Corneal Confocal Microscopy is Superior to Skin Biopsy as a Surrogate Marker of Human Diabetic Neuropathy

Abstract

Aim and Method: There is an urgent need to develop surrogate markers to aid in the diagnosis and assessment of treatment efficacy in human diabetic neuropathy. 22 diabetic patients underwent detailed assessment of Neuropathy Impairment Score in the lower limbs (NIS-LL), electrophysiology (EMG), quantitative sensory testing (QST), corneal confocal microscopy (CCFM) and skin biopsy. On the basis of NIS-LL, EMG and QST assessment 11 patients were classed to have no neuropathy (age= 5010 years, NIS-LL= 4+/-4, Peroneal Nerve Conduction Velocity (PNCV) 50+/- 10 m/s, Sural Nerve Amplitude Potential (SNAP) 11 +/- 7 mA, Tibial Nerve Onset Latency (TNOL) 15+/-2 ms, Vibration Perception Threshold (VPT) 7+/- 4 V, Cooling Detection Threshold (CDT) 57 +/- 28 pc, Deep Breathing Heart Rate Variability (DBHRV) 67+/-28 pc, and 11 had neuropathy (age 5311 years, NIS-LL=17+/-12, PNCV 39+/- 8 m/s, SNAP 5+/- 3 mA, TNOL 18+/-2 ms, VPT 30+/- 13 Volts, CDT 97+/- 2 pc, DBHRV 13+/-29 pc. CCFM was performed to quantify corneal nerve morphology: nerve fibre density (NFD), tortuosity (NFT), and branch density (NFBD) were compared with normative values from healthy volunteers (n=23, age 5215). 3 mm punch skin biopsies were performed from the dorsum of the foot and immunohistochemical staining was performed with PGP 9.5 to quantify dermal nerve fibre density (DNFD) and compared with normative values from 22 volunteers (age and sex matched). Results: 1) CCFM demonstrated differences between non neuropathic diabetic, neuropathic diabetic and control groups. Respective results are given as mean +/- SD: NFD 29+/-10, 35+/-13, and 46+/-16 nerves/mm2 (ANOVA P= 0.02), NFT 8+/-8, 19+/-14, 24+/-10 (tortuosity index, ANOVA P=0.002), NFBD 6+/-2, 6+/-3, 33+/-26 branches/mm2 (ANOVA P< 0.001). Post-hoc analysis showed a significant difference between both diabetic groups and controls for NFBD (P< 0.01), but only for non neuropathic diabetic groups compared to controls for NFD (P< 0.05), NFT (P< 0.01). 2) DNFD was reduced in all diabetic patients, non neuropathic and neuropathic, compared to controls (respectively 203+/-114, 190+/-58, and 414+/-196 fibres/mm2: mean + SD, ANOVA P< 0.001). No differences between the two patient groups were showed by post-hoc analysis. 3) Positive correlations were found between NFBD and SNAP (Pearson r= 0.55, P= 0.01), NFT and SNAP (r= 0.55, P= 0.004), and DNFD and SNAP (r=0.53, P= 0.01). NFD and NFBD inversely correlated with Peroneal F waves Latencies at (r=-0.63, P= 0.002) and r=-0.64, P= 0.001, respectively. Conclusions: Whilst both CCFM and DNFD were more reduced in diabetic patients with compared to those without neuropathy this did not reach significance. Corneal nerve morphology and skin DNFD correlated with electrophysiological parameters. However, they did not correlate to each other. This preliminary study demonstrates significant associations between corneal nerve morphology, ENFD and conventional measures of neuropathic severity suggesting that they may be reliable surrogate measures of human diabetic neuropathy. The correlation between some of the CCFM parameters and F waves suggests that CCFM may be particularly important in the very early stages of diabetic neuropathy.