Comparison of dynamic 2D and 3D brain PET for kinetic analysis of C-11 win 35,428 binding to dopamine transporters

3D PET data acquisition is now being used for clinical FDG-PET brain studies. However, its validity for studying kinetics of receptor-ligand interaction has not been fully evaluated. To examine the feasibility and quantitative accuracy of dynamic 3D brain PET for kinetic analysis of radioligand binding, 4 healthy volunteers and 1 Parkinson's disease patient were studied. Each subject received two 555 MBq C-11 WIN 35,428 injections and a dynamic sequence of 31 scans were acquired until 90 min p.i. in 2D and 3D mode using a GE Advance PET scanner. Both 2D and 3D image sets were reconstructed employing a Hanning filter 4.5 mm and resulting axial resolution was about 7 mm for both 2D and 3D images. The 3D data were corrected for scatter employing a method using 2D fitted Gaussian functions. Attenuation correction was performed using a 20 min transmission scan. Tissue time-activity curves were generated in the striatum and cerebellum. The forward (k3) and dissociation (k4) rate constants were calculated in the striatum using a two-compartment model, which consists of free plus nonspecifically bound (cerebellar) and specifically bound (striatal-cerebellar) compartments. The 3D kinetic results were comparable to the 2D results and within the expected range. The 3D %SE was less than 2D %SE. Striatal-to-cerebellar ratios from 2D and 3D images showed excellent correlation and agreement (r=0.994, p<0.0001, slope=0.952). The plot of striatal-to-cerebellar ratios versus time showed a linearly increasing pattern, with significantly lower RMSE in 3D than in 2D studies in linear regression analysis (0.119±0.025 vs. 0.276±0.087, p=0.028). These preliminary data suggest that 3D PET provides more reliable tissue kinetic data for the analysis of C-11 WIN 35,428 kinetics. Improved sensitivity in 3D may allow more accurate receptor characterization, especially in small brain structures or in low specific binding areas