Comparison of bone marrow cellularity measurements using proton nuclear magnetic resonance spectroscopy (H-NMRS) and histology performed at the same location on bone

Objectives Bone marrow (BM) toxicity limits the therapeutic dose in radionuclide therapy patients. Calculation of BM dose requires BM mass, which is unknown in a patient. We have previously developed equations to predict BM mass in adult patients given a few skeletal pelvic measurements (Pichardo et al 2007). These equations presently rely on average and incomplete bone marrow cellularity (BMC) values published in ICRP Publication 70 (1995) that are most likely not representative of a given patient. H-NMRS provides a non-invasive method for measuring BMC in a patient, but its accuracy in vivo remains unclear. In vivo accuracy requires a comparison of BMC measured by H-NMRS to BMC measured from histology at the same site in the bone. Since bone extraction is required, this analysis cannot be performed on live humans. We propose using limbs of dogs undergoing amputation as treatment for osteosarcoma.

Methods Dog limbs were scanned prior to surgery and histology was performed at the same site on the amputated limb. We measured T1 and T2 for BM water and lipid using standard SE sequences with water and lipid suppression. BMC was calculated as one minus the lipid fraction, where the lipid fraction is given by the ratio of the lipid signal amplitude to the sum of the lipid and water amplitudes at echo, accounting for relaxation.

Results Accuracy and precision were characterized using Bland-Altman plots.

Conclusions Extensions to human studies are discussed.