Hyperbaric Oxygenation Increases Available Oxygen and Resistance to Enzymatic Digestion in Accelerated Corneal Collagen Crosslinking

Abstract

Purpose: The present study aimed to determine whether hyperbaric oxygenation increases intrastromal oxygen concentration during accelerated corneal collagen crosslinking (CXL) and whether this enhances CXL effects.

Methods: Oxygen concentration was measured in porcine corneas during epithelium-off CXL with UV-A intensity of 9 mW/cm² for 10 minutes. Experiments were performed under normobaric and hyperbaric (2.4 bar) conditions, with and without supplemental oxygen. Stromal resistance to lysis by collagenase A and biomechanical changes were measured.

Results: Oxygen concentrations before irradiation averaged 11% in the normobaric group (CXL_norm), 37% in the normobaric group with supplemental oxygen (CXL_norm + O₂), 13% in the hyperbaric group (CXL_hyp), and 43% in the hyperbaric group with supplemental oxygen (CXL_hyp + O₂). After the onset of UV-A, oxygen concentration decreased to 1.5% in CXL_norm, 3% in CXL_hyp, 21% in CXL_norm + O₂, and 38% in CXL_hyp + O₂. Differences between groups were statistically significant (P < 0.0001). Corneal lenticules from the CXL_hyp + O₂ group were completely digested by collagenase after 5.4 days (standard deviation = 1.34). This was significantly later (P < 0.05) than in other groups. Biomechanical testing did not reveal meaningful differences between the treatment protocols.

Conclusions: These data suggest that hyperbaric oxygenation increases intrastromal oxygen in porcine corneas, ensuring constant availability during accelerated CXL. This leads to enhanced resistance to collagenase digestion of anterior corneal stroma.

Translational Relevance: More intrastromal oxygen and resistance to proteolytic enzymes may make CXL with hyperbaric oxygenation an improved treatment for progressive keratoconus and (infectious) corneal ulcers.

Introduction

Riboflavin–UV-A corneal collagen crosslinking (CXL) can halt progression of keratoconus, reducing the number of corneal transplants required.¹ The classic (“Dresden”) protocol relies on UV-A irradiance of 3 mW/cm² for 30 minutes.² To shorten this procedure, modified protocols have been proposed that use increased irradiance for a shorter duration (e.g., 9 mW/cm² for 10 minutes).³ According to the Bunsen–Roscoe law of reciprocity, as long as the product of irradiance and time of exposure remains the same, the photochemical effect should be the same.⁴ However, accelerated protocols may still be less effective due to the fact that oxygen is consumed faster at increased irradiance.^5,6 Since oxygen is a crucial factor for the photochemical reaction, reduced oxygen availability may lead to a lower CXL effect.⁷ Consistent with this notion, Wang et al.⁸ demonstrated an improvement in stiffness of porcine corneas when oxygen was supplemented during CXL. 

To study the dynamics of stromal oxygen levels during CXL, intracorneal oxygen measurements can be performed using a fiber-optic microsensor.^9–11 Previous studies have demonstrated a rapid decline in oxygen concentration within seconds of the onset of UV-A irradiation.^9–11 After switching off UV-A light, the oxygen concentration returns to baseline within a few minutes. Seiler et al.¹¹ demonstrated that oxygen concentration decreases with tissue depth, underscoring the importance of tissue penetration of oxygen during the course of CXL. 

Hyperbaric oxygenation utilizes a medical treatment chamber to raise the partial pressure of oxygen. This has been used therapeutically to increase the diffusion rate of oxygen into different (ocular) tissues.¹² We have previously shown that corneal stromal oxygen concentration can be enhanced by hyperbaric oxygenation during CXL using the Dresden protocol.¹³ Here, we report that hyperbaric oxygenation results in a meaningful elevation of stromal oxygen levels during accelerated CXL and provide evidence that it can improve the effect of accelerated CXL on corneal tissue resistance to enzymatic degradation. 

