|Home | About | Journals | Submit | Contact Us | Français|
Previous studies suggest that in Alzheimer's disease (AD) the Apolipoprotein E (APOE) ε4 allele is associated with greater vulnerability of medial temporal lobe structures. However, less is known about its effect on the whole cortical mantle. Here we aimed to identify APOE-related patterns of cortical atrophy in AD using an advanced computational anatomy technique. We studied 15 AD patients carriers (ε4+, age:72±10SD years, MMSE:20±3SD) and 14 non-carriers (ε4-, age:69±9, MMSE:20±5) of the ε4 allele and compared them to 29 age-and-sex matched controls (age:70±9, MMSE:28±1). Each subject underwent a clinical evaluation, a neuropsychological battery, and high-resolution MRI. UCLA's cortical pattern matching technique was used to identify regions of local cortical atrophy. ε4+ and ε4- patients showed similar performance on neuropsychological tests (p>.05, t-test). Diffuse cortical atrophy was detected for both ε4+ (p=.0001, permutation test) and ε4- patients (p=.0001, permutation test) relative to controls, and overall gray matter loss was about 15% in each patients group. Differences in gray matter loss between carriers and non-carriers mapped to the temporal cortex and right occipital pole (20% greater loss in carriers) and to the posterior cingulate, left orbitofrontal and dorsal fronto-parietal cortex (5-15% greater loss in non-carriers). APOE effect in AD was not significant (p>.74, ANOVA), but a significant APOE by region (temporal vs fronto-parietal cortex) interaction was detected (p=.002, ANOVA), in both early and late-onset patients (p<.05, ANOVA). We conclude that the ε4 allele modulates disease phenotype in AD, being associated with a pattern of differential temporal and fronto-parietal vulnerability.
Alzheimer disease (AD) is a clinically heterogeneous disease, neuropathologically characterized by the accumulation of beta-amyloid plaques (Abeta) and neurofibrillary tangles (NFT) in the brain (Braak and Braak, 1991; Delacourte et al, 1999; Price and Morris, 1999; Thal et al., 2002; Haroutunian et al., 1999; Haroutunian et al., 1998). Multiple genes and environmental factors are believed to be involved in the pathogenesis and development of the disorder, through a complex interplay still largely unknown. To date the major genetic risk factor known for AD is the ε4 allele of the Apolipoprotein E (APOE) gene, that is present with a higher frequency in AD subjects than in the normal population (Strittmatter et al., 1993; Poirier et al., 1993; Corder et al., 1993; Tang et al., 1998; Kukull et al., 2002; Saunders et al., 2000), and lowers the mean age at onset of the disease in a dose-dependent fashion (Poirier et al., 1993; Blacker et al., 1997; Meyer et al., 1998; Goldstein et al., 2001).
ApoE, the protein coded by the APOE gene, is a lipid transport protein implicated in maintenance and repair of neuronal cells (Mahley, 1988), and current in vitro and animal model data strongly suggest that the ε4 allele is less efficient than other isoforms in these functions, through mechanisms that involve neuronal growth (Nathan et al., 1994), synaptic remodeling (Buttini et al., 2002) and cholinergic function (Buttini et al., 2002). The findings of faster brain atrophy rate (Chen et al., 2007; Moffat et al., 2000) and reduced hipppocampal volumes in healthy APOE ε4 carriers (Burggren et al., 2008; Plassman et al., 1997) seems to provide some in vivo evidence to the hypothesis of a pathogenic role of this allele in AD.
