Introduction

We have recently developed a new method for the treatment of keratoconus by inducing collagen crosslinking in the cornea using ultraviolet A (UVA) and the photosensitizer riboflavin. Collagen crosslinking leads to an increase in intra- and interfibrillar covalent bonds by photosensitized oxidation and causes a biomechanical stabilization of the cornea. In a prospective clinical pilot study including 22 patients with moderate or advanced progressive keratoconus and with a follow-up time of up to 4 years, the progression of keratoconus could thus be stopped in all treated eyes. Regression with a reduction of the maximal keratometry readings by 2 diopters was achieved in 70% of patients.1

In stress–strain measurements of the human,2 porcine,2, 3 and rabbit4 cornea, a significant increase of mechanical rigidity was found. An increased resistance to collagenases could be demonstrated.5 Besides keratoconus and corneal ulcers, the new treatment can be used in corneal melting processes6 and in the field of refractive surgery to reduce the risk for iatrogenic keratectasia after LASIK (laser assisted in situ keratomileusis).7

The impact of various treatment modalities like corneal abrasion, PRK (photorefractive keratectomacy),8 LASIK,8, 9 and epithelial injury10, 11, 12 on keratocytes has gained considerable interest in recent years having a possible influence on scarring or corneal thinning.10 Therefore, as part of a comprehensive safety evaluation, we have undertaken the following experimental study to assess the possible damage to corneal keratocytes after combined riboflavin/UVA-treatment and to compare it to the effect of UVA-irradiation alone.

Material and methods

Materials

All primary cultures and serial passaging were carried out in growth media consisting of Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS, from Sigma, Deisenhofen, FRG), phenol red, and antibiotics (penicillin–streptomycin). Trypan-blue solution13 (Biochrome AG, Berlin, FRG) and Yopro (Molecular probes, Eugene, OR, USA)-fluorescent stain were used for the viability assay.14

Explant culture conditions

Explant cultures13 were established from porcine eyes, which were obtained from a local slaughterhouse within 3–5 h postmortem. The epithelium was removed mechanically using a blunt hockey knife. Pieces 6 × 6 mm of mid-stromal lamellae were dissected by a pair of scissors, and transferred into a 25 cm2 cell culture flask (Nunc, Wiesbaden, FRG). The flask was filled with 2 ml DMEM plus 10% foetal calf serum and placed into an incubator gassed with carbon dioxide 6% at 37°C. Cell outgrowth from the stromal explants started after 10–12 days, and confluence with about 2.5 × 106 cells per flask was reached after 21 days. The media were exchanged every week. The confluent stock cultures were passaged every 3 weeks with a split ratio of 1 : 3. For dissociation and detachment, the cultures were treated with 0.05% trypsin–EDTA solution. The free cells were centrifuged at 230 g, transferred into new culture flasks and resuspended. The cell growth was evaluated regularly using an inverted phase-contrast microscope at × 100 magnification (Zeiss Axiovert L5). Before the irradiation, 5 × 104 cells were plated per well in eight-well tissue plates, where they reached confluence after 8–10 days. After another week, the UVA-irradiation was performed.

Treatment groups

The cells were exposed to three different treatments:

  1. 1

    Riboflavin alone: The cells were exposed to 25, 50, 100, and 500 μM riboflavin solution.

  2. 2

    Riboflavin plus UVA: The cells were exposed to 0.025% (=500 μM) riboflavin solution plus UVA-irradiances ranging from 0.4 to 1 mW/cm2 (Table 1).

    Table 1 Cytotoxicity for porcine keratocytes after combined riboflavin/UVA-treatment and after UVA alone
    Full size table
  3. 3

    UVA alone: The cells were exposed to UVA-irradiances ranging from 2 to 9 mW/cm2 (Table 1).

Except for the varying UVA-irradiances, the other parameters like riboflavin concentration or the UVA-wavelength of 370 nm were kept identical to the clinical treatment in the second treatment group.

