Introduction

    Diabetic retinopathy (DR) is one of the most common microvascular complications of diabetes and a primary cause of blindness among working-age adults [1]. In 2020, 103.12 million individuals lived with DR worldwide, cor-responding to a global prevalence of 22.27% among people with diabetes [1]. Prevalence is expected to increase; by 2045, an estimated 160.50 million adults worldwide will be afflicted by DR, and 44.82 million will experience vision-threatening DR [1].

    Early detection is necessary for reducing DR progres-sion. Despite major risk factors such as hyperglycemia, hypertension, and dyslipidemia, considerable variability in DR onset and progression remains unexplained. This sug-gests a need to explore pathophysiological mechanisms and biomarker identification. Emerging evidence from metabo-lomics—the study of small-molecule metabolites in biologi-cal systems—has shown promise in uncovering molecular alterations associated with diabetic complications, includ-ing DR [2]. Previous studies identified metabolites like 12-hydroxyeicosatetraenoic acid (12-HETE) and 3,4-dihy-droxybutyrate (3,4-DHBA) as DR-related [2], pointing to inflammation and altered lipid metabolism. However, few studies include healthy controls (HC) in addition to DR and no-DR groups [2]. Recognizing the need for more compre-hensive designs, this study uses untargeted metabolomics to examine DR, no-DR, and HC groups to identify biomarkers and gain insight into DR pathogenesis.

Materials and methods

    We analyzed baseline fasting serum samples from 77 par- ticipants aged 40–65 years, drawn from a cohort who had previously enrolled in sleep and circadian rhythm studies conducted at a tertiary academic hospital. Post-illumination pupillary light reflex (PIPR), a measure of intrinsically photosensitive retinal ganglion cells (ipRGCs) located in the inner retina, and nocturnal urinary 6-sulfatoxymelato-nin (aMT6s); both previously shown to be reduced in DR, were collected using standardized protocols, previously described[3] Participants included 26 T2D without DR (no-DR), 36 T2D with moderate-severe nonproliferative DR, and 15 HCs. DR severity was graded per the Early Treatment Diabetic Retinopathy Study classification [4]. Inclusion required a recent ophthalmologic evaluation. Key exclusion criteria included ocular disease unrelated to dia-betes, systemic conditions affecting the retina, and recent melatonin or illicit drug use. Obstructive sleep apnea, known to be associated with DR [5], was assessed using the apnea-hypopnea index (AHI) via an overnight home diag nostic device, the WatchPAT 200/300®.

    Blood samples were collected following an overnight fast and assayed for hemoglobin A1C (HbA1c), lipids and serum creatinine (Quest Diagnostics), and serum was stored at -80 °C until analyzed. Metabolomic profiling was con-ducted using an untargeted LC-MS (Agilent 6545 Q-TOF, 1290 UPLC system). Raw data files were processed using the Profinder software (vB.10.00, Agilent Technologies). Features were extracted and filtered based on quality met-rics and retention time. Metabolite data were normalized and analyzed using the limma package with empirical Bayes adjustment, including covariates (age, sex, BMI, blood pressure, LDL, triglycerides, serum creatinine, AHI, smoking status, medication use). Exploratory correlations between serum metabolites and urinary aMT6s and PIPR were assessed using biweight midcorrelation while control-ling for age, sex, BMI, and pAHI. Adjusted p-values (q-val-ues) were calculated using the Benjamini–Hochberg false discovery rate correction. Log-transformed, scaled data were subjected to principal component analysis (PCA) in R to explore global metabolic differences, with plots dis-playing the first two components for DR, no-DR, and HCs. Pathway enrichment between DR and no-DR was explored using Ingenuity Pathway Analysis (IPA; Qiagen). Data are presented as mean (SD), median (IQR), or frequency (%), and group comparisons were performed using Wilcoxon rank-sum, Welch’s t-test, or Chi-square as appropriate.

Results

    Table 1 shows participant demographics, which were com-parable across DR and no-DR groups in age, sex, BMI, and lipid levels. The DR group had longer diabetes duration and greater insulin use than the no-DR group, consistent with disease severity. Systolic blood pressure was higher in DR

participants while urinary aMT6s and PIPR were lower. HCs demonstrated less medication use, lower triglycerides, blood pressure, HbA1c, and fewer AHI events than the total T2D group (DR and no-DR).

    PCA showed partial clustering of three groups, with DR samples tending to separate from no-DR and HCs, although some overlap remained (Fig. 1). No-DR and HCs largely overlapped, underscoring that the most pronounced metabo-lomic differences were between DR and the other groups. IPA revealed no significantly enriched metabolic pathways, and no metabolites were correlated with either PIPR or uri-nary aMT6s in nominal or adjusted analyses. 

    Differential abundance analysis identified 387 metabo-lites significantly altered between DR and no-DR, while no differences were observed between total T2D and HCs. Filtering for log2 fold change≥1.5 yielded 58 metabolites, with candidate biomarkers summarized in Table 2 and the full list provided in Supplementary Table S1. A heatmap of their relative abundances (Supplementary Figure S1) illus-trates differences between DR and no-DR. Comparison of DR versus HCs showed 61 significantly altered metabolites, all of which overlapped with differentially abundant metab-olites between DR and no-DR (Supplemental Table S2).

    Findings included higher 5-tetradecenoylcarnitine and phosphatidylcholine (24:1/24:1) and lower long-chain fatty acids (6-hydroxypentadecanedioic acid (15:0) and 10,16-dihydroxy-palmitic acid (16:0)). Monomethyl phthal-ate, a metabolite linked to plastic exposure and endocrine disruption, was significantly elevated in DR. Two metabo-lites, 12-HETE and 3,4-DHBA, previously implicated in DR pathogenesis, were also significantly higher, although not above the log2 fold change≥1.5 threshold.

