Abstract
Objective: To determine whether the effects of APOE promoter polymorphisms on AD are independent of the APOE-ε4 allele.
Background: Recently, the −491 A→T and −219 G→T polymorphisms located in the APOE promoter have been suggested to be risk factors for AD. However, the effects of these polymorphisms have not always been reproduced in case-control studies, possibly because of the strong linkage disequilibrium existing at this locus or the characteristics of the populations studied.
Methods: Data collection was performed from six independent samples (1,732 patients with AD and 1,926 control subjects) genotyped for APOE exon 4 and the two APOE promoter polymorphisms. The risks associated with the APOE polymorphisms for developing AD were estimated using logistic regression procedures and calculation of odds ratios with 95% CI adjusted by age, sex, and collection center. Independence of the APOE promoter polymorphisms was tested by stratification for APOE-ε4 and tertile design was used for age stratification.
Results: The independence of the −491 AA genotype was observed in the whole sample whereas the independence of the −219 TT genotype was observed only in the oldest population.
Conclusion: The −491 and −219 APOE promoter polymorphisms incur risk for AD in addition to risk associated with the APOE-ε4 allele, with age accentuating the effect of the −219 TT genotype. Because these polymorphisms appear to influence apoE levels, these results suggest that APOE expression is an important determinant of AD pathogenesis.
AD is a multifactorial disease implicating interactions between environmental and genetic factors. To date, the only recognized risk factor for the most common forms of AD—defined as sporadic or complex (i.e., without obvious mendelian inheritance)—is the APOE gene on chromosome 19.1 This gene exists as three major alleles in the general population: ε2, ε3, and ε4, resulting from amino substitutions (Arg and Cys) at positions 112 and 158 of the protein. It is now well established that carrying an ε4 allele increases the risk for AD in an allele dose–dependent manner and is associated with an earlier age at onset of the disease.2 Conversely, bearing an ε2 allele confers protection against the disease.3,4⇓ Thus, APOE genotype appears to be an important biologic marker for AD susceptibility accounting for between 45 and 60% of AD genetic variability.5 However, possession of the ε4 allele is not sufficient to develop the disease, because as many as 50% of people who have two copies of ε4 and survive to age 80 years are not cognitively impaired. It has been proposed that other environmental determinants such as serious head injury, smoking, and cholesterol level may modify the APOE-related risk.6-8⇓⇓
Previous reports have suggested that additional factors within the APOE locus itself might also modulate this risk. Genetic studies of markers in the vicinity of the APOE locus have demonstrated that a combination of several polymorphisms increases the APOE-associated risk for AD, compared with the APOE polymorphism alone.4 Furthermore, the observation of a distortion of the allelic expression of the APOE gene suggests that in addition to a qualitative effect of the APOE-ε2/ε3/ε4 polymorphisms on the occurrence of AD, the quantitative expression of these alleles may also be a key determinant.9 This hypothesis was reinforced by the discovery of APOE promoter variants, in particular, −491 A→T and −219 G→T (also called Th1/E47cs), which are capable of modulating the risk for developing AD.10-13⇓⇓⇓ These polymorphisms seem to be functional and modulate APOE expression as described by in vitro and in vivo studies.10,11,13-16⇓⇓⇓⇓⇓ These observations may be particularly relevant to AD pathogenesis, because it has recently been shown that the amount of Aβ deposition in amyloid precursor protein (APP)V717F transgenic mouse brain correlates with the copy number of the APOE gene,17-19⇓⇓ i.e., the level of APOE expression.
However, the contribution of these promoter polymorphisms to AD susceptibility is controversial20 for two reasons. First, because strong linkage disequilibrium exists between the APOE promoter polymorphisms and the ε4 allele, it is difficult to detect an independent effect of these polymorphisms compared with the ε4 allele-related risk.12 Second, even though the association of these polymorphisms with increased risk for AD has been replicated or confirmed in some studies,15,21-24⇓⇓⇓⇓ other studies failed to reproduce this observation.20,24-28⇓⇓⇓⇓⇓ Discrepancies between these studies may result from several biases including ethnic heterogeneity, weak effects not detected in small populations, or specific confounding effects that are not restricted to a particular age group.