Less clear is whether the APOE ε4 allele has a role in modulating the expression of the disease. Post-mortem sudies reported a greater accumulation of AD pathological hallmarks in the neocortex of patients carrying the ε4 allele than those with no ε4 allele (Polvikoski et al., 1995; Tiraboschi et al., 2004; Nagy et al., 1995). In vivo data however agree only partially with these findings. Indeed, they reported greater atrophy in carriers in specific regions of the brain, namely in the hippocampus, entorhinal cortex, and temporal pole (Hashimoto et al., 2001; Geroldi et al., 1999; Juottonen et al., 1998; Lehtovirta et al., 1995). Conversely, frontal (Geroldi et al., 1999) and whole brain (Hashimoto et al., 2001; Yasuda et al., 1998) volumes seem to be relatively preserved in carriers than non-carriers. These findings, which seem to suggest a region-specific effect of the ε4 allele on brain atrophy rather than an overall greater disease severity in carriers (Hashimoto et al., 2001), are not conclusive (Lehtovirta et al., 1995; Jack et al., 1998) and are limited by the small number of studies available.
Here we aim to resolve discordances in previous MRI studies by investigating APOE-related patterns of atrophy over the whole cortex. Compared with prior studies, which were based on manually outlined regions of interest, here we used a recently developed semi-automated MRI analysis technique (Thompson et al., 2004) able to analyze the whole cortical mantle at thousands of homologous cortical points. We hypothesized that patterns of atrophy in AD patients would differ according to APOE genotype, with carriers showing greater involvement of the temporal lobe and non-carriers in the remainder of the cortex.
Subjects were recruited among ouptatients seen at the IRCCS Centro San Giovanni di Dio Fatebenefratelli (National Center for Alzheimer's Disease), in Brescia, Italy, between November 2002 and August 2005. Patients were enrolled in a study on neurodegenerative dementias aimed at detecting in vivo structural brain changes in various diseases, including Alzheimer's disease, frontotemporal dementia, Parkinson's disease, and dementia with Lewy bodies.
All patients underwent a standardized protocol including clinical, physical, neurological and neuropsychological evaluations. Each subject underwent MRI scan and laboratory exams, comprising complete blood count, chemistry profile, thyroid function, B12 and folic acid, and EKG. History was taken with a structured interview from patients' relatives (usually spouses or children), and was focused on those symptoms that might help in the differential diagnosis of the dementias (hallucinations, gait, language, and behavioural disturbances). Physical and neurologic examinations were performed by a geriatrician and a neurologist. Neuropsychological assessments were performed by a psychologist and included the evaluation of global cognitive status by Mini Mental State Examination (MMSE; Folstein et al., 1975) and a neuropsychological battery assessing: verbal and non-verbal memory with Rey's word list immediate and delayed recall tests (Carlesimo et al., 1996) and Rey figure delayed recall test (Caffarra et al., 2002); attention and executive functions with the Trail Making Test (Reitan, 1958; Amodio et al., 2002); language with the Phonological and Semantic fluency (Novelli et al., 1986) and Token (De Renzi and Vignolo, 1962; Spinnler and Tognoni, 1987) tests; and visuo-spatial abilities with the Rey figure copy (Caffarra et al., 2002). Severity of dementia was measured by the Clinical Dementia Rating score (Hughes et al., 1982).
Genomic DNA was extracted from whole-blood samples of subjects according to standard procedures. APOE genotyping was carried out by PCR amplification and HhaI restriction enzyme digestion. The genotype was resolved on 4% Metaphor Gel (BioSpa, Italy) and visualized by ethidium bromide staining (Hixson and Vernier, 1990).
Twenty-nine patients were diagnosed with AD according to the NINCDS criteria (McKhann et al., 1984) and were included in the study. Subjects were divided into two groups based on the presence (ε4+, n=15) or absence of the ε4 allele (ε4-, n=14). All the carriers were heterozygous (ε3ε4) for the ε4 allele whereas non-carriers were homozygous for the ε3 allele (ε3ε3) except for one subject who carried one ε2 allele.
