Note the Nuance—Examining the Association of Alzheimer-Protective APOE Variation With Risk for Age-Related Macular Degeneration

Abstract

Apolipoprotein E (apoE), the protein encoded by the APOE gene, combines with lipids to form lipoproteins responsible for packaging and transporting cholesterol throughout the body. APOE is well known for its association with Alzheimer disease (AD); high plasma apoE and specific APOE alleles (ε2) associate with lower AD risk and, conversely, lower apoE levels, and opposing APOE alleles (ε4) associate with higher AD risk. 1 In age-related macular degeneration (AMD), APOE ε2 alleles and higher plasma apoE (and lower risk for AD) associate with AMD risk. 2 It has been suggested that AD-targeted pharmacotherapy should focus on altering apoE plasma levels; 3 however, as blindness is certainly an undesirable adverse effect, this possibility may be limited. A contemporary debate on the influence of rare APOE variation on potential AD-focused drug targets in APOE highlights an area for concern. 3 The complex relationship between apoE levels, APOE variation, AD risk, and AMD risk is the premise for the study presented in this issue of JAMA Ophthalmology by Rasmussen and colleagues, 4 wherein the authors evaluate these factors in a large, longitudinal cohort of more than 100 000 Europeans.These vast data, previously accessed to study AD and related comorbidities, 5 enabled the examination of rare variants for which alleles were present as few as less than 10 times in the sample and/or in publicly available data sets. These variants are truly on the rare or singleton end of the spectrum, including many not reported in previous AMD studies. Evaluating several rare variants implicated in AD, 5 but not previously reported in AMD, the authors detected an average 8% to 10% reduction in AMD risk moving from ε22 to ε32 to ε42 to ε33 to ε43 to ε44, which is the opposite direction of AD risk. In addition, some of the rarer isoforms like ε22, ε42, and to a lesser extent ε44, are not individually significant in all comparisons, which means that these would need further examination for their effect. Currently, it is impossible to account for all rare or structural variations in the APOE extended locus, and to imply otherwise warrants caution.Interestingly, there is some evidence that the rare APOE variants have different effects in AMD subtypes, but these are not discussed. The association between weighted allele score and "not neovascular AMD" strengthens substantially when adjusting for the more common APOE ε2/ε3/ε4 variants, despite loss of power from the addition of 5 parameters to estimate. At the same time, increases for "neovascular AMD" are modest, and likely temper the increase in "all AMD." This is supported by single-variant tests of association, although the rarity of these variants results in many insignificant or incalculable hazard ratios. A subgroup analysis in those with ε33 only further attempts to separate the collective effects of the rare variants from APOE ε2/ε3/ε4 in the weighted allele score,