Amyloid‐PET Centiloid quantification predicts cognitive functioning in a pre‐dementia population: findings from AMYPAD‐PNHS

Abstract

AbstractBackgroundThe prevention of symptomatic Alzheimer’s disease (AD) is a major endeavor currently in the field, with the prevalence of subjects on the AD continuum (i.e. presence of amyloid‐β [Aβ] pathology) being much higher than previously estimated. However, the extent to which Aβ pathology and its interaction with common risk factors in the aging population relates to disease progression is still limited. The AMYPAD Prognostic and Natural History Study (PNHS) aims to evaluate the value of amyloid‐PET for predicting AD‐related disease progression.MethodA total of 1423 non‐demented subjects were included from the AMYPAD‐PNHS pan‐European cohort (17 sites across 10 Parent Cohorts), for who a baseline [18F]flutemetamol or [18F]florbetaben amyloid‐PET scans that passed quality control was available (Table‐1). Amyloid burden was assessed using the Centiloid (CL) method (whole cerebellum reference region). Independent linear mixed‐effect models were used to assess the predictive value of continuous CL, and its interaction with time, across cognitive domains: mini‐mental state examination (MMSE), memory, attention, language fluency, visuospatial, and executive functions. Years from baseline, age, sex, years of education, cognitive dementia rating (CDR), and APOE‐ε4 were used as independent variables. Clinical follow‐up was 3.4 ± 1.8 years (up to ±10 years) on average.ResultAt baseline, 90% of the subjects were classified as cognitively unimpaired (CDR = 0). Based on CL, most subjects were categorized as amyloid‐negative (59%, 0±9CL), followed by grey‐zone (24%, 25±11CL), and amyloid‐positive (17%, 82±21CL). Baseline CL and its interaction with time were predictive of MMSE scores, memory, attention, and language fluency. Baseline CL, but not its interaction with time, was predictive of executive function. Finally, the interaction of baseline CL with time, but not its main effect, was predictive of visuospatial functioning (Figure‐1). Age and sex were significant contributors to nearly all models.ConclusionHigher baseline CL was associated with lower cognitive performance in all explored domains, suggesting that emerging cerebral amyloid accumulation in non‐demented individuals is predictive of overall cognitive function. Also, higher CL values were predictive of decline across cognitive domains, except for executive function. Further analysis is ongoing, including different confounding factors, longitudinal PET outcomes, and other biomarkers of the disease.