Ichise M, Becker G, Barthel H, Patt M, Luthardt J, Gertz H-, Schultze-Mosgau M, Rohde B, Reininger C, Sabri O.
Kinetic modeling of florbetaben PET data to quantify β-amyloid binding in the human brain.
Human Amyloid Imaging 2010 Meeting Abstracts. 2010 April 9;
Objectives: Florbetaben is a promising PET tracer for imaging of human β-amyloid deposits. This 18F-ligand is
rapidly metabolized in vivo, producing a low molecular weight, polar metabolite, which appears to enter the brain.
We modeled the presence of the metabolite in the brain in quantifying β-amyloid binding.
Methods: PET data and multiple arterial samples were collected over 260 min after i.v. injection of florbetaben
(300 MBq) for 10 Alzheimer’s disease (AD) and 10 healthy controls (HC). PET data were analyzed by 1) “full” two
input (parent and metabolite) model (2F) with 6 and 4 kinetic parameters in target and reference tissue (cerebellar
cortex), respectively, 2) “simplified” two input model (2S) in which the metabolite delivery rate constant was fixed,
3) one input (parent) model (1), and 4) reference tissue model, to estimate the β-amyloid binding parameters, BPP
(2F), BPP (2S), BPP (1) and BPND, respectively. Target-to-cerebellar cortex ratio (RT) at 90 min was also calculated. To evaluate the impact of the metabolite, these 5-binding parameters were compared.
Results: There were no differences between BPP (2F) (2.65±1.82) and BPP (2S) (2.67±2.22) (p=0.7). BPP (1)
(3.80±2.55) and RT (0.57±0.37) were 44% and 32% higher than BPP (2F) and BPND, respectively (n=20). There were
strong correlations between BPP (1) or RT and BPP (2F) (r2=0.82 and 0.89, pConclusion: The florbetaben metabolite effect can be modeled in quantifying β-amyloid binding. However, the
binding parameters that ignore the metabolite correlate very well with those that account for it and all binding
parameters discriminate effectively between amyloid positive and negative scans.