A group of twenty-nine healthy persons was selected for comparisons with patients from those enrolled in a study on normal brain structure with MRI (ArchNor, Normative Archive of Structural Brain Magnetic Resonance Imaging), as described in detail elsewhere (Riello et al., 2005). Subjects were outpatients of the Neuroradiology Units of the Città di Brescia Hospital in Brescia, undergoing brain MR scan for reasons other than memory disturbance, cognitive impairment, degenerative diseases, or head trauma, and whose MR scan was negative. In detail, clinical exclusion criteria were: MR scan for memory problems or cognitive impairment, MR scan for clinical suspicion of neuro-degenerative diseases (Parkinson's disease, progressive supranuclear palsy, Huntington's disease, multiple system atrophy, etc), patient undergoing MR for suspected stroke, history of transient ischemic attack or stroke, head trauma, alcohol and substance abuse, cortico-steroid therapy, and loss of weight greater than 5 kilograms in the last 6 months, and cognitive impairment on neuropsychological testing. Radiological exclusion criteria included: brain mass, white matter hyperintensities with signs and symptoms of multiple sclerosis, aneurysm larger than 10 mm, arteriovenous malformation, malformations of the central nervous system, enlarged cisterna magna, meningioma, severe cerebrovascular disease, severe atrophy, large arachnoid cysts. Each subject underwent multidimensional assessment including clinical, neurological, and neuropsychological evaluation. Controls were matched 1:1 to AD patients according to age and sex.
Written informed consensus was obtained by all the subjects. No compensation was provided for study participation. The local ethics committee approved the study.
MR images were acquired at the Neuroradiology Unit of the Città di Brescia Hospital in Brescia with a Philips Gyroscan 1.0 T scanner. The acquisition protocol included T1-weighted and FLAIR sequences. High-resolution gradient echo T1-weighted sagittal 3D sequences were acquired using the following parameters: TR=20 ms, TE=5 ms, flip angle=30°, field of view=220 mm, acquisition matrix=256×256 and slice thickness=1.3 mm. Axial dual echo FLAIR sequences were acquired as follows: TR=5000 ms, TE=100 ms, flip angle=90°, field of view=230 mm, acquisition matrix=256×256, slice thickness=5 mm. T1-weighted images were used for cortical gray matter analyses, FLAIR sequences to assess subcortical cerebrovascular disease with the age-related white matter changes scale (Wahlund et al., 2001, total score ranging between 0 and 30).
Cortical gray matter was studied using the cortical pattern matching technique developed at the Laboratory of Neuroimaging (LONI) of the University of California at Los Angeles (Thompson et al., 2004).
The 3D images were reoriented along the AC-PC line and voxels below the cerebellum were removed with the MRIcro software (www.psychology.nottingham.ac.uk/staff/cr1/mricro.html). The anterior commissure was manually set as the origin of the spatial coordinates for an anatomical normalization algorithm implemented as part of the Statistical Parametric Mapping (SPM99) software package (www.fil.ion.ucl.ac.uk/spm). A 12-parameter affine transformation was used to normalize each image to a customized template in stereotaxic space, created from the MRI scans of the first 40 consecutively enrolled controls of the ArchNor project.
Individual brain masks for each hemisphere were extracted from normalized images with the automatic method EMS (expectation maximization segmentation; www.medicalimagecomputing.com/EMS) (Van Leemput et al., 1999a; Van Leemput et al., 1999b), visually inspected, and manually corrected with DISPLAY, a three-dimensional visualization program that enables simultaneous viewing of sagittal, coronal and axial slices of the brain (http://www.bic.mni.mcgill.ca/software/Display/Display.html), and allows the manual correction of errors between brain and non-brain tissue. The resulting masks were applied to normalized images to obtain ‘skull-stripped’ images of each hemisphere. A 3D model of hemispherical cortical surfaces was automatically extracted using intensity information (MacDonald et al., 1994). Normalized images were segmented into gray matter (GM), white matter and cerebrospinal fluid using an algorithm that employs partial volume correction and bias field correction (Shattuck et al., 2001).
