The ε2 and ε4 Alleles of the Apolipoprotein Gene Are Associated with Age-Related Macular Degeneration

Abstract

purpose. To date, of all the genes studied in relation to age-related macular degeneration (AMD), the alleles of the apolipoprotein (apoE) gene have been the most consistently associated with disease. However, not all apoE studies have found an association, and among these the associations differ. The current study was conducted to investigate further the association of this gene in AMD.

methods. Three hundred twenty-two unrelated individuals with diagnosed AMD and 123 unrelated but ethnically matched control subjects were analyzed. All subjects completed a standard questionnaire and were given a fundus examination. A blood sample was collected for DNA extraction. The common allelic variants of apoE were screened through the use of polymerase chain reaction (PCR) and restriction enzyme digestion followed by statistical analysis.

results. Individuals with the ε3 ε4 genotype of apoE had an approximate halving of disease risk for late (end-stage) AMD (odds ratio [OR] 0.58, 95% confidence interval [CI] 0.34–0.98) relative to the ε3 ε3 genotype at age of ascertainment. Stratification of late AMD into atrophic and neovascular disease revealed that the greatest protective effect for the ε3 ε4 genotype was in individuals with atrophic disease (OR 0.35, 95% CI 0.13–0.92). Men with the ε3 ε4 genotype also showed almost a threefold reduction in risk of disease in late AMD (OR 0.36, 95% CI 0.16–0.82). However, individuals with late AMD and the ε2 ε3 genotype had a significantly earlier mean age of diagnosis of disease (3.4 years, P = 0.015) compared with those with the ε3 ε3 genotype, and this was most evident in women (3.9 years, P = 0.011) and in individuals with neovascular disease (4.7 years, P = 0.003).

conclusions. The alleles of apoE appear to have a role in the etiology of AMD, with the ε4 allele being protective, or at the very least, delaying the age of diagnosis of disease, whereas the ε2 allele appears to have a modifier effect by bringing forward the mean age of disease diagnosis.

Age-related macular degeneration (AMD) is the leading cause of blindness in the developed world. ^{1 2} It occurs in less than 1% of people younger than 65, but increases dramatically with age so that late AMD affects more than 25% in the 90-plus age group. ² It is a progressive, late-onset disease in which central vision is lost, either through atrophy (dry AMD) or by the growth of new blood vessels, choroidal neovascular membranes (wet AMD). These changes are thought to occur as a result of the accumulation of lipid rich debris, in Bruch’s membrane, the semipermeable membrane that separates the retina from its blood supply in the choroid. ³  

The etiology of AMD is poorly understood, but it is most likely a complex disease in which environmental risk factors impact on an individual’s genetic background. ^{4 5} Potential risk factors have been identified from epidemiologic studies. Of particular interest to this study are the risk factors associated with cardiovascular disease (CVD). These include smoking, which is strongly associated with a two- to fourfold increased risk of disease ⁶ ; a high-fat diet, which is associated especially with neovascular disease ^{7 8 9} ; high serum cholesterol; elevated blood pressure; atherosclerosis; and a history of CVD. ^{10 11 12 13} The association of one or more of these risk factors in most epidemiologic studies on AMD raised the possibility that some pathogenic mechanisms may be shared between CVD and AMD. 

The common allelic variants (ε2, ε3, ε4) of the apolipoprotein E (apoE) gene have been implicated in several diseases including CVD and ¹⁴ Alzheimer’s disease, ¹⁵ as well as AMD. Indeed, to date it has been the most consistently associated gene with AMD. ApoE is involved in cholesterol transport throughout the body ¹⁶ and facilitates the binding of lipoproteins to low-density lipoprotein (LDL) receptors, thereby maintaining the lipoprotein cholesterol available to the cell. ¹⁶ ApoE is expressed in retinal tissue, in particular the retinal pigment epithelium (RPE) and Bruch’s membrane. ¹⁷ Animal studies have also shown that induction of a hypercholesterolemic state after either a high-fat diet or apoE deficiency resulted in a significant thickening of Bruch’s membrane ¹⁸ as well as accumulation of retinal debris. ¹⁹  

