Comparison of reference region stability for longitudinal amyloid PET in Down syndrome

Abstract Background The cerebellum is frequently used as the reference region for amyloid PET analysis. However, this reference region has been shown to demonstrate longitudinal variability, particularly with [ 18 F]florbetapir (FBP) PET (Landau, JNM 2015). For investigations in individuals with Down syndrome (DS), cerebellar atrophy and rapid disease progression may increase these longitudinal variabilities. Although white matter possesses different non‐displaceable uptake properties, the relative lack of specific binding makes white matter a suitable reference region for longitudinal studies. This work compares the observed longitudinal change when using whole cerebellum and white matter reference regions in [ 18 F]FBP and [ 11 C]PiB scans of adults with DS. Method Participants with DS, recruited through the ABC‐DS study, underwent longitudinal PiB or FBP PET imaging and T1w MRIs (Table 1 lists cohort differences). PET images were smoothed to 8mm resolution, summed 50‐70 min, co‐registered with the MRI, and normalized to a common DS MRI template (LeMerise, 2022). GAINN whole cerebellum (WC) VOI was applied to create SUVR WC . Whole brain white matter was segmented in native space using SPM, smoothed to PET resolution, and eroded to 90% tissue probability. The resulting eroded white matter (EWM) mask was used as reference to create SUVR EWM . Average SUVR was calculated for GAINN global cortex (CTX). Longitudinal scans were assessed for correlations between reference region strategies and average rate of SUVR change: % Change/year = (SUVR 2 ‐SUVR 1 )/(SUVR 1 *Δt). Result Figure 1 displays the averaged EWM reference template. Figure 2 displays longitudinal PET data and regressions between SUVRs. Across participants, SUVR WC shows 78/90 (PiB) and 50/83 (FBP) between‐scan increases. SUVR EWM shows 66/90 (PiB) and 71/83 (FBP) between‐scan increases. For A+ individuals (18CL cutoff), the average difference (% Change EWM ‐ % Change WC )/year is ‐0.5%/year [‐1.2,0.3] (PiB) and 1.9%/yr [0.5,3.3]** (FBP). FBP group SD in % Change/year decreases from 5.6% (WC) to 2.9% (EWM). Conclusion As observed in LOAD, SUVR EWM demonstrates lower group variability and greater longitudinal change in FBP. SUVR EWM shows strong agreement with SUVR WC in PiB. These data suggest that an EWM reference region can reduce variability in longitudinal FBP studies in DS.