Apolipoprotein E ϵ4 Allele, Temporal Lobe Atrophy, and White Matter Lesions in Late-Life Dementias

Objective  To examine the relationship between the apolipoprotein E (APOE) ϵ4 genotype, medial temporal lobe atrophy, and white matter hyperintensities on magnetic resonance imaging in late-life dementias.

Design  Structural magnetic resonance imaging study using T₂-weighted and proton density–weighted axial scans and T₁-weighted coronal scans.

Setting  Community-dwelling population of elderly patients prospectively chosen from a clinical case register of consecutive referrals to old age psychiatry services.

Subjects  Twenty-five subjects with Alzheimer disease (by criteria of the National Institute of Neurological and Communication Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association; mean age, 77.8 years), 22 subjects with dementia with Lewy bodies (consensus criteria; mean age, 77.2 years), and 24 subjects with vascular dementia (by criteria of the National Institute of Neurological Disorders and Stroke and the Association International pour la Recherche et l'Enseignement en Neurosciences; mean age, 76.9 years) were selected. Subjects were well matched for age, sex, duration of illness, and cognitive function.

Main Outcome Measures  The APOE genotype was determined using the polymerase chain reaction method, and medial temporal lobe atrophy and white matter hyperintensities (periventricular and deep white matter) were visually rated using standardized scales.

Results  In all subjects with dementia, no significant associations were noted between APOE ϵ4 status and medial temporal lobe atrophy (mean score: 0 ϵ4=4.5, 1 ϵ4=4.5, and 2 ϵ4=4.3; P=.90), periventricular hyperintensities (0 ϵ4=3.3, 1 ϵ4=3.1, and 2 ϵ4=2.9; P=.83), and white matter hyperintensities (0 ϵ4=5.3, 1 ϵ4=4.9, and 2 ϵ4=4.9; P=.79).

Conclusions  The APOE ϵ4 allele does not determine medial temporal lobe atrophy or white matter lesions, as measured by magnetic resonance imaging in patients with Alzheimer disease, vascular dementia, or dementia with Lewy bodies. Although APOE ϵ4 may modify the risk for acquiring dementia, this finding provides further evidence that APOE ϵ4 does not influence pathological processes thereafter.

THE PRESENCE of the apolipoprotein E (APOE) ϵ4 allele has been identified as a major risk factor for both sporadic and familial late-onset Alzheimer disease (AD). Located on chromosome 19 at q13.2, the APOE gene codes for 3 major isoforms: ϵ2, ϵ3, and ϵ4. An increased frequency of the APOE ϵ4 allele has been observed in subjects with AD¹ and those with other types of dementia, including dementia with Lewy bodies (DLB)^2-5 and vascular dementia (VaD).^6-8 The pathological pathway through which the apolipoprotein E exerts influence has yet to be determined. In particular, it is not known whether the effects of the apolipoprotein E are specific to AD or expressed through a common pathway, independent of the diagnosis.

Recently, several studies have reported a link between the presence of APOE ϵ4 and specific morphologic changes identified on neuroimaging. Lehtovirta and colleagues,^9,10 Tanaka et al,¹¹ and Juottonen et al¹² studied subjects with AD and found increased atrophy in medial temporal lobe structures on magnetic resonance imaging (MRI) in those possessing an APOE ϵ4 allele. The relationship between medial temporal lobe atrophy (MTA) on MRI and the APOE ϵ4 status in subjects with DLB and VaD has not yet been examined.

Imaging studies have also reported a high prevalence of white matter changes in subjects with AD^13,14 and VaD.^15,16 These lesions have been linked to vascular causes and diseases, and in postmortem studies, APOE ϵ4 has been strongly associated with cerebral amyloid angiopathy,^17,18 suggesting the possibility of an association between APOE ϵ4 status, microvascular ischemia, and white matter lesions.

The primary aim of this study was to examine the relationship between the APOE ϵ4 status, MTA, and white matter lesions on MRI and to investigate whether any association was specific to AD or present in other dementias when associated with APOE ϵ4. Our hypotheses were that APOE ϵ4 would be associated with increased prevalences of MTA and white matter lesions and that this association, if observed, would be independent of the diagnosis. To test these hypotheses, the APOE genotype was determined in a representative sample of subjects with AD, VaD, or DLB, and MTA and white matter lesions were rated on MRI in a manner blinded to the diagnosis and APOE status.

