The Role of Apolipoprotein E Gene Polymorphisms in Primary Open-angleGlaucoma FREE

Primary open-angle glaucoma (POAG) is one of the most common causesof blindness and affects approximately 70 million people worldwide.¹ It is thought to be a neurodegenerative disease witha significant hereditary element.² It is likelythat POAG is a complex trait with multiple genes contributing to the phenotype.³ Defining the genetic factors behind glaucoma willadvance our understanding of the pathogenesis of visual failure and may leadto the development of novel treatments.

Clinical studies^4- 5 haveshown an association between glaucoma and Alzheimer disease (AD), which isalso a complex trait. The most important identified genetic risk factor forthe neurodegeneration in sporadic AD is the apolipoprotein E (APOE) genotype.⁶ Apolipoprotein E isthe major apolipoprotein of the central nervous system, where it is synthesizedby glia, macrophages, and neurons.⁷ ApolipoproteinE exists in 3 common isoforms, E2, E3, and E4, encoded by the corresponding APOE gene alleles, ϵ2, ϵ3, and ϵ4.^7- 8 It is therefore possible that the APOE genotype is a common risk factor for neurodegenerationand explains the association between AD and glaucoma. This hypothesis wassupported by the results of a single study⁹ reportingan association between normal-tension glaucoma and the APOE ϵ4 allele. However, normal-tension glaucoma only accountsfor approximately 30% of glaucoma cases,¹⁰ andthe intraocular pressure (IOP) cutoff defining normal-tension glaucoma isin fact arbitrary. To clarify the role of APOE inglaucoma, we investigated the role of APOE polymorphismsas risk factors for POAG in a population in northeast England, using a statisticalapproach that controls for the effects of IOP without making a distinctionbetween low and high pressure.

Having obtained ethics approval from our local research ethics committee,blood samples were analyzed from an unrelated cohort of 137 POAG patientsand 75 control subjects, from the same population.

The definition for POAG included characteristic cupping of the opticdisc, open iridocorneal angle, and typical glaucomatous visual field defects.An experienced ophthalmologist (M.B.), specializing in glaucoma, clinicallyexamined the patients and controls. The highest pretreatment IOP and the cup-discratio were recorded and analyzed as variables. We excluded patients with IOPshigher than 30 mm Hg and secondary types of glaucoma (pseudoexfoliative, pigmentdispersion syndrome, and trauma- or corticosteroid-induced). Patients withIOPs higher than 30 mm Hg were excluded to maximize the chance of detectingpolymorphisms that exert their effect by increasing the vulnerability of theoptic nerve to damage. All controls were older than 65 years and had normalIOPs and optic discs.

The genomic DNA was amplified by polymerase chain reaction of the relatedlocus (19q13.2). For each patient, 1 µL of DNA was mixed with 1 U of Taq DNA polymerase (Promega, Madison, Wis), 10× Promegabuffer, 2-mmol deoxyribonucleoside triphosphate, 10% dimethyl sulfoxide, 0.25µmol/L of each oligonucleotide (primer), and water to a total volumeof 50 µL. The primers used were forward (5'-TCC AAG GAG CTG CAG GCGGCG CA-3') and reverse (5'-ACA GAA TTC GCC CCG GCC TGG TAC ACT GCC A-3').Reactions were treated in a thermal cycle machine to incubation at 94°Cfor 2 minutes, followed by 40 cycles at 65°C for 1 minute, 72°C for1 minute, 94°C for 30 seconds, and a final incubation at 72°C for10 minutes.

The products were then digested separately per sample with 2 restrictionenzymes, Hae II and Afl III(New England Biolabs Inc, Beverly, Mass). The Hae IIdigestion mixture contained 20 µL of polymerase chain reaction product,5 U of enzyme, 0.3 µL of bovine serum albumin, and 3 µL of buffer(buffer 4, New England Biolabs Inc). The Afl IIIdigestion mixture contained 20 µL of polymerase chain reaction product,1 U of enzyme, 0.3 µL of bovine serum albumin, and 3 µL of buffer(buffer 3, New England Biolabs Inc). Water was added to create a total volumeof 30 µL per sample. Both reactions were allowed to proceed for at least12 hours at 37°C.

The resulting fragments were separated by electrophoresis on a 4% MicroSieveagarose gel (Flowgen; Ashby de la Zouch, Leicestershire, England) and visualizedby ethidium bromide staining with a digital camera.

APOE ϵ3 alleles were digested by bothenzymes (Afl III results in 50- and 177–basepair [bp] fragments, and Hae II results in 32- and195-bp fragments) where ϵ4 alleles lack the restriction site for Afl III and ϵ2 alleles lack the restriction site for Hae II. Therefore, the genotype was obtained by combiningthe band patterns for the 2 enzymes per sample.¹¹

To minimize the chance of detecting a spurious statistical association,we used a logistic regression model to simultaneously study the effect ofmultiple variables and their interactions when comparing POAG patients withcontrols.¹² The model assumes that the logarithmof the odds ratio (OR) is a linear function of the variables included in themodel:

Grahic Jump Location

where P indicates the probability of beingaffected; X₁, X₂, . . . ...X_n, the chosen predictor variables; B₀, the intercept for the regression equation; and B₁, B₂, . . . ...B_n, the coefficients reflectingthe nature of each predictor. For the analysis, the predictor variables wereage, IOP, cup-disc ratio, and APOEgenotype. All the variables were added tothe model with stringent forward-selection criteria using Minitab version13.1 statistical package (Minitab Inc, State College, Pa).

