Alzheimer’s disease (AD), the most common neurocognitive disorder, is characterized by a progressive loss of neurons and synapses, with cerebral deposits of amyloid-β (Aβ) senile plaques [21
] and neurofibrillary tangles (NFTs) [18
]. The apolipoprotein E (APOE) polymorphism, specifically the ε
4 allele, is widely accepted as the most robust genetic risk factor for late onset Alzheimer’s disease (LOAD) [4
]. APOE ε
4 has been reported as being related to a lower level of Aβ42
in normal elderly adults [25
] and a high level of tau in AD [17
]. Both of these appear to contribute to the development of AD [20
A growing number of studies have indicated that APOE ε
4 modulates brain morphology specifically by a loss in hippocampal volume [6
] and by causing abnormal white matter integrity [27
]. The correlation between APOE ε
4 and cortical thickness has also been increasingly reported. For example, healthy adults with the ε
4 allele showed age-related reductions in cortical thickness in the medial prefrontal and pericentral cortex [9
]. Thinner cortical thickness was also observed in the entorhinal cortex and the subiculum in healthy adults who had the ε
4 allele [2
Just as the APOE ε
4 allele has been identified as a major risk factor for LOAD, other studies have found that the ε
2 allele may have some protective qualities, since carriers appear to have a lower risk of developing Alzheimer’s disease [10
]. One study [32
] observed increased cortical thickness in healthy children and adolescents who were ε
2 carriers. Other researchers [6
], however, did not find any effect of ε
2 on brain volume in healthy elderly subjects.
Since APOE has been identified as major risk factor in LOAD, finding the effects of this gene on cortical thickness in elderly adults is important. Our goal was, therefore, to examine the cortical thickness across the entire cortex of healthy, elderly adults to see if it correlates with the various APOE genotypes, especially with the ε2 allele. We expected that we would find increased cortical thickness for the ε2 allele carriers and decreased cortical thickness for the ε4 allele carriers, each compared to the ε3 homozygotes.
Data used in this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative
) database (www.loni.ucla.edu/ADNI
). The initial goal of ADNI was to recruit 800 adults, ages 55–90, to participate in the research—approximately 200 cognitively normal older individuals to be followed for 3 years, 400 people with MCI to be followed for 3 years, and 200 people with early AD to be followed for 2 years. For up-to-date information see www.adni-info.org
Our downloaded data initially included 225 baseline NC scans. We excluded 5 subjects because they converted to MCI or AD in 36 months. To ensure the accuracy of the segmentation, we manually checked the interface between the grey matter and white matter as well as the one between the white matter and the cerebrospinal fluid (CSF), and on this basis excluded 46 additional subjects due to segmentation errors. The resulting data for the cortical thickness analysis included 164 healthy, cognitively normal, elderly subjects, all collected from ADNI sites. The individuals were grouped into ε2 carriers (21), ε3 homozygotes (103) and ε4 carriers (37), with ε2ε4 individuals excluded from consideration. The Mini Mental State Examination (MMSE) and the education level had also been recorded as baseline data. Sample characteristics are presented in . In addition, those individuals in this study group for whom an intact record of their CSF biomarkers, including total tau (t-tau), hyperphosphorylated tau (p-tau), and Aβ42, was available were separately grouped into APOE ε2 carriers (14), ε3 homozygotes (50) and ε4 carriers (14).
Sample demographic characteristics.
The data sets included standard T1-weighted MR images acquired sagittally using volumetric 3D MPRAGE with 1.25 mm × 1.25 mm in-plane spatial resolution, TE = 3.9 ms, TR = 90 ms and 1.2 mm thick sagittal slices (8° flip angle). All of the images were obtained using 1.5 T scanners. The MRI scans were acquired at multiple sites using a GE, Siemens or Philips 1.5 T system. The images were preprocessed using a number of steps detailed on the ADNI website. This preprocessing was done to correct for differences across scanners used at various ADNI sites.
We processed each scan using FreeSurfer [5
) with its volume and surface pipeline. Starting with the segmentation of white matter and the tessellation of the grey/white matter boundary, an initial surface was obtained after an automated topological correction. We used this surface as the initial shape for a deformable model that was used to reconstruct the pial surface.