Several case–control studies have now been undertaken to examine the association of the common allelic variants (ε2, ε3, ε4) of the apoE gene with AMD. These studies have concentrated on white individuals from mainly European backgrounds, ^{20 21 22 23 24 25} although one study has been undertaken in ethnic Chinese. ²⁶ Only three of the seven previous studies have shown a significant protective effect for the apoE ε4 allele with late (end-stage) AMD (odds ratio [OR] 0.16, 95% confidence interval [CI] 0.05–0.43 ²⁰ ; OR 0.43, 95% CI 0.21–0.88 ²¹ ; and a pooled study OR 0.54, 95% CI 0.41–0.70, ²⁴ ), whereas the protective nature of this allele was more modest in two other studies (OR 0.88, 95% CI 0.58–1.35 ²² ; and OR 0.4, 95% CI 0.1–1.2 ²³ ). In individuals of Chinese ethnicity, there was little evidence to suggest an involvement of the ε4 allele of this gene in AMD (OR 0.83, 95% CI 0.23–2.53). ²⁶ In the studies in which association was identified with the ε4 allele, the findings were inconsistent, with protection being found when both exudative or atrophic disease were pooled, ^{21 24} or when only exudative disease was analyzed, ²⁰ or in individuals with familial disease who were younger than 70 years. ²² In addition, these studies had a small number of individuals carrying the ε4 allele, and that limited the investigators’ ability to perform disease subanalysis. Three of the studies also suggested that the ε2 allele may increase the risk of AMD, although none of these reached statistical significance. ^{21 23 24}  

As apoE is the one gene that has so far shown the most consistent replication in genetic association studies with AMD and, given the inconsistencies in the previous studies, it was imperative to investigate this association further. We performed a study of power sufficient to allow subclassification of AMD into different subsets to explore further any associations of the alleles of this gene and AMD. 

Methods

Subjects

All individuals were recruited as part of our AMD inheritance study (AMDIS) and were Anglo-Celtic in origin, with all four of their grandparents born in either the United Kingdom or Ireland. All study participants were examined by a medical retinal ophthalmologist (RHG) and fundus photographs were taken. All participants were given our standard risk factor questionnaire. Of particular interest for this study were the questions regarding the age of diagnosis and the age of ascertainment. The age of diagnosis referred to when the patient was told for the first time that he or she had any clinical sign of macular degeneration. The age of ascertainment was when blood was collected from individuals for DNA analysis. There may or may not have been accompanying symptoms. Lesions were classified based on clinical examination and fundus photographs. In some cases, family members were examined in their own homes with a portable fundus camera and indirect ophthalmoscopy. Information on age, sex, ethnicity, and eye disease was available for analysis in this study. Written informed consent was obtained from all individuals, and ethics approval for the project was provided by the Human Research and Ethics Committee of the Royal Victorian Eye and Ear Hospital, Melbourne (RVEEH). The study was conducted in accordance with the Declaration of Helsinki and the National Health and Medical Research Council of Australia’s statement on ethical conduct in research involving humans, revised in 1999. 

All patients were identified through either outpatient clinics at the RVEEH, private ophthalmology practices in Melbourne, or low-vision clinics. Late AMD was classified as either neovascular (wet) or atrophic (dry). Individuals who presented with both neovascular and atrophic forms of AMD were classified as having neovascular AMD. Early AMD was defined as the presence of soft drusen (>63 μm), with or without regions of hyperpigmentation. People with only hard drusen or with only pigmentary changes were not classified as having AMD, as it was imperative that classifications of individuals as positive for disease be beyond doubt. Cases were classified as familial AMD if two or more first-degree relatives were documented as having disease. Sporadic disease was classified as an individual having no known family history of AMD. 

Control subjects were collected from the same community as part of the large population-based epidemiologic eye study, the Melbourne Visual Impairment Project (VIP), ²⁷ or through aged-care nursing homes. Individuals were selected as control subjects based on the presence of a normal fundus (<10 hard drusen classified as drusen <63 μm in size). 