Methods

Experiments used porcine eyes from a local slaughterhouse (Theo Keinhörster Großschlachterei mit Fleischhandel e.K., Recklinghausen, Germany). In using this tissue, this investigation adheres to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Animals were approximately 6 months old. Experiments were performed within 36 hours after tissue collection. Four different CXL treatment protocols were used; these are shown in the Table. An untreated control group was analyzed for comparison. During CXL, stromal oxygen concentration was measured as described below. Following CXL, enzymatic testing was performed on five corneas, and biomechanical testing was performed on five corneas in each group (except CXL_hyp, where no enzymatic testing was performed to safe resources). 

Table.

 

Treatment Conditions for Each of the Four Experimental Groups Plus the Control Group

 

Hyperbaric conditions of 2.4 bar were achieved using a cylindrical multiperson medical pressure chamber (Sayers/Hebold Druckkammersysteme GmbH, Cuxhaven, Germany; modernized by Haux-Life-Support GmbH, Karlsbad, Germany). A confined space for controlled oxygen application was created using a plastic box with connections for oxygen supply, an opening for the UV-A light source, and access for riboflavin application and oxygen probe. The experimental setup is shown in Figure 1. In experiments with supplemental oxygen, the box was first flooded with 100% oxygen (2 minutes; 10 L/min), and then a constant oxygen supply of approximately 5 L/min was maintained throughout the experiment. 

Figure 1.

 

Experimental setup. The eye is placed inside the acrylic box underneath the UV-A light source (1). Oxygen (100%) or ambient air passes through an inlet on the left (arrow), while excess gas can leave the box through an outlet on the right (dashed arrow). Riboflavin (asterisk) drops are applied with a needle through a rubber plug in the lid. The oxygen sensor is held in place by a micromanipulator (2); its cable is connected to an oximeter outside the box (not shown).

Using a femtosecond laser (Ziemer Ophthalmic Systems AG, Port, Switzerland), a tunnel for intrastromal oxygen measurement was created at a depth of 300 µm in the stroma of each cornea to ensure that measurements always took place at the same depth. The tunnel was 1.2 mm in horizontal width and 5 mm in length (Fig. 2A). Oxygen was measured in the cornea using an oximeter with a 50-µm diameter needle-type housing fiber-optic oxygen microsensor (PreSens Precision Sensing GmbH, Regensburg, Germany) that was placed inside the tunnel with the tip positioned in the central cornea (Fig. 2B). 

Figure 2.

 

Tunnel for intrastromal oxygen measurement. Schematic representation of the tunnel with exact dimensions: the tunnel was of 1.2 mm horizontal width/diameter, 5 mm length, and placed at a depth of 300 µm below the corneal surface (A). Photo of the tunnel with needle-type housing of the oxygen probe (B). This needle acts as a mandrin, which is retracted, leaving only the oxygen probe in place.

An accelerated CXL treatment protocol was used. The corneal epithelium was removed with a hockey knife and 0.1% riboflavin-5-phosphate (Sigma-Aldrich, St. Louis, MO, USA) in 20% dextran 500 (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) applied every 5 minutes for 30 minutes before irradiation and throughout the subsequent irradiation. Specimens were irradiated using a CCL-365 vario crosslinking system (MLase AG, Germering, Germany) at 9 mW/cm² for 10 minutes, with the light beam being adjusted to a diameter of 11 mm. 

Following CXL, resistance of the tissue to enzymatic digestion was tested. For this purpose, five corneas from the experimental groups CXL_norm, CXL_norm + O₂, CXL_hyp + O₂, and control were dissected using the femtosecond laser to create corneal lenticules with a thickness of 80 µm and a diameter of 7.5 mm, taken from the anterior corneal stroma. Lenticules were dyed with Trypan blue solution (0.4%; Sigma-Aldrich). Each specimen was incubated in 5 mL of 0.1 U/mL collagenase A (Roche, Basel, Switzerland, enzyme activity 0.223 U/mgL) in phosphate-buffered saline (pH 7.4) at 37°C. Specimens were inspected daily to document the day of complete digestion. 

