DeAngelis MM, Ji F, Kim IK, Adams S, Capone A, Ott J, Miller JW, Dryja TP. Cigarette Smoking, CFH, APOE, ELOVL4, and Risk of Neovascular Age-Related Macular Degeneration. Arch Ophthalmol. 2007;125(1):49-54. doi:10.1001/archopht.125.1.49
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
To examine if the genes encoding complement factor H (CFH), apolipoprotein E (APOE), and elongation of very-long-chain fatty acids–like 4 (ELOVL4) confer risk of neovascular age-related macular degeneration (AMD) in an independent or interactive manner when controlling for smoking exposure.
We studied 103 unrelated patients with neovascular AMD who each had at least 1 sibling with normal maculae. Smoking histories were obtained. Genotyping was performed by analyzing amplified genomic fragments from CFH, APOE, and ELOVL4 by direct sequencing or by restriction endonuclease digests. Conditional logistic regression analysis was used to build a multifactor model.
For CFH, only the CC genotype carried a statistically significant elevation of disease risk (odds ratio, 49.37; 95% confidence interval, 6.20-393.22; P<.001). No significant association was observed between neovascular AMD and APOE or ELOVL4. No significant interactions were found between smoking and having the CFH or APOE genotype nor were significant interactions found between the CFH, ELOVL4, and APOE genotypes.
Smoking and having the CFHCC genotype independently increase risk of neovascular AMD. APOE and ELOVL4 genotypes do not seem to modify risk.
Smoking 10 pack-years or more and having the CFHCC genotype increase one's risk of neovascular AMD 144-fold compared with smoking less than 10 pack-years and having the CT or TT genotype.
Age-related macular degeneration (AMD) is a common form of legal blindness, with 1.75 million US citizens having advanced AMD in at least 1 eye.1 Approximately 10% of the population aged 43 years and older are affected with some form of the disease, and 30% of the population aged 75 years and older are affected.2 It is estimated that 2.95 million individuals in the United States will have advanced AMD by 2020.1
Previous findings of social, medical, genetic, environmental, and constitutional factors have not been in agreement as to predictors of AMD. Cigarette smoking seems to be the only epidemiological risk factor generally accepted as being associated with an increased risk of AMD,3 while an allelic variant in the complement factor H gene (CFH) seems to be the strongest genetic risk factor. Several independent articles have shown the polymorphism T1277C in CFH to be associated with increased risk of early and late stages of AMD (neovascular and geographic atrophy).4- 11 These findings suggest that the C allele, or disease allele, contributes to almost half of all AMD cases. Some studies12- 14 have demonstrated that the ε4 allele of the apolipoprotein E gene (APOE) decreases risk of dry or neovascular AMD, but another study15 did not support this finding. In addition, a recent article suggests that the A895G variant of the elongation of the very-long-chain fatty acids–like 4 gene (ELOVL4) may increase risk of neovascular AMD,9 although an earlier study16 did not find this association. Given the multifactorial and heterogeneous nature of AMD, methods that detect weak to moderate associations may be required to identify the contribution that smoking and genetic risk factors such as those associated with CFH, APOE, and ELOVL4 make independently and in combination to risk of AMD. This information may help to more accurately determine the overall risk of neovascular AMD.
Siblings who are discordant for a quantitative trait tend not to share alleles at genetic loci that govern that trait. Sets of siblings who are extremely discordant for a quantitative trait can be a powerful resource for identifying those genes. The prototype of an extreme sibpair consists of one member with a trait value in the top 10% and the other member in the bottom 10% of the population distribution, although pairs in the top 10% and the bottom 30% can be almost as valuable.17,18 Such sibpairs are more likely than pairs of siblings with intermediate trait values to differ at many of the multiple genetic loci that govern a multifactorial trait.