DNA Analysis

Genomic DNA was isolated from venous blood leukocytes using a standard phenol-chloroform extraction procedure. ²⁸ Polymerase chain reaction (PCR) amplification was undertaken with the primers AF(5′GCCTCCCCACTGTGCGA3′), AR(5′GGCCGAGCATGGCCTG3′) to produce a 243-bp product encompassing amino acid position 112 (site A) and BF(5′ACCGAGGAGCTGCGGG3′), BR(5′CTCGCGGATGGCGCTGA3′), to produce a 150-bp product encompassing amino acid position 158 (site B). PCR conditions used in amplification were 94°C for 3 minutes, followed by 40 cycles of 94°C for 30 seconds, 61°C for 30 seconds, and 72°C for 30 seconds and a final extension cycle of 72°C for 10 minutes. Each of sites A and B contained a single nucleotide polymorphism (SNP), either being a C or T at site A or a C or T at site B, that resulted in either a cysteine (cys) or arginine (arg). Digestion of each PCR product was undertaken with the restriction enzymes AflIII for site A and HaeII for site B. Digested products were size separated on agarose gels. The combination of amino acids at sites A and B (cys/cys, cys/arg, or arg/arg) allowed the common allelic variants of either ε2, ε3, or ε4 of the apoE gene to be scored. Dideoxynucleotide sequencing was undertaken on a representative sample to confirm the results of each restriction digestion profile. 

Statistical Analysis

The Fisher exact test was used to calculate the probabilities for the tests of whether there was any (unadjusted) association between various kinds of AMD and different genotypes or alleles. Odds ratios adjusted for age and gender were obtained from logistic regressions, and t-tests were used to calculate the probabilities for the tests of whether there was any difference in the mean ages of cases and control subjects or the mean ages at diagnosis of the two different genotypes (SAS, ver. 8.02; SAS, Cary, NC). Analysis of AMD was undertaken by subclassification into early, late, neovascular, and atrophic, as well as familial and sporadic disease. 

Results

Three hundred twenty-two unrelated individuals aged more than 50 years a diagnosis of AMD were analyzed in this study 1 . Seventy individuals (22%) had early AMD and 252 (78%) had late AMD 1 . Of those individuals with late AMD, 200 (79%) had neovascular disease, and 52 (21%) had atrophic disease 1 . Fifty-three individuals (21%) had late AMD that was familial in origin and 199 (79%) had sporadic AMD. An additional 123 unrelated but ethnically matched control subjects were also recruited 1 from Melbourne. There was no statistical difference between the ages of late-onset cases and control subjects and no evidence of departure from Hardy Weinberg equilibrium in cases (χ² = 2.29, df = 3, P = 0.515) or control subjects (χ² = 5.42, df = 3, P = 0.143). 

Because of the large sample size, we were able to classify AMD into a number of subgroups including early, late, neovascular, and atrophic, as well as familial and sporadic disease. We identified a significantly lower frequency of the apoE ε4 allele in late AMD (P = 0.037, Fisher exact test) than in control subjects. A significant result was also found when late disease was subgrouped into either neovascular or atrophic AMD (P = 0.091 and 0.018, respectively, for the apoE ε4 allele; Fisher exact test). Of note, was the total absence of the ε4 allele (0/14 individuals) in individuals who had atrophic disease with a family history of AMD 2 . We did not detect any significant reduction in ε4 allele frequency in individuals with early AMD when compared with control subjects (data not shown). 

In the cohort of late AMD cases and control subjects, the ε3 ε4 genotype constituted approximately 80% of all ε4 alleles of apoE 2 . We identified an approximate halving of disease risk in individuals with late AMD when the ε3 ε4 genotype (OR 0.58, 95% CI 0.34–0.98) was compared with individuals with the ε3 ε3 ancestral genotype adjusted for both age and gender 3 . Classification of late disease into either neovascular or atrophic disease indicated that this protective effect was significant only in the atrophic disease group, showing an approximate threefold reduction in disease risk (OR 0.35, 95% CI 0.13–0.92; 3 ) when the ε3 ε4 genotype was compared with the ε3 ε3 genotype. 

Classification of late AMD by gender indicated a significantly lower frequency of the ε4 allele in men with late AMD (10.3%) than in control subjects (23.6%; P = 0.010, Fisher exact test). We identified an almost threefold reduction in late AMD risk for men with the ε3 ε4 genotype (OR 0.36, 95% CI 0.16–0.82) in comparison with the ε3 ε3 genotype. However, no significant association was found in women, although a protective trend was evident (OR 73, 95% CI 0.37–1.43) for the ε3 ε4 genotype. 

