Relevance of Complement Factor H–Related 1 (CFHR1) Genotypes in Age-Related Macular Degeneration

Abstract

Purpose.: Age-related macular degeneration (AMD) has a strong genetic component with a major locus at 1q31, including the complement factor H (CFH) gene. Detailed analyses of this locus have demonstrated the existence of one SNP haplotype block, carrying the CFH 402His allele, which confers increased risk for AMD, and two protective SNP haplotypes, one of them carrying a deletion of the CFHR1 and CFHR3 genes (Δ _CFHR3-CFHR1 ). The purpose of these studies was to evaluate the contribution of newly described CFHR1 alleles to the association of the 1q31 locus with AMD.

Methods.: Two hundred fifty-nine patients and 191 age-matched controls of Spanish origin were included in a transversal case–control study using multivariate logistic regression analysis and ROC (receiver operating characteristic) statistics to generate and test models predictive of the development of AMD.

Results.: This study showed for the first time that a particular CFHR1 allotype, CFHR1*A, is strongly associated with AMD (odds ratio, 2.08; 95% confidence interval, 1.59–2.73; P < 0.0001) and illustrate a peculiar genotype–phenotype correlation between the CFHR1 alleles and different diseases that may have important implications for understanding the pathophysiology of AMD. It also shows that CFHR1*A is in strong linkage disequilibrium with the CFH 402His allele, which provides additional candidate variants within the major risk haplotype at 1q31, promoting its association with AMD. Further, using the Spanish population as a model, the results showed that analysis of the CFHR1 genotypes provide sufficient information to delineate the individual risk of developing AMD.

Conclusions.: The results support a relevant role of CFHR1 in the pathogenesis of AMD.

Age-related macular degeneration (AMD) is the most common cause of visual disability in the elderly in developed countries. ¹ It is a multifactorial disease, influenced by age, ethnicity, and a combination of environmental and genetic risk factors. ² Genetic predisposition to AMD has been suggested based on familial segregation and twins studies. Several candidate genes with a relatively minor contribution and two major susceptibility loci (1q31, CFH/CFHR1/CFHR3; 10q26, HTRA1/ARMS2) that independently contribute to the risk of AMD have been identified by candidate region studies and whole-genome association analyses. ^{3 –10}  

The 10q26 locus includes two nearby genes, ARMS2 (age-related maculopathy susceptibility 2, also known as LOC387715) ¹¹ and HTRA1 (high-temperature requirement factor A1). ¹¹ The most studied SNP at this locus is rs10490924, which causes the Ala69Ser amino acid substitution in ARMS2, a constituent of the extracellular matrix. The risk-associated allele ARMS2 69Ser is in strong linkage disequilibrium with a genetic variant resulting from a combination of a deletion and insertion (*372_815delins54) in the 3′ untranslated region of the ARMS2 mRNA, which affects the stability of the ARMS2 mRNA and supports a role for ARMS2 in AMD pathogenesis. ¹¹  

Several independent studies have shown that the CFH 402His allele (rs1061170), at the CFH locus in 1q31, confers a significantly increased risk to AMD with an odds ratio (OR) between 2.1 and 7.4. ^{3,5 –7} The association of CFH with AMD strengthened the implication of the complement system in the pathogenesis of AMD ⁵ and has prompted subsequent candidate gene approaches that identified additional associations of complement genes with AMD. ^{4,9,10,12 –15} As a whole, these genetic data reinforce the concept that complement dysregulation is a major player in AMD pathogenesis. This conclusion is supported by the functional analysis of the AMD-associated SNPs in the CFH, ¹⁶ CFB, ¹⁷ and C3 ¹⁸ genes. The CFH Tyr402His polymorphism has also been found to have functional implications. Notably, it alters the binding specificity of factor H for different glycosaminoglycans ^19,20 and decreases its binding to retinal pigment epithelial cells ²¹ and Bruch's membrane. ²² However, the physiological relevance of these observations is still unclear. 

