Breast Cancer is a complex disease characterized by multiple variables obtained from several data-sources, such as clinical, genetic or image sources. Over the past decades, various studies have tried to predict the outcome of breast cancer with the support of these data, and big advances have been done in this direction. However, only a few reports describe the causal relationships among the variables and outcomes, such as adverse events and survival rate, and usually they are very limited to a specific dataset. This research work presents a novel system that using data mining and visual analytics tools depicts in an intuitive way the patterns associated with different outcomes, such as treatment response and adverse events related to a treatment. For that the system processes heterogenous data coming from a real setting for primary breast cancer. This way clinicians can explore in a dynamic, fast and intuitive way whether certain group of patients are prone to certain outcomes.