WHY TECHNICAL INNOVATIONS ARE NEEDED TO FACILITATE BETTER CHARACTERIZATION OF PATHOLOGICAL FRACTURES IN METASTATIC SPINAL DISEASE

Due to an ageing population and improved oncological treatments, the number of patients living with metastasized cancer is increasing. After lungs and liver, the spinal column is most often affected by metastases. The natural course of spinal metastases is often insidious and may over time lead to pain, vertebral fractures, and neurological deficits. These symptoms collectively are known as metastatic spinal disease (MSD) and have a strong impact on quality of life. The treatment for MSD is dependent on the clinical stage and patient condition/wishes. Isolated pain in MSD can, for example, effectively be treated by palliative radiotherapy while neurological deficits, the final consequence of MSD, generally require urgent and highly demanding decompressive surgery. The appropriate treatment for (pending) vertebral fractures without neurological deficits depends on clinical symptoms, current weight-bearing capacity, and potential cause for future vertebral collapse, spinal deformity and development of new neurological deficits. As such there is a need to better understand vertebral morphology and biomechanics in MSD to predict these events and hopefully prevent them by intervening in time and with the appropriate measure. The spinal instability neoplastic score (SINS score) was developed by spine experts for non-experts to assess the degree of spinal instability in MSD and help timely referral to a spine surgeon. Although the SINS score has been proven extremely useful as a referral tool for patients with evidence of spinal instability, it has only been moderately successful in stratifying (pending) fractures into categories with specific treatments. Currently, patients with vertebral fractures are at risk of being under- or overtreated as no reliable prediction system is available to help guide healthcare professionals in choosing the optimal treatment. Signs of a pending or progressive vertebral fracture are often present on CT/MRI images but the biomechanical implications of these radiographical findings are difficult to interpret and quantify as the variability in presentation and subsequent outcome is typically large. Finding patterns in the radiographic and clinical data and correlating these patterns to biomechanical characteristics, with the end goal of stratifying patients into distinct treatment categories has, until now, proven extremely difficult. It is hypothesized that a model combining artificial intelligence with physiology-based spinal biomechanics may be able to use radio-graphical and clinical data to stratify individual MSD patients-at-risk and help support doctors in making better treatment decisions. Acknowledgements: METASTRA project funded by the European Union HEU topic HLTH-2022-12-01 grant 101080135