To address these concerns, we performed an analysis on patients with AD and control subjects from six participating centers. In this study, we confirm an association between both APOE promoter polymorphisms and risk for AD. The size of the cohort revealed that the association was independent of the ε4 allele-related risk despite strong linkage disequilibrium.
Materials and methods.
In October 1999, we selected from the published literature10-14⇓⇓⇓⇓ those studies that had genotype data available for all APOE coding, −491 A→T and −219 G→T polymorphisms. Unfortunately, most others only tested the −491 A→T polymorphism, thus limiting available data for studying them in combination. We contacted laboratories willing to share their data for a meta-analysis. From an initial population of 1,882 patients with AD and 2,006 control subjects, we obtained a final sample of 1,732 patients with AD and 1,926 control subjects for whom we possessed all relevant clinical and APOE locus genotype data.
The characteristics of the six populations are described in table 1 and detailed as described elsewhere.11,13,24,29-31⇓⇓⇓⇓⇓ These independent white populations were recruited from clinical, community samples. The diagnoses of probable AD were established at all sites according to the Diagnostic and Statistical Manual III–Revised (DSM-III-R) and National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Assoications criteria, whereas the control subjects were defined as subjects without DSM-III-R dementia criteria and with integrity of their cognitive functions. Informed consent had been obtained from each individual or their relatives. For each subject, the following information was sought: a unique identifier, diagnostic status, age at sampling or death, age at disease onset, family history (defined as the presence of at least one first-degree relative with dementia), −491 A→T, −219 G→T, and APOE genotypes.
Characteristics of patients with AD and control subjects
Genetic analysis of APOE polymorphisms.
The APOE genotypes were identified by PCR followed by digestion with HhaI, as slightly modified from the method of Hixson and Vernier.32 The −491 A→T genotyping was performed in all studies by a nested PCR design, as described by Bullido et al.11 A 1,423-bp fragment was amplified containing the transcriptional regulatory region, spanning −1017 to +406 relative to the transcriptional start site with primers 5′-CAAGGTCACACAGCTGGCAAC-3′ and 5-TCCAATCGACGGCTAGCTACC-3′. The 1,423-bp PCR product was therefore used as template using as forward mismatched primer: 5′-TGTTGGCCAGGCTGGTTTTAA-3′, and as reverse primer: 5′-CCTCCTTTCCTGACCC-TGTCC-3′. The PCR products were digested with DraI restriction enzyme.
The −219 G→T genotyping was determined by different methods depending on the center: A restriction digest method was based on the amplification of a 236-bp fragment using mismatched forward primer 5′-AGAATGGAGGAGGGTGCCTG-3′ and reverse primer 5′-ACTCAAGGATCCCAG-ACTTG-3′. The PCR product was subsequently digested with BstNI and separated on a 12% polyacrylamide gel. A second restriction digest method amplified a 220-bp fragment using mismatched forward primer (5′-AGAATGGAGGAGGGTGTCCG-3′) and reverse primer (5′-ACTTGTCCAATTATAGGGCTCC-3′). The PCR product was digested with Hpa II and separated on a 3% agarose gel. A third restriction digest method amplified a 231-bp fragment using mismatch forward primer (5′-AGAATGGAGGAGGGTGCCTG-3′) and reverse primer (5′-TCAAGGATCCCAGACTTGTCG-3′). The PCR product was digested with TaqI and separated on 2.5% agarose gel. All restriction fragment length polymorphisms digest products were visualized by ultraviolet after ethidium bromide staining.
Statistical analysis.