Within the CFH gene, downstream of the Tyr402His polymorphism, there are three polymorphisms, a synonymous SNP in exon 11 (rs2274700) and two intronic SNPs (rs1410996 and rs7535263), showing stronger association with disease susceptibility than the CFH 402His variant. ^14,15 These polymorphisms and the CFH 402His variant form part of the same CFH haplotype conferring increased risk to AMD (haplotype H1). ²³ Two additional haplotypes in the CFH gene (haplotypes H2 and H4) are markedly decreased in AMD and therefore have been described as associated with lower risk for AMD. ^13,15,24 CFH haplotype H2 carries the Ile62 factor H variant showing increased complement regulatory activity, ¹⁶ which likely confers lower risk to AMD by reducing complement activation. CFH haplotype H4 is also associated with decreased risk of AMD. ^13,25 Interestingly, this CFH haplotype is also unique because it carries a deletion of the CFHR1 and CFHR3 genes (Δ _CFHR3-CFHR1 ). ¹³ Although it has been indicated that the CFHR1 and CFHR3 proteins have the potential to compete with factor H and interfere with its complement regulatory activity, the potential benefit of the absence of CFHR1 and CFHR3 proteins is still puzzling, as these two proteins are considered complement regulators. ²⁶ Recently, it has been proposed that complement activity is determined by a homeostatic balance between CFHR1, CFHR3, and factor H and that loss of CFHR1 and CFHR3, affecting local factor H binding, may increase local factor H-mediated regulation and protecting against AMD development. ²⁷  

We provide further insights into the gene variants encoded by the AMD locus at 1q31. We show that CFHR1*A, a recently described allele encoded by the CFHR1 gene, is associated with AMD and in strong LD with the CFH 402His allele. These data add to previous results and support a relevant role of CFHR1 in AMD pathogenesis. To further illustrate the relevance of the CFHR1 gene, we tested the value of a model of disease risk based on the CFHR1 genotypes. This model provides meaningful predictive values for the development of AMD that are comparable to those obtained with models based on SNPs at the CFH gene. 

Materials and Methods

Patients and Controls

Our study included 259 patients of Spanish origin with advanced AMD and a cohort of 191 age-matched controls. The cases were recruited from the Department of Ophthalmology, Clínica Universidad de Navarra, Pamplona, Spain. The general inclusion criteria for the patients with advanced AMD and the controls were as follows: age 60 years or older; absence of other retinal diseases related to choroidal neovascularization (CNV), such as angioid streaks, a nevus in the macular area, toxoplasma scars, photocoagulation scars in the posterior pole, or polypoidal choroidal vasculopathy; and less than 6 D of myopia. The inclusion criteria for patients with advanced AMD included drusen and a chorioretinal macular atrophy involving central macular signs related to CNV at least in one eye (category 4 in the AREDS study). ²⁸ The inclusion criteria for controls were an absence of drusen or no more than five small drusen (none exceeding 65 μm), an absence of retinal pigment abnormalities in the macular area, and an absence of chorioretinal macular atrophy or any form of CNV (category 1 of AREDS study). ²⁸ All samples were obtained with informed consent, in accordance with the Declaration of Helsinki and our institutional review boards. 

Genotyping

Genomic DNA was extracted from peripheral blood using standard procedures or from oral swabs (QIAcube; Qiagen, Valencia, CA). DNA samples were genotyped for seven SNPs in four of the genes previously shown to be associated with AMD (CFH Ile62Val, CFH Tyr402His, CFH c.2237-543A>G, CFB Leu9His, CFB Arg32Gln/Trp, C3 Arg102Gly, and ARMS2 Ala69Ser). The genotyping was performed using multiplex PCR and primer extension methodology (ABI Snapshot; Applied Biosystems, Inc. [ABI], Foster City, CA) in an automated sequencer (model 3730; ABI), and the fragments were analyzed on computer (GeneMapper Software ver, 4.0; ABI). The analysis of the Δ _CFHR3-CFHR1 polymorphism and the genotyping of CFHR1 was performed as described previously. ²⁹ All the polymorphisms analyzed in this report were in Hardy-Weinberg equilibrium. 

