Complement Factor H Polymorphism, Complement Activators, and Risk of Age-Related Macular Degeneration FREE

Age-related macular degeneration (AMD) is the most important cause of irreversible visual loss in the elderly of the Western world.¹ This late-onset disorder causes focal deposition of extracellular material (drusen) underneath the retinal pigment epithelium, ultimately leading to geographic atrophy or subretinal neovascularization. Recent studies provide increasing evidence that inflammation is an important disease mechanism. Drusen were shown to contain complement components and regulators, immunoglobulins, and anaphylatoxins²; C-reactive protein (CRP) was associated with AMD³; and a mouse model lacking the gene for monocyte chemoattractant protein appeared to develop hallmarks of AMD.⁴

It has long been recognized that hereditary factors play a role in AMD. First-degree relatives were shown to have an increased risk,^5- 6 and segregation analysis suggested the presence of a major gene.⁷ Genome-wide linkage analyses identified a disease locus on 1q25-q31,^8- 14 and case-control studies recently identified complement factor H (CFH) as the responsible gene.^15- 21 This gene has many frequent polymorphic variants that relate to AMD.¹⁸ The CFH Y402H variant, located within a binding site for CRP, was consistently shown to have the strongest association in the coding region.^15- 21

Complement factor H is an important inhibitor of the complement pathway. Activation of this pathway initiates a proteolytic cascade that releases proinflammatory anaphylatoxins and causes formation of a membrane-attack complex ultimately leading to cell lysis. Complement factor H preferentially binds and inactivates complement component C3b, and prevents the production of C3 convertase and progression of the cascade.²² The association between CFH and AMD emphasizes an inflammatory pathogenesis of AMD and suggests that triggering the complement cascade in genetically predisposed individuals promotes development of AMD.

The purpose of this study was 3-fold. First, we examined the associations between the CFH Y402H polymorphism and early (less severe) as well as late (vision threatening) AMD in a general population. Second, we investigated whether smoking and other proinflammatory markers may modify the relationship between CFH and AMD. And third, we assessed whether genetic variants of CRP interact with this CFH polymorphism. We investigated these issues in the population-based Rotterdam Study. The large study sample, the variety of risk factors determined at baseline, and the unbiased diagnosis of AMD during a long follow-up particularly addressed the multifactorial origin of AMD and facilitated the study of gene-environment interaction.

The Rotterdam Study is a prospective, population-based cohort study of chronic diseases in the elderly. The eligible population comprised all 10 275 inhabitants aged 55 years or older living in Ommoord, a suburb of Rotterdam, the Netherlands. Inhabitants were ascertained from the municipal register, were invited by mail, and contacted by telephone for a home interview and examinations at the research center. Of the eligible population, 7983 (78%) individuals participated (58% female and 98% white).²³ The ophthalmologic part of the study became operational after the pilot phase of the study had started and consisted of 9774 eligible individuals, of whom 7598 (78%) participated. The investigation was approved by the medical ethics committee of Erasmus University (Rotterdam, the Netherlands), and all participants provided signed, informed consent for participation in the study, publication of obtained data, retrieval of medical records, and the use of blood and DNA for scientific purposes.

Baseline examinations took place from March 20, 1990, through July 31, 1993. One follow-up examination was performed between September 1, 1993, and December 31, 1994, and had a mean (SD) time between baseline and follow-up of 1.98 (0.64) years; another examination was performed between April 15, 1997, and December 31, 2000, and had a mean (SD) time of 6.50 (0.35) years; and the third examination was performed between April 23, 2002, and December 31, 2004, and had a mean (SD) time of 11.08 (0.53) years. At baseline, 6418 participants underwent an eye examination and had gradable fundus photographs; 5681 of these had a successful assessment of the CFH gene polymorphism (88.6% of persons with AMD and 88.4% of those without AMD) and were therefore available for prevalence analyses. Seventy-eight persons with prevalent late AMD (stage 4) were excluded from further incidence analyses. At first follow-up examination, 270 persons had died and 691 were not included in the analyses due to refusal, lost to follow-up, or ungradable fundus photographs, resulting in 4642 individuals with complete data of whom 12 had late AMD. At the second follow-up examination, 663 persons had died and 561 were not included due to refusal, lost to follow-up, or ungradable photographs, leaving 3406 with complete data of whom 32 had late AMD. At the third follow-up examination, 738 persons had died, 249 were not included due to refusal, lost to follow-up, or ungradable photographs, and 2387 individuals had complete data of whom 49 had late AMD. The total number of person-years on which incidence analyses were based was 30 621. Data analysis took place from May 10, 2005, through May 30, 2006.