Irradiation procedure

A riboflavin concentration as close as possible to the conditions of the clinical treatment of human corneas was chosen and calculated as follows. In humans, 0.1% riboflavin solution is applied. Using the so-called diffusion equation and the diffusion coefficient (D=6.5 × 10−7 cm2/s) of the related dye fluorescein, we calculated the average riboflavin concentration over 30 min in the stroma as 0.024% riboflavin solution. Therefore, we used 0.025% riboflavin solution (=500 μM) by adding 57 μl of 0.2% riboflavin stock solution to 400 μl colourless culture medium without phenol red. No riboflavin was added to the cultures with UVA-treatment alone. The riboflavin solution for the UVA-irradiation was added to the wells 5 min before the irradiation and was replaced by the cell medium after the irradiation. To avoid UVA-absorption by riboflavin solution overlying the monolayer of keratocytes attached to the floor of the wells, we irradiated the wells from underneath fixating the UVA-double diode (370 nm) 1 cm under the respective wells (Figure 1) with the help of a stand. The UVA-absorption by the 100 μm thin floor of the wells made of borsilicone was measured to be only 2% and was negligible. Before the treatment, the required UVA-irradiances ranging from 0.4 to 9 mW/cm2 were controlled with a UVA-meter (LaserMate-Q, LASER 2000, Wessling, Germany) at a 1 cm distance and if necessary regulated with a potentiometer in series. The irradiation itself lasted 30 min, which is conform to the clinical setting. Five wells were tested for each irradiance level. The irradiation doses (J/cm2) were calculated from the UVA-irradiances (mW/cm2) by multiplying the value with the irradiation time in seconds (=30 × 60).

Figure 1
figure 1

Irradiation procedure with double UVA-diode under the wells of the culture plate.

Full size image

Supravital staining

To determine possible cell damage after treatment, 100 μl of 0.25% trypan-blue solution dissolved in colourless culture medium was applied per well for 15 min followed by twofold rinsing with culture medium. After microscopic evaluation of the trypan-blue staining,13 the cells were subsequently stained with Yopro adding 1 μl per well followed by one rinse with culture medium.14 For trypan blue (Figures 2 and 3), cultures were examined in an inverse microscope (Leica DMIR) at 100–400-fold magnification using differential interference contrast and for Yopro (Figure 4) using fluorescence at 488 nm (N2.1 filter). Photos were taken with a digital camera attached to the microscope (Nikon coolpix 950). Only the nuclei of damaged cells were labelled with the stains.

Figure 2
figure 2

Representative morphology of the cultured porcine keratocytes at confluence (phase-contrast, × 100).

Full size image
Figure 3
figure 3

Peripherial sector of the circular treatment area (riboflavin plus 0.5 mW/cm2 UVA) with trypan-blue-positive keratocytes and loss of cells (trypan blue, × 400).

Full size image
Figure 4
figure 4

Peripherial sector of the circular treatment area (5 mW/cm2 UVA) with Yopro-stained keratocyte nuclei (Yopro, × 400).

Full size image

Results

Cellular characteristics

Before confluence, the scattered cells appeared dendritic, whereas they assumed a spindle-shaped appearance when confluence was reached (Figure 2). About 10% of the cells contained some brown-coloured, autofluorescent granules as is typical for lipofuscein deposits. After cytotoxic doses, the damaged cell nuclei stained positively with trypan blue (Figure 3) and the green-fluorescent Yopro (Figure 4). The cell damage detected by the two stains was always observed in the same cells and at the same irradiance level. In addition, dead cells became more globular in shape and about 5–10% of the damaged dead cells lost adhesion floating off the bottom of the wells. The area of the damaged cells was shaped like a circle quasi like an imprint of the UVA-beam with surviving cells only in the nonirradiated periphery (Figures 3 and 4).

Cytotoxicity

The cytotoxic irradiance level was found to be 10-fold lower at 0.5 mW/cm2. (=0.9 J/cm2) for keratocytes after riboflavin combined with UVA-irradiation compared to 5 mW/cm2 (=9 J/cm2) for cells with UVA-irradiation alone (Table 1). The cytotoxic effect occurred in a threshold-like manner with a sharp and abrupt transition from damaging to nontoxic irradiance levels. There was no variability in the five different wells per irradiance level and always all cells irradiated with the toxic dose were damaged. There was no cytotoxicity for the cells treated with riboflavin alone without UVA-irradiation.

