Accuracy of CT and MRI to assess resection margins in primary malignant bone tumours having histology as the reference standard

Abstract

Aim: To evaluate the accuracy of magnetic resonance imaging (MRI) and computed tomography (CT) in assessing the resection margins of primary malignant bone tumours.

Materials and methods: Resected primary malignant bone tumour specimens removed from 46 patients (27 male; mean age: 48±22 years) were imaged using MRI (fat-saturated proton density-weighted and three-dimensional fat-suppressed T1-weighted gradient-recalled-echo) and CT immediately after surgery. A radiologist and an orthopaedist evaluated bone and soft-tissue margins of the specimens on both examinations. Histological evaluation was performed by a senior orthopaedic oncology pathologist. Margins were classified as R0 (safe margins), R1 (residuals between 0 and 1 mm), and R2 (macroscopic residuals). Cohen's k, chi-square, and McNemar's statistics were used.

Results: Having histology as the reference standard, reproducibility of the radiologist ranged from moderate (k=0.544) to substantial (k=0.741) for bone and soft-tissue margins on CT, respectively, while that of the orthopaedist ranged from fair (k=0.316) to moderate (k=0.548). When comparing R2 and R0+R1 scores, reproducibility of readers' evaluation of bone margins increased ranging from substantial (k=0.655) to perfect (k=1.000). Inter-reader agreement ranged from fair (k=0.308) to substantial (k=0.633). Accuracy of the radiologist and orthopaedist ranged from 76% to 83% and from 68% to 72%, respectively. When comparing R2 and R0+R1 scores, the accuracy of both readers ranged from 83% to 100%. There was no association between local recurrence and margin scores of histology, MRI, and CT (p≥0.058).

Conclusions: MRI and CT may be useful for extemporaneous analysis of resection margins of primary malignant bone tumours, although wide accuracy variability between the different imaging methods was observed.