Haplotype Determinations

Linkage disequilibrium analysis of the variants within CFH and CFHR1 polymorphic loci was performed with MIDAS software. ³⁰ Haplotype frequencies in the controls and patients were estimated by using the expectation maximization algorithm implemented by the SNPstats software (http://bioinfo.iconcologia.net/snpstats/start.htm). Calculation of P values between groups was performed by Pearson's χ² test of association and considered significant if the two-sided P was less than 0.05. Odds ratios (ORs) and 95% confidence intervals (CIs) were also calculated. 

Statistical Analysis

Allele frequencies differences between cases and controls were assessed by performing a Pearson's χ² test of association and OR and 95% CI were calculated. CFH, CFHR1, CFB, and ARMS2 genotypes for each individual were included in a transversal, comparative case–control study using multivariate logistic regression analysis and ROC (receiver operating characteristic) statistics to generate and test models predictive of development of AMD, based exclusively on the genotypes of the genes studied. Cumulative risk scores (z) were determined by the equation z = α + Σβ_i X _i where the regression coefficients α (constant) and β_i (risk score specific for each genotype X) are taken directly from the multivariate logistic regression analysis. Probabilities of developing AMD (P) were calculated as P = e ^Z/(1 + e ^Z) and were categorized in four risk groups: very low (P < 25%), low (25% > P < 50%), high (50% > P < 75%), and very high (P > 75%). A two-sided P < 0.05 was considered statistically significant (SPSS software for Windows, ver. 14.0; SPSS, Chicago, IL). 

Results

The CFHR1*A Allele Is Strongly Associated with AMD

We have recently described a novel polymorphism of the CFHR1 gene with two alleles, CFHR1*A and CFHR1*B, encoding CFHR1 proteins that present different degree of similarity to factor H (Fig. 1). The CFHR1*B allotype, with greater sequence similarity to factor H, is strongly associated with atypical hemolytic uremic syndrome (aHUS), perhaps suggesting that increased competition between CFHR1*B and factor H decreases the protection of cellular surfaces from complement damage. ²⁹ To study whether this novel CFHR1 polymorphism also influences predisposition to AMD, we genotyped our AMD cohort and the matched control population for the CFHR1 allotypes. In addition, we included in these studies the analysis of other polymorphisms that had been associated with AMD ^{3 –10,13} (see the Materials and Methods section). The allelic frequencies of each of these polymorphisms were determined and compared between the two groups (Table 1). In agreement with previous studies, we found strong positive associations between the alleles CFH 402His, CFH c.2237-543G, and ARMS2 69Ser and AMD and a strong protective association of CFH 62Ile, CFB 9His, and CFB 32Gln/Trp against AMD. C3 102Gly, previously found to be associated with AMD, shows a similar positive trend in our population that was not statistically significant. The three CFHR1 alleles show very distinct associations with AMD. A very strong and significant association was found between AMD and the CFHR1 allotype CFHR1*A (OR, 2.08; 95% CI, 1.59–2.73), whereas no significant differences in the frequency of the CFHR1*B allele were found between AMD patients and controls (0.34 vs. 0.38). As previously reported, the third CFHR1 allele Δ _CFHR3-CFHR1 showed a very significant protective effect against AMD (OR, 0.34; 95% CI, 0.23–0.50; Table 1). The associations between the six CFHR1 genotypes and AMD are depicted in Table 2. CFHR1*A showed a very strong positive association with AMD, but only when this allele is homozygous (OR, 3.08; 95% CI, 1.92–4.92). The positive association of the CFHR1*A allele with AMD disappeared in the CFHR1*A-CFHR1Δ _CFHR3-CFHR1 heterozygous carriers. In fact, the frequency of these heterozygotes was significantly decreased in AMD, indicating that this genotype protects against AMD (OR, 0.40; 95% CI, 0.24–0.69). Therefore, the CFHR1Δ _CFHR3-CFHR1 allele has a dominant effect over the CFHR1*A allele. The CFHR1 genotype data shown in Table 2 confirm the neutral role of the CFHR1*B allele in AMD. In agreement with previous reports, the CFHR1Δ _CFHR3-CFHR1 homozygous genotype had a strong protective effect against AMD (OR, 0.48; 95% CI, 0.17–1.37), although it is not statistically significant in our study because of the reduced size of our control and patient cohorts. 

