Diagnostic accuracy of imaging methods for the diagnosis of skeletal malignancies: A retrospective analysis against a pathology-proven reference

Objectives: To examine the diagnostic accuracy of imaging modalities in skeletal tumours versus pathology reports.

Materials and methods: Pathology reports of bone biopsies were compared to diagnostic imaging with X-ray, computed tomography (CT), magnetic resonance imaging (MRI), bone scintigraphy (BS), and (18)F-fluorodeoxyglucose positron emission tomography (FDG-PET/CT) performed within 6 months of biopsy.

Results: A total of 409 biopsies were included. Sensitivity and specificity were significantly different among the five modalities (p<0.0001). The sensitivity of MRI and PET/CT was better than CT, but CT had a better specificity than PET/CT. In general, these methods outperformed BS and X-ray. The sensitivity for osteolytic lesions varied significantly between modalities (p<0.0001), with MRI and PET/CT being more sensitive than CT. Differences in sensitivity were also observed in mixed lesions (p=0.0002) but not in osteosclerotic lesions. In spine lesions, MRI showed the best sensitivity followed by PET/CT and CT (p<0.0005 vs. MRI). There was no significant differences among non-spine lesions.

Conclusions: MRI and FDG-PET/CT showed comparable diagnostic characteristics in general, in individual tumour types, and in different bone lesions and locations. Nominally, they outperformed CT in most situations. The diagnostic accuracy of X-ray and BS were notably inferior to other modalities.