Comparison of Invasive and Non-invasive Estimation of [ 11 C]PBR28 Binding in Non-human Primates

Purpose To identify a reliable alternative to the full blood [ 11 C]PBR28 quantification method that would be easily replicated in multiple research and clinical settings. Procedures Ten [ 11 C]PBR28 scans were acquired from 7 healthy non-human primates (NHP). Arterial input functions (AIFs) were averaged to create a population template input function (TIF). Population-based input functions were created by scaling the TIF with injected activity per body weight (PBIF) or unmetabolized tracer activity in blood at 15-,30-, and 60-min post-injection (PBIF15, PBIF30, and PBIF60). Two additional input functions were used: the native unmetabolized total plasma activity (Totals) and the Totals curve metabolite corrected by a scaled template parent fraction from a 30-min sample (TPF30-IF). Total distribution volumes (V T s) were calculated using PBIF, PBIF30, PBIF15, PBIF60, Totals, TPF30-IF, and the individual AIF (V T AIF ). Distribution volume ratios (DVR) were computed using the cerebellum and the centrum semiovale (CSO), as pseudo-reference regions (DVR Cereb , DVR CSO ). Results obtained with each method were compared to V T AIF . Applicability of these alternative methods was tested on an independent pharmacological challenge dataset of microglial activation and depletion. Evaluation was carried at baseline, immediately after intervention (acute), and weeks post-intervention (post-recovery). Results V T s computed using PBIF15 and PBIF30 showed the best correlation to V T AIF ( r > 0.90), while V T derived from the blood-free-scaled PBIF showed poor correlation ( r = 0.46) and DVR CSO correlated the least ( r = 0.26). In the pharmacological challenge study, most population-derived V T values were comparable to V T AIF at baseline and showed varied sensitivity to challenges at acute and post-recovery evaluation. DVR values did not detect relevant changes. Conclusions Population-based input functions scaled with a single blood sample might be a useful alternative to using AIF to compute [ 11 C]PBR28 binding in healthy NHPs or animals with comparable metabolism and overall perform better than pseudo-reference regions approaches.