Differentiating high-grade from low-grade chondrosarcoma with MR imaging

Abstract

The purpose of the study was to evaluate the MR imaging features that differentiate between low-grade chondrosarcoma (LGCS) and high-grade chondrosarcoma (HGCS) and to determine the most reliable predictors for differentiation. MR images of 42 pathologically proven chondrosarcomas (28 LGCS and 14 HGCS) were retrospectively reviewed. There were 13 male and 29 female patients with an age range of 23-72 years (average age 51 years). On MR images, signal intensity, specific morphological characteristics including entrapped fat, internal lobular architecture, and outer lobular margin, soft tissue mass formation and contrast enhancement pattern were analysed. MR imaging features used to identify LGCS and HGCS were compared using univariate analysis and multivariate stepwise logistic regression analysis. On T1-weighted images, a central area of high signal intensity, which was not seen in LGCS, was frequently observed in HGCS (n = 5, 36%) (p < 0.01). Entrapped fat within the tumour was commonly seen in LGCS (n = 26, 93%), but not in HGCS (n = 1, 4%) (p < 0.01). LGCS more commonly (n = 24, 86%) preserved the characteristic internal lobular structures within the tumour than HGCSs (n = 4, 29%) (p < 0.01). Soft tissue formation was more frequently observed in HGCS (n = 11, 79%) than in LGCS (n = 1, 4%) (p < 0.01). On gadolinium-enhanced images, large central nonenhancing areas were exhibited in only two (7.1%) of LGCS, while HGCS frequently (n = 9, 64%) had a central nonenhancing portion (p < 0.01). Results of multivariate stepwise logistic regression analysis showed that soft tissue formation and entrapped fat within the tumour were the variables that could be used to independently differentiate LGCS from HGCS. There were several MR imaging features of chondrosarcoma that could be helpful in distinguishing HGCS from LGCS. Among them, soft tissue mass formation favoured the diagnosis of HGCS, and entrapped fat within the tumour was highly indicative of LGCS.