Figure 1.

 

Structural similarities between the CFHR1 alleles and factor H. Structural organization of factor H and CFHR1*A and CFHR1*B. Short consensus repeats (SCRs) are represented by circles and are numbered from the N-terminal end. Homologous SCRs are aligned and the amino acid differences from factor H indicated for SCR-3, -4, and -5 of CFHR1*A and CFHR1*B. SCR-1 and -2 of CFHR1 show only a 41% homology with factor H. The homologous position to factor H 402His in both CFHR1*A and CFHR1*B is 100Ser.

Table 1.

 

Polymorphisms in the CFH, CFB, C3, CFHR1, and ARMS2 Genes Associated with AMD

Gene (SNP)	Allele	Allele Frequencies		P	OR (95% CI)
		Controls (n = 191)	AMD (n = 259)
CFH (rs800292)	c.62Ile	0.25	0.14	<0.0001	0.48 (0.34–0.67)
CFH (rs1061170)	c.402His	0.29	0.49	<0.0001	2.27 (1.71–3.00)
CFH (rs1410996)	c.2237–543G	0.55	0.76	<0.0001	3.96 (2.98–5.30)
CFB (rs4151667)	c.9His	0.04	0.02	0.0087	0.34 (0.14–0.79)
CFB (rs12614; rs64115; rs641153)	c.32Gln/Trp	0.26	0.19	0.0093	0.66 (0.48–0.90)
ARMS2 (rs10490924)	c.69Ser	0.2	0.45	<0.0001	3.32 (2.45–4.51)
	CFHR1*A	0.39	0.57	<0.0001	2.08 (1.59–2.73)
CFHR1	CFHR1*B	0.38	0.34	0.176	0.83 (0.63–1.09)
	Δ CFHR1-CFHR3	0.23	0.09	<0.0001	0.34 (0.23–0.50)
C3 (rs4151667)	c.102Gly	0.2	0.24	0.094	1.32 (0.95–1.82)

 The nomenclature of the polymorphisms is referred to the transcription start site (+1) of each gene. Alleles at the CFHR1 gene are as described in the text. The P value is the result of a two-sided Pearson's χ² test of association for the comparison of the allelic frequencies between the AMD and control groups.

Table 2.

 

CFHR1 Genotypes Associated with AMD

Genotype	Controls (n = 191)		AMD (n = 259)		P	OR (95% CI)
	n	Frequency	n	Frequency
CFHR1*A/CFHR1*A	29	0.15	92	0.36	<0.0001	3.08 (1.92–4.92)
CFHR1*B/CFHR1*B	32	0.17	38	0.15	0.55	0.85 (0.51–1.43)
Δ CFHR1-CFHR3 /Δ CFHR1-CFHR3	9	0.05	6	0.02	0.16	0.48 (0.17–1.37)
CFHR1*A/Δ CFHR1-CFHR3	40	0.21	25	0.1	0.0008	0.40 (0.24–0.69)
CFHR1*B/Δ CFHR1-CFHR3	30	0.16	11	0.04	<0.0001	0.39 (0.12–0.49)
CFHR1*A/CFHR1*B	51	0.27	87	0.34	0.12	1.39 (0.92–2.09)

