Session Type: Abstract Submissions (ACR)
Background/Purpose: Peri-articular fractal signature analysis (FSA) and bone mineral density (paBMD) are potentially cost-effective prognostic imaging markers for osteoarthritis (OA) progression. While we know that paBMD relates to trabecular morphology, it remains unclear whether FSA, which assesses the texture of bone on radiographs, measures a similar construct of bone structure as paBMD. Understanding this association is paramount for determining if these two measurements are redundant. Therefore, we evaluated the association between FSA, paBMD, and trabecular morphology.
Methods: The sample comprised 450 participants in the Osteoarthritis Initiative (OAI) progression cohort who had weight-bearing radiographs, magnetic resonance imaging (MRI), and knee dual-energy x-ray absorptiometry (DXA) at the 48-month OAI visit. The right knee was assessed unless MRI was contraindicated. Three separate readers measured the medial proximal tibia to generate 1) paBMD on DXA, 2) fractal dimensions with FSA on digital radiographs, and 3) MRI-based apparent trabecular morphology: bone volume fraction (aBV/TV), trabecular number (aTb.Tn), spacing (aTb.Sp), and thickness (aTb.Th). Fractal dimension data in the horizontal “tension” and vertical “compression” directions were reduced into 3 piece-wise, centered linear regression from 0 to 3mm radii. We assessed univariate associations with Spearman correlations and conducted 5 robust regression models to determine the association between FSA-based data (compression and tension together; predictors) and the five other bone measures (paBMD and trabecular morphology) as outcomes. These models were adjusted for OAI clinical site.
Results: The 450 participants were 47% female, 72% radiographic knee OA in the study knee (Kellgren-Lawrence > 2), and on average 66 (SD=9) years of age with a mean body mass index of 29.6 (SD=4.7) kg/m2. Table 1 shows that FSA-based data were not associated with paBMD. The significant associations between FSA-based data and trabecular morphology indicated that lower tension and compression gradients may be related to greater aBV/TV, aTb.Th, aTb.Tn, but lower Tb.Sp (Table 1). In the multivariate analyses we found that FSA-based compression offsets had a small relationship with paBMD (R2 = 0.03). Furthermore, we found that lower FSA-based compression gradients were associated with greater aBV/TV (p=0.04) and aTb.Th (p=0.01) but lower aTb.Sp (p=0.01; R2 = 0.04-0.05). Lower FSA-based tension gradients were associated with greater aTb.Th (p= 0.02) but lower aTb.Sp (p=0.04).
Conclusion: We found that lower peri-articular FSA values were not associated with paBMD but had some small associations with trabecular morphology that is typical of greater OA severity (greater aBV/TV, aTb.Th, and aTb.N and lower aTb.Sp). This suggests that these are not redundant measures and both imaging markers warrant further evaluation.