The SAS software release 6.11 was used (SAS Institute, Cary, NC). Univariate analysis was performed using the Pearson χ2 test and the Fisher Exact test when necessary. In the multivariate analysis, we coded the genotypes of each APOE promoter polymorphism as dummy variables according to the hypothesis for a recessive model, i.e., TT vs GT+GG genotype for the −219 G→T polymorphism and AA vs AT+TT genotype for the −491 A→T polymorphism. Homogeneity of odds ratios (OR) between populations was tested using Breslow-Day computation.33
Logistic regression analysis was performed to control for age, sex, and center. Ages at last examination among patients with AD and control subjects were assigned to the age variable. The percent of the maximum possible value of linkage disequilibrium between the APOE coding and promoter polymorphisms was computed as outlined by Thompson et al.34 Because the APOE coding and APOE promoter polymorphisms were in strong linkage disequilibrium, the independence of the APOE promoter polymorphisms was tested by stratification for APOE-ε4. Extended haplotype distribution of all the markers was estimated on collapsed data using the myriad haplotype algorithm described by McLean and Morton,35 implemented in a computer program by Cox et al.36
Interactions between age, sex, −491 A→T, −219 G→T, and APOE polymorphisms were tested by logistic regression. Possible effects of age were sought by stratification using tertile design (age ≤ 71 years, 71 years to ≤ 81 years, and > 81 years), which were chosen in order to define the age- and size-matched groups.
Results.
None of the polymorphism distributions deviated from Hardy–Weinberg equilibrium in control subjects.
The APOE coding polymorphism.
The distribution of the APOE-ε2/ε3/ε4 polymorphism in both control and AD samples is shown in table 2. As expected in the whole sample, the ε4 allele was a major risk factor for AD (OR = 4.6; 95% CI = 4.0 to 5.3), whereas the ε2 allele exhibited a protective effect (OR = 0.5; 95% CI = 0.4 to 0.6). However, when the ε4 allele-related risk was compared between the centers, the range of OR was heterogeneous (Breslow-Day, p = 0.007), mainly due to the data from the Spanish study yielding an OR of 11.6, the range of other OR being narrow (3.4 to 5.3) (Breslow-Day, p < 0.17).
APOE allele frequency in patients with AD and control subjects by center and odds ratios for developing AD according to the presence of the ε4 allele adjusted for sex, age, and center
As previously described, we also observed an effect of age on APOE-ε4–related risk4-5,35⇓⇓ (interaction, p = 0.05). When dividing the population in three equivalent classes (age ≤ 71 years, 71 years to ≤ 81 years, and >81 years), the stronger effect was observed in the middle age class, but in subjects > 81 years old, an important decrease of the APOE-ε4 effect was detected compared with younger age groups, in agreement with the data in the literature5,37⇓ (OR = 5.0, 95% CI = 4.0 to 6.4 for age ≤ 71 years; OR = 5.6, 95% CI, 4.2 to 7.4 for age from 71 to 81 years; and OR = 3.2, 95% CI = 2.4 to 4.2 for age > 81 years).
APOE promoter polymorphisms.
The distributions of the APOE promoter polymorphisms are shown in tables 3 and 4⇓ and compared between the different centers. The −219 TT genotype–related risk (OR = 1.3 to 3.3; see table 4) was homogeneous across all six studies (Breslow-Day p < 0.21). However, the range of the −491 AA genotype–related risk was greater (OR = 0.8 to 3.4; see table 3) and was heterogeneous (Breslow-Day p < 0.001). Multiple Breslow-Day tests for the −491 A→T polymorphism using several sample combinations did not allow us to define the center mainly responsible for this difference, confirming the strong heterogeneity of the effect of the −491 AA genotype on risk for AD between centers. To minimize the effect of this variability, we systematically adjusted OR for all logistic regression models by centers, age, and sex. The previously described linkage disequilibrium for the ε4 allele and promoter polymorphisms was verified in the current study.12 In the whole sample, the coefficients of linkage disequilibrium between the ε4 and the −491 A alleles was of 72.8% (range, 49.9 to 100%; p < 1.10-6) and between ε4 and 219 T alleles it was of 50.2% (range, 23 to 72%; p < 1.10-6). Therefore, we estimated the effect of each polymorphism on AD risk in the groups of carriers and noncarriers of the ε4 alleles. Finally, we also examined whether the impact the APOE promoter polymorphisms was closely related to the variation of the ε4 allele–related risk according to age, in order to clarify their independence from the ε4 allele (table 5).
