Design: Pooled analysis of four prospective cohorts, Cox models meta-analysed with false discovery rate correction
Population: 3,263 individuals with and without type 2 diabetes at profiling; 674 incident neuropathy and 639 incident nephropathy cases
Primary outcome: Incident diabetic nephropathy and neuropathy after diabetes diagnosis
Effect: Nephropathy: IgG-GP3 hazard ratio 1.13 (95% CI 1.04-1.24), IgG-GP4 1.15 (1.05-1.26); protective peaks IgG-GP14 0.86 (0.78-0.94), IgG-GP18 0.85 (0.77-0.94). Neuropathy associations non-significant after correction

Antibody molecules carry sugar chains that shift with inflammation, and the pattern can be read from stored blood. Four prospective cohorts — EPIC-Potsdam, DiaGene, GenoDiabMar and Hoorn DCS — were pooled to ask whether those patterns precede microvascular complications in type 2 diabetes. Among 3,263 people profiled, 674 developed neuropathy and 639 nephropathy after their diabetes diagnosis.

The glycan peaks that indicate missing terminal sugars went with higher kidney risk (hazard ratios 1.13 and 1.15 per peak), and the galactosylated and sialylated peaks went with lower risk (0.86 and 0.85), all surviving correction for multiple testing. The neuropathy associations did not survive that correction, and the authors say plainly that the neuropathy evidence remains inconclusive.

This is research-stage and not a test to start ordering. Its value is directional: it supports an immune contribution to diabetic kidney disease specifically, which is the same tissue where anti-inflammatory mechanisms are already being pursued therapeutically. Nothing here outperforms albumin-to-creatinine ratio and eGFR, and nothing here should displace them.

Keep risk stratification on albuminuria and eGFR — no glycan measure is validated against them
Note the split result: the kidney signal held after correction, the nerve signal did not
Treat the effect sizes as mechanistic evidence, not as a screening threshold
Watch this class of marker where an immune mechanism in diabetic kidney disease is being argued

Why it matters

It argues that diabetic kidney disease has an immune component that diabetic neuropathy may not share.

The statistics, in plain English

Hazard ratios of 1.13 to 1.15 per glycan peak are small, and confidence intervals that sit just clear of 1.0 (1.04-1.24, 1.05-1.26) are consistent with a real but modest association. More importantly, nothing was tested for added value over the measures already in use, so a marker like this can be genuinely associated with the outcome and still change no decision. The neuropathy results becoming non-significant after correcting for multiple comparisons is the expected fate of findings that were never strong.