Apolipoprotein E Genotype Is Related to Progression of White Matter Lesion Load

Epidemiological studies have shown that the apolipoprotein E (ApoE) genotype is a major genetic risk factor for Alzheimer disease (AD).¹ The underlying mechanisms remain incompletely understood. Both animal and neuropathological studies suggest that the ApoE-ε4 allele enhances brain amyloid beta (Aβ) deposition, which induces amyloid plaques formation.^2–4

At a macroscopic level, the impact of ApoE on brain atrophy in both AD and healthy elderly has been reported.⁵ Similar to the ApoE genotype, results from recent epidemiological studies emphasize the predictive value of white matter lesions (WMLs) load on the risk of AD.⁶ However, relatively little is known about how ApoE and WML load are interrelated.

Reports on the association between WML severity and ApoE are controversial, some cross-sectional studies showing no association^7,8 and others reporting a positive association between the presence of ε4 allele and greater WML volumes^9,10 There has been, to date, no longitudinal study exploring the association between ApoE genotype and WML change in nondemented people.

In a large population-based cohort of elderly, free of dementia at enrollment, who had a cerebral MRI at both study entry and 4-year follow-up, we assessed, cross-sectionally and longitudinally, the relationship between ApoE genotype and WML volumes.

Materials and Methods

Study Population

The Three-City (3C) Study is a multicenter population-based cohort study conducted in 3 French cities (Bordeaux, Dijon, and Montpellier) and designed to estimate the risk of dementia and cognitive impairment attributable to vascular factors as described elsewhere.¹¹ The protocol was approved by the Ethical Committee of the University Hospital of Kremlin-Bicêtre. Between March 1999 and March 2001, 9293 noninstitutionalized individuals aged ≥65 years, selected from the electoral rolls of the 3 cities, agreed to participate in this project. Each participant signed an informed consent and was followed-up every 2 years during 4 years. In Dijon (n=4931), a cerebral MRI examination was proposed to those aged 65 to 80 years old who were enrolled between June 1999 and September 2000 (N=2763). Although 2285 subjects (82.7%) agreed to participate, because of financial limitations, 1779 examinations were performed and interpretable. Compared to the rest of the 3C-Dijon participants, those who had both valid MRI and DNA available (N=1779) were on average younger (72.4 years; SD, 4.1) versus 75.9 years (SD,6.1; P<0.0001), and the proportion of participants carrying ε4ε4 alleles did not differ (1.0% versus 1.2%; P=0.24). After 4-year follow-up, 1319 subjects had a second valid MRI. Subjects without follow-up MRI (N=460, among whom 54 died before the second MRI examination) were older and they did not differ for ApoE genotype distribution (data not shown).

Data Collected

At each study wave, data were collected at the participants’ homes by a trained psychologist during a face-to-face interview using a standardized questionnaire. Information about demographic background, occupation, medical history, drug use, and personal habits was collected.

ApoE Genotyping

ApoE genotyping was performed using the fluorogenic 5′-nuclease assay with TaqMan chemistry. The sequences of the primers and probe oligonucleotides were designed as already described elsewhere.¹²

MRI Scans

MRI acquisition was performed using a 1.5-Tesla Magnetom (Siemens, Erlangen, Germany) as described elsewhere.¹³ Fully automatic image processing software was developed to detect, measure, and localize WML.¹³ Among the various tissue segmentation techniques available, we selected a multispectral approach that uses different MRI sequences. Indeed, because of the heterogeneity of gray matter signals on the T2 images, the white matter mask derived with a 3-class-only segmentation was suboptimal for accurate WML detection. Although such inhomogeneities have negligible impact when calculating the bias field correction, they could be at the origin of false-positive WML on the T2 images. Therefore, T1, T2, and plasma desorption bias-corrected volumes were segmented into 7 classes using the same multispectral algorithm: (1) cerebrospinal fluid; (2) gray matter; (3) caudate nucleus; (4) lenticular nucleus; (5) thalamus; (6) white matter; and (7) WML. The MRI analysis contained 3 major steps: (1) preprocessing, including registration, nonbrain tissue removal, and bias field correction; (2) detection of white matter hyperintensities in T2 images, including removal of false-positives; and (3) postprocessing, including generation of WML probability maps at the individual and sample levels, morphometry, localization, and classification of WML. Morphological parameters were computed for each detected WML, including center of mass coordinates, Euclidian distance to the ventricular system, and principal axes dimension. When its distance to the ventricular system was <10 mm, a WML was labeled as periventricular; otherwise, it was labeled as deep. Periventricular WML and deep WML volumes were calculated by summing the volumes of all the lesions detected in each area.