Using such types of sibpairs to search for epidemiological risk factors associated with the neovascular form of AMD, it has been demonstrated that smoking is associated with increased risk of neovascular AMD,19 in support of many previous studies. In the study reported herein, using a phenotypically well-characterized and documented cohort of extremely discordant sibpairs, we examined the roles of CFH, APOE, and ELOVL4 independently and together with smoking history.
The protocol was reviewed and approved by the institutional review board at Massachusetts Eye and Ear Infirmary, Boston, Mass, and the William Beaumont Hospital, Royal Oak, Mich, and conforms to the tenets of the Declaration of Helsinki. Eligible patients were enrolled in this study after they gave informed consent in person, through the mail, or over the telephone, before responding to a standardized questionnaire and donating 10 to 50 mL of venous blood.
Index patients with neovascular AMD were recruited from the retina service of the Massachusetts Eye and Ear Infirmary and from Associated Retinal Consultants, PC, William Beaumont Hospital. Details of the recruitment and of the clinical description of the patients are given elsewhere.19 In brief, all index patients had the neovascular form of AMD in at least 1 eye, defined by fibrosis, subretinal hemorrhage, or fluorescein angiographic presence of neovascularization documented at the time of or before enrollment in the study (AMD level 4b on the Age-Related Eye Disease Study scale).20 The unaffected siblings had normal maculae at an age older than that at which the index patient was first diagnosed as having neovascular AMD. Normal maculae (defined as the zone centered at the foveola and extending 2 disc diameters, or 3000 μm, in radius) fulfilled the following criteria: no pigment abnormalities, no geographic atrophy, 0 to 5 small drusen (all <63 μm in diameter), and no neovascularization as defined previously20- 22 (AMD level 0-1 on the Age-Related Eye Disease Scale). Disease status of every participant was confirmed by at least 2 of the investigators (I.K.K., J.W.M., or T.P.D.) by evaluation of fundus photographs or fluorescein angiograms, except when an investigator directly examined an unaffected sibling during a home visit (6 subjects).
We administered a standardized questionnaire to all eligible participants in person or over the telephone to ascertain smoking exposure, with the age of the index patient at the time of the fundus photographs as our cutoff reference age for smoking exposure for all members in a sibship. In most cases, the diagnosis of AMD was made simultaneously with the diagnosis of neovascular AMD. If a participant ever smoked, we recorded the age at which he or she started smoking, the age at which the participant quit smoking (if he or she had quit), and the number of packs of cigarettes smoked per day, on average. Based on the responses, the number of pack-years of cigarettes smoked was calculated for each smoker. Participants who had smoked fewer than 100 cigarettes during their lifetime (ie, <1/73 of a pack-year) were categorized as having never smoked. A pack-year was defined as 1 pack of cigarettes per day for 1 year, with 1 pack defined as 20 cigarettes. For our statistical analysis, the reference cutoff for smoking was defined as 10 pack-years or more vs fewer than 10 pack-years. With this cutoff, our subjects were divided into 2 approximately equal groups (Figure).
Leukocyte DNA was purified using standard phenol-chloroform or DNAzol (Invitrogen Corporation, Carlsbad, Calif) extraction protocols. Previously reported oligonucleotide primers based on the flanking intron sequences of exon 9 for CFH, exon 4 for APOE, and exon 6 for ELOVL4 were selected.6,23,24 The primer pairs were as follows (sense primer, antisense primer, both written in the 5"-3" direction): CFH: GGTTTCTTCTTGAAAATCACAGG, CCATTGGTAAAACAAGGTGACA; APOE: TAAGCTTGGCACGGCTGTCCAAGGA, ACAGAATTCGCCCCGGCCTGGTACAC; and ELOVL4: GTTGTTAAAAGTTGTTTACTATTC, TCAACAACAGTTAAGGCCCAGTTC.
