Cross-Compartment Proteomic Signatures in Human Diabetic Retinopathy: A Systematic Review and Meta-Analysis.
Soumya Behera, Nibedita Sahoo, Subhangi Sahu
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis synthesized proteomic findings from 28 human diabetic retinopathy studies spanning vitreous, aqueous, plasma or serum, tears, and urine. Random-effects quantitative synthesis found overall protein upregulation in diabetic retinopathy (Log2FC = 1.49; 95% CI: 0.72-2.27). Effects were stronger in vitreous (Log2FC = 2.41) and aqueous (Log2FC = 1.28) than in plasma, where estimates were weaker and more heterogeneous. Fibrinogen chains were consistently upregulated across compartments, while qualitative findings identified additional stage-specific candidates and potential noninvasive markers.
Key findings
- Across studies, protein levels were higher in diabetic retinopathy (Log2FC = 1.49; 95% CI: 0.72-2.27).
- Subgroup estimates were stronger in vitreous (Log2FC = 2.41) and aqueous (Log2FC = 1.28); plasma findings were weaker and more heterogeneous.
- Fibrinogen chains (FGA, FGB, FGG) were robustly upregulated across compartments and exceeded complement proteins (β = +1.41; p < 0.001).
- After trim-and-fill adjustment, the effect was reduced but remained significant (Log2FC = 1.10); tears and urine showed potential for noninvasive screening.
Why it matters
The review brings together evidence from multiple biological compartments, identifying fibrinogen chains as a recurring signal rather than a finding confined to one sample type. This cross-compartment consistency, alongside candidate stage-specific proteins and possible tear- or urine-based markers, helps prioritize targets for validation. The findings support further biomarker research but do not establish diagnostic performance or clinical utility.
What to keep in mind
The evidence combines studies across different biological compartments, and plasma estimates were weaker and more heterogeneous. Publication-bias adjustment reduced the overall effect, indicating that the pooled estimate may be sensitive to reporting bias.