Longitudinal Patterns of Brain Changes in a Community Sample in Relation to Aging and Cognitive Status

Abstract

Background:

Aging and Alzheimer’s disease (AD) are characterized by widespread cortical and subcortical atrophy. Though atrophy patterns between aging and AD overlap considerably, regional differences between these two conditions may exist. Few studies, however, have investigated these patterns in large community samples.

Objective:

Elaborate longitudinal changes in brain morphometry in relation to aging and cognitive status in a well-characterized community cohort.

Methods:

Clinical and neuroimaging data were compiled from 72 participants from the Cardiovascular Health Study-Cognition Study, a community cohort of healthy aging and probable AD participants. Two time points were identified for each participant with a mean follow-up time of 5.36 years. MRI post-processing, morphometric measurements, and statistical analyses were performed using FreeSurfer, Version 7.1.1.

Results:

Cortical volume was significantly decreased in the bilateral superior frontal, bilateral inferior parietal, and left superior parietal regions, among others. Cortical thickness was significantly reduced in the bilateral superior frontal and left inferior parietal regions, among others. Overall gray and white matter volumes and hippocampal subfields also demonstrated significant reductions. Cortical volume atrophy trajectories between cognitively stable and cognitively declined participants were significantly different in the right postcentral region.

Conclusion:

Observed volume reductions were consistent with previous studies investigating morphometric brain changes. Patterns of brain atrophy between AD and aging may be different in magnitude but exhibit widespread spatial overlap. These findings help characterize patterns of brain atrophy that may reflect the general population. Larger studies may more definitively establish population norms of aging and AD-related neuroimaging changes.

Keywords

Aging Alzheimer’s disease cognition cognitive dysfunction neuroimaging

INTRODUCTION

Brain atrophy in aging is well known. Mechanisms of this phenomenon include microglial degeneration [1], neuronal DNA damage [2], and synaptic alterations [3]. Alzheimer’s disease (AD) is a neurodegenerative disorder marked by memory loss, confusion, and impaired executive cognitive function [4]. Advanced age is the most significant risk factor in the development of AD [5], and thus it is important to elucidate the relationship between age and AD with regards to each of their effects on brain morphometry.

Healthy aging and AD are characterized by diffuse atrophy of numerous cortical and subcortical regions, including the hippocampus, posterior cingulate cortex, dorsal parietal cortex, and dorsal frontal cortex [6, 7]. Trajectories of cortical and subcortical atrophy have been shown to differ between clinical or pre-clinical AD and healthy aging, especially in AD-specific regions, such as the hippocampus and medial temporal lobes [8 –11]. However, in other regions, trajectories of age-related atrophy and AD-related atrophy are less pronounced [7]. Currently, age is an independent factor from AD with regards to brain atrophy, but AD serves as an accelerating role in atrophy in certain regions [8, 11]. All these findings are well-corroborated among studies utilizing cohorts that have been selected based on cognitive variables and AD biomarkers [7].

However, few studies have explored longitudinal changes in brain atrophy with a well-characterized large community cohort of participants throughout a clinically relevant time frame. Such studies are vital to generalizing widespread findings of brain atrophy in relation to healthy aging and cognitive status. The community cohort utilized in this study was selected independently from cognitive variables, and thus investigation of this sample may provide a more direct understanding of the effects of aging and cognitive decline among the general population. This study attempts to explore longitudinal changes in brain morphometry of healthy aging and cognitively impaired participants from the Cardiovascular Health Study-Cognition Study (CHS-CS), a subset of the Cardiovascular Health Study [12]. We utilized longitudinal data from this cohort to investigate cortical and subcortical morphometric changes to investigate the independent and combinatory effects of aging and AD.

METHODS

Participant selection

Clinical and neuroimaging data were obtained from 72 participants from the CHS-CS, a community cohort study of participants with both healthy aging and probable AD spanning a period of 15–20 years [12]. As the data had already been collected and de-identified, this study was IRB exempt. Participants were selected based on availability of scans at baseline, which were dated between 1997–1999, and at a second time point between 3 and 7 years past the baseline timepoint. Each participant underwent psychiatric, neuropsychological, and neurological examinations, including an extensive neuropsychological battery [13 –15]. Among these, the Modified Mini-Mental Status Examination (3MSE) was tracked throughout the follow-up period as a screening tool for dementia [16]. The results from these examinations were assessed by a consensus of neurologists, psychiatrists, and neuropsychologists, who then classified participants’ cognition into normal, mild cognitive impairment (MCI), or dementia. Those who were classified into dementia or MCI were then reviewed and confirmed by an adjudication committee [13 –15].

Utilizing this methodology, 347 available CHS-CS scans were identified between 1997–1999, 101 of which underwent 1 or more follow-up scans. 10 were excluded due to a follow-up time of more than 7 years. Additionally, 14 were excluded due to missing either baseline or 2nd time point scan files, and 5 failed quality control despite appropriate corrective measures, with 72 for final analyses.

