Effect of partial volume correction on estimates of the influx and cerebral metabolism of 6-[(18)F]fluoro-L-dopa studied with PET in normal control and Parkinson's disease subjects.

The poor spatial resolution of positron emission tomography (PET) is a limiting factor in the accurate assay of physiological processes investigated by compartmental modeling of tracer uptake and metabolism in living human brain. The radioactivity concentration in a region-of-interest is consequently altered by loss of signal from that structure and contamination from adjacent brain regions, phenomena known as partial volume effects. We now apply an MRI-based algorithm to compensate for partial volume effects in the special case of compartmental modeling of the cerebral uptake of 6-[(18)F]fluoro-L-dopa (FDOPA), an exogenous substrate of dopa decarboxylase. High-resolution MRI scans were obtained from normal volunteers (n = 4) and patients with Parkinson's disease (n = 4) in order to segment specific brain regions and calculate the partial volume correction factors. Dynamic 2D PET scans were acquired during 90 min following intravenous infusion of FDOPA. After partial volume correction, the apparent net blood-brain clearance of FDOPA (K(i)) was greatly increased in caudate and putamen of normal subjects and in caudate of Parkinson's disease patients. The equilibrium distribution volume of FDOPA (V(D)(e)) in cerebral cortex increased by 35% in all subjects. Using a two-compartment model, the relative activity of dopa decarboxylase with respect to FDOPA (k(D)(3)) in the basal ganglia was increased 2-3 times in normal subjects, to the range obtained previously in brain of living rat. The partial volume correction also increased the magnitude of k(D)(3) in caudate of Parkinson's disease patients, but did not alter k(D)(3) in putamen. A three-compartment model correcting for elimination of decarboxylated metabolites also yielded higher estimates of k(D)(3), but with a penalty in precision of the estimates. Together, these observations suggest that the limited spatial resolution of PET results in substantial underestimation of the true rate of FDOPA uptake and metabolism in vivo, and may also tend to obscure regional heterogeneity in the neurochemical pathology of Parkinson's disease. Copyright 2000 Wiley-Liss, Inc.