Inter-Cell Interference Coordination (ICIC) techniques are proposed as solutions to alleviate the negative impact of interference on system performance, while enhancing the provided Quality of Service (QoS). Typically, the available bandwidth is divided into inner and edge sub-bands. Users are also classified into interior and edge users. The available resources in each zone are exclusively allocated to users belonging to this zone. Mobile users classification is usually based on a threshold that can be either a given mean SINR value or a given distance. However, ICIC approaches based on these static parameters cannot efficiently manage non-homogeneous distribution of users. In this paper, we introduce a dynamic handoff algorithm that aims to adapt static ICIC schemes to uneven distribution of users. Our new solution dynamically computes the classification of active users into interior and edge users, based on a heuristic load balancing algorithm. In our proposal, each cell autonomously reconfigures its bandwidth allocation constraints without modifying bandwidth repartition across the cellular network. This makes the solution well adapted to the non-uniform repartition of users at the multicell scale. Simulation results show that the proposed scheme improves bandwidth usage, reduces packet delay, and increases user satisfaction compared to state-of-the-art ICIC techniques.