These 72 participants were included in longitudinal morphometric comparisons between the first and second time point. As an adjunctive analysis to assess the effect of worsening cognitive status as a covariate in the longitudinal studies, between-group volume and thickness comparisons were also performed utilizing a similar cluster-wise methodology. For this analysis, groups were divided into participants with normal cognition who stayed normal (n = 35) and those with normal cognition who subsequently developed MCI or dementia (n = 29), for a total of 64 participants.

MRI data acquisition

All MRI data were acquired at the University of Pittsburgh Medical Center MR Research Center using a 1.5 T GE Signa Scanner (GE Medical Systems, Milwaukee, MI, LX Version). A 3-dimensional volumetric spoiled gradient recalled acquisition (SPGR) was obtained (echo time/repetition time = 5/25, flip angle = 40°, number of excitations = 1, slice thickness = 1.5 mm/0 mm interslice gap), with an in-plane acquisition matrix of 256×256×124 image elements, 250×250 mm field of view, and an in-plane voxel size of 0.98 mm³.

2.3 MRI processing and cortical morphometry

MRI post-processing and analyses were performed at the Neuroimaging Lab (NIL) at the Mallinckrodt Institute of Radiology (MIR) at the Washington University School of Medicine in Saint Louis. Each scan underwent cortical reconstruction and segmentation through FreeSurfer (version 7.1, http://surfer.nmr.mgh.harvard.edu), in which each subject’s cortical surface and thickness at each vertex were measured through a semi-automated pipeline [17 –21]. Cortical surface reconstruction through FreeSurfer consists of 1) transformation of image coordinates into Talairach coordinates, 2) correction and normalization of intensity variations due to magnetic field inhomogeneities, 3) stripping of the skull, 4) segmentation at the gray-white matter boundary, 5) delineation of subcortical and cortical structures through the usage of cutting planes, 6) generation of connected components within white matter, and 7) tessellation and deformation of the resulting volume to generate an accurate and smooth surface. Volumes and thicknesses of cortical and subcortical structures were computed automatically through FreeSurfer. Methodologies of computing these variables have previously been well-described, but in summary, thickness was calculated as the difference between the pial and white matter boundaries, while volume was calculated utilizing a probabilistic atlas and Bayesian classification to assign voxels to neuroanatomical labels [17, 20]. Hippocampal subfield volumes were measured similarly using a probabilistic atlas built from two manually delineated datasets, one comprised of postmortem ex vivo images, and the second from T1-weighted in vivo images. These two datasets were combined by a Bayesian algorithm into the atlas carrying probabilistic information on each hippocampal subregion[22 –24].

The resulting segmentations were visually inspected on a slice-by-slice basis to assess for any incorrect delineations of pial from dura surfaces and any deformities in the parcellations and segmentations of cortical and subcortical structures. Those with substantial problems in cortical segmentation were excluded from the study, while those with less egregious segmentations were addressed in two ways: 1) if the segmentation error was due to an intensity normalization, control points were manually added to correctly classify appropriate brain matter, and 2) if the segmentation error was due to skull-stripping and pial surface errors, the surface was manually stripped of any skull or dura artifacts. Cortical volume and thickness measures were smoothed with a 10-mm full width at half maximum Gaussian kernel for statisticalanalysis.

Statistical analysis

Cortical, subcortical, and hippocampal subfield changes were analyzed through FreeSurfer’s longitudinal two-stage model, which is applicable for samples with the same number of time points with similar follow up times [25]. Volume and thickness changes were quantified through symmetrized percent change (SPC), measured by: $SPC = \frac{100 * rate}{average}$ (1)

where $rate = \frac{measure at 2 nd time point - measure at baseline}{time at 2 nd time point - time at baseline}$ (2)

and $average = \frac{sum of measures at both time points}{2}$ (3)

SPC is considered a robust measure of percent change, due to a multitude of factors: 1) measures at time point 1 is considered noisier than the average, 2) SPC is symmetric, allowing consistent quantification of changes, regardless of direction, and 3) SPC tends to confer more statistical power compared to standard percent change [25, 26].

Surface-based longitudinal comparisons were performed through FreeSurfer’s general linear modeling tool (mri_glmfit) with covariates including age at baseline, gender, body mass index (BMI), baseline cognition, cognitive decline, and, for volume analyses, estimated total intracranial volume (eTIV) as quantified by the cortical reconstruction pipeline. Cognitive decline was defined as participants’ cognitive status declining from normal to cognitively impaired, whether MCI or dementia, and also from MCI to dementia. Age, BMI, and eTIV were centered for analysis, due to potential multicollinearities [27]. For cortical volumes and thicknesses, statistical significance of within-subject differences were processed through cluster-wise correction, based on the Monte Carlo Z simulation, at an absolute cluster-forming threshold of 0.05 [28]. Subcortical, ventricular, total brain matter, and hippocampal subfield volumes were corrected via a voxel-wise threshold of p < 0.05 and a false discovery rate = 0.05 [29]. Between-group volume and thickness comparisons were also performed utilizing a similar cluster-wise methodology, while excluding baseline cognition and cognitive decline as covariates. Additionally, linear regression was performed to assess the relationship of morphometric changes within significant brain clusters with regards to aging and cognitive decline.