For all 3 genes, polymerase chain reaction (PCR) was used to amplify genomic DNA fragments from 20 ng of leukocyte DNA in a solution of 10× PCR buffer containing 25mM magnesium chloride, 0.2mM cytidine 5-triphosphate, 0.2mM deoxyadenosine triphosphate, 0.2mM deoxyguanosine triphosphate, 0.2mM deoxythrombotic thrombocytopenic purpura, and 0.5 U of Taq DNA polymerase (USB Corporation, Cleveland, Ohio). For APOE and ELOVL4, 5 M of betaine (Sigma-Aldrich Inc, St Louis, Mo) was added to each PCR. The temperatures used during PCR were as follows: for APOE, 95°C for 5 minutes, followed by 35 cycles at 66°C for 30 seconds, 72°C for 30 seconds, and 95°C for 30 seconds, with a final annealing at 66°C for 1.5 minutes and extension at 72°C for 5 minutes; and for CFH and ELOVL4, 95°C for 5 minutes, followed by 35 cycles at 56°C for 30 seconds, 72°C for 30 seconds, and 95°C for 30 seconds, with a final annealing at 56°C for 1.5 minutes and extension at 72°C for 5 minutes. For sequencing reactions, PCR products were digested according to the manufacturer's protocol using ExoSAP-IT (USB Corporation) and were then subjected to a cycle sequencing reaction using the Big Dye Terminator version 3.1 cycle sequencing kit (Applied Biosystems, Foster City, Calif) according to the manufacturer's protocol. Products were purified using 96-well plates (Performa DTR Ultra; Edge Biosystems, Gaithersburg, Md) to remove excess dye terminators. Samples were sequenced on a DNA sequencer (ABI Prism 3100; Applied Biosystems). Electropherograms generated from the DNA sequencer were analyzed using Lasergene DNA and protein analysis software (DNASTAR, Inc, Madison, Wis). All patients were sequenced in the forward direction (5′-3′), unless variants, mutations, or polymorphisms were identified, in which case confirmation was obtained in some instances by sequencing in the reverse direction.
Conditional logistic regression (CLR) analysis (SAS version 8.0; SAS Institute Inc, Cary, NC) was performed to identify factors associated with neovascular AMD. Potential risk factors of interest, as already defined, were evaluated initially 1 at a time. A multiple CLR model was built using those factors from the single-factor model that seemed to be associated with neovascular AMD at P ≤.12. Nonsignificant variables were then dropped from the multiple CLR model to create the most parsimonious statistical model predictive of neovascular AMD. Using the McNemar test, we performed a separate analysis based on only 1 extremely discordant sibpair per family. We chose the unaffected sibling most discordant in age (the oldest) from the affected member (index patient) in sibships in which there was more than 1 discordant pair. The discordant sibpairs analyzed by the McNemar test were also used to calculate the genotype and allele frequencies for CFH, APOE, and ELOVL4 in affected siblings and, separately, in unaffected siblings. For our 1:1 matched discordant sibpair data, we computed the population attributable risk as θ[(RR−1)/RR], where the relative risk on us is RR is approximated by the odds ratio (OR),25 and θ is the proportion of cases exposed to the factor.
We recruited 143 extremely discordant sibpairs derived from 103 sibships. Seven of 103 sibships had more than 1 affected member, and 19 had more than 1 unaffected member. All subjects were 50 years old or older on entrance into the study. As described in the “Methods” section, to ascertain smoking exposure, the mean reference ages of affected and unaffected subjects were calculated based on the date of neovascular AMD diagnosis of the affected sibling. The mean ± SD age of 115 affected siblings was 71.7 ± 7.6 years (age range, 41.3-90.9 years), and the age of 127 unaffected siblings was 72.8 ± 8.8 years (age range, 49.0-86.4 years). Forty percent of the unaffected siblings were male, and 43% of the matching affected siblings were male. All participants were white.
