One-Year Outcomes After Conventional vs Accelerated Epi-Off Corneal Cross-Linking in Progressive Keratoconus (Observational Cohort)

This prospective, single-center observational cohort extends the follow-up of a previously reported 6-month series from the same center to a 12-month horizon. The study population comprises consecutive patients with progressive keratoconus managed with standard-of-care, epi-off corneal collagen cross-linking (CXL) at the University Hospital Centre Zagreb. Patients were treated according to routine clinical decision-making by the attending surgeon; no prospective allocation or randomization to a protocol was performed. In routine care, some eyes underwent conventional CXL30 (Dresden-type; UVA 370 nm, 3 mW/cm² for 30 min) and others underwent CXL10 (accelerated; UVA 370 nm, 9 mW/cm² for 10 min). Riboflavin 0.1% solution was used per institutional practice. All procedures were epi-off.

Rationale. While both conventional and accelerated CXL are widely used in daily practice, comparative one-year data that integrate tomographic metrics from the ABCD grading system with traditional surrogates (e.g., Kmax) help clarify the balance between anterior corneal flattening, posterior remodeling, and pachymetric behavior under real-world conditions. This observational design captures effectiveness and safety without altering clinical decision pathways.

Design and cohorts. Consecutive eligible eyes were enrolled at the time of surgery and followed prospectively at baseline (day of surgery) and 1, 3, 6, and 12 months. Two observational cohorts are defined post hoc by the protocol actually received in standard care:

CXL30 (Conventional): continuous UVA 3 mW/cm² × 30 min (Dresden-type).

CXL10 (Accelerated): continuous UVA 9 mW/cm² × 10 min.

Assignment to CXL30 vs CXL10 was at the surgeon's discretion based on routine clinical considerations; the study did not prospectively assign interventions.

Eligibility. Inclusion required progressive keratoconus documented on serial Scheimpflug tomography within the prior 12 months; a clear corneal axis (no visually significant scarring); and corneal thickness 370 µm or more per institutional safety protocol. Exclusions comprised no central corneal scars, prior chemical burns, severe corneal infection, or ocular surface disease. Pregnancy and lactation during treatment were also contraindications.

Examinations and outcomes. Outcomes were assessed using Scheimpflug tomography (e.g., Pentacam or equivalent) and standard visual testing. The primary outcome was change in ABCD-A (anterior corneal curvature metric) from baseline to 12 months. Secondary outcomes included Kmax, anterior and posterior radii of curvature (ARC/PRC), thinnest pachymetry, and UDVA/BCVA at 12 months. Safety endpoints encompassed delayed epithelial healing (>7 days), infectious keratitis, clinically significant corneal haze, endothelial decompensation, and need for re-treatment.

Follow-up timeline. Enrollment/treatment occurred May 2021-December 2021; 12-month follow-up was completed by January 2023. Of the 28 eyes in the previously reported 6-month cohort, 22 completed the 12-month visit and were included in this analysis.

Data handling and oversight. Data were collected prospectively in a standardized fashion and analyzed as observational (no masking, no allocation). The protocol was reviewed and approved by the institutional Ethics Committee; all analyses used de-identified data. No data monitoring committee was required given the non-interventional, standard-of-care nature of the study.

Statistical approach (summary). Effectiveness outcomes are summarized descriptively; within-cohort change from baseline to 12 months is evaluated with paired tests on continuous variables as appropriate to distributional assumptions. Between-cohort comparisons (CXL30 vs CXL10) at 12 months and for change scores are explored with two-sample tests or generalized linear models with adjustment for prespecified covariates (e.g., baseline severity indices) where applicable. Statistical significance is assessed at a two-sided α = 0.05, with corresponding confidence intervals to aid interpretation.