CFHR1*A Is in Strong Linkage Disequilibrium with CFH 402His

Strong linkage disequilibrium across the CFH-CFHR3-CFHR1 genes has been described earlier. ^24,27 Consistent with these findings, we found strong linkage disequilibrium between SNPs in the CFH gene and the CFHR1 alleles in both the AMD and control populations (Fig. 2), which further illustrates that the CFH haplotypes associated with AMD extend away from the CFH gene to include the CFHR1 alleles. Comparison of the frequencies of these extended haplotypes (including the CFH, CFHR3, and CFHR1 genes) between the AMD and the control cohorts demonstrated three major extended haplotypes associated with risk for or protection against AMD that correlate well with the CFHR1 alleles. Accordingly, CFHR1Δ _CFHR3-CFHR1 occurs predominantly on an extended haplotype that includes the AMD inhibitor CFH haplotype H4; CFHR1*A is in strong linkage disequilibrium with the AMD risk allele CFH 402His and occurs predominantly in the AMD risk CFH haplotype H1; and CFHR1*B distributes almost equally between two extended haplotypes: one including the protective CFH haplotype H2 and another including the neutral CFH haplotype H3 (Fig. 2). The CFH 402His and CFHR1*A alleles also showed much stronger linkage disequilibrium in the controls than in the AMD patients (0.90 vs. 0.78; Fig. 3). This increase could be related to the slight increase of some minority haplotypes carrying CFHR1*A without CFH 402His in the AMD cohort (Fig. 4). If confirmed, these data would emphasize the relevance of CFHR1 for the association of this extended CFH haplotype with AMD. 

Figure 2.

 

CFH-CFHR3-CFHR1 haplotypes and their association with AMD. The exon structure of CFH, CFHR3, and CFHR1 showing the location of the polymorphism included in these studies. Haplotype frequencies in the control and patient cohorts were estimated using the EM algorithm (expectation maximization algorithm) implemented by the SNPStats software. CFH haplotypes with a frequency >1% are shown. The frequency of each CFH haplotype was compared between the controls and the AMD cohorts and the P-values and the OR were calculated. Red: risk haplotypes; green: protective haplotypes. P values were derived using Pearson's χ² test of association. OR and 95% CI are shown. The nucleotide and amino acid numbers are in reference to the translation start site (A in ATG is +1; Met is +1) as recommended by the Human Genome Variation Society (HGVS).

Figure 3.

 

Linkage disequilibrium analysis between CFH SNPs and CFHR1 alleles.

Figure 4.

 

Model of disease risk based on CFHR1, CFB, and ARMS2 genotypes. (a) Comparison of disease risk models. Model 1 (based on genotypes of the CFH Tyr402His, Δ _CFHR3-CFHR1 , CFB Leu9His, CFB Arg32Gln/Trp, and ARMS2 Ala69Ser polymorphisms). Model 2 (based on genotypes of the CFHR1, CFB Leu9His, CFB-Arg32Gln/Trp, and ARMS2-Ala69Ser polymorphisms). The ROC curve analysis showed no significant differences between both models (AUC values: 0.79 vs. 0.78) in assessing the predictability of AMD. Optimal risk cutoffs provide sensitivities and specificities of approximately 70% for both models. (b) Distribution of cases and controls for model 2. (c) Distribution of Spanish AMD cases and controls according to their frequencies in the risk groups. Risk categories were established as described in Materials and Methods.

Usefulness of the CFHR1 Genotypes in the Calculation of AMD Risk Scores and Probability of AMD Development

To evaluate the relevance of the CFHR1 genotypes in assessing the predictability of AMD development, we compared two alternative models including different sets of polymorphisms associated with AMD in the Spanish population. Model 1 considers the CFH Tyr402His genotypes together with those of the CFB Leu9His, CFB Arg32Gln/Trp, ARMS2 Ala69Ser, and Δ _CFHR3-CFHR1 polymorphisms. Model 2 is based on the genotypes at the triallelic locus CFHR1 in addition to the genotypes of the CFB Leu9His, CFB Arg32Gln/Trp, and ARMS2 Ala69Ser polymorphisms. We used multivariate logistic regression analysis and ROC statistics for the calculation of the AMD risk scores and the probability of AMD development using these two models (Table 3; Fig. 4). Regression coefficients (risk scores) for all the genotypes were determined comparing their frequencies in cases and controls for the two models (Table 3). To assess the cumulative AMD risk scores (z) for each individual, we added the scores (α and β regression coefficients) obtained in the multivariate logistic regression analysis (Table 3) for each of the genotypes carried by the individual. The ROC curve analysis showed no significant differences between both models (AUC 0.79 vs. 0.78) in assessing the predictability of AMD (Fig. 4). Furthermore, at the appropriate cumulative risk score cutoffs, both models yielded virtually identical sensitivities and specificities of approximately 70%. Therefore, the CFHR1 genotypes provide as meaningful predictive values for AMD as the CFH Tyr402His and Δ _CFHR3-CFHR1 genotypes. 