RESULTS

Participant demographics and neuropsychological testing

As summarized in Table 1, 72 participants, 28 males and 44 females, were included in the longitudinal study. 53 participants identified as Caucasian and 19 identified as African American. With regards to cognition at baseline, 64 were classified as cognitively normal; 8 as MCI; and none as dementia. Follow-up time to the second time point varied between 3.7 and 7.0 years, with a mean of 5.3 years. The mean age at baseline was 72.7 years, and the mean BMI at baseline was 26.8. Out of the 64 participants who were initially classified as normal cognition, 12 declined in cognition to MCI, and 17 declined to dementia by the second time point. Of the other 8 participants, who were initially classified as MCI, 6 cognitively declined to dementia by the second time point. By the second time point, 35 participants were classified as cognitively normal; 19 as MCI; and 18 as dementia. In total, 35 out of 72 participants declined in cognition between the two time points, and specifically, 29 out of 64 originally cognitively normal participants declined in cognition in the same time period. Two participants had a past history of stroke, though none of the participants experienced an interval stroke during the study’s time period.

Table 1

Demographic and clinical data of healthy aging and cognitively impaired participants selected for data analysis

Category	n
Total number of participants	72 (100%)
Age at baseline (y)	78.3±3.4 (70–88)
Follow-up time (y)	5.3±0.7 (3.7–7.0)
Biological sex
Male	28 (39%)
Female	44 (61%)
Race
Caucasian	53 (74%)
African American	19 (26%)
BMI	26.8±3.6 (19.6–37.1)
Cognitive progression
Normal at baseline	64 (89%)
Stable	35 (49%)
Normal → MCI	29 (40%)
MCI at baseline	8 (11%)
Stable	2 (3%)
MCI → dementia	6 (8%)
3MSE, stable (n = 37)
Baseline	96.4±4.4 (84–100)
Follow up	24.7±4.7 (84–100)
SPC (% /y)	–0.34^*±0.7 (–2.35–0.63)
3MSE, CD (n = 35)
Baseline	94.3±4.4 (84–100)
Follow up	89.3±8.5 (70–100)
SPC (% /y)	–1.49^**±2.8 (–14.6–0.8)

Percentage or standard deviation and range included for appropriate demographic variables. MCI, mild cognitive impairment; BMI, body mass index; 3MSE, Modified Mini Mental State Examination; CD, cognitively declined; SPC, symmetrized percent change. ^*p-value: 0.003; ^**p-value: 0.0003.

All 72 participants completed the 3MSE at baseline, and of those, 71 of them completed the 3MSE at follow-up. 3MSE scores significantly decreased in the group who were cognitively stable (–0.34% /year, p = 0.003) as well as in the group who cognitively declined (–1.49% /year, p = 0.0003), according to the aforementioned dementia/MCI classifications (Table 1).

Longitudinal differences in cortical volume

Cortical volume was significantly atrophied at the second time point compared to baseline in multiple regions, after controlling for age at baseline, gender, BMI, baseline cognition, cognitive decline, and eTIV (Table 2, Fig. 1). Significant regions included the left superior frontal (–2.322% /year, p = 0.0002), 2 left superior parietal (–3.451% /year, p = 0.0002; –3.04% /year, p = 0.0002), left inferior parietal (–3.129% /year, p = 0.0002), left lateral occipital (–2.696% /year, p = 0.0002), left medial orbitofrontal (–1.786% /year, p = 0.0004), left insular cortex (–1.144% /year, p = 0.01112), right superior frontal (–3.03% /year, p = 0.0002), right lateral occipital (–3.153% /year, p = 0.0002), right caudal middle frontal (–2.506% /year, p = 0.0002), and right inferior parietal (–3.656% /year, p = 0.0003) regions. Linear regressions of significant clusters correlating average age demonstrated significant positive correlations in the left superior frontal (p = 0.014, R² = 0.08) and right superior frontal (p = 0.027, R² = 0.07) regions (Supplementary Figure 1). Additional clusters with significant positive correlations with age were one of the left superior parietal regions (p < 0.001, R² = 0.19) and the right caudal middle frontal region (p = 0.004, R² = 0.11) (Supplementary Figure 1). No other clusters exhibited significant associations in rate of atrophy with age.

Fig. 1

Longitudinal Changes in Cortical Volume. Regional differences in cortical volume, controlling for age, gender, BMI, eTIV, baseline cognition, and cognitive decline. Significant clusters include on the left hemisphere: superior frontal (A), superior parietal (B, C), inferior parietal (D), lateral occipital (E), medial orbitofrontal (F), and insular cortex (G), and on the right: superior frontal (H), lateral occipital (I), caudal middle frontal (J), and inferior parietal (K) regions. Quantitative data listed in Table 2.

