DCE-MRI-DERIVED TISSUE PERFUSION IS DIFFERENTIALLY ASSOCIATED WITH TIBIAL AND FEMORAL BONE DENSITY IN PEOPLE WITH EARLY KNEE OSTEOARTHRITIS.
Authors
Affiliations (4)
Affiliations (4)
- Schroeder Arthritis Institute, University Health Network, Toronto, Canada; Joint Department of Medical Imaging, University Health Network, Toronto, Canada.
- Schroeder Arthritis Institute, University Health Network, Toronto, Canada; Joint Department of Medical Imaging, University Health Network, Toronto, Canada; Division of Epidemiology, Dalla Lana School of Public Health, University of Toronto, Toronto, Canada.
- Joint Department of Medical Imaging, University Health Network, Toronto, Canada.
- Schroeder Arthritis Institute, University Health Network, Toronto, Canada; Joint Department of Medical Imaging, University Health Network, Toronto, Canada; Division of Epidemiology, Dalla Lana School of Public Health, University of Toronto, Toronto, Canada. Electronic address: [email protected].
Abstract
Bone mineral density (BMD) and structure of the knee have been implicated in the pathogenesis of knee OA, but it remains unclear how it is affected by local periarticular inflammation. Dynamic contrast-enhanced (DCE)-MRI quantifies tissue-level perfusion kinetics that may reflect inflammation in knee OA. However, the quantification of periarticular tissue perfusion remains challenged by the lack of automated tools, limiting the ease of bone-inflammation investigations. To understand how periarticular perfusion (between, within and just around the tibia and femur) relates to bone structural and densitometric properties in the knee joint in individuals with knee OA. This study included females and males 40-85 years old with knee pain that occurred ≥ 3 times a week, each occurrence > 3 hours. DCE-MRI scans were obtained using a 3.0T MRI with a quadrature knee coil following intravenous administration of 0.1 mmol/kg Gd-DTPA at 2 mL/s. Sagittal T1-weighted fast spoiled gradient echo sequences (TR/TE = 8/3 ms, flip angle = 13°, voxel size: 0.8 × 0.8 × 3.0 mm, 0 mm gap) were obtained at 5.1s temporal resolution for a total of 7.5 minutes. The referent arterial input function (AIF, represented by the femoral artery) was identified by selecting three consecutive slices demonstrating the greatest total amount of contrast enhancement as determined across the full timespan. The periarticular region, excluding the AIF, was represented by combining femoral and tibial bone masks generated using a deep learning model (nnUNet), and dilated for connectivity. Contrast enhanced regions within this periarticular mask were identified by selecting top 5% of voxel intensities, verified by visual inspection (Figure 1). Arterial (C<sub>p</sub>) and tissue (C<sub>t</sub>) contrast concentrations over time were measured within the AIF and periarticular contrast regions, respectively. These values were fitted to a standard Toft's model to estimate the rate of contrast perfusion from the plasma to extracellular space (K<sup>trans</sup>, volume transfer constant, relative units min<sup>-1</sup>), representing periarticular inflammation. Volumetric(v)BMD and apparent microstructure were obtained from four separate transaxial peripheral(p) QCT (38 kVp, voxel size: 0.2 × 0.2 × 2.3 mm) scans, each aligned against medial and lateral, femoral and tibial compartments, including total(Tt) and trabecular(Tb).vBMD, and trabecular separation (Tb.Sp), all derived from segmentations achieved by an iterative threshold-based analysis with Laplacian filtering. Associations between periarticular K<sup>trans</sup> and bone parameters were evaluated using multivariable linear regression, adjusting for age, body mass index (BMI), KLG, glucocorticoid, and antiresorptive medication use. Among 60 participants (mean age: 61.6(SD:8.0) yrs, BMI of 23.0 (SD 3.4), and 15 (25%) participants had a KLG ≥ 2). In the adjusted multivariable model, a 0.1/min higher periarticular K<sup>trans</sup> was associated with lower Tt. and Tb.vBMD in the medial and lateral femur; but higher Tt. and Tb.vBMD in the medial and lateral tibia (Table 1). Effect sizes in the medial compartments were in general larger than the lateral; but for lateral Tb. analyses, were not statistically significant. Although higher periarticular perfusion related to larger separation in trabeculae within the medial tibia, the effect was generally weak. Higher periarticular inflammation may be associated with compromised femoral bone density but higher tibial bone density among individuals with early knee OA and established symptoms, suggesting site-specific relationships. Longitudinal studies are needed to establish temporality and whether accumulated inflammation contributes to progressive bone deterioration.