Plasma NFL Biomarker Predicts Diabetic Retinopathy from Blood

SNIPPET: Plasma neurofilament light chain (NFL) is a blood-detectable biomarker that distinguishes diabetic retinopathy across all stages, with a hazard ratio of 2.01 for predicting incident disease over 12 years. Combined with proteomic panels including plexin B2, GDF15, and renin, minimally invasive blood tests may soon replace or supplement retinal imaging for early diabetic eye disease detection.

THE PROTOHUMAN PERSPECTIVE#

Here's what most people with diabetes don't realize: by the time you notice vision changes, retinal damage is already well underway. Diabetic retinopathy affects roughly one-third of all people with diabetes globally, and current screening depends on periodic eye exams that many patients skip or delay. The gap between what's happening inside your retina and what a standard blood panel tells your doctor has been, frankly, enormous.

That gap may be closing. A convergence of three recent proteomic studies — spanning Chinese, US, and UK populations — points toward a future where a simple blood draw flags retinal neurodegeneration years before symptoms appear. For anyone tracking biomarkers as part of a longevity or metabolic health protocol, this changes the screening calculus entirely. We're not talking about replacing your ophthalmologist. We're talking about knowing when to see one — and knowing it from a tube of blood rather than a dilated eye exam you keep postponing.

The implications extend beyond the eye. NFL and its companion markers are tied to broader microvascular and macrovascular risk, which means this data touches metabolic optimization at its core.

THE SCIENCE#

What Proteomics Actually Found#

Diabetic retinopathy (DR) is the progressive destruction of retinal microvasculature and neurons driven by chronic hyperglycemia. It is the leading cause of blindness in working-age adults in developed countries, and up to 90% of DR-related vision loss is preventable with early intervention^[3]. The problem has always been detection — specifically, the absence of blood-based markers that track disease across its full continuum. Three recent proteomic studies have converged on a potential solution.

The headline finding comes from a March 2026 multi-fluid biopsy study published in Diabetologia^[1]. Researchers from the Guangdong DR Multiple-Omics Study used SomaScan v4.1 high-throughput proteomics on aqueous humour samples from 32 patients across diabetic retinopathy stages. They identified 40 candidate proteins showing monotonic concentration changes as DR progressed. Of these, 25 showed conserved directional patterns when validated in a separate US dataset — 17 increasing and 8 decreasing, all at p<0.05.

Single-cell RNA sequencing in an oxygen-induced retinopathy mouse model localized 15 of these candidates, including NFL, to retinal neurons and glia. Which is annoying, actually — because it means this signal originates from actual neural tissue damage, not just vascular leakage. That's a different biology than what most DR biomarker research has focused on.

NFL in the UK Biobank: The Numbers That Matter#

The clinical validation happened in 2,495 individuals with diabetes from the UK Biobank. Baseline plasma NFL distinguished those with DR from controls with an odds ratio of 1.98 (95% CI: 1.61–2.42). More importantly for early detection, plasma NFL predicted incident diabetic retinopathy with a hazard ratio of 2.01 (95% CI: 1.48–2.73) over a median 12-year follow-up^[1].

The data extends beyond the retina. Elevated plasma NFL was associated with microvascular complications broadly (HR 2.28, 95% CI: 1.94–2.69) and macrovascular complications (HR 1.49, 95% CI: 1.26–1.77). Adding NFL to a conventional risk factor model improved net reclassification by 0.194 (95% CI: 0.042–0.297).

I want to be careful here. An NRI of 0.194 is statistically significant but modest in absolute terms. It means NFL meaningfully reclassifies some patients into correct risk categories, but it's not replacing HbA1c and fundoscopy tomorrow. It's an incremental gain — a clinically useful one, but let's not overstate it.

The Three-Protein Panel: Plexin B2, GDF15, and Renin#

A complementary study published in Nature Communications in October 2025 profiled approximately 3,000 proteins in 10,873 individuals with diabetes or prediabetes across two ethnically distinct cohorts^[3]. They identified 668 proteomic associations with DR, but the key finding was a parsimonious three-protein panel: plexin B2, growth differentiation factor 15 (GDF15), and renin.

This panel alone achieved predictive performance comparable to the full proteomic profile. All three proteins were validated across cohorts and linked to retinal microvascular degeneration via swept-source OCTA imaging in the Guangzhou Diabetic Eye Study. Renin was further confirmed as a causal promoter of DR through Mendelian randomization — which is the strongest evidence tier short of a randomized trial for establishing causality from observational data.

The renin finding is particularly interesting for anyone already monitoring their RAAS (renin-angiotensin-aldosterone system) activity. It suggests that the vascular damage pathway in DR may be more tightly coupled to systemic blood pressure regulation than previously appreciated at the proteomic level.

Vitreous Fluid Tells a Different Story#

A third study, published in Scientific Reports in February 2026, took a complementary approach — analyzing undiluted vitreous fluid from 8 PDR patients and 6 controls using TMT-tagged proteomics combined with untargeted metabolomics^[2]. The sample size is small, so I'd frame these findings as hypothesis-generating rather than confirmatory.

They identified CD5L as upregulated in proliferative DR, with functional validation showing it promotes endothelial cell proliferation and migration. CLU (clusterin) and SERPINF1 (PEDF) were downregulated. The pathways implicated — complement and coagulation cascades, AMPK-related energy dysregulation via creatine metabolism disturbances — overlap with what we know about mitochondrial efficiency decline in diabetic tissues.

