Addendum: 2014 Studies

Summary
In September 2013, SBH addressed the available evidence regarding the genetic testing of AMD patients conducted for the purpose of “tailoring” AREDS supplements based on an individual’s genetic risk profile. The resulting report explains the research, offers a glossary of genetic terms, and provides a prospective on the issues related to customized supplementation based on genetic make-up. 

Since that time, two additional studies have been published in 2014: A study by Chew et al1, and a second study by Awh et al2.  These two studies are discussed below.  In addition, the new studies were independently assessed by Elizabeth Johnson, PhD of the Jean Mayer USDA Human Nutrition Research Center on Aging, and Professor, Friedman Scahool of Nutrition and Science Policy at Tufts University. After careful review, SBH continues to conclude that current information is insufficient to make personalized supplement recommendations based on individual genotype.

New Research
Background: The first study by Chew and colleagues (published June, 2014) was conducted to reexamine the previous findings of Awh et al, who in 2013 reported an increased risk of disease progression associated with zinc treatment in patients with CFH risk genotypes, while antioxidants appeared to be of benefit. The opposite relationships were observed for the ARMS2 risk genotypes: maximum benefit was associated with zinc supplementation, while a detrimental response was linked to antioxidant supplements. 

Chew et al examined the same AREDS trial participant subgroups as did Awh et al in their 2013 study, and came to a different conclusion. Chew et al found that AREDS supplements (zinc plus antioxidants) reduced the rate of disease progression in all genotype groups—i.e. having a different genetic background had no impact on their response to AREDS supplements. Having CFH and ARMS2 risk genotypes did not significantly alter the benefits of the zinc and antioxidant combination AREDS formula. In contrast to the recommendations of Awh et al, these authors conclude that customized supplements (i.e. a zinc free or antioxidant only formulas) based on CFH and ARMS2 risk genotypes is unwarranted.

The second study conducted by Awh and colleagues was published in September 2014.  This retrospective statistical analysis compared disease progression rates within four selected CFH and ARMS2 risk genotype subgroups that received placebo, antioxidants, zinc or combined zinc and antioxidants in the AREDS trial. They then estimated or projected the potential impact of treatment recommendations on these genotype groups.

According to the authors, the benefit of the AREDS formula (combined zinc and antioxidants) seems the result of a favorable response by patients in only one subgroup, balanced by neutral or unfavorable responses in the three other pre-selected genotype groups. Those with 2 CFH risk alleles and no ARMS2 risk genotypes receiving zinc fared worse than those in the placebo group according to their projections, and the authors conclude that people with these risk alleles should avoid taking the combined AREDS formula.

Why the Discrepancy in Findings?
There are differences in methodological approaches that could explain the contrasting results of these two studies that evaluated the same study population. Some of the notable differences and concerns include:

• Overall, the lack of statistical power is the biggest concern. An evaluation of combinations of CFH and ARMS2 alleles, as in the Awh et al paper, could make this problem worse. For example if a given SNP is relatively frequent then groups carrying the SNP are large, but if this group is combined with another SNP of low frequency (that is, individuals need to have both SNPs) then the second SNP could limit the size of the groups that are compared. (A SNP is a site in the genome where a single DNA nucleic acid base--C, G, T, or U—often differs from person to person).

• In contrast to Awh et al, Chew et al. independently established risk groups, which reduces bias.  Chew et al grouped genotypes into 3 risk categories--low risk, medium risk, and high risk according to their combined risk effect—based upon a separate analysis of another group of individuals (the Genome-Wide Association Study cohort) with DNA available for analysis. 

• Although both papers analyzed some of the same CFH genotypes, each analyzed two additional genotypes. This may also account for the different sample sizes in each study (larger in the Chew et al study), which may also contribute to the differing results.

• The same ‘rs’ (SNP identification numbers) were not always evaluated in the two studies (akin to comparing apples with oranges), and the differences in SNP choices could be a source of differing results.

• It should be noted that the statistical values (‘p’ values) of the Awh paper were not corrected for the multiple tests performed. Significant findings may be lost when the customary correction for multiple testing is applied. (To reduce the chances of obtaining false-positive results when certain statistical tests are performed on a single set of data, an adjustment to p values, called the Bonferroni correction, is applied).

Additionally, Awh et al suggest an insufficient sample size for the Chew et al paper in which 27 subgroups (9 genetic subgroups x 3 active AREDS treatments) were analyzed. In other words, Awh et al indicated that Chew et al looked at too many subgroups and included too many variables, resulting in a lack of statistical power. While this is true, the initial statistical analysis of Chew et al found no significant results. Only then did Chew et al perform the group interaction analysis to mimic what was done in the Awh et al paper.

Bottom Line?
In any event, these studies as well as the two previous studies discussed in the earlier SBH report had limited sample sizes.  Due to the inconsistent methods, results, and sample size limitations, SBH concludes that information remains insufficient at this time to recommend personalized AREDS supplements based on different genetic profiles.

An overall concern is the individual variability in progression of this disease. AMD is a complex condition that likely results from a combination of genetic and environmental factors. One or two loci (location of a gene on a chromosome) may be insufficient to account for differences in response to a set of nutrients. Importantly, the discrepancies in the results of these two analyses conducted in the same population warrants an investigation of possible relationships in more independent populations. It would be premature to draw conclusions from only one dataset. 

References:

• Chew EY et al. No clinically significant association between CFH and ARMS2 genotypes and response to nutritional supplements. AREDS report number 38. Ophthalmol; [Epub ahead of print], June 2014.
• Awh CC et al. Treatment response to antioxidants and zinc based on CFH and ARMS2 genetic risk allele number in the age-related eye disease study. Ophthalmol; [Epub ahead of print], Sept 2014.