Introduction

Osteoarthritis (OA) is a degenerative joint disease characterized by degradation of articular cartilage and structural changes in subchondral bone that lead to unusual loading and mechanical stress on the joint. Once cartilage loss and bone structural changes are detected on MRI, they are practically irreversible. Quantitative MRI techniques such as elevated T2 relaxation times have been used to detect cartilage degradation prior to volumetric loss¹. Additionally, unusual mechanical loading after an exercise intervention can be detected using [¹⁸F]sodium fluoride PET imaging, as increased standard uptake values (SUV) correspond to increased bone metabolism². These techniques are promising to help detect and understand early mechanisms contributing to OA progression. In this work, we use hybrid PET-MRI to evaluate the relationship between subchondral bone response to exercise and joint changes over two years, including changes in cartilage T2 relaxation times and structural changes in bone and cartilage.

Methods

Both knees of five subjects with knee osteoarthritis were scanned with a 3.0T PET-MR system. Two sets of images were taken, with a two-year gap between the initial and final visits (Figure 1). At the initial visit, subchondral bone response to exercise was assessed using [¹⁸F] sodium fluoride ([¹⁸F]NaF) uptake. Upon injection of a 2.5 mCi dose of [¹⁸F]NaF, PET and bilateral MRI data were obtained over 50 minutes. Subjects then performed a single-leg squat exercise, with the other leg supported (Figure 1). After the exercise, the PET-MRI scan was repeated. Two years later, only a single bilateral MRI scan was repeated. MRI sequences included a quantitative double-echo in steady-state (qDESS) sequence to assess cartilage T2 relaxation times^3,6 and 2D FSE sequences for morphology. MRI Osteoarthritis Knee Score (MOAKS)⁴ assessment of each knee was performed by a musculoskeletal radiologist for both the initial and final scans.
The change in the maximum standard uptake value (ΔSUV_max) associated with exercise was measured in bone regions representing the patella; medial and lateral tibia; and anterior, central, and posterior regions of the medial and lateral femur. The ΔSUV_max was compared between regions that experienced a MOAKS change after two years and regions that did not, using a Wilcoxon Rank Sum test (𝛂= 0.05). The femoral, tibial, and patellar cartilage were automatically segmented from qDESS images using a deep-learning based algorithm⁵ and manually corrected as necessary. The cartilage surfaces were divided into nine regions corresponding to the bone regions described above. The ΔSUV_max values at baseline were also compared to the change in cartilage T2 relaxation times over two years using a Spearman’s correlation coefficient in legs that exercised and those that rested (ρ, 𝛂= 0.05).

Results

A representative image of MOAKS progression and baseline SUV maps before and after exercise is shown in Figure 2. A difference in baseline ΔSUV_max was seen between regions that had a MOAKS change after two years and those that did not (Figure 3). Regions that had a MOAKS change had a mean ΔSUV_max of 3.29 after exercise, while regions that did not have MOAKS change had a mean ΔSUV_max of 2.16 (p = 0.08). There was a moderate positive correlation between ΔSUV_max and ΔT2 for the exercised legs (ρ = 0.459, p < 0.05) (Figure 4, left). A nonsignificant poor correlation was observed between ΔSUV_max and ΔT2 for the rested legs (ρ = -0.196, p = 0.198) (Figure 4, right).

Discussion

This study showed that for osteoarthritic knees, in regions that experience MOAKS changes after two years, there is increased metabolic bone response to exercise at baseline compared to regions that do not have MOAKS progression, with this difference trending towards statistical significance in this small feasibility study.
A moderate correlation was seen between ΔSUV_max and ΔT2 after two years for exercised OA knees, suggesting that a large metabolic bone response to loading is correlated with cartilage microstructure changes that are often used as early markers of cartilage degeneration. However, the observed ΔT2 was negative in numerous regions which was unexpected. This may be due to fibrosis and cartilage loss that was observed in several subjects at two years. No correlation between ΔSUV_max and ΔT2 was observed in the rested knees.
Prior studies suggest that this metabolic bone response is load-dependent. Thus, high [¹⁸F]NaF PET response to exercise may reflect joint dysfunction that is predictive of future joint degeneration. Further, larger studies evaluating an early OA cohort would be informative.

Conclusion

Subchondral bone response to exercise, as measured by the change in maximum standard uptake value of [¹⁸F]NaF, may be a useful metric to predict joint structural changes and cartilage degradation in OA knees after two years.

Acknowledgements

This work was funded by research support from GE healthcare and NIH Grants R00EB022634, R21EB030180 and R01AR074492.