Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Apr 27.
Published in final edited form as: Ophthalmology. 2017 Sep 12;125(2):245–254. doi: 10.1016/j.ophtha.2017.08.008

Human Plasma Metabolomics Study across All Stages of Age-Related Macular Degeneration Identifies Potential Lipid Biomarkers

Inês Laíns 1,2,3,4, Rachel S Kelly 5, John B Miller 1, Rufino Silva 2,3,4, Demetrios G Vavvas 1, Ivana K Kim 1, Joaquim N Murta 2,3,4, Jessica Lasky-Su 5, Joan W Miller 1,#, Deeba Husain 1,#
PMCID: PMC8077680  NIHMSID: NIHMS959525  PMID: 28916333

Abstract

Purpose:

To characterize the plasma metabolomic profile of patients with age-related macular degeneration (AMD) using mass spectrometry (MS).

Design:

Cross-sectional observational study.

Participants:

We prospectively recruited participants with a diagnosis of AMD and a control group (>50 years of age) without any vitreoretinal disease.

Methods:

All participants underwent color fundus photography, used for AMD diagnosis and staging, according to the Age-Related Eye Disease Study classification scheme. Fasting blood samples were collected and plasma was analyzed by Metabolon, Inc. (Durham, NC), using ultrahigh-performance liquid chromatography (UPLC) and high-resolution MS. Metabolon’s hardware and software were used to identify peaks and control quality. Principal component analysis and multivariate regression were performed to assess differences in the metabolomic profiles of AMD patients versus controls, while controlling for potential confounders. For biological interpretation, pathway enrichment analysis of significant metabolites was performed using MetaboAnalyst.

Main Outcome Measures:

The primary outcome measures were levels of plasma metabolites in participants with AMD compared with controls and among different AMD severity stages.

Results:

We included 90 participants with AMD (30 with early AMD, 30 with intermediate AMD, and 30 with late AMD) and 30 controls. Using UPLC and MS, 878 biochemicals were identified. Multivariate logistic regression identified 87 metabolites with levels that differed significantly between AMD patients and controls. Most of these metabolites (82.8%; n = 72), including the most significant metabolites, belonged to the lipid pathways. Analysis of variance revealed that of the 87 metabolites, 48 (55.2%) also were significantly different across the different stages of AMD. A significant enrichment of the glycerophospholipids pathway was identified (P = 4.7 × 10−9) among these metabolites.

Conclusions:

Participants with AMD have altered plasma metabolomic profiles compared with controls. Our data suggest that the most significant metabolites map to the glycerophospholipid pathway. These findings have the potential to improve our understanding of AMD pathogenesis, to support the development of plasma-based metabolomics biomarkers of AMD, and to identify novel targets for treatment of this blinding disease.

Age-related macular degeneration (AMD) is the leading cause of adult blindness in developed countries. Worldwide, it ranks third and is expected to affect 288 million people by 2040.1 Even when it does not cause blindness, AMD often leads to altered central visual function and significant impairment in patient quality of life.2 The natural history of AMD typically comprises early and intermediate forms, which can progress to atrophic lesions (i.e., geographic atrophy) or neovascular lesions (i.e., choroidal neovascularization, wet AMD), or both in some.3,4

Age-related macular degeneration usually is asymptomatic in its early stages and is diagnosed only on routine eye examination; thus, it often remains undetected until it is more advanced and accompanied by loss of vision. It is important to find tools to detect AMD earlier and to develop treatments to slow progression and vision loss. Ideally, a screening test for AMD should be performed easily and should be able to predict disease progression. Serologic biomarkers have been studied, mostly related to biomarkers of inflammation5–8 or lipid levels,9 because both seem to play a role in AMD pathogenesis. However to date, results have been inconsistent.10 Therefore, we continue to rely on fundus examination to diagnose and monitor progression, and useful biofluid biomarkers are still lacking in AMD. Metabolomics, the global profiling of all the small molecules (<1 kDa) constituting a biological system,11 is a new approach that is now recognized for its promising role in identifying biomarkers.3,4,12 Metabolites are the downstream products of the cumulative effects of the genome and its interactions with environmental exposures.13 Therefore, the metabolome is thought to represent closely the so-called true functional state of the biological system and thus disease phenotype in multifactorial diseases, such as AMD.14

The vital role of metabolomics as a translational tool for the clinical setting has been demonstrated through several studies in other medical disciplines,13 including cancer and prenatal diseases.15–19 Metabolomics studies can be performed readily on easily accessible biological samples, such as plasma, serum, and urine. Two main techniques can be used for metabolomic profiling: mass spectrometry (MS) and nuclear magnetic resonance spectroscopy.20,21 Both approaches have strengths and weaknesses, and currently they are considered complementary.22 Mass spectrometry presents a high sensitivity and selectivity, which renders it increasingly popular in large-scale metabolomics studies.22–24 Recent work also has highlighted the potential and versatility of metabolomics for the study of eye diseases.25–27 However, to our knowledge, only 1 study has been published on AMD to date,28 comparing plasma metabolomic profiles of participants with neovascular AMD and those of a control group.

The current study aimed to characterize the plasma metabolomic profiles of patients with AMD and to compare them with those of participants with no AMD, and also to compare the findings across the different stages of AMD (early, intermediate, and late disease) using MS-based metabolomics. Ultimately, we aim to support the development of novel metabolic biomarkers for this blinding disease to aid in diagnosis and prognosis, as well as to understand disease mechanisms better and to identify new druggable targets.

