Parameter and index images of benzodiazepine receptor concentration in the brain

In vivo studies of ligand-receptor interactions with PET data are based on different approaches that provide either quantitative results (receptor density and affinity) or indices that are assumed to be correlated with the receptor concentration. The aims of this study are to obtain parametric images of benzodiazepine receptor concentration and of flumazenil affinity and to study the validity of two receptor concentration indexes.

A three-compartment ligand-receptor model, [11C]flumazenil, and experimental data obtained using a three-injection protocol in human volunteers were used to acquire parametric images. The delayed activity method and the apparent distribution volume (estimated using a two-compartment model) were also tested and their results compared with those of the multi-injection approach.

Parametric images of receptor density, affinity and all kinetic parameters were obtained with acceptable variation coefficients. A correlation between receptor density and apparent affinity was found (r = 0.83; p < 0.0005). The correlation between receptor concentration and apparent distribution volume (estimated with three- and two-compartment models, respectively) was accessed using both a linear (the usual hypothesis) and a nonlinear correlation derived from the relationship between the receptor density and the affinity.

In spite of the complexity of this protocol (three injections, a 2-hr experiment, blood sampling and a metabolite study), we showed that the multi-injection approach is suitable for parametric brain imaging. By using this approach as a reference, we deduced that the distribution volume and delayed activity images are valid methods in the usual range of the benzodiazepine receptor concentrations found in the human brain.