Kinetic modeling and long-term test-retest of ¹⁸F-FPEB mGluR5 PET in healthy volunteers.

Objectives The metabotropic glutamate subtype 5 receptor (mGluR5) is involved in many neuropsychiatric disorders. ¹⁸F-FPEB is a promising PET tracer for studying mGluR5 expression and as aid in drug development. In order to assess the potential for long-term evaluation in longitudinal patient studies, we evaluated full kinetic modeling with long-term test-retest variability (VAR) in healthy volunteers.

Methods The regional brain kinetics of ¹⁸F-FPEB were tested in 8 healthy volunteers (age 38±15, 4M/4F) who received two consecutive PET scans separated by 176±11 days. Dynamic PET scans were combined with arterial sampling and radiometabolite analysis. Total distribution volume (V_T) and binding potential (BP_nd) values were derived from a two-tissue compartment model, both with (2TCM-C) and without (2TCM) constraining the K1/k2 ratio to either to cerebellum (CBL), pons or to a coupled value from all regions. Values were evaluated in 13 brain regions obtained from the Hammers template in PMOD v3.6.

Results 2TCM was the model of choice according to the Akaike Information Criterion. V_T from 2TCM-C with CBL as reference region showed negligible bias compared to 2TCM (e.g. mean for all VOIs = -0.07%±0.5%). VAR of V_T was similar for 2TCM and 2TCM-C (e.g. VAR frontal lobe (FL) = 6%; range all VOIs : 0.1-23%). For 2TCM-C, the lowest VAR for BP_nd was achieved by fixing K1/k2 to the value of the CBL. TRV of BP_nd was smaller for 2TCM-C (VAR for FL = 5%; range : 0.7-11%) than for 2TCM (mean VAR for FL = 15%, range 2-57%). Relative to 2TCM, BP_nd using 2TCM-C showed a positive bias (e.g. bias FL = 23±25%). Plasma-to-blood ratio analysis suggests that metabolites do not cross the blood brain barrier.

Conclusions ¹⁸F-FPEB brain PET has good long-term test-retest variability and therefore allows monitoring of mGluR5 expression in longitudinal clinical studies in different neuropsychiatric conditions. BP_nd derived from 2TCM-C with K1/k2 fixed to the value of the cerebellum shows lowest long-term test-retest variability.