Methods

Study Design

This study was part of a cross-sectional observational study on AMD biomarkers at Massachusetts Eye and Ear, Harvard Medical School, Boston, Massachusetts. The research protocol was conducted in accordance with Health Insurance Portability and Accountability Act requirements and the tenets of the Declaration of Helsinki and was approved by the Massachusetts Eye and Ear Institutional Review Board. All included participants provided written informed consent and were recruited prospectively.

Study Protocol

From January 2015 through July 2016, we recruited participants with a diagnosis of AMD at the time of their regular appointments at the Massachusetts Eye and Ear Retina Service. Exclusion criteria included diagnosis of any other vitreoretinal disease, active uveitis or ocular infection, significant media opacities that precluded the observation of the ocular fundus, refractive error of 6 diopters or more of spherical equivalent, history of retinal surgery, history of any ocular surgery or intraocular procedure (such as laser and intraocular injections) within the 90 days before enrollment, and diagnosis of diabetes mellitus, with or without concomitant diabetic retinopathy. Additionally, a control group of participants 50 years of age or older without any evidence of AMD in either eye was identified and gave informed consent at the Massachusetts Eye and Ear Comprehensive Ophthalmology and Optometry Services. The same exclusion criteria were applied.

All participants underwent a comprehensive eye examination, including best-corrected visual acuity assessment, current refraction, intraocular pressure measurement, slit-lamp biomicroscopy, and dilated fundus examination. A standardized medical history questionnaire was designed specifically for this study, including, among others, self-reported data on smoking habits (smokers were considered those who reported current smoking and former smokers were considered those who have ever smoked, regardless of when they stopped, but who do not currently smoke), and weight and height, which were used to calculate body mass index (BMI). If the study participants did not know their current height or weight, these were obtained by a study investigator.

Nonstereoscopic 7-field color fundus photographs (Topcon TRC-50DX [Topcon Corporation, Tokyo, Japan] or Zeiss FF-450Plus [Carl Zeiss Meditec, Dublin, CA]) were obtained at the same visit. These were used to diagnose and grade AMD according to the Age-Related Eye Disease Study (AREDS) classification system.29,30 Two independent experienced graders analyzed all field 2-color fundus photographs (https://www.ncbi.nlm.nih.gov/pubmed/28316876) on IMAGEnet 2000 software (version 2.56; Topcon Medical Systems). In cases of disagreement, a senior author (D.H.) established the final categorization. All graders were masked to the patients’ clinical and demographic characteristics during this process.

We adopted the most recent AREDS 2 definitions,30 namely that the standard disc diameter equals 1800 μm (rather than 1500 μm), which affects the size of the Early Treatment Diabetic Retinopathy Study grid and of the standard drusen circles, and that geographic atrophy is present if the lesion has a diameter of 433 μm or more (AREDS circle I-2) and has at least 2 of the following features: absence of retinal pigment epithelium (RPE) pigment, circular shape, or sharp margins (foveal involvement not a requirement). With these criteria, we established the following groups29,30: controls (AREDS stage 1), presence of drusen maximum size less than circle C0 and total area less than C1; early AMD (AREDS stage 2), drusen maximum size C0 or more but less than C1 or presence of AMD characteristic pigment abnormalities in the inner or central subfields; intermediate AMD (AREDS stage 3), presence of drusen maximum size of C1 or more or of drusen maximum size of C0 or more if the total area occupied was more than I-2 for soft indistinct drusen and more than 02 for soft distinct drusen; and late AMD (AREDS stage 4), presence of geographic atrophy according to the criteria described above or evidence of neovascular AMD. For participants with different severity stages in the 2 eyes (for example, early AMD in one eye and intermediate in the other eye), the more advanced stage was assumed.

Sample Collection and Mass Spectrometry Analysis

This study used a single plasma collection per individual. For all participants, after confirmed overnight fasting, blood samples were collected in the morning, into sodium–heparin tubes and centrifuged within 30 minutes (1500 rpm, 10 minutes, 20°C). Plasma aliquots of 1.5 ml were transferred into sterile cryovials and immediately stored at −80°C. When all participants had been recruited, plasma samples were shipped to Metabolon, Inc., in dry ice (through TNT Express, Inc., Melville, NY). Samples arrived frozen in less than 24 hours and were stored immediately at −80°C until processing, which was performed according to the protocol described in the Supplemental Material (available at www.aaojournal.org). Nontargeted MS analysis was performed by Metabolon, Inc., using ultrahigh-performance liquid chromatography–tandem MS, according to protocols that have been described previously31 and are summarized in the Supplemental Material (available at www.aaojournal.org).

Statistical and Data Analysis

Traditional descriptive methods were used to describe the clinical and demographic characteristics of the included study population: mean and standard deviation for continuous variables and percentages for dichotomous or categorical variables. The 4 study groups were compared using analysis of variance and chi-square tests.

In this data set, 878 named metabolites were identified, of which 384 (44%) mapped to the lipid pathways. A total of 173 metabolites determined to be exogenous to humans (including medications, food additives, and buffering agents) were excluded from analysis because we were interested in endogenous metabolites that could be driving systemic biological factors. As part of our quality-control procedures, we observed that 1 participant (a man with early AMD) had missing or undetectable levels for more than 30% of metabolites and therefore was excluded. Any missing values for the remaining participants were imputed with half the minimum detected level for that metabolite. To ensure that only the most informative metabolites were included in the analysis, those metabolites with an interquartile range of 0 were excluded. This left 698 metabolites that were Pareto scaled and log-transformed for analysis. Figure S1 (available at www.aaojournal.org) presents the included metabolites and samples.

