Reproducibility and Reliability of Subbasal Corneal Nerve Parameters of the Inferior Whorl in the Neurotoxic and Healthy Cornea

Purpose: The aim of this study was to investigate the reliability of subbasal corneal nerve plexus parameters of the inferior whorl compared with the central cornea with in vivo corneal confocal microscopy and to investigate the impact of inferior whorl pattern complexity on reproducibility. Methods: Subbasal corneal nerves of healthy controls (n = 10) and patients with chemotherapy-induced peripheral neuropathy (n = 10) were imaged with a laser scanning confocal microscope. Two masked, experienced observers and the original image taker were tasked with selecting representative images of the central cornea and inferior whorl for each participant. This was conducted on 2 occasions 1 week apart. Corneal nerve fiber length (CNFL) and fractal dimension (CNFrD) [central cornea: CNFL and CNFrD; inferior whorl region: inferior whorl length (IWL) and inferior whorl fractal dimension (IWFrD)] were analyzed. Intraclass correlation coefficient (ICC) was analyzed for interobserver and intraobserver reliability. Inferior whorl complexity was classified according to the ease of identification of the center point of convergence. Results: Interobserver ICC was 0.992 for CNFL, 0.994 for CNFrD, 0.980 for IWL, and 0.954 for IWFrD. When analyzed by inferior whorl complexity, the interobserver reliability was similar for simple (0.987 for IWL; 0.960 for IWFrD) and complex patterns (0.967 for IWL; 0.949 for IWFrD). However, intraobserver ICC were reduced for complex (IWL 0.841-0.970; IWFrD 0.830-0.955) compared with simple patterns (IWL 0.931-0.970; IWFrD 0.921-0.969). Conclusions: Although the overall interobserver reliability was excellent for the central corneal and inferior whorl parameters, there was lower intraobserver reliability for the inferior whorl parameters for complex morphological patterns. To improve reliability, more sophisticated wide-field imaging of the inferior whorl may be needed.