In model-driven engineering (MDE), a particular MDE setting of employed languages and automated and manual activities has major impact on productivity. Furthermore, it has been observed that such MDE settings evolve over time. However, currently not much is known about this evolution and its impact on the MDE setting's maturity, i.e., on changeability and other productivity dimensions. Research so far focuses on evolution of separate building blocks, such as (modeling-) languages, tools, or transformation, only. In this article, we address the lack of knowledge about evolution of MDE settings by investigating case studies from different companies. The first results reveal (1) that there is evolution that affects the composition of an MDE setting (structural evolution) and has the potential to strongly impact aspects, such as changeability and (2) that this structural evolution actually occurs in practice. Based on these first results, we investigated (3) whether there are cases in practice, where structural evolution already altered the risks of changeability given by the respective MDE setting. Therefore, we search and identify examples for such evolution steps on MDE settings from practice and collected six case studies on evolution histories in detail. As a result, we show in this paper that structural evolution (a) is not seldom in practice and (b) sometimes leads to the introduction of changeability risks.