Owing to the large number of metabolites that can now be measured in the human body, specialized statistical methods are required to analyze high-dimensional metabolomic data sets.32 One of the most commonly used dimensionality reduction techniques is principal component analysis, which we performed in our study.33 Principal component analysis is an unsupervised clustering approach that assesses how participants cluster based on their metabolome. Basically, it relies on the transformation of metabolites into a set of linearly uncorrelated variables known as principal components, which summarize a large number of metabolites with a smaller number of variables. This decomposition method maximizes the variance explained by the first component, whereas the subsequent components explain increasingly reduced amounts of variance.34

To isolate the metabolites significantly associated with AMD compared with normal macular health, a multivariate logistic regression was performed to account for potential confounding factors (age, gender, BMI, and smoking status). The discriminatory ability of the significant metabolites then was assessed using a summary score based on their first principal component and receiver operating characteristic (ROC) curve analyses. The significant metabolites were studied further using and analysis of variance to see whether they differed between early, intermediate, and late AMD cases. Pathway analysis using MetaboAnalyst 3.0 (available at: http://www.metaboanalyst.ca/faces/home.xhtml)35 was performed on the significant metabolites to interpret these findings biologically. This combines overrepresentation analysis with topology analysis to identify pathways that are dysregulated in AMD based on (1) the number of metabolites from our significant metabolites that fall within Kyoto Encyclopedia of Genes and Genomes–defined metabolic pathways and (2) the positional importance of our metabolites within these pathways. It generates a pathway impact score and the associated P value.

Results

Study Population

We recruited 120 participants, 25% (n = 30) with normal macular health (control group) and 75% (n = 90) with AMD. As described, 1 patient with AMD (early AMD) was considered an outlier and was excluded from further analyses. Table 1 presents the clinical and demographic characteristics of the study group. Among the potential confounders evaluated, only age showed a statistically significant difference among the different study groups (P = 0.0005; Table 1).

Table 1.

Clinical and Demographic Characteristics of the Study Population

Total Population (n = 119) Control (n = 30) Early (n = 29) Intermediate (n = 30) Late (n = 30) P Value
Age (yrs), mean (SD) 70 (8) 68 (10) 68 (7) 70 (5) 76 (8) 0.0005*
Gender, no. (%)
 Male 43 (36) 12 (40) 9 (31) 9 (30) 13 (43) 0.640
BMI (kg/m2), mean (SD) 27 (5) 26 (4) 27 (5) 29 (7) 27 (4) 0.070
Ethnicity, no. (%)
 White 101 (85) 22 (73) 24 (83) 30 (100) 25 (83) 0.176
 Black/Hispanic/Asian 7 (6) 2 (7) 3 (10) 0 2 (7)
 Unknown 11 (9) 6 (20) 2 (7) 0 3 (10)
Smoking, no. (%)
 Nonsmoker 57 (48) 18 (60) 18 (62) 13 (43) 8 (27) 0.057
 Former smoker 56 (47) 10 (33) 11 (38) 15 (50) 20 (67)
 Smoker 3 (3) 1 (3) 0 2 (7) 0
 NA 3 (3) 1 (3) 0 0 2 (7)
Age started smoking (yrs), mean (SD) 19 (7) 20 (7) 16 (4) 22 (10) 17 (4) 0.055
Age stopped smoking (yrs), mean (SD) 39 (13) 37 (11) 34 (16) 42 (12) 42 (13) 0.315
No. of cigarettes per day, mean (SD) 18 (14) 18 (10) 9 (8) 20 (18) 20 (14) 0.244
AMD subtype, no. (%)
 Choroidal neovascularization (wet) 25 (28) n/a n/a n/a 25 (83) 1.8 × 10−17
 Geographic atrophy (dry) 4 (5) n/a n/a n/a 4 (13)
Open in a new tab

AMD = age-related macular degeneration; BMI = body mass index; NA = not available; n/a = not applicable; no. = number; SD = standard deviation.

*

Significant differences (P < 0.05), analysis of variance for continuous variables and chi-square test for dichotomous variables.

Principal Component Analysis

In our data set, the first 2 principal components (PC1 and PC2) accounted for 20% of the variance in the data (Fig 1). As shown, there is a suggestion of a shift between the late-stage patients (blue) and the controls (black).

Figure 1.

Figure 1.

Open in a new tab

Scatterplot showing principal component 1 (PC1) and principal component 2 (PC2) with controls and age-related macular degeneration (AMD) groups. The x-axis corresponds to PC1 and the y-axis to PC2.

Multivariate Logistic Regression Analysis

To identify the metabolites driving the differences in the metabolome of AMD cases and controls observed in the principal component analysis, we conducted multivariate regression analyses controlling for potential confounders. We considered a dichotomous outcome: normal macular health (controls; n = 30) versus AMD (n = 89). Our results revealed that, after controlling for age, gender, BMI, and smoking status, 87 metabolites were associated with AMD (Table 2). Most of these metabolites (82.8%; n = 72) belonged to the lipid superpathway, followed by amino acids (5.7%; n = 5; including N-acetylasparagine, a component of alanine and aspartate metabolism). The remaining metabolites were peptides (4.6%; n = 4), cofactors and vitamins (2.3%; n = 2), and metabolites involved in purine and pyrimidine metabolism (4.6%; n = 4). Of the 7 most significant metabolites (P < 0.001; Table 3), all but 1 (adenosine) were lipids (4 diacylglycerols and 2 phosphatidylcholines). Pathway analysis of these 87 metabolites confirmed the importance of lipid metabolism, specifically glycerophospholipid metabolism, in AMD. Indeed, this pathway was highly enriched among the significant metabolites (P = 4.7 × 10−9; Fig 2).

Table 2.

