CFH and ARMS2 Polymorphisms Interact with Zinc Supplements in Cognitive Impairment in the Women’s Health Initiative Hormone Trial

Abstract

Background:

An interaction between genetic variants in complement factor H (CFH) and age-related maculopathy susceptibility 2 (ARMS2) and high-dose zinc supplementation on progression to advanced age-related macular degeneration (AMD) exists. Because cognitive impairment (CI) is associated with AMD, we used data from the Women’s Health Initiative (WHI) to search for a zinc/genetics interaction.

Objective:

To study the interaction of chronic zinc supplementation with genetic variants in CFH and ARMS2 on the development of CI.

Background:

Zinc dietary supplements, CFH and ARMS2 genotypes, and serial mental status was analyzed in participants with available genetic data (n = 7,483). Cognition was assessed using the Modified Mini-Mental State Examination. The development of CI over 5 years was analyzed by genotype and zinc intake using a repeated measures logistic regression model.

Results:

Zinc supplementation of approximately 15 mg/day was associated with decreased development of CI in women with 1 or 2 CFH and no ARMS2 risk alleles (OR = 0.46: 1 CFH risk allele; 0.20: 2 CFH risk alleles; p = 0.002).

Conclusion:

Low-dose zinc (approximately 15 mg) is associated with reduced CI in women with 2 CFH and 0 ARMS2 AMD risk alleles. This interaction is opposite in direction to that observed in AMD, where patients with 2 CFH and 0 ARMS2 risk alleles had increased progression to neovascular AMD if treated with 80 mg/day of zinc. This may be due to a zinc dose-response or to a fundamental difference in the role of zinc in the progression of early CI versus advanced AMD.

Keywords

Cognitive dysfunction complement factor H gene-environment interaction genotype macular degeneration zinc

INTRODUCTION

Cognitive impairment (CI) and age-related macular degeneration (AMD) frequently occur together and share similar risk factors and some histopathological features [1]. Patients with reduced visual acuity due to AMD have an increased risk of Alzheimer’s disease (AD) (odds ratio (OR) = 2.88, 95% confidence interval, 1.75–4.76) and have amyloid-β deposits in areas of degenerating photoreceptors and retinal pigment epithelium cells [2, 3]. AMD risk is heavily influenced by genetics within genes coding classical members of the alternative complement cascade and other cell surface activators of complement. Individuals with one risk haplotype within complement factor H, an inhibitor of C3 convertase experience an OR of 2.63 for the development of advanced AMD with independent risk (OR = 2.81) conferred by each risk allele at the age-related maculopathy sensitivity (ARMS) 2 genes [4]. Such AMD-associated polymorphisms at rs1061170 and rs800292 in the complement factor H (CFH) gene also correlate with premorbid entorhinal cortex abnormalities and higher atrophy rates in AD [3, 5]. CFH AMD risk-allele carriers with AD have higher cerebrospinal fluid tau and amyloid-β levels, and experience deeper cognitive decline than non-risk carriers [5].

The Age Related Eye Disease Study (AREDS) found that daily supplementation with approximately 10×the recommended daily allowance (RDA) for zinc in women (80 mg), combined with copper (2 mg), β-carotene (15 mg), vitamin E (400 mg), and vitamin C (500 mg) moderately reduces progression from intermediate to advanced AMD [6 –11]. The Age-Related Eye Disease Study 2 (AREDS2) found that a dose of 25 mg of zinc per day was as effective as 80 mg/day for AMD progression prophylaxis [12], but did not evaluate lower doses of zinc. More recent analysis of AREDS data revealed that individuals with low CFH risk and high ARMS2 risk derive over twice the prophylactic benefit from AREDS supplements providing 80 mg/day of zinc, while individuals with high CFH risk and low ARMS2 risk derive no benefit and have an increased risk of progression to neovascular advanced AMD [11]

We investigated the interaction of CFH and ARMS2 gene polymorphisms with zinc supplementation on the development of CI using data from The Women’s Health Initiative Hormone Trial (WHI HT). We hypothesized that zinc supplementation may influence the risk of CI with pharmacogenetic determinants similar to those previously identified in AMD [11]. The WHI HT included longitudinal Modified Mini-Mental State Examination (3MS) scores and nutrient supplementation surveys allowing the identification of daily zinc supplementation levels [13, 14].