When all the surfaces had been reconstructed, the cortical thickness was computed. Each subject’s cortical thickness was measured in native space [11
]. Surface based registration [13
] was used to construct an average template, and all the individual reconstructed cortical surfaces were aligned to it. The cortical thickness measurements at the interconnection between the two hemispheres were masked because there is very little grey matter there. Finally, we smoothed the thickness using a heat kernel [3
] 30 mm wide to increase the signal-to-noise ratio and improve the ability to detect morphometric variations.
We performed an Analysis of Covariance (ANCOVA) using SPSS 13.0 statistical software to assess the effects of APOE genotypes on biomarker levels (tau, Aβ42), controlling for age and gender. After that, a post hoc multiple comparison using Fisher’s LSD was conducted when the ANCOVAs were significant. In all tests, results with probability values less than 0.05 were considered statistically significant.
We used SurfStat (http://www.math.mcgill.ca/keith/surfstat/
) toolbox for Matlab (R2007a, The Mathworks, Natick, MA, USA) to perform statistical analyses of the cortical thicknesses. Statistical tests were performed at every unmasked point on the pial surface. To test for variability in the thickness of the cortex, we applied a general liner model to check point-wise thickness differences using the APOE genotypes as fixed factor, and age and gender as covariates. The ensuing p
-values were adjusted for multiple comparisons on the cortical surface to control for the false positive rates by the random field theory [39
]. We discarded statistically significant clusters containing fewer than 50 points, in order to reduce the possible influence of noise.
We tested the effects of the APOE genotypes on the CSF biomarkers t-tau, p-tau and Aβ42 level on a smaller subset of 78 samples. The ANCOVA showed a significant effect of APOE on Aβ42 level (p < 10−4). In addition, a post hoc comparison showed a statistically significant stepwise trend toward lower Aβ42 levels, with ε2 being the highest, ε4 the lowest, and ε3 homozygotes occupying an intermediate position (p = 0.034 for the step from ε2 to ε3, and p < 10 e–4 for the one from ε3 to ε4) (). In contrast to Aβ42, the effects of APOE on CSF t-tau or p-tau were only a trend (p = 0.063 and p = 0.054 respectively), and this was limited to comparisons of APOE ε4 and ε2 carriers (p = 0.024 for t-tau, and p = 0.016 for p-tau) (). No significant association between the APOE genotype and the MMSE score was observed (p = 0.170).
Fig. 1 Mean biomarker levels (t-tau, p-tau and Aβ42) for the APOE genotype groups. The APOE ε2 carriers are represented in black, the ε3 homozygotes in grey and the ε4 carriers in white. The CSF Aβ42 levels show a significant (more ...)
The results of the group comparisons of cortical thickness are displayed in . No increases, only decreases, in cortical thickness were found in any region in the APOE ε4 carriers and the ε3 homozygotes when compared with the same regions in the ε2 carriers. This was also true in the APOE ε4 carriers when they were compared with the ε3 homozygotes. The APOE ε2 allele carriers had significantly increased cortical thickness primarily in the bilateral superior temporal cortices, bilateral dorsolateral prefrontal cortices, left supramarginal gyrus, left precentral and postcentral gyri, left parietal operculum and right parahippocampal region compared to the APOE ε3ε3 subjects. The APOE ε2 carriers also had a significantly thicker cortex primarily in the bilateral lateral and medial frontal regions, the right parahippocampal cortex, right temporoparietal cortex and bilateral temporal regions than did subjects who were ε4 allele carriers. The APOE ε4 carriers showed significant atrophy in the left medial prefrontal cortex and left orbitofrontal cortices compared with the ε3 homozygotes. After performing multiple comparison corrections using random field theory, we determined that the APOE ε2 carriers had significantly thicker cortical thickness in the left superior temporal cortex compared with the ε3 homozygotes. Moreover, the cortices in the left superior temporal and left dorsolateral prefrontal region were thicker in the APOE ε2 carriers than in the ε4 carriers. However, no significant differences remained when comparing the APOE ε3 homozygotes with the ε4 carriers after applying the multiple comparison correction.
Fig. 2 Cortical thinning maps (in percentage). Only points with statistically significant differences (p < 0.05) are displayed. The top shows uncorrected p < 0.05, and the bottom shows p < 0.05 corrected. The color bar displays the T (more ...)
