A pitfall of white matter reference regions used in [18F] florbetapir PET: a consideration of kinetics

Objective Many studies have demonstrated the superiority of white matter (WM) reference regions (RR) in amyloid PET studies in comparison to cerebellar RR. However, the principle behind its improved measurement stability is yet to be elucidated. Our study aimed to determine the origin of WM stability; stability over cerebral blood flow and input function fluctuation or the greater statistical noise in the cerebellum due to its smaller size and its location in the axial periphery of the PET scanner bore. Methods We conducted simulations of [ 18 F] florbetapir using in-house program varying K 1 and input function, and adding statistical noise. Results Our simulations revealed that WM RR were more susceptible to CBF variation and input function fluctuation than cerebellar RR. WM RR did not gave superior measurement stability unless cerebellar statistical noise exceeded 4.55 times that in WM, a figure often surpassed in traditional amyloid PET studies. The greater statistical noise in cerebellum is likely the etiology for improved measurement stability of WM RR. Conclusion A longitudinal [ 18 F] florbetapir PET study should be conducted with a long bore PET. It can also be hypothesized that a second scan with the cerebellum in the axial center of a 3D PET, using a cerebellar RR to calculate changes in tracer concentration may improve the measurement stability of longitudinal [ 18 F] florbetapir studies.