−491 A→T allele and genotype distributions in patients with AD and control subjects by center and odds ratios for developing AD according to the −491 AA genotype adjusted for sex, age, and center
−219 G→T allele and genotype distributions in patients with AD and control subjects by center and odds ratios for developing AD according to the −219 TT genotype adjusted for sex, age, and center
Odds ratios for developing AD according to the −491 A→T, −219 G→T, and APOE genotype ≤81 and >81 years old adjusted for sex, age, and center
Impact of the −491 A→T polymorphism.
The −491 AA genotype was associated with an increased risk for developing AD (OR = 1.7, 95% CI = 1.5 to 1.9, p < 0.0001; see table 3). This effect appeared to be independent of the ε4 allele, because this association is preserved in both bearers and nonbearers of the ε4 (OR = 1.3, 95% CI = 1.1 to 1.6, p < 0.003 in non-ε4 bearers and OR = 1.4 = 95% CI = 1.0 to 1.8, p < 0.04 in the ε4 bearers). Sustaining the increased risk associated with the −491 A allele independently of the ε4 allele, we observed an increased risk associated with the −491A-ε3 haplotype compared with the −491 T-ε3 haplotype (OR = 1.3, 95% CI = 1.1 to 1.5, p < 0.003).
No interaction between the −491 A→T polymorphism and sex (p = 0.29) was detected.
We divided the sample into three equivalent age classes (age ≤ 71 years, 71 years to ≤ 81 years, and > 81 years). The effect of the −491 AA genotype was similar in the two younger age classes and we merged these data. Before 81.1 years of age, the −491 AA genotype was a risk factor for AD (OR = 1.8, 95% CI, 1.5 to 2.1, p < 0.0001), this association being independent of the APOE genotype (OR = 1.4; 95% CI = 1.1 to 1.7, p < 0.009 in non-ε4 bearers and OR = 1.6, 95% CI = 1.1 to 2.1, p < 0.007, in ε4 bearers; see table 5). The effect of the −491 A-ε3 haplotype was also visible in the sample before ≤ 81 years old (OR = 1.3; 95% CI = 1.1 to 1.4, p < 0.003).
Impact of the −219 G→T polymorphism.
The −219 TT genotype was associated with an increased risk for developing AD (OR = 1.6, 95% CI = 1.3 to 1.8, p < 0.0001; see table 4). We did not detect an independent effect of the −219 TT genotype on risk for AD, when stratified for APOE genotype (OR = 1.2, 95 CI% = 1.0 to 1.5, p < 0.07 in ε4 nonbearers; OR = 1.2, 95 CI% = 1.0 to 1.5, p < 0.16 in ε4 bearers). No interaction between the −219 G→T polymorphism and sex was detected (p < 0.43).
When stratifying for age using the same three age classes (age ≤ 71 years, 71 years to ≤ 81 years, and > 81 years), the −219 TT genotype was associated with an increased risk for developing AD in the > 81 years class (OR = 2.0, 95 CI% = 1.5 to 2.8, p < 0.0001) this association being independent of the ε4 allele (see table 5). When stratifying for ε4 status, the effect of the −219 TT genotype appeared independent of the ε4 allele (OR = 1.6, 95 CI% = 1.0 to 2.3, p < 0.03 in ε4 nonbearers; OR = 1.9, 95 CI% = 1.2 to 3.2, p < 0.01 in ε4 bearers). Supporting this independent effect, the association of the −219 TT genotype was observed despite the approximately twofold decrease of the ε4 allele–related risk in the elderly group compared with the youngest group. These data suggested that the risk associated with the −219 TT genotype may be obscured by the strong effect of the ε4 allele–related risk in the youngest group. Supporting the effect of age, an increased risk was observed for the carriers of the −219 T-ε4 compared with the −219 G -ε4 haplotype in the sample > 81 years old (OR = 2.1, 95% CI = 1.7 to 2.6, p < 10−7).
