Breast cancer shows plasticity in terms of classification. Upon drug treatment and metastasis some tumors switch to another subtype leading to loss of response to therapy. In this study, we ask the question which molecular subclasses of breast cancer are more switchable upon drug therapy and metastasis. We used in silico data to classify breast cancer tumors in PAM50 molecular classes before treatment and after treatment using gene expression data. Similar analysis was performed for primary tumors and their metastatic growth. Our analysis showed that in both scenarios some breast tumors shift from one class to another. This suggests that patients who underwent chemotherapy but resulted in relapse or/and metastasis should be retyped for molecular subclass so that treatment protocol should be adopted according to those subtypes. Additionally, 20 genes were identified as biomarkers for metastasis in breast cancer.