Predictive Factors for Disease Progression Following Corneal Crosslinking in Keratoconus: a Novel Approach Using...

BACKGROUND: To identify predictive factors for disease progression following corneal crosslinking (CXL) in keratoconus patients, focusing on tomographic, pachymetric, and patient-specific characteristics.

PATIENTS AND METHODS: This retrospective study analyzed 147 eyes from 123 keratoconus patients who underwent conventional epithelium-off CXL (3 mW/cm² for 30 min) between 2011 and 2021, with a minimum follow-up of 12 months. Baseline parameters included preoperative maximum keratometry (pre-Kmax), dynamic preoperative annualised Kmax progression (pre-ΔKmax), corneal thickness at the thinnest point (pre-CTT), age, sex, and atopic dermatitis. A Linear Mixed Model (LMM) identified predictors of postoperative annualised ΔKmax progression (diopters/year), while a Cox regression assessed risk factors for a single postoperative Kmax increase of > 1 diopter (D), and both approaches accounted for inter-eye correlation.

RESULTS: In the LMM, atopic dermatitis showed a trend toward a higher rate of postoperative Kmax progression (β = 0.00379, p = 0.093), corresponding to an estimated increase of + 1.38 D per year. In contrast, a higher rate of preoperative progression (pre-ΔKmax; β = - 0.0684, p = 0.074) and higher baseline Kmax (pre-Kmax; β = - 0.000173, p = 0.070) demonstrated trend-level associations with reduced annual postoperative progression, reflecting greater corneal flattening over time. However, in Cox regression, higher pre-Kmax demonstrated a trend-level association with an increased risk of a single postoperative progression event (HR = 1.05; 95% CI: 0.99 - 1.10; p = 0.095). Age, sex, and pre-CTT showed no significant associations in either model.

CONCLUSION: Atopic dermatitis was identified as a potential risk factor for keratoconus progression following CXL, and showed a trend-level association with a higher annual Kmax increase. Paradoxically, while higher pre-Kmax increased the risk of isolated progression events, it was simultaneously associated with a greater long-term flattening trend. Additionally, a rapid rate of preoperative progression was associated with a greater stabilising effect, suggesting that these corneas may be particularly responsive to treatment. These findings underscore the importance of integrating event-based and trend-based analytical approaches when interpreting post-CXL outcomes and their associated predictive factors.