Biomedical *K*nowledge *O*rganization *S*ystems (KOS) and their mappings are widely adopted nowadays and have a growing relevance for different software applications. By virtue of the size of the biomedical domain, these KOS are huge, complex and semantic correspondences are established between them for integration purpose. However, the highly dynamic aspect of the biomedical knowledge leads to the frequent release of new versions of KOS impacting the validity of mappings already determined. An open research problem is how to adapt these mappings maintaining their reliability over time. Approaches regenerating mappings between evolved KOS are still very costly solutions in terms of human efforts and expertise. We have investigated the most adequate mapping adaptation actions to apply according to different types of change operations identified from KOS evolution. In this talk we present an initial framework for coping with mapping adaptation at KOS evolution time. Results from empirical experiments conducted for understanding mappings evolution are also showed.