Large acetabular defects can be treated effectively through the use of acetabular cages combined with bone grafts, with the formation of living bone facilitated by mechanical stimulus. The mechanical stimulus on the graft is highly dependent on the design of the acetabular cage. Topology optimization offers a means to create conceptual designs of acetabular cages, which can then be assessed for their impact on strain energy density (SED) distribution within the graft. This study aims to compare various acetabular cage designs generated through multiple optimization constraints, with a focus on analyzing the SED distribution in the graft. A virtual bone defect was modeled, and a graft was virtually implanted within it, followed by the creation of a design space for the acetabular cage. Different acetabular cup designs were then generated using volume minimization as the objective function, along with varying displacement constraints, namely the global displacement of the cage center or its relative displacement to the pelvis constrained. Linear static simulations were performed on the hemipelvis model, and the results were filtered using different relative densities to evaluate SED distribution in the graft. The results revealed that the acetabular cage designs produced qualitatively similar strain energy density distributions. Both types of optimization are worth using because together they were able to reduce the element-wise average SED relative errors to below 14%. The model using relative displacement constraints can produce more diverse acetabular cage variants than the model with global displacement constraints.  Kulcsszavak: acetabular cage, topology optimization, finite element method, graft transformation