All participants were genotyped for the CFH Y402H polymorphism (1277 T→C, rs1061170) in 2-ng genomic DNA extracted from leukocytes with the Taqman assay (Applied Biosystems, Foster City, Calif).

To assess variation in the CRP gene, we genotyped single-nucleotide polymorphisms (rs1130864 C→T, rs1205 C→T, rs3093068 C→G), enabling stratification into the 4 haplotypes that are present in persons of European descent (SeattleSNPs, http://pga.gs.washington.edu).²⁴ Overlap between these 3 tagging single-nucleotide polymorphisms was only present in 9 of 10 800 alleles (<0.001%). Haplotypes were estimated with the PHASE program²⁵ and the probability of the estimated haplotypes was 0.999 or higher in all individuals. Haplotypes were coded as 1 to 4 in order of decreasing frequency in the population (coding from rs1130864 C→T, rs1205 C→T, and rs3093068 C→G; haplotype 1 = C-T-C; haplotype 2 = T-C-C; haplotype 3 = C-C-C; and haplotype 4 = C-C-G). In the analysis of interaction between the CRP gene and CFH, not all strata had sufficient cases for haplotype 4. These analyses were restricted to the 3 most common haplotypes, which describe 94.2% of the study population.

The inflammatory mediators erythrocyte sedimentation rate (ESR), CRP serum level, and leucocyte count were evaluated at baseline. Blood was drawn by venous puncture and the ESR was read after 60 minutes. Leukocyte count was directly assessed using a Coulter Counter T540 (Coulter Electronics, Luton, England). Serum samples were initially stored at −20°C, and were thawed and assayed for CRP by Rate Near Infrared Particle Immunoassay method (IMMAGE high sensitive CRP, Beckman Coulter, Fullerton, Calif). Information on cigarette smoking (never, past, and current) was obtained by interview at baseline.²³

Fundus photographs covering a 35° field centered on the macula were taken at each visit (Topcon TRV-50VT fundus camera, Topcon Optical Co, Tokyo, Japan) after pharmacologic mydriasis. All signs of AMD were graded according to the international classification and grading system for AMD under the supervision of senior ophthalmologists (J.R.V. and P.T.V.M.D.J.), and graders were tested regularly for interrater and intrarater agreement.^26- 27 We categorized these signs into 5 exclusive stages that represent an increasing severity of AMD. In short, stage 0 contains no or only small hard drusen; stage 1, either small soft distinct drusen or pigmentary irregularities; stage 2, either soft distinct drusen with pigmentary irregularities or soft indistinct drusen without pigmentary irregularities; stage 3, soft indistinct drusen with pigmentary irregularities; and stage 4, geographic atrophy (dry AMD), neovascular AMD (wet AMD), or mixed AMD (both dry and wet AMD). Persons were classified in stage 0 to 4 based on the eye with the more severe stage. Unaffected participants were defined as those who remained in stage 0 throughout the follow-up period. Incident cases were defined as the absence of a stage in both eyes at baseline and the first appearance of AMD in at least 1 eye at follow-up.

Baseline characteristics of participants were compared among CFH genotypes with analysis of covariance for continuous variables and with logistic regression analysis for discrete variables and were adjusted for age and sex. Natural, log-transformed CRP levels, ESRs, and leukocyte counts were used to normalize a skewed distribution. P<.05 was considered statistically significant along with 95% confidence intervals (CIs) excluding the null hypothesis.

Prevalences of AMD stages were calculated as percentages of the study population at baseline and incidence rates were calculated by dividing the number of incident cases by the number of person-years of follow-up. Duration of follow-up was calculated from the date of fundus photography at baseline until diagnosis of AMD, lost to follow-up, last visit before death, or end of follow-up, whichever came first.

