Predictive Factors For The Progression Of The Fellow Eye In Unilateral Keratoconus: A Cohort Study

To investigate the independent predictive value of corneal tomography parameters and Corvis ST biomechanical parameters for progression in the fellow eye of patients with unilateral keratoconus（KC）, and to develop a nomogram prediction model based on these intuitive parameters.

Single-center retrospective cohort study

This study enrolled 66 fellow eyes from patients with strictly unilateral keratoconus. The mean follow-up time was 34.24 ± 22.59 months. Based on predefined tomographic progression criteria (a decrease in thinnest corneal thickness of ≥20 µm, an increase in maximum anterior corneal curvature of ≥1.5 D, or an increase in BAD-D value of ≥0.42), the eyes were categorized into a progression group (n=14) and a non-progression group (n=52). Baseline parameters from corneal topography and raw Corvis ST measurements were compared between the two groups. Independent risk factors were screened using univariate and multivariate Logistic regression analyses, and a nomogram prediction model was constructed based on these factors. 

The progression and non-progression groups showed no significant differences in baseline demographic or corneal tomographic parameters but did differ significantly in several raw Corvis ST biomechanical parameters. Multivariable regression identified corneal astigmatism, first applanation time, and second applanation length as independent predictors of progression. The nomogram based on these three variables demonstrated excellent performance, with a training area under the curve (AUC) of 0.876, a mean cross-validated AUC of 0.841, good calibration, and clinical net benefit across a wide threshold probability range.

This study is the first to develop a predictive model based on corneal biomechanical and morphological parameters for assessing progression risk in the fellow eye of unilateral keratoconus. Our findings demonstrate that corneal astigmatism, first applanation time (A1 time), and second applanation length (A2 length) hold independent predictive value for disease progression. The established nomogram model provides a practical tool for individualized risk quantification in clinical settings.