Vertebral fractures near the upper end of long-segment spinal fusions represent a clinically significant problem, driven by complex load redistribution and the highly heterogeneous microstructure of vertebral bone. Understanding the underlying mechanisms requires numerical models that integrate organ-scale biomechanics with high-resolution imaging data. Such models must capture geometric complexity, material heterogeneity, and damage-related processes. This talk presents ongoing work toward a computational framework for vertebral analysis based on the Finite Cell Method (FCM). The method employs CT and micro-CT data to represent trabecular architecture in a voxel-based manner. By embedding this geometry into a structured high-order discretization, FCM avoids labor-intensive meshing and enables detailed mechanical analysis even in regions with highly irregular topology, such as around screw channels or thin trabecular plates. The presentation will outline current development steps, including image segmentation, voxel aggregation, and the setup of FCM vertebral models. The methodology emphasizes efficient numerical integration and solution strategies to handle the large, complex systems arising from voxel-based FCM discretizations. In particular, advanced integration techniques such as pre-integration combined with non-negative moment fitting (NNMF) are employed to accelerate computations, including nonlinear material behavior. Furthermore, numerical homogenization is used to derive effective anisotropic material properties from the vertebral microstructure, which can then be transferred to macroscopic finite element models. An adaptive selection of representative volume elements balances computational cost and accuracy. This hierarchical multiscale approach enables a spatially resolved material description at the organ scale, improving the fidelity of structural analyses and enhancing mechanistic understanding of vertebral load-bearing behavior.