Genotype distributions for CFH were tested for Hardy-Weinberg equilibrium using the χ² test. Prevalence odds ratios (ORs) were estimated with logistic regression analysis and incidence relative risks were estimated with Cox proportional hazards analysis, both analyses were adjusted for age and sex. The proportional hazards assumption was tested by drawing log minus log plots of the survival function, which was verified by visual inspection. The ORs based on both prevalent and incident cases were estimated with logistic regression analyses with age, sex, and follow-up time in the model. Cumulative risks of late AMD were calculated using Kaplan-Meier product-limit analysis in the presence of competing risks. Participants who died, and those who were free of AMD and were lost to follow-up, were censored at the time of their last visit. P values from log-rank tests of equality were used to test significant differences between CFH genotypes. The formula to calculate population attributable risk is: population attributable risk = (relative risk −1/relative risk) × proportion of exposed. Relative risk of late AMD in this formula is estimated by the OR. The proportion exposed is the proportion of participants with late AMD carrying the Y402H allele.

To maximize contrast between affected and healthy individuals, the disease outcome for the interaction analyses was defined as late AMD (stage 4) vs no AMD (stage 0). Risk assessments were initially performed for prevalent and incident cases separately. However, because risk estimates were similar, interaction analyses were subsequently performed using pooled data of prevalent and incident cases to increase statistical power. For the pooled data, ORs were used as representative for relative risks because pooling allows a case-control analysis. The ESRs were evaluated using categories of normal and elevated rates (cutoff of 12 mm per hour for men and 20 mm per hour for women). Because CRP levels and leukocyte counts were within normal limits (CRP<8 mg/L; leukocyte count, 3.5-10.0 × 10⁹/L) for all participants, we evaluated these mediators in tertiles. Smoking was stratified as never, past, and current. Biological interaction with CFH was assessed by calculating the synergy index.²⁸ The synergy index measures deviation from additivity of 2 risk factors and is based on the ratio of the combined effect to the sum of the separate effects. A synergy index exceeding 1.00 suggests the presence of at least 1 pathway in the pathogenesis of the disease for which both risk factors are required.

Interaction with the CRP gene was studied with homozygous haplotype 1 carriers (1-1) as the reference category because this genotype was associated with the lowest CRP levels. We pooled all haplotype 2 carriers (2*) in 1 group and all haplotype 3 carriers in another group (3*). Individuals with haplotypes 2 and 3 (2-3) were present in both groups 2* and 3*.

At baseline, we identified 2062 (36.3%) of 5681 persons with any type of AMD and 78 (1.4%) with late AMD (stage 4). During follow-up (mean, 7.85 years; median, 10.31 years), we identified 1858 (40.0%) of 4642 persons who had remained in stage 0 and 1649 (35.5%) who progressed to a higher stage of AMD. Of these 1649, 93 persons (5.6%) had late AMD. The baseline characteristics and frequency of AMD stratified by CFH genotypes are presented in Table 1.

Genotype frequencies of CFH were analogous for participants and nonparticipants and were in Hardy-Weinberg equilibrium. The risks of AMD for the CFH Y402H genotypes appear in Table 2. The prevalence of the CFH Y402H allele was 36.2% (4116/11 362). Risks were consistently higher for homozygous than for heterozygous persons, representing an allele-dose effect. Overall, carriers of the CFH Y402H allele had higher risks of all types of AMD than noncarriers and risks increased with each stage of AMD.

The ORs of bilateral late AMD (17.93; 95% CI, 9.00-35.70) were higher than of unilateral late AMD (6.58; 95% CI, 3.47-12.48) in homozygotes and the same trend was seen in heterozygotes. Within AMD subtypes adjusted for bilaterality, homozygotes had higher ORs of geographic atrophy (17.2; 95% CI, 6.7-44.2) and mixed AMD (19.2; 95% CI, 6.1-60.2) than of neovascular AMD (12.4; 95% CI, 2.4-65.4), although differences between subtypes did not reach statistical significance (geographic atrophy vs neovascular AMD, P = .63; mixed vs neovascular AMD, P = .66; geographic atrophy vs mixed AMD, P = .56). The population attributable risk of late AMD for CFH Y402H was 54.0% (ie, 23.2% for those carrying 1 allele and 30.8% for those carrying 2 alleles).