Table 3.

 

Logistic Regression Models

Gene (Polymorphism)	Genotype	Model 1 (402His; Δ CFHR3-CFHR1 )			Model 2 (CFHR1)
		Regression Coefficient (β)	P	OR (95% CI)	Regression Coefficient (β)	P	OR (95% CI)
CFB (L9H)	L/L	0			0
	L/H	−1.114	0.018	0.32 (0.12–0.82)	−1.084	0.023	0.34 (0.13–0.86)
CFB (R32Q/W)	R/R	0.632	0.252	1.88 (0.64–5.54)	0.552	0.317	1.73 (0.60–5.11)
	R/Q or W	−0.284	0.613	0.75 (0.25–2.26)	−0.359	0.524	0.70 (0.23–2.10)
	Q or W/Q or W	0			0
ARMS2 (A69S)	A/A	0			0
	A/S	0.960	<0.001	2.61 (1.64–4.16)	0.963	<0.001	2.62 (1.65–4.15)
	S/S	2.148	<0.001	8.57 (3.89–18.8)	2.223	<0.001	9.23 (4.19–20.3)
CFH (Y402H)	Y/Y	0			—	—	—
	Y/H	0.585	0.017	1.79 (1.11–2.90)	—	—	—
	H/H	1.178	0.001	3.25 (1.60–6.61)	—	—	—
Δ CFHR3-CFHR1	Del/Del	−1.577	0.025	0.21 (0.05–0.80)	—	—	—
	Del/wt	−1.029	<0.001	0.36 (0.21–0.60)	—	—	—
	wt/wt	0			—	—	—
CFHR1 (A/B/Δ CFHR3-CFHR1 )	A/A	—	—	—	0.992	0.004	2.69 (1.37–5.31)
	B/B	—	—	—	0
	Del/Del	—	—	—	−1.041	0.102	0.35 (0.10–1.23)
	A/B	—	—	—	0.525	0.111	1.69 (0.88–3.23)
	A/Del	—	—	—	−0.538	0.157	0.58 (0.28–1.23)
	B/Del	—	—	—	−1.012	0.026	0.36 (0.15–0.89)
	Constant (α)	−0.678			−0.658

Figure 4b shows a plot of the individual probabilities (P) of developing AMD, calculated from the cumulative AMD risks scores for model 2 (see the Materials and Methods section). In general, cases had higher probabilities than controls. To illustrate this better, we categorized the probabilities in four groups of risk: very low (VL; P < 25%), low (L; 25% > P < 50%), high (H; 50% > P < 75%), and very high (VH; P > 75%). As expected, the controls with no sign of AMD tended to cluster in the low risk groups, whereas our AMD cases fell predominantly into the high-risk groups (Fig. 4c). 

AMD shows a prevalence of 25% in the general Spanish population older than 80 years. ³¹ We used this statistic and the frequency distribution of cases and controls in the four risk categories to simulate the frequencies of AMD cases within each risk category in the general population and thus to provide an estimation of the relative risk of having AMD by 80 years of age of the individuals in each category (Fig. 5). With this model, an individual in the very low-risk (VL) category has a 10-fold lower relative risk for AMD than one in the very high-risk (VH) category. The overall population prevalence of 25% was exceeded only in the two highest risk groups (H and VH), with an estimated AMD prevalence rising to 61% for those in the VH risk group. This risk group represents approximately 18% of the general population over 80 years of age. 

Figure 5.

 

Distribution of AMD cases and healthy individuals in the >80-year-old Spanish population according to their predicted distribution in the risk groups and on a prevalence of AMD of 25%. Percentages of AMD (black) cases and healthy individuals (white) within each group are indicated.

