Predictive model for proliferative diabetic retinopathy using single-cell transcriptomics.

Area of Science:

• Ophthalmology
• Genomics
• Computational Biology

Background:

• Proliferative diabetic retinopathy (PDR) is an advanced complication of diabetes, a leading cause of blindness.
• Fibrovascular membranes (FVMs) in PDR contract and cause retinal traction, severely impacting vision.
• Understanding PDR's molecular basis is crucial for effective treatment strategies.

Purpose of the Study:

• To investigate the cellular and molecular mechanisms underlying fibrovascular membrane (FVM) formation in proliferative diabetic retinopathy (PDR).
• To identify key cell types and molecular pathways involved in PDR pathogenesis.
• To develop a machine learning model for PDR detection and monitoring.

Main Methods:

• Integrated scRNA-seq data from PDR patient FVMs and healthy retinas.
• Performed differential gene expression analysis on astrocytes, microglia, and rod cells.
• Utilized Gene Ontology (GO) and KEGG pathway analyses.
• Developed and validated a machine learning-based predictive model.

Main Results:

• Identified distinct gene expression patterns in astrocytes, microglia, and rod cells from PDR patients.
• GO and KEGG analyses revealed specific pathway enrichments in PDR.
• The predictive model achieved high AUC values (0.81-0.96) in test and validation cohorts.

Conclusions:

• This study provides novel insights into the cellular and molecular drivers of FVM development in PDR.
• The findings highlight potential therapeutic targets for PDR.
• A validated machine learning model shows promise for PDR detection and monitoring.