Automated CT-based analysis to detect changes in the prevalence of lytic bone metastases from breast cancer

摘要

The spinal column is the most frequent site of bone metastasis in patients with breast cancer. It is important to understand how the pattern of vertebral lesions may be affected by the introduction of modern cancer therapies. The purpose of this study was to characterize changes in the radiological appearance of spinal column metastases over the past decade using highly automated Computed Tomography (CT) based computational analysis methods. Two case series studies were performed using CT scans of patients with confirmed spinal metastases secondary to breast cancer: Cohort A with CT scans acquired between 1998 and 2001 and Cohort B with CT scans acquired between 2004 and 2007. Diseased vertebrae were classified as lytic, blastic, or mixed based on CT scan intensity through an automated 3D computer algorithm. The relative incidence of lytic vertebral metastases decreased in comparing Cohort B to Cohort A (12% vs. 49%) with a corresponding increase in mixed lesions (51% vs. 18%) Significant associations were found between the percentage of lytic lesions in number of diseased vertebrae measured per patient and lack of bisphosphonate use (RR = 2.6) and for membership in Cohort A vs. Cohort B (RR = 5.9). This work highlights a change in the CT appearance of vertebral metastases from breast cancer during the past decade toward a lower proportion of lytic disease. Observation of patient therapies suggests that differences in radiological assessment may be linked, at least in part, to bisphosphonate use. These findings have important implications for both clinical practice and research strategies involving vertebral metastases.