Discussion

Our findings suggest that DR is associated with metabolic alterations beyond those seen in T2D alone. Unsupervised PCA revealed modest group separation between DR versus no-DR and HCs, supporting global metabolic differences in DR Metabolites differentially abundant in DR versus HCs completely overlapped with those identified in DR versus no-DR, supporting the robustness of these DR-associated metabolic alterations.

    We observed elevated levels of 5-tetradecenoylcarnitine. Although prior studies have reported mixed findings regard- ing acylcarnitine levels in DR, our results add to this evolv-ing literature. Concurrent reductions in long-chain fatty acids in DR may reflect broader disruptions in lipid metabo-lism, though mechanistic interpretations remain specula-tive. The observed increases in 12-HETE and 3,4-DHBA, help validate our findings, as they are well documented in metabolomic studies and agree with previous literature [2]. Both metabolites have been found to be independent risk markers for DR progression, thus making them possible therapeutic targets [2].

    Of interest, monomethyl phthalate exhibited significant upregulation in DR versus no-DR and HCs, consistent with its potential role as a risk factor for diabetes as reported in meta-analysis [6]. Since monomethyl phthalate, a chemical derived from plastics, has been acknowledged as an endo-crine disruptor correlated with increased serum TNFa and decreased adiponectin levels [6], it suggests a link between inflammation and the development of diabetic complica-tions such as DR.

pRGC function is known to be lower in DR [3, 5], though no significant correlations were observed between metabo-lites and PIPR, suggesting that metabolic alterations seen in DR may not be secondary to ipRGC dysfunction. Similarly, negative findings were found between metabolites and uri-nary aMT6s. These findings are hypothesis-generating and should be interpreted cautiously given study limitations. A strength of this study is the inclusion of HCs, which allowed differentiation of metabolic changes specifically associated with DR from those related more generally to T2D. This three-group design offered a broader perspective on disease-specific perturbations. This study is limited by its smaller sample size, cross-sectional nature, and potential residual confounding despite covariate adjustment. None-theless, the observed metabolic differences warrant fur-ther investigation in larger, prospective cohorts to validate potential biomarkers and clarify future avenues of research.

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00592-0 26-02678-5.

Author contributions M.S.: analyzed data, wrote, reviewed and edited the manuscript. L.Y., B.T.L.: wrote, reviewed and edited the manu-script. J.J.M, J.P, F.C., B.P., S.P., E.C.H., E.V.C., K.K.D., S.J.C.: re-searched data, reviewed and edited the manuscript. G.E.C, C.M., H.C.: analyzed data, wrote, reviewed and edited the manuscript. S.R.: conceptualization, researched and analyzed data, wrote, reviewed and edited the manuscript.

Funding This study was funded by the National Eye Institute (R01EY029782, R01EY026004, P30EY001792), NIDDK P30 DK020595, and The Center for Clinical and Translational Science (CCTS) UL1TR002003.

Data availability The data that support this study’s findings are not openly available due to participant consent limitations and are avail-able upon reasonable requests to the authors. Data are located in the Research Resources Center (RRC) data portal at the University of Illi-nois at Chicago, which is a controlled access data storage.

Declarations

Conflict of interest S.R. received speaker fees from Eli Lilly. S.P. served on advisory boards for Novo Nordisk and Eli Lilly. Other au-thors report no conflicts of interest.

Ethical approval The protocol was approved by the Institutional Re-view Board at the University of Illinois at Chicago and adhered to the tenets of the Declaration of Helsinki. No animals were involved in the study.

Consent to participate Written informed consent to participate was obtained by all participants.

Trial Registration Part of the data came from the clinical study, reg-istered at www.clinicaltrials.gov, NCT04547439, date of registration 9/14/2020.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.o rg/licenses/by/4.0/.

References

1. Teo ZL, Tham YC, Yu M et al (2021) Global Prevalence of Diabetic Retinopathy and Projection of Burden through 2045: Systematic Review and Meta-analysis. Ophthalmology 128(11):1580–1591. https://doi.org/10.1016/j.ophtha.2021.04.027

2. Jian Q, Wu Y, Zhang F (2022) Metabolomics in Diabetic Retinop-athy: From Potential Biomarkers to Molecular Basis of Oxidative Stress. Cells 11(19):3005. https://doi.org/10.3390/cells11193005

3. Reutrakul S, Crowley SJ, Park JC et al (2020) Relationship between Intrinsically Photosensitive Ganglion Cell Function and Circadian Regulation in Diabetic Retinopathy. Sci Rep 10(1):1560. https://doi.org/10.1038/s41598-020-58205-1

4. Davis MD, Fisher MR, Gangnon RE et al (1998) Risk factors for high-risk proliferative diabetic retinopathy and severe visual loss: Early Treatment Diabetic Retinopathy Study Report #18. Invest Ophthalmol Vis Sci 39(2):233–252

5. Simonson M, Li Y, Zhu B et al (2024) Multidimensional sleep health and diabetic retinopathy: Systematic review and meta-analysis. Sleep Med Rev 74:101891. https://doi.org/10.1016/j.sm rv.2023.101891 Zhang H, Ben Y, Han Y, Zhang Y, Li Y, Chen X (2022) Phthalate exposure and risk of diabetes mellitus: implications from a sys tematic review and meta-analysis. Environ Res 204(Pt B):112109. https://doi.org/10.1016/j.envres.2021.112109

Publisher’s Note Springer Nature remains neutral with regard to juris-dictional claims in published maps and institutional affiliations.

  This article is excerpted from the 《Acta Diabetologica》 by Wound World.