Multivariate Logistic Regression Analysis for Age-Related Macular Degeneration Patients versus Controls

Metabolites in AMD Groups Compared with Controls
Decreased in AMD Patients (Odds Ratio <1) Increased in AMD Patients (Odds Ratio >1) Total
Significance Level No. % No. % No. %
P < 0.05 59 8.5 28 4.0 87 12.5
P < 0.01 24 3.4 9 1.3 33 4.7
P < 0.001 6 0.9 1 0.1 7 1.0
Open in a new tab

AMD = age-related macular degeneration; No = number.

Table 3.

Significantly Different Metabolites (P < 0.001) between Patients with Age-Related Macular Degeneration and Controls

Biochemical Superpathway Subpathway Metabolites in AMD Patients vs. Controls Odds Ratio* P Value
Linoleoyl-arachidonoyl-glycerol (18:2/20:4) [2]* Lipid Diacylglycerol Decreased 0.0961 0.0008
Stearoyl-arachidonoyl-glycerol (18:0/20:4) [1]* Lipid Diacylglycerol Decreased 0.0411 0.0009
Oleoyl-arachidonoyl-glycerol (18:1/20:4) [2]* Lipid Diacylglycerol Decreased 0.0463 0.0002
Oleoyl-arachidonoyl-glycerol (18:1/20:4) [1]* Lipid Diacylglycerol Decreased 0.111 0.0007
1-Palmitoyl-2-arachidonoyl-GPC (16:0/20:4n6) Lipid Phosphatidylcholine Decreased 0.0004 0.0006
1-Stearoyl-2-arachidonoyl-GPC (18:0/20:4) Lipid Phosphatidylcholine Decreased 0.0002 0.0005
Adenosine Nucleotide Purine metabolism, adenine containing Increased 3.7422 0.0009
Open in a new tab

AMD = age-related macular degeneration; GPC = glycero-3-phosphocholine; PC = phosphatidylcholine.

*

Reference term for odds ratios is the control group, which means that values of less than 1 represent a decrease in patients with age-related macular degeneration as compared with controls (and the opposite for values of more than 1).

Figure 2.

Figure 2.

Open in a new tab

Graph showing pathway analysis based on the 87 metabolites associated significantly with age-related macular degeneration. −log(p) = minus logarithm of the P value.

Receiver Operating Characteristic Curve Analysis

The significant metabolites identified in the multivariate logistic regression were used to create a predictive model for AMD, which was tested using ROC curve analyses. When a summary score based on the first principal component of these 87 metabolites was included as a model predictor, ROC analysis (area under the ROC curve, 0.80; 95% confidence interval, 0.71–0.90) demonstrated that outperformed (P = 0.142) a baseline model including only age, gender, BMI, and smoking status (area under the ROC curve, 0.71; 95% confidence interval, 0.59–0.85; Fig 3).

Figure 3.

Figure 3.

Open in a new tab

Receiver operating characteristic (ROC) curve analysis of model including the 87 significant metabolites compared with the baseline model. The results of the ROC analysis, including metabolite data, which outperformed (area under the ROC curve [AUC], 0.80; 95% confidence interval [CI], 0.71–0.90) a baseline model including only age, gender, body mass index, and smoking status (AUC, 0.71; 95% CI, 0.59–0.85), appear in red.

Analysis of Variance

We further explored whether the identified significant metabolites (n = 87) were able to discriminate between early, intermediate, and late AMD cases. Analysis of variance revealed that, of the 87 metabolites, 48 (55.2%) were significantly different across the different stages of AMD. Consistent with the previous data, all but 1 of the 13 most significant metabolites (P < 0.01) belonged to the lipid pathways (diacylglycerol, n = 4; phosphatidylcholine, n = 3; fatty acid metabolism, n = 3; and phosphatidylinositol, n = 2). Figure 4 displays the mean peak intensity of these 13 metabolites across the 3 AMD groups. Similar to what was observed between AMD and controls, metabolite set enrichment analysis of the 48 metabolites that differed significantly among the 3 AMD groups revealed an enrichment of the glycerophospholipids pathway (P = 0.01).

Figure 4.

Figure 4.

Open in a new tab

Boxplots of the 13 most significant metabolites (P < 0.01) in analysis of variance results comparing severity stages of age-related macular degeneration (AMD). E = early; I = intermediate; L = late.

Discussion

Using a broad-based MS platform, we assessed the plasma metabolomic profiles of a cohort of patients with AMD compared with participants with healthy maculae. After controlling for age, gender, BMI, and smoking status, 87 metabolites differed significantly between AMD patients and controls. Indeed, a summary score based on these 87 metabolites increased the ability to predict AMD cases relative to clinical covariates alone. Of these metabolites, more than half (48 metabolites) also differed significantly across AMD severity stages. Most of the significant metabolites are involved in lipid metabolism, in particular glycerophospholipid metabolism. These metabolites include stearoyl-arachidonoyl-glycerol, a diacylglycerol, and 1-stearoyl-2-arachidonoyl-glycerophosphacholine, a phosphatidylcholine.

Our work presents a pioneer assessment of MS plasma metabolomics across all stages of AMD. Osborn et al28 applied MS plasma metabolomics to the study of this disease, but focused their investigation on a single subtype of advanced AMD (choroidal neovascularization) and described differences primarily related to peptides and modified amino acids. Similarly, we identified significantly increased levels of metabolites linked to dipeptides and amino acids in patients with AMD, including a significant role for alanine and aspartate metabolism. Although lipids were present in the samples from Osborn et al,28 they could not be distinguished and therefore were not analyzed, rendering a direct comparison with our lipid findings impossible.