Single-factor CLR analysis demonstrated associations at the P <.12 significance level between neovascular AMD and having at least 1 APOE ε4 allele, being homozygous or heterozygous for the CFHC (His) allele, and smoking 10 pack-years or more (Table 1). No significant association was observed between neovascular AMD and having the APOE ε2 or APOE ε3 allele in single-factor analysis. We found no statistically significant association between the ELOVL4G allele for the homozygous (GG) or heterozygous genotype and risk of neovascular AMD. There was no association found with respect to sex (data not shown).
Multiple CLR analyses were conducted to determine which risk factors might be independently associated with neovascular AMD. Our multiple CLR model was done in 2 ways (Table 2). For model A, we examined (1) smoking 10 pack-years or more vs fewer than 10 pack-years, (2) having the APOE ε4 allele in the heterozygous (ε2ε4 or ε3ε4) or homozygous (ε4ε4) state vs having no ε4 allele, and (3) having the CFHC (His) allele in the homozygous state (CC) vs having no C allele (TT) or having the C allele in the heterozygous state (CT). Being homozygous for the CFHC allele was significantly associated with neovascular AMD (OR, 49.37; 95% CI, 6.20-393.22; P<.001). Smoking 10 pack-years or more became more significant than in the single-factor analysis (OR, 2.92; 95% CI, 1.41-6.03; P = .004). The presence of at least 1 APOE ε4 allele was not significantly associated with neovascular AMD in the multiple CLR model (OR, 0.71; 95% [confidence interval] CI, 0.29-1.74; P = .45). For model B, we examined (1) having at least 1 APOE ε4 allele, (2) smoking 10 pack-years or more vs fewer than 10 pack-years, and (3) having the CFHC (His) allele separately in the homozygous or heterozygous state vs having no C allele. In this model, having the CFHC (His) allele in the homozygous state was statistically significant (OR, 96.18; 95% CI, 9.98-926.82; P<.001), along with smoking 10 pack-years or more, which (as in model A) became more significant than in the single-factor analysis (OR, 2.95; 95% CI, 1.41-6.15; P = .004). As in model A, the findings in model B suggested that the presence of at least 1 APOE ε4 allele was not significantly associated with risk of neovascular AMD (OR, 0.65; 95% CI, 0.26-1.65; P = .37). Although the ORs for the APOE ε4 allele in model A and model B (0.71 and 0.65, respectively) indicated a trend toward a protective effect with respect to risk of neovascular AMD, this effect was not statistically significant.
After introducing an interaction term into model A, further analysis suggested no interaction between smoking and having the CFHCC genotype, which suggests that these 2 factors affect risk of having neovascular AMD independently. Therefore, the effect of both factors is the product of the risk of the individual factors (ie, a multiplicative effect). For example, homozygotes with the CFHC allele who smoke 10 pack-years or more, compared with individuals with the CFH CT or TT genotype who smoke fewer than 10 pack-years, have a 144-fold risk of neovascular AMD.
Similar to the results from the single-factor analyses, we found using the McNemar test that the CFHCC genotype was much more highly associated with risk of neovascular AMD than the CT genotype. For example, when the CC genotype was compared with the CT or TT genotype (Table 3), there were 26 informative sibpairs (counting only 1 sibpair per family). In 25 of these 26 sibpairs, the affected sibling had the CC genotype (P<.001). In contrast, when individuals with the CC or CT genotype were combined and were compared with those with the TT genotype, there was a much less striking association of neovascular AMD with the CC or CT genotype (P = .03). The McNemar test also showed no significant association between the APOE ε4 allele and neovascular AMD or between ELOVL4 genotypes and neovascular AMD. With regard to ELOVL4, other previously reported variations16 in exon 6 were found, but the number of informative sibpairs was too small for an appropriate statistical analysis. For example, only 1 of 3 sibpairs was discordant for the Ile267Thr variation; the affected member of the sibpair had the sequence change, while the unaffected member did not. The affected member of 1 sibpair exclusively had the Glu272Gln variant, while the unaffected members of 2 other sibpairs exclusively had this sequence change. Both members of 1 sibpair had the Asn284Asp change.