But here's where it gets complicated. The vitreous findings and the plasma findings don't map neatly onto each other. Vitreous CD5L upregulation speaks to local proliferative processes in advanced disease. Plasma NFL elevation speaks to neurodegeneration detectable even in early stages. These are different biological stories happening in the same organ, and any clinical protocol needs to account for which question you're actually trying to answer.

COMPARISON TABLE#

THE PROTOCOL#

A proteomic blood panel for DR prediction isn't something you can order from a standard lab yet. But the data points toward actionable steps for anyone with diabetes or prediabetes who takes biomarker tracking seriously.

Step 1: Request plasma neurofilament light chain (NFL) testing through your physician or a research-oriented lab. Several clinical labs now offer NFL assays (originally developed for neurodegenerative disease monitoring). Establish a baseline value. Interpret any elevation in context — NFL rises with age and neurological conditions, so a single high reading doesn't confirm retinal damage.

Step 2: Track GDF15 if you have access to expanded proteomic panels. GDF15 is increasingly available through longevity-oriented blood testing services. Elevated GDF15 in the context of diabetes may indicate broader mitochondrial stress and microvascular risk beyond what HbA1c captures alone.

Step 3: Monitor your renin levels, particularly if you have hypertension or are on ACE inhibitors/ARBs. The Mendelian randomization data from the Nature Communications study^[3] confirms renin as a causal contributor to DR. If your renin is elevated and you have diabetes, this is a conversation to have with your physician about RAAS modulation — not a signal to self-medicate.

Step 4: Maintain strict adherence to dilated eye exams at the intervals your ophthalmologist recommends. Blood biomarkers may enhance risk stratification, but they do not replace direct retinal visualization. The data shows NFL improves net reclassification by about 19% — useful, but not sufficient on its own.

Step 5: If you're tracking metabolic health biomarkers already (fasting glucose, insulin, HbA1c, lipid panel, hsCRP), consider adding NFL and GDF15 to your annual or biannual panel. The longitudinal trend matters more than any single value — the UK Biobank data used baseline measurements, but in practice, watching the trajectory over 2–3 years gives you signal that a snapshot cannot.

Step 6: For those with proliferative DR already diagnosed, discuss with your retinal specialist whether any of the emerging pathway targets (complement modulation, AMPK activation) are relevant to your treatment plan. This is firmly in the "ask your doctor" category, not a self-optimization play.

What is neurofilament light chain and why does it matter for diabetic retinopathy?#

Neurofilament light chain (NFL) is a structural protein found in neurons. When retinal neurons and glial cells are damaged — as happens progressively in diabetic retinopathy — NFL is released into the aqueous humour and eventually into the bloodstream. According to Lai et al. (2026), plasma NFL can distinguish DR patients from controls and predict new-onset DR with a hazard ratio of 2.01 over 12 years^[1]. It was previously known mainly as a marker for multiple sclerosis and Alzheimer's; its role in DR is a newer finding.

How does the three-protein panel compare to standard retinal screening?#

The plexin B2/GDF15/renin panel identified by the Nature Communications proteome atlas achieved predictive performance comparable to a full ~3,000-protein profile, which is impressive for parsimony^[3]. However, it hasn't been tested head-to-head against fundoscopy in a clinical screening trial. The honest answer is we don't know yet whether it could reduce the need for eye exams or merely supplement them. I'd bet on supplementation, at least for the next decade.

Who should consider getting plasma NFL tested?#

Anyone with type 2 diabetes, especially those with suboptimal glycemic control (HbA1c consistently above 7%), a family history of diabetic complications, or who have missed recommended eye screening appointments. NFL is also relevant if you're already tracking neurodegeneration markers for longevity purposes. The key limitation: NFL isn't DR-specific, so elevated values need clinical context.

When might proteomic DR screening become clinically available?#

The technology (SomaScan, Olink) exists and is commercially available for research. Clinical adoption depends on regulatory approval of specific panels, standardized reference ranges, and health economic analyses proving cost-effectiveness. I'd estimate 3–5 years for specialist adoption and longer for primary care integration. The UK Biobank validation is a strong signal, but it's not a clinical trial endpoint.

Why is renin's causal role in diabetic retinopathy significant?#

Most DR biomarker candidates are associative — they correlate with disease but may not drive it. Renin was confirmed as a causal promoter through Mendelian randomization, meaning genetic variants that raise renin levels also increase DR risk independently of confounders^[3]. This matters because it suggests that RAAS-targeted therapies (ACE inhibitors, ARBs) may have direct retinal protective effects beyond blood pressure control — a hypothesis that existing trial data partially supports but hasn't definitively proven.

VERDICT#

Score: 7.5/10

The convergence of three independent proteomic studies on blood-detectable DR biomarkers is genuinely encouraging. NFL stands out as the most clinically actionable marker, with UK Biobank validation across 2,495 individuals and 12-year follow-up providing the kind of evidence that actually moves clinical practice. The three-protein panel (plexin B2, GDF15, renin) from the larger 10,873-person study adds mechanistic depth and the renin causality finding is a real contribution.

Where I'm less convinced: the discovery cohort for the Diabetologia study was 32 people. Thirty-two. The UK Biobank validation rescues this somewhat, but the initial signal came from a very small sample. The vitreous study had 14 participants total. These are early-stage findings dressed up in impressive multi-omics methodology. The data is promising, not settled.

For the biohacker or quantified-self crowd: NFL is worth adding to your panel now, especially if you have metabolic risk factors. For the clinical community: we're still a few validation steps away from changing screening guidelines. That's not a criticism — that's how science is supposed to work.