Sulcal lines were traced by a single tracer (MP) on the cortical surfaces according to a previously validated anatomical delineation protocol (http://www.loni.ucla.edu/~khayashi/Public/medial_surface, http://www.loni.ucla.edu/~esowell/new_sulcvar.html). For each subject, 17 sulci were manually outlined on the lateral surface of each brain hemisphere, and a set of 12 sulci were traced on the medial surface; additional 3D lines were drawn to outline interhemispheric gyral limits. The reliability of manual outlining was assessed prior to experimental subject tracing with a standard protocol requiring the same rater to trace all lateral and medial sulci of 6 test brains (Sowell et al., 2002). At the end of the reliability phase, the mean 3D difference of the tracer from the gold standard was <3 mm everywhere for the medial sulci and <4.5 mm everywhere for the lateral sulci.
Sulcal curves and cortical surfaces were flattened and averaged across subjects to create a population specific template based on all subjects in the study (Thompson et al., 2000). Averaged sulci were then used as landmarks to warp each subject's anatomy into the template. The same deformation was applied to the segmented images, thus allowing measurement of GM at thousands of homologous cortical locations. Gray matter density (GMD) was computed at each cortical point as the proportion of GM tissue classified as GM in a sphere centered at that point, with a radius of 15 mm, and then averaged within each group to obtain the GMD mean. All morphological measurements were performed without knowledge of genotype.
A map of the average percentage GM reduction was created for each AD patient group (ε4+ and ε4-) computing the ratio at each cortical point between the mean GMD value at that point in the patient group and the GMD mean of the controls. This ratio allows visualization of the relative deficit in GM of APOE groups as a proportion, or percentage, of the normal values seen in healthy controls. Differences in severity of GM reduction between ε4 carriers and non-carriers was assessed by computing at each cortical point, the absolute difference between percentage GM reduction in ε4+ and ε4-. The resulting map shows the percentage differences between APOE groups.
A deformable Brodmann area (BA) atlas (Rasser et al., 2004; Rasser et al., 2005) was applied to the left and right hemisphere average models. Briefly, this involved the extraction of a three-dimensional model of the right cerebral hemisphere (MacDonald et al., 1994) of the Colin27 single-subject average brain MRI template (Holmes et al., 1998) followed by labeling with BA as delineated by the Caret software package (http://brainmap.wustl.edu/caret) (Van Essen et al., 2001; Van Essen, 2002). BAs were then deformed to the average right and left hemisphere templates using cortical pattern matching, followed by the tabulation of each subject's average gray matter loss in each BA.
Cortical pattern matching analyses were carried out in two steps, first comparing each patient group (APOE ε4+ and APOE ε4-) with controls and then comparing ε4+ patients with ε4-patients. For the first set of comparisons, we ran a regression at each cortical point between GMD and diagnosis. For the second comparison, we ran a regression entering percentage GM deficits and APOE (ε4+ and ε4-) as binary variables. The p-values representing the significance of the variable effect (diagnosis, APOE) were mapped onto the whole cortex, after setting a significance threshold of p=.05. Overall p-values for the uncorrected statistical maps were computed by running permutation tests. This test computes the chance of the observed pattern occurring by accident by running n=10,000 permutations of the variables of interest (diagnosis and APOE) at a threshold of p=.01 (Thompson et al., 2003).
In order to investigate the hypothesis of a region-specific effect of APOE on brain atrophy, we defined two regions on the BA atlas: (1) the temporal cortex, including the medial and lateral temporal lobes (superior, middle and inferior temporal gyrus, entorhinal, perirhinal and presubicular cortex, anterior temporal pole, and the fusiform gyrus), (2) the frontal and parietal neocortex (dorsolateral frontal and parietal cortex, orbitofrontal and subgenual cortex, anterior and posterior cingulate, and retrosplenial cortex; Figure 1). Main effect of APOE and cortical region on brain atrophy was investigated using an analysis of variance (ANOVA) model, where APOE (ε4+ vs ε4-) was the between-subject factor, region (temporal vs fronto-patietal cortex) was the within-subject factor, and percentage GM reduction was the dependent variable.
In order to assess whether age at onset of the disease had an impact on APOE effect, the analyses were repeated separately in early (<65 years, n=14) and late-onset (>65 years, n=15) patients, by comparing younger and older patients with their age-matched controls.
