Predicting the Efficacy of Orthokeratology for Myopia Control: Effects of Corneal Zones and Topographic Powers

Abstract

Purpose : To assess how essential biometric parameters, in terms of the Relative Corneal Refractive Power Shift (RCPRS) calculated from corneal topographic maps, can predict the treatment efficacy of orthokeratology (OK).

Methods : Clinical records from 2012 to 2018 were retrieved from the university optometry clinic. Topographic maps of right eyes from 204 Chinese children (6 to 11 years) who started OK treatment with at least 1 year of follow-up were analyzed. The treatment efficacy of OK was determined by the rate of axial length elongation (ALE) in the first year of treatment. The summations of tangential, axial, and refractive RCRPS within various corneal radii were analyzed up to 4 mm in a 0.1 mm step. The robustness of these biometric parameters in predicting the ALE was assessed using a 3-fold randomized cross-validation with 50 repetitions. In each iteration, the training set was fitted by a non-positive linear model to predict the ALE in the validation set. The validation R²was averaged across the 150 iterations. After identifying the most robust RCRPS parameter based on the validation R², a multilinear regression was applied to predict the ALE based on this RCRPS parameter and participants’ initial age and axial length (AL).

Results : The summation of tangential RCRPS within the 2.2-mm corneal radius best predicted children’s ALE after wearing OK lenses (R² = 0.022). Note that regardless of the analyzing radius, neither axial nor refractive RCRPS could benefit the prediction of ALE (all R² < 0). The multilinear regression indicated that the 2.2-mm-summed tangential RCRPS (β* = –0.149, p < 0.001), initial age (β* = –0.368, p < 0.01), and initial AL (β* = –0.180, p = 0.02) were significantly and negatively correlated with ALE (R² = 0.199).

Conclusions : RCRPS derived from tangential power within the 2.2 mm corneal radius most robustly predicted the rate of axial elongation in the first year of OK treatment in 6- to 11-year-old children. Children with more positive RCRPS, older initial age, and longer initial AL had slower axial elongation.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

 

Fig.1A) The mean RCRPS profiles calculated from different topographic powers, the summation of RCRPS within various analyzing zones (shaped areas in the example) were then analyzed. B) The mean validation R²for predicting the 1-year ALE from different combinations of topographic powers and analyzing zone.