Posters

Classification of keratoconus: suggestions for optimizing reference ranges for corneal thickness

Poster Details

First Author: M.Habeeb THE NETHERLANDS

Co Author(s):    A. Geerards   M. Meester-Smoor   C. Klaver   W. Ramdas   B. van Dooren        

Abstract Details

Purpose:

To assess the use of (tomo-/topo- graphic) parameters from both Amsler-Krumeich (AK) and Belin ABCD (ABCD) keratoconus classification systems and to investigate whether these parameters could be optimized.

Setting:

Patients diagnosed with keratoconus from the Erasmus MC Rotterdam and the Rotterdam Eye Hospital who underwent Pentacam imaging were included.

Methods:

Pentacam scans from patients with concurrent ocular disease(s) or previous refractive, corneal, or ocular surgeries were excluded. The severity of keratoconus, assessed by AK and ABCD, was compared. Cohen's kappa was used to assess the proportion of agreement between different classification parameters.

Results:

A total of 1444 Pentacam scans (707 left eyes) of 815 patients (age [28.72±10.89] years; 66% males) were included in this analysis. According to the AK classification (1-4), 70% of the scans were stage 1, 25% stage 2, and less than 5% stage 3 or 4. Using the ABCD classification, the frequency of the different stages (0-4) were: 1) parameter "A" (anterior radius of curvature in 3.0 mm), 30%, 40% and 15% of the scans were stage 0, 2, 4 respectively; 2) parameter "B" (posterior radius of curvature in 3.0 mm), 15%, 28% and 36% of the scans were stage 0, 2 and 4 3) parameter "C" (thinnest pachymetry), 31%, 35%, 26%, 6% of the scans were stage 0, 1, 2, 3 respectively while 0% were stage 4. Agreement between the ABCD parameters was moderate between "A" and "B" (cohen's k=0.57), fair between "A" and "C" (cohen's k=0.32) and low between "B" and "C" (cohen's k=0.18).

Conclusions:

The ABCD, especially the "B" parameter, reflected early corneal changes in keratoconus more than the AK. However, the ABCD "C" parameter, did not reflect similar disease stages as the concomitant “A” and “B” parameters. We suggest to adjust the thinnest pachymetry reference values in both the AK and ABCD classifications. This would allow changes in the corneal thickness to be simultaneously reflected with changes in the "A" and "B" parameters. Also, current pachymetry safety margins (~400um) for treatment decisions i.e. collagen cross-linking, maybe more relevantly reflected. These suggested alterations need to be reproduced and validated or adjusted, using other datasets.

Financial Disclosure:

None