Table 1 shows that APOE ε4 carriers and non-carrier patients were comparable in age (p=.358 on t-test), duration of dementia (p=.519), cerebrovascular burden (p=.203), and severity of the disease (p=.287 on chi-square), CDR scores being indicative of mild dementia in the most of the patients (Table 1). The ε4+ group had a greater proportion of women than ε4- (p=.009) and a lower educational level (p=.033; Table 1). Because some previous studies showed gender differences in gray matter volumes of healthy and AD subjects (Luders et al., 2006; Sowell et al., 2007; Juottonen et al., 1998), we replicated the analyses considering only women in order to ensure that these differences did not influence our results.
Controls were similar to ε4+ and ε4- patients with regard to age (p>.457 on t-test) and sex (p>.114 on chi-square test, Table 1). Years of education were comparable to that of ε4 non-carriers (p=.572) and higher than that of ε4 carriers (p<.001, Table 1). Global cognition was lower in patients compared with controls on the MMSE test (p<.001, Table 1) and in all the domains investigated by the neuropsychological battery (p<.038, Table 2). The degree of impairment was similar in ε4+ and ε4- patients on both MMSE scores (p=.862) and neuropsychological tests (p>.063; Table 2).
Both ε4+ and ε4- showed widespread cortical atrophy, involving most of the neocortex and sparing only the somatosensory and motor areas, the anterior cingulate gyrus, and the medial orbitofrontal cortex (thresholded at p<.01 uncorrected, Figure 2, top panel). In ε4+ patients, regions of significant atrophy mapped to the whole left and right temporal lobes, and to the occipital lobes, retrosplenial and posterior cingulate area (Figure 2, top left panel). The same regions were affected in the ε4- group, but involvement of the temporal lobe was less selective and conversely fronto-parietal atrophy appeared more diffuse (Figure 2, top right panel). Patterns of atrophy in the medial wall involved the posterior cingulate and retrosplenial cortex in both groups. The comparisons between patients and controls were highly significant after correction for multiple comparisons (p=.0001 for both ε4+ vs controls and ε4- vs controls, permutation test).
Percentage maps showed that overall severity of atrophy was similar, with ε4+ and ε4- showing an average GM reduction of 15% vs 14% in the left, and 16% vs 17% in the right hemisphere respectively (p>.80 on t-test). Regionally, the most severe GM reductions (>20%) in carriers were in the entorhinal cortex, anterior temporal pole, superior and middle temporal gyrus, and in the ventral and dorsal occipital lobe bilaterally (Figure 2, bottom left panel). Non-carriers were more severely affected bilaterally in the superior and middle frontal gyri, superior temporal gyrus, retrosplenial cortex, posterior cingulate and orbitofrontal medial right cortex (Figure 2, bottom right panel).
Direct comparisons between maps of percentage reduction in the ε4+ and ε4- groups revealed APOE-associated regions of atrophy. These analyses detected greater involvement of the medial and lateral temporal lobes, and of the right occipital pole (p<.01, uncorrected; Figure 3A, left) in the ε4+ group. The opposite effect (greater atrophy in non-carriers than carriers) was detected in the posterior cingulate and left lateral orbitofrontal cortex (p<.01 uncorrected; Figure 3A, right), and in the right dorsolateral cortex (p<.05 uncorrected). Differences significant at the uncorrected threshold of p<.01 corresponded to approximately 20% greater GM loss in the ε4 carriers than non-carriers (Figure 3B, left), whereas the opposite effect amounted to about 5-15% greater GM loss for non-carriers than carriers (Figure 3B, right). Statistical maps were not significant after controlling for multiple comparisons (p>.33; permutation test).