Discussion

CFHR1*A, a recently described allele in the CFHR1 gene, is overrepresented in AMD patients (OR, 2.08; 95% CI, 1.59–2.73; P < 0.0001) and is included in the risk haplotype carrying the CFH 402His allele. The identification of CFHR1*A as a risk factor for AMD is consistent with the strong protective effect against AMD associated with the deletion of this gene and suggests a relevant role for CFHR1 in AMD pathogenesis. Moreover, our findings illustrate a distinct association of the CFHR1 alleles with AMD and other diseases, like atypical hemolytic uremic syndrome, which may provide insights into AMD pathogenesis. Further, we demonstrate the value of CFHR1 genotyping in prediction of AMD risk. 

AMD has a strong genetic component with a major susceptibility locus encompassing the CFH, CFHR3, and CFHR1 genes within the regulators of complement activation (RCA) gene cluster in 1q31. Strong linkage disequilibrium at this genomic region limits the genetic variability to a small set of extended CFH haplotypes, which includes one haplotype conferring increased risk and two haplotypes that protect from AMD. CFH 402His, CFH 62Ile and Δ _CFHR3-CFHR1 , respectively, are considered the genetic variations within each of these extended CFH haplotype responsible for the association with AMD. ²³ The CFH Tyr402His polymorphism, in particular, has been extensively studied from a functional point of view. These studies indicated that this amino acid substitution alters the binding specificity of factor H for different glycosaminoglycans ^19,20 and decreases its binding to retinal pigment epithelial cells ²¹ and Bruch's membrane. ²² However, the physiological relevance of these observations is still unclear. In contrast, the factor H 62Ile variant, showing increased complement regulatory activity, most likely confers a lower risk for AMD by reducing complement activation. ¹⁶  

In close proximity to CFH, within the extended CFH haplotypes, are CFHR1 and CFHR3, two genes that code for proteins showing sequence and structural homology to factor H (Fig. 1). CFHR3 and CFHR1 are both complement regulators that compete with factor H for binding to C3b. CFHR1 acts downstream of factor H and inhibits the C5-convertase and the lytic pathway, whereas CFHR3 acts as a cofactor for factor I in inactivating C3b. ^26,27 Based on the reported functional activities of CFHR1 and CFHR3 it seems paradoxical that the loss of these two complement regulators strongly protects against AMD. On the other side, it has been consistently found that Δ _CFHR3-CFHR1 is an independent protective factor for AMD, which suggest that CFHR3 and/or CFHR1 protein function confers risk for AMD. 

The concept that CFHR1 confers risk to AMD is further supported by our findings that CFHR1 encodes two major alleles, CFHR1*A and CFHR1*B, with different degrees of similarity with factor H, which associates differentially with disease. ²⁹ We have shown earlier that homozygosity for the CFHR1*B allotype, with greater sequence similarity to factor H, is strongly associated with atypical hemolytic uremic syndrome (aHUS) (OR, 2.54; 95% CI, 1.36–4.73; P = 0.0032) and suggests that increased competition between CFHR1*B and factor H decreases protection by factor H of cellular surfaces against complement damage. ²⁹ CFHR1*A, which is neutral in aHUS, has a strongly homozygous association with AMD (OR, 3.08; 95% CI, 1.92–4.92; P < 0.0001). The association of both CFHR1*B with aHUS and CFHR1*A with AMD seems to be dose dependent, as they disappear if only one allele is preset (CFHR1*B/CFHR1Δ _CFHR3-CFHR1 or CFHR1*A/CFHR1Δ _CFHR3-CFHR1 heterozygous). This indicates that the CFHR1Δ _CFHR3-CFHR allele is protective against both aHUS and AMD, even though homozygous CFHR1Δ _CFHR3-CFHR is increased in aHUS because of the association with anti-factor H autoantibodies. ^26,27 A last relevant observation from our association studies is that heterozygous CFHR1*A and CFHR1*B are not increased in AMD and aHUS. If we assume that CFHR1 proteins associate with disease because of competition with factor H, these data suggest that the CFHR1*A and CFHR1*B alleles may compete with distinct functional activities in factor H that are essential in the pathogenesis of one or the other disorder. Detailed functional analyses of the CFHR1*A and CFHR1*B alleles would be needed to determine whether this hypothesis is correct. 