Our cohort consisted of 137 POAG patients and 75 controls. The median± SD age was 73.0 ± 8.0 years (range, 51-87 years) for the POAGpatients and 78.0 ± 4.4 years (range, 68-90 years) for the controls.The mean ± SD IOP was 20.8 ± 2.6 mm Hg for the POAG patientsand 16.2 ± 3.4 mm Hg for the controls (Figure 1). The median cup-disc ratios were 0.8 and 0.3 for POAGpatients and controls, respectively (Figure2).

The frequency of ϵ2, ϵ3, and ϵ4 alleles in affected andunaffected individuals is shown in Table1. Logistic regression analysis confirmed that our patient and controlgroups were matched for age and sex but showed no evidence of an associationbetween ϵ2, ϵ3, and ϵ4 allele frequency and POAG (Table 2), irrespective of IOP (ϵ2OR, 0.82; 95% confidence interval, 0.12-5.79 [P =.84]; ϵ3 OR, 0.39; 95% confidence interval, 0.10-1.49 [P = .17]; and ϵ4 OR, 3.84; 95% confidence interval, 0.80-18.49[P = .09]).

Apolipoprotein E, a 36-kDa glycoprotein, is the major apolipoproteinof the central nervous system, where it is synthesized by retinal Müllercells, glia, macrophages, and neurons.^13- 14 ApolipoproteinE plays an important role in neural function and is involved in neurite outgrowthand repair from injury.¹⁵ It is up-regulatedin response to oxidative stress and appears to act as an antioxidant. Apolipoproteinisoforms may affect the onset and severity of several neurodegenerative disorders,and the strongest association is with AD. The APOE ϵ4allele is associated with 40% to 50% of sporadic and familial AD, comparedwith a 30% frequency of the allele in the general population.⁸APOE alleles have been reported to have an effect on recoveryfrom head injury.^8,15 There isalso association between the ϵ4 allele and severe progression of multiplesclerosis and reduced survival time in amyotrophic lateral sclerosis.⁷

Primary open-angle glaucoma can be considered a neurodegenerative disease¹⁶ with a significant hereditary element.² Thestrongest risk factors for glaucomatous optic atrophy are raised IOP and age.However, it is clear that multiple risk factors operate, some of which areas yet undetermined.¹⁷ There is evidence fromtwin studies, case-control studies, and population studies of a heritableelement to POAG.² The risk of glaucoma in first-degreerelatives is 2 to 4 times greater than that in the general population.¹⁸ Recently, several genes have been identified thatmay contribute to POAG,^3,19 includingmyocilin,²⁰OPA1,²¹ and OPTN.²² Itis likely that POAG is a complex trait with multiple genes contributing tothe phenotype, along with environmental factors.

There is evidence that the prevalence of POAG is higher in AD patients.A retrospective study⁴ of medical records ofpatients with POAG found that those with AD had more rapid progression oftheir disc cupping than those without AD. Another study⁵ showedthat AD and Parkinson disease occur more commonly in patients with POAG. Furthermore,at the cellular level, similar neurofilament triplet proteins susceptibleto neurofibrillary tangle formation were identified in AD and POAG.²³ This raises the interesting possibility that the APOE genotype is a common genetic risk factor for the neurodegenerationin POAG and other neurodegenerative diseases.

This study showed no association between APOE genotypeand POAG. The frequency of the APOE ϵ4 allelein our control population was 13%. Power calculations indicate that we had87% power to detect an OR of 3.00 at the P = .05significance level. Our findings are in agreement with a previous study²⁴ carried out on a French cohort with POAG, but theycontrast with those of a previous study⁹ thatreported an increased frequency of the APOE ϵ4allele in a Tasmanian population with normal-tension glaucoma. There are severalpossible explanations for this discrepancy. It is possible that APOE might have a more obvious effect in populations exposed to differentenvironmental factors or with a different genetic background. An alternativeexplanation is that our patients were defined differently compared with theprevious study. The choice of a normotensive cohort, which has also been usedby other authors,²¹ eliminates the strongestrisk factor (IOP) and may make a study more sensitive to underlying neurodegenerativerisk factors. However, the arbitrary division of POAG into normal- and high-tensiontypes may be unhelpful and misleading because POAG represents a spectrum ofphenotypes. Such a distinction based on a specific level of IOP, if wronglychosen, could conceal the true risk factors that operate in POAG or couldreveal spurious associations through artificial stratification.¹⁶ Toresolve these issues, we used a logistic regression model that controls forany effect of IOP during the analysis, without drawing an arbitrary cutoff,enabling a direct analysis of the effects of APOE genotypeirrespective of the IOP. This approach also minimizes the chance of detectinga false-positive result (type I error). This statistical approach may proveto be useful in further studies of POAG, as it allows for a more holisticview of the disease.

There is a strong rationale for pursuing further studies of apolipoproteinE as an etiological factor in glaucoma. Our study suggests that it is notthe isoform itself that is important, but the effect may be more subtle, possiblyinvolving different levels of APOE expression inthe optic nerve. This is supported by the recent finding of an associationbetween POAG and polymorphisms in the APOE promoterregion.²⁴ Further investigations are requiredto confirm these findings in other populations.

Corresponding author: Thomas Ressiniotis, MRCOphth, Department ofOphthalmology, Royal Victoria Infirmary, Queen Victoria Road, Newcastle uponTyne NE1 4LZ, England (e-mail: tomres@doctors.org.uk).

Submitted for publication February 25, 2003; final revision receivedJuly 8, 2003; accepted October 1, 2003.

The study was funded by educational grants from Pharmacia Ltd, Buckinghamshire,England, and from the Special Trustees of the Newcastle upon Tyne HospitalsNHS Trust, Tyne and Wear, England.

The study was performed in partial fulfillment of requirements for amaster of philosophy degree for Dr Ressiniotis.

We thank Richard Andrews, FRCOphth, PhD, for his contribution to thework.