Aims: To establish the comparative utility of corneal confocal microscopy (CCM) and skin biopsy in differentiating painless (pDPN-) from painful diabetic peripheral neuropathy (pDPN+).
Methods: A systematic review and multiple meta-analysis was conducted following the PRISMA guidelines. This work formed part of the AGORA, an international multidisciplinary working group tasked to systematically evaluate surrogate markers of neuropathic pain including genetics, epigenetics, neuropsychological, neurophysiological and neuroimaging markers; which are being addressed in a series of systematic reviews and meta-analyses followed by a structured DELPHI consensus to update clinical guidelines. A comprehensive search strategy was implemented. Studies reporting on corneal nerve fiber density (CNFD), and/or branch density (CNBD), and/or fiber length (CNFL), and/or intra-epidermal nerve fiber density (IENFD) in pDPN- vs. pDPN+ were included. The final included studies were assessed for risk of bias and heterogeneity i2 (low: 0%–40%; moderate: 30%–60%; substantial: 50%–90%; and considerable: 75%–100%) according to the Cochrane handbook guidelines. The results are presented as [standardized mean difference (SMD), 95% confidence interval, p-value, i2).
Results: For the first part of the meta-analysis related to CCM, 7/59 studies (339 pDPN-; 344 pDPN+) were selected (according to the use of Heidelberg Retina Tomograph III with manual analysis). In pDPN+ compared to pDPN- CNFD (-0.50, -0.96 to -0.03, p=0.03, I2=88%) and CNFL (-0.32, -0.59 to -0.05, p=0.02, I2 66%) were significantly lower. For the second part related to IENFD, 10/168 studies (299 pDPN-; 365 pDPN+) were selected. In pDPN+ vs. pDPN-, IENFD (-0.31, -0.59 to -0.02, p=0.03, I2=69%) was significantly lower. Furthermore, the authors listed studies reporting on additional skin parameters, which did not fulfill the criteria for meta-analysis such as: higher GAP-43/PGP9.5 ratio (p<0.005); calcitonin gene related peptide (p<0.005), substance-P (p<0.005) and total peptidergic (p<0.005) fibers in pDPN+ vs. pDPN-. For the final part comparing CCM vs. IENFD, 3/283 studies were selected (for CCM: 87 pDPN-; 74 pDPN+; for IENFD: 75 pDPN-; 77 pDPN+ respectively). CNFD (-0,64, -0.96 to -0.32, p<0.0001), CNBD (-0.39, -0.71 to -0.08, p=0.01), CNFL (-0.63, -0.95 to -0.31, p=0.0001), and IENFD (SMD −0.46, 95% CI −0.80, −0.12, p = 0.008) were significantly lower in pDPN+ compared to pDPN-.
Conclusions: Both CCM and skin biopsy demonstrate utility in differentiating painful from painless DPN.
Comments: Neuropathic pain is defined as "pain arising as a direct consequence of a lesion or disease affecting the somatosensory system" (Treede RD et al Neurology 2008;70:1630-5). Painful neuropathy complicates 1/3 of patients with DPN; has a complex etiology involving several factors; and like painless neuropathy it lacks a gold standard test. Small fibers are sensory unmyelinated C or thinly myelinated Aδ fibers, which can be structurally examined by skin biopsy and CCM. The present systematic review with multiple meta-analysis demonstrates strong utility for both endpoints to distinguish painful from painless neuropathy. It also highlights strengths and opportunities for future research. Skin biopsy is the histological gold standard, however intraepidermal fiber loss alone cannot fully account for pain intensity. Thus, quantification of additional morphological parameters in the skin (GAP-43/PGP9.5, calcitonin gene related peptide fibers etc.) may aid to better understand pain. CCM offers a reliable non-invasive alternative for pain-related phenotyping, which may benefit from greater standardization of acquisition and analysis. The limited number of head-to-head studies comparing skin biopsy and CCM highlights the need for larger longitudinal confirmatory trials.
Ioannis N. Petropoulos
Reference. Gad HY, Devigili G, Merkies I, Gilron I, MacDonald R, Lauria G, Malik RA. Corneal Confocal Microscopy and Skin Biopsy to Differentiate Painful From Painless Diabetic Neuropathy: A Systematic Review With Multiple Meta-Analyses. Eur J Neurol. 2026 Mar;33(3):e70576. doi: 10.1111/ene.70576. PMID: 41858110; PMCID: PMC13093296.
