Temporal assessment of bone marrow lesions on magnetic resonance imaging in a canine model of knee osteoarthritis: impact of sequence selection.

OBJECTIVE

To assess the evolution of bone marrow lesions (BMLs) in a canine model of knee osteoarthritis (OA) using three different magnetic resonance imaging (MRI) sequences.

DESIGN

Three MRI sequences [coronal, T1-weighted three-dimensional fast gradient recalled echo (T1-GRE), sagittal fat-suppressed 3D spoiled gradient echo at a steady state (SPGR), and sagittal T2-weighted fast spin echo with fat saturation (T2-FS)] were performed at baseline, and at week 4, 8 and 26 in five dogs following transection of the anterior cruciate ligament. The same reader scored (0-3) subchondral BMLs twice, in blinded conditions, according to their extent in nine joint subregions, for all imaging sessions, and independently on the three MRI sequences. Correlation coefficients and Bland-Altman plots evaluated intra-reader repeatability. Readings scores were averaged and the nine subregions were summed to generate global BML scores.

RESULTS

BMLs were most prevalent in the central and medial portions of the tibial plateau. Intra-reader repeatability was good to excellent for each sequence (r(s)=0.87-0.97; P<0.001). Maximal intra-reader variability (24%) was reached on T2-FS and was associated to higher scores (P<0.05). Global BML scores increased similarly on all three sequences until week 8 (P<0.05). At week 26, score on T2-FS was decreased, being lower when compared to T1-GRE and SPGR (P<0.05).

CONCLUSION

In this canine OA model, the extent of BMLs varies in time on different MRI sequences. Until the complex nature of these lesions is fully resolved, it is suggested that to accurately assess the size and extent of BMLs, a combination of different sequences should be used.