Noninvasive Relative Quantification of [11C]ABP688 PET Imaging in Mice Versus an Input Function Measured Over an Arteriovenous Shunt

Jeroen Verhaeghe, Daniele Bertoglio, Lauren Kosten, David Thomae, Marleen Verhoye, Annemie Van Der Linden, Leonie Wyffels, Sigrid Stroobants, John Wityak, Celia Dominguez, Ladislav Mrzljak, Steven Staelens

Impairment of the metabotropic glutamate receptor 5 (mGluR5) has been implicated with various neurologic disorders. Although mGluR5 density can be quantified with the PET radiotracer [11C]ABP688, the methods for reproducible quantification of [11C]ABP688 PET imaging in mice have not been thoroughly investigated yet. Thus, this study aimed to assess and validate cerebellum as reference region for simplified reference tissue model (SRTM), investigate the feasibility of a noninvasive cardiac image-derived input function (IDIF) for relative quantification, to validate the use of a PET template instead of an MRI template for spatial normalization, and to determine the reproducibility and within-subject variability of [11C]ABP688 PET imaging in mice. Blocking with the mGluR5 antagonist MPEP resulted in a reduction of [11C]ABP688 binding of 41% in striatum (p < 0.0001), while no significant effect could be found in cerebellum (-4.8%, p > 0.99) indicating cerebellum as suitable reference region for mice. DVR-1 calculated using a noninvasive IDIF and an arteriovenous input function correlated significantly when considering the cerebellum as the reference region (striatum: DVR-1, r = 0.978, p < 0.0001). Additionally, strong correlations between binding potential calculated from SRTM (BPND) with DVR-1 based on IDIF (striatum: r = 0.980, p < 0.0001) and AV shunt (striatum: r = 0.987, p < 0.0001). BPND displayed higher discrimination power than VT values in determining differences between wild-types and heterozygous Q175 mice, an animal model of Huntington's disease. Furthermore, we showed high agreement between PET- and MRI-based spatial normalization approaches (striatum: r = 0.989, p < 0.0001). Finally, both spatial normalization approaches did not reveal any significant bias between test-retest scans, with a relative difference below 5%. This study indicates that noninvasive quantification of [11C]ABP688 PET imaging is reproducible and cerebellum can be used as reference region in mice.