Changes in biomechanical properties of the corneas after CXL were tested using a material testing machine (ZwickiLine, ZwickRoell GmbH & Co. KG, Ulm, Germany). First, a pressure test was performed on the whole eye. For this purpose, the treated eye was placed in a bulb holder. The compression test started at a preload of 0.003 newton and was performed at a rate of 1 mm/min up to a maximum deformation of 2 mm. After the compression test, a 5.5-mm-wide central corneal strip was cut using a punch. This cut was oriented perpendicularly to the femtosecond laser tunnel, so that any potential effect of the laser incision on corneal biomechanics would be standardized. These strips were stored overnight in deswelling medium (Dulbecco’s Modified Eagle's Medium [Sigma-Aldrich]; 1% penicillin/streptomycin [Sigma-Aldrich]; 6% dextran 500 [Carl Roth]). On the following day, tensile testing was performed. The corneal strip was inserted vertically into the material testing machine with a clamp distance of 6 mm and a preload of 5000 Pa. Further strain was applied at a rate of 1.5 mm/min up to a maximum strain of 15%. 

Data presentation and statistical analysis were performed using GraphPad Prism statistical software version 9.1.0 for macOS (GraphPad Software, San Diego, CA, USA). 

Results

Intrastromal oxygen measurements are shown in Figure 3. Baseline values were 11% ± 1.4% (mean ± standard deviation) in the CXL_norm group, 13% ± 1.9% in the CXL_hyp group, 37% ± 9.4% in the CXL_norm + O₂ group, and 43% ± 4% in the CXL_hyp + O₂ group. In all treatment groups, oxygen concentration decreased at the onset of UV-A irradiation. In the CXL_norm group, oxygen concentration dropped to approximately 1.5% within a few seconds. In the CXL_hyp group, oxygen concentration fell more slowly within about 1 minute to a value of approximately 3%. The drop in the CXL_norm + O₂ group was also slower. Within about 3 minutes, oxygen concentration in this group dropped to approximately 21%. In the CXL_hyp + O₂ group, the oxygen concentration fell by only a few percent within approximately 3 minutes to a value of approximately 38%. 

Figure 3.

 

Oxygen concentrations during CXL. Mean oxygen concentration in percent O₂ is plotted against time, with the colored area representing standard deviation. Riboflavin was applied 5 minutes after the start of irradiation (UV on). After a total of 10 minutes of irradiation, the light source was switched off (UV off). Results of the multiple group comparison (one-way analysis of variance) of the AUC of the oxygen concentration over time are shown in the table below the diagram. ****P < 0.0001.

At the time of additional riboflavin application, 5 minutes after the start of irradiation, a mechanically induced deflection of the oxygen measurement curves was observed in all treatment groups. 

To determine the cumulative oxygen uptake throughout the observed temporal span, the area under the curve (AUC) was calculated with 0% O₂ as baseline. The mean AUC over time was 1659, 16,410, 3603, and 26,158 for group CXL_norm, CXL_norm + O₂, CXL_hyp, and CXL_hyp + O₂, respectively. Analysis of variance of the AUC showed a significant difference for each group comparison (P < 0.0001). Hence, measured oxygen concentration over time in the CXL_hyp + O₂ group was significantly higher than in all other experimental groups. This was followed in descending order by the oxygen concentrations in the CXL_norm + O₂, CXL_hyp, and CXL_norm groups. 

The resistance of corneas to degradation by collagenase A is shown in Figure 4. Corneal lenticules treated under hyperbaric conditions with additional oxygen (CXL_hyp + O₂) were completely digested after 5.4 ± 1.34 days. Thus, their enzymatic digestion occurred significantly later (P < 0.05) than in untreated control corneas (2.4 ± 2.07 days) and corneas treated with CXL under normobaric conditions with (2.4 ± 1.52 days) and without (2.2 ± 1.30 days) additional oxygen supply. 

Figure 4.

 

Collagenase assay. Mean values (± standard deviation) of time (in days) until complete digestion for stromal lenticules from untreated control corneas, from specimens crosslinked under normobaric conditions with and without supplemental oxygen, and from specimens crosslinked under hyperbaric conditions plus oxygenation. *P < 0.05; two-way analysis of variance.

Results of compression tests are shown in Figure 5. Compared to all other groups, significantly more force was required to deform the CXL_hyp + O₂ group (all P values <0.001). No statistically significant difference was found between the control group and the CXL_norm group or between the CXL_norm + O₂ group and the CXL_hyp group. Interestingly, greater forces were required for corneal indentation in the control group than in the CXL_norm + O₂ group (P = 0.0063) and the CXL_hyp group (P = 0.0022). Similarly, greater forces were needed to indent corneas in the CXL_norm group compared with the CXL_norm + O₂ group (P = 0.0013) and with the CXL_hyp group (P = 0.0004). 

