In this paper, we propose a distributed load balancing algorithm with adaptive bias adjustment for LTE/LTE-A HetNets. We have formulated the problem as a local sum utility maximization. The distributed algorithm is capable of fairly distributing the load among base stations with a minimum level of coordination. The performance of the algorithm was evaluated with a network topology subject to the conditions of a real HetNet deployment, in an urban environment, and applying a realistic traffic distribution during a period of peak usage. Our simulation results show that 350% higher data rate can be achieved for users with the worst 5% rates, and an overall average data rate gain of 23% can be obtained. Furthermore, the load balancing algorithm applied in combination with our adaptive bias adjustment method, provides an additional average gain of 50% in the rates for the 25th percentile.