Osteoarthritis (OA) is a complex disease characterized by degeneration affecting all tissues in the joint. While great advances have been made in MRI of soft tissues, evaluation of bone health is still limited to gross signal changes, such as bone marrow lesions and osteophytes, on conventional MRI scans. To better understand the physiological processes that drive these changes, [¹⁸F]-fluoride is a promising PET tracer that reveals bone remodeling within OA lesions [1]. Most clinical PET scans look at static accumulation of a radiotracer after an appropriate post-injection delay. However, kinetic modeling of dynamic PET data enables quantification of new rate constants that describe how the tracer is metabolized by tissue. This study investigates dynamic uptake of [¹⁸F]-fluoride in OA-related bone marrow lesions and osteophytes observed on MRI. We aim (1) to characterize rate constants of bone metabolism in bone pathology relative to healthy bone; and (2) to determine whether increased blood flow delivery or specific binding to bone mineral drives elevated accumulation of [¹⁸F]-fluoride in OA.

Both knees of 4 subjects with knee OA (Kelgren-Lawrence grade 1 - 3) were imaged immediately after injection of 2.5 - 5 mCi of [¹⁸F]-fluoride on a 3T time of flight PET-MRI hybrid system (GE Healthcare, Milwaukee, WI). An experienced radiologist identified regions of interest (ROIs) within bone marrow lesions (BMLs) and osteophytes on T₂-weighted fast spin echo MRI scans that were co-registered to the PET images. An ROI of normal tissue in each knee was also identified for all subjects.

Arterial input functions were determined in the popliteal artery from dynamic PET frames binned with 10-s temporal resolution (Figure 1). Tissue time activity curves (TACs) of [¹⁸F]-fluoride uptake for each area of bone pathology and in normal bone were created from 1-minute frames for the first 15 minutes and 5-minute frames thereafter (total of 45 minutes). In each ROI, we used a two-compartment kinetic model in PMOD software to fit the TAC (PMOD Technologies, Zurich). This model provides quantitative values of K₁ (blood flow delivery of tracer); k₂ (reverse transport back to blood plasma); and k₃ (rate of bone mineralization) that describe the dynamic uptake of the [¹⁸F]-fluoride tracer. For all ROIs, the macroparameter K_i = (K₁ x k₃) / (k₂ + k₃) was calculated as an overall metric of bone metabolism and osteoblast activity [2, 3]. We compared the K_i, K₁, and k₃ rate constants of different bone pathologies with values from normal bone through ANOVA statistics.

We identified 9 BMLs, 11 osteophytes of grade 1, and 6 osteophytes of grade 2 or 3. These MRI findings corresponded to areas of high [¹⁸F]-fluoride uptake relative to surrounding bone. The TACs from areas of bone pathology showed increased tracer uptake relative to normal tissue that continued over the duration of the scan. Furthermore, this accumulation was typically faster in BMLs compared to osteophytes (Figure 2), and faster in osteophytes of grade 2 or 3 compared to grade 1 (Figure 3).

Kinetic modeling provided good fits of the TACs, with chi-squared values of less than 12.1 for all regions. Mean quantitative Ki in units of ml/min/ml (representing total bone metabolism) was 0.062 ± 0.04 for BMLs; 0.057 ± 0.02 for osteophytes of grade 2 or 3; 0.016 ± 0.01 in osteophytes of grade 1; and 0.007 ± 0.01 in normal tissue. BMLs and grade 2-3 osteophytes showed higher total bone metabolism K_i (P < 0.01) and higher bone mineralization rate k₃ (P < 0.01) relative to grade 1 osteophytes and normal bone (Figure 4).

Kinetic modeling of [¹⁸F]-fluoride revealed elevated bone metabolism K_i and bone mineralization rate k₃ in BMLs and osteophytes compared to healthy bone. Our findings are consistent with high standardized uptake value (SUV) values previously observed in OA [4]. While the abnormal blood flow delivery K₁ to these regions also followed a similar trend, the differences from normal tissue were subtler and did not reach statistical significance. This observation suggests that while increased blood flow may contribute to high SUV, the rate of mineralization k₃ is a key driver of [¹⁸F]-fluoride accumulation in OA lesions. Quantification of rate constants may also help stratify different OA phenotypes, as K_i and k₃ were higher for osteophytes of grade 2 or 3 than for grade 1 osteophytes.Kinetic modeling of dynamic [¹⁸F]-fluoride PET scans provides quantification of bone metabolism rate constants that highlight underlying mechanisms of OA pathology on MRI scans. These new physiological parameters may help differentiate between different grades of OA lesions or identify which lesions are active parts of the disease.