New model to better diagnose dry eye disease integrating OCT corneal epithelial mapping.

To optimise the objective diagnosis of dry eye disease (DED), the capabilities of wide corneal epithelial mapping using optical coherence tomography (OCT) were studied and subsequently integrated into a new scoring method. Fifty-nine patients (118 eyes) with DED and 55 control subjects (110 eyes) were included. All patients underwent a complete ocular surface evaluation. Corneal epithelial thickness was collected using OCT for seven zones. DED and the control group were compared using a t-test, and univariate receiver operating characteristic (ROC) curves were calculated to define the diagnostic ability of OCT epithelial mapping. Multivariate analyses were performed using artificial intelligence (random forest) and logistic regression approaches to define the best way to integrate OCT mapping in the diagnosis of DED. Then, a final multivariable model for diagnosing DED was validated through a bootstrapping method. The DED group had significant epithelial thinning compared with the controls, regardless of location. Superior intermediate epithelial thickness was the best marker for diagnosing DED using OCT (binormal estimated area under ROC: 0.87; best cut-off value: 50 µm thickness). The difference between the inferior and superior peripheral zones was the best marker for grading the severity of DED (analysis of variance, p=0.009). A multivariate approach identified other significant covariables which were integrated into a multivariate model to improve the sensitivity (86.4%) and specificity (91.7%) of this innovative diagnostic method. Including OCT corneal epithelial mapping in a new diagnostic tool for DED could allow optimisation of the screening and staging of the disease in current practice as well as for clinical research purposes.

Informations de copyright

© Author(s) (or their employer(s)) 2022. No commercial re-use. See rights and permissions. Published by BMJ.

Déclaration de conflit d'intérêts

Competing interests: None declared.