The cumulative incidence of late AMD for the CFH genotypes appears in Figure 1. Risk differences between carriers and noncarriers became statistically significant after age 75 years (log-rank P<.01). At age 95 years, risks had increased to 48.3% (95% CI, 28.8%-67.7%) for homozygous and 42.6% (95% CI, 24.3%-60.8%) for heterozygous carriers compared with 21.9% (95% CI, 5.4%-38.4%) for noncarriers.

Interaction between ESR, CRP level, leukocyte count, smoking, and CFH Y402H is demonstrated in Figure 2. The joint effect of each determinant with CFH Y402H was significantly greater than the sum of the independent effects except for leukocyte count. An elevated ESR augmented the association to an OR of 20.2 (95% CI, 9.5-43.0). Higher serum CRP levels increased the association for the second tertile to an OR of 27.7 (95% CI, 10.7-72.0) while the third tertile did not further increase the OR. Neither ESR nor CRP levels were significantly associated with AMD in noncarriers.

Smoking had a large influence on the risk of AMD related to CFH. Current smokers homozygous for CFH Y402H had an OR of 34.0 (95% CI, 13.0-88.6) for late AMD compared with individuals who never smoked without the risk allele. In the absence of CFH Y402H, current smoking increased the OR for AMD to 3.36 (95% CI, 1.14-9.86).

Serum levels of CRP varied among the CRP haplotypes in this study population. Haplotype 1 (the most common) had a frequency of 32.7% and the lowest CRP levels; haplotype 2 had a frequency of 31.6% and the highest CRP levels; and haplotype 3 had a frequency of 29.9% and intermediate levels of CRP.²⁹ The CRP haplotypes per se were not related to AMD. We tested the hypothesis whether these haplotypes influenced the effect of CFH Y402H on AMD (Figure 3). Compared with noncarriers of CFH Y402H with CRP haplotype 1, noncarriers with CRP haplotype 2 had an OR of AMD of 0.17 (95% CI, 0.06-0.46; P<.001) and noncarriers with CRP haplotype 3 had an OR of 0.25 (95% CI, 0.09-0.64; P = .004). In contrast, in homozygous CFH Y402H carriers, the OR of AMD was 3.32 (95% CI, 1.38-8.01; P = .007) for haplotype 2 and 3.86 (95% CI, 1.56-9.53; P = .003) for haplotype 3. The highest difference in ORs between homozygous carriers and noncarriers was observed for CRP haplotype 2, which is the haplotype with the highest CRP levels. Our results show that those participants homozygous for CFH Y402H with an additional genetic predisposition to high serum CRP levels were at higher risk of developing AMD.

In this prospective study, which was based on an older, white population in the middle socioeconomic class in the Netherlands, we find that the CFH gene is a major risk factor for AMD. The gene was implicated in all stages of AMD from early hallmarks such as drusen to vision-disabling late AMD and the risks increased with each successive stage to a high of 11 for late AMD. We calculated that individuals homozygous for the CFH Y402H polymorphism had a 48% risk of developing late AMD by age 95 years while this risk did not exceed 22% for noncarriers. These data suggest that CFH Y402H may be a causal factor in more than 50% of all AMD cases in the general population.

Previous reports on the association between CFH and AMD were from clinic or family-based, case-control studies with cross-sectional designs. This hampers extrapolation of the role of CFH in AMD development for the population at large. Strengths of our current study are the population-based prospective design, the large study sample, and the use of standardized procedures for AMD diagnosis by experienced graders.²⁷ We restricted the analyses of potential modifiers to individuals with late AMD (stage 4) and those who remained free of any type of AMD (stage 0) throughout the study. To maximize statistical power and enable precise risk estimates, we pooled prevalent and incident cases with late AMD. Prevalent cases showed similar risk estimates as incident cases so the exposures to inflammation and smoking likely preceded AMD and pooling did not jeopardize causal inference. Although not negligible, our a priori hypothesis and assessment of only 1 single-nucleotide polymorphism in the CFH gene make it unlikely that our findings are falsely positive.

Complement factor H was associated with both late AMD subtypes in this study. Homozygous CFH Y402H carriers had a higher risk of bilateral than of unilateral late AMD, and risks of geographic atrophy and mixed AMD were slightly but not significantly higher than neovascular AMD. This is in agreement with other studies that reported higher frequencies of CFH Y402H carriers in persons with geographic atrophy,^19,30 and 1 study that suggested a lower risk of geographic atrophy for a CFH haplotype containing the nonrisk allele.¹⁸ Nevertheless, the high risk for both subtypes signifies a common inflammatory pathogenesis.

