Selectivity requirements for diagnostic imaging of neurofibrillary lesions in Alzheimer's disease: a simulation study

Whole-brain imaging is a promising strategy for premortem detection of tau-bearing neurofibrillary lesions that accumulate in Alzheimer's disease. However, the approach is complicated by the high concentrations of potentially confounding binding sites presented by beta-amyloid plaques. To predict the contributions of relative binding affinity and binding site density to the imaging-dynamics and selectivity of a hypothetical tau-directed radiotracer, a nonlinear, four-tissue compartment pharmacokinetic model of diffusion-mediated radiotracer uptake and distribution was developed. Initial estimates of nonspecific binding and brain uptake parameters were made by fitting data from a previously published kinetic study of Pittsburgh Compound B, an established amyloid-directed radiotracer. The resulting estimates were then used to guide simulations of tau binding selectivity while assuming early-stage accumulation of disease pathology. The simulations suggest that for tau aggregates to represent at least 80% of specific binding signal, binding affinity or density selectivities for tau over beta-amyloid should be at least 20- or 50-fold, respectively. The simulations also suggest, however, that overcoming nonspecific binding will be an additional challenge for tau-directed radiotracers owing to low concentrations of available binding sites. Overall, nonlinear modeling can provide insight into the performance characteristics needed for tau-directed radiotracers in vivo.