Biomechanical and clinical evidence of vertebral metastases to predict the risk of fracture

Cancer is becoming a chronic disease increasing the numbers of cancer survivors which are at risk of developing metastases. Spine is among the most common sites affected by bone metastases. The current patients’ stratification to assess the risk of spinal instability is only partially reliable, leading most patients to be under or over-treated. In order to understand how the metastases are responsible for the altered microstructural and mechanical properties of the metastatic vertebrae, this PhD project focused on a comprehensive experimental biomechanical characterization of the human metastatic spine. This aim was reached investigating the microarchitectural alteration in metastatic vertebrae and assessing the contribute of the metastatic lesions and of the adjacent structures (e.g. IVDs) in the mechanical behaviour of the vertebrae. More than 80 human (metastatic/healthy) vertebrae were scanned with clinical and high-resolution imaging, and were biomechanically tested, in different loading conditions, while experimental strain measurements were performed with Digital Volume Correlation (inside the vertebrae) and Digital Image Correlation (on the surface of the vertebrae and of the adjacent intervertebral discs). This work created an unprecedented experimental dataset essential to identify those features associated with the risk of vertebral failure. The results pointed out that the microstructural alterations within metastatic vertebrae affected the mechanical behaviour differently for each metastatic vertebra and also for the adjacent vertebrae. Additionally, the different role played by degenerated/non-degenerated intervertebral discs in driving the vertebral failure was identified. These findings highlight the need for a more specific clinical stratification system based on digital tools (i.e. Digital Twin, Explainable Artificial Intelligence, mechanistic models). Moreover, the collected data can serve as a benchmark for their initial implementation.