Discussion

The present study has shown an abrupt threshold-like cytotoxic irradiance level of combined riboflavin/UVA-treatment at 0. 5 mW/cm2. Using the standard surface irradiance of 3 mW/cm2 and the absorption coefficient of 53 cm−1, it can be calculated according to the Lambert–Beer equation Idepth=Isurfacee(−dμ) that in human corneas the cytotoxic keratocyte UVA-irradiance of 0.5 mW/cm2 is reached down to a stromal depth of 300 μm.4, 15 These data fit well to our in vivo results in riboflavin/UVA-treated rabbit corneas, where we found exactly the same cytotoxic irradiance of 0.5 mW/cm2 with a keratocyte loss down to a depth of 300 μm after 3 mW/cm2 surface irradiance.16

UV-induced damage of keratocytes has also been reported by others and depends on the ultraviolet wavelength and dose. While ultraviolet C (UVC) (100–290 nm) is completely absorbed at the corneal surface and 80% of ultraviolet B (UVB) (290–315 nm) in the corneal epithelium, 25–34% of UVA (315–400 nm) is absorbed in the stroma.17 Accordingly, massive keratocyte damage was observed especially in the anterior portion of the rabbit cornea following exposure to UVA-irradiation at 350 nm.18, 19

In the present in vitro study, the cytotoxic irradiance level after combined riboflavin/UVA-treatment was about 10-fold lower than that after treatment with UVA alone because the cytotoxic effect, which is due to the oxidant effect of UVA-light, is multiplied by the photosensitizer riboflavin due to increased UVA-absorption. Concurrent with this observation, UVA-absorption was shown by us to be increased to 95%4 in the cornea after riboflavin treatment compared to 25–35% without riboflavin.17

Riboflavin (vitamin B2) alone produced no cell damage as is also known from other experiments20 and is not surprising because riboflavin is also present in the retina, liver, and heart, being an essential element in normal nutrition.21

In terms of the clinical problems of keratocyte loss, only long-term imbalances of the keratocyte reconstitution bear a risk for corneal thinning.8, 10 However, this is not to be expected because the keratocyte loss can be repaired rapidly by repopulation from migrating keratocytes of the adjacent cornea.8, 22 Loss of keratocytes has been observed already in other corneal procedures without having dramatic consequences. So after PRK,8 LASIK,8, 9 or epithelial injury10, 11, 12 keratocyte apoptosis of variable degree was described. However, treatment-related keratocyte apoptosis might be an issue because of the presumptive role of keratocyte apoptosis in the pathogenesis of keratoconus.8

The state of the keratocytes was assessed 24 h following UVA-irradiation because then there is the maximum degree of cell damage as has been shown previously.16 To determine possible cell death, we first conducted trypan-blue staining followed by Yopro staining.14 Both stains are positive in the nuclei of damaged cells without interfering with cell viability, whereas undamaged endothelial cells are impervious to these dyes. The DNA-intercalant dye Yopro is also able to detect minor damage like in apoptotic cells due to its relatively low molecular weight (MW 629 vs 961).14 Yet, as the threshold for both Yopro and trypan blue was congruent, and at the same dose level, the cellular damage in the present in vitro experiment is obviously mainly due to necrosis because otherwise the Yopro stain should have been positive already at lower irradiance levels than the one found by trypan blue.

In conclusion, we have shown that combined riboflavin/UVA-treatment leads to a 10-fold lower threshold for keratocyte cytotoxicity at 0.5 mW/cm2 compared to 5 mW/cm2 after UVA-irradiation alone. Using riboflavin/UVA in human corneas, keratocyte damage can be expected down to a depth of 300 μm using 3 mW/cm2 surface irradiance. Although no corneal thinning or scarring due to keratocyte loss has been observed so far in a 4-year-clinical trial,1 a long-term study including confocal in vivo microscopy in humans after riboflavin/UVA-treatment is currently underway to evaluate the depth of the keratocyte loss, the repopulation process, and to exclude reliably the development of treatment-related stromal scarring, haze, or thinning in humans.