Predicting the Effects of Customized Corneal Cross-Linking on Corneal Geometry

To validate an existing finite element model (FEM) for predicting the flattening effect of corneal cross-linking (CXL) in a clinical scenario and to use this model to investigate the parameters that most influence CXL-induced flattening effects.

Retrospective data were collected from two clinical cohorts, each with 20 patients receiving either standard or customized CXL. Data were collected before surgery and at the six-month follow-up. Both CXL treatments were simulated with a FEM calibrated on experimental data. Standard anterior corneal geometry indexes (e.g., sphere, cylinder), as well as the curvature changes observed at follow-up were compared to those predicted by FEM simulations.

At follow-up, patients who underwent customized CXL exhibited more corneal flattening compared to those who received standard CXL (Kmax-t: -2.28 ± 1.4 D vs. -0.81 ± 1.5 D; P < 0.001). The FEM-predicted curvature reduction in the central CXL regions showed a significant correlation with the follow-up data for both standard (R2 = 0.48, P < 0.01) and customized CXL (R2 = 0.59, P < 0.01). Compared to follow-up data, standard CXL model showed concordance correlation coefficients > 0.9 for nine corneal geometry parameters and customized CXL model for three. Sensitivity analysis demonstrated that a 3 mm Hg increase in intraocular pressure (IOP) combined with a 10% weaker keratoconus region alters flattening outcomes by up to 20%.

Customized CXL induces a flattening of about 2 diopters in the cone region six months after surgery. The model adequately captured the curvature corrections induced by the treatment in the keratoconus cone region, but showed reduced accuracy in predicting global corneal metrics, particularly for customized CXL. The induced flattening effects depend on the IOP, keratoconus-induced biomechanical weakening, and the fluence delivered to the cone.