Urine Nuclear Magnetic Resonance (NMR) Metabolomics in Age-Related Macular Degeneration

Abstract

Biofluid biomarkers of age-related macular degeneration (AMD) are still lacking, and their identification is challenging. Metabolomics is well-suited to address this need, and urine is a valuable accessible biofluid. This study aimed to characterize the urinary metabolomic signatures of patients with different stages of AMD and a control group (>50 years). It was a prospective, cross-sectional study, where subjects from two cohorts were included: 305 from Coimbra, Portugal (AMD patients n = 252; controls n = 53) and 194 from Boston, United States (AMD patients n = 147; controls n = 47). For all participants, we obtained color fundus photographs (for AMD staging) and fasting urine samples, which were analyzed using ¹H nuclear magnetic resonance (NMR) spectroscopy. Our results revealed that in both cohorts, urinary metabolomic profiles differed mostly between controls and late AMD patients, but important differences were also found between controls and subjects with early AMD. Analysis of the metabolites responsible for these separations revealed that, even though distinct features were observed for each cohort, AMD was in general associated with depletion of excreted citrate and selected amino acids at some stage of the disease, suggesting enhanced energy requirements. In conclusion, NMR metabolomics enabled the identification of urinary signals of AMD and its severity stages, which might represent potential metabolomic biomarkers of the disease.

Keywords: age-related macular degeneration; metabolomics; nuclear magnetic resonance spectroscopy; urine.

Figure 1.

Average ¹H NMR spectra of urine of controls (a,c) and late AMD groups (b,d) for Coimbra (a,b) and Boston (c,d) cohorts. Peak numbering identifies the main assignments: 1, ile; 2, leu; 3,val; 4, β-HBA (β-hydroxybutyrate); 5, 3-HIVA (3-hydroxyisovalerate); 6, lactate; 7, thr; 8, 2-HIBA (2-hydroxyisobutyrate); 9, ala; 10, acetate; 11, PAG (phenylacetylglutamine); 12, p-CS (p-cresol sulfate); 13, citrate; 14, DMA (dimethylamine); 15, creatine; 16, creatinine; 17, carnitine; 18, TMAO (trimethylamine-N-oxide); 19, tau; 20, gly; 21, hippurate; 22, trigonelline; 23, NMND (N-methyl nicotinamide), 24, ascorbate; 25, glucose; 26, furoylglycine; 27, 2-Py (N-methyl-2-pyridone-5-carboxamide); 28, 4-HPA (4-hydroxyphenylacetate); 29, tyr; 30, 4-OH-hippurate (4-hydroxyhippurate); 31, formate. Remaining assignments are shown in Table S2. The three-letter code is used for amino acids. Arrows indicate some visual changes. *: removed water and urea spectral regions.

Figure 2.

PLS-DA scores plot for pairwise models for variable-selected spectra of urine from the a) Coimbra cohort (circles): left, controls (blue, n = 52) vs early AMD (red, n = 54); middle, early AMD (red, n = 56) vs intermediate AMD (green, n = 141); right, intermediate AMD (green, n = 141) vs late AMD (purple, n = 54), and b) Boston cohort (diamonds): left, controls (blue, n = 46) vs early AMD (red, n = 30); middle, early AMD (red, n = 33) vs intermediate AMD (green, n = 64); right, intermediate AMD (green, n = 63) vs late AMD (purple, n = 43). Q² value and MCCV results are shown for each model.

Figure 3.

Heatmap of urine (left) and plasma (right) based on previous results metabolite changes related to AMD in (a) Coimbra and in (b) Boston cohorts. C: controls, E: early AMD, I: intermediate AMD, L: late AMD, a.a.: amino acids, o.a.: organic acids, Un.: unassigned peaks and regions, 3-letter code used for amino acids, 2-HIBA: 2-hydroxyisobutyrate, 4-HPA: 4-hydroxyphenylacetate, cho: choline, DMA: dimethylamine, β-HBA: β-hydroxybutyrate, FA: fatty acids, NMND: N-methylnicotinamide. ^a: metabolite whose level may be affected by age to some extent, ^b: metabolite related to diet and/or gut microflora. * p-value <0.05; ** p-value <0.01; *** p-value <0.001.