DOI: 10.1002/alz.075194 ISSN: 1552-5260

Associations of peripheral and neuronal insulin resistance with cognition and functional brain networks in older adults with obesity

Clayton Connor McIntyre, Robert Lyday, Shannon L Macauley, Gagan Deep, Yixin Su, Barbara Nicklas, Christina E Hugenschmidt
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Geriatrics and Gerontology
  • Neurology (clinical)
  • Developmental Neuroscience
  • Health Policy
  • Epidemiology

Abstract

Background

Insulin resistance (IR) in peripheral tissue is a risk factor for Alzheimer’s disease. A proposed method for isolating neural‐enriched extracellular vesicles (nEVs) may allow measurement of neuronal IR in living humans. The hippocampus may be particularly sensitive to insulin signaling. We assessed associations of peripheral and neuronal IR with cognition and resting‐state functional brain networks in sedentary adults aged 65+ with obesity.

Methods

Analyses were performed on baseline data in a subset of participants (n = 75) from the INFINITE study who had MMSE ≥ 24 and had obesity but not type 2 diabetes. HOMA2‐IR quantified peripheral IR. The proposed insulin resistance index for nEVs (nEV‐IR) quantified neuronal IR. Executive function was measured by a composite score calculated from the digit symbol coding task, Trail Making Test parts A and B, Stroop task, phonemic fluency, and semantic fluency. Immediate learning was measured by the summed score of Auditory Verbal Learning Test (AVLT) trials 1‐5. Associations between cognition and HOMA2‐IR or nEV‐IR were evaluated using linear regression adjusted for age, sex, education, and BMI. Functional networks were created from resting‐state fMRI. Associations between brain network features in the hippocampus (global efficiency (Eglob), local efficiency (Eloc), first‐step connections, community structure) and HOMA2‐IR or nEV‐IR were evaluated using a novel regression framework for brain network distance metrics. Brain network regression models controlled for age, sex, education, and BMI.

Results

HOMA2‐IR and nEV‐IR were not correlated (p = .682). HOMA2‐IR was negatively associated with the executive function composite (β = ‐0.752, SE = 0.329, p = .026) but not AVLT trials 1‐5 (p = .185) or MMSE (p = .099). nEV‐IR was not associated with the executive function composite (p = .861), AVLT trials 1‐5 (p = .725), or MMSE (p = .262). HOMA2‐IR was positively associated with hippocampal Eloc (β = 0.0035, SE = 0.0014, p = .011) but not Eglob (p = .724). nEV‐IR was not associated with Eloc (p = .782) or Eglob (p = .752). Neither HOMA2‐IR nor nEV‐IR was associated with hippocampal connectivity (HOMA2‐IR: p = .419, nEV‐IR: p = .187) or community structure (HOMA2‐IR: p = .670, nEV‐IR: p = .404).

Conclusion

Measures of peripheral – but not neuronal – IR are associated with unfavorable cognitive outcomes and with higher local efficiency in the hippocampus. These findings add to ongoing debate over nEV‐IR as a neuronal IR assessment.

More from our Archive