Introduction

Osteoarthritis (OA) is a whole joint disease in which degeneration of articular cartilage, are accompanied by changes in subchondral and trabecular bone. The changes in bone structure occur over longer time periods, but are mediated by faster processes of bone remodeling^1,2. The aim of this study was to analyze bone structure, and cartilage degeneration with bone remodeling and functional changes that may potentially be markers of early OA. Using state of the art simultaneous Time-of-flight (TOF) Positron Emission Tomography/Magnetic Resonance (PET/MR) imaging, the goal was to evaluate the association between quantitative PET/MR biomarkers with patient reported outcomes (PROMs), an important step in the validation of this technique as a tool for assessing OA.

Methods

Fourteen subjects with radiographic or symptomatic OA were recruited for this study (age 57.79±8.58 years, BMI 25.147±4.26 Kg/m3, 10 males, 12(85.8%) Kellgren-Lawrence (KL) 0-1, 2(14.2%) Kellgren-Lawrence (KL) 2-4). Patients reported outcome was measured with KOOS³ and WOMAC⁴. Using an intravenous catheter, patients were placed feet first into a 3T TOF PET-MR scanner (GE Healthcare, Waukesha, WI). Patients were then injected with an average of 340.4 Mbq [¹⁸F]-NaF through an intravenous catheter at the onset of scan. A dynamic PET scan was acquired the first 60 minutes and a static 10 min scan was acquired at 90 minutes after the initial time of injection, while a simultaneous MR scan was conducted. For PET analysis bone regions were manually segmented using circular ROIs in PMOD (PMOD technologies Ltd, Zurich). Average and maximum Standardized Uptake Values (SUV) were computed for 60 and 90 min data Figure 1 shows the SUV at 60 minute computed from one of the subjects. Dynamic PET data Ki defined as the slope of the Patlak plot was derived from the assessment of the 60 min PET data using a linear Patlak graphical analysis method. The MRI imaging protocol included simultaneous T1ρ/T2 MAPSS⁵ for cartilage compositional assessment and variable flip angle 3D fast spin-echo⁶ (3DVFA-FSE: pixel-spacing 0.19mm, slice-thickness 0.5mm TR/TE = 400/9.9ms, ETL=8) used for the trabecular structure quantification. Trabecular number: Tb.N and trabecular volume BV/TV were obtained within manually segmented ROIs (Figure 2). Fuzzy-C means clustering was used to segment the ROI into bone and marrow⁷. Pearson partial correlation adjusted for age, gender and BMI were used to study the associations between bone remodeling, trabecular bone structure data and PROMs. Voxel-based relaxometry (VBR)⁸ for T_1ρ and T₂ was then used to study cartilage biochemical composition.

Results

Figure 3 shows the overall correlation matrix computed on the 3 data domains considered (PET, Tb and PROMs). SUV^avg₆₀ showed moderate correlations between different compartments (average R = 0.31), generally much stronger was the inter-compartmental correlations when the SUV^avg₉₀ was considered, with stronger correlation observed between adjacent compartments (SUV^avg₆₀.MT vs SUV^avg₆₀.MF R=0.90). Inter-compartmental SUV^avg₉₀ correlation was much weaker when the patella compartment was considered (average R=0.35). This lack of correlation between patella and other compartments was also observed in the analysis SUV^max₆₀ and SUV^max₉₀. Generally positive association was observed between SUV^avg and Ki values (average R=0.35) showing similar behavior for 60 and 90 min data. Moderate negative correlation was observed between the SUV^avg₆₀ in femoral trochlea compartment and trabecular bone structure (Tb.N average R=-0.46, BV/TV average R=-0.26). This relationship was stronger for the SUV^max₉₀ (Tb.N average R=-0.85, BV/TV average R=-0.81). Trabecular bone data did not demonstrate direct association with PROMs. However, SUV^avg₆₀ in patella showed a negative association with all the KOOS and WOMAC sub-scores with strongest correlation observed with KOOS pain (R=-0.49). In addition to the association with PROMs SUV^avg₆₀ in patella was observed to be positively associated with cartilage T_1ρ especially in trochlea and medial tibia (Figure 4).

Discussion and Conclusions

In this study correlations were seen between bone trabecular structure measured with MRI and bone remodelling, as well as a relationship between [¹⁸F]-NaF PET uptake and patient reported outcomes. Our results showed stronger inter-compartmental association in SUV^avg₉₀ then SUV^avg₆₀ , suggesting that by 90 minutes the local differences are no longer visible as the uptake has equilibrated. While strong associations were observed between trabecular bone data and bone uptake (SUV), the association observed with Ki was weak. SUV^avg₆₀ was observed to be associated with the level of pain measured by KOOS, speculatively this may indicate that bone remodelling and blood flow may be markers for pain, and also associated cartilage matrix degeneration as indicated by the correlations with T_1ρ. These results demonstrate the potential of [¹⁸F]-NaF PET-MR to extract quantitative imaging biomarkers potentially useful in studying osteoarthritis and relating imaging findings with patient outcomes.

Acknowledgements

NIH-P50AR060752 and GE Healthcare.

References

[1] E. Radin, R. Rose, Role of subchondral bone in the initiation and progression of cartilage damage. Cinical orthopedics and Related research 213, 34 (1986). [2] D. B. Burr, M. A. Gallant, Bone remodelling in osteoarthritis. Nature reviews. Rheumatology 8, 665 (Nov, 2012). [3] Roos, E.M., Roos, H.P., Lohmander, L.S., Ekdahl, C. & Beynnon, B.D. Knee Injury and Osteoarthritis Outcome Score (KOOS)--development of a self-administered outcome measure. The Journal of orthopaedic and sports physical therapy 28, 88-96 (1998). [4] Ackerman, I. Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). The Australian journal of physiotherapy 55, 213 (2009).