Quantification of Bone Metabolism in the Hip and Spine Using PET/MRI/CT

Objectives: ¹⁸F-sodium fluoride (NaF) is a readily available radiotracer, and its use in PET/CT has shown great potential to study bone metabolism associated with osseous diseases such as osteoporosis¹. While NaF uptake should occur only in the remodeled regions of bone, PET imaging lacks the resolution needed to localize such regions. Consequently, the presence of bone marrow and bone tissue in the same analysis region underestimates the NaF uptake measures. MRI and CT can be used to quantify bone marrow content and bone volume fraction (BVF), respectively. Our goal was to enhance the utility of NaF-PET for bone metabolism studies by incorporating yellow bone marrow fraction (YBMF), red bone marrow fraction (RBMF), and BVF to partial-volume correct the SUV_mean data of the hip and spine.

Methods: We obtained NaF-PET/CT and MRI (Siemens 3T, TSE, 0.4 x 0.4 x 3 mm³) images from 27 male patients (age 36-82 years). Each patient’s scans were subject to the following segmentation procedures. OsiriX software (Pixmeo, Bemex, Switzerland) was used to define a volume of interest (VOI) of the right femoral neck region using anatomical landmarks from CT scans: the epiphyseal line as the proximal boundary and the intertrochanteric ridge as the distal boundary² (Figure 1a). Two operators each completed this PET/CT segmentation independently. PMOD Version 3.7 (PMOD Technologies, Zurich, Switzerland) was used to define a VOI for each lumbar and thoracic vertebra (Figure 1c). Within each VOI, PET and CT statistics yielded SUV_mean and BVF, respectively. A semi-automatic segmentation algorithm developed in the authors’ laboratory was used to define a similar VOI on MRI images and calculate a corresponding YBMF and RBMF (Figure 1b). Each patient’s SUV_mean was then divided by the corresponding BVF, YBMF, and RBMF to obtain partial-volume corrected SUV_mean values.

Results: The SUV_mean data independently obtained by two operators were strongly correlated (R² = 0.99, p < 0.0001). SUV_mean decreases with increasing age for both hip and spine (R² = 0.32, p = 0.0018 and R² = 0.407276, p = 0.0189, respectively). For each patient, the SUV_mean measured at the hip is strongly correlated to the SUV_mean measured in the spine (R² = 0.42, p = 0.0158). Hip partial-volume correction for BVF, YBMF, and RBMF maintained the negative correlation with (R² = 0.08, p = 0.0004), (R² = 0.34, p = 0.0016), and (R² = 0.04, p = 0.0004), respectively. Furthermore, each individual adjustment led to an increase in SUV_mean’s by as much as an order of magnitude. We were unable to collect MRI data for the spine, and thus those partial-volume corrections could not be made.

Conclusion: Our study demonstrated the feasibility and repeatability of using NaF-PET to study bone metabolism. Also, a region’s SUV_mean proved to be very sensitive to its composition of bone and marrow. For future studies, the behavior of NaF in the presence of yellow bone marrow will be of interest because yellow marrow contains mesenchymal stem cells which can differentiate into osteoblasts and affect bone metabolism.