Novel High-Resolution Lipidomes Could Serve as New Biomarkers for Diabetic Retinopathy: A Bidirectional and Mediated Mendelian Randomization Study

Abstract

Although lipid metabolism is a critical factor in the pathogenesis of diabetic retinopathy (DR), the connection between lipidome and DR is still a subject of debate. We aimed to demonstrate that lipidome could serve as novel biomarkers for DR and elucidate the mediating role of inflammatory factors. Data for our investigation are available from the GWAS catalogue and FinnGen Biobank. The bidirectional Mendelian randomization (MR) analyses were conducted to assess the "total effect" between lipidome and DR and its subtypes. Subsequently, the mediation analyses were performed to explore the involvement of circulating inflammatory proteins in mediating the connection between them. Mediation proportion was calculated to measure the contribution of inflammatory factors to the overall effect. Ultimately, a battery of sensitivity tests proceeded to examine the dependability of the findings. This study has revealed a causal relationship between lipidome and different stages of DR. Additionally, we have successfully discovered a range of new lipids that protect against DR and have the potential to serve as new markers. This study also highlights the important role of inflammatory factors in elucidating the protective mechanisms of lipids against DR and provides new perspectives on lipidomic-based treatments and cytokine-targeted interventions for DR.

Keywords: Mendelian randomization; circulating inflammatory protein; diabetic retinopathy; interleukin‐10; lipidome; mediation analysis; shotgun lipidomics technique; triacylglycerol.

FIGURE 1

The design and flow chart of the investigation. The research aimed to determine whether circulating proteins of inflammation (mediators) play a causal role in mediating the effects of lipidome (exposures) on DR, NPDR, and PDR (outcomes). Initially, a two‐sample bidirectional MR analysis was performed to investigate the causal relationship between lipidome and retinopathy. The mediated MR used a two‐step approach: the first step was to determine the causal relationship between lipids and inflammatory proteins in the blood. The next step is to assess the causative influence of circulating inflammatory proteins on DR, NPDR, and PDR.

FIGURE 2

A heatmap displays bidirectional causality between lipidome and DR, NPDR, and PDR. Statistical significance is indicated by an asterisk (*) for a p‐value of < 0.05. The cell's colour represents the β value linked to the pair, with red indicating positive values and blue signifying negative values. BWMR, Bayesian weighted Mendelian randomization; DG, Diacylglycerol; IVW, inverse variance weighted; PC, Phosphatidylcholine; PE, Phosphatidylethanolamine; rMR, reverse Mendelian randomization; SE, Sterol ester; SM, simple mode; SM, Sphingomyelin; TG, Triglyceride; WM1, weighted median; WM2, weighted mode.

FIGURE 3

Causal effects of lipidome on mediators. (a) Heat map illustrating the causative impact of lipids on the levels of inflammatory proteins in the bloodstream. The cell's colour signifies the p‐value relating to the pairing, and the blue colour corresponding to a value of 0 indicates the absence of the causal relationship. (b) Forest plot showing the causality of lipidome on circulating inflammatory proteins using the IVW method. The IVW‐derived OR and 95% CI have been rigorously validated through four complementary analytical approaches (MR‐Egger, weighted median, simple mode, and weighted mode), with full verification details provided in Table S4. β, genetic effect sizes; CI, confidence interval; OR, odds ratio; pval, p value; Q_value, p value for Cochran's Q test; Se, standard error; SNPs, single nucleotide polymorphisms.

FIGURE 4

Forest plot showing the causality of inflammatory proteins (mediators) on outcomes. CI, confidence interval; DR, diabetic retinopathy; IVW, inverse variance weighted; NPDR, non‐proliferative diabetic retinopathy; OR, odds ratio; PDR, proliferative diabetic retinopathy; pval, p value; Q_value, p value for Cochran's Q test; Se, standard error; SNPs, single nucleotide polymorphisms; β, genetic effect sizes.

FIGURE 5

Mediation analysis of two‐step Mendelian randomization. The diagram illustrates the involvement of circulating inflammatory proteins in mediating the causative effect of exposures on outcomes. β, the total causal effect of exposures with outcomes; β1, the effect of exposures on mediators; β2, the impact of mediators on outcomes; DR, diabetic retinopathy; NPDR, non‐proliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy.