In this paper, we consider the probabilistic channel contention resolution problem for net revenue maximization in multi-hop wireless ad-hoc networks (MHAHNs) under collision-rate-constrained opportunistic spectrum access (OSA) approach. Specifically, we focus on the interference-dependent contention model, in which secondary users (SUs) must coordinate to each other to simultaneously balance between interference and collision, leading a more efficient MAC protocol than the location-dependent one proposed in the literature. By introducing some auxiliary variables and noisy channel estimations, we can then develop a novel heuristic cross-layer cognitive MAC protocol (HCC-MAC) in OSA-based MHAHNs to solve the formulated MAC optimization problem which is shown non-convex and inseparable. More importantly, our proposed protocol can achieve near-optimal throughput in a distributed manner without control overhead. Finally, the numerical results show that HCC-MAC can outperform the existing MAC protocols under OSA paradigm.