Plasma Metabolomics Reveals Metabolic Profiling For Diabetic Retinopathy and Disease Progression

Abstract

Backgrounds: Diabetic retinopathy (DR), the main retinal vascular complication of DM, is the leading cause of visual impairment and blindness among working-age people worldwide. The aim of this study was to investigate the difference of plasma metabolic profiles in patients with DR to better understand the mechanism of this disease and disease progression.

Methods: We used ultrahigh-performance liquid Q-Exactive mass spectrometry and multivariate statistical analyses to conduct a comprehensive analysis of plasma metabolites in a population with DR and proliferative DR (PDR). A risk score based on the level of the selected metabolite was established and evaluated using the least absolute shrinkage and selection operator regularization logistic regression (LASSO-LR) based machine learning model.

Results: 22 differentially expressed metabolites which belonged to different metabolic pathway were identified and confirmed to be associated with the occurrence of DR. A risk score based on the level of the selected metabolite pseudouridine was established and evaluated to strongly associated with the occurrence of DR. Four circulating plasma metabolites (pseudouridine, glutamate, leucylleucine and N-acetyltryptophan) were identified to be differentially expressed between patients with PDR and other patients, and a risk score formula based on these plasma metabolites was developed and assessed to be significantly related to PDR.

Conclusions: Our work highlights the possible use of the risk score assessment based on the plasma metabolites not only reveal in the early diagnosis of DR and PDR but also assist in enhancing current therapeutic strategies in the clinic.

Keywords: biomarkers; diabetes mellitus; diabetic retinopathy; machine learning; plasma metabolomics.