Seventy-one subjects older than 60 years who fulfilled the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,¹⁹ criteria for dementia were recruited from a community-dwelling population of patients with an informant in regular contact. Subjects were prospectively chosen from a clinical case register of consecutive referrals to Old Age Psychiatry Services, Newcastle General Hospital, Newcastle upon Tyne, England. The recruitment of subjects with DLB was supplemented by referrals to a specialist dementia clinic. The research was approved by the local ethics committee, and all subjects, as well as the nearest relative for patients, gave informed consent after the nature of the procedures had been fully explained.

Standardized clinical diagnostic criteria were used to characterize the type of dementia. The diagnoses of AD, VaD, and DLB were made in accordance with criteria of the National Institute of Neurological and Communication Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association,²⁰ the National Institute of Neurological Disorders and Stroke and the Association International pour la Recherche et l'Enseignement en Neurosciences,²¹ and the Consortium on DLB International Workshop,²² respectively, by consensus among 3 experienced raters (J.O'B., C.B., and I.McK.). The diagnosis was made in a manner blinded to MRI findings and APOE status. Applying these criteria, 25 subjects had AD, 24 had VaD, and 22 had DLB. Cognitive function was measured using the Mini-Mental State Examination²³ within 3 months of obtaining an MRI scan.

All scans were performed on a 1.0-T MRI scanner (Siemens Magnetom Impact; Siemens Analytical X-ray Instruments, Erlangen, Germany). To assess MTA, T₁-weighted coronal images of the whole temporal lobe were obtained with a slice thickness of 5 mm and no interslice gap. These images were acquired by reformatting a 3-dimensional data set perpendicular to the long axis of the hippocampus (magnetization prepared rapid-acquisition gradient echo: repetition time, 11.4 milliseconds; echo time, 4.4 milliseconds; inversion time, 400 milliseconds; time delay, 50 milliseconds; matrix, 256×256; and slice thickness, 1 mm).

To assess white matter lesions, whole-brain axial images of 5-mm thickness (0.5-mm gap) were acquired using proton density–weighted and T₂-weighted turbo or fast-spin echo sequences (rapid acquisition with relaxation enhancement: repetition time, 2800 milliseconds; echo time, 14/85 milliseconds; matrix, 256×256; field of view, 230 mm, giving a pixel size of 0.92×0.92 mm; and acquisition time, 4 minutes 13 seconds).

All scans were rated with the diagnosis and the APOE genotype blinded. A standardized scale²⁴ was used to rate left and right MTA from copies of T₁-weighted coronal images. This scale rates MTA as 0 (indicating absent) to 4 (indicating severe) according to the width of the surrounding cerebrospinal fluid spaces (choroidal fissure and temporal horns) and the height of the hippocampal formation, which includes the hippocampus proper, the subiculum, and the parahippocampal and dentate gyri. The scans were scored using the following criteria: 0=no change; 1=mild increase in the width of the choroid fissure, normal temporal horn width, and the height of hippocampal formation; 2=moderate increase in the width of the choroid fissure, mild increase in the width of the temporal horn, and mild decrease in the height of the hippocampal formation; 3=severe increase in the width of the choroid fissure, moderate increase in the width of the temporal horn, and moderate decrease in the height of the hippocampal formation; and 4=severe increase in the width of the choroid fissure and temporal horn and severe decrease in the height of the hippocampal formation. Scans were rated by consensus among 3 experienced raters (J.O'B., A.G., and P.S.). For the purposes of analysis, left and right MTA scores were summed to give a combined score, ranging from 0 to 8.

White matter lesions were rated using a semiquantitative scale²⁵ from copies of proton density– and T₂-weighted axial images by an experienced rater (P.S.). The scale provided a measurement of periventricular hyperintensities and deep white matter hyperintensities. Periventricular hyperintensities (maximum score=6) represented the combined score for bands and frontal and occipital caps according to their size (0=absent; 1=≤5 mm in width; and 2=>5 mm and <10 mm in width). White matter hyperintensities were rated in 4 regions: frontal, temporal, parietal, and occipital (0-6 in each area, maximum total score=24) according to the size and number of lesions (0=absent; 1=<3 mm, ≤5 lesions; 2=<3 mm, ≥6 lesions; 3=4-10 mm, ≤5 lesions; 4=4-10 mm, ≥6 lesions; 5=≥11 mm, ≥1 lesions; and 6=confluent).

The APOE genotypes were analyzed using the standard polymerase chain reaction method. Genomic DNA was isolated from whole blood using a proprietary extraction method according to the manufacturer's instructions (QIAamp Blood Kit; QIAGEN, Crawley, England). Amplification of the APOE gene containing the allelic sites was performed by a modification of the method of Wenham et al²⁶ and Hixson and Vernier.²⁷ The following polymerase chain reaction primers were used in the amplification: forward: 5′-TCCAAGGAGCTGCAGGCGGCGCA-3′, and reverse: 5′-ACAGAATTCGCCCCGGCCTGGTACACTGCCA-3′. The reactions were performed in a final 50-mL volume of standard buffer containing 15 pmol of each primer; Taq polymerase (Pharmacia Biotech Ltd, St Albans, England), 1.0 U; 10% dimethyl sulfoxide; 200 mmol/L of each deoxynucleotide; and DNA, 200 ng. Reaction conditions were an initial denaturation at 95°C for 5 minutes, followed by 40 cycles of annealing at 65°C for 30 seconds, an extension at 70°C for 90 seconds, and denaturation at 94°C for 30 seconds. The resultant amplification products were then digested overnight at 37°C with CfoI enzyme (Boehringer, Lewes, England), 2.5 U, in the buffer supplied. Following digestion, the polymerase chain reaction products were electrophoresed through a composite 3% Nu Sieve and 1% standard agarose (Flowgen, Lichfield, England) gel, and the bands were visualized using ethidium bromide fluorescence.

A commercial statistical software package (Statistical Package for the Social Sciences for Windows, release 7.5; SPSS Institute, Inc, Chicago, Ill) was used for data analysis. Differences between groups on continuous variables were assessed using analysis of variance with the post hoc Scheffé test. For nonparametric data, the Kruskal-Wallis test, Mann-Whitney test, or Pearson χ² statistic was used as appropriate, with the Fisher exact probability test calculated for 2×2 tables when the expected cell frequency was less than 5. All statistical tests were 2-tailed and were regarded as significant at P<.05.

The subject characteristics and frequencies of APOE alleles are shown in Table 1. The dementia groups were matched for age, sex, age at onset of illness, duration of illness, and cognitive function. No significant differences were found in the frequency of APOE ϵ2, ϵ3, and ϵ4 among the dementia groups, although subjects with AD had the highest frequency of APOE ϵ4.

When the subjects with dementia were divided according to the presence (n=38) or absence (n=33) of the APOE ϵ4 allele, no significant differences were noted among the groups in age (ϵ4 present vs ϵ4 absent: 77.8 vs 76.7 years), age at onset of illness (74.7 vs 73.0 years), duration of illness (36.1 vs 44.2 months), education (9.6 vs 9.6 years), or Mini-Mental State Examination scores (mean±SD, 15.4 ± 6.1 vs 16.7 ± 5.5). Subjects possessing an APOE ϵ4 allele, however, were more likely to be female (23 vs 11, χ²₁=5.2; P=.02) and to have a family history of dementia (14 vs 4 subjects; Fisher exact test, P=.03).

Figure 1 summarizes the results of the ratings of MTA, periventricular hyperintensities, and white matter hyperintensities for all subjects with dementia according to the absence of an APOE ϵ4 allele (ϵn=33) and the presence of 1 (ϵn=31) or 2 APOE ϵ4 alleles (ϵn=7). No significant associations were noted between APOE ϵ4 status and MTA (mean score: 0 ϵ4=4.5, 1 ϵ4=4.5, and 2 ϵ4=4.3; P=.90), periventricular hyperintensities (0 ϵ4=3.3, 1 ϵ4=3.1, and 2 ϵ4=2.9; P=.83), and white matter hyperintensities (0 ϵ4=5.3, 1 ϵ4=4.9, and 2 ϵ4=4.9; P=.79).

When each dementia group was analyzed separately according to the presence or absence of the APOE ϵ4 allele, no significant differences were noted between MRI ratings and the presence of ϵ4 in subjects with AD, VaD, or DLB, as shown in Table 2.

No association was found between the presence of the APOE ϵ4 allele and MTA and white matter lesions on MRI in subjects with late-onset AD, VaD, or DLB. The frequency of APOE ϵ4 in the different dementia groups was consistent with that in previous studies,^1,3,5,7,8 and in common with other findings, subjects with APOE ϵ4 were more likely to have a family history of dementia in line with other reports.¹ Overall, subjects were well matched for age, age at onset of illness, duration of illness, and the severity of cognitive impairment.

