MR-based spatial normalization improves [18F]MNI-659 PET regional quantification and detectability of disease effect...

The positron emission tomography (PET) tracer [18F]MNI-659, selective for phosphodiesterase 10A (PDE10A), is a promising tool to assess an early biomarker for Huntington's disease (HD). In this study we investigated [18F]MNI-659 uptake in the Q175 mouse model of HD. Given the focal striatal distribution of PDE10A as well as the striatal atrophy occurring in HD, the spatial normalization approach applied during the processing could sensibly affect the accuracy of the regional quantification. We compared the use of a magnetic resonance images (MRI) template based on individual MRI over a PET and CT templates for regional quantification and spatial normalization of [18F]MNI-659 PET images. We performed [18F]MNI-659 PET imaging in six months old heterozygous (HET) Q175 mice and wild-type (WT) littermates, followed by X-ray computed tomography (CT) scan. In the same week, individual T2-weighted MRI were acquired. Spatial normalization and regional quantification of the PET/CT images was performed on MRI, [18F]MNI-659 PET, or CT template and compared to binding potential (BPND) using volumes manually delineated on the individual MR images. Striatal volume was significantly reduced in HET mice (-7.7%, p<0.0001) compared to WT littermates. [18F]MNI-659 BPND in striatum of HET animals was significantly reduced (p<0.0001) when compared to WT littermates using all three templates. However, BPND values were significantly higher for HET mice using the PET template compared to the MRI and CT ones (p<0.0001), with an overestimation at lower activities. On the other hand, the CT template spatial normalization introduced larger variability reducing the effect size. The PET and CT template-based approaches resulted in a lower accuracy in BPND quantification with consequent decrease in the detectability of disease effect. This study demonstrates that for [18F]MNI-659 brain PET imaging in mice the use of an MRI-based spatial normalization is recommended to achieve accurate quantification and fully exploit the detectability of disease effect.