Brain regions vulnerable and resistant to aging without Alzheimer's disease

Abstract

'Normal aging' in the brain refers to age-related changes that occur independent of disease, in particular Alzheimer's disease. A major barrier to mapping normal brain aging has been the difficulty in excluding the earliest preclinical stages of Alzheimer's disease. Here, before addressing this issue we first imaged a mouse model and learn that the best MRI measure of dendritic spine loss, a known pathophysiological driver of normal aging, is one that relies on the combined use of functional and structural MRI. In the primary study, we then deployed the combined functional-structural MRI measure to investigate over 100 cognitively-normal people from 20-72 years of age. Next, to cover the tail end of aging, in secondary analyses we investigated structural MRI acquired from cognitively-normal people, 60-84 years of age, who were Alzheimer's-free via biomarkers. Collectively, the results from the primary functional-structural study, and the secondary structural studies revealed that the dentate gyrus is a hippocampal region differentially affected by aging, and that the entorhinal cortex is a region most resistant to aging. Across the cortex, the primary functional-structural study revealed and that the inferior frontal gyrus is differentially affected by aging, however, the secondary structural studies implicated other frontal cortex regions. Together, the results clarify how normal aging may affect the brain and has possible mechanistic and therapeutic implications.

Fig 1. Mapping dendritic spine loss with CBV-fMRI and volumetric MRI.

(A) 3D mouse cortical surface maps of the ex vivo neuron density (left image) and in vivo cerebral blood volume (CBV) density (right image). Cortical surface maps were generated from the template images using our proposed MouseStream software suite by projecting the mean density along each streamline (see Method for details). (B) CBV density map in a single coronal slice of the right hemisphere (dashed line in panel A right image) going through the barrel cortex (left image) and the zoom-in view of the barrel cortex (right image). The CBV density map was generated as the population average of 14 co-registered CBV scans. The whole cortex was delineated evenly into 8 depths by solving the Laplace’s Equation. Cortical layers were defined using the Allen Brain Atlas anatomical CCF v3 template. Individual barrel modules with high basal metabolism were observed at layer-IV of the barrel cortex. (C) Result of voxel-based analysis of percent CBV density (%CBV) across the whole cortex. The t-value distribution of CBV decrease during 30-day trimming period across cortical regions shows that the barrel cortex on the control side (right hemisphere) has significant CBV decreases (p<0.05, cluster>20) and the barrel cortex contralateral to the trimmed side (left hemisphere) has no clear CBV decreases. (D) Total CBV (tCBV) and volume changes over 30-day trimming period in the depth 1 (belongs to layer I) of the barrel cortex. At the depth that corresponds to layer I, both tCBV and volume have significant decreases on the control side and whisker trimming significantly reduces both tCBV and volume decreases over 30 days on the trimmed side. (E) Total CBV (tCBV) and volume changes over 30-day trimming period in the depth that corresponds to layer IV of the barrel cortex. Both tCBV and volume have no significant changes on the control side and there is no significant whisker trimming effect on tCBV or volume changes over 30 days on the trimmed side. (F) Changes in total CBV (tCBV) is tightly correlated with changes in volume over the 30-day trimming period at the depth that corresponds to layer I of the barrel cortex (R = 0.7187, p = 0.0038). (G) Changes in total CBV (tCBV) is independent of changes in volume over 30-day trimming period at the depth that corresponds to layer IV of the barrel cortex (R = 0.0045, p = 0.9878).

Fig 2. Mapping aging with CBV-fMRI from 20–72 years of age.

(A) A vertex-based analysis of the cortex (VBA; left image) and a region-of-interest analysis across cortical regions (ROI; right image) identified the greatest age-related decrease in cerebral blood volume (CBV) in the inferior frontal gyrus. (B) The t-value distribution of age-related CBV decreases across cortical regions shows that two regions of the inferior frontal gyrus (indicated in red, the pars orbitalis and the pars triangularis) are most reliably vulnerable to aging. The entorhinal cortex (indicated in blue) was found most resistance to aging. The dashed red line indicates the t-value threshold at α = 0.05 adjusted for Šidák multiple comparison. (C) A voxel-based analysis of the hippocampus (VBA; left image) and a region-of-interest analysis across hippocampal regions (ROI; right image) identified the greatest age-related CBV decrease in the dentate gyrus. (D) The t-value distribution of age-related CBV decrease across hippocampal regions, shows that the dentate gyrus (indicated in red) is most reliably vulnerable to aging. The dashed red line indicates the t-value threshold at α = 0.05 adjusted for Šidák multiple comparison.

Fig 3. Mapping aging with volumetric MRI from 20–72 years of age and its relationship to CBV-fMRI.

(A) The t-value distribution of age-related decrease in volume, measured by structural MRI, across cortical regions in 20-72-year old healthy subjects (indicated in red is the arrow illustration above) While not most reliably affected, volume in regions of the inferior frontal gyrus (indicated in red) show a decrease with age. The volume of the entorhinal cortex (indicated in blue) was found most resistance to aging. The dashed red line indicates the t-value threshold at α = 0.05 adjusted for Šidák multiple comparison. (B) A significant relationship between CBV and volume observed for the dentate gyrus and the inferior frontal gyrus, a joint MRI biomarker which best correlates with dendritic spine loss, but not for the entorhinal cortex.

Fig 4. Mapping aging with volumetric MRI from 62–85 years of age and across the age-span.

(A) The t-value distribution of age-related decreases in volume across cortical and hippocampal region in Alzheimer’s-free 62–85 year old subjects (indicated in red in the arrow illustration above), shows that the dentate gyrus is most vulnerable to aging (although not crossing threshold of multiple comparisons) and the inferior frontal gyrus (indicated in red, left graph) is not reliably associated with aging. The entorhinal cortex (indicated in blue, left graph) is the region least affected by aging. The dashed red line indicates the t-value threshold at α = 0.05 adjusted for Šidák multiple comparison. (B) The t-value distribution of age-related decrease in volume across cortical and hippocampal region in Alzheimer’s-free subjects across the full age-span, of both the ADNI and Columbia aging cohort, (indicated in red in the arrow illustration above), shows that the dentate gyrus (indicated in red, right graph) is most vulnerable to aging and the inferior frontal gyrus (indicated in red, left graph) is reliably associated with aging. The entorhinal cortex (indicated in blue, left graph) is the region least affected by aging. The dashed red line indicates the t-value threshold at α = 0.05 adjusted for Šidák multiple comparison.

Fig 5. Trajectories of brain regions vulnerable and resistant to normal aging across the age-span.

(A) The aging trajectory of dentate gyrus volume (left image) shows a linear decrease across the age-span. The trajectory of inferior frontal gyrus volume (middle image) shows a curvilinear decrease. The trajectory of entorhinal cortex volume (right image) shows that it unaffected by aging across the age-span. (B) A graphic summary of the two regions differentially vulnerable to normal aging, the dentate gyrus and the inferior frontal gyrus (red), and the region most resistant to normal aging, the entorhinal cortex (blue).