The role of corneal epithelial thickness map in detecting early keratoconus.

Abstract

Purpose: To detect subtle changes in early keratoconus by evaluating corneal epithelial thickness differences among patients with binocular very asymmetric ectasia (VAE) and normal subjects.

Methods: Corneal epithelial thickness was measured using the Fourier-domain AS-OCT system RTVue^® 100 (Optovue, Fremont, CA, USA). 152 eyes from 76 patients were divided into three groups: Very asymmetry ectasia-ectasia (VAE-E, n = 38), Very asymmetry ectasia-normal topography (VAE-NT, n = 38), and Normal control (NC, n = 76). Discrimination capacity was assessed using areas under the curve (AUC) of receiver operator characteristic (ROC) curves.

Results: In the keratoconus group, the epithelial Min (minimum), central, midperipheral I (inferior), midperipheral IT (inferior-temporal), peripheral IT, and midperipheral T (temporal) were thinner (all P < 0.05). The topography normal group had thinner midperipheral IN (inferior-nasal), peripheral IN, midperipheral T, and peripheral T, and larger Max-Min and Std. Dev (P < 0.05). For diagnosing typical keratoconus, Std. Dev (AUC = 0.982, sensitivity 97.4%, specificity 92.1%) had the highest diagnostic efficiency. Combining four variables (Minimum, Max-Min, Midperipheral IT, and Midperipheral I) performed well in distinguishing topography normal eyes (AUC = 0.896, sensitivity 76.3%, specificity 89.5%). Multivariable analysis using epithelial parameters combined with Pentacam random forest index (PRFI) yielded the best results (AUC = 0.951, sensitivity 90.6%, specificity 89.5%).

Conclusions: The corneal epithelial parameters play an important auxiliary role in the diagnosis of keratoconus and the screening of subclinical keratoconus. Combination of epithelial parameters and tomographic parameters can improve the sensitivity of early stage keratoconus detection.