Derivation of [C-11]WAY-100635 binding parameters with reference tissue models: Effect of violations of model assumptions

In several positron-emission tomography studies of human subjects, analyses of data from the 5-hydroxytryptamine(1A) (5-HT1A) receptor radioligand, [C -11]WAY-100635 {[carbonyl-C-11]N-(2-(4-methoxy-phenyl)-1-piperazin-1-yl)eth yl)-N-(2-pyridyl)cyclohexanecarboxamide} have shown a discrepancy between t he outcome measure k(3)/k(4) (binding potential normalized to cerebellum) a s estimated by the simplified reference region method and results obtained by conventional kinetic modeling with an arterial input function. The refer ence region method has yielded results that are lower than the conventional approach, with the relative underestimation appearing to be an increasing function of k(3)/k(4). We performed simulations on idealized data to identi fy the source of the discrepancy, Both the simplified reference tissue mode l (SRTM) and the original full reference tissue model (FRTM) were tested to determine (a) if the error in estimated k(3)/k(4) is dependent on the bloo d flow in the region of interest relative to the blood flow in the region o f reference (R-1) and on the receptor density in the region of interest (tr ue k(3)/k(4)), and (b) which violation of the reference model assumptions w ere responsible for this effect. FRTM returned parameter estimates that wer e independent and accurate if the reference region was constructed precisel y as a one-tissue compartment model. SRTM overestimated k(3)/k(4) when the reference region was constructed as a one-tissue compartment model and unde restimated k(3)/k(4) when the reference region was constructed as a two-tis sue compartment model (which is the case for [C-11]WAY-100635). In both cas es, the magnitude of the error in k(3)/k(4) returned by SRTM was dependent on true R-1 and true k(3)/k(4). In conclusion, the SRTM is associated with a bias in the derivation of k(3)/k(4) that is not a simple scaling factor. This magnitude of these errors should be carefully evaluated for each new r adioligand. NUCL MED BIOL 27;5:487-492, 2000. (C) 2000 Elsevier Science Inc . All rights reserved.