Prospective Objective Analysis of Corneal Haze Following Customized Transepithelial PRK Without Mitomycin C Combined With Accelerated Corneal Cross-Linking Versus Corneal Cross-Linking Alone

Abstract

PURPOSE: To compare haze and refractive outcomes in patients undergoing combined accelerated corneal cross-linking (A-CXL) and selective wavefront-guided transepithelial photorefractive keratectomy (WG-transPRK) without mitomycin C (MMC) versus those undergoing A-CXL. METHODS: This prospective study analyzed 95 eyes (86 patients) with progressive keratoconus from October 2018 to October 2022. The first group underwent CXL combined with corneal or ocular WG-transPRK (CXL+PRK, n = 52), targeting higher order aberrations (HOAs). The second underwent CXL only (n = 43), both following the same accelerated CXL protocol without MMC on the SCHWIND Amaris laser platform (SCHWIND eye-tech-solutions). Baseline and postoperative evaluations (1, 3, 6, and 12 months) included uncorrected (UDVA) and corrected (CDVA) distance visual acuity, manifest refraction, tomography, corneal HOAs, and optical coherence tomography (OCT) scans. A patented machine learning algorithm objectively detected and quantified stromal haze on OCT scans in grayscale units. RESULTS: In both groups, anterior corneal haze reflectivity and subepithelial haze peaked at 3 months postoperatively, then progressively decreased at 6 and 12 months. Haze did not differ between groups at any time point. By 12 months, CDVA increased by 2.5 lines in the CXL+PRK group (P < .001) and by 0.7 lines in the CXL group (P = .10), and maximum keratometry decreased from 51.70 ± 5.10 to 47.90 ± 7.90 diopters (D) (CXL+PRK group) (P < .001) and from 51.20 ± 5.10 to 50.30 ± 4.60 D (CXL group) (P = .004). Corneal HOAs decreased in both groups but more in the CXL+PRK group. CONCLUSIONS: Combining CXL with WG-transPRK without MMC does not result in increased haze when compared to A-CXL alone. This combined approach achieves greater improvements in visual, topographic, and aberrometric parameters.