The genotype and allele frequencies for CFH, APOE, and ELOVL4 in affected and unaffected siblings are given in Table 4. No significant deviations from Hardy-Weinberg equilibrium for any of the genotypes studied in CFH, APOE, and ELOVL4 were observed in the affected or unaffected sets of siblings. When testing for significant departures from Hardy-Weinberg equilibrium, 1 df was used for the 3 genotypes of CFH and ELOVL4, while 3 df were used for the 6 genotypes of APOE.
The population attributable risk for smoking and having the CFH genotypes is summarized in Table 5. Risk conferred by having the CFHCC genotype is 36%, and risk conferred by having the CFHCC or CT genotype is 49%. Smoking 10 pack-years or more explains 28% of risk in the total population. Because smoking and having the CFHCC or CT genotype are not mutually exclusive, the population attributable risk of smoking 10 pack-years or more and having the CFHCC or CT genotype is 56%.
No interaction was found between smoking and having the CFHCC genotype; therefore, the effect of both factors is multiplicative to the risk of neovascular AMD.11 Thirteen of 14 sibships in which both the affected and unaffected siblings had the CFHCC genotype (Table 3) were also concordant for their smoking history, suggesting that other risk factors are yet to be discovered. The single sibship in which both siblings had the CFHCC genotype and were not concordant for their smoking history, the affected sibling had smoked 46 pack-years, while the unaffected sibling had smoked 2 pack-years. Only 1 sibship in our extremely discordant sibpair cohort had an unaffected sibling with the CFHCC genotype and an affected sibling with the CFHCT genotype. This sibpair had discordant smoking histories; the affected sibling had smoked 34 pack-years, while the unaffected sibling had smoked 9 pack-years.
Although having at least 1 CFH C allele (CT or CC) compared with having no CFHC allele (TT) was significantly associated with neovascular AMD in single-factor analysis (Table 1, P = .02), this result was due to the highly significant effect of the CC genotype. Compared with individuals who had no CFHC allele (TT), individuals who were heterozygous for the CFHC allele (CT) had no significant increase in risk of neovascular AMD in multiple CLR analyses (Table 2, model B). These results suggest that CFHCT heterozygotes are not at increased risk of neovascular AMD. This is in contrast to what other groups have reported for neovascular AMD.6,9- 11
The association of CFH alleles with neovascular AMD suggests a role for inflammation and innate immunity in the pathophysiology of the disease. The association with cigarette smoking lends support to a role for oxidative stress and injury. Although our analysis demonstrated that these 2 risk factors are independent of each other, they may work through similar mechanisms to increase risk of neovascular AMD.26
In summary, our analyses of extremely discordant sibpairs provide evidence that having the CFH CC genotype and smoking 10 pack-years or more are independent risk factors for neovascular AMD. APOE and ELOVL4 genotypes do not modify the risk, in agreement with what others have reported.15,16 Smokers of 10 pack-years or more who also have the CFH CC genotype have approximately a 144-fold increase in disease risk compared with individuals who smoke fewer than 10 pack-years and have the CFHCT or TT genotype.
Correspondence: Margaret M. DeAngelis, PhD, Department of Ophthalmology, Harvard Medical School, and Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114 (email@example.com).
Submitted for Publication: June 30, 2006; final revision received September 25, 2006; accepted September 26, 2006.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grants from the Lincy Foundation, Massachusetts Lions, Ruth and Milton Steinbach Fund, Friends of the Massachusetts Eye and Ear Infirmary, and Genetics of Age-Related Macular Degeneration Fund; Research to Prevent Blindness; as well as by grants EY014458, EY14104, and HG00008 from the National Institutes of Health.
Acknowledgments: We thank the patients and their siblings for their participation in this study. We also thank Amanda J. Harring and Tammy Osentoski for their recruitment efforts. Finally, we thank Anne Marie Lane, MPH, for critical and helpful discussions.