GM loss in the temporal cortex region defined on BAs was 19% in carries and 16% in non-carriers relative to controls. In the fronto-parietal cortex, carriers exhibited a reduction of 14.5% vs 16% in non-carriers (Figure 3C). The ANOVA model showed in the whole sample a significant effect of region (p=.001) but not of APOE (p=.740) on GM loss. However, the interaction between APOE status and brain region was significant (p=.002,ANOVA; Figure 3C). When patients were separated into younger (ε4+: n=7, ε4-: n=7) and elderly (ε4+: n=8, ε4-: n=7), the interaction term in the ANOVA model was significant in both early (p=.013) and late-onset (p=.050) patients (Supplementary Figure S3).
The analyses carried out on the women samples (ε4+: n=14; ε4-: n=7) provided similar results. Patterns of atrophy remained highly significant in both carriers (p=.0001 on permutation test) and non-carriers (p=.0003) relative to controls, and confirmed that regional differences in GM loss between carriers and non-carriers mapped to the temporal and fronto-parietal cortex (p<.01 uncorrected). These differences were not significant after controlling for mutiple comparisons (p>.57, permutation test). The interaction term between APOE and region remained significant (p=.008, ANOVA. Supplementary materials).
In the present study we found that APOE modulates AD pathology, the ε4 allele being associated with a pattern of increased susceptibility of the temporal cortex together with lower vulnerability in the fronto-parietal neocortical regions.
The finding of a significant interaction between APOE and region affected is in line with previous studies suggesting a region-specific effect of APOE on brain atrophy (Hashimoto et al., 2001; Geroldi et al., 1999), rather than greater disease severity in ε4 carriers. This modulating effect became evident when the analyses were driven by a-priori hypotheses consistent with previous data (Geroldi et al., 1999; Hashimoto et al., 2001). Although early studies did not assess the interaction between APOE and region, the findings of greater atrophy of the medial and lateral temporal cortex in ε4 carriers (Juottonen et al., 1998; Lehtovirta et al., 1995; Pennanen et al., 2006; Hämäläinen et al., 2008; Thomann et al., 2008) together with milder involvement of whole brain volumes (Yasuda et al., 1998; Lehtovirta et al., 1995) are consistent with the hypothesis of a modulating effect of APOE. Lehtovirta did not actually report a significant difference between carriers and non-carriers in the frontal lobe, notwithstanding there was a trend for larger frontal volumes together with significantly smaller hippocampal volumes in carriers (Lehtovirta et al., 1995). It is likely that the interaction effect detected here would be stronger if patients carrying two ε4 alleles were included, as APOE effect on brain atrophy has been reported to be proportional to allele dose (Yasuda et al., 1998; Geroldi et al., 1999). The lack of ε4 homozygous in our sample may indeed have reduced the statistical power of our findings.
The mechanism through which APOE affects regional atrophy might involve the role of apoE on the deposition of NFT. The temporal cortex is highly susceptible to NFT deposition whereas other neocortical areas are less vulnerable (Braak and Braak, 1991; Arnold et al., 1991). An association between APOE ε4 and increased NFT deposition in the temporal cortex has been demonstrated in vivo in transgenic mice (Brecht et al., 2004; Tesseur et al., 2000), and from neuropathological examinations (Tiraboschi et al., 2004; Nagy et al., 1995). The effect of APOE on atrophy in the temporal region might therefore be mediated by greater NFT deposition. Furthermore, the ε4 allele in the temporal lobe has been associated with impaired synaptic activity (Buttini et al., 2002), reduced neurite outgrowth (Sun et al., 1998), more severe Abeta deposition (Tiraboschi et al., 2004; Nagy et al., 1995) and cholinergic deficits (Buttini et al., 2002). There are thus several possible mechanisms in addition to (or concurrent with) NFT deposition through which the ε4 allele may modulate temporal lobe atrophy in AD. Less straightforward is the interpretation of how APOE ε4 may be associated with a lesser degree of atrophy in the frontal and parietal neocortex; previous studies indeed reported a detrimental effect of the ε4 allele on Abeta deposition and cholinergic/synaptic activity in the fronto-parietal cortex as well (Buttini et al., 2002; Soininen et al., 1995; Beffert et al., 1999). These data may thus indicate that the effect of APOE ε4 on atrophy may not be a simple consequence of Abeta deposition and of cholinergic/synaptic deficits. Furthermore, the relative preservation of the fronto-parietal cortex may alternatively suggest that APOE ε4 could be less detrimental in some respects. The ε4 allele indeed may be more efficient than other isoforms in aspects related to development, such as in promoting cognitive development (Yu et al., 2000), although region-specific effects in the frontal and parietal neocortex have never been reported. A recent study has shown that – contrary to expectation – ε4 carriers had a better long-term outcomes after brain injury (Willemse-van Son et al., 2008). These data, together with findings of longer survival, slower cognitive and functional decline, slower atrophy rate in AD patients who were ε4 carriers (Frisoni et al., 1995; Hoyt et al., 2005; Stern et al., 1997; Sluimer et al., 2008), may indicate that ε4 effect may be less detrimental than other isoforms in the long term (Teasdale, 2008). As experiments from basic and molecular studies are usually carried out over a relatively short period of time, they may lack the ability to address some aspects of the diseases that can be captured in vivo.