MATERIALS AND METHODS

Source of study subjects

Study data from the WHI HT [13] was provided through an investigator agreement with the NCBI database of Genotypes And Phenotypes (dbGAP) [15]. The WHI HT was sponsored by the National Heart, Lung, and Blood Institute (ClinicalTrials.gov Identifier: NCT00685009) and was a randomized, double blind, community-based, placebo-controlled clinical trial of postmenopausal women recruited from June 1995 to July 2002 (estrogen plus progestin versus placebo) or to February 2004 (estrogen-alone versus placebo) to investigate the effects of postmenopausal hormone replacement therapy (HRT) on human health. Subjects were between 50 and 79 years of age and did not have medical conditions that predicted a survival of less than three years or had characteristics predicted to diminish study adherence [13]. In total, 27,347 subjects were recruited by WHI HT investigators. We restricted our analysis to women for whom genotyping data and serial mental status examinations were available, leaving 7,700 women for the analysis of the interaction of zinc supplements and genetics on cognitive decline. This sample is believed to be representative of the study population as a whole.

Assessment of cognitive status

Cognitive function was determined by trained surveyors using the 3MS [16]. The 3MS score assesses attention, concentration, orientation, memory, language, constructional praxis, abstract thinking, and list generation, generating a score from 0 to 100. While cognitive function is a continuous variable, its measurement can fluctuate transiently. Scores <79 have been validated previously as consistent with cognitive impairment and were used to dichotomize mental status for the purpose of this analysis similar to a previous study of AMD patients [2, 17].

Supplement use history

During the WHI HT period of follow up the reported daily-ingested quantities of over 25 individual dietary supplements, including elemental zinc, were recorded. This information was provided as part of an investigator agreement by the National Institutes of Health, dbGAP [15].

Ethics and oversight

The WHI HT received ongoing oversight by a Data and Safety Monitoring Committee and the US National Institutes of Health and the Office for Human Research Protection [13]. Each study center (forty) provided independent IRB approval and elicited individual participant consent [13]. Ethics review of our analysis of genotype interaction with zinc supplements in cognitive impairment was provided by the University of Toronto, Ontario, Canada.

Genotype assignment and interpretation

Genotyping for study subjects was performed by WHI HT investigators using a combination of array-based, PCR and next-generation sequencing techniques (dbGAP Study accession: phs000200.v10.p3). Genotype information was available for 7,700 subjects at the ARMS2 AMD risk polymorphism (rs10490924), for the CFH AMD risk-associated SNPs (rs3766405 and rs412852) [7], and for 2 SNPs (rs7412 and rs429358) which determine APOE alleles [18].

CFH AMD risk-haplotype definition was described previously [7]. Individuals with homozygous CC at both CFH SNPs were considered high risk, those with heterozygous CT genotypes at each locus or those homozygous CC at rs3766405 and heterozygous CT at rs412852 were considered AMD intermediate risk, and all other genotype combinations at these 2 CHF loci were considered AMD low risk.

For the ARMS2 locus, patients with homozygous TT genotypes at rs10490924 were considered to have a high-risk AMD genotype (2 risk alleles), GT combinations were considered intermediate AMD risk (1 risk allele) and individuals with homozygous GG were low risk AMD (0 risk allele). For the purpose of analysis, ARMS2 intermediate and high risk were combined because of the relative rarity of homozygous risk alleles and considered collectively as the “high risk” group [7]. APOE alleles were determined though genotyping at rs7412 and rs429358 with the APOE allele defined as a haplotype with cytosine at each of these loci [19].

In our main model of CI risk, we considered allelic dosage modeling of the number of CFH risk haplotypes and binary ARMS2 status (combining 1 and 2 risk ARMS2 allele groups because of rarity of the risk allele in the population) resulting in 6 genetic groups which we refer to as C_XA_Y, where C refers to CFH where X is 0, 1, or 2 risk alleles and A represents ARMS2 where Y is 0 or 1/2 risk alleles.

Statistical analysis

7,700 women had both 3MS scores and genetic data available, 7,486 of which self-reported ‘White’ or ‘Hispanic/Latino’ as their ethnicity. This group was selected for analyses because AMD genetics have been studied primarily in these ethnic groups. Twenty-two women (0.29%) had missing Education status which was imputed using the majority group (“post-High School”). The status of APOE mutation could not be ascertained for 2 women (0.03%) and these were excluded from analysis. Out of 25,805 CI assessments available for this set of 7,486 women, 15 (0.06%) had missing 3MS score (including one subject with a singular observation) and these data points were dropped from analysis. After these exclusions, 7,483 subjects with 25,786 data points were available for analysis.

To examine the effect of zinc supplementation on CI we used a repeated-measures logistic regression model with CI as an outcome. In the main-effects model, we considered the risk of CI as functions of age (with cubic splines), HRT (dichotomous), education (5 levels), zinc supplementation (dichotomized to ≤7.5 mg /day (“un-supplemented”) or >7.5 mg/day (“supplemented”), CFH risk alleles (using allelic coding), and ARMS2 risk alleles (dominant coding) (Table 2). The gene×treatment model added the effect of interaction between Zinc and ARMS2 and CFH polymorphisms.

Table 1

Distribution of analytical variables within the studied subset of WHI HT subjects

Variable	Number (Percent of study population)
HRT:true	3,738 (50.0%)
CFH 0	2,502 (33.4%)
CFH 1	3,729 (49.8%)
CFH 2	1,252 (16.7%)
ARMS2 0	4,554 (60.8%)
ARMS2 1/2	2,929 (39.2%)
Zinc supplemented	3,065 (41.0%)
Zinc unsupplemented	4,418 (59.0%)
APOE4 (0/1/2 alleles)	5,617/1,744/122 (75/23/0.02%)

Table 2

No-interactions model: Average effects of main covariates on developing Cognitive Impairment, adjusted for education, age, APOE4, and HRT

Variable	OR	p
Zinc supplementation	0.70	0.02
CFH:		0.03
CFH: 1 versus 0 risk alleles	0.73	0.05
CFH: 2 versus 0 risk alleles	0.57	0.02
ARMS2 risk alleles (1 or 2 versus 0)	0.98	0.90

Zinc supplementation has an overall beneficial effect on the development of cognitive impairment (OR = 0.70). CFH AMD risk alleles were associated with a lower risk of CI. ARMS2 AMD risk alleles had no effect on the risk of CI.

Zinc supplementation status

Zinc intake (mg/day) was determined from micronutrient supplements survey data, and was collected without knowledge of individual CFH and ARMS2 genotypes (WHI form 45). No information was available on the temporal relationship between zinc and food intake. Dietary zinc intake was not considered in our analysis because of the confounding effects of recall bias. Micronutrient surveys were completed at approximately 3-year intervals during the study, though not always at the time of the 3MS score determination. To associate zinc intake with 3MS scores we used a heuristic algorithm, which prioritized the temporal proximity of zinc and 3MS score information, considering the determinations to be coupled if they were sampled within one year of each other. For more temporally separated zinc and 3MS scores, the algorithm computed a weighted average of 2 zinc intake determinations flanking the 3MS score assessment, giving twice the weight to the zinc intake reported prior to 3MS score. This averaging heuristic affected 119 (0.46%) of the observations.

Half of 25,786 recorded zinc intake values (longitudinal observations) were zero (12,951 or 50.2%). RDA-level supplementation (7.5 mg–15 mg/d) was reported in 39.1% of surveys. Supplementation levels >15 mg were rare and reported in only 8.2% of surveys. Zinc intake associated with each 3MS score was used to consider subjects as “supplemented” or “un-supplemented” (at a given assessment) for analysis, using a 7.5 mg/day threshold. Those taking at least 7.5 mg/day were considered “supplemented” while those reporting less than 7.5 mg of daily zinc consumption were considered “un-supplemented”. We performed initial analysis of potential confounding of zinc intake by other covariates using ANOVA models of per-subject mean zinc intake and an F-test.

Age and education

Age and education correlate strongly with the development of cognitive impairment as measured using 3MS scores [20]. Education levels were reported in 11 categories of increasing schooling years, which we re-categorized into 5 levels. These were included in subsequent multivariate models in an unrestricted fashion allowing for education level adjustment of 3MS scores. For each 3MS score we determined the age of the subject at time of cognitive assessment, using “Age at Screening” and “Days since Screening” information provided and rounded to the nearest year. The age effect was modeled using penalized natural cubic splines with 3 degrees of freedom using Generalized Additive Modeling approach [21]. Briefly, the age range is divided into a small number of panels, and within each panel age is allowed to take a cubic shape, with a restriction that overall shape of age effect across the whole range is continuous and smooth. The degree of smoothness is further controlled by penalizing the parameters underlying the individual cubic functions, forcing them to be smaller than they would be if no penalization was invoked. The continuity and smoothness restrictions, coupled with the amount of penalization determines the equivalent degrees of freedom utilized to model age, which we set here to be three. This allows the effect of age on risk of CI to be modeled as a smooth, non-linear curve, which is a more flexible and realistic assumption compared to usual linear model parameterization and increases the chance that any observed effect of zinc and genetics is properly controlled for potential age confounding.

To account for reported association between HRT and increased risk of CI, we utilized a binary HRT status variable (pooling estrogen-only and progesterone + estrogen arms together) in our model [20].

In our repeated-measures logistic regression model, CI (3MS score <79) was used as the outcome, age (using cubic spline expansion), 5-level education status, number of APOE risk alleles, and HRT status were considered as potential confounders. The number of CFH risk alleles, the number of ARMS2 risk alleles, and zinc intake (“un-supplemented” or “supplemented”) were the primary variables of interest. A time-to-event analysis, such as with Cox Proportional Hazards model, was considered but rejected due to long assessments intervals, the infrequency of assessments, and occasional reversion of mental status scores. The time-to-event analysis and repeated measures logistic regression allows us to perform classical and well-accepted statistical inference on assessing genetic effect modification of zinc supplementation on CI and make the results directly comparable to similar studies in AMD which also used repeated measures logistic regression [6] and Cox regression [11]. Our modeling was done using the “gam” function in R within an “mgcv” package [22] which allows for both repeated measures and non-linear age effects, while providing a conventional Generalized Linear Models hypothesis testing framework. Less conventional models such as Latent Growth Curve models focusing more on analysis of changes in cognitive function may offer an alternative approach [23, 24].

RESULTS

Data was initially available for 2,314 women treated with estrogen and progesterone, 1,427 treated with estrogen alone, and 3,745 receiving placebo. Three subjects were dropped due to missing data, as described above, leaving data on 7,483 subjects with a total of 25,786 longitudinal observations. Educational attainment categories were: 1) grade school (n = 425, 5.7%), 2) high school (n = 1,692, 22.6%), 3) some post-secondary education (n = 3,056, 40.8%), 4) university (n = 1,399, 18.7%), and 5) graduate or post-doctoral/professional (n = 911, 12.2%). The distribution of study variables within the WHI HT participants available for study is shown in Table 1.

The per-subject mean zinc intake did not differ significantly among education groups (p = 0.14), age-at-screening using a 3DF spline model (p = 0.74), CFH/ARMS2 genetic groupings (p = 0.133), or by APOE genotype (p = 0.144).

On average, 3.45 (median = 4) 3MS scores were obtained from each subject, over an average time range of 5.45 years (SD = 1.95) consistent with the duration of follow-up determinations within the WHI HT [13].

The distribution of genotype groups in the study population did not differ from that of a representative European population provided within the 1000 Genomes database (data not shown) [25].

Regression analysis

On average, across zinc supplementation groups, an increased number of CFH risk alleles was associated with a decreased risk of CI (p = 0.03). Compared to subjects with no CFH risk alleles, the OR for developing CI was 0.73 (p = 0.05) for those with one risk allele, and 0.57 (p = 0.02) for those with two risk alleles, showing an allelic dosage effect. On average, across genetic groups, zinc supplementation was associated with reduced risk for CI (OR 0.70, p = 0.023). The number of ARMS2 risk alleles had no significant impact on the risk of CI (Table 2).

We analyzed the genetic modification of zinc supplementation on the risk of developing CI by considering a model with interaction effects between zinc supplementation and genotype groups defined by CFH and ARMS2 risk allele number. We previously reported that genotype groups comprised of CFH and ARMS2 variants correlate with the treatment effect of zinc on progression from intermediate AMD to the neovascular form of advanced AMD. Individuals in some genotype groups experience an increased risk for progression to choroidal neovascular disease if using the AREDS dietary supplement compared to placebo [11].

We identified a statistical interaction between CFH and ARMS2 risk allele genotype groups and zinc supplementation (p = 0.028) in a model of CI. We then examined the association of zinc on the development of CI in 6 predefined genotype groups. We found that the overall beneficial zinc effect appears to be confined to those individuals who have at least one risk CFH risk haplotype and no ARMS2 risk alleles. In this genotype group, a significant reduction in CI risk was observed (OR = 0.456 for C₁A₀, OR = 0.206 for C₂A₀, p = 0.002). The C_XA_Y nomenclature is explained above in “Genotype Assignment and Interpretation (Table 3). For other genetic subgroups, zinc supplementation was not associated with either reduced or increased risk of CI.

Table 3

The effect of zinc supplementation on CI by genetic risk groups derived from our main model

	OR	95% CI Lower	95% CI Upper	p
C0A0	1.007	0.606	1.670	0.979
C1A0	0.456	0.274	0.756	0.002
C2A0	0.206	0.075	0.563	0.002
C0A1/2	0.727	0.364	1.453	0.357
C1A1/2	0.831	0.501	1.378	0.464
C2A1/2	0.949	0.405	2.223	0.902

The overall p value for the interaction of genetic risk group with zinc supplementation is 0.028 suggesting significant zinc effect modification by genetics. These results further suggest that significant reduction of the rate of CI is possible with zinc supplementation in individuals with at least 1 CFH risk allele and normal ARMS2 function.

The effects of known risk factors for CI and confounders were significant with the direction of effect as expected, with the exception of HRT, which was not significant (p = 0.087) (Table 4 and Fig. 1). The effects of education and age were both significant (p < 0.001 with 4 and 3 degrees of freedom, respectively) and showed a beneficial effect of higher education, and a detrimental effect of increasing age. The number of APOE4 alleles was also associated with increased risk of CI (OR = 2.80, p < 0.001).

Table 4

Effects of other risk factors in our main interaction model. The values are adjusted to the following base line: grade school education only, zero APOE4 alleles, and absence of HRT treatment

Variable	OR	p
Education (4DF)		<0.001
Edu: high school	0.42	0.001
Edu: post-High	0.42	<0.001
Edu: University	0.31	<0.001
Edu: post Graduate	0.23	<0.001
HRT	1.30	0.087
Age (3DF non-linear effect)		<0.001
APOE4 (per allele)	2.80	<0.001

Fig.1

Effect of Age on risk of CI using a natural cubic spline model with 3 degrees of freedom, adjusted for HRT, education, zinc use, and genetic effects.

DISCUSSION

Using data from the WHI HT, we have demonstrated that zinc supplementation at a dose of approximately 15 mg/day is associated with reduced risk of CI in a fashion that is heavily influenced by CFH and ARMS2 genotype groups previously shown to interact with high dose zinc supplementation (80 mg/day) in patients with AMD. While variants of CFH which are associated with increased AMD risk are themselves associated with reduced risk of CI (OR 0.57, p = 0.02), this effect is most marked in the presence of zinc supplementation. In the presence of intact complement pathway activator, ARMS2, women receiving zinc supplementation who have one CFH risk allele have half the observed risk of CI of unsupplemented women. In those with 2 CFH risk alleles, the observed OR for the development of CI is reduced by almost 80% (Table 3).

CFH influences the alternative pathway of complement activation by inhibiting the conversion of complement component 3 into its active form, C3b, which is a critical regulator of the cascade [26]. CFH also binds and inhibits monomeric C-Reactive Protein (mCRP) which is very proinflammatory when unbound. mCRP upregulates IL-8 and CCL2 levels in retinal pigment epithelial cells leading to tissue destruction. CFH variants associated with AMD have impaired binding to mCRP, and therefore these proinflammatory effects of mCRP remain unrestrained [27].

We demonstrate that variants of CFH and ARMS2 that maximize complement activation are associated with a decreased risk of cognitive decline, a benefit which seems augmented by zinc supplementation. Animal data suggests that complement activation can protect against amyloid-β-induced neurotoxicity and may reduce the accumulation of or may promote the clearance of amyloid from the midfrontal cortex and hippocampus [28, 29]. Inhibition of C3 activation in transgenic mice expressing a complement inhibitor results in increased amyloid-β deposition and neuronal degeneration [28]. Similarly, increased amyloid-β and fibrillar amyloid plaque burden were observed in a C3-/- mouse model of AD [29]. This suggests a beneficial role for complement C3 in plaque clearance as well as neuronal health, which may be important early in the development of cognitive impairment. Our results are consistent with these observations and suggest that early cognitive decline may be ameliorated through increased complement activation. In contrast, age related macular degeneration appears to be worsened by complement activation, possibly indicating different pathophysiology [27].

The ARMS2 gene is an important genetic mediator of AMD. It regulates complement by binding to damaged human cells, initiating complement activation by recruiting properdin [30]. ARMS2-properdin complexes increase C3b cell surface opsonization, promoting phagocytosis. The ARMS2 rs10490924 AMD risk allele results in absent protein expression, which may impede complement activation [31]. The combination of impeded complement inhibition by high CFH genetic risk in the absence of ARMS2 genetic risk may promote supra-normal complement activation, resulting in decreased development of cognitive decline among those women without obvious baseline impairment.

While this is the first report of the interaction of complement-regulatory gene polymorphisms with zinc supplementation and CI, previous work has demonstrated an interaction between progression of AMD and high dose zinc supplementation (80 mg/day) [7–9 , 11]. Individuals with 2 CFH risk alleles and 0 ARMS2 risk alleles derive no benefit from high-dose zinc supplementation as prophylaxis against AMD progression to the neovascular form of advanced AMD (nvAMD). In fact, such individuals have an increased risk of progression to nvAMD if treated with high-dose zinc [7–10 , 32]. These observations may indicate different inflammatory pathophysiology in AMD and cognitive decline or may be related to the substantially different levels of zinc supplementation evaluated in these groups. It is conceivable that for both diseases the lower doses of zinc supplementation typical of participants in the WHI HT (approximately 15 mg/day) are beneficial for those with 2 CFH and 0 ARMS2 risk alleles, while the 80 mg/day dose of zinc evaluated in the AREDS could be deleterious. This dose response hypothesis remains to be tested.

The work we present here describes the association of lower-dose zinc supplementation (>7.5 mg/day and approximately 15 mg/day) on the development of CI, while the AMD literature has examined the effect of much higher doses (80 mg/day). The influence of zinc on the complement system is concentration dependent, with low doses being associated with CFH oligomerization and precipitation, resulting in an increase in the activated form C3b. Higher zinc levels (100μM) causes C3b to precipitate, resulting in reduced complement activation [33]. These large zinc exposure differences prevent a direct comparison of the effects of zinc on risk of CI versus risk of AMD progression in those with 2 CFH risk alleles and 0 ARMS2 risk alleles.

The WHI HT is among the largest available longitudinal data sources documenting mental status, genetic variability, and use of nutritional supplements such as elemental zinc [20, 34]. The WHI HT prospectively recorded the formulations and quantities of micronutrient supplements consumed, making this a reliable and valuable resource for genetic epidemiological research in cognition.

We did not evaluate the impact of dietary zinc, nor do we know the temporal relationship between zinc supplementation and food intake, which may affect zinc bioavailability. It is possible that some women in WHI HT had zinc-deficient diets, and that the relatively low doses of zinc supplementation simply brought them to the RDA of zinc intake. Given that the recommended daily allowance (RDA) for zinc ranges from 8–15 mg/day, our analysis of WHI HT participants might demonstrate the importance adequate zinc intake, rather than a need for excess zinc supplementation. Participants in the AREDS were offered the option of taking a multivitamin that contained approximately 8 mg of zinc, and two-thirds of AREDS participants opted to take the multivitamin through the course of the study. Therefore, most AREDS participants randomized to non-zinc containing treatments were taking as much zinc as the WHI HT participants who we considered to be zinc “supplemented”.

In our analysis, we show that low dose zinc supplementation helps slow or prevent CI in general, which could have important public health consequences. This effect is stronger in women with 1 or 2 CFH risk alleles and 0 ARMS2 risk alleles, suggesting a potential for personalized approaches to nutritional prophylaxis. Further studies are indicated to validate these findings, to evaluate a potential dose-dependent effect of zinc supplementation, and to study these issues in men. We present level IIB evidence for the use of zinc as a prophylaxis against cognitive decline [35]. A prospective evaluation of this pharmacogenetic phenomenon might provide evidence to support public health policy recommendations.

Footnotes

ACKNOWLEDGMENTS

The Women’s Health Initiative (WHI) program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts N01WH22110, 24152, 32100-2, 32105-6, 32108-9, 32111-13, 32115, 32118-32119, 32122, 42107-26, 42129-32, and 44221. This manuscript was not prepared in collaboration with investigators of the WHI, has not been reviewed and/or approved by the WHI, and does not necessarily reflect the opinions of the WHI investigators or the NHLBI. No financial support was provided to the authors for the conduct of this study.

Authors’ disclosures available online ().

References

Kaarniranta

, Salminen

, Haapasalo

, Soininen

, Hiltunen

(2011) Age-related macular degeneration (AMD): Alzheimer’s disease in the eye? J Alzheimers Dis 24, 615–631.

Clemons

, Rankin

, McBee

(2006) Cognitive impairment in the Age-Related Eye Disease Study: AREDS report no. 16. Arch Ophthalmol 124, 537–543.

Chouraki

, Seshadri

(2014) Genetics of Alzheimer’s disease. Adv Genet 87, 245–294.

Fritsche

, Igl

, Bailey

, Grassmann

, Sengupta

, Bragg-Gresham

, Burdon

, Hebbring

, Wen

, Gorski

, Kim

, Cho

, Zack

, Souied

, Scholl

, Bala

, Lee

, Hunter

, Sardell

, Mitchell

, Merriam

, Cipriani

, Hoffman

, Schick

, Lechanteur

, Guymer

, Johnson

, Jiang

, Stanton

, Buitendijk

, Zhan

, Kwong

, Boleda

, Brooks

, Gieser

, Ratnapriya

, Branham

, Foerster

, Heckenlively

, Othman

, Vote

, Liang

, Souzeau

, McAllister

, Isaacs

, Hall

, Lake

, Mackey

, Constable

, Craig

, Kitchner

, Yang

, Su

, Luo

, Chen

, Ouyang

, Flagg

, Lin

, Mao

, Ferreyra

, Stark

, von Strachwitz

, Wolf

, Brandl

, Rudolph

, Olden

, Morrison

, Morgan

, Schu

, Ahn

, Silvestri

, Tsironi

, Park

, Farrer

, Orlin

, Brucker

, Li

, Curcio

, Mohand-Said

, Sahel

, Audo

, Benchaboune

, Cree

, Rennie

, Goverdhan

, Grunin

, Hagbi-Levi

, Campochiaro

, Katsanis

, Holz

, Blond

, Blanche

, Deleuze

, Igo

Jr. , Truitt

, Peachey

, Meuer

, Myers

, Moore

, Klein

, Hauser

, Postel

, Courtenay

, Schwartz

, Kovach

, Scott

, Liew

, Tan

, Gopinath

, Merriam

, Smith

, Khan

, Shahid

, Moore

, McGrath

, Laux

, Brantley

Jr. , Agarwal

, Ersoy

, Caramoy

, Langmann

, Saksens

, de Jong

, Hoyng

, Cain

, Richardson

, Martin

, Blangero

, Weeks

, Dhillon

, van Duijn

, Doheny

, Romm

, Klaver

, Hayward

, Gorin

, Klein

, Baird

, den Hollander

, Fauser

, Yates

, Allikmets

, Wang

, Schaumberg

, Klein

, Hagstrom

, Chowers

, Lotery

, Leveillard

, Zhang

, Brilliant

, Hewitt

, Swaroop

, Chew

, Pericak-Vance

, DeAngelis

, Stambolian

, Haines

, Iyengar

, Weber

, Abecasis

, Heid

(2016) A large genome-wide association study of age-related macular degeneration highlights contributions of rare and common variants. Nat Genet 48, 134–143.

Zhang

, Li

, Wu

, Cui

, Bi

, Zhou

, Wang

, Zhang

, Wang

, Kong

, Li

, Fang

, Jiang

, Yao

(2016) CFH variants affect structural and functional brain changes and genetic risk of Alzheimer’s disease. Neuropsychopharmacology 41, 1034–1045.

(2001) A randomized, placebo-controlled, clinical trial of high-dose supplementation with vitamins C and E, beta carotene, and zinc for age-related macular degeneration and vision loss: AREDS report no. 8. Arch Ophthalmol 119, 1417–1436.

Awh

, Hawken

, Zanke

(2015) Treatment response to antioxidants and zinc based on CFH and ARMS2 genetic risk allele number in the Age-Related Eye Disease Study. Ophthalmology 122, 162–169.

Awh

, Lane

, Hawken

, Zanke

, Kim

(2013) CFH and ARMS2 genetic polymorphisms predict response to antioxidants and zinc in patients with age-related macular degeneration. Ophthalmology 120, 2317–2323.

Seddon

, Silver

, Rosner

(2016) Response to AREDS supplements according to genetic factors: Survival analysis approach using the eye as the unit of analysis. Br J Ophthalmol 100, 1731–1737.

10.

Klein

, Francis

, Rosner

, Reynolds

, Hamon

, Schultz

, Ott

, Seddon

(2008) CFH and LOC387715/ARMS2 genotypes and treatment with antioxidants and zinc for age-related macular degeneration. Ophthalmology 115, 1019–1025.

11.

Vavvas

, Small

, Awh

, Zanke

, Tibshirani

, Kustra

(2018) CFH and ARMS2 genetic risk determines progression to neovascular age-related macular degeneration after antioxidant and zinc supplementation. Proc Natl Acad Sci U S A 115, E696–E704.

12.

(2013) Lutein + zeaxanthin and omega-3 fatty acids for age-related macular degeneration: The Age-Related Eye Disease Study 2 (AREDS2) randomized clinical trial. JAMA 309, 2005–2015.

13.

(1998) Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group. Control Clin Trials 19, 61–109.

14.

Shumaker

, Legault

, Rapp

, Thal

, Wallace

, Ockene

, Hendrix

, Jones

, Assaf

, Jackson

, Kotchen

, Wassertheil-Smoller

, Wactawski-Wende

(2003) Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in postmenopausal women: The Women’s Health Initiative Memory Study: A randomized controlled trial. JAMA 289, 2651–2662.

15.

Women’s Health Initiative Clinical Trial and Observational Study SHARe, https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000200.v1.p1.

16.

Teng

, Chui

(1987) The Modified Mini-Mental State (3MS) examination. J Clin Psychiatry 48, 314–318.

17.

Assel

, Li

, Wang

, Allen

, Baggerly

, Vickers

(2018) Genetic polymorphisms of CFH and ARMS2 do not predict response to antioxidants and zinc in patients with age-related macular degeneration: Independent statistical evaluations of data from the Age-Related Eye Disease Study. Ophthalmology 125, 391–397.

18.

Seripa

, D’Onofrio

, Panza

, Cascavilla

, Masullo

, Pilotto

(2011) The genetics of the human APOE polymorphism. Rejuvenation Res 14, 491–500.

19.

Corder

, Saunders

, Strittmatter

, Schmechel

, Gaskell

, Small

, Roses

, Haines

, Pericak-Vance

(1993) Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261, 921–923.

20.

Shumaker

, Legault

, Kuller

, Rapp

, Thal

, Lane

, Fillit

, Stefanick

, Hendrix

, Lewis

, Masaki

, Coker

(2004) Conjugated equine estrogens and incidence of probable dementia and mild cognitive impairment in postmenopausal women: Women’s Health Initiative Memory Study. JAMA 291, 2947–2958.

21.

Hastie

, Tibshirani

(1990) Generalized Additive Models, Chapman and Hall/CRC.

22.

Wood

(2011) Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J R Stat Soc Series B Stat Methodol 73, 3–36.

23.

Robitaille

, Garcia

, McIntosh

(2015) Joint trajectories of cognitive functioning and challenging behavior for persons living with dementia in long-term care. Psychol Aging 30, 712–726.

24.

Duncan

, Duncan

(2009) The ABC’s of LGM: An introductory guide to latent variable growth curve modeling. Soc Personal Psychol Compass 3, 979–991.

25.

Abecasis

, Auton

, Brooks

, DePristo

, Durbin

, Handsaker

, Kang

, Marth

, McVean

(2012) An integrated map of genetic variation from 1,092 human genomes. Nature 491, 56–65.

26.

Clark

, Bishop

(2018) The eye as a complement dysregulation hotspot. Semin Immunopathol 40, 65–74.

27.

Molins

, Fuentes-Prior

, Adan

, Anton

, Arostegui

, Yague

, Dick

(2016) Complement factor H binding of monomeric C-reactive protein downregulates proinflammatory activity and is impaired with at risk polymorphic CFH variants. Sci Rep 6, 22889.

28.

Wyss-Coray

, Yan

, Lin

AH-T

, Lambris

, Alexander

, Quigg

, Masliah

(2002) Prominent neurodegeneration and increased plaque formation in complement-inhibited Alzheimer’s mice. Proc Natl Acad Sci U S A 99, 10837–10842.

29.

Maier

, Peng

, Jiang

, Seabrook

, Carroll

, Lemere

(2008) Complement C3 deficiency leads to accelerated amyloid beta plaque deposition and neurodegeneration and modulation of the microglia/macrophage phenotype in amyloid precursor protein transgenic mice. J Neurosci 28, 6333–6341.

30.

Tan

, Zhao

, Zhu

, Wu

, Li

, Shen

, Lei

, Chen

, Peng

(2018) The Akkermansia muciniphila is a gut microbiota signature in psoriasis. Exp Dermatol 27, 144–149.

31.

Fritsche

, Loenhardt

, Janssen

, Fisher

, Rivera

, Keilhauer

, Weber

(2008) Age-related macular degeneration is associated with an unstable ARMS2 (LOC387715) mRNA. Nat Genet 40, 892–896.

32.

, Reynolds

, Fagerness

, Rosner

, Daly

, Seddon

(2011) Association of variants in the LIPC and ABCA1 genes with intermediate and large drusen and advanced age-related macular degeneration. Invest Ophthalmol Vis Sci 52, 4663–4670.

33.

Nan

, Tetchner

, Rodriguez

, Pao

, Gor

, Lengyel

, Perkins

(2013) Zinc-induced self-association of complement C3b and Factor H: Implications for inflammation and age-related macular degeneration. J Biol Chem 288, 19197–19210.

34.

Rapp

, Espeland

, Hogan

, Jones

, Dugan

(2003) Baseline experience with Modified Mini Mental State Exam: The Women’s Health Initiative Memory Study (WHIMS). Aging Ment Health 7, 217–223.

35.

Simon

, Paik

, Hayes

(2009) Use of archived specimens in evaluation of prognostic and predictive biomarkers. J Natl Cancer Inst 101, 1446–1452.