I am not familiar with boost implementation, but as far as I am concerned, the algorithm finds the maximum clique in a greedy way. Hence to find subgraph with clique number k one have to find a subgraph with clique number k-1. So, the algorithm "knows" about cliques you are looking for.

As a workaround you can change the implementation to make it capable of enumerating all cliques in a graph. I dont think it is a hard task.