Our hypothesis of an association between the presence of APOE ϵ4 and MTA was not supported by the findings of this study. Furthermore, subjects homozygous for the APOE ϵ4 allele had similar MTA ratings as those with 1 allele, and, therefore, we did not observe an ϵ4 allele dosage effect, as reported by Lehtovirta and colleagues.^9,10

Several factors may explain the discrepancy between the findings of this study and those^9-12 reporting an association between APOE ϵ4 and atrophy of medial temporal lobe structures. Lehtovirta et al^9,10 and Juottonen et al¹² used volumetric measurements to assess atrophy of the hippocampus and the entorhinal cortex, respectively. By comparison, our ratings of MTA may have lacked the sensitivity to detect small changes in volume. Alternatively, subjects in our study were older, and the association between the possession of the APOE ϵ4 allele and subsequent disease may weaken with advancing age, in parallel with the diminishing effect of APOE ϵ4 on the risk of dementia developing. Although subjects in a study by Tanaka et al¹¹ were of an equivalent age or even older, they were fewer (n=34 vs 71), and the authors used a different scanning protocol (acquiring thicker slices from a low-resolution scanner) and method to rate atrophy.

The results of this study are consistent with the findings of Jack et al.²⁸ Examining subjects of a similar age, they found that the APOE ϵ4 genotype and hippocampal atrophy were independently linked with AD. They suggested that the lack of an association between APOE ϵ4 and hippocampal atrophy could be explained by the differential association between hippocampal atrophy and the neurofibrillary tangle burden on the one hand and a putative link between APOE ϵ4 and plaque burden on the other.

Our hypothesis of an association between the possession of APOE ϵ4 and white matter lesions was, again, not supported by our findings. Skoog et al,²⁹ likewise, found no direct link between APOE ϵ4 and white matter lesions on computed tomography of subjects aged 85 years with AD or VaD. Other studies have also failed to find an association between APOE ϵ4 and white matter lesions on MRI in normal elderly subjects³⁰ and those with Dutch hereditary cerebral angiopathy.³¹ Overall, this finding indicates that the APOE ϵ4 allele is unlikely to play a direct role in the pathogenesis of white matter lesions in late-onset dementias, although given the variety of disorders underlying the development of white matter lesions,³² confirmatory clinicopathological studies are necessary. Interestingly, the data are consistent with reports postulating that the APOE ϵ4 allele and cerebrovascular disease, including white matter lesions, are independent but synergistic risk factors for the development of dementia.^29,33

A possible criticism of this and other antemortem studies would be the reliance on standardized clinical, rather than pathological, diagnoses. This might account for some of the homogeneity among the diagnostic groups, given that a proportion of patients with clinical diagnoses of late-onset dementias turn out to have overlapping features on pathological examination.

The role of APOE in the pathogenesis of dementias remains to be determined, and so far, postmortem studies have produced conflicting results. Although several studies have failed to find a specific association between APOE ϵ4 status and the type or amount of disease,^34-37 others have reported isoform-specific associations with impaired neural plasticity,³⁸ neurofibrillary tangle formation,³⁹ amyloid deposition,^40,41 and cholinergic dysfunction.⁴² In contrast, a consistent pattern is emerging from antemortem studies indicating that a range of clinical changes, including the severity and pattern of cognitive impairment, noncognitive symptoms, and neurologic features, are not influenced by APOE ϵ4 genotype.^43-48 Although controversial, taken together, these clinical studies suggest that APOE ϵ4 can modify the risk of acquiring dementia but not subsequent pathological processes.

The APOE ϵ4 allele was not associated with either MTA or white matter lesions in subjects with late-onset AD, DLB, or VaD. Furthermore, in this cross-sectional study, we found no evidence of a specific association between the type of dementia, the APOE ϵ4 genotype, and MRI ratings. Longitudinal studies may provide further insights into the pathogenic role of APOE ϵ4, but the results of this study are consistent with other studies, suggesting that APOE ϵ4 exerts an early influence, modifying the risk of acquiring dementia and age at onset, but appears not to have significant pathological effects thereafter.

Accepted for publication January 15, 1999.

This research was supported by a grant from the Northern and Yorkshire Regional Health Authority, Durham, England.

We thank Messrs P. English and A. Gray and Mss K. Lowry and D. Roberts for facilitating the scanning of patients.

Corresponding author: Robert Barber, MRCPsych, Institute for the Health of the Elderly, Newcastle General Hospital, Newcastle upon Tyne, NE4 6BE, England (e-mail: Robert.Barber@ncl.ac.uk).