The potential role of lipids in the pathophysiologic process of AMD has been proposed by several investigators, but their exact role remains somewhat controversial. Candidate gene and genome-wide association studies also have pointed to the role of lipid pathways in the pathogenesis of AMD.12 At a histopathologic level, similarities have been described between the aging process of Bruch’s membrane and atherosclerosis, with deposition of lipids and lipoproteins and associated parainflammation.36,37 Lipids were identified early as a component of drusen, the visible hallmark of AMD.36 It has been postulated by Curcio and others36,37 that these derive from a imbalance of lipoprotein influx and efflux from the RPE,36,37 coming from phagocytosis of the lipid-rich outer photoreceptor segments, RPE synthesis, and only a minor contribution from plasma lipoproteins.38–41 Indeed, pooled data from large epidemiologic studies have failed to show associations between serum cholesterol levels and AMD incidence and progression.10 Clearly, a complete understanding of the role of lipids in AMD pathogenesis remains to be elucidated.42 Metabolomic profiling may provide novel insights into these relationships.

Our data support the relevance of lipid-related metabolites in AMD and, in particular, a significant dysregulation of the glycerophospholipids pathway. Glycerophospholipids are a major component of cell membranes and are especially enriched in neural membranes, accounting for up to 25% of the dry weight of the adult brain. They provide structural stability and membrane fluidity. They also participate in forming ion channels and receptors, generating second messengers in signal transduction, and regulating neuro-transmitter release.43,44 Additionally, the metabolites of glycerophospholipids (together with sphingolipids) seem to play an important role in initiating and promulgating oxidative stress in neurologic disorders, as well as a role in neural cell proliferation, differentiation, and apoptosis.45

Alterations in glycerophospholipids and their metabolism have been investigated extensively in neurodegeneration and in several chronic neurologic diseases.43,46 In particular, glycerophospholipids have been shown to be reduced in the plasma of participants with Alzheimer disease and to play a central role in the pathogenesis of this condition.47,48 Metabolomics currently is considered a promising tool to identify valid biomarkers and new targets in Alzheimer disease.49–51 Using MS metabolomics and lipidomics, Mapstone et al52 identified a panel of 10 plasma lipids, most of them phosphatidylcholines (a class of glycerophospholipids), that predicted with high accuracy conversion to mild cognitive impairment and to frank Alzheimer disease in elderly participants. Changes in plasmalemmas of Alzheimer-affected brain tissue, alterations in local cellular glycerophospholipid metabolism, and increased free-radical–mediated lipid peroxidation all have been suggested as potential causes for the observed glycerophospholipid depletion in this disease. These findings in Alzheimer disease are interesting in the context of our findings in AMD, because both are neurodegenerative diseases and have been noted to share other features and pathologic mechanisms.53–56

The photoreceptors and the RPE are rich in phospholipids, which are important for the transduction of visual stimuli.57,58 As mentioned (Table 3), our study revealed that metabolites linked to the key glycerophospholipids, such as diacylglycerols and phosphatidylcholines, are found at lower levels in participants with AMD, suggesting that impaired cell membrane structure and function may be important components of AMD pathogenesis. The observed depletion of glycerophospholipids in AMD patients could be explained by decreased levels of parent molecules or a change in metabolism or lipid peroxidation. Decreased levels of parent molecules could happen locally in the eye or systemically, because fatty acids required for synthesis in the central nervous system are transported from the gastrointestinal tract (coming from the diet or being produced by the liver).59 Decreased levels of glycerophospholipids also could be the result of altered catabolism because of a change in phospholipases, the enzymes responsible for the catabolism of glycerophospholipids in the central nervous system and the retina.60 Phospholipase C is involved in the regulation of phototransduction and is responsible for the hydrolyzation of phospholipids into inositol 1,4,5-triphosphate and diacylglycerol.43 Phospholipase A2 is another phospholipase that seems to play a role in apoptosis, inflammation, and neurodegeneration.43 One of the catabolic metabolites of phospholipase A2 is glycerolphosphocholine, a metabolite that was decreased in participants with AMD in our cohort. Interestingly, changes in this metabolite also have been identified in Alzheimer disease,50,61,62 once again supporting the hypothesis that there are similarities in the pathogenesis of AMD and Alzheimer disease. Further investigation is needed to understand fully and to validate the role of enzymes and metabolites in AMD.

The current study has a number of limitations, including a relatively small sample size. It was well powered to evaluate the difference between patients with AMD and controls,33 but its relatively small sample size precluded a thorough analysis of AMD subtypes especially the late AMD subgroup - specifically, geographic atrophy and choroidal neovascularization.40 Additionally, our cohort comprised almost solely white participants, which is related in part to the epidemiologic factors of AMD3 and in part to the population served by the enrolling site of our tertiary care hospital. Age differed significantly across the AMD subtypes, which can have an important effect on plasma metabolomic profiles.63 We accounted for this by performing multivariate analysis, controlling for age and other relevant confounding factors. We did not have access to a validation cohort in this study, and this will be sought out for a future study. Additionally, it would be interesting to analyze how other parameters, namely, dietary patterns, as well as conventional measures of lipid levels (such as serum cholesterol) and other serologic biomarkers, relate to our findings. Data on smoking and BMI were collected through self-reported questionnaires; this means that there is a potential response bias. Another interesting analysis for the future will be to correlate these findings with the genetic risk profiles of patients and controls. Finally, this was a cross-sectional study and provided a snapshot of the metabolome for the participants studied. Yet the metabolome is highly dynamic and susceptible to external factors. Longitudinal studies are needed to confirm our findings and to assess the evolution of the metabolome with the progression of AMD. Future work should address these limitations and explore how the identified plasma metabolomic profiles relate to the genetic risk factors that have been linked with AMD. This likely will provide important insights into the pathogenesis of the disease.

