Due to sample availability and the desire to treat all corneas within 12 h of death to minimise the effects of posthumous corneal swelling, this study was conducted over the course of three separate experimental runs. Run 1 comprised a preliminary study to determine the impact of riboflavin treatment (without UVA exposure) on corneal enzymatic resistance and Runs 2 and 3 assessed CXL effectiveness in corneas treated with and without a pre-UVA rinse and ensured reproducibility of the most clinically relevant findings.

In total, 68 porcine eyes with clear, intact corneas were received on ice from a licenced European abattoir within 6 h of death. The eyes were returned to room temperature immediately prior to use and divided into the six groups detailed below. In all eyes, the corneal epithelium was removed using a single edged razor blade.

Group 1: Untreated control (Run 1: n = 7; Run 2: n = 10; Run 3: n = 6): De-epithelialised cornea that received no riboflavin or UVA irradiation.

Group 2: Riboflavin only control (Run 1: n = 5): 0.1% riboflavin in 1.0% HPMC (Mediocross M, Avedro Inc., USA) was applied to the de-epithelialised corneal surface at 2-min intervals for 16 min.

Group 3: Cross-linking (CXL) with no BSS rinse (Run 2: n = 5; Run 3: n = 10): 0.1% riboflavin in 1.0% HPMC was applied to the de-epithelialised corneal surface at 2-min intervals for a total of 16 min. This was immediately followed by a 10-min exposure to 365 nm UVA light with an irradiance of 9 mW/cm2 using a CCL-Vario cross-linking device (Peschke M, Peschke trade GmbH, Huenenberg, Switzerland) set to a beam diameter of 11 mm.

Group 4: CXL with a 0.25 mL pre-UVA BSS rinse (Run 2: n = 5; Run 3: n = 10): 0.1% riboflavin in 1.0% HPMC was applied to the de-epithelialised corneal surface at 2-min intervals for a total of 16 min. This was followed by a rinse of the corneal surface with 0.25 mL BSS (Dulbecco’s Phosphate buffered saline), and a 10-min exposure to 365 nm UVA light with a fluence of 9 mW/cm2.

Group 5: CXL with a 1 mL pre-UVA BSS rinse (Run 2: n = 5): 0.1% riboflavin in 1.0% HPMC was applied to the de-epithelialised corneal surface at 2-min intervals for a total of 16 min. This was followed by a rinse of the corneal surface with 1 mL BSS (Dulbecco’s Phosphate buffered saline), and a 10-min exposure to 365 nm UVA light with a fluence of 9 mW/cm2.

Group 6: CXL with a 10 mL pre-UVA BSS rinse (Run 2: n = 5): 0.1% riboflavin in 1.0% HPMC was applied to the de-epithelialised corneal surface at 2-min intervals for a total of 16 min. This was followed by a rinse of the corneal surface with 10 mL BSS, and a 10-min exposure to 365 nm UVA light with a fluence of 9 mW/cm2.

In each experimental run, one eye from each group was treated in sequence to ensure that posthumous corneal swelling effects were evenly distributed amongst the groups. The riboflavin instillation time was increased from the manufacturer recommended 10 to 16 min to account for the greater thickness of porcine corneas compared to human corneas and ensure maximal riboflavin penetration throughout the thicker porcine corneal stroma. For corneas undergoing a surface rinse as part of their treatment (Groups 4–6), BSS was used to avoid any potential changes in stromal pH, which might adversely affect oxygen availability and hinder the cross-linking process [19], and this was applied in a dropwise manner (at a rate of approx. 0.5 mL/s). The central corneal thickness (CCT) of each eye was measured using a SP-100 portable pachymeter (Tomey GmbH Technology and Vision, Nurnberg, Germany) before and after epithelium debridement, and following each stage of treatment (post-riboflavin application, post-rinse, and post-UVA exposure, where applicable). Immediately after completion of each treatment, the surface of the cornea was gently wiped with a tissue to remove any excess riboflavin/BSS and an 8.0-mm full-tissue thickness disk was trephined from the centre of each cornea. The corneal disks were then wrapped tightly in catering film and refrigerated until all corneas had been treated, after which they were returned to room temperature.

Corneal disks were placed into individual wells each containing a 2 mL solution of 0.3% Collagenase A obtained from Clostridium histolyticum (Sigma, UK) and incubated at 37 °C. In Runs 1 and 2, samples were maintained in a static incubator. In Run 3, to ensure continuous sample agitation and expedite the rate of digestion, a 37 °C/200 rpm incubator-shaker (Incu-Shake MIDI, SciQuip, Newtown, United Kingdom), was used. At regular intervals (every 1.5 to 2 h during the day and at 8-hourly intervals during the night) the samples were removed from their respective incubators and examined under a microscope to assess their integrity. In each run, the time required for total tissue digestion was recorded for each sample.

A further 15 porcine eyes with clear, intact corneas were obtained on ice from the abattoir within 6 h of death and divided into five groups. The corneas underwent the same 16-min riboflavin application and corneal rinse protocols described previously for Groups 1, 3 to 6 (n = 3 per group) but importantly, these samples remained unirradiated. Post treatment, the corneas were immediately excised, immersed in liquid nitrogen for 5 min and transferred to storage at − 80 °C. A 1-mm wide, full-tissue thickness corneal strip was cut from the center of each cornea (along the horizontal meridian), and the cross-sectional surface of each strip was placed face-down onto a microscope slide. A CoverWell™ Imaging Chamber Gasket (Thermofisher, UK) was placed on top of the sample. To minimise the migration of riboflavin within the corneal strip prior to imaging, no immersion fluid was used. The time taken to excise and mount the tissue prior to imaging was under 2 min.

Riboflavin penetration was evaluated by measuring the fluorescence intensity of riboflavin within the anterior-most region of the stroma. Samples were imaged on a Zeiss LSM 880 META NLO microscope (Carl Zeiss Ltd, Welwyn Garden City, UK). Based on the absorption spectrum for riboflavin, an excitation wavelength of 458 nm was used. A single image (using a 20X/0.8 NA air objective) was obtained at a resolution of 1024 by 1024 pixels per image and with a lateral resolution of 0.42 μm per pixel. Experimental settings (including detector gain) were unchanged for all samples to ensure comparable intensity values.

Grayscale images were exported and analysed using ImageJ imaging software [20]. The plot profile tool in ImageJ was used to calculate the fluorescence intensity of riboflavin as a function of tissue depth. A line profile was drawn normal to the anterior-most surface of the cornea down to a depth of 300 µm, and the fluorescence intensity profiles for each image were exported into Excel. The fluorescence intensity data for each image was then averaged within each treatment group (n = 3 images per treatment group) and plotted as intensity profiles to show the fluorescence intensity of riboflavin as a function of tissue depth.

Changes in CCT between the different stages of treatment were statistically assessed by means of paired t-tests. Comparisons between groups in terms of the time required for total digestion were performed using single-factor ANOVA and post hoc least significant difference testing. Data are shown as mean measurements (± SD) for corneal thickness and digestion times. All statistical analyses were performed using Microsoft Excel. A probability value of less than 0.05 was considered significant.