Preclinical drug testing commonly relies on the use of two-dimensional (2D) cultures, which allow for rapid drug screening in vitro. However, 2D cultures are unable to capture the complexity of the native three-dimensional (3D) tumor microenvironment, resulting in a lack of correspondence between preclinical data and clinical trial outcomes. The establishment of high-throughput 3D models able to describe distinctive aspects of the tumor niche would advance our understanding of tumor biology and would allow for drug testing in a physiologically relevant setup. Along this rationale, this thesis focuses on the development of a high-throughput 3D tumor model of bone sarcoma based on tissue-engineered polymeric scaffolds in combination with a flow perfusion bioreactor. First, we investigated the effects of flow perfusion on a 3D culture of Ewing sarcoma (ES) cells. We found that increasing levels of flow-derived shear stress promoted the secretion of insulin-like growth factor-1 (IGF-1) which, in turn, resulted in shear stress-dependent cell sensitivity to the IGF-1 receptor (IGF-1R) blockade, a central player in ES progression. We then leveraged these findings to culture ES cells on 3D-printed scaffolds under flow perfusion conditions. By designing 3D scaffolds with a defined porosity gradient, ES cells were exposed to a shear stress gradient that resulted in a gradient in cell response. In this way we sought to model variable levels of shear stress present within ES tumors due to intratumoral heterogeneity. In the final part of this thesis we investigated how the simultaneous presence of mesenchymal stem cells (MSCs) and flow perfusion affected drug sensitivity and phenotype of ES cells. We showed that the presence of MSCs within the coculture induces a progressive inhibition of cell growth and resistance to the IGF-1R blockade, highlighting the role of mechanically-sensitive mesodermal stroma on ES drug resistance. Overall, in this thesis we present a tissue-engineered tumor model that reliably mimics key features of the bone microenvironment, specifically the effects of biomechanical stimulation and of tumor-stroma interactions. The model hereby developed is amenable to further mechanistic studies on tumor biology and allows for a more accurate high-throughput screening of novel drug candidates.