Table 2

Significant clusters of cortical volume changes

Cluster	Hemisphere	Region	MNI - x	MNI - y	MNI - z	Size (mm²)	p	SPC (% /y)	Cohen’s D
A	Left	Superior frontal	–8.7	32.8	52.7	4659.21	0.0002	–2.322	–0.6953
B	Left	Superior parietal	–12.5	–53.3	65.6	2934.77	0.0002	–3.451	–0.7082
C	Left	Superior parietal	–22.5	–61.6	545	1110.24	0.0002	–3.04	–0.6639
D	Left	Inferior parietal	–37.9	–78	27	1237.47	0.0002	–3.129	–0.7353
E	Left	Lateral occipital	–17.7	–98.8	–13.1	918.44	0.0002	–2.696	–0.6232
F	Left	Medial orbitofrontal	–6.6	15.2	–16	760.65	0.0004	–1.786	–0.7569
G	Left	Insular cortex	–37.1	0.3	–14.4	538.33	0.0112	–1.144	–0.6527
H	Right	Superior frontal	10.4	45.7	42.5	7654.91	0.0002	–3.03	–0.7055
I	Right	Lateral occipital	21.9	–91.8	19	1432.73	0.0002	–3.153	–0.6229
J	Right	Caudal middle frontal	35.5	25.3	46.7	1094.84	0.0002	–2.506	–0.6647
K	Right	Inferior parietal	36.6	–76.1	37.4	788.62	0.003	–3.656	–0.7728

Cluster letters correspond to those in Fig. 1. MNI, Montreal Neurological Institute 305 coordinates; SPC, symmetrized percent change.

Longitudinal differences in cortical thickness

Cortical thickness was significantly reduced at the second time point compared to baseline in multiple regions, after controlling for age at baseline, gender, BMI, baseline cognition, and cognitive decline (Table 3, Fig. 2). Significantly thinned regions included the left superior frontal (–1.435% /year, p = 0.0064), left inferior parietal (–1.581% /year, p = 0.0247), right postcentral (–2.245% /year, p = 0.0002), and right superior frontal (–1.615% /year, p = 0.0002) regions. The cortical thickness of the left fusiform gyrus (1.228% /year, p = 0.0376) was significantly increased at the second time point.

Fig. 2

Longitudinal Changes in Cortical Thickness. Regional differences in cortical thickness, controlling for age, gender, BMI, baseline cognition, and cognitive decline. Significant clusters include on the left hemisphere: superior frontal (A), inferior parietal (B), and fusiform region (C), and on the right: postcentral (D) and superior frontal (E) regions. Significant reductions in cortical volume were observed in all these clusters, except in the fusiform region, where cortical volume was increased. Quantitative data listed in Table 3.

Table 3

Significant clusters of cortical thickness changes

Cluster	Hemisphere	Region	MNI - x	MNI - y	MNI - z	Size (mm²)	p	SPC (% /y)	Cohen’s D
A	Left	Superior frontal	–22.8	14.7	51	749.74	0.0064	–1.435	–0.5621
B	Left	Inferior parietal	–36.9	–77.5	29.9	654.08	0.0247	–1.581	–0.6754
C	Left	Fusiform	–27.2	–44.9	–18.2	613.34	0.0376	1.228	0.6406
D	Right	Postcentral	43.3	–17.1	33.3	2553.64	0.0002	–2.245	–0.6405
E	Right	Superior frontal	11.5	48	39.4	2158.35	0.0002	–1.615	–0.6016

Cluster letters correspond to those in Fig. 2. MNI, Montreal Neurological Institute 305 coordinates; SPC, symmetrized percent change.

Longitudinal differences in subcortical structures, ventricles, and hippocampal subfields

Numerous subcortical regions demonstrated a significant reduction in volume from baseline to the second time point, while ventricular regions significantly increased in volume. In addition, total gray matter volume (–1.090% /year, p = 4.771E-07), cerebral white matter volume (–0.492% /year, p = 0.0215), and cortex volume (–1.201% /year, p = 9.397E-06) exhibited significant reductions. A full list of subcortical structures, ventricles, and overall brain measures that significantly changed are listed inTable 4.

Table 4

Significant subcortical, ventricular, and overall brain regions of segmentation

Segmentation	p	SPC (% /y)
Left lateral ventricle	7.345E-05	4.1575
Left inferior lateral ventricle	2.037E-05	6.9814
Left cerebellum, white matter	9.616E-09	–1.9952
Left cerebellum, cortex	0.0033	–0.5568
Left thalamus	5.636E-05	–1.2618
3rd ventricle	0.0002	2.3577
4th ventricle	7.063E-05	1.5879
Brain stem	3.444E-05	–0.4978
Left hippocampus	0.0004	–1.7662
Right lateral ventricle	0.0157	3.2701
Right inferior lateral ventricle	3.926E-05	7.8912
Right cerebellum, white matter	2.576E-05	–2.3043
Right cerebellum, cortex	2.564E-05	–0.8591
Right thalamus	0.0005	–1.5022
Right hippocampus	0.0013	–1.8639
Right ventral dorsal column	0.0065	–0.7824
White matter hypointensities	0.0019	4.4610
Non-white matter hypointensities	0.0092	9.7692
Corpus callosum, mid-posterior	0.0002	–1.8158
Corpus callosum, central	0.0188	–2.6712
Corpus callosum, mid-anterior	0.0074	–2.2381
Brain segmentation volume	1.742E-06	–0.6755
Brain segmentation volume (no ventricles)	2.218E-08	–0.8806
Left hemisphere cortex volume	0.0002	–1.1747
Right hemisphere cortex volume	7.516E-05	–1.2226
Total cortex volume	9.397E-06	–1.2006
Left cerebral white matter volume	0.0215	–0.5950
Cerebral white matter volume	0.0251	–0.4922
Subcortical gray volume	0.0002	–0.7724
Total gray volume	4.71E-07	–1.0900
Supratentorial volume	3.162E-05	–0.6174
Supratentorial volume (no ventricles)	4.406E-07	–0.8494
Brain segmentation volume-to-eTIV ratio	1.742E-06	–0.6755