In this study, we used a surface-based approach to quantify the local cortical thickness in healthy nondemented adults with different APOE genotypes. We observed a stepwise lower Aβ42 level for different APOE genotypes, and thicker cortical thickness when comparing APOE ε2 carriers to ε4 carriers or ε3 homozygotes in specific regions.
We examined the effect of APOE ε
2 on brain morphology by comparing subjects with this genotype with ε
3 homozygotes and ε
4 carriers. Interestingly, when we compared the others with the ε
2 carriers, the ε
3 homozygotes and the ε
4 carriers had a similar thinner cortical area, which was primarily located in the bilateral lateral prefrontal cortices, the bilateral temporal cortices, and the right parahippocampal region. The differences in the left superior temporal gyrus, which is associated with speech production [22
], remained significant after a multiple comparison correction. A loss of the speech production function is a symptom in the development of AD [16
]. Moreover, we observed a significant stepwise lowering of the Aβ42
levels across the APOE genotypes and a significantly higher t-tau or p-tau level when comparing APOE ε
4 carriers. Importantly, the accumulation of Aβ peptide and tau in neurons has been implicated in the development of AD [20
]. Aβ peptide deposition [8
] and neurofibrillary tangle development [36
] have been found to occur early in the development of AD. Thus, the same biological markers, that is, those for tau and Aβ42
, which contribute to the Aβ peptide and neurofibrillary tangles could also induce a metabolic decline leading to neuronal loss and thus perhaps cortical thinning [19
In addition to thinning in the superior temporal cortex, the ε
4 allele was associated with greater atrophy in the bilateral pre-frontal cortex when compared with the ε
2 allele. Moreover, after applying a multiple comparison correction, the differences in the left dorsolateral prefrontal cortex, which has been suggested as being correlated with cognition control [26
] and long term memory formation [1
], remained significant. Again, the failure of these functions is characteristics of AD [15
]. Thus, we can conclude that the finding that APOE ε
2 allele carriers possess a thicker left dorsolateral prefrontal cortex may indicate a specific protective role against possible memory loss in the development of AD.
We also found that APOE ε
4 carriers displayed atrophy (uncorrected p
< 0.05) in the left prefrontal cortex and the left orbitofrontal cortex compared to ε
3 homozygotes. Although when comparing ε
3 with ε
4 individuals we did not observe atrophy in the parahippocampal cortex, which is the earliest region affected in AD [23
], atrophy was present in this region in the ε
3 homozygotes or ε
4 carriers when compared with the APOE ε
2 carriers (uncorrected p
< 0.05). Published studies about APOE ε
4s effect on brain morphology, especially in the parahippocampal region, are controversial. Several studies have suggested that APOE ε
4 is related to hippocampal volume loss [6
], but one recent study [29
] which also used ADNI data found no effect of APOE ε
4 on hippocampal volume loss in the normal control group. Because most published studies have subdivided APOE genotypes into APOE ε
4 non-carriers and carriers, the inclusion of the ε
2 carriers may have caused a continuum of thicknesses that caused any e4-related thinning of the parahippocampal cortex to be lost in the data.
The results of the present study may also indicate that APOE genotypes affect cognitive function in normal aging. A large meta-analysis study [38
] investigated the effects of the APOE genotype on cognition in the nondemented population. Specifically, in the non-demented population the APOE ε
4 allele carrier often had impaired cognitive functioning, and ε
2 allele carriers showed better performance on episodic memory. Moreover, increased cortical thickness, including that in the frontal and temporal regions, was associated with higher cognitive performance in older healthy adults [14
]. Our results are compatible with these findings. In this study, we did not find a significant relationship between the APOE genotypes and the MMSE scores. One of major reasons may be a ceiling effect that appeared because the healthy nondemented individuals had high scores that were similar to those of the subjects on these measures. However, in a previous study, APOE ε
4 carriers were observed to have a greater rate of cognitive decline as shown by their MMSE scores using nondemented individuals [7
]. Thus, the effects of APOE genotypes on cognitive decline in normal aging require further investigation.
In conclusion, the APOE ε2 allele may have a specific protective role in the development of AD. To our knowledge, this is the first study to explore the cortical thickness pattern between APOE genotypes in nondemented elderly subjects. Further studies are needed to clarify the exact mechanism and role of the APOE genotypes in the cognitive decline associated with normal aging as well as in the development of AD.