Despite these uncertainties, our genetic analysis of CFHR1 in AMD suggests that models based on CFHR1 genotypes should provide meaningful predictive values for the development of AMD. The results prove this suggestion to be correct. In fact, our model of disease risk based on the CFHR1, CFB, and ARMS2 genotypes works well in assessing the predictability of AMD (Fig. 4) with AUC values of 0.78 and sensitivities and specificities of approximately 70%. These are significant values for a model based exclusively on genetic markers. Relatives of AMD patients and individuals with very early stage disease signs may benefit from knowing their genetic risk, as this may anticipate diagnosis and potential treatments. 

Conclusions and Limitations of the Study

The CFHR1*A allele is strongly associated with AMD. These novel data and previous studies in our laboratory ²⁹ demonstrate a differential association of the two major CFHR1 allotypes, CFHR1*A and CFHR1*B, with AMD and aHUS, respectively, which suggests that CFHR1 plays a relevant role in these pathologies. In fact, and even though CFHR1*A is in strong linkage disequilibrium with the CFH 402His allele, our data support that CFHR1*A is a strong candidate within the major risk haplotype at 1q32 for promoting AMD. However, our study has limitations. It included a relatively small case–control cohort, which prevented us from determining whether the CFHR1*A variant remained significantly associated with AMD after removing the CFH 402His–carrying haplotypes. Additional studies and further understanding of the functional activities of CFHR1, including the potential differences between the CFHR1*A and CFHR1*B allotypes, may help to determine whether the CFHR1 genotypes are more relevant to AMD than the CFH Tyr402His variant. Meanwhile, we have tested the usefulness of CFHR1 to generate models of disease risk and demonstrated that the CFHR1 genotypes provide meaningful data that are predictive of the development of AMD. 

Footnotes

 Supported by Grant SAF2008-00226 from the Spanish Ministerio de Ciencia e Innovación, the Ciber de Enfermedades Raras, and the Fundación Renal Iñigo Alvarez de Toledo (SRdeC); Grants RTICS RD07/0062, FIS PI 08/1705, and FIS 11/00898 from the Instituto de Salud Carlos III (AG-L); and Grant 35/2008 from the Comunidad Autónoma de Madrid, Dirección General de Innovación Tecnológica of the Consejeria Economía e Innovavión Tecnológica (JP-P).

Footnotes

 Disclosure: R. Martínez-Barricarte, P; S. Recalde, P; P. Fernández-Robredo, P; I. Millán, P; L. Olavarrieta, Secugen SL (E), P; A. Viñuela, Secugen SL (E), P; J. Pérez-Pérez, Secugen SL (E), P; A. García-Layana, Secugen SL (C), P; S. Rodríguez de Córdoba, Secugen SL (I, C), P

The authors thank the patients and their relatives for their participation in this study and the members of Secugen SL DNA sequencing and Diagnostics laboratories for invaluable technical assistance with the sequencing and genotyping. 

Appendix

Members of the Spanish Multicenter Group on AMD

Miguel Ángel Zapata, Hospital Vall d'Hebron, Barcelona; José María Ruiz-Moreno, Universidad Castilla-La Mancha, Albacete; Rosa Coco, Instituto de Oftalmolobiología Aplicada, Universidad de Valladolid, Valladolid; Lluís Arias, Hospital de Bellvitge, Barcelona; Clemencia Torrón and Oscar Ruiz-Moreno, Hospital Miguel Servet, Zaragoza; Henar Heras, Complejo Hospitalario de Navarra, Pamplona; María Isabel López-Gálvez, Hospital Clínico Universitario, Valladolid; Juan Donate, Hospital Clínico, Madrid; Miguel Ángel de la Fuente, Fundación Jiménez Díaz, Madrid; Ana María Gómez-Ramírez, Hospital Reina Sofía, Murcia; and Rosa Sanabría, Hospital San Telmo, Palencia.