Association between brain [18F]FDG‐PET signal and genetic cell markers in the brain tissue
Luiza Santos Machado, Marco Ant, De Bastiani, Andreia Silva da Rocha, Bruna Bellaver, Carolina Soares, Guilherme Povala, Nesrine Rahmouni, Débora Guerini de Souza, Tharick A. Pascoal, Diogo O. Souza, Pedro Rosa‐Neto, Andrea Lessa Benedet, Kaj Blennow, Henrik Zetterberg, Nicholas J. Ashton, Pamela C.L. Ferreira, Eduardo R. Zimmer- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Brain glucose hypometabolism, indexed by [18F]FDG‐PET, is considered a biomarker of neurodegeneration in Alzheimer’s disease (AD). However, other brain cells, such as astrocytes and microglia, also consume considerable amounts of glucose and may contribute to the [18F]FDG‐PET signal in the brain. Thus, the cellular source of the [18F]FDG‐PET signal remains controversial. Here, we aimed to evaluate whether canonical markers of different brain cell types (neuron, oligodendrocyte, astrocyte, and microglia) associate with brain [18F]FDG‐PET signal in humans and a rat model of human amyloidosis.
Method
[18F]FDG‐PET images were acquired in ten‐month‐old APP/PS1 (TgF344‐AD, n = 8) and wild‐type (WT, n = 6) rats. Next, we evaluated the gene expression of NeuN, Olig2, GFAP, S100B, and IBA1 in the frontal and temporoparietal cortices and cerebellum. Association maps integrating mRNA with brain [18F]FDG‐PET signal were conducted at the voxel level using RMINC. [18F]FDG‐PET and brain tissue mRNA (GFAP, ACSA2, S100B, GLAST, IBA1, CSF1R, TREM2, Olig2, Cldn11, NeuN) of cognitively unimpaired patients were obtained from ADNI and Allen Brain Atlas database, respectively. In humans, mean regional [18F]FDG‐PET SUVr (reference region: pons) and mRNA expression were used to evaluate their association using a gaussian model. Differences were considered statistically significant at p<0.05 (t>2).
Result
GFAP and S100B mRNA levels in the temporoparietal and frontal cortices positively correlated with brain [18F]FDG‐PET signal (local maxima, GFAPt(13) = 9.4, S100Bt(13) = 6.6; Fig1B‐C). No associations were found with TREM2, Olig2 and NeuN (t(13)<2; Fig1D‐F). We expanded our analysis by comparing gene astrocyte markers and astrocyte gene cell markers in humans. At the regional level, we found that FDG‐PET SUVr and astrocyte mRNA were positively associated: S100B (R‐squared = 0.34), GFAP (R‐squared = 0.33), GLAST (R‐squared = 0.31) and ACSA‐2 (R‐squared = 0.31; Fig1G).
Conclusion
Our findings suggest astrocyte mRNA markers are more closely associated with brain [18F]FDG‐PET signal than neuronal, oligodendrocyte, and microglial mRNA markers.