Genetics and Patients’ Response to AMD Nutritional Supplements

One size does not fit all!

By Carl C. Awh, MD

My colleagues and I recently published what we believe to be compelling evidence that almost one out of seven individuals with moderate age-related macular degeneration in the Age-Related Eye Disease Study (AREDS) were more likely to have been harmed than helped by the AREDS formulation.1 With this knowledge, what should we do when counseling patients about nutritional supplements for AMD? In this article, I review the evidence that easily measured markers of AMD genetic risk predict the response to components of the AREDS formulation and that a personalized approach to supplement therapy could lead to improved outcomes for patients with AMD.


Although there is disagreement regarding the value of genetic testing to better understand an individual’s risk of progression to advanced AMD, there is no argument about the validity of identified genetic risk markers for AMD.

AMD is the most genetically influenced of any human multigenic disease. Genetic risk marker identification has provided an unprecedented ability to calculate the risk of developing AMD. Because each risk marker acts independently, genetic risk in an individual having more than one marker can be estimated by the summed probabilities of the individual risk markers. Given the number and magnitude of known risk markers, the resultant summed risk score can vary tremendously. For example, an individual with two unfavorable copies of the complement factor H (CFH), age-related maculopathy susceptibility 2 (ARMS2) and complement component 3 (C3) genes has a 400-fold increased risk (odds ratio) of developing advanced AMD compared to an individual without any of these risk markers. The degree to which genetics contributes to the risk of AMD has not been demonstrated in any other common human disease of aging (diabetes, heart disease, hypertension, or stroke). Physician-ordered tests for predicting AMD risk use genetic risk markers as data for algorithms that also incorporate nongenetic factors such as body mass index, smoking history, age; and, most importantly, the clinical stage of AMD.2


About 5 years ago I was invited to become a consultant to ArcticDx, a company with a prognostic AMD test based upon genetic markers. Although I knew little about the genetics of AMD, it seemed that genetic testing and personalized management of patients with AMD was the way of the future, so I accepted. I found the science compelling and became convinced of the value of prognostic AMD genetic testing. As mentioned previously, physician-ordered genetic tests rely not just on genetic risk markers, but also on nongenetic risk variables such as AMD phenotype, age, smoking history, etc. In studying the algorithms used in the prognostic tests, I realized that whether a patient took an AREDS supplement was not an incorporated variable. This seemed odd, as ophthalmologists have been strongly encouraged to promote the AREDS recommendations. Why wasn’t this part of the algorithms?

I learned that the impact of the AREDS supplement had been studied, but had been found to add no predictive power to the algorithms. I wondered whether the apparent quite modest benefit of AREDS supplements might actually represent the product of opposing individual responses, and whether this might be related to differences in genetic risk.

Together with collaborators more expert in genetics and statistics than I, we performed a statistical analysis of publically available AREDS data and DNA from 989 AREDS patients.3 We identified CFH and ARMS2 genetic risk markers as significant predictors of the response to AREDS formulation components. Most importantly, we found that CFH risk was inversely associated with the response to the high-dose zinc used in the AREDS formulation.

A relationship between CFH risk and the effectiveness of AREDS supplements had been previously reported by Klein et al, who identified an interaction between CFH risk and the zinc component of the AREDS supplement.4 However, Klein et al stated that the AREDS formulation was beneficial for all patients, regardless of genotype, with some patients benefitting more than others. Our analysis was the first to suggest that some patients may actually be harmed by high-dose zinc, alone or as a component of the AREDS formulation. Our genotype-directed treatment recommendations, based upon our statistical model, suggested that almost half of eligible patients would have better outcomes if treated with something other than the AREDS formulation (eg, zinc, antioxidants without zinc, or nothing at all).

Following our publication, Chew et al, offered an alternative statistical model in which they found no association between CFH and ARMS2 genotypes and response to AREDS components.5 Although their analysis was technically accurate, their statistical model was underpowered to refute our analysis, akin to tossing a coin five times to test the hypothesis that heads comes up 50% of the time. Scrutiny of their study, which divided 1,237 patients into 27 subgroups for independent analysis, reveals a statistical method underpowered to demonstrate the generally accepted benefits of the AREDS formulation for any subgroup or even for their entire data set.1

For the sake of argument, however, let us accept that the Chew et al analysis is accurate. That leaves us with two statistical models, one (ours) that identifies CFH and ARMS2 as significant predictors of response to nutritional supplements and another (Chew et al) that states that CFH and ARMS2 have no impact. How can we resolve these opposing opinions?

Sometimes, as with the former question of whether bevicizumab (Avastin; Genentech) can penetrate the retina, it is better to simply test the hypothesis than continue a theoretical debate.


Based on our conclusions that CFH and CFH genetic risk alleles were indeed significant predictors of response to AREDS components, we looked at the actual 7-year outcomes of AREDS patients as influenced by differences in CFH and ARMS2 genetic risk.1 This analysis was also done using DNA from 989 AREDS patients, which represented all DNA available from the public repository. DNA sequencing, as in our first study, was performed by an independent reference laboratory.

There are nine possible combinations of zero, one, or two CFH and ARMS2 risk alleles. Because some of the combinations occur infrequently, it is impossible to conduct a statistically meaningful analysis based upon each of the nine genotype combinations, even with almost 1,000 patients. Therefore, we combined rare genotypes (eg, two copies of the ARMS2 risk allele) with more common heterozygous genotypes, defining four genotype groups based on relatively higher or lower CFH and ARMS2 risk allele number. This resulted in groups sufficiently sized to perform a meaningful outcomes analysis.

We found dramatic differences in the actual 7-year outcomes of AREDS patients based upon genetic risk and assigned treatment (Figure).

Patients with high ARMS2 risk and low CFH risk were significantly benefited by zinc or the zinc-containing AREDS formulation, but patients with high CFH and low ARMS2 risk did much worse. Patients with CFH risk alleles and zero ARMS2 risk alleles had a 135% increase in AMD progression if treated with the AREDS formulation, compared with placebo (40% vs 17%).

Patients with both low CFH and ARMS2 risk did slightly better if treated with the AREDS formulation, but these individuals had the best outcomes if they were treated with antioxidants alone. Patients with both high CFH and ARMS2 risk, due to the impact of their high genetic risk on the disease’s progression, had uniformly poor outcomes regardless of treatment.


Although our findings are supported by convincing statistics, it is equally important that research regarding the role of CFH and ARMS2 supports the argument for an interaction with AREDS formulation components. Zinc has complex physiologic influences on oxidative stress, oxidative damage, and activation of the complement system. Zinc has been shown to reduce AMD progression, but has also been implicated in the pathogenesis of AMD.6 Zinc has been specifically demonstrated to inactivate CFH, thus resulting in increased complement-mediated inflammation.7 These effects could be more pronounced in patients with abnormal CFH. The role of ARMS2 is less understood, but abnormal ARMS2 has been demonstrated to impair the ability of the retinal pigment epithelium to defend itself against aging-related oxidative stress because of impaired mitochondrial superoxide dismutase (SOD) 2 activity.8 We hypothesize that zinc supplementation may limit this damage through activation of SOD1, which is zinc and copper dependent, or other parallel pathways.


Based on the outcomes of these AREDS patients, we propose genotype group-based treatment recommendations for patients with AMD (Table). We found that 65% of patients would do better with something other than the AREDS formulation.

Our findings are based on a post-hoc retrospective subgroup analysis. We have not “proved” that CFH and ARMS2 risk alleles have an impact on the response to AREDS formulation components. A true validating study would take years, however, and the available evidence is strong; clinicians are continually compelled to make treatment decisions with imperfect data. In fact, the AREDS recommendations were based on a subgroup analysis of 2,516 patients and were never validated by a second trial, as there was no placebo group in ARMS2.

Our outcomes analysis is based on almost 40% of the 2,516 patients who provided the basis of the AREDS recommendations. We found dramatic differences in response to the AREDS formulation based on easily measured genetic risk. As a practicing retina specialist, I find these differences too large to ignore, and I choose to follow the treatment guidelines as presented in the table. At a minimum, I think doctors should consider withholding supplements containing highdose zinc from patients in genotype group 2 (2 CFH and 0 ARMS2 risk alleles) until further studies are done.Primum non nocere (first, do no harm).


The AREDS2 formulation has the same zinc dose (80 mg) as the original AREDS formulation, so we think our treatment recommendations remain valid for patients who wish to take the newer supplement. We have no way of knowing whether the substitution of lutein and zeaxanthin for beta-carotene will alter the response to antioxidants seen in our outcomes analysis. Because the addition or removal of zinc seems to be the critical variable, however, we predict that outcomes for patients treated with a zinc-free ARMS2 formulation will be similar to those in our study treated with the original AREDS antioxidants.

Carl C. Awh, MD, practices at Tennessee Retina in Nashville. He is an Arctic Dx advisory board member, consultant, and shareholder. Dr. Awh may be reached at

  1. Awh CC, Hawken S, Zanke BW. Treatment response to antioxidants and zinc based on CFH and ARMS2 genetic risk allele number in the Age Related Eye Disease Study [published online ahead of print September 4, 2014]. Ophthalmology. 2014. doi: 10.1016/j.ophtha.2014.07.049.
  2. Yu Y, Reynolds R, Rosner B, Daly MJ, Seddon JM. Prospective assessment of genetic effects on progression to different stages of age-related macular degeneration using multistate Markov models. Invest Ophthalmol Vis Sci. 2012;53(3):1548-1556.
  3. Awh CC, Hawken,S., Lane, AM., Zanke, B., Kim, I. CFH and ARMS2 Genetic polymorphisms predict response to antioxidants and zinc in patients with age-related macular degeneration. Ophthalmology 2013;120:2317-2323.
  4. Klein ML, Francis PJ, Rosner B, et al. CFH and LOC387715/ARMS2 genotypes and treatment with antioxidants and zinc for age-related macular degeneration. Ophthalmology 2008;115(6):1019-1025.
  5. Chew EY, Klein ML, Clemons TE, et al. No clinically significant association between CFH and ARMS2 genotypes and response to nutritional supplements: AREDS Report Number 38. Ophthalmology 2014.
  6. Lengyel IP, T. Cure or cause: The opposing roles for zinc in age-related macular degeneration. Expert Rev Ophthal 2008;3:1-4.
  7. Nan R, Gor J, Lengyel I, Perkins SJ. Uncontrolled zinc- and copper-induced oligomerisation of the human complement regulator factor H and its possible implications for function and disease. J Mol Biol 2008;384(5):1341-1352.
  8. Imamura Y, Noda S, Hashizume K, et al. Drusen, choroidal neovascularization, and retinal pigment epithelium dysfunction in SOD1-deficient mice: a model of age-related macular degeneration. Proc Natl Acad Sci U S A 2006;103(30):11282-11287.