Developing blood fatty acid-based algorithms as early predictors of macular degeneration and glaucoma: Applying machine learning to harmonized data from prospective cohort studies
openNEI - National Eye Institute
Project summary Two of the most common and debilitating eye-related diseases are age-related macular
degeneration (AMD; ~20 million cases in the US) and glaucoma (~4 million cases in the US). The economic
impact of these conditions is substantial and growing as our population continues to age, with an estimate of
over $373 billion in annual lost productivity by 2050 in the US alone. Over the last 60 years multiple risk factors
(RFs) for AMD and/or glaucoma have been discovered, with smoking, blood pressure, obesity, high
cholesterol, cardiovascular disease, diabetes, poor diet, and sun exposure now considered ‘standard’ and, in
some cases, modifiable. But even when combined with age, sex, race/ethnicity and genetics, the predictive
ability of these RFs is still lower than desired. Novel biomarkers of risk that can better predict who is at high
risk for AMD and/or glaucoma, and/or changes in optical coherence tomography angiography (OCTA)
measures (e.g., retinal thickness; vessel density) may allow for earlier and more targeted intervention and are
of high interest. There is considerable evidence that circulating fatty acids (FAs; esp. n3-FA - see the recent
JAMA Ophthalmology paper by co-I Sala-Vila et al. on n-3s and diabetic retinopathy) can provide prognostic
information regarding risk for eye health, independent of standard RFs. But, emerging data supports a role for
other FAs as well. Data suggest that trans FA are potentially harmful, omega-6 levels are largely inconclusive,
and saturated fats and mono-unsaturated fats (e.g., oleic acid, common in the mediterranean diet) have
yielded widely disparate findings to date. As a clinical laboratory that has specialized in providing FA
measurements, interpretation and customized behavioral interventions for the last 15 years, OmegaQuant
Analytics (OQA) supports a large and growing customer base of researchers, clinicians, businesses, and
individuals- including an increasing number of optometrists, ophthalmologists and other eye health experts.
OQA is a leader in the at-home FA testing market through its innovative dried blood spot collection system,
testing ~40,000 samples annually. In this project we will: Aim 1. Use machine learning to define RBC FA
patterns that predict risk for 1) incident AMD 2) incident glaucoma or 3) related OCTA measures. We will begin
by harmonizing eye health outcomes, fatty acids and covariate data from the FHS, WHIMS, MESA, and
BPRHS yielding sample sizes of up to 19,922 total individuals with information on AMD or glaucoma outcomes
over an average of 10+ years of follow-up. We will then apply statistical / machine learning algorithms to
determine (Aim 1a) the extent to which we can separately predict changes in AMD and glaucoma from
baseline RBC FA metrics. These analyses will lead to 2 unique sets of FA metrics that will predict risk for 1)
incident AMD (the macular degeneration FA index, FAMADI), and 2) incident glaucoma (Glaucoma Fatty Acid
Index, GAFI). Aim 2. Explore how FAMADI and GAFI can be leveraged to profitability. Proof of concept
feasibility will set us up for larger-scale prospective studies and improved modelling in Phase II.
Up to $321K
health research