Accuracy of magnetic resonance imaging for estimating intramedullary osteosarcoma extent in pre-operative planning of canine limb-salvage procedures

The objective of this work was to compare the accuracy of radiographs and magnetic resonance imaging (MRI) for estimating appendicular osteosarcoma margins. The accuracy of computed tomography (CT) and bone scintigraphy was also assessed when these studies were available. Eight dogs with appendicular osteosarcoma underwent radiographic and MRI of affected limbs. In addition, bone scintigraphy was performed in six dogs and CT examination was performed in five dogs. Two observers jointly measured tumor length on all imaging studies. Correlative gross and histologic evaluation of all affected limbs was performed to determine tumor extent as measured from the nearest articular surface. Results from imaging studies were compared to gross and microscopic morphometry findings to determine the accuracy of each modality for determining tumor boundaries. MRI images were accurate with a mean overestimation of actual tumor length of 3 ± 13%. T1-weighted non-contrast images were superior in identifying intramedullary tumor margins in most instances whereas contrast-enhanced images provided supplemental information in two dogs. Lateromedial and craniocaudal radiographs overestimated tumor length by 17 ± 28% and 4 ± 26%, respectively. Scintigraphy and CT overestimated tumor margins by 14 ± 28% and 27 ± 36%, respectively. MRI appears to be an accurate diagnostic imaging modality in determining intramedullary osteosarcoma boundaries. MRI should be considered as part of a pre-operative assessment of appendicular osteosarcoma, particularly when a limb-sparing procedure is contemplated.