Imaging Bone-Cartilage Interactions in Osteoarthritis Using [18F]-NaF PET-MRI

Abstract

Purpose: Simultaneous positron emission tomography-magnetic resonance imaging (PET-MRI) is an emerging technology providing both anatomical and functional images without increasing the scan time. Compared to the traditional PET/computed tomography imaging, it also exposes the patient to significantly less radiation and provides better anatomical images as MRI provides superior soft tissue characterization. Using PET-MRI, we aim to study interactions between cartilage composition and bone function simultaneously, in knee osteoarthritis (OA).

Procedures: In this article, bone turnover and remodeling was studied using [¹⁸F]-sodium fluoride (NaF) PET data. Quantitative MR-derived T_1ρ relaxation times characterized the biochemical cartilage degeneration. Sixteen participants with early signs of OA of the knee received intravenous injections of [¹⁸F]-NaF at the onset of PET-MR image acquisition. Regions of interest were identified, and kinetic analysis of dynamic PET data provided the rate of uptake ( K_i) and the normalized uptake (standardized uptake value) of [¹⁸F]-NaF in the bone. Morphological MR images and quantitative voxel-based T_1ρ maps of cartilage were obtained using an atlas-based registration technique to segment cartilage automatically. Voxel-by-voxel statistical parameter mapping was used to investigate the relationship between bone and cartilage.

Results: Increases in cartilage T_1ρ, indicating degenerative changes, were associated with increased turnover in the adjoining bone but reduced turnover in the nonadjoining compartments. Associations between pain and increased bone uptake were seen in the absence of morphological lesions in cartilage, but the relationship was reversed in the presence of incident cartilage lesions.

Conclusion: This study shows significant cartilage and bone interactions in OA of the knee joint using simultaneous [¹⁸F]-NaF PET-MR, the first in human study. These observations highlight the complex biomechanical and biochemical interactions in the whole knee joint in OA, which potentially could help assess therapeutic targets in treating OA.

Keywords: PET-MRI; T1ρ relaxation times; bone remodeling; bone–cartilage interactions; cartilage degeneration; knee osteoarthritis; multimodality imaging; sodium fluoride imaging; voxel-by-voxel statistical parameter mapping.

Figure 1.

Voxel-based T_1ρ maps. An atlas-based voxel-by-voxel T_1ρ map that shows regions in the medial tibia (MT), the medial femur (MF), patella (P), and in the trochlea (TrF). The T_1ρ color maps are overlaid on a gray scale magnetic resonance (MR) image and the color bar represents relaxation times from 0 ms (blue) to 70 ms (red).

Figure 2.

Correlations between quantitative and semiquantitative positron emission tomography (PET) measures. A, PET-magnetic resonance (MR) fused image in the lateral and medial side of the knee. Bone compartments are showed as regions of interest (ROIs) in the lateral and medial side of the knee in compartments—patella (P), trochlea (TrF), lateral femur (LF), lateral tibia (LT), medial femur (MF), medial tibia (MT). B and C, Correlations between NaF uptake at 60 minutes (standardized uptake value [SUV]) in all bone compartments plotted with the uptake rate (K_i) in the same compartments. SUV_avg and SUV_max have high correlations with slope K_i (R ² = .71 and 0.69).

Figure 3.

Voxel-based T_1ρ maps from the cartilage correlated with slope K_i from the bone of the medial tibia (MT). The color map represents the Pearson correlation R values representing positive and negative correlations. The right side shows the scatterplots of the average value of the voxels in white dashed boxes from the magnetic resonance (MR) images. The ellipse illustrates the 95% confidence interval. (Top) Anterior MT cartilage shows a strong positive correlation with the slope K_i from MT bone (R = .68, P = .0014). (Middle) The patella cartilage shows a strong positive correlation with slope K_i from MT bone (R = .75, P = .0045). (Bottom) The trochlea (TrF) cartilage shows a moderate negative correlation with slope K_i from MT bone (R = −.48, P = .066).

Figure 4.

Voxel-based T_1ρ maps from the cartilage correlated with standardized uptake value (SUV_max) in the bone of the patella (P). The color map represents the Pearson correlation R values representing positive and negative correlations. The right side shows the scatterplots of the average value of the voxels in white dashed boxes from the magnetic resonance (MR) images. The ellipse illustrates the 95% confidence interval. (Top) The bone layer of the P cartilage shows a moderate positive correlation with SUV_max from the P bone (R = .42, P = .10). (Middle) The articular layer of the P cartilage shows a moderate negative correlation with SUV_max from the P bone (R = −.50, P = .048). (Bottom) The trochlea (TrF) cartilage shows a strong negative correlation with SUV_max from the P bone (R = −0.72, P = .002).

Figure 5.

Voxel-based T_1ρ maps from the cartilage correlated with standardized uptake value (SUV_max) from lateral tibia (LT) bone. The color map represents the Pearson correlation R values representing positive and negative correlations. The right side shows the scatterplots of the average value of the voxels in white dashed boxes from the magnetic resonance (MR) images. The ellipse illustrates the 95% confidence interval. (Top) The cartilage of the lateral tibia (LT) shows a strong positive correlation with SUV_max from the LT bone (R = .74, P = .0042). (Middle) The cartilage in the posterior lateral femur shows a strong positive correlation with SUV_max from the LT bone (R = .71, P = .0062). (Bottom) The cartilage in the medial femur (MF) shows a strong negative correlation with SUV_max from the LT bone(R = −.61, P = .012).