Association of peripheral inflammatory markers with connectivity in large-scale functional brain networks of non-demented older adults

Background: Systemic inflammation has emerged as a risk factor for cognitive decline and Alzheimer's disease, but inflammation's effect on distributed brain networks is unclear. We examined the relationship between peripheral inflammatory markers and subsequent functional connectivity within five large-scale cognitive networks and evaluated the modifying role of cortical amyloid and APOE epsilon 4 status. Methods: Blood levels of soluble tumor necrosis factor-alpha receptor-1 and interleukin 6 were assessed in 176 participants (at baseline mean age: 65 (SD 9) years; 63% women; 85% cognitively normal, 15% mild cognitive impairment (MCI)) and were combined to derive an Inflammatory Index. Approximately six years later, participants underwent resting-state functional magnetic resonance imaging to quantify functional connectivity; a subset of 137 participants also underwent C-11 Pittsburgh compound-B (PiB) PET imaging to assess cortical amyloid burden. Results: Using linear regression models adjusted for demographic characteristics and cardiovascular risk factors, a higher Inflammatory Index was associated with lower connectivity within the Default Mode (beta=-0.013; 95% CI:-0.023,-0.003) and the Dorsal Attention Networks (beta=-0.017; 95% CI:-0.028,-0.006). The strength of these associations did not vary by amyloid status (positive/negative). However, there was a significant interaction between Inflammatory Index and APOE epsilon 4 status, whereby epsilon 4-positive participants with a higher Inflammatory Index demonstrated lower connectivity. Inflammatory Index was unrelated to connectivity within other large-scale cognitive networks (Control, Limbic, and Salience/Ventral Attention networks). Conclusion: Peripheral pro-inflammatory signaling in older adults without dementia, especially among APOE epsilon 4-positive individuals, is associated with altered connectivity within two large-scale cognitive networks.