Significance level as calculated at a false discovery rate threshold of 0.05 was 0.0251. eTIV, estimated total intracranial volume; SPC, symmetrized percent change. Pink rows indicate increases with p < 0.05. Red rows indicate increases with p < 0.001. Light blue rows indicate decreases with p < 0.05. Dark blue rows indicate decreases with p < 0.001.

Hippocampal subfield regions that demonstrated a significant reduction in volume included the left granule cell and molecular layer of the dentate gyrus (GC-ML-DG) head (–1.023% /year, p = 0.0006), left cornu ammonis (CA) 3 body (–1.789% /year, p = 0.0022) and head (–1.067% /year, p = 0.0001), left CA4 head (–0.9506% /year, p = 0.0024), and right GC-ML-DG body (–1.362% /year, p = 0.0048). The right presubiculum body demonstrated a significant increase in volume (1.172% /year, p = 0.0029).

Between-group differences in cognitive decline

Participants who cognitively declined had more cortical thinning in the left lateral occipital region than participants who stayed cognitively stable (–0.324% /year versus 0.917% /year, n = 29, n = 35, p = 0.013), after controlling for age at baseline, gender, BMI, and eTIV. Likewise, participants who cognitively declined had more cortical volume atrophy in the right postcentral region than participants who stayed cognitively stable (–1.601% /year versus –0.893% /year, n = 29, n = 35, p = 0.0239), after controlling for age at baseline, gender, and BMI. There were no statistically significant differences in hippocampal subfield or subcortical volume atrophy rates between the two groups. Detailed information for each cluster is listed in Table 5.

Table 5

Significant clusters of between-group analyses

Measure	Hemisphere	Region	MNI - x	MNI - y	MNI - z	Size (mm²)	p	SPC, stable (% /y)	SPC, CD (% /y)	Difference	Cohen’s D
Volume	Right	Postcentral	62.6	–15.7	20	569.32	0.0239	–0.8927	–1.6011	–0.7084	0.6586
Thickness	Left	Lateral occipital	–24.5	–80.9	–7.2	711.71	0.013	0.9172	–0.3241	–1.2413	0.9143

MNI, Montreal Neurological Institute 305 coordinates; CD, cognitively declined; SPC, symmetrized percent change.

DISCUSSION

Longitudinal changes in cortical volume

We observed cortical volume reductions consistent with previous studies of brain atrophy after controlling for age and cognitive decline. Our findings overlap with previous observations of brain atrophy with aging, particularly with regards to widespread atrophy of frontal and prefrontal cortices [30, 31]. The magnitude of atrophy is higher than previous studies investigating healthy aging, where annual decline in similar regions was 0.5% /year [11]. Similarly, atrophy of the inferior parietal lobule is consistent with previous studies of cortical atrophy with aging and MCI/AD [8]. Atrophy of the superior parietal region have also been reported in preclinical AD participants, in which one study reported comparable atrophy rates, with an approximately 2.5% atrophy rate in a 2-year time frame [32]. These areas of volume reduction correlate well with clinical courses of individuals with AD, as the frontal and parietal lobes play a vital role in executive function and emotional regulation, which are core cognitive features impaired in AD. Mechanistically, these areas are spatially associated with accumulation of neurofibrillary tangles and increasing amyloid burden, resulting in neuronal degeneration [33]. Moreover, the positive correlation between age and rate of atrophy in bilateral superior frontal regions, which were the most significant and largest clusters of the analyses, is consistent with previous studies, which suggest that earlier onset of AD is associated with greater cognitive deficits and greater regional vulnerability to atrophy, especially in the precuneal, parietal, and frontal lobe regions [34]. Many of our participants developed late-onset MCI/AD, at which phenotypic expression of cognitive decline is relatively less pronounced [35]. Hence, the age at which cognitive decline occurs may play a factor into the extent of brain atrophy.

In general, the extent of atrophy in this study lay closer to MCI/AD trajectories than healthy aging. As the regions that had significantly different atrophy trajectories between cognitively stable and cognitively declining participants lacked any major overlap with the regions that atrophied in the overall cohort, these widespread volume reductions in the overall cohort may signify an overlap between AD-prone regions and the effects of healthy aging [36]. Notably, cognitive decline seems to play a significant role in cortical atrophy in the right postcentral region, which is an area that previous studies have demonstrated major age-related brain changes, independently of neurodegenerative pathologies [37]. These findings imply that among the general population, morphometric cortical brain changes in healthy aging and MCI/AD persons may primarily differ in the magnitude of atrophy, rather than the localization of atrophy. Given these results, delineating the divergent roles of aging and cognitive decline should be further explored in future studies.

