Purpose
Although keratoconus (KC) is regarded as a bilateral disease, in some instances, patients initially present unilaterally, with the fellow eye appearing normal clinically and topographically. Such eyes are considered to represent the gold standard for evaluation of methods for early KC detection. In this study we assessed the effectiveness of a KC detection algorithm derived from Artemis very high frequency ultrasound epithelial thickness maps in 12 unilateral KC subjects.
Methods
Subjects were recruited from a population of candidates for corneal refractive surgery presenting at the London Vision Clinic. The study included clinically and topographically normal fellow eyes of 12 patients with unilateral moderate to advanced KC. We acquired Artemis scan data and applied a previously developed linear discriminant analysis multivariate model for separation of normal and advanced KC based on Artemis epithelial thickness maps to the asymptomatic fellow eyes. Pentacam Belin-Ambrosio Display (BAD) data were available for 7 of the 12 eyes.
Results
Four of the 12 fellow eyes were classified as KC by the Artemis model. Three cases classified as normal by the algorithm, however, exhibited an epithelial pattern suggestive of KC to an expert observer. Of the 7 fellow eyes with both Pentacam and Artemis data, only one was classified as KC by the Artemis algorithm, a different eye by Pentacam (BAD-D >1.5) and yet another by a combined Artemis plus Pentacam model. Three of the 7 fellow eyes with Pentacam data were classified as normal by all three methods (BAD-D values of -0.39, +0.07, +0.5).
Conclusions
We know that in a significant proportion of cases, epithelial remodeling manifests prior to changes in topography and masks early KC topographically. Assuming that KC is a bilateral disease and that its expression may be asymmetric, we have demonstrated that one third of ‘normal’ fellow eyes could be found to have KC by using epithelial thickness maps. The combination of topographic BAD criteria with epithelial maps did not appear to perform better. Possibly this population included patients who were truly monocularly keratoconic. Alternatively, it may be that the phenotypically unexpressed form of KC might actually be undetectable by anatomical analysis: this would explain the rare cases of post-LASIK ectasia “without a cause”. Perhaps in vivo localized biomechanical evaluation may be required to detect such cases.