In the present study, carriers and non-carriers showed comparable cognitive deficits in all the domains investigated. This is to some extent in contrast with earlier studies reporting greater memory deficits in ε4 carriers (Smith et al., 1998), and conversely more severe impairment in non-memory domains in AD patients lacking the ε4 allele (Lehtovirta et al., 1996; Van Der Vlies et al., 2007). The majority of previous studies that reported neuropsychological testing scores together with MRI indexes of atrophy found lower performance on memory tests in ε4 carriers (Lehtovirta et al., 1995; Juottonen et al., 1998), and one reported a significant association between number of ε4 alleles and score on attention and intelligence scales (Hashimoto et al., 2001). A possible explanation for the discrepancies with previous studies may lie in differences in the sampling of the ε4 group: the association between the ε4 allele and atrophy seems to be gene-dose dependent (Yasuda et al., 1998; Geroldi et al., 1999), and previous studies included a subgroup of patients who were homozygous for the ε4 allele. Thus, the lack of patients carrying two ε4 alleles in our study may explain why we did not find a significant effect of APOE on cognition in our sample.
This study has both strengths and limitations. The major strength involves the spatial accuracy of our mapping technique that allows to compare anatomical features over the whole cortical mantle between subjects. Commonly used methods such as manual regions of interest tracing suffer the disadvantage of being spatially less detailed and typically they are strongly dependent on a priori hypothesis. As for limitations, although the sample size of our AD groups was comparable to that of previous studies, confirmation is required in larger samples to take into account factors that may have influenced our results to some degree, such as age, sex, and education differences. In this study indeed there was a larger proportion of women in carriers than non-carriers, and educational level differed between groups. A previous study (Juottonen et al., 1998) reported that atrophy of the entorhinal cortex was more pronounced in female patients carrying at least one ε4 allele than in men. Conversely, some authors showed greater vulnerability of the frontal and parietal cortex in men and persons with a higher educational level (Kidron et al., 1997). Although it is generally agreed that these factors may play a role in modulating susceptibility to AD (Lahiri et al., 2004; Azad et al., 2007), it is less clear whether they can affect disease phenotype. In the present study, control subjects were overall matched to patients and this should have attenuated differences due to sociodemographic differences. Secondly, when we replicated the analyses excluding men, the results remained unchanged thus confirming that the effect observed was not due to differences in sex and/or education. Clearly, further studies including groups well-balanced in their sociodemographic features are recommended. Another limitation of the study is that APOE genotype was not available for all controls. Thus, we could not investigate changes modulated by APOE from those representing APOE-related morphological traits.
In the present study, we provided an independent confirmation of previous findings about a modulating effect of APOE on brain atrophy, the ε4 allele being associated with greater susceptibility of the temporal cortex, and conversely less vulnerability in the frontal-parietal cortex. These data suggest that the ε4 allele modulate AD phenotype. The mechanism underlying APOE ε4 effect on cortical atrophy however may be quite complex and involve several processes related to AD pathology.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.