Longitudinal changes in cortical thickness

The observed regions of reduction in cortical thickness, specifically the frontal and parietal regions, are largely consistent with healthy aging [38]. Moreover, AD studies demonstrate cortical thinning in similar regions [39]. The extent of cortical thinning found in this study, as in cortical volume atrophy, are closer to MCI/AD trajectories [32]. As cortical thinning is one of the primary contributors to cortical volume reductions [40], this analysis provides confirmatory evidence for the patterns of atrophy associated with healthy aging and MCI/AD. The fusiform gyrus increased in thickness over time, though significant morphometric trends in this region have not been widely reported previously [41]. However, the lack of a corresponding volume increase in this region makes it a possibility that this result may not be a significant contributory factor in the manifestation of brain atrophy patterns.

Longitudinal changes in hippocampal subfields

With regards to hippocampal subfields, we observed significant volume reductions in dentate gyri, CA3, and CA4 regions. These findings are consistent with previous studies, which have indicated that the dentate gyrus is the most vulnerable subfield to the effects of aging while CA regions are AD-specific targets [22, 42]. Furthermore, the magnitude of these changes is also similar under the effects of aging and AD and may be predictive of preclinical AD [42]. Such processes are suggested to be driven by accumulation of tau, leading to neuronal degeneration. Clinically, hippocampal alterations manifest into amnestic syndrome, a capital feature of AD [43]. These results may implicate generalizable trends in hippocampal subfields for both healthy aging as well as pathophysiological processes of AD.

Strengths and limitations

Among the strengths of this project is that the sample we utilized was a well-characterized community cohort with thorough demographic information, and we were able to perform a comprehensive quantitative analysis with this selection. Cortical thickness and hippocampal subfields are measures that are not as widely studied in the context of aging and AD, and thus our methodology incorporates a multidimensional approach to investigating diffuse cortical and subcortical atrophy. Another strength of this study lies in the longitudinal nature of the data collection and analysis, in that each participant had a comparable follow-up time, which allowed us to utilize the longitudinal two-stage module of FreeSurfer as a robust statistical approach.

However, the study is not without limitations. We recognize one limitation is the lack of AD biomarker screenings, partly because this community cohort was originally selected as a subset of an investigation of coronary heart disease and stroke [12, 44]. Commonly-measured AD biomarkers, such as amyloid-β and tau, were not tracked during the CHS-CS as these metrics were not available for characterization at those time points. Thus, while the study evaluations strongly suggest Alzheimer’s dementia, they do not identify Alzheimer’s disease. This nuance should be considered when interpreting our study results. Study restrictions also prevented a more nuanced investigation into the development of pre-clinical and clinical AD with regards to patterns of brain atrophy. The lack of quantitative neurological data raises the possibility that cognitively normal participants could have been in the preclinical AD stage, and thus possibly mask differences between aging and cognitive impairment. Furthermore, perfusion or diffusion parameters were not included in this study. A previous study investigating cortical perfusion reported areas of loss of cerebral blood flow did not correspond to either the cortical thickness or volume changes observed in this study [45]. However, the study also suggested that cortical thinning may only play a minimal role in AD progression and cerebral blood flow was a surface measurement rather than a full-thickness measurement. In addition, differences in study design may make it difficult to ascertain discrepancies in results as the nature of our sample characterization did not provide for precise staging of cognitive impairment ordementia.

Conclusion

The morphometric cortical and subcortical changes over time that were observed may reflect early changes in dementia combined with effects of healthy aging. Though regional changes in aging and AD have been thoroughly documented, this study marks the first time such changes were quantified in a well-characterized community cohort. Patterns of brain atrophy between AD and aging may be different in magnitude but exhibit widespread spatial overlap. Larger studies replicating these findings in conjunction with investigation of neurologic biomarkers will be important to establish population norms of brain change.

Footnotes

ACKNOWLEDGMENTS

This work is dedicated to the fond and sincerest admiration for the memory of Dr. Lewis Kuller, without whom none of this research would be possible. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

FUNDING

This research was supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, 75N92021D00006, and grants U01HL080295 and U01HL130114 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org.

CONFLICT OF INTEREST

Dr. Raji is an Editorial Board Member of this journal but was not involved in the peer-review process nor had access to any information regarding its peer-review. Dr. Raji does consult to Brainreader ApS, Apollo Health, Neurevolution LLC, and the Pacific Neuroscience Foundation.

DATA AVAILABILITY

The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

The supplementary material is available in the electronic version of this article: .

References

Conde

, Streit

(2006) Microglia in the aging brain. J Neuropathol Exp Neurol 65, 199–203.

Rutten

, Schmitz

, Gerlach

, Oyen

, de Mesquita

, Steinbusch

, Korr

(2007) The aging brain: Accumulation of DNA damage or neuron loss? Neurobiol Aging 28, 91–98 .