An important advantage of our study is that it was designed prospectively, and all data collection was standardized according to a predefined protocol. In addition, participants underwent a complete ophthalmologic examination performed by a retina specialist, ensuring excellent phenotypic characterization. This is particularly important because many other metabolomic studies rely on established repositories and databases that often lack good phenotypic characterization of ophthalmic disease. Our samples were collected after fasting, were processed within 30 minutes, and were stored immediately for metabolomic profiling, which was performed using a state-of-the-art platform that covers a wide range of the metabolome and identifies metabolites using a chemocentric approach with standards for each identified metabolite.

In conclusion, our data provide for the first time a comprehensive overview of AMD metabolomics and suggest that MS plasma metabolomic profiling is a powerful tool to identify participants with AMD and to distinguish the different stages of disease. Our findings also contribute to the current knowledge of AMD pathophysiology by highlighting the role of lipid metabolism. In particular, our work points to the relevance of the glycerophospholipid pathway and the need for further research into its role in AMD. These findings offer potential novel targets for early diagnosis and screening, for providing prognostic information and aiding in the monitoring of disease progression, and for identifying druggable targets for treatment of AMD. This work has the potential to lead us into an era of precision medicine in AMD.

Supplementary Material

Supplementary Methods
Supplementary Figure

Financial Disclosure(s):

The author(s) have made the following disclosure(s): I.L.: Patent - Biomarkers for Age-Related Macular Degeneration

R.S.K.: Patent - Biomarkers for Age-Related Macular Degeneration

R.S.: Consultant - Novartis, Bayer, Thea, Allergan, Alcon, aLIMERA

I.K.K.: Financial support - Genentech, Allergan, Iconic Therapeutics

J.N.M.: Financial support - Alcon

J.L.-S.: Consultant - Metabolon; Patent - Biomarkers of AMD pending

J.W.M.: Consultant - Amgen, Inc., KalVista Pharmaceuticals, Ltd., Maculogix, Inc., Biogen Idec, Inc., Alcon Research Council; Financial support - Lowy Medical Research Institute, Ltd.; Patents - Valeant Pharmaceuticals, ONL Therapeutics LLC

D.H.: Patent - Biomarkers for Age-Related Macular Degeneration

Supported by the Miller Retina Research Fund (to Massachusetts Eye and Ear); the Champalimaud Vision Award (to J.W.M.); Research to Prevent Blindness, Inc., New York, New York (unrestricted departmental grant); and the Portuguese Foundation for Science and Technology/Harvard Medical School Portugal Program (grant no.: HMSP-ICJ/006/2013). None of the aforementioned funding organizations had any role in the design or conduct of this research.

Abbreviations and Acronyms

AMD

age-related macular degeneration

AREDS

Age-Related Eye Disease Study

BMI

body mass index

MS

mass spectrometry

ROC

receiver operating characteristic

RPE

retinal pigment epithelium

Footnotes

Supplemental material available at www.aaojournal.org.

Presented at: Association for Research in Vision and Ophthalmology Annual Meeting, Baltimore, MD, May 2017; Macula Society Annual Meeting, Singapore, June 2017.