Using voxel-based morphometry techniques, total intracranial volume (TIV) was computed as the sum of the gray matter, white matter, and cerebrospinal fluid volumes.

Covariates

Selected covariates were among those measured at study entry. Education level was defined in 4 categories ranging from primary certificate level (low) to baccalaureate or university degree (high). Subjects were considered as hypertensive if systolic blood pressure was ≥160 mm Hg or diastolic blood pressure was ≥95 mm Hg or if they were on antihypertensive medication. Hyperglycemia was defined as 6.1 mmol/L ≤ glycemia <7 mmol/L and diabetes as fasting blood glucose ≥7 mmol/L or diabetes medication intake. Subjects were considered as having a history of cardiovascular disease if they reported a history of stroke or myocardial infarcts or arteritis. Tobacco and alcohol consumptions were collected from a questionnaire on diet and lifestyle. Weight and height were measured and body mass index was computed.

Statistical Analysis

Cross-sectional and longitudinal analyses were respectively based on samples of 1779 and 1319 subjects. The relationship between baseline WML volume and ApoE genotype was computed using analysis of covariance adjusting for sex, age, education level, and total intracranial volume (Model 1). We performed analysis of covariance adjusted for age, sex, education level, TIV, and delay between the 2 MRI examinations to assess the longitudinal changes in WML volume over 4-year follow-up and ApoE genotype (Model 1). For both cross-sectional and longitudinal analyses, additional adjustments for vascular factors (hypertension, diabetes, history of cardiovascular disease, alcohol and tobacco consumptions, and body mass index) and baseline WML volume were also performed (Model 2). Linear trend tests were performed to assess a potential ε4 allele dose-dependent association.

We used SAS (Release 9.1; SAS Statistical Institute, Cary, NC) for the analyses.

Results

Principal characteristics of the participants and factors associated with ApoE genotype are displayed in Table 1. Subjects’ mean age was 72.4 years (SD, 4.1) and 60.5% were women. There were no significant differences between number of ε4 alleles for ApoE genotype and demographic status or vascular risk factors frequencies.

Cross-Sectional Findings

After adjusting for age, sex, education level, and TIV, baseline total WML volume was significantly higher in subjects who were homozygous for the ApoE ε4 allele compared with the others (Table 2). This finding was consistent for both periventricular and deep WML volumes. Adjustment for additional potential confounders (Model 2, Table 2) led to similar findings. There was no trend for a ε4 allele dose-dependent association with total WML volume (P=0.51).

Longitudinal Findings

Over 4-year follow-up, mean increase in total WML volume was +1.4 cm³ (SD, 2.8) in the entire sample. It differed significantly by ApoE genotype, subjects who were ε4ε4 having higher increase in WML volume (Table 3). Carriers of ε4ε4 had an increase on average 1.8 cm³ higher in periventricular WML volume and 0.30 cm³ higher in deep WML volume than the rest of the participants. Further adjustments for vascular factors did not modify the findings for total WML or periventricular WML volumes (Model 2, Table 3). For deep WML volumes, after full adjustment, the probability value was 0.40.

In multivariable models (Model 2), test for a ε4 allele dose-dependent association almost reached statistical significance for total WML volume increase (P=0.07) and periventricular WML volume increase (P=0.08).

Stratified analyses by sex, age, or vascular risk factors were not in favor of any significant interaction or even trend (data not shown).