Morrison

, Hof

(2007) Life and death of neurons in the aging cerebral cortex. Int Rev Neurobiol 81, 41–57.

Khachaturian

(1985) Diagnosis of Alzheimer’s disease. Arch Neurol 42, 1097–1105.

Guerreiro

, Bras

(2015) The age factor in Alzheimer’s disease. Genome Med 7, 106.

Mueller

, Schuff

, Yaffe

, Madison

, Miller

, Weiner

(2010) Hippocampal atrophy patterns in mild cognitive impairment and Alzheimer’s disease. Hum Brain Mapp 31, 1339–1347.

Pini

, Pievani

, Bocchetta

, Altomare

, Bosco

, Cavedo

, Galluzzi

, Marizzoni

, Frisoni

(2016) Brain atrophy in Alzheimer’s disease and aging. Ageing Res Rev 30, 25–48.

Bakkour

, Morris

, Wolk

, Dickerson

(2013) The effects of aging and Alzheimer’s disease on cerebral cortical anatomy: Specificity and differential relationships with cognition. Neuroimage 76, 332–344.

Barnes

, Bartlett

, van de Pol

, Loy

, Scahill

, Frost

, Thompson

, Fox

(2009) A meta-analysis of hippocampal atrophy rates in Alzheimer’s disease. Neurobiol Aging 30, 1711–1723.

10.

Dickerson

, Bakkour

, Salat

, Feczko

, Pacheco

, Greve

, Grodstein

, Wright

, Blacker

, Rosas

, Sperling

, Atri

, Growdon

, Hyman

, Morris

, Fischl

, Buckner

(2009) The cortical signature of Alzheimer’s disease: Regionally specific cortical thinning relates to symptom severity in very mild to mild AD dementia and is detectable in asymptomatic amyloid-positive individuals. Cereb Cortex 19, 497–510.

11.

Fjell

, Walhovd

, Fennema-Notestine

, McEvoy

, Hagler

, Holland

, Brewer

, Dale

(2009) One-year brain atrophy evident in healthy aging. J Neurosci 29, 15223–15231.

12.

Lopez

, Kuller

, Fitzpatrick

, Ives

, Becker

, Beauchamp

(2003) Evaluation of dementia in the cardiovascular health cognition study. Neuroepidemiology 22, 1–12.

13.

Lopez

, Jagust

, DeKosky

, Becker

, Fitzpatrick

, Dulberg

, Breitner

, Lyketsos

, Jones

, Kawas

, Carlson

, Kuller

(2003) Prevalence and classification of mild cognitive impairment in the Cardiovascular Health Study Cognition Study: Part 1. Arch Neurol 60, 1385–1389.

14.

Lopez

, Kuller

, Becker

, Dulberg

, Sweet

, Gach

, Dekosky

(2007) Incidence of dementia in mild cognitive impairment in the cardiovascular health study cognition study. Arch Neurol 64, 416–420.

15.

Lopez

, Becker

, Jagust

, Fitzpatrick

, Carlson

, DeKosky

, Breitner

, Lyketsos

, Jones

, Kawas

, Kuller

(2006) Neuropsychological characteristics of mild cognitive impairment subgroups. J Neurol Neurosurg Psychiatry 77, 159–165.

16.

Teng

, Chui

(1987) The Modified Mini-Mental State (3MS) examination. J Clin Psychiatry 48, 314–318.

17.

Dale

, Fischl

, Sereno

(1999) Cortical surface-based analysis: I. Segmentation and surface reconstruction. Neuroimage 9, 179–194.

18.

Fischl

, Dale

(2000) Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad Sci U S A 97, 11050–11055.

19.

Fischl

, Salat

, Busa

, Albert

, Dieterich

, Haselgrove

, van der Kouwe

, Killiany

, Kennedy

, Klaveness

, Montillo

, Makris

, Rosen

, Dale

(2002) Whole brain segmentation: Automated labeling of neuroanatomical structures in the human brain. Neuron 33, 341–355.

20.

Fischl

, Sereno

, Dale

(1999) Cortical surface-based analysis. II: Inflation, flattening, and a surface-based coordinate system. Neuroimage 9, 195–207.

21.

Fischl

, Sereno

, Tootell

, Dale

(1999) High-resolution intersubject averaging and a coordinate system for the cortical surface. Hum Brain Mapp 8, 272–284.

22.

Iglesias

, Augustinack

, Nguyen

, Player

, Wright

, Roy

, Frosch

, McKee

, Wald

, Fischl

, Van Leemput

(2015) A computational atlas of the hippocampal formation using, ultra-high resolution MRI: Application to adaptive segmentation of in vivo MRI. Neuroimage 115, 117–137.

23.

Iglesias

, Van Leemput

, Augustinack

, Insausti

, Fischl

, Reuter

(2016) Bayesian longitudinal segmentation of hippocampal substructures in brain MRI using subject-specific atlases. Neuroimage 141, 542–555.

24.

