<p>Elastic Optical Networks (EONs) offer a more efficient solution for handling variable traffic demands compared to traditional wavelength division multiplexing (WDM) networks. However, EONs face critical constraints of spectrum contiguity and continuity, which pose challenges in resource allocation. The inherent variability in link lengths and nodal degrees in physical network topologies leads to uneven link betweenness centrality (LBC) across the network. Links with higher LBC are more prone to congestion, potentially increasing request-blocking and bandwidth-blocking probabilities. This study proposes a novel optimization approach using a non-dominated sorting genetic algorithm (NSGA) to reduce the variance of LBC across network links by determining optimum link costs while preserving the benefits of adaptive modulation. Uniform distribution of traffic due to optimized link costs, the proposed method effectively reduces both request-blocking and bandwidth-blocking probabilities. The efficiency of the approach is validated through simulations on two practical topologies: the German network and NSFNET. The results demonstrate significant performance improvements compared to optimizations of link costs focusing solely on minimizing LBC variance.</p>