Clinical and experimental imaging of breast cancer metastases in the spine.

An investigation was undertaken to determine the characteristics of spinal breast cancer metastases with respect to MR signal intensity (SI), patho-anatomy and uptake of the bone-seeking radionuclide 18F measured with positron emission tomography (PET). Patients with spinal metastases from breast cancer, or spinal specimens, were examined with MRI and the results were correlated to histopathological findings, or they were examined with conventional radiography and CT in correlation with cryomicrotomical images, with CT and dynamic 18F-PET; or with MRI, CT, skeletal scintigraphy and conventional radiography, compared with one another. Metastases were detected in all anatomical parts of the vertebrae. The areas with bone marrow replacement by tumour were larger in the cryosectional images than was apparent on CT. Metastases were often in contact with the vertebral cortex or end-plates, and fractures occurred in destructive lesions. Neurovascular compromise was detected only at few levels and was caused by vertebral collapse rather than epidural tumour growth. Metastases (in vivo) displayed low SI on T1-weighted, low or intermediate SI on proton density-weighted, and high or intermediate SI on T2-weighted and "phase contrast" images, except for highly sclerotic metastases, which showed low SI on all sequences. Sensitivity was high and specificity limited since connective tissue in the vertebrae and bone marrow with high cellularity had similar SI. Both MRI and CT were more sensitive than conventional radiography and skeletal scintigraphy for revealing metastases in the cervical spine. On PET there was an increased uptake of 18F in metastases, both in osteosclerotic lesions and in osteolytic defects in the bone. Post-mortem MR examinations showed different SIs than MRI in vivo. The T1-and T2-relaxation times and SI were dependent on tissue temperature. Reversal of contrast between some tissues occurred at 5 degrees C in T1-weighted images.