Aims/hypothesis: The individual risk of progression of diabetic peripheral neuropathy is difficult to predict for each individual. Mutations in proteins that are responsible for the process of myelination are known to cause neurodegeneration and display alteration in experimental models of diabetic neuropathy. In a prospective observational human pilot study, we investigated myelin-specific circulating mRNA targets, which have been identified in vitro, for their capacity in the diagnosis and prediction of diabetic neuropathy. The most promising candidate was tested against the recently established biomarker of neural damage, neurofilament light chain protein.

Methods: Schwann cells were cultured under high-glucose conditions and mRNAs of various myelin-specific genes were screened intra- and extracellularly. Ninety-two participants with type 2 diabetes and 30 control participants were enrolled and evaluated for peripheral neuropathy using neuropathy deficit scores, neuropathy symptom scores and nerve conduction studies as well as quantitative sensory testing at baseline and after 12/24 months of a follow-up period. Magnetic resonance neurography of the sciatic nerve was performed in 37 individuals. Neurofilament light chain protein and four myelin-specific mRNA transcripts derived from in vitro screenings were measured in the serum of all participants. The results were tested for associations with specific neuropathic deficits, fractional anisotropy and the progression of neuropathic deficits at baseline and after 12 and 24 months.

Results: In neuronal Schwann cells and human nerve sections, myelin protein zero was identified as the strongest candidate for a biomarker study. Circulating mRNA of myelin protein zero was decreased significantly in participants with diabetic neuropathy (p < 0.001), whereas neurofilament light chain protein showed increased levels in participants with diabetic neuropathy (p < 0.05). Both variables were linked to altered electrophysiology, fractional anisotropy and quantitative sensory testing. In a receiver-operating characteristic curve analysis myelin protein zero improved the diagnostic performance significantly in combination with a standard model (diabetes duration, age, BMI, HbA_1c ) from an AUC of 0.681 to 0.836 for the detection of diabetic peripheral neuropathy. A follow-up study revealed that increased neurofilament light chain was associated with the development of a hyperalgesic phenotype (p < 0.05), whereas decreased myelin protein zero predicted hypoalgesia (p < 0.001) and progressive loss of nerve function 24 months in advance (HR of 6.519).

Conclusions/interpretation: This study introduces a dynamic and non-invasive assessment strategy for the underlying pathogenesis of diabetic peripheral neuropathy. The diagnosis of axonal degeneration, associated with hyperalgesia, and demyelination, linked to hypoalgesia, could benefit from the usage of neurofilament light chain protein and circulating mRNA of myelin protein zero as potential biomarkers.