Comparison of accelerated corneal collagen cross-linking under normal and increased oxygen concentrations: A pilot study

Abstract

Objective:

To evaluate the role of local oxygen supplementation in accelerated collagen cross-linking (CXL) in human subjects.

Method:

Patients with progressive keratoconus were enrolled and randomized to undergo accelerated CXL in either local oxygen supplementation (Group 1, Cases, n = 10) or room-air oxygen (Group 2, Controls, n = 10). Baseline uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), manifest refractive spherical equivalent (MRSE), corneal tomographic parameters, corneal biomechanics, and wavefront analysis were documented. Patients were followed up at 1 week, 1 month, 3 months, and 6 months.

Results:

Twenty eyes of 20 patients (Group 1, Cases, n = 10; Group 2, Controls, n = 10) were enrolled. At 6 months, both groups exhibited significant flattening of steepest keratometry, increased corneal optical densitometry, and reduced thinnest pachymetry. No eye had keratoconus progression in either group. There was no significant change in UDVA, CDVA, and MRSE in both groups. No intergroup significant differences were found in UDVA, CDVA, MRSE, steepest keratometry (Kmax), or corneal astigmatism. Demarcation line depth was comparable between groups at 6 months.

Conclusion:

Accelerated CXL is a useful modality to manage progressive keratoconus. However, in short term, local oxygen supplementation does not seem to have any additional benefit over room-air oxygen. Larger studies with longer follow-up period are required to further investigate the same.

The introduction of collagen cross-linking (CXL) in the late 1990s revolutionized the management of progressive keratoconus. This minimally invasive technique utilizes ultraviolet-A (UV-A, wavelength = 370 nm) light in combination with a photosensitizing agent, riboflavin (Vitamin B2), to form covalent bonds between collagen fibrils, thereby enhancing the biomechanical stability of the cornea and halting the progression of ectasia. However, as demonstrated by several authors, the efficacy of CXL is oxygen-dependent.[1] McCall et al.[2] reported that cross-linking could be suppressed by adding reactive oxygen species (ROS) inhibitors. Subsequent studies confirmed the necessity of oxygen, showing that cross-linking does not occur in oxygen-free environments, such as under helium[3] or argon[4] atmospheres, where hydroxyl radicals are absent. Moreover, reduced oxygen conditions, such as in contact-lens-assisted CXL, result in lower biomechanical effects.[5] Additionally, a few laboratory studies have demonstrated that the efficiency of CXL can be strengthened by oxygen supplementation. However, evidence of the effect of additional oxygen supplementation on the outcomes of accelerated CXL in human eyes is scarce. Therefore, we conducted a pilot study to compare the outcomes of accelerated CXL under normal and increased local oxygen concentration in human participants with progressive keratoconus.

Methods

The study adhered to the tenants of Declaration of Helsinki and was approved by the Institutional Ethics Committee before initiation (IECPG-100/27.02.2020, RT-03/20.05.2020). The study was registered in the Clinical Trials Registry of India (CTRI/2020/11/029122). Patients older than 18 years of age, with progressive keratoconus, with thinnest pachymetry ≥400 microns, and willing to participate were included in the study. “Progression” was defined as change in steepest keratometry (Kmax) greater than 0.5 D, reduction in thinnest pachymetry greater than 5%, and change in manifest refractive spherical equivalent (MRSE) greater than 1 D over 6 months. Those with a history of past ocular surgery, any ocular comorbidity other than keratoconus, corneal scarring, and one-eyed patients were excluded from the study.

Enrolled patients were randomized into the increased oxygen group (Group 1, Cases) and the room air oxygen group (Group 2, Controls). Participants were allocated into either of these two groups using a computer-generated random number table prepared prior to recruitment to ensure unbiased randomization. In both groups, 20% ethyl alcohol assisted epithelium debridement was performed in the central 8 mm of the cornea. Iso-osmolar riboflavin solution (0.1%) was used to presoak the cornea for 20 minutes, 2 drops being instilled every 2 minutes.

In Group 1 (Cases), our indigenous goggle-apparatus for oxygen delivery was fitted over the drape 20 minutes after soaking the cornea with riboflavin and before initiating irradiation of UV-A light [Fig. 1].[6] This apparatus has two opposing nozzles on the side. Oxygen was delivered into the goggle-chamber through one of the lateral nozzles at a rate of 5L/min. Oxygen sensors of the anesthesia workstation were plugged into the opposite nozzle of the apparatus to measure the oxygen concentration within the goggle chamber in real time. This step of oxygen supplementation was skipped in Group 2 (controls). UV-A irradiation was performed as per the accelerated protocol summarized in Table 1. In Group 1 (Cases), oxygen supplementation was carried out throughout the procedure.

Figure 1.

Comparison of thinnest pachymetry in both Group 1 (Cases) and Group 2 (Controls) at baseline, 1 week, 1 month, 3 months, and 6 months

Table 1.

Collagen cross-linking method

Parameter		Variable
Treatment target		Ectasia in progressive keratoconus
Total fluence (J/cm2)		5.4
Soak time (intervals)		20 (2) minutes
Intensity (mW)		9
Treatment time		10 minutes
Epithelium status		Off
Chromophore		Riboflavin (source name)
Chromophore carrier		Dextran
Chromophore osmolarity		Iso-osmolar
Chromophore concentration		0.1%
Light source		PXL Platinum 330 (Peschke GmbH, Hünenberg, Switzerland)
Irradiation mode		Continuous
Protocol modifications		Accelerated CXL with (Group 1) and without (Group 2) oxygen supplementation

Values are expressed as median (inter-quartile range). CDVA=Corrected distance visual acuity; MRSE=Manifest Refractive Spherical Equivalent; UDVA=Uncorrected Distance Visual Acuity

At the end of the procedure, a bandage contact lens (BCL) was placed over the cornea and a topical antibiotic drop (E/D Moxifloxacin 0.5%) was instilled. In the postoperative regime, E/D Moxifloxacin hydrochloride QID for 2 weeks, E/D Prednisolone phosphate 6t/D tapered 1t/week over 6 weeks, and E/D carboxymethylcellulose 0.5% 6t/d for 6 weeks were prescribed.

Follow-up was done on postoperative day (POD) 1, POD 7, 1 month, 3 months, and 6 months. On follow-up, the assessment included uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), intraocular pressure (IOP), anterior segment evaluation, subjective refraction, anterior segment OCT (ASOCT), Pentacam™ (Oculus Pentacam, Wetzlar, Germany), Corvis-ST (Oculus Corvis-ST, Wetzlar, Germany), and specular microscopy. A blinded observer evaluated the postoperative findings.

The primary outcome measures were change in visual acuity (LogMAR units) and reduction in keratometric astigmatism at 6 months postoperative period. The secondary outcome measures were change in biomechanical strength on Corvis ST, tomographic changes on Pentacam^TM, and endothelial cell count and morphology on final follow-up.

Statistical analysis

The statistical analysis was done using Statistical Package for Social Sciences (SPSS) software (IBM manufacturer, Chicago, USA, ver 21.0). The categorical variables were presented as numbers and percentages (%). Quantitative data with normal and non-normal distribution were presented as mean ± standard deviation (SD) and median [interquartile range (IQR)], respectively. The data normality was checked using the Kolmogorov–Smirnov test. Parametric tests were used for normative data, and nonparametric tests were used for non-normative data distribution. The comