Abstract
Purpose:
Statistically independent phenotypes and genotypes related to macular pigment optical density (MPOD) levels have not reflected joint and synergistic factors.
Methods:
Partial least squares (PLS) regression, a data-dimension reduction technique, was used to select and identify MPOD determinants which were the strongest predictors conditional on (rather than independent from) other variables in the model. A total of 28 phenotypes and 439 tag SNPS were evaluated. The dataset included 1,056 women, 55-84 years of age, in the Carotenoids in Age-Related Eye Disease Study with complete data. MPOD was assessed in 2001-2004, using heterochromatic flicker photometry at 0.5 degrees. Intakes of lutein (L) and zeaxanthin (Z), which comprise MP, and other diet components were estimated food frequency and supplement questionnaires. Anthropometrics and sun exposure history were obtained. Lifestyle and health history variables and blood specimens, analyzed for serum carotenoids, triglycerides, and SNPs from carotenoid candidate genes, were obtained in the Women's Health Initiative visit (1994-1998). The most stable model was determined by leave-one-out, cross-validation testing in a model with two factors and included variables with variable importance in the projection (VIP) scores >1.2.
Results:
The model (Q2=7.4%) contained 68 covariates (VIP scores 5.3 to 1.2) and included phenotypes and genotypes previously found to be independently related to MPOD: (serum LZ , waist circumference, intakes of fruit and vegetables, beta-carotene, dietary fiber, polyunsaturated, saturated and omega-3 fats and variants in BCMO1, RPE65, SCARB1, LIPC, ABCA1, ABCG5 and ELOVL2. Novel genotypes were also identified and included 10 SNPS within the StARD3, StAR-related lipid transfer domain containing 3 gene, encoding a human ocular lutein specific binding protein, 9 SNPS within the fatty acid elongase 5 ELOVL5 gene, related to elongation of very long chain fatty acids, and rs217434 of the Niemann-Pick C1-Like 1gene. In addition, novel phenotypes were identified (VIP score 2.4-1.2) and included low levels of serum triglycerides, high levels of physical activity and light exposure and not using cholesterol lowering medication or having hypertension.
Conclusions:
PLS identified novel genotypes and phenotypes related to MPOD, suggesting additional potential biological mechanisms which influence MPOD accumulation.