Histological changes in human cornea after cross‐linking with riboflavin and ultraviolet A

Editor, Cross-linking treatment of progressive keratoconus (CXL) using the photosensitizer riboflavin and ultraviolet (UV) A was introduced by Wollensak in 2003 and is becoming increasingly popular (Wollensak et al. 2003a). A long-term effect could be demonstrated recently (Wollensak & Iomdina 2008). Extensive studies have been performed in vitro and in animals to evaluate the potential cytotoxicity increasing the safety of the new treatment modality (Wollensak et al. 2003b, 2004; Wollensak 2006). However, no human microscopic specimens have been studied so far. Recently, I had the opportunity to examine histologically the eye of a patient whose cornea had been cross-linked 24 hr before enucleation. The 23-year-old man with a blind painful left eye (caused by status post-severe perforating sclera and lens injury in 1992 and subsequent unsuccessful operations for visual restoration including vitrectomy, encircling band and silicone oil) had a visual acuity of 20/20 in his healthy right eye. With the patient’s consent, a standard cross-linking treatment was performed in the left eye one day before the scheduled enucleation. As described previously, the central 8 mm of the epithelium was removed and the cornea soaked with the 0.1% riboflavin solution for 20 min. Irradiation was performed for 30 min using a UVA double diode (3 mW/cm2) at a distance of 1 cm. Antibiotic ointment was administered after the cross-linking procedure and the eye was enucleated 24 hr later. The retrieved eye was fixed in neutral buffered 4% formalin. Paraffin sections (4 μm thin) were stained with haematoxylin and eosin. On light microscopy, absence of the epithelium in the debrided central cornea was noted. Bowman’s layer was unremarkable. There was complete loss of keratocytes down to a depth of 250–280 μm with normal keratocyte density beneath. Descemet’s membrane and the endothelial cells were complete and intact (Fig. 1). There were no inflammatory cells and only minimal oedema. Apart from the corneal findings, the lens was partially calcified and the retina atrophic with silicone oil inclusions and a funnel-shaped retinal detachment. Photomicrograph of cross-linked human cornea 24 hr after cross-linking. Note the absence of epithelium, loss of keratocytes down to 250 μm (zone A), intact keratocytes in posterior stroma (zone B) forming a ‘safety zone’, intact endothelium and the absence of inflammatory cells (haematoxylin and eosin stain, × 200). Concurrent findings have been reported after standard CXL applying 3 mW/cm2 of UVA for 30 min: keratocyte damage has been observed down to 300 μm using in vivo confocal microscopy (Mazzotta et al. 2007) like in the animal experiments. Accordingly, a treatment effect as evaluated by hydration studies (Wollensak et al. 2007) and slit-lamp biomicroscopy (Seiler & Hafezi 2006) is also noted only in the anterior stroma down to 250–300 μm. Interestingly, the cytotoxic threshold for the endothelium is only 0.36 mW/cm2 (Wollensak et al. 2003b) compared to 0.5 mW/cm2 (Wollensak et al. 2004) for keratocytes. There is a relatively large safety gap towards the endothelium with still-intact keratocytes: otherwise, the endothelium with its lower cytotoxic threshold would be destroyed despite keratocyte survival directly anterior to the endothelium. In conclusion, the present article confirms earlier findings in animals and demonstrates a high degree of safety for the endothelium in standard CXL as long as a stromal thickness of at least 400 μm is present in the irradiated target area of the cornea.