Figure 5.

 

Indentation tests. Deformation of corneas plotted against force applied for this purpose. (A) Mean values of all groups. (B) Control group with standard deviation (SD). (C) CXL_norm with SD. (D) CXL_norm + O₂ with SD. (E) CXL_hyp with SD. (F) CXL_hyp + O₂ with SD. Results of the multiple group comparison (one-way analysis of variance) are shown in the table below the diagram. ns = not significant; **P < 0.01; ***P ≤ 0.001; ****P < 0.0001.

Results of tensile tests are shown in Figure 6. Significantly less force was required for stretching in the control group compared to all treatment groups (P < 0.0001). Similarly, the force required for stretching was significantly lower in the CXL_norm group than in the CXL_norm + O₂ group (P < 0.0001) and the CXL_hyp group (P = 0.0229). In the CXL_norm + O₂ group, significantly more force was also required for stretching than in the CXL_hyp group (P = 0.0004) and the CXL_hyp + O₂ group (P < 0.0001). Significantly higher forces were measured for the CXL_hyp group than for the CXL_hyp + O₂ group (P = 0.0137). Only between the CXL_norm and CXL_hyp + O₂ groups was there no statistically significant difference. 

Figure 6.

 

Tensile tests. Elongation of corneal strips plotted against force applied for this purpose. (A) Mean values of all groups. (B) Control group with standard deviation (SD). (C) CXL_norm with SD. (D) CXL_norm + O₂ with SD. (E) CXL_hyp with SD. (F) CXL_hyp + O₂ with SD. Results of the multiple group comparison (one-way analysis of variance) are shown in the table below the diagram. ns = not significant; *P < 0.05; ***P < 0.001; ****P < 0.0001.

Discussion

Accelerated CXL protocols have not persistently shown the same effectiveness as the standard (“Dresden”) protocol. Some authors have attributed this to insufficient availability of oxygen.¹⁴ Increasing oxygen levels could improve the effectiveness of modified CXL protocols, potentially making keratoconus treatment safer and less time-consuming.^7,9 To this end, other authors have provided oxygen curves from experiments under normobaric conditions with and without supplemental oxygen, which, despite different CXL protocols, measurement methods, and absolute values, are in line with the data presented here.^9,11 We extend these works by showing that corneal stromal oxygen availability during accelerated CXL can be further increased using hyperbaric oxygenation. 

Crosslinks induced by the photochemical process cannot be detected by light microscopy, which is why other methods are used to demonstrate the effectiveness of CXL in vitro.¹⁵ Crosslinked corneas show a higher resistance to degradation by proteolytic enzymes.^16–18 The results presented here suggest slower degradation by collagenase A in the hyperbaric treatment group with supplemental oxygen and thus evidence of greater corneal stability. This may be clinically meaningful, since it has been reported that collagenases and other proteolytic enzymes play a role not only in corneal ulceration¹⁹ but also to some degree in keratoconus.^20,21 The effect is caused by a change in the tertiary structure of collagen fibrils, leading to inhibition of specific cleavage sites.²² If CXL not only halts progression of keratoconus but can also decelerate corneal melting,²³ this effect may be reinforced by hyperbaric oxygenation. In addition, CXL may have antimicrobial effects that are attributed, among other factors, to destruction of nucleic acids by reactive oxygen species.²⁴ It is tempting to speculate whether this effect may be strengthened by the increased availability of oxygen during CXL with hyperbaric oxygenation. Note that we did not investigate corneal resistance to enzymatic degradation following hyperbaric treatment only (without oxygen supplementation). This decision was made based on our findings that oxygen measurements in this experimental group were only slightly higher than under normobaric conditions. It is unlikely that in clinical practice, the hyperbaric treatment chamber would be used when supplemental oxygen at ambient pressure can achieve higher corneal stromal oxygen levels. 

