Short-term retinal microvascular and structural changes after epi-on and epi-off corneal cross-linking assessed by SD-OCTA.

PURPOSE: To evaluate and compare short-term retinal microvascular and structural changes following ultraviolet-A (UVA) exposure in corneal cross-linking (CXL) procedures with and without epithelial removal using spectral-domain OCT angiography (SD-OCTA). METHODS: This single-center, retrospective, non-randomized, contralateral-eye comparative study included 62 eyes of 31 patients. Group 1 received accelerated epi-off CXL in eyes with advanced keratoconus; Group 2 received epi-on CXL in eyes with good visual acuity. Because allocation was based on clinical severity rather than randomization, between-procedure comparisons were considered exploratory. Pre- and postoperative (1st week, 1st and 3rd month) SD-OCTA and Sirius imaging were used to assess BCVA, central corneal thickness (CCT), foveal and parafoveal retinal thicknesses, ganglion cell complex (GCC), superficial and deep capillary plexus vessel densities, foveal avascular zone (FAZ), outer retinal flow area (OR-FA), and choriocapillaris flow area (CC-FA). RESULTS: At baseline, Group 1 had worse BCVA and thinner CCT than Group 2, reflecting clinically determined allocation according to disease severity. Group 1 showed a significant reduction in BCVA at 1 week (p = 0.005) and significant CCT thinning postoperatively (p < 0.05). Foveal and parafoveal thicknesses and ganglion cell complex decreased significantly in Group 1. SCP and DCP vessel densities decreased in Group 1, with minimal change in Group 2. FAZ and CC-FA values were more affected in the epi-off group. Statistically greater changes were observed over time in OR-FA and CC-FA in Group 1 (p = 0.010, p = 0.023). These statistically significant changes were generally small in absolute magnitude and should be interpreted cautiously. CONCLUSIONS: Selected short-term structural and OCTA-derived microvascular parameters changed after CXL, more prominently after epi-off treatment. However, because treatment allocation was non-random and based on clinical severity, these findings should be interpreted as exploratory short-term associations rather than proof of UVA-related retinal phototoxicity or causal retinal damage.