Age Effects on Cortical Thickness in Cognitively Normal Elderly Individuals

Abstract

Background/Aims

Atrophy in both grey and white matter is found in normal aging. The prefrontal cortex and the frontal lobe white matter are thought to be the most affected regions. Our aim was to examine the effects of normal aging on cortical grey matter using a 3D quantitative cortical mapping method.

Methods

We analyzed 1.5-tesla brain magnetic resonance imaging data from 44 cognitively normal elderly subjects using cortical pattern matching and cortical thickness analyses. Linear regression analysis was used to study the effect of age on cortical thickness. 3D map-wide correction for multiple comparisons was conducted with permutation analyses using a threshold of p < 0.01.

Results

We found a significant negative association between age and cortical thickness in the right hemisphere (p_corrected = 0.009) and a trend level association in the left hemisphere (p_corrected = 0.081). Age-related changes were greatest in the sensorimotor, bilateral dorsal anterior cingulate and supplementary motor cortices, and the right posterior middle and inferior frontal gyri. Age effects greater in the medial than lateral visual association cortices were also seen bilaterally.

Conclusion

Our novel method further validates that normal aging results in diffuse cortical thinning that is most pronounced in the frontal and visual association cortices.

Introduction

According to the Department of Health and Human Services, the aging population in the United States will grow to an estimated 72.1 million by 2030, more than twice the number of older adults in 2000 [1]. Therefore, it is imperative to prepare to foster the wellbeing of older adults. Currently, many studies investigating cognitive and functional changes in normal aging are being conducted. Researchers agree that the degree with which cognition is affected varies from one individual to another [2], and this variability may in part reflect incipient disease processes in a subset of individuals. Aging is commonly associated with loss of structural integrity of the cortical grey matter and expansion of the ventricles due to loss of tissue. The most commonly reported cortical region that is affected in aging is the prefrontal cortex, an area responsible for attention and executive functions [3,4,5,6,7].

A number of studies have investigated the effects of aging on cerebral grey matter using advanced neuroimaging techniques. By far the most widely replicated and arguably most susceptible cortical region seems to be the frontal cortex [5,6,8,9,10,11], although associations with other brain regions have been reported including the parietal [5,6,7,8,9] and temporal [6,8,9] cortices, the primary motor and sensory cortices [9,11], the secondary visual cortex [6,8], and the anterior cingulate [9].

A better understanding of these age-associated changes of the human cortex will advance our knowledge of the normal aging process and help to distinguish it from the neurodegenerative effects of many neurological disorders of the elderly, such as Alzheimer's disease (AD). The aim of our study was to assess the effects of age on cortical thickness using another well-established quantitative cortical mapping method in a well-defined cognitively normal elderly cohort.

Materials and Methods

Subjects

We analyzed SPGR (spoiled gradient echo) T1-weighted magnetic resonance imaging (MRI) data for 44 cognitively normal elderly controls from the University of California, Los Angeles (UCLA) Alzheimer's Disease Research Center database, an ongoing longitudinal study of AD, other dementias, and normal cognitive aging. All participating subjects provided informed consent following the Declaration of Helsinki and the restrictions and policies of the UCLA Institutional Review Board.

Demographic and cognitive data were collected through detailed neurological examinations, interviews as well as a comprehensive neuropsychological evaluation including the following tests: Mini-Mental State Examination (MMSE) [12], the Wechsler Adult Intelligence Scale – 3rd edition (WAIS-III: 8 subtests) [13], the Boston Naming Test [14], the Controlled Oral Word Association Test (FAS and Animals) [15], the Wechsler Memory Scale – 3rd edition (WMS-III: Logical Memory I and II and Visual Reproduction I and II) [16], California Verbal Learning Test – 1st or 2nd edition (CVLT-I or CVLT-II) [17,18], the Rey-Osterrieth Complex Figure test – copy and 3-min recall (ROCFT) [19], the Stroop Interference Test [20], Trailmaking A and B [21], and the Wisconsin Card Sorting Test-64 (WCST-64) [22]. Diagnosis was reached by consensus among neurologists, psychiatrists, and neuropsychologists. Normal aging was defined as scoring within normal limits based on age- and education-adjusted norms on a predefined subset of the tests from the neuropsychological battery. Exclusion criteria included age <50, history of significant drug or alcohol abuse, and concurrent psychiatric or neurological illness.

Image Acquisition and Analysis

Structural SPGR T1-weighted images were acquired using a 1.5-tesla Signa scanner (GE Medical Systems, Milwaukee, Wisc., USA) with the following protocol: gapless coronal acquisition with 28 ms repetition time, 6 ms echo time, 220 mm field of view, 256 × 192 voxels in acquisition matrix, and 1.5 mm slice thickness. The MRI scans were intensity normalized using a regularized tricubic B-spline approach and spatially scaled using the Minctracc algorithm and a 9-parameter linear transformation (3 translations, 3 rotations, and 3 scales) to fit the International Consortium for Brain Mapping 53 (ICBM53) average brain imaging template [23,24].

