Kinetic modeling and long-term test-retest reproducibility of the mGluR5 PET tracer 18F-FPEB in human brain

(18)F-FPEB is a promising PET tracer for studying the metabotropic glutamate subtype 5 receptor (mGluR5) expression in neuropsychiatric disorders. To assess the potential of (18)F-FPEB for longitudinal mGluR5 evaluation in patient studies, we evaluated the long-term test-retest reproducibility using various kinetic models in the human brain. Nine healthy volunteers underwent consecutive scans separated by a 6-month period. Dynamic PET was combined with arterial sampling and radiometabolite analysis. Total distribution volume (V(T)) and nondisplaceable binding potential (BP(ND)) were derived from a two-tissue compartment model without constraints (2TCM) and with constraining the K(1)/k(2) ratio to the value of either cerebellum (2TCM-CBL) or pons (2TCM-PONS). The effect of fitting different functions to the tracer parent fractions and reducing scan duration were assessed. Regional absolute test-retest variability (aTRV), coefficient of repeatability (CR) and intraclass correlation coefficient (ICC) were computed. The 2TCM-CBL showed best fits. The mean 6-month aTRV of V(T) ranged from 8 to 13% (CR < 25%) with ICC > 0.6 for all kinetic models. BPND from 2TCM-CBL with a sigmoid fit for the parent fractions showed the best reproducibility, with aTRV ≤ 7% (CR < 16%) and ICC > 0.9 in most regions. Reducing the scan duration from 90 to 60 min did not affect reproducibility. These results demonstrate for the first time that (18)F-FPEB brain PET has good long-term reproducibility, therefore validating its use to monitor mGluR5 expression in longitudinal clinical studies. We suggest a 2TCM-CBL with fitting a sigmoid function to the parent fractions to be optimal for this tracer.