APOE ε4 gene dose effect on imaging and blood biomarkers of neuroinflammation and beta-amyloid in cognitively unimpaired elderly

Anniina Snellman, Laura L. Ekblad, Jouni Tuisku, Mikko Koivumäki, Nicholas J. Ashton, Juan Lantero-Rodriguez, Thomas K. Karikari, Semi Helin, Marco Bucci, Eliisa Löyttyniemi, Riitta Parkkola, Mira Karrasch, Michael Schöll, Henrik Zetterberg, Kaj Blennow, Juha O. Rinne

Neuroinflammation, characterized by increased reactivity of microglia and astrocytes in the brain, is known to be present at various stages of the Alzheimer's disease (AD) continuum. However, its presence and relationship with amyloid pathology in cognitively normal at-risk individuals is less clear. Here, we used positron emission tomography (PET) and blood biomarker measurements to examine differences in neuroinflammation and beta-amyloid (Aβ) and their association in cognitively unimpaired homozygotes, heterozygotes, or non-carriers of the APOE ε4 allele, the strongest genetic risk for sporadic AD. Sixty 60-75-year-old APOE ε4 homozygotes (n = 19), heterozygotes (n = 21), and non-carriers (n = 20) were recruited in collaboration with the local Auria biobank. The participants underwent 11C-PK11195 PET (targeting 18-kDa translocator protein, TSPO), 11C-PiB PET (targeting Aβ), brain MRI, and neuropsychological testing including a preclinical cognitive composite (APCC). 11C-PK11195 distribution volume ratios and 11C-PiB standardized uptake value ratios (SUVRs) were calculated for regions typical for early Aβ accumulation in AD. Blood samples were drawn for measuring plasma glial fibrillary acidic protein (GFAP) and plasma Aβ1-42/1.40. In our cognitively unimpaired sample, cortical 11C-PiB-binding increased according to APOE ε4 gene dose (median composite SUVR 1.47 (range 1.38-1.66) in non-carriers, 1.55 (1.43-2.02) in heterozygotes, and 2.13 (1.61-2.83) in homozygotes, P = 0.002). In contrast, cortical composite 11C-PK11195-binding did not differ between the APOE ε4 gene doses (P = 0.27) or between Aβ-positive and Aβ-negative individuals (P = 0.81) and associated with higher Aβ burden only in APOE ε4 homozygotes (Rho = 0.47, P = 0.043). Plasma GFAP concentration correlated with cortical 11C-PiB (Rho = 0.35, P = 0.040), but not 11C-PK11195-binding (Rho = 0.13, P = 0.47) in Aβ-positive individuals. In the total cognitively unimpaired population, both higher composite 11C-PK11195-binding and plasma GFAP were associated with lower hippocampal volume, whereas elevated 11C-PiB-binding was associated with lower APCC scores. Only Aβ burden measured by PET, but not markers of neuroinflammation, differed among cognitively unimpaired elderly with different APOE ε4 gene dose. However, APOE ε4 gene dose seemed to modulate the association between neuroinflammation and Aβ.