Discussion.
The purpose of this study was to collect data from a large number of white patients with AD and control subjects, all genotyped for the APOE, −219, and −491 polymorphisms to assess the variability in the strength of the association between the −491 and −219 APOE promoter polymorphisms and AD and the degree of independence of their effects compared with the ε4 allele–related risk. Indeed, the degree of linkage disequilibrium of these two polymorphisms with the ε4 allele is important and does not allow simple evaluation of their respective contributions even by stratifying the population. Furthermore, the size of the sample studied is a critical parameter; a recent analysis of both these promoter polymorphisms did not confirm their independent association compared with the ε4 allele,38 but trends toward an excess of the −219 TT genotype (27 vs 23%) and of the −491 AA genotype (75 vs 67%) were observed in patients compared with control subjects. Our analysis of 3,658 individuals demonstrates a moderate but significant effect of the −491 A→T polymorphism on the risk for AD, independent of the ε4 allele. Similarly, we detect an effect of the −219 G→T polymorphism independent of the ε4 allele, but this was restricted to the older age group. Because one-third of our oldest population came from the Finnish population, in which the −219 G→T polymorphism is associated with the highest effect, this may suggest that our observation was due only to this population. However, when the Finnish population was excluded from the analysis, similar effects of the −219 TT genotype were observed after APOE stratification in the oldest group (OR = 1.4, 95% CI = 0.9 to 2.2, p < 0.17 in non-ε4 bearers and OR = 1.7, 95% CI = 1.0 to 3.1, p < 0.07 in ε4 bearers). Because the lack of significance may result from the reduction in size of the population (almost 300 patients with AD and control cases), this observation indicated that the impact of the −219 G→T polymorphism in the oldest subjects may be independent of the effect of this polymorphism in the Finnish population. Supporting this observation, an increased risk remained in carriers of the −219T-ε4 compared with the −219 G-ε4 haplotype in the sample > 81 years of age when the Finnish population was excluded (OR = 1.8, 95% CI = 1.4 to 2.4, p < 9.10-7).
Furthermore, our observations were supported by a recent study that reported an independent effect of the −219 TT genotype on the risk for developing dementia using a population-based cohort of 648 subjects aged ≥ 85 years (Leiden 85-plus study).39
It is of note that the clinical diagnosis of AD in the oldest group (> 81 years) may be problematic because these subjects often have other simultaneous neuropathologic processes (e.g., vascular) contributing to dementia. This may result in incorrect clinical definition of both AD and control subjects.40 This problem may be illustrated in the Vantaa Finnish population, which consists of individuals aged ≥ 85 years of age. In the subgroup we used for the meta-analysis, there was a significant effect of the −219 genotype on the risk for clinically diagnosed AD (see table 4). When neuropathologic criteria were used to define AD and control subjects, in another subgroup, the overall association with the −219 genotype was not found.41 However, in haplotype analysis of neuropathologically verified ε3/ε3 subjects, the −219 polymorphism significantly modulated the risk for neuropathologic AD and β-amyloid deposition, consistent with the view that the haplotype background of APOE influences AD risk independent of apoE isoform.
The risks imposed by these specific genotypes may be difficult to assess without a more thorough analysis of the larger APOE haplotypes. A recent study examining APOE variation at the sequence haplotype level observed 22 biallelic polymorphisms and 31 haplotypes within a 5.5-kb stretch of DNA, including the four exons of the gene and some upstream sequence.42 Seven of the 22 variants occurred within the sequence 5′ of the transcriptional start site and thus have the potential to alter levels of APOE transcription. Because the observed haplotypes showed important interpopulation differences in frequency, it might be expected that the relative risk for AD associated with these promoter polymorphisms would vary substantially, as it does for the APOE-ε4 allele. Among the different polymorphisms analyzed, the −219 site was one that defined major subtypes of both ε3 and ε4 haplotypes in multiple populations, and the −491 site was the most homoplastic site, which is the inferred occurrence of multiple independent evolutionary events giving rise to the same allelic state at a variable site.42 Indeed in our study, the magnitude of the effect of the −491 A→T polymorphism in all age groups combined is heterogeneous between the different centers. Interestingly, the frequency of the −491 A→T polymorphism in control subjects varies not only in the six populations reported here but also in recent publications.20,24⇓ This variability may also result in part from differences in control ascertainment or genotyping errors.
Despite Breslow-Day analysis results, bias may appear from the heterogeneity of our six populations. In order to restrict these potential problems we systematically adjusted for age, sex, and centers in our logistic model but also used an age tertile design to homogenate our sample. For instance, the tertile distribution of the sample as a function of sex was homogeneous (data not shown). Nonetheless, it is of note that we were limited because only a few studies have performed the genotyping on both promoter polymorphisms, whereas a number of reports examine the potential involvement of the −21925 or the −49126-28⇓⇓ promoter polymorphisms.
Although both promoter polymorphisms exert effects that are independent of ε4 status, it is not clear how they influence the AD process. Evidence has been put forward in support of an allele-specific effect of the polymorphisms on APOE expression. In vitro, nuclear protein binding as well as reporter gene expression is modified by the −491 A→T and −219 G→T polymorphism alleles.10,11,14⇓⇓ In vivo, allelic mRNA expression in AD brains, as well as apoE protein concentration in plasma, is correlated with these polymorphisms, although it is recognized that we do not know whether the same relationships exist to modify the amount of apoE protein in brain tissue.15,16⇓ A study of APOE expression in the brain and its modulation by promoter polymorphisms may lead to a better understanding of the role of apoE in the AD process. Recently, we have reported that polymorphisms in the promoter of APOE are associated with the amount of amyloid peptide (Aβ) in the brains of patients with AD, an effect that is independent of APOE genotype.43 Interestingly, the effect of the −219 TT genotype on Aβ load is stronger in the oldest age group, consistent with the findings of this analysis. Furthermore, in a large cohort of control brains, the number of senile plaques in two brain regions (CA1, subiculum), was significantly greater in individuals homozygous for the T allele of the −219 G→T, although it is not clear whether this effect was independent of the ε4 allele.44 Interestingly, recent studies in transgenic mice expressing APPV717F and human APOE genes17-19⇓⇓ show that the amount of human APOE transgene product in the brain, particularly in ε4 mice, correlates with Aβ deposition as the mice get older. Collectively, these data demonstrate the importance of the correlation between the APOE expression and Aβ peptide deposition in humans.
Acknowledgments
Supported by INSERM, Institut Pasteur de Lille, the Conseil Régional du Nord-Pas de Calais (“Funds for Neurodegenerative Research” 9926002), The South Birmingham Mental Health Trust, Fundación Ramon Areces, The Academy of Finland (project number 48173), and Helsinki University Central Hospital (EVO Trust Funding), The Finnish Cultural Foundation, the NIH/NIA (grants #AG05681, AG03991 to J.C.M. 38 A.M.G., AG16208 to A.M.G. 38 J.H.), the Mayo Foundation and an investigator-initiated grant and Zenith award from the Alzheimer’s Association (A.M.G., J.H.), with fellowships from Marie Curie (J.C.L.), Institut Pasteur-Conseil Régional Nord-Pas de Calais (L.A.G.) and poste-vert INSERM (M.E.T.)
Acknowledgment
The authors thank John Coates, Birmingham University, for technical assistance, and David Neary and Julie Snowden, Department of Neurology at Manchester Royal Infirmary, for sample collection.
- Received December 24, 2001.
- Accepted in final form March 14, 2002.
References
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.