Abstract
442
Objectives Kinetic modeling for the assessment of tumor response demands an understanding of the repeatability of the underlying quantitative parameters in the absence of therapy. Establishing that changes in parameter values during treatment reflect changes in underlying biology rather than statistical fluctuations in the measurement is crucial. In this study we determine reproducibility of kinetic parameters (KPs) for 18F-fluoromisonidazole (18FMISO), deduced from kinetic modeling of dynamic PET (dPET) images.
Methods Ten patients diagnosed with non-small cell lung cancer (NSCLC) underwent 2 free-breathing (3 days apart) pre-treatment 18FMISO dPET. Data for dPET were acquired as follows: 12x10 sec, 10x1min, and 5x5min (37 m). Two 10-min frames were acquired at 90 +/- 10 min and 150 +/- 10 min, respectively. Data were reconstructed in 128x128x47 matrices and a700-mm FOV. Pixel-wise and VOI-average time-activity curves (TACs) were obtained for each lesion. Input functions were derived from VOIs over the aorta for each session. Modeling was performed on all TACs with a 2-tissue,3-compartment (plasma, free, trapped) model with irreversible binding (k4=0).Reproducibility was assessed for all KP (K1,Vd,k3,Vb) by calculating the respective coefficients of repeatability (CR) between the 1st and 2nd scans for KPs from average TACs.
Results The median KPs for the two pre-treatment data sets are: k3) 0.0023 min-1; K1) 0.32 min-1; Vd) 0.70 ;Vb) 0.12. The corresponding CRs for these are: k3) 0.0034 min-1; K1) 0.13 min-1; Vd) 0.34 ;Vb) 0.13. Comparison of KPs calculated from averaging all TACs in the VOI,the % difference in the parameters are 35.1, 6.8, 32.2, and 28.8% for k3,Vd,K1 and Vb, respectively.
Conclusions This study shows that of performing kinetic analysis of non-gated (free-breathing) PET-derived TACS in lung lesions may be very susceptible to anomalies introduced by breathing motion. In future,the same method will be used to compare results for breath-holding. These are expected to have higher repeatability.