An intracranial mask of the brain was generated using an automated brain surface extraction algorithm tool in BrainSuite [25]. The outputs were visually inspected and all segmentation errors (i.e., mislabeled brain or nonbrain tissue) were manually corrected. Hemispheric masks were produced. Bias field correction was applied to correct intensity errors due to magnetic field inhomogeneities. Each voxel was classified as cerebrospinal fluid, grey matter, and white matter based on signal intensity values [25]. The grey matter data were interpolated in 0.33 mm isotropic ‘subvoxel’ resolution. Resampling of the maps to subvoxel spatial resolution reduces subsequent inaccuracies in computing thickness fields by calculating distances to sets of voxels. After automated 3D hemispheric reconstruction, 39 sulci per hemisphere were traced following a well-established protocol with high interrater reliability [26]. All individual sulcal maps were averaged to create a representative study-specific sulcal map. Next, cortical surfaces were parameterized, flattened, and warped to align all individual sulci with the average sulcal models assuring that all analogous gyri were as explicitly matched as possible [27] (fig. 1). Cortical grey matter thickness (the distance in millimeters between the grey/white and the grey/cerebrospinal fluid boundaries) was computed at each surface point and smoothed using a surface-based kernel with 10 mm radius [28].

Fig. 1.

Schema of cortical pattern matching method.

Linear regression analysis was conducted with age as the predictor variable and cortical thickness as the outcome variable. Both statistical significance and correlation maps were created to calculate the effect of age on cortical thickness at every surface point. We investigated this association by correcting for multiple comparisons using permutation testing with a threshold of p < 0.01. In our permutation analyses, we created 100,000 random simulations in which we computed the fraction of experiments that would be observed by chance. These methods are standardized and have been described in greater detail elsewhere [29].

Results

Our sample consisted of 26 males and 18 females with a mean age of 66.09 ± 7.23 years (range: 51-79), a mean education of 17.25 ± 2.40 years (range: 13-24) and an average MMSE score of 29.34 ± 0.78 (range: 27-30). Follow-up data were available for 37 subjects (mean duration of follow-up 6 ± 4 years, range: 1-11). Of these 37, 2 subjects had converted to dementia in years 2 and 10 of follow-up. Our findings were reported for all subjects and also after excluding the 2 subjects who later converted to dementia. The mean cortical thickness for the final cohort was 2.5 mm for the right hemisphere and 2.6 mm for the left hemisphere, which corresponds well with the previously reported range of 1.5-3.4 mm [30].

In figure 2, the statistical and correlation maps show the relationship between age and cortical thickness in our elderly adult cohort. The global permutation-corrected significance of the association between age and cortical thinning was p_corrected = 0.009 for the right and p_corrected = 0.081 for the left hemisphere. The cortical areas showing significant associations included the sensorimotor [Brodmann area (BA) 1-4], supplementary motor cortices (medial BA 4 and 6), dorsal anterior cingulate (BA 24), posterior inferior temporal (BA 37) and visual association cortices (BA 18-19) bilaterally, as well as the premotor (lateral BA 4 and 6) and the posterior parts of the prefrontal cortex (BA 8, 9 and 44) on the right, and the precuneus/posterior cingulate (BA 23 and 31) and the primary visual cortex (BA 17) on the left. Of note, the entorhinal/parahippocampal area – the cortical region affected earliest by Alzheimer's pathology – did not show significant age-associated atrophy. This area showed a nonsignificant effect in the opposite direction. After excluding the two normal control subjects who later converted to dementia, our maps remained largely unchanged; therefore, only the maps from the entire cohort are shown. The map-wise permutation-corrected significance of the maps after excluding the 2 subjects who progressed to dementia during the follow-up period were p_corrected = 0.013 for the right and p_corrected = 0.089 for the left hemispheres.

Fig. 2.

Statistical (left) and correlation maps (right) demonstrating the association between age and cortical grey matter thickness in cognitively normal subjects.

Discussion

Our findings suggest that the different divisions of the cortex have differential susceptibility to aging, both anatomically and phylogenetically. In general, we found that the neocortical areas were profoundly affected, while the allocortical areas were largely preserved. This pattern is distinct from the one observed in AD in two very important ways. Our work and the work of others clearly show that aging has a significant impact on the primary sensory, motor, and visual cortices, which are relatively spared in AD, while at the same time normal aging spares the entorhinal/parahippocampal regions, which are affected earliest in AD. Our findings extend the findings of other research studies [2,3,4,5,6,7,8,9,11] by capturing a distinct pattern of cortical thinning with advancing age using a novel cortical mapping method. By doing so, our study improves our understanding of the structural effects of aging on the brain. Such work is important, as it will allow us to more insightfully interpret future imaging studies, especially when considering cognitively normal subjects with positive amyloid positron emission scans. This is an important problem given that approximately one third of all cognitively normal elderly have positive amyloid scans [31], and it remains unclear who of these subjects will progress to dementia and who will not. In these cases, determining whether these subjects have atrophy in the primary sensory, motor, and visual cortices versus the entorhinal/parahippocampal regions might provide an important disease biosignature to determine a better prognosis.

Several strengths and limitations of our study should be recognized. Although the majority of subjects had longitudinal clinical data and therefore were not at imminent risk for clinical dementia, the imaging analyses were cross-sectional. A longitudinal approach would be even more informative with potential to show the trajectory of gradual age-associated structural changes in the brains of elderly individuals over time. The relatively small sample size and narrow age variability in our cohort are additional limitations of our study. However, the narrow age span and relatively younger mean age of this cohort allowed for the examination of earlier aging effects that are quite distinct from what is commonly seen in AD – a disease with exponential increase in prevalence with age. Despite these limitations, this study was strengthened by the use of explicit cortical morphology mapping and advanced imaging software allowing for the detection of subtle structural effects associated with normal aging.

Disclosure Statement

The authors have no conflicts of interest to disclose.