Quantification of 5-HT 1a receptors in temporal lobe epilepsy with PET and [ 18 F]FCWAY: Effect of modeling methods on group discrimination

Abstract

Previous PET studies with [¹⁸F]FCWAY were modeled with a 120-min kinetic analysis, using arterial input function with corrections for uptake of the metabolite [¹⁸F]fluorocyclohexanecarboxylic acid ([¹⁸F]FC) and vascular activity. We evaluated whether logistically less demanding analysis approaches can be used as alternatives to the full analysis. Dynamic frames were acquired for 120-min in 13 temporal lobe epilepsy (TLE) patients and 10 controls, and modeled with a simplified 3-parameter 2-tissue compartment. The gold standard binding index was BP_F. Images were also generated with a 90-min (BP_{F_90}) and 60-min (BP_{F_60}) data analysis interval, without correction for [¹⁸F]FC (BP_{F_NOFC}) and without neither [¹⁸F]FC correction nor vascular correction (BP_{ND_NOFCVA}). f_P was significantly higher in patients than in controls. BP_{F_NOFC} had very similar statistical power as BP_F, except in the raphe where the effect size of BP_{F_NOFC} was about 40% lower than that of BP_F. BP_{ND_NOFCVA} had poorest performance. Negligible loss of statistical significance in detected differences occurred in the cortex with both BP_{F_90} and BP_{F_60}, but BP_{F_60} was associated with some time instability in the raphe. PET scans can be reduced to 90-min in future, arterial line-based, TLE studies. Lack of correction for [¹⁸F]FC mildly reduced the statistical power only in the raphe.