A therapeutic strategy, biomarker-driven and response-guided approach has been investigated in cancer therapy where the treatment targets heterogeneous and unstable disease. Neoadjuvant chemotherapy, for instance, is indicated based on tumor stage and subtype and its therapeutic outcomes like pathological responses are associated with the long-term prognostic probability in subgroups such as hormone receptor (HR) negative and HR-positive patients with high-grade cancers. Therefore, it would be reasonable to consider a treatment plan according to the short-time response in the stratified subgroups. It is also applicable for new therapy development, and in fact many clinical trials are under investigation in the post-neoadjuvant setting. In order to increase the therapeutic efficacy, it is recognized as necessary to incorporate biomarkers that enable us to classify conventional subtypes further including genetic mutations and epigenetic phenotypes into the planning of treatment. It is also crucial to analyze tumor biology particularly tumor evolution in the metastasis and the clonal selection by the treatment in these clinical settings.