Complement factor H is an important regulator of the complement system. Three enzyme cascades exist: the classical complement pathway, initiated by antigen-antibody complexes and surface-bound CRP; the lectin, turned on by mannose groups of microbial carbohydrates; and the alternative complement pathway, activated by surface-bound C3b. The pathways converge at the point in which C3 is cleaved into C3a and C3b by C3 convertase, which initiates C5 convertase, resulting in the formation of the membrane-attack complex with the terminal components (C5b-C9). Complement factor H specifically inhibits the alternative complement cascade but also regulates the common pathway. It binds C3b and acts as a cofactor in the proteolysis of C3b by factor I, resulting in an inactive C3b molecule. This prevents the production of C3 convertase in the alternative cascade as well as the production of C5 convertase in the common pathway. As a result, CFH interferes with progression of the entire cascade.^22,31

Genetic predisposition to a malfunctioning CFH can only be of importance when the complement system is switched on. Our data provide strong evidence that onset of this cascade leads to AMD in persons with the CFH Y402H polymorphism. This is demonstrated by the significant interaction between chronic as well as acute inflammation and CFH Y402H. Elevated baseline ESRs considerably increased the risk of AMD in carriers and a similar trend was observed for serum CRP levels. Neither ESRs nor CRP levels increased the risk significantly in noncarriers. Earlier studies reported a relationship between serum CRP level and progression of AMD.³ However, our results imply that this relationship is mostly determined by the CFH polymorphism. Increased leukocyte counts did not contribute to an additional effect, possibly due to the absence of clinically elevated levels in our study.

Smoking was considered the highest risk factor for AMD prior to the introduction of CFH in AMD. Our data show that the combined effect of both exposures exceeds the sum of the independent effects. Compared with no exposure, smoking increased the risk of AMD 3.3 times, the presence of 2 CFH Y402H alleles increased the risk 12.5 times, while the combination of both determinants increased this risk 34-fold. Smoking increases cytokines and inflammatory cells and has been shown to activate the complement pathway by weakening the susceptibility of C3 to CFH and factor I.^32- 33 When CFH function is genetically impaired, progression may be further accelerated. However, the elevated risk of smoking in noncarriers suggests that smoking may have an alternative mechanism in AMD pathogenesis.

We further explored the relationship with CRP for 2 reasons. First, the CFH Y402H variant represents an amino acid change in the SCR7 domain, which contains a binding site for CRP, heparin, and M-protein, prompting a functional interaction with these proteins.²² Second, CRP not only triggers the classical complement cascade by binding to C1q, it also limits the amount of complement activation by its ability to interact with CFH, thereby reducing the complement-associated damaging effect.³⁴ Because serum CRP levels are known to fluctuate, a single measurement of CRP may not accurately reflect a continuous baseline level nor adequately represent the possible response after an inflammatory stimulus. This motivated us to examine the CRP gene. Our data suggest that CRP haplotypes, which increase serum levels, modify the effect of CFH. These haplotypes decrease the risk of AMD in noncarriers but increase the risk in persons homozygous for CFH Y402H. We propose the following as a plausible biological mechanism. Genetic variants of CRP have been shown to determine serum levels especially in response to inflammatory stimuli.³⁵ The CFH Y402H polymorphism may impair CRP binding, decrease CFH inhibition, and lead to destruction of host cells in particular in those individuals who are genetically predisposed to high CRP levels. By contrast, normal binding of CFH with CRP may increase inhibition and decelerate complement activation in those who are hyperresponsive to inflammation.³¹

In conclusion, CFH, an inhibitory gene of the complement pathway, is a major risk factor for AMD in this population. It is involved in early as well as late disease pathogenesis and markedly increases risk of late AMD in the very old. The effect of CFH is significantly influenced by environmental and genetic factors that determine the inflammatory response and activate the complement pathway. Future research on therapeutic modalities that help regulate the terminal complement pathway, thereby sparing host tissue, may provide an approach for preventing sight-threatening AMD in genetically predisposed individuals.