References

  • 1.Wong WL, Su X, Li X, et al. Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Glob Heal. 2014;2(2):e106–e116. [DOI] [PubMed] [Google Scholar]
  • 2.Handa JT. How does the macula protect itself from oxidative stress? Mol Aspects Med. 2012;33(4):418–435. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Sobrin L, Seddon JM. Nature and nurture–genes and environment–predict onset and progression of macular degeneration. Prog Retin Eye Res. 2014;40:1–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Yonekawa Y, Miller JW, Kim IK. Age-related macular degeneration: advances in management and diagnosis. J Clin Med. 2015;4(2):343–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Mitta VP, Christen WG, Glynn RJ, et al. C-reactive protein and the incidence of macular degeneration: pooled analysis of 5 cohorts. JAMA Ophthalmol. 2013;131(4):507–513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Hong T, Tan AG, Mitchell P, Wang JJ. A review and meta-analysis of the association between C-reactive protein and age-related macular degeneration. Surv Ophthalmol. 2011;56(3):184–194. [DOI] [PubMed] [Google Scholar]
  • 7.Klein R, Myers CE, Cruickshanks KJ, et al. Markers of inflammation, oxidative stress, and endothelial dysfunction and the 20-year cumulative incidence of early age-related macular degeneration: the Beaver Dam Eye Study. JAMA Ophthalmol. 2014;132(4):446–455. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Uğurlu N, Aşık MD, Yülek F, et al. Oxidative stress and anti-oxidative defence in patients with age-related macular degeneration. Curr Eye Res. 2013;38(4):497–502. [DOI] [PubMed] [Google Scholar]
  • 9.Chakravarthy U, Wong TY, Fletcher A, et al. Clinical risk factors for age-related macular degeneration: a systematic review and meta-analysis. BMC Ophthalmol. 2010;10:31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Klein R, Myers CE, Buitendijk GHS, et al. Lipids, lipid genes, and incident age-related macular degeneration: the Three Continent Age-Related Macular Degeneration Consortium. Am J Ophthalmol. 2014;158(3):513–524.e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Fiehn O Metabolomics—the link between genotypes and phenotypes. Plant Mol Biol. 2002;48(1–2):155–171. [PubMed] [Google Scholar]
  • 12.Fritsche LG, Igl W, Bailey JNC, et al. A large genome-wide association study of age-related macular degeneration highlights contributions of rare and common variants. Nat Genet. 2016;48(2):134–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Nicholson JK, Holmes E, Kinross JM, et al. Metabolic phenotyping in clinical and surgical environments. Nature. 2012;491(7424):384–392. [DOI] [PubMed] [Google Scholar]
  • 14.Nicholson JK, Lindon JC, Holmes E. “Metabonomics”: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica. 1999;11:1181–1189. [DOI] [PubMed] [Google Scholar]
  • 15.Dessì A, Marincola FC, Fanos V. Metabolomics and the great obstetrical syndromes—GDM, PET, and IUGR. Best Pract Res Clin Obstet Gynaecol. 2015;29(2):156–164. [DOI] [PubMed] [Google Scholar]
  • 16.Kelly RS, Virkud Y, Giorgio R, et al. Metabolomic profiling of lung function in Costa-Rican children with asthma. Biochim Biophys Acta Mol Basis Dis. 2017;1863(6):1590–1595. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Wishart DS. Emerging applications of metabolomics in drug discovery and precision medicine. Nat Rev Drug Discov. 2016;15(7):473–484. [DOI] [PubMed] [Google Scholar]
  • 18.Lima AR, Bastos M de L, Carvalho M, Guedes de Pinho P. Biomarker discovery in human prostate cancer: an update in metabolomics studies. Transl Oncol. 2016;9(4):357–370. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Kalita-de Croft P, Al-Ejeh F, McCart Reed AE, et al. ‘Omics approaches in breast cancer research and clinical practice. Adv Anat Pathol. 2016;23(6):356–367. [DOI] [PubMed] [Google Scholar]
  • 20.Junot C, Fenaille F, Colsch B, Bécher F. High resolution mass spectrometry based techniques at the crossroads of metabolic pathways. Mass Spectrom Rev. 2014;33(6):471–500. [DOI] [PubMed] [Google Scholar]
  • 21.Beckonert O, Keun HC, Ebbels TMD, et al. Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nat Protoc. 2007;2(11):2692–2703. [DOI] [PubMed] [Google Scholar]
  • 22.Psychogios N, Hau DD, Peng J, et al. The human serum metabolome. PLoS One. 2011;6(2):e16957. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Suhre K, Gieger C. Genetic variation in metabolic phenotypes: study designs and applications. Nat Rev Genet. 2012;13(11):759–769. [DOI] [PubMed] [Google Scholar]
  • 24.Emwas A-HM. The strengths and weaknesses of NMR spectroscopy and mass spectrometry with particular focus on metabolomics research. In: Bjerrum J, ed. Metabonomics. Methods in Molecular Biology vol 1277. Humana Press, New York, NY: 2015:161–193. Available at: https://link.springer.com/protocol/10.1007%2F978-1-4939-2377-9_13. [DOI] [PubMed] [Google Scholar]
  • 25.Young SP, Wallace GR. Metabolomic analysis of human disease and its application to the eye. J Ocul Biol Dis Infor. 2009;2(4):235–242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Midelfart A Metabonomics—a new approach in ophthalmology. Acta Ophthalmol. 2009;87(7):697–703. [DOI] [PubMed] [Google Scholar]
  • 27.Tan SZ, Begley P, Mullard G, et al. Introduction to metabolomics and its applications in ophthalmology. Eye (Lond). 2016;30(6):773–783. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Osborn MP, Park Y, Parks MB, et al. Metabolome-wide association study of neovascular age-related macular degeneration. PLoS One. 2013;8(8):e72737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.The Age-Related Eye Disease Study system for classifying age-related macular degeneration from stereoscopic color fundus photographs: the Age-Related Eye Disease Study report number 6. Am J Ophthalmol. 2001;132(5):668–681. [DOI] [PubMed] [Google Scholar]
  • 30.Danis RP, Domalpally A, Chew EY, et al. Methods and reproducibility of grading optimized digital color fundus photographs in the Age-Related Eye Disease Study 2 (AREDS2 report number 2). Invest Ophthalmol Vis Sci. 2013;54(7):4548–4554. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Mazzone PJ, Wang X-F, Beukemann M, et al. Metabolite profiles of the serum of patients with non–small cell carcinoma. J Thorac Oncol. 2016;11(1):72–78. [DOI] [PubMed] [Google Scholar]
