ABSTRACT
BACKGROUND AND AIMS: The diagnosis of small fiber neuropathy (SFN) is supported by reduced intraepidermal nerve fiber density (IENFD). The noninvasive method corneal confocal microscopy (CCM) has the potential to be a practical alternative. We aimed to estimate the diagnostic accuracy of CCM compared with IENFD and cold detection thresholds (CDT) in SFN and mixed fiber neuropathy (MFN). METHODS: CCM was performed in an unselected prospective cohort of patients with a clinical suspicion of polyneuropathy. Predefined criteria were used to classify SFN and MFN. Neuropathy scores, including the Utah early neuropathy scale (UENS), were used to describe severity. Patients with established other diagnoses were used for diagnostic specificity calculations. RESULTS: Data were taken from 680 patients, of which 244 had SFN or MFN. There was no significant difference in sensitivities [95%CI] of CCM (0.44 [0.38-0.51]), IEFND (0.43 [0.36-0.49]), and CDT (0.34 [0.29-0.41]). CCM specificity (0.75 [0.69-0.81]) was lower (p = .044) than for IENFD (0.99 [0.96-1.00]) but not than for CDT (0.81 [0.75-0.86]). The AUCs of the ROC curves of 0.63, 0.63 and 0.74 respectively, was lower for corneal nerve fiber density (p = .0012) and corneal nerve fiber length (p = .0015) compared with IENFD. While UENS correlated significantly with IENFD (p = .0016; R2 = .041) and CDT (p = .0002; R2 = .056), it did not correlate with CCM measures. INTERPRETATION: The diagnostic utility of CCM in SNF and MFN is limited by the low specificity compared with skin biopsy. Further, CCM is less suitable than skin biopsy and CDT as a marker for neuropathy severity.
