Evaluating the global cortical Centiloid value for predicting functional decline

Abstract

AbstractBackgroundPrevious results on amyloid‐PET imaging for predicting functional decline are conflicting. Given that the Centiloid approach (CL) may serve for inclusion in anti‐amyloid trials, we investigated the extent to which CL may predict functional decline in the AMYPAD‐Prognostic and Natural History Study (PNHS), including 1094 participants.MethodWe analyzed, cross‐sectionally and longitudinally, the associations between CL and global and sum‐of‐boxes (SOB) Clinical Dementia Rating (CDR), and Instrumental Activity of Daily Living (IADL) scores. Mean follow‐up duration ranged from 2.7‐3.0 years (±1.1‐1.6) depending on the functional score. Longitudinal CDR‐SOB and IADL changes were modeled using linear mixed‐effects models with random intercepts and slopes, covarying age and sex, stratified by CDR global score. Analyses included comparisons between three CL groups (Aβ‐: CL<12; grey‐zone: CL = 12‐50; Aβ+: CL>50) and correlation between the baseline CL and the time for progression on the CDR global score.ResultThe baseline CL was significantly higher in CDR = 0 (Mdn = 6.6) than in CDR = 0.5 participants (Mdn = 17.2; p<.001). Baseline CDR‐SOB was significantly higher in Aβ+ (M = 0.48) than in Aβ‐ (M = 0.12) and grey‐zone participants (M = 0.13, p‐values <.001). Baseline IADL was (marginally) significantly lower in Aβ+ (Mdn = 81.3) than in Aβ‐ (Mdn = 82.8, p = .02) and grey‐zone participants (Mdn = 83.5, p = .08, Table 1). Baseline CL significantly predicted subsequent changes in CDR‐SOB and IADL, both in CDR = 0 and CDR = 0.5 participants, but only in the CL>50 groups (βCL12‐50 = 0.03, p = .32,βCL>50 = 0.14, p<.001 for CDR‐SOB in CDR = 0 participants;βCL12‐50 = 0.19, p = .31,βCL>50 = 1.04, p = .01 for CDR‐SOB in CDR = 0.5 participants;βCL12‐50 = ‐0.002, p = .99, βCL>50 = ‐1.00, p = .08 for IADL in CDR = 0 participants; Fig.1). The participants who progressed from global CDR = 0 to CDR = 0.5 (N = 90) had a higher baseline CL (Mdn = 15.16) than stable participants (Mdn = 5.95, p = .002). However, the baseline CL was not associated with a shorter time of clinical progression (rs= ‐.02, p = .8; Fig.2).ConclusionCross‐sectionally, amyloid burden expressed in CL units was associated with functional outcomes. A baseline CL>50 predicted subsequent functional decline in CDR‐SOB and IADL scores in clinically normal and mildly clinically impaired participants. At lower CL values, no functional decline was observed during a three‐year follow‐up. Clinically normal individuals with increasing global CDR scores had higher baseline CL values, but these values were not associated with faster clinical progression.