Investigating spinal biomechanics and mechanobiology to predict disease progression and treatment effects in oncology patients.

Primary cancers, such as prostate cancer and multiple myeloma (MM), frequently metastasise to the skeleton, disrupting bone remodelling, reducing vertebral strength, and increasing fracture risk. As treatments extend survival, more patients are living with structurally compromised vertebrae. To assess the vertebral strength in cancer patients using clinical CT scans, a finite element (FE) pipeline was first developed. Its development highlighted the impact of image quality on segmentation reproducibility and showed comparable results between phantom and phantomless calibration methods.

In Chapter 4, the effect of androgen deprivation therapy (ADT) on vertebral strength in prostate cancer patients was investigated using a combination of DXA-derived areal bone mineral density (aBMD), CT-derived volumetric BMD (vBMD), and mechanical properties from the FE pipeline. Significant reductions in densitometric and mechanical properties were found after 12 months of ADT. Furthermore, vBMD outperformed aBMD in predicting vertebral strength, suggesting that these clinically accessible measures may better assess fracture risk.

Chapter 5 applied the FE pipeline to MM patients treated non-surgically. Changes in CT derived densitometric and FE-derived mechanical properties over time were quantified. In vertebrae with large lesions (>50% of the vertebral body), remineralisation around the lesion was observed, resulting in increased strength. To investigate the mechanisms behind this, a mechanobiological model was developed in Chapter 6. This multi-scale model combined organ-level FE simulations with a cell-level bone adaptation algorithm and was optimised for each patient. While the model accurately predicted mineral and strength changes in vertebrae with smaller lesions, it failed in vertebrae with extensive lesions, suggesting that standard bone remodelling principles are insufficient in myeloma-affected bone. Additionally, its reliance on longitudinal scans limited its prospective use.

To enable prospective use, Chapter 7’s pilot study incorporated serum bone turnover markers at 1, 2, and 3 months into the mechanobiological model, allowing for successful prediction of 12-month mechanical outcomes in 3 out of 5 patients from baseline imaging alone. This biomarker-informed model marks the first use of biomarkers to drive personalised vertebral strength prediction. Together, these innovations offer a foundation for improved treatment monitoring, early fracture risk prediction, and decision-making in metastatic bone disease.