Modeling alternatives for cerebral carbon-11-iomazenil kinetics

The in vivo binding kinetics of [11C]iomazenil, a central benzodiazepine antagonist, were analyzed using PET and compartmental modeling. This method is of interest because it allows validation of the SPECT tracer [123I]iomazenil.

The experimental protocol consisted of serial PET imaging following a single bolus injection of the serial PET imaging following a single bolus injection of the radioligand. Imaging was performed on five healthy young volunteers over 106 min. The tissue time-activity curves of various brain regions were analyzed with models consisting of two (K1, k2") and three (K1, k2', k3', k4) compartments. Some of the methods use simultaneous fitting of the data from multiple brain regions coupled with common parameters. Distribution volumes and k3-based parameters [(K1/k2') k3' and k3')] were chosen to represent receptor density. Goodness of fit was assessed with F-test statistics and chi-square analysis.

Compared with the two-compartment model, goodness of fit was significantly improved by all three-compartment configurations. Of the three-compartment models, goodness of fit was similar for the configurations with K1/k2', k4 or no parameter coupled, and slightly worse when both parameters were coupled. The most reliable estimates of receptor density were obtained from the specific distribution volumes (DVs) calculated with the three-compartment model, and the coupling of k4 or both k4 and K1/k2'. Due to oversimplification of the kinetics, the DV values calculated with the two-compartment model were underestimated.

Reliable quantitative information regarding benzodiazepine receptor density following bolus injection of iomazenil is best obtained by tracer kinetic modeling that uses a three-compartment model and parameter coupling.