Background While the association of BMLs with joint space narrowing and pain in knee OA is verified, the correlation between BMLs and local BMD is a topic of major interest with a potential to strongly improve the understanding of the pathogenesis of the disease.
Objectives To investigate whether trabecular BMD in BMLs is increased compared to the surrounding tissue.
Methods 18 BMLs (WORMS grade ≥1) detected on coronal turbo spin echo MR images of 14 patients with knee OA (KL grade 2 or 3) have been automatically 3D segmented (left figure, BML). To compare these volumes of interest (VOI) with the circumjacent lesion-free trabecular bone tissue, a neighboring VOI was defined by a dilation of the BML VOI (left figure, Neighborhood). Via a 3D multimodality registration between MR and CT datasets of the same patient both VOIs were transferred to the CT datasets (right figure). Mean BMD values were calculated in both VOIs and their difference was determined for each BML. To account for the preferred occurrence of BMLs in proximity of the bone surface and the presence of a BMD gradient in the epiphysis, reference CT datasets (reference group) of patients with knee OA without BMLs were used. In these datasets, a BMD analysis was performed in VOIs corresponding to the BML and Neighborhood VOIs in the patients with BMLs (BML group). Further, the effect of a change from fatty to water equivalent tissue within the BML VOI was estimated. This effect has to be considered due to dependence of single energy QCT accuracy on fat concentration.
Results Despite different spatial resolutions between MR and CT, the registrations showed excellent overlap. %BMD differences between BML- and neighborhood-VOIs were significant in both groups, but differences were significantly higher (p=0.01) in the BML (37.6%±32.6%) compared to the reference group (11.9%±22.2%). Under the assumption of a maximum 29% contribution of water equivalent material in the BML group, the %BMD difference would decrease to 30.1±30.7%, which would still be significantly different from the reference group. (Assumptions: fat value of -100 HU and 68% fat in the BML VOI (80% fat in 85% yellow marrow) of the reference group).
Conclusions Highly accurate segmentation and registration methods ensure the precise investigation of BMD at BML locations and circumjacent bone regions. Locally increased BMD at the BML locations seems to be present in patients with knee OA. However, the effect of compositional changes within the BML needs to be further discussed as a high content of water equivalent material may impair the BMD results. The outcome may reflect increased loading conditions within the knee joint. In the presented study, the benefits of both MR and CT acquisitions regarding OA changes in soft tissue and bone, respectively, were utilized to gain information about the pathogenesis of knee OA and therefore help to improve its diagnosis.
Disclosure of Interest None Declared