Prediction Of Posterior Corneal Surface Parameters From Epithelial Parameters In Keratoconus

To analyze the association between epithelial parameters and posterior corneal surface parameters in keratoconus and to develop predictive models for posterior corneal elevation and posterior Kmax based exclusively on epithelial measurements.

Department of Ophthalmology, Faculty of Medicine and University Hospital Carl Gustav Carus, TU Dresden, Dresden, Germany.

In this monocentric cross-sectional study, 60 eyes of 60 patients with keratoconus were included, with one randomized eye per patient. Anterior segment OCT epithelial mapping was performed using the Anterion (Heidelberg Engineering). Epithelial parameters included zonal thickness, inter-zonal thickness differences, thinnest and thickest points, and the standard deviation of mean epithelial thickness both globally and by zone. Correlations with posterior tomographic parameters were assessed using Pearson correlation. Predictive models for maximum posterior elevation and posterior Kmax were developed in 42 eyes using LASSO-based predictor selection and ordinary least squares regression, and validated in 18 eyes.

Standard deviation of mean epithelial thickness correlated with posterior Kmean (r=-0.54), posterior astigmatism (r=-0.49), posterior Kmax (r=-0.74), and maximum posterior elevation (r=0.70) (all P<0.01). Superior-temporal interzonal epithelial thickness differences showed the strongest correlation with posterior Kmax (r=-0.79, P<0.01). The maximum posterior elevation model, based on four epithelial parameters, achieved R=0.88 and mean absolute error (MAE) 10.7µm in development dataset and R=0.69 and MAE 20.9µm in validation dataset. The posterior Kmax model, also based on four epithelial parameters, achieved R=0.90 and MAE 0.4D in development dataset and R=0.91 and MAE 0.5D in validation dataset, explaining 81-83 % of variance.

Epithelial remodeling is strongly associated with posterior corneal surface changes in keratoconus. Epithelial parameters enabled prediction of maximum posterior elevation and posterior Kmax, with the best performance observed for posterior Kmax. These parameters may complement keratoconus staging and support the evaluation of disease progression.