In real-world social systems, individual interactions are frequently shaped by reputation, which not only influences partner selection but also affects the nature and benefits of the interactions themselves. We propose a heterogeneous game transition model that incorporates a reputation-based dynamic threshold mechanism to investigate how reputation regulates game evolution. In our framework, individuals determine the type of game they engage in according to their own and their neighbors' reputation levels. In turn, the outcomes of these interactions modify their reputations, thereby driving the adaptation and evolution of future strategies in a feedback-informed manner. Through simulations on two representative topological structures, square lattice and small-world networks, we find that network topology exerts a profound influence on the evolutionary dynamics. Due to its localized connection characteristics, the square lattice network fosters the long-term coexistence of competing strategies. In contrast, the small-world network is more susceptible to changes in system parameters due to the efficiency of information dissemination and the sensitivity of strategy evolution. Additionally, the reputation mechanism is significant in promoting the formation of a dominant state of cooperation, especially in contexts of high sensitivity to reputation. Although the initial distribution of reputation influences the early stage of the evolutionary path, it has little effect on the final steady state of the system. Hence, we can conclude that the ultimate steady state of evolution is primarily determined by the reputation mechanism and the network structure.