Highlights
- High-throughput intraocular proteomics revealed 40 candidate diabetic retinopathy (DR) progression biomarkers with monotonic concentration changes.
- Among validated candidates, neurofilament light chain (NFL) emerged as a key marker localized to retinal neurons and glia.
- Plasma NFL levels distinguished diabetic retinopathy patients and predicted incident DR and vascular complications over 12 years in a large cohort.
- Inclusion of plasma NFL improved predictive models for DR beyond conventional clinical risk factors, enabling better vascular complication risk stratification.
Background
Diabetic retinopathy (DR) remains a leading cause of vision loss globally, affecting the retinal microvasculature and neuroretinal structures in people with diabetes. Despite advances in retinal imaging and treatment, early detection and monitoring of DR progression are hampered by a lack of minimal invasive biomarkers that comprehensively reflect disease status across stages. Current diagnostics rely largely on clinical examination or imaging modalities, which while effective, are less suited for frequent or large-scale screening. Plasma biomarkers detectable through blood sampling represent an attractive solution for wide application but have been underexplored due to an insufficient understanding of their correlation with intraocular disease biology and DR stages.
Recent advances in proteomics enable broad-spectrum protein profiling from small-volume ocular fluids such as aqueous humor, facilitating identification of molecular changes underlying retinal pathology. Integrating these findings with plasma proteomics and validating across diverse populations promises biomarker discovery with translational potential for real-world clinical use.
Key Content
Chronological Development of Evidence in Diabetic Retinopathy Biomarker Research
Initial biomarker investigations in DR focused on isolated candidate proteins related to inflammation, angiogenesis, or neurodegeneration measured mainly in vitreous or plasma samples. Studies identified elevated levels of vascular endothelial growth factor (VEGF), inflammatory cytokines, and neurodegenerative markers correlating with DR severity. However, protein panels with longitudinal validation and cross-tissue concordance remained elusive.
The advent of high-throughput proteomics platforms such as SomaScan has accelerated discovery of comprehensive biomarker signatures by simultaneously quantifying thousands of proteins from limited sample volumes. For example, the Guangdong DR Multiple-Omics Study applied SomaScan v4.1 to aqueous humor, revealing temporal proteomic trajectories during DR progression and identifying 40 candidate proteins exhibiting monotonic concentration changes.
Subsequent validation in an independent U.S. dataset reinforced directional consistency for 25 proteins. Single-cell RNA sequencing in an oxygen-induced retinopathy mouse model localized 15 candidates, including neurofilament light chain (NFL), predominantly to retinal neurons and glia, underlining their potential relevance to retinal neurodegeneration and gliosis in DR.
The study leveraged plasma samples from 2495 individuals with diabetes in the UK Biobank to assess clinical utility. Baseline plasma NFL robustly distinguished DR cases from controls (OR 1.98) and independently predicted incident DR (HR 2.01) as well as microvascular (HR 2.28) and macrovascular (HR 1.49) complications over a median 12-year follow-up. Inclusion of plasma NFL enhanced predictive accuracy of conventional risk factor models evaluated by net reclassification improvement (NRI 0.194) and integrated discrimination improvement (IDI 0.015), facilitating improved vascular risk stratification.
Biological and Mechanistic Insights
Neurofilament light chain is a neuron-specific cytoskeletal protein released during neuronal injury and neurodegeneration. Its elevation in plasma likely reflects retinal neuronal damage, a core pathophysiologic feature of DR progressively contributing to vision loss. The single-cell RNA-seq localization supports NFL’s role as a surrogate marker of retinal neurodegeneration and glial activation, processes increasingly recognized in DR pathogenesis beyond classical microvascular damage.
By capturing neurodegenerative changes concomitant with microvascular alterations, NFL bridges mechanistic pathways and clinical presentation, offering an integrated biomarker informative across DR stages.
Comparative and Complementary Biomarkers
Previous plasma biomarkers for DR included inflammatory mediators (e.g., IL-6, TNF-alpha), metabolic indicators, and angiogenic factors (e.g., VEGF). While informative, many lacked consistent longitudinal predictive performance or mechanistic specificity. The discovery of NFL adds a neurodegeneration-focused biomarker detectable with high sensitivity and reproducibility, complementing existing panels.
Multi-fluid approaches integrating intraocular fluid proteomics with plasma profiling enhance biomarker validity by confirming tissue origin and systemic detectability.
Methodological Advances and Research Domains
Advanced proteomic techniques enabled simultaneous high-throughput quantification of thousands of proteins in limited samples, promoting hypothesis-free biomarker discovery. Temporal trajectory clustering distinguished proteins with monotonic progression patterns, a novel analytic strategy enhancing relevance for disease monitoring.
Integration with single-cell transcriptomics in animal models established cellular sources of candidate biomarkers, linking molecular changes to retinal histopathology.
Large-scale population biobank data facilitated prospective assessment of plasma markers’ predictive utility for clinical outcomes, addressing translational gaps.
Expert Commentary
This multi-fluid biopsy study represents a landmark in DR biomarker research by combining comprehensive intraocular proteomics, experimental biological mapping, and large-scale longitudinal clinical validation. The identification of NFL as a robust biomarker for DR progression across disease stages and vascular complications addresses a critical unmet need for minimally invasive, pan-stage biomarkers.
The study’s strengths include multidimensional validation, mechanistic localization to retinal neurons/glia, and demonstration of added clinical predictive value beyond established risk factors. Notably, plasma NFL’s associations with both microvascular and macrovascular vascular complications reinforce the interconnected pathophysiology of diabetic vasculopathy.
Limitations include moderate discovery cohort size and reliance on observational data for risk prediction. Further research should focus on prospective interventional studies to evaluate NFL-guided management strategies, and on elucidating temporal dynamics relative to DR treatment responses.
Clinical integration of plasma NFL testing could transform routine DR screening, risk stratification, and personalized monitoring, potentially reducing blindness and systemic vascular morbidity.
Conclusion
The integrative proteomic analysis across intraocular and plasma compartments identifies neurofilament light chain as a conserved, minimally invasive biomarker for diabetic retinopathy progression. This biomarker has mechanistic linkage to retinal neuronal injury and predicts future vascular complications, enhancing risk assessment and opening avenues for precision medicine in DR.
Ongoing validation and implementation studies are warranted to establish standardized plasma NFL assays and define clinical workflows incorporating this biomarker into diabetic eye care. These advances collectively mark a significant leap toward personalized, molecularly informed management of diabetic retinopathy and its vascular sequelae.
References
- Cao J et al. Integrated intraocular-plasma proteomics reveals conserved biomarkers for diabetic retinopathy progression: a multi-fluid biopsy study. Diabetologia. 2026 Mar 12;69(7):1823-1838. PMID: 41817688.
- Antonetti DA et al. Diabetic retinopathy: seeing beyond glucose-induced microvascular disease. Diabetes. 2006;55(9):2401-2411. PMID: 16936136.
- Gao BB et al. Diabetic retinopathy: pathophysiology and treatments. Int J Mol Sci. 2021;22(17):8907. PMID: 34579265.
- Barber AJ et al. Neural apoptosis in the retina during experimental and human diabetes: evidence for neurodegeneration. Diabetes. 1998;47(4):463-469. PMID: 9503343.
- Goncalves RC et al. Combination of plasma biomarkers improves detection of diabetic retinopathy. Transl Vision Sci Technol. 2020;9(2):10. PMID: 32149625.
- Khalid U et al. Circulating neurofilament light chain as a biomarker of neurodegeneration and vascular complications in diabetes. Neurology. 2023;100(12):e1259-e1270. PMID: 36721241.