Abstract
177
Objectives: Hybrid PET/MR imaging is a promising technique to improve understanding of bone metabolism in joint diseases such as osteoarthritis. Quantitative analysis of dynamic Na[18F]F- ([18F]fluoride) uptake utilizes a bone time activity curve (TAC) and an arterial input function to model tracer delivery to the bone and identify mechanisms of uptake such as bone perfusion (K1), extraction fraction of [18F]fluoride into the bone, and the total rate of [18F]fluoride clearance from the blood (Ki). Tracer clearance has been shown to have good reproducibility, with a coefficient of variation of less than 15% when calculated from 60 min of dynamic PET data1-4. Shorter PET/MRI scan times could improve the ability to assess acute changes in physiologic conditions, reduce impacts of motion, and improve subject comfort. The goal of this work was to assess the impact of abbreviated dynamic TAC data on the reproducibility of dynamic [18F]fluoride uptake parameters in the knee.
Methods: The knees of six individuals with and without osteoarthritis were scanned using a 3 Tesla PET/MRI system. List mode PET data of both knees was acquired for 50 minutes (53 ± 3.3 minutes) following injection of 2.5 mCi [18F]fluoride. MR angiography and a short time-frame PET angiogram were used to calculate an arterial input function5. TAC data of bone tracer uptake were determined using dynamic PET data. TAC and arterial input function data were fitted to a 2-tissue compartment model using a nonlinear regression method to estimate rate parameters K1, k2, and k3 (COMKAT). The rate parameters were calculated for the full-duration TAC data, as well as for retrospectively shortened TAC data of 20, 25, 30, 35, 40, 45, and 50 minutes duration. The parameters K1, the extraction fraction (k3/[k2+k3]), and Ki (K1*extraction fraction) were determined for each TAC duration for bone regions representing cortical and trabecular bone of the medial tibia and medial femur and the cortical bone of the patella on the right knee. The coefficient of variation (CV, %) of the measured kinetic parameters derived from the shorter duration TAC curves were calculated with reference to parameters from the full-duration TAC curve.
Results: Estimated K1, Ki, and extraction fraction values were relatively stable over the TAC durations (representative results in Figures 1 and 2). Compared to the full-duration TAC data across all bone regions, the average CV of K1 and Ki was less than 10% for all TAC durations except 20 minutes (Table 1); the average extraction fraction CV was less than 5% for all TAC durations.
Conclusions: The average coefficient of variation was less than 10% for kinetic uptake parameters estimated using at least 25 minutes of dynamic data and data acquired over 50-55 minutes. The average coefficient of variation associated with shortening the TAC data is comparable to previously reported reproducibility results for Ki using 60-minute PET data1-4. Reducing TAC length increases the CV of kinetic measures. However, results suggest that the length of dynamic data acquisition can be reduced to potentially as short as 25 minutes without appreciable effects on kinetic parameters of 18F uptake derived from 2-tissue compartment models.