Contrary to our expectations, our method was not able to detect differences between untreated corneas and those treated with a protocol other than CXL_hyp + O₂. This contrasts with reports from the literature, where the effect of CXL²⁵ could be detected using enzymatic digestion. However, similar to our work, the study by Aldahlawi et al.²⁵ recorded the progress of digestions once daily; this may be one reason why they were able to detect differences only between nonirradiated corneas and crosslinked corneas but not between different CXL protocols. It may be that this method allows only the detection of gross differences. This is suggested also by looking at a study by Hafezi and colleagues,²⁶ who inspected corneas hourly to determine the time until complete digestion, allowing them to detect more subtle differences between different CXL protocols. 

The force required to indent the corneas was measured in the present study using a material testing machine with the objective to demonstrate increased rigidity attained by CXL. To our knowledge, this form of material testing has not been used so far to verify the effectiveness of CXL, so our results cannot be compared to those of others. Other researchers have also used indentation to assess corneal biomechanics following CXL²⁷; however, their method uses triangular corneal segments placed on a plastic plate rather than whole globes. More force was required for indentation in the CXL_hyp + O₂ group than for any of the other groups. This indicates that the combination of hyperbaric conditions and supplemental oxygen may have produced the best CXL outcome. However, the finding that more force was required for deformation of untreated controls compared with other CXL groups suggests that additional factors are at play. One main influencing factor will certainly be intraocular pressure. Since intraocular pressure was not controlled for in the present study, these results should be interpreted with caution. 

For additional biomechanical testing, a tensile strength test was performed. This method has been relied upon in a number of previous works to test the effectiveness of CXL.^28–30 Wollensak et al.³⁰ used it to show that “standard” CXL significantly increases corneal stiffness. Other authors demonstrated an increase in tensile strength at an irradiance of 9 mW/cm², as was used in the present work.^5,31,32 Furthermore, some authors investigated the effect of supplemental oxygen during CXL on the biomechanical properties of the cornea.^8,9,33 These results of studies involving supplemental oxygen are, however, inconclusive. Wang et al.⁸ showed that at 6% and 8% elongation, the Young's modulus of corneas from the CXL group with supplemental oxygenation is higher than in the standard CXL group. This suggests that increased intrastromal oxygen availability may increase the efficacy of CXL. Similarly, Hill et al.⁹ were able to show that increasing the oxygen supply for transepithelial CXL resulted in an improved CXL effect. In contrast to these two studies, Diakonis et al.³³ could not demonstrate any benefit of supplemental oxygen in their work. There was no significant difference between human corneas treated with accelerated CXL with a UV-A intensity of 30 mW/cm² and treatment with the same protocol with supplemental oxygen. In addition, both treatment groups were inferior to the standard protocol. This underscores the difficulty in assessing modifications of CXL protocols in vitro using biomechanical testing. Wollensak and colleagues³⁰ suggest that, in porcine corneas, the effects of CXL on stress–strain measurements are less distinct than in human corneas because of the larger thickness of the specimens, of which probably only the anterior 300 µm are effectively crosslinked. This may be one reason why we did not find any differences between the different CXL protocols using tensile testing. Data from the present study confirm that CXL increases corneal stiffness, as significantly more force was required for stretching all treatment groups than for the control group. However, comparison of the different treatment groups was not consistent with the expected results, as no difference could be confirmed between the CXL_norm and CXL_hyp + O₂ groups, and in addition, more force was required to stretch the corneas from group CXL_norm + O₂ than for group CXL_hyp + O₂. 

In summary, we provide evidence that in this porcine ex vivo model, hyperbaric oxygenation sustains elevated corneal stromal oxygen levels during accelerated, epi-off CXL with UV-A irradiation of 9 mW/cm². This may enhance the CXL effect, as indicated by higher resistance of anterior corneal stromal tissue to enzymatic digestion. Whether this may reduce the risk of perforation from corneal melt in microbial keratitis remains to be established. Also, whether it improves the clinical effect of accelerated CXL on corneal biomechanics in keratoconus warrants further research. 

Acknowledgments

Supported by Rolf M. Schwiete Foundation, Mannheim, Germany (grant 2020-021). 

Disclosure: J. Menzel-Severing, None; T. Streit, None; J. Schmiedel, None; S. Dreyer, None; T.G. Seiler, None; J. Witt, None; G. Geerling, None