Search

Pathologic vertebral compression fractures (PVCF) cause significant morbidity in patients with bone metastases from breast cancer and other malignancies. Due to limitations of existing biochemical and imaging biomarkers, clinicians currently have no reliable metrics to identify patients with impending PVCF, impeding efforts to prevent this severe complication. To establish the feasibility of a new method for defining risk of PVCF, we retrospectively analyzed serial CT scans from five breast cancer patients using parametric response mapping (PRM) to quantify dynamic bone density changes that preceded an event. Vertebrae segmented from each scan were registered to vertebrae at the earliest time point (i.e. furthest from PVCF) and voxel classification accomplished using a predetermined threshold of change in HU values, resulting in relative volumes of increased (PRM HU+ ), decreased (PRM HU- ), or unchanged (PRM HU0 ) attenuation. A total of seven PVCF were compared to un-diseased vertebrae in each patient serving as controls. Receiver operator curve (ROC) analysis identified optimal image acquisition and analysis times for group stratification. Bone density changes were visualized by an increasing trend in PRM HU+ as early as one year before fracture. PRM HU- demonstrated negligible changes over the course of the study. These observations were consistent with ROC results, showing poor performance of PRM HU- in stratifying PVCF versus control. As early as 6 months prior to PVCF, PRM HU+ was significantly larger (12.9 ± 11.6%) compared to control vertebrae (2.3 ± 2.5%), with an AUC of 0.918 from a receiver operator curve analysis. Mean HU changes were also significant between PVCF (+26.8 ± 26.9%) and control (-2.2 ± 22.0%) over the same period. PRM analysis of bone density changes using standard CT imaging was sensitive for spatially resolving bone remodeling which preceded structural failure in patients with breast cancer vertebral metastases.