A service choreography is a distributed service composition in which services interact without a centralized control. Adequate adaptation strategies are required to face complex and ever-changing business processes, given the collaborative nature of choreographies. Choreographies should also be able to adapt to changes in its non-functional requirements, such as response time, and especially for large scale choreographies, adaptation strategies need to be automated and scale well. However, the body of knowledge regarding choreography adaptation approaches has not yet been consolidated and systematically evaluated. By means of a systematic literature review, in which we examined seven scientific paper sources, we identified and analyzed the state-of-the-art in choreography adaptation. We found 24 relevant primary studies and grouped them into six categories: model-based, measurement-based, multi-agent-based, formal-methods-based, semantic-reasoning-based, and proxy-layer-based. We analyzed (i) how each strategy deals with different types of requirements, (ii) what their required degree of human intervention is, (iii) how the different studies considered scalability, (iv) what implementations are currently available, and (v) which choreography languages are employed. From the selected studies, we extracted key examples of choreography adaptation usage and analyzed the terminology they adopted with respect to dynamic adaptation. We found out that more attention has been devoted to functional requirements and automated adaptation; only one work performs scalability evaluation; and most studies present some sort of implementation and use a specific choreography notation.