Cooperative spectrum sensing is an efficient solution to mitigate the primary hidden terminal problem through the involvement of the sensing result for cooperative users. Clustering is introduced to tackle the degradation in the performance of cooperative spectrum sensing caused by fading and shadowing of reporting channel, and also to reduce the control channel overhead when the number of cooperative users becomes very large. In practice, some cluster heads might be far away from the fusion centre, which leads to more energy consumption for reporting their results to the fusion centre, especially in large-scale cognitive radio networks. Furthermore, this case also may leads to a degradation of sensing performance. However, existing studies have not focused on these problems. In this paper, we propose a multi-hop cluster-based cooperative spectrum sensing scheme to reduce the power consumption, which also improves the sensing performance. The simulation results show that our algorithm can achieve better energy gains than conventional cooperative spectrum sensing algorithms.