Semantic heterogeneity is a key issue in the deployment of medical applications. In this paper, we examine solutions to address semantic heterogeneity for a complex organ transplantation information system. The information system that has been developed by the French agency for transplantation (EfG) has to gather medical information concerning patients and donors for organ allocation, epidemiological studies, and public health decisions. We analyze in this context the limits of the traditional approach based on a standard vocabulary and a relational database for the transplantation domain. We present its evolution towards a system that combines a terminology server and a data warehouse. The perspective is now to build a formal ontology that may support semantic integration in an advanced transplantation information system. Two open issues in formal ontology design are discussed. First, providing definitions of all medical concepts is not an obvious task if we want those definitions be meaningful, formal, and compatible with other ontologies. Second, propagation of properties along relations raises the needs for extending description logics with rules.