Complex three-dimensional tumor models to study and treat sinonasal cancers

This PhD project focused on developing advanced in vitro tumor models to overcome the limitations of conventional culture systems in reproducing the complexity of human solid tumors. The aim was to engineer physiologically relevant models capable of replicating key features of the tumor microenvironment, enabling patient-specific approaches for precision medicine. Special attention was given to rare and aggressive sinonasal (SN) cancers. SN osteosarcoma (KHOS) was cultured in mono-, co-, and tri-culture systems with mesenchymal stromal cells (MSCs) and endothelial cells (HUVECs). Models progressively transitioned from 2D to 3D systems, including spheroids and PVA/gelatin porous scaffolds, with the evaluation of different chemotherapeutic agents. Tri-culture PVA/gelatin constructs emerged as the most biomimetic model, showing increased viability and chemoresistance. Additional SN tumor types (ITAC, NC4, TCS627) were studied in PVA/gelatin 3D scaffolds, demonstrating improved differentiation and tumor-specific gene expression compared to 2D counterparts. Finally, SN cancer constructs were integrated into a dynamic microfluidic cancer-on-a-chip platform to replicate physiological stimuli. Dynamic cultures showed higher predictive performance than static cultures, with perfusion optimized for tissue-specific environments. The platform was further extended to hypopharynx, lung, pancreas, colon, and bone cancers, all displaying native-like behavior. Overall, this work shows the transition from 2D systems to advanced 3D biomimetic tumor models. The microfluidic platform proved to be the most versatile and effective microphysiological system for tumor modeling and preclinical drug screening, enabling predictive, patient-specific strategies in precision oncology.