Amyloid aggregation and subsequent memory decline over time in cognitively intact older identical twins

Abstract

AbstractBackgroundOlder individuals with intact cognition and pathological amyloid depositions are at increased risk for memory decline. At this point the precise biological processes leading to cognitive decline are unclear. We tested the role of genetic influences on the relationship between amyloid pathology and subsequent memory decline over time using a monozygotic twin approach, and whether these relationships were dependent on the methodology used to determine amyloid aggregation.MethodWe selected 78 monozygotic twins with at baseline normal cognition from the EMIF‐AD PreclinAD study (Table 1), who had completed 4 years of follow‐up. We defined baseline amyloid status using either visual read of dynamic [18F]flutemetamol PET images or CSF amyloid‐β 1‐42/1‐40 (Aβ42/40) ratio (ADx Neurosciences/Euroimmun assays) <0.066 (based on Gaussian mixture modelling). Memory was tested at baseline, 2‐year and 4‐year, with six tests combined into one composite score. Associations between baseline amyloid status and subsequent decline on memory was tested using linear mixed models with main effect amyloid status, time and amyloid*time, adjusted for age, sex, education and genetic relatedness. We repeated analyses taking continuous CSF Aβ42/40 ratio and PET binding potential (BPND) values as predictors. To examine the role of genetic influences on observed associations we performed cross‐twin cross‐trait analysis.ResultNine individuals had abnormal Aβ at baseline and did not differ in age, sex and education from Aβ‐ subjects. They had a faster decline in memory than Aβ‐ subjects (ß=‐0.127(SE=0.041), pamyloid*time=0.002, Figure 1). Continuous measures of amyloid burden also predicted memory decline at follow‐up (CSF Aβ42/40 ratio ß=0.055(SE=0.016), PET BPND ß=‐0.040(SE=0.014), both p<0.005). Baseline CSF Aβ42/40 ratio in one twin could predict memory decline in its co‐twin (r=0.31,p=0.04), which suggests similar genetic factors underlie these processes, however this relation did not reach significance when using amyloid‐PET BPND (r=‐0.12,p=0.35).ConclusionIn our study CSF Aβ42/40 ratio was more sensitive for detecting early pathophysiological changes compared to PET BPND. A larger sample is needed to confirm this finding. Our twin approach suggests that in preclinical AD amyloid burden and memory performance may share common genetic pathways. The next step will be to identify what these underlying shared biological mechanisms are.