Neuroinflammation plays a central role in a variety of neurological diseases. Over the last decade, PET imaging of neuroinflammation proved useful for non-invasive identification and quantification of the inflammation cascade. [18F]DPA714 is the most promising ligand pushed forward into clinical studies. This radiotracer binds with a high specificity to the 18kDa Translocator Protein (18kDa TSPO), a molecular biomarker of neuroinflammation. Nevertheless, quantification of TSPO-PET remains a topic under debate because of the complex binding properties of TSPO-PET ligands. For clinical practice, fully non-invasive quantification methods without any additional blood measurements are preferable, but their diagnostic performance has never been compared.

Researchers of the CNRS Laboratory of Neurodegenerative Diseases at MIRCen created an animal model with controlled and localized TSPO expression using a viral-vector strategy, on which they performed a [18F]DPA714 PET scan before 3-D histological staining of TSPO in the entire brain. The PET images were analyzed according to 9 different quantification methods.  When compared to the reference TSPO-histological 3D brain, each PET voxel was classified as true or false, positive or negative. Using this approach, it was shown that standard uptake value ratios using the cerebellum as a reference region has the most optimal score amongst all non-invasive quantification methods.

This study will help guide future clinical studies that increasingly use PET-TSPO ligands.

Immunohistochemical analysis of the lenti-CNTF NHP model. Immunostainings for TSPO (a, b) and VIM (d, e) were rendered three-dimensional, which allowed to appreciate a major overlap between VIM-IHC (f) and TSPO-IHC (c). Higher magnifications confirmed that TSPO-IHC was only detected in activated astrocytes (g, h, i), and not in the microglia (j, k, l).