Sämann

, Iglesias

, Gutman

, Grotegerd

, Leenings

, Flint

, Dannlowski

, Clarke-Rubright

, Morey

, van Erp

TGM

, Whelan

, Han

LKM

, van Velzen

, Cao

, Augustinack

, Thompson

, Jahanshad

, Schmaal

(2022) FreeSurfer-based segmentation of hippocampal subfields: A review of methods and applications, with a novel quality control procedure for ENIGMA studies and other collaborative efforts. Hum Brain Mapp 43, 207–233.

25.

Reuter

, Schmansky

, Rosas

, Fischl

(2012) Within-subject template estimation for unbiased longitudinal image analysis. Neuroimage 61, 1402–1418.

26.

Berry

, Ayers

(2006) Symmetrized percent change for treatment comparisons. Am Stat 60, 27–31.

27.

Iacobucci

, Schneider

, Popovich

, Bakamitsos

(2017) Mean centering, multicollinearity, and moderators in multiple regression: The reconciliation redux. Behav Res Methods 49, 403–404.

28.

Hagler

Jr. , Saygin

, Sereno

(2006) Smoothing and cluster thresholding for cortical surface-based group analysis of fMRI data. Neuroimage 33, 1093–1103.

29.

Benjamini

, Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc B Methodol 57, 289–300.

30.

DeCarli

, Massaro

, Harvey

, Hald

, Tullberg

, Au

, Beiser

, D’Agostino

, Wolf

(2005) Measures of brain morphology and infarction in the framingham heart study: Establishing what is normal. Neurobiol Aging 26, 491–510.

31.

Salat

, Buckner

, Snyder

, Greve

, Desikan

, Busa

, Morris

, Dale

, Fischl

(2004) Thinning of the cerebral cortex in aging. Cereb Cortex 14, 721–730.

32.

Pegueroles

, Vilaplana

, Montal

, Sampedro

, Alcolea

, Carmona-Iragui

, Clarimon

, Blesa

, Lleó

, Fortea

(2017) Longitudinal brain structural changes in preclinical Alzheimer’s disease. Alzheimers Dement 13, 499–509.

33.

Whitwell

(2010) Progression of atrophy in Alzheimer’s disease and related disorders. Neurotox Res 18, 339–346.

34.

Fiford

, Ridgway

, Cash

, Modat

, Nicholas

, Manning

, Malone

, Biessels

, Ourselin

, Carmichael

, Cardoso

, Barnes

(2018) Patterns of progressive atrophy vary with age in Alzheimer’s disease patients. Neurobiol Aging 63, 22–32.

35.

Holland

, Desikan

, Dale

, McEvoy

(2012) Rates of decline in Alzheimer disease decrease with age. PLoS One 7, e42325.

36.

Fjell

, Westlye

, Grydeland

, Amlien

, Espeseth

, Reinvang

, Raz

, Dale

, Walhovd

(2014) Accelerating cortical thinning: Unique to dementia or universal in aging? Cereb Cortex 24, 919–934.

37.

Peelle

, Cusack

, Henson

(2012) Adjusting for global effects in voxel-based morphometry: Gray matter decline in normal aging. Neuroimage 60, 1503–1516.

38.

Thambisetty

, Wan

, Carass

, An

, Prince

, Resnick

(2010) Longitudinal changes in cortical thickness associated with normal aging. Neuroimage 52, 1215–1223.

39.

Cho

, Jeon

, Kang

, Lee

, Kim

, Shin

, Kim

, Noh

, Im

, Kim

, Chin

, Seo

, Na

(2013) Longitudinal changes of cortical thickness in early- versus late-onset Alzheimer’s disease. Neurobiol Aging 34, 1921.e1929–1921.e1915.

40.

Storsve

, Fjell

, Tamnes

, Westlye

, Overbye

, Aasland

, Walhovd

(2014) Differential longitudinal changes in cortical thickness, surface area and volume across the adult life span: Regions of accelerating and decelerating change. J Neurosci 34, 8488–8498.

41.

Salat

, Greve

, Pacheco

, Quinn

, Helmer

, Buckner

, Fischl

(2009) Regional white matter volume differences in nondemented aging and Alzheimer’s disease. Neuroimage 44, 1247–1258.

42.

Nadal

, Coupé

, Helmer

, Manjon

, Amieva

, Tison

, Dartigues

, Catheline

, Planche

(2020) Differential annualizedrates of hippocampal subfields atrophy in aging and futureAlzheimer’s clinical syndrome. Neurobiol Aging 90, 75–83.

43.

Halliday

(2017) Pathology and hippocampal atrophy in Alzheimer’s disease. Lancet Neurol 16, 862–864.

44.

Fried

, Borhani

, Enright

, Furberg

, Gardin

, Kronmal

, Kuller

, Manolio

, Mittelmark

, Newman

, O’Leary

, Psaty

, Rautaharju

, Tracy

, Weiler

(1991) The Cardiovascular Health Study: Design and rationale. Ann Epidemiol 1, 263–276.

45.

Ashford

, Shih

, Coupal

, Shetty

, Schneider

, Cool

, Aleem

, Kiefer

, Mendiondo

, Schmitt

(2000) Single SPECT measures of cerebral cortical perfusion reflect time-index estimation of dementia severity in Alzheimer’s disease. J Nucl Med 41, 57–64.