  • 32.Chi Y-Y, Gribbin MJ, Johnson JL, Muller KE. Power calculation for overall hypothesis testing with high-dimensional commensurate outcomes. Stat Med. 2014;33(5):812–827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Barnes S, Benton HP, Casazza K, et al. Training in metabolomics research. II. Processing and statistical analysis of metabolomics data, metabolite identification, pathway analysis, applications of metabolomics and its future. J Mass Spectrom. 2016;51(8):535–548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Yin P, Peter A, Franken H, et al. Preanalytical aspects and sample quality assessment in metabolomics studies of human blood. Clin Chem. 2013;59(5):833–845. [DOI] [PubMed] [Google Scholar]
  • 35.Xia J, Wishart DS. Using MetaboAnalyst 3.0 for comprehensive metabolomics data analysis. Curr Protoc Bioinforma. 2016;55:14.10.1–14.10.91. [DOI] [PubMed] [Google Scholar]
  • 36.Curcio CA, Millican CL, Bailey T, Kruth HS. Accumulation of cholesterol with age in human Bruch’s membrane. Invest Ophthalmol Vis Sci. 2001;42(1):265–274. [PubMed] [Google Scholar]
  • 37.Curcio CA, Johnson M, Rudolf M, Huang J-D. The oil spill in ageing Bruch membrane. Br J Ophthalmol. 2011;95(12):1638–1645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Fliesler SJ, Bretillon L. The ins and outs of cholesterol in the vertebrate retina. J Lipid Res. 2010;51(12):3399–3413. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Pikuleva IA, Curcio CA. Cholesterol in the retina: the best is yet to come. Prog Retin Eye Res. 2014;41:64–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Miller JW. Beyond VEGF. The Weisenfeld Lecture. Invest Ophthalmol Vis Sci. 2016;57(15):6911–6918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Miller JW. Age-related macular degeneration revisited—piecing the puzzle: the LXIX Edward Jackson Memorial Lecture. Am J Ophthalmol. 2013;155(1):1–35.e13. [DOI] [PubMed] [Google Scholar]
  • 42.Wang Y, Wang M, Zhang X, et al. The association between the lipids levels in blood and risk of age-related macular degeneration. Nutrients. 2016;8(10):E663. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Farooqui AA, Horrocks LA, Farooqui T. Glycerophospholipids in brain: their metabolism, incorporation into membranes, functions, and involvement in neurological disorders. Chem Phys Lipids. 2000;106(1):1–29. [DOI] [PubMed] [Google Scholar]
  • 44.Hopiavuori BR, Agbaga M-P, Brush RS, et al. Regional changes in CNS and retinal glycerophospholipid profiles with age: a molecular blueprint. J Lipid Res. 2017;58(4):668–680. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Farooqui AA, Horrocks LA, Farooqui T. Interactions between neural membrane glycerophospholipid and sphingolipid mediators: a recipe for neural cell survival or suicide. J Neurosci Res. 2007;85(9):1834–1850. [DOI] [PubMed] [Google Scholar]
  • 46.Kosicek M, Hecimovic S. Phospholipids and Alzheimer’s disease: alterations, mechanisms and potential biomarkers. Int J Mol Sci. 2013;14(1):1310–1322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Farooqui AA. Lipid mediators and their metabolism in the nucleous: implications for Alzheimer’s disease. J Alzheimers Dis. 2012;30(Suppl 2):S163–S178. [DOI] [PubMed] [Google Scholar]
  • 48.Frisardi V, Panza F, Seripa D, et al. Glycerophospholipids and glycerophospholipid-derived lipid mediators: a complex meshwork in Alzheimer’s disease pathology. Prog Lipid Res. 2011;50(4):313–330. [DOI] [PubMed] [Google Scholar]
  • 49.Mapstone M, Lin F, Nalls MA, et al. What success can teach us about failure: the plasma metabolome of older adults with superior memory and lessons for Alzheimer’s disease. Neurobiol Aging. 2017;51:148–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Whiley L, Sen A, Heaton J, et al. Evidence of altered phosphatidylcholine metabolism in Alzheimer’s disease. Neurobiol Aging. 2014;35(2):271–278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Proitsi P, Kim M, Whiley L, et al. Association of blood lipids with Alzheimer’s disease: a comprehensive lipidomics analysis. Alzheimers Dement. 2017;13(2):140–151. [DOI] [PubMed] [Google Scholar]
  • 52.Mapstone M, Cheema AK, Fiandaca MS, et al. Plasma phospholipids identify antecedent memory impairment in older adults. Nat Med. 2014;20(4):415–418. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Kaarniranta K, Salminen A, Haapasalo A, et al. Age-related macular degeneration (AMD): Alzheimer’s disease in the eye? J Alzheimers Dis. 2011;24(4):615–631. [DOI] [PubMed] [Google Scholar]
  • 54.Ohno-Matsui K Parallel findings in age-related macular degeneration and Alzheimer’s disease. Prog Retin Eye Res. 2011;30(4):217–238. [DOI] [PubMed] [Google Scholar]
  • 55.Sivak JM. The aging eye: common degenerative mechanisms between the Alzheimer’s brain and retinal disease. Invest Opthalmol Vis Sci. 2013;54(1):871–880. [DOI] [PubMed] [Google Scholar]
  • 56.Ermilov VV, Nesterova AA. β-amyloidopathy in the pathogenesis of age-related macular degeneration in correlation with neurodegenerative diseases. Adv Exp Med Biol. 2016;854:119–125. [DOI] [PubMed] [Google Scholar]
  • 57.Giusto NM, Pasquaré SJ, Salvador GA, Ilincheta de Boschero MG. Lipid second messengers and related enzymes in vertebrate rod outer segments. J Lipid Res. 2010;51(4):685–700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Delgado R, Muñoz Y, Peña-Cortés H, et al. Diacylglycerol activates the light-dependent channel TRP in the photosensitive microvilli of Drosophila melanogaster photoreceptors. J Neurosci. 2014;34(19):6679–6686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Horrocks LA, Yeo YK. Health benefits of docosahexaenoic acid (DHA). Pharmacol Res. 1999;40(3):211–225. [DOI] [PubMed] [Google Scholar]
  • 60.Frank RN. Potential new medical therapies for diabetic retinopathy: protein kinase C inhibitors. Am J Ophthalmol. 2002;133(5):693–698. [DOI] [PubMed] [Google Scholar]
  • 61.Klein J Membrane breakdown in acute and chronic neurodegeneration: focus on choline-containing phospholipids. J Neural Transm. 2000;107(8–9):1027–1063. [DOI] [PubMed] [Google Scholar]
  • 62.Walter A, Korth U, Hilgert M, et al. Glycerophosphocholine is elevated in cerebrospinal fluid of Alzheimer patients. Neurobiol Aging. 2004;25(10):1299–1303. [DOI] [PubMed] [Google Scholar]
  • 63.Nicholson JK, Wilson ID. Opinion: understanding “global” systems biology: metabonomics and the continuum of metabolism. Nat Rev Drug Discov. 2003;2(8):668–676. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Methods
NIHMS959525-supplement-Supplementary_Methods.pdf (76.8KB, pdf)
Supplementary Figure
NIHMS959525-supplement-Supplementary_Figure.pdf (127.6KB, pdf)

RESOURCES