Impact of FDG Dose Reduction on Lesion Quantification in Dynamic PET - A Simulation Study Based on Clinical Trial Data

Objectives To investigate the impact of tracer dose reduction on dynamic 18F-FDG PET lesion quantification in order to explore the ability of low dose dynamic PET for both accurate quantification and reduction of tracer dose burden to the patient especially for response assessment at multiple time points of treatment.

Methods 20 min-long dynamic FDG PET scans were acquired 30min post-injection on a Philips Gemini TF 64. Dynamic list mode PET data were reconstructed using different subset numbers (Ns) (33s, 29s, 25s, 21s, 15s, 13s, 9s and 5s) and following a 20-frame (fr) protocol (60 sec ×20 fr), a 40-frame protocol (30 sec ×40 fr) and an 80-frame protocol (15 sec × 80 fr) to simulate full tracer dose of 13 mCi (481 MBq) FDG and 100%, 50% and 25% dose administrations, respectively. Maximum activity concentrations (Bq/mL) of target tumor lesions and plasma in the descending aorta were obtained with 3D Volume of Interest (VOI) placement along the different time frames. The average uptake value (Lesion: L_Ave, Artery: A_Ave), median uptake value (Lesion: L_Med, Artery: A_Med) and the standard deviation (SD) of dose uptake (Lesion: L_SD, Artery: A_SD) of both lesion and artery were calculated. The ratio of artery average uptake to lesion average uptake (A/L_Ave) and the ratio of artery median uptake to lesion median uptake (A/L_Med) were also calculated for comparison. Parameters calculated from the data reconstructed with 60 sec ×20 fr and with 33 subsets were taken as reference (gold) standard. L_Ave, L_Med, A_Ave and A_Med values generated with different doses were compared to gold standard using Student’s t-test with statistical significance being set at p<0.05. A total of 5 patient data sets were evaluated.

Results All lesions were readily identifiable even when the simulated FDG dose was decreased by 50%. However, obvious noise artifacts appeared on PET images in 3 out of 5 data sets with 33 subsets and 2 out of 5 data sets with 21 subsets when simulated FDG dose was decreased to 1/4 of the original dose. Artifacts on PET images at this 25% level could be largely eliminated when the number of subsets was decreased to be less than 15. Dose reduction greatly increased L_SD and A_SD of dose uptake time activity data. However, both L_SD and A_SD could be reduced to even lower than the gold standard values when the number of subsets was decreased (100% dose: SN<15 for L_SD, SN<29 for A_SD; 50% dose: SN<9 for L_SD, SN<25 for A_SD; 25% dose: SN=5 for L_SD, SN<9 for A_SD). L_Ave, L_Med, A_Ave and A_Med were slightly increased but not significantly affected by dose reduction (p > 0.05). A/L_Ave and A/L_Med remained consistent when dose decreased (A/L_Ave=0.22±0.007; A/L_Med=0.22±0.006), revealing a robust relationship between lesion uptake and artery uptake with dose reduction.

Conclusions A 50% dose reduction in dynamic FDG PET can be readily achieved without negative impact providing that the reconstruction is modified to use 9 subsets. This study demonstrates dynamic PET can be performed even with low tracer doses and accurate quantification realized when the appropriate number of subsets is adjusted in PET reconstruction to account for the difference in iterative convergence.