Introduction
[11C](R)-PK11195 is a tracer of microglial activation in the brain. In previous studies PK11195 binding was analysed using reference tissue models 1 , showing increased binding in the thalamus with age 2 . Up to date PK11195 binding amongst young, old and AD subjects has not yet been analysed using statistical parametric mapping (SPM). Therefore, a study was performed to evaluate SPM analysis of PK11195 binding in AD versus old and young subjects. SPM analysis was based on: (1) binding potential (BP) images obtained with a reference tissue input (Ichise multilinear regression) and (2) volume of distribution (Vd) images obtained with plasma input (Logan plots).
Methods
Thirty [11C](R)-PK11195 studies with on-line arterial sampling were included (8 young (<40 y) and 12 old controls (>60 y), and 10 AD patients). Using the on-line whole blood curve and a number of manual arterial samples, a metabolite corrected plasma input curve was derived. A reference tissue input curve was obtained from a region of interest over cerebellum. PET scans were reconstructed using FBP Hanning 0.5. Additional smoothing with a 10 mm FWHM Gaussian filter was performed to reduce noise to acceptable levels, thereby avoiding noise induced bias 3 during Ichise and Logan plot analysis. Subsequently, SPM analysis of both Ichise and Logan images was performed without further smoothing, i.e. the standard smoothing in SPM was omitted. Because Vd images represent total distribution volume, including free and non-specific binding, SPM analysis of Vd data was performed with and without proportional scaling.
Results
SPM analysis without proportional scaling based on Vd images did not show any regions with differences in PK11195 binding amongst all groups. When proportional scaling was applied, only a trend (p<0.1) of increased PK11195 binding was observed in both thalami. When SPM analysis was performed using BP images, significantly (P<0.001) increased PK11195 binding in the thalamus was observed in both older and AD subjects, compared with young controls. Furthermore, SPM of BP images showed additional areas (e.g. occipital, lateral temporal areas and subcortical structures (presumably nucleus subthalamicus)) with increased binding in elderly and AD subjects.
Discussion
Logan plots provide Vd values, which depend on the sum of free, specifically and non-specifically bound ligand. As non-specific binding of PK11195 is relatively high, SPM analysis of Vd images without proportial scaling may be confounded by intersubject variability in non-specfic binding. To some extent this may be compensated by proportional scaling, explaining the observed trend of increased uptake in the thalamus after such scaling. SPM analysis of BP images was found to be most powerful. As BP images only represent increased (specific) binding, it does not suffer from the large contribution and variability of non-specific binding. In addition, reference tissue input models may provide better precision than plasma input models, because they do not need (noisy) metabolite corrections.