Can imaging criteria distinguish enchondroma from grade 1 chondrosarcoma?

To minimize systematic bias and optimize agreement on imaging criteria in order to better define the accuracy of imaging criteria in the diagnosis of grade 1 chondrosarcoma.

Study was IRB-approved and HIPAA compliant; informed consent was waived. Records were reviewed and disclosed 53 cases (38 women, 15 men ages 21-76) which were diagnosed as enchondroma or grade 1 chondrosarcoma and had available radiographs, contrast-enhanced MRI, and definitive diagnosis by histology or 5-year follow-up. 2 MSK radiologists read the studies independently after a session where they agreed on criteria for malignancy. Interobserver variability was determined as raw variability and with the kappa statistic. Accuracy was determined compared to final diagnosis. Reliability of imaging features of chondrosarcoma was determined using regression analysis.

The correct diagnosis of enchondroma was made on radiographs in 43 (67.2%) of readings, and on MRI in 37/64 (57.8%). The correct diagnosis of chondrosarcoma was made on radiographs in 5/24 (20.8%) of readings, and on MRI in 14/24 (57.8%). A diagnosis of borderline lesion was made in 19/64 (29.7%) of enchondromas on radiographs and 18/64 (28.1%) on MRI. The false positive rate of radiographs for chondrosarcoma was 2/64 (3.1%) and the false positive rate of MRI was 9/64 (14.1%). There was substantial interobserver variability. Cortical thickening and bone expansion were rare but specific signs of chondrosarcoma.

Both radiographs and MRI have limitations in the evaluation of low-grade cartilage lesions. MRI has an increased rate of both true-positive and false-positive diagnosis compared to radiographs. Differences in the findings of this study compared to previous literature may reflect the influence of systematic biases.