Development and validation of the QUT corneal nerve grading scale

PURPOSE: The aim of this study was to develop and validate a grading scale to facilitate the estimation of the amount of nerve tissue in images of the corneal subbasal nerve plexus captured using in vivo laser scanning corneal confocal microscopy (LSCCM).

METHODS: Images of the corneal subbasal nerve plexus obtained using a Heidelberg LSCCM were sourced from a large image bank at the Queensland University of Technology (QUT). These images were used to construct a grading scale for depicting the amount of nerve tissue, ranging from 0 (sparse) to 4 (extensive). Twenty-five observers graded 20 images of a known corneal nerve fiber length (defined as the total length of nerves per unit area) on 2 occasions, at least 2 weeks apart. An equivalent calculated grade was determined for each test image from known values of corneal nerve fiber length.

RESULTS: The intraclass correlation coefficient for repeat gradings was 0.88 (P < 0.001). Intraobserver and interobserver repeatabilities were unrelated to the calculated grade (P = 0.467 and P = 0.530, respectively). Grading can be performed with average 95% confidence limits of ±1.2 grading units. Overall grading estimates did not differ between observers (P = 0.998). There was a strong agreement between the estimated and calculated grades (intraclass correlation coefficient = 0.92, P < 0.001).

CONCLUSIONS: We have developed the QUT Corneal Nerve Grading Scale, which is demonstrated to be repeatable, reliable, precise, and accurate. This tool provides clinicians and researchers with a simple and convenient pictorial reference for assessing, comparing, and monitoring the corneal subbasal nerve plexus with reference to LSCCM images.