Discussion

In a large population-based sample of elderly, we found from both cross-sectional and longitudinal analyses the significant impact of ApoE genotype on WML load. Indeed, subjects who were ε4 homozygous for ApoE genotype had on average significantly higher WML volumes at baseline than ε4 heterozygous or ε4 allele noncarriers (cross-sectional analyses) but also exhibited higher WML increases over 4-year follow-up compared with the others. A similar pattern of associations was observed whatever the WML localization (periventricular or deep). Controlling for vascular factors (hypertension, diabetes, history of cardiovascular disease, alcohol and tobacco consumptions, and body mass index) or WML load at baseline (for longitudinal analysis) did not affect the overall findings.

Previous reports on the association between ApoE genotype and WML load have provided inconsistent results. Most of them have concluded there was no association, whereas other studies, mainly in nondemented individuals, have reported a positive association.^9,10,14 However, it should be noted that these studies have compared ε4 noncarriers with individuals carrying at least one ε4 allele, which does not allow comparisons with our findings. Only one study, in a sample of 60 patients with AD, also reported that ε4ε4 carriers had significantly more extensive WMLs than ε4ε3 and ε3ε3 carriers.¹⁵

Our cross-sectional data are not in favor of a dose-dependent effect between number of ε4 alleles carried and lesion load, whereas in longitudinal data analysis, a trend close to statistical significance in favor of a dose-dependent effect is observed mainly for increase in total and periventricular WML volumes. One hypothesis could be that with the 3C participants being very healthy, the association between ApoE genotype and higher WML volume is not yet visible among ε4 heterozygous, hence the small trend in longitudinal analysis. We have investigated this hypothesis further by splitting the ε4 heterozygous group according to the median age, but the results did not reveal a difference in WML load between the 2 age groups either in cross-sectional or in longitudinal analysis. Another hypothesis is that carrying one allele other than ε4 might be enough to slow the process leading to WML progression.

ApoE genotype being a well-established risk factor for dementia and WML being also related to dementia risk, it could be argued that incipient AD in ε4ε4 carriers explains our findings. We have explored that assumption in 2 ways; first, we have excluded the 7 subjects who had converted to dementia after baseline examination and second, we have rerun the analyses on the subsample of subjects having Mini Mental State Examination score >24 at baseline. In both scenarios, results were unchanged so we can exclude the hypothesis that our findings are driven by dementia. Some limitations of our study should be considered; our findings are based on a relatively small number of ε4ε4 carriers that require being cautious in interpreting and generalizing the results.

The strengths of our study include the sample size, the population-based setting, the repeated cerebral MRI, and the fully automated quantification of WML volume using a validated and reproducible algorithm.¹⁶

This study is the first showing, both longitudinally and cross-sectionally, the relationship between WML load and ApoE genotype in a population-based sample. Various mechanisms could be considered to explain this association. The association between ε4ε4 and increased WML volumes could reflect cerebral blood flow reduction. Indeed, on the one hand, cerebral blood flow has been found to be decreased in subjects who are homozygous for the ε4 allele,¹⁷ and, on the other hand, there are arguments showing that cerebral blood flow might play a role in the occurrence of WML.¹⁸ It also has been suggested, from a molecular perspective, that the presence of the apolipoprotein ε4 allele may cause an increased vulnerability to slight chronic hypoperfusion of the white matter by reducing the range of mechanical and chemical flexibility of the glial cytoskeleton.¹⁹

However, the most likely scenario linking WML load to the ApoE genotype is through cerebral amyloid angiopathy (CAA). CAA is caused by the deposition of amyloid within the media and adventitia of small- to medium-sized cerebral arteries, which may lead to vessel fragility. It has been shown that the ε4 allele promotes vascular deposition of the ß-amyloid peptide and therefore CAA^4,20 and vascular amyloid deposition may alter white matter perfusion through vascular stenosis dysfunction.²¹ Therefore, one could hypothesize that CAA is an intermediate factor in the association between ApoE and WML severity. It could also be hypothesized that WML is intermediate in the association between ApoE genotype and CAA, although no biological basis for such a pathway is currently available. If the hypothesis that CAA is an intermediate factor in the association between ApoE and WML severity is true, this could have important consequences to elucidate more globally the link between WML and dementia, WML being potentially a marker of CAA disease activity. In a broader sense, WML load could be considered as a surrogate marker of both CAA disease activity and dementia risk in future clinical trials.

Footnotes