver metastasis and bone metas-tasis.Patients were clustered into 2 categories based on hierarchical clustering.Kaplan-Meier sur-vival analysis showed that there were significant differences in the prognosis between two categories of patients(x2=397.00,P＜0.001).[Conclusion]Breast cancer patients exhibit high heterogeneity,and patients with varying clinicopathological features have different survival prognosis.Therefore,it is necessary to classify patients based on their clinical and pathological characteristics in order to develop personalized treatment plans.
breast cancerLasso-Cox regressionhierarchical clustering analysisprognosis
NETL
NSTL
万方数据
