Comparison of group‐level and individualized ROIs for predicting change in longitudinal tau‐PET in preclinical and prodromal AD. (20th December 2022)

• Record Type: Journal Article
• Authors: Leuzy, Antoine; Binette, Alexa Pichet; Vogel, Jacob W; Klein, Gregory; Borroni, Edilio; Tonietto, Matteo; Strandberg, Olof; Mattsson‐Carlgren, Niklas; Palmqvist, Sebastian; Pontecorvo, Michael J.; Stomrud, Erik; Ossenkoppele, Rik; Hansson, Oskar
• Is Part Of: Alzheimer's & dementia. Volume 18(2022)Supplement 10
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-20
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.063057
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Longitudinal tau‐PET is increasingly used as an outcome in clinical trials evaluating disease‐modifying therapies in Alzheimer's disease (AD). In order to quantify change in tau‐PET signal over time, regions of interest (ROIs) are used; to date, these have been based on neuropathological studies or data‐driven approaches where the same ROI is used for each subject (group‐level ROI). However, given the inter‐individual heterogeneity in spatial patterns of tau‐PET, a key question is whether the use of subject‐specific (individualized) ROIs might offer any advantages over group‐level ROIs in AD clinical trials. We aimed to address this question by comparing group‐level and individualized ROIs across several metrics using longitudinal tau‐PET in preclinical and prodromal AD. Methods: We included 174 participants with [ 18 F]RO948 (97 preclinical AD [Aβ‐positive cognitively unimpaired], 77 prodromal AD [Aβ‐positive MCI] from BioFINDER‐2; average follow‐up 1.90±0.50 years). Group‐level ROIs included i) Braak and ii) event‐based modelling (EBM) stages and iii) temporal and iv) whole‐brain meta‐ROIs. Individualized ROIs included i) epicenter (regions with earliest tau), ii) Q1 (quartile closest to subject‐specific epicenter based on functional connectivity) and iii) probability‐based approach (regions selected based on their probability of being tau‐positivity at baseline using Gaussian mixture modelling). Group‐level and individualized ROIs were compared using yearly percent‐change in SUVR (ΔSUVR) and required sample size in a simulated intervention that reduces tau accumulation by 20, 30 or 40%. Results: In preclinical AD (Figure 1A), EBM stage‐I had the greatest ΔSUVR among group‐level ROIs (4.27%). Based on 95% confidence intervals, ΔSUVR was significantly higher using the probability‐based approach (5.51%). A similar pattern was seen in prodromal AD (Figure 1B), where ΔSUVR for the best individualized ROI (probability‐based, 7.86%) was significantly greater than that for the best group‐level ROI (EBM stage‐II, 5.45%). Across interventions (20/30/40%), sample size differences were on average ∼6% lower in preclinical AD and ∼20% lower in prodromal AD using individualized ROIs, relative to group‐level ROIs (Table 1). Further analysis are ongoing using a validation cohort (n = 278) scanned with [ 18 F]flortaucipir. Conclusion: Preliminary findings using [ 18 F]RO948 suggest that individualized ROIs carry an advantage over group‐level ROIs in AD clinical trials using longitudinal tau‐PET as outcome.