Validation Of The SUVr As A Non-invasive Quantification Method Of [¹¹C]UCB-J Binding In Alzheimer’s Disease

Inès Arino¹, Albertine Dubois¹, Pauline Testard², Pauline Olivieri², Matteo Tonietto³, Gregory Klein³, Philippe Gervais¹, Antoine Pinon¹, Thibaud Lebouvier⁴, David Wallon⁵, Bastian Zinnhardt³, Irene Gerlach³, Myriam Ly Le Moal⁶, Alexandra Petiet⁶, Guillaume Dorothée⁷, Marie-Claude Potier⁸, Aurélie Kas⁹, Michel Bottlaender¹, Marie Sarazin², Julien Lagarde²

1 Université Paris-Saclay, BioMaps, CEA, Inserm Service Hospitalier Frédéric Joliot, Orsay, France,
2 Neurologyof Memory and Language, GHU Paris Psychiatrie & Neurosciences - Sainte-Anne Hospital, Paris, France,
3 Roche Pharma Research and Early Development, Neuroscience and Rare Diseases, Roche Innovation Center, Basel, Switzerland,
4 Univ Lille, Inserm U1172, Lille, France ; CHU de Lille, DISTALZ, Lille, France,
5 Normandie Univ, UNIROUEN, Inserm U1245; CHU Rouen, Department of Neurology and CNR-MAJ, Rouen,France,
6 Institut Roche, 4 cours de l'île Seguin, Boulogne Billancourt, France,
7 Sorbonne Université, Inserm ; Centre de Recherche Saint-Antoine, CRSA, Immune System and Neuroinflammation Laboratory, Hôpital Saint-Antoine, Paris, France,
8 Paris Brain Institute (ICM), Centre National de la Recherche Scientifique (CNRS)UMR 7225, INSERM U1127, Hôpital de la Pitié-Salpêtrière, Sorbonne Université, Paris, France,
9 Service demédecine nucléaire, Hôpital Pitié-Salpêtrière, Paris, France.

Abstract:
Background:
The non-invasive quantification method of [¹¹C]UCB-J binding (that reflects synaptic density) based on the standardized uptake value ratio (SUVr) has so far been relatively uncommon in Alzheimer’sDisease (AD). It has rarely been compared with other methods considered more reliable, such as the distribution volume ratio (DVR), which requires acquisition of the entire dynamic, resulting in much longer acquisition times that are often difficult for patients to tolerate. We compared SUVr with DVR using different reference regions in order to validate its use in AD.
Methods:
Forty-four patients with a clinical-biological diagnosis of AD (CSF biomarkers) and 26 age-matched healthy controls were included and underwent [¹¹C]UCB-J PET imaging (Table 1). Dynamic PET acquisition lasted 70 min. The SUVr was calculated between 50 and 70 minutes post-injection. We considered three reference regions: the centrum semi-ovale (CS), the whole cerebellum (WC), and the 2mm-eroded white matter (WM). We defined 14 cortical meta-regions-of-interest (left and right) covering the entire associative cortex. Correlations between DVR and SUVr were analyzed in all participants combined. We also compared the ability of DVR and SUVr obtained with the different reference regions to distinguish AD patients and controls based on the regional effect sizes (expressed as Cohen's d).
Results:
By combining all participants and regions-of-interest, we found strong correlations between SUVr and DVR (r=0.92, p<0.001), with all reference regions (Figure 1). Significantly lower SUVr and DVR values were found in patients compared to controls, particularly in the lateral/medial parietal, and lateral temporal cortices, with all reference regions. Cohen's d values were higher in DVR than in SUVr when CS was the reference region, with no significant difference with WC and WM. For SUVr, Cohen's d values were higherwith WC and WM than with CS (Figure 2).
Conclusion:
This study validates the use of SUVr (50-70 min) in AD. The WC appears to be the most appropriate reference region, and WM, which has several advantages over CS, has characteristics very similar to those of WC, and could be considered, especially when a true reference region (i.e., with no or low specific binding) is required.