Characteristics and quantitative impact of off-target skull binding in tau PET studies of Alzheimer disease

Background: Binding of 18F-Flortaucipir (FTP) has qualitatively been noted in the skull. We investigate the characteristics of this off-target signal, as well as its impact on regional PET analyses. Method: FTP tau PET, 18F-Florbetapir (FBP) amyloid PET, head CT, and T1-weighted MR images were acquired for 313 cognitively unimpaired and impaired older participants. A subset (n=152) also received a 11C-PiB PET scan. Another subset (n=14) received longitudinal FTP scans. MR images were segmented into ROIs using FreeSurfer and a skull ROI for each participant was defined from the head CT after removing non-skull bone and tissue. Standard uptake value ratios (SUVRs) were calculated for all ROIs. Skull binding was correlated across FTP, FBP, and PiB to determine if binding is radioisotope specific. Longitudinal persistence of FTP skull signal was tested in the subset with longitudinal data. Finally, participants were grouped by amyloid positivity based on FBP and linear models controlling for age and sex predicted regional non-partial volume corrected (non-PVC) and PVC SUVR from skull SUVR to assess the regional impact of skull signal. Result: FTP skull binding was significantly correlated with both FBP (ρ=0.48, p<0.001, Figure 1A) and PiB (ρ=0.34, p<0.001, Figure 1B), suggesting the binding is not due to tracer defluorination. In the longitudinal subset, baseline SUVR was significantly correlated with follow-up SUVR (Figure 2). Higher skull SUVR resulted in erroneously elevated regional SUVRs for areas in the frontal, parietal, and temporal lobes in amyloid negative individuals. After applying PVC, only medial temporal, inferior parietal, and lateral occipital continued to show significant positive relationships with skull SUVR (Figure 3). Decreased skull bone density was significantly related to higher skull SUVR in women but not men (see Figure 4). Conclusion: Skull signal from FTP is a stable, subject-specific phenomenon that can persist across longitudinal visits and tracers. This off-target signal can potentially bias quantitative PET measurements in AD tauopathy regions in amyloid negative but not positive individuals even after PVC is applied. The bone binding was overwhelmingly driven by women, suggesting at least some proportion of previously observed sex effects may be driven by bone uptake.