When the ε2 allele was considered in this study, there was a slight but not significant increase in the frequency of this allele in late AMD (9.9%) compared with control subjects (6.9%). There was no significant increased risk of disease with the ε2 ε3 genotype in late AMD (OR 1.69, 95% CI 0.79–3.61), although there was a trend for an increased risk of late AMD. In addition a threefold increased risk of atrophic disease was found when comparing those with familial disease with control subjects. However, this finding was of borderline significance when compared with those with the ε3 ε3 genotype (OR 3.58, 95% CI 0.98–13.1; 3 ). 

We investigated whether the ε2 ε3 and the ε3 ε4 genotypes had any effect on the mean age of diagnosis of disease compared with the reference ε3 ε3 genotype. In individuals with the ε2 ε3 genotype, a significantly younger age of diagnosis was detected (3.4 years, P = 0.015) in late cases 4 . This finding was most evident in women (3.9 years, P = 0.011) and in individuals with neovascular disease (4.7 years, P = 0.003). When we considered the ε3 ε4 genotype, we did not observe a significant difference in age of diagnosis for late cases. 

Discussion

Association studies provide a powerful approach to the analysis and identification of genetic variants involved in disease susceptibility. However, one of the main criticisms of these studies has been that they are not always reproducible. In the case of the common allelic variants of apoE and AMD, a significant protective effect for the ε4 allele has been reported in two population-based studies, ^{20 21} as well as in a pooled case–control study, ²⁴ yet not in other studies. In our study we were able to replicate the finding of a significant protective effect for the ε4 allele of apoE and late AMD. We were also able to subclassify late AMD into either atrophic or neovascular disease and show a significant protective effect for the ε4 allele in the atrophic group. 

In the two previous studies that found a similar protective effect for the ε4 allele and late AMD, either all AMD cases were analyzed together ²¹ or only neovascular disease was considered. ²⁰ The pooled AMD study ²⁴ classified disease based on disease type and reported a protective effect in both atrophic (OR = 0.40) and neovascular disease (OR = 0.61), similar to our findings of OR = 0.35 and OR = 0.65, respectively. When we subclassified individuals into those who had atrophic disease as well as disease that was familial in origin, we noted a total absence of the ε4 allele of apoE in these individuals—an observation that has not been reported before. Although the number of individuals in this group was small (n = 14), the finding adds further support to the notion that the ε4 allele provides a greater protective effect in individuals with atrophic disease than in those with neovascular disease. 

The protective effect that we observed for the apoE ε4 allele in familial index cases has also been reported in familial individuals under the age of 70 years. ²² However, in another study, no such protective effect of this allele was reported in familial cases. ²⁵ The apoE ε4 allele frequency was 8.7% in our atrophic AMD compared with frequencies of 8.9% ²² and 12.4% ²⁵ in the previous studies, whereas in our control subjects, it was 17.9% compared with 12.4% ²⁵ and 14.6% ²² in these studies. The variation in ε4 allele frequency in the three studies presumably reflects ethnic differences between the populations. It is known that the ε4 allele of apoE exhibits a population gradient across Western Europe, with the frequency of this allele varying from 20% in Finland and Sweden to 8% in Italy. ²⁹ Matching of cases and control subjects based on ethnic origin is therefore critical in any association study. In the present study, we sought to minimize problems of ethnic mismatching by including only those cases and control subjects where individuals could report the ancestry of all four of their grandparents back to the United Kingdom or Ireland. 

We noted a significantly lower frequency of the apoE ε4 allele in men with late AMD compared with control subjects. This was estimated to lead to an almost threefold reduction in disease risk in men with the ε3 ε4 genotype compared with men with the ε3 ε3 genotype. This finding differed to the previous pooled study of AMD, ²⁴ in which no significant association of the apoE ε4 allele was detected with gender. The ε4 allele of apoE has been reported to be associated with an increased risk of coronary heart disease in men but not in women. ³⁰ Further investigation of this observation would be useful in determining how the ε4 allele may influence risk of AMD, depending on gender. 

We confirmed the consistent trend of an increased risk of end-stage AMD when the ε2 allele of apoE was present. We were also able to demonstrate the novel finding of an earlier age of diagnosis of disease of approximately 3.4 years in individuals in whom an ε2 ε3 genotype was present. This finding has not been documented before in AMD and further study are needed to replicate it. 

Association studies designed to dissect the components of a complex disease such as AMD are difficult, and we took several steps to improve the design and minimize potential pitfalls. These included the selection of ethnically matched cases and control subjects; the examination of all individuals in the study, including control subjects; and the identification of a sample size large enough to provide the power to undertake the association study. 

In conclusion, we were able to demonstrate a protective association for the ε4 allele in late AMD and an apparent earlier age of diagnosis of disease in individuals with an ε2 ε3 genotype compared with ε3 ε3. We were also able to show that the protective effect of ε4 was greatest in men and in individuals with atrophic disease. However, it still remains to be determined how the alleles of the apoE gene influence the outcome of AMD and why the association of alleles with disease is the opposite of that in CVD and Alzheimer’s disease, in which the ε4 allele is a risk factor for disease and the ε2 allele is protective. Once the pathogenic mechanisms of AMD are better understood, it is hoped that these enigmas can be resolved. 

 Supported by the Ophthalmic Research Institute of Australia; the Helen Macpherson Smith Trust; the L. E. W. Carty Charitable Fund; the Eye, Ear, Nose and Throat Research Institute, Perpetual Trustees; the Royal Victorian Institute for the Blind; the Macular Vision Loss Support Society of Australia; and the Royal Victorian Eye and Ear Hospital Research Committee.

 Submitted for publication October 10, 2003; revised December 2, 2003, January 20, 2004; accepted January 22, 2004.

 Disclosure: P.N. Baird, None; E. Guida, None; D.T. Chu, None; H.T.V. Vu, None; R.H. Guymer, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Table 1.

 

 

Characteristics of Study Population: Age and Gender Distributions of AMD and Control Subjects

Table 1.

 

Characteristics of Study Population: Age and Gender Distributions of AMD and Control Subjects

	Sporadic AMD					Familial AMD					Total Late AMD (n = 252)	Control Subjects (n = 123)
		Early (n = 56)	Neovascular (n = 161)	Atrophic (n = 38)	Early (n = 14)	Neovascular (n = 39)	Atrophic (n = 14)
Male	15 (26.8)	39 (24.2)	14 (36.8)	3 (21.4)	11 (28.2)	4 (28.6)	68 (27.0)	53 (43.1)
Female	41 (73.2)	122 (75.8)	24 (63.2)	11 (78.6)	28 (71.8)	10 (71.4)	184 (73.0)	70 (56.9)
Mean age (y) at ascertainment of male and females ± (SD)	73.0 (6.80)	77.5 (7.08)	76.7 (6.35)	70.9 (8.12)	78.5 (6.30)	78.0 (6.77)	77.6 (6.82)	76.6 (4.86)
P	0.001	0.20	0.91	0.02	0.086	0.47	0.11

 Data are the number of individual cases with percentage of total sample in parentheses. Probabilities were calculated for each group versus controls using t-tests.

Table 2.

 

 

ApoE Genotype and Allele Frequency Distribution for Late AMD and Control Subjects

Table 2.

 

ApoE Genotype and Allele Frequency Distribution for Late AMD and Control Subjects

	Atrophic AMD			All Atrophic AMD	Neovascular AMD			All Neovascular AMD	Total Late AMD	Control Subjects
		Sporadic	Familial			Sporadic	Familial
ApoE genotype
ε2ε2	1 (2.6)	0 (0)	1 (1.9)	1 (0.6)	0 (0)	1 (0.5)	2 (0.8)	2 (1.6)
ε2ε3	6 (15.8)	5 (35.7)	11 (21.2)	22 (13.7)	4 (10.3)	26 (13.0)	37 (14.7)	10 (8.1)
ε2ε4	1 (2.6)	0 (0)	1 (1.9)	6 (3.7)	2 (5.1)	8 (4.0)	9 (3.6)	3 (2.4)
ε3ε3	23 (60.5)	9 (64.3)	32 (61.5)	99 (61.5)	26 (66.7)	125 (62.5)	157 (62.3)	69 (56.1)
ε3ε4	6 (15.8)	0 (0)	6 (11.5)	32 (19.9)	7 (17.9)	39 (19.5)	45 (17.9)	37 (30.1)
ε4ε4	1 (2.6)	0 (0)	1 (1.9)	1 (0.6)	0 (0)	1 (0.5)	2 (0.8)	2 (1.6)
Total	38 (100)	14 (100)	52 (100)	161 (100)	39 (100)	200 (100)	252 (100)	123 (100)
ApoE allele
ε2	9 (11.8)	5 (17.9)	14 (13.5)	30 (9.3)	6 (7.7)	36 (9.0)	50 (9.9)	17 (6.9)
ε3	58 (76.3)	23 (82.1)	81 (77.9)	252 (78.3)	63 (80.8)	315 (78.8)	396 (78.6)	185 (75.2)
ε4	9 (11.8)	0 (0)	9 (8.7)	40 (12.4)	9 (11.5)	49 (12.3)	58 (11.5)	44 (17.9)
Total	76 (100)	28 (100)	104 (100)	322 (100)	78 (100)	400 (100)	504 (100)	246 (100)

 Data are the number of cases with the percentage of the total in parentheses.

Table 3.

 

 

Odds Ratio of Late AMD Stratified by Type of Disease

Table 3.

 

Odds Ratio of Late AMD Stratified by Type of Disease

AMD Subtype	Subgroup	Allele Type	Sample Size (n)	OR (95% CI)*	P
All late disease		ε2 ε3	37	1.69 (0.79–3.61)	0.18
		ε3 ε4	45	0.58 (0.34–0.98)	0.04
Neovascular	All cases	ε2 ε3	26	1.59 (0.71–3.55)	0.26
		ε3 ε4	39	0.65 (0.37–1.13)	0.13
	Sporadic	ε2 ε3	22	1.72 (0.75–3.94)	0.20
		ε3 ε4	32	0.69 (0.39–1.24)	0.22
	Familial	ε2 ε3	4	0.89 (0.24–3.24)	0.85
		ε3 ε4	7	0.53 (0.20–1.36)	0.19
Atrophic	All cases	ε2 ε3	11	2.30 (0.88–6.04)	0.09
		ε3 ε4	6	0.35 (0.13–0.92)	0.03
	Sporadic	ε2 ε3	6	1.78 (0.58–5.52)	0.32
		ε3 ε4	6	0.48 (0.18–1.30)	0.15
	Familial	ε2 ε3	5	3.58 (0.98–13.1)	0.05
		ε3 ε4	0	NA	NA

 NA not applicable (no familial individuals with the ε4 allele).

*  ε3 ε3, reference genotype; adjusted for age and gender.

Table 4.

 

 

Mean age of Diagnosis in Individuals with Either the ε2 ε3 or ε3 ε4 Genotype Relative to ε3 ε3 in Late AMD

Table 4.

 

Mean age of Diagnosis in Individuals with Either the ε2 ε3 or ε3 ε4 Genotype Relative to ε3 ε3 in Late AMD

Subgroup	Allele Type	Sample Size (n)	Mean Age of Diagnosis (SD)	P
All	ε2 ε3	37	70.5 (6.74)	0.015
	ε3 ε4	45	71.8 (7.78)	0.107
	ε3 ε3	157	73.9 (7.64)	ref
Male	ε2 ε3	10	71.3 (4.88)	0.546
	ε3 ε4	12	72.8 (7.75)	0.910
	ε3 ε3	44	73.1 (8.85)	ref
Female	ε2 ε3	27	70.3 (7.38)	0.011
	ε3 ε4	33	71.5 (7.88)	0.058
	ε3 ε3	113	74.2 (7.13)	ref
Neovascular (n = 190)	ε2 ε3	26	69.8 (6.99)	0.003
	ε3 ε4	39	71.3 (7.99)	0.021
	ε3 ε3	125	74.5 (7.15)	ref
Atrophic (n = 49)	ε2 ε3	11	72.2 (6.11)	0.868
	ε3 ε4	6	75.0 (5.76)	0.398
	ε3 ε3	32	71.7 (9.09)	ref

 ε3 ε3 reference (ref) genotype. Probabilities calculated using a two-sided t-test.