Age-Related Regional Network Covariance of Magnetic Resonance Imaging Gray Matter in the Rat

Abstract

Healthy human aging has been associated with brain atrophy in prefrontal and selective temporal regions, but reductions in other brain areas have been observed. We previously found regional covariance patterns of gray matter with magnetic resonance imaging (MRI) in healthy humans and rhesus macaques, using multivariate network Scaled Subprofile Model (SSM) analysis and voxel-based morphometry (VBM), supporting aging effects including in prefrontal and temporal cortices. This approach has yet to be applied to neuroimaging in rodent models of aging. We investigated 7.0T MRI gray matter covariance in 10 young and 10 aged adult male Fischer 344 rats to identify, using SSM VBM, the age-related regional network gray matter covariance pattern in the rodent. SSM VBM identified a regional pattern that distinguished young from aged rats, characterized by reductions in prefrontal, temporal association/perirhinal, and cerebellar areas with relative increases in somatosensory, thalamic, midbrain, and hippocampal regions. Greater expression of the age-related MRI gray matter pattern was associated with poorer spatial learning in the age groups combined. Aging in the rat is characterized by a regional network pattern of gray matter reductions corresponding to aging effects previously observed in humans and non-human primates. SSM MRI network analyses can advance translational aging neuroscience research, extending from human to small animal models, with potential for evaluating mechanisms and interventions for cognitive aging.

Keywords: aging; behavior; perirhinal cortex; prefrontal cortex; scaled subprofile model; structural covariance.

Figure 1

Behavioral performance comparing young adult (n = 10) and aged (n = 10) Fischer 344 male rats on the spatial Morris swim task. The bar graph shows means and standard error of the means (SEM) in the young (white bars) and old (gray bars) groups for the spatial swim task for each of the four consecutive days of testing indicated on the x-axis. Performance on the corrected integrated path length (CIPL) measured in meters*seconds (m*s) is shown on the y-axis. *After adjusting for visual swim task performance averaged across 2 days of testing, a mixed-design analysis of covariance (ANCOVA) tested for age group, days of testing, and age group × days of testing interaction effects for the spatial swim task. The aged rats showed poorer spatial learning compared to the young adult rats (p = 0.036) and both groups showed increasingly shorter CIPL distances traveled over the 4 days of testing (p = 1.30E-6), but there was no group by days of testing interaction (p = 0.48).

Figure 2

Scaled Subprofile Model (SSM) subject scores from the network analysis of magnetic resonance imaging (MRI) voxel-based morphometry (VBM) in young adults (n = 10) and aged (n = 10) Fischer 344 male rats. The scatterplot shows that the old group had a higher expression of the MRI age-related network pattern than the young group. The group difference between the young and aged rats in the SSM analysis of the MRI VBM gray matter maps was tested with a multiple regression model using the Akaike Information Criterion (AICc), which selected the first SSM component that accounted for 85.5% of the variance in distinguishing the young from aged groups (r = 0.93, p ≤ 5.52E-9).

Figure 3

Magnetic resonance imaging (MRI) gray matter pattern reflecting the first Scaled Subprofile Model (SSM) component whose subject scores predicted age group in the Fischer 344 male rats. Voxels with SSM pattern weights are superimposed on horizontal slices from the statistical parametric mapping (SPM8) average MRI scans. The blue end of the color scale indicates brain regions showing lower gray matter volume with older age, whereas the orange end of the scale shows areas of relatively increased gray matter with increasing age. A rat with a high positive score for this age-related pattern has relatively greater reductions in the blue areas and relatively greater covarying increases in the orange areas. Only voxels with z scores ≤ −3.5 or ≥ + 3.5 and a 50 voxel extent after bootstrap re-sampling with 500 iterations to provide robust regional pattern weights are shown. Notable regions of reduction (blue) with older age are observed (from top-bottom rows) mainly in the vicinities of the bilateral temporal association/perirhinal, prefrontal/insula, olfactory bulb, and cerebellum; whereas relative increases (orange) are seen in the vicinities of the bilateral somatosensory, selected hippocampal, septal, thalamus, and midbrain regions. That all the aged rats had higher network age pattern scores than the young adult rats and in the positive range is consistent with a higher expression of the hypothesized age-related regional pattern in that group, including greater reductions in the blue areas than in the young.

Figure 4

Regression analyses showing the association between behavioral performance on a corrected integrated path length (CIPL) learning index score of the spatial Morris swim task with the subject scores from the age-related network Scaled Subprofile Model (SSM) pattern. The learning index is derived from subtracting average CIPL performance on Day 4 from Day 1 and dividing by the sum of the 2 days. The aged rats are shown by filled circles and the young rats have open circles in the scatterplot. A higher expression of the network age pattern is associated with poorer task performance for both the young and old groups combined, accounting for 26% of the variance (r = −0.51, p ≤ 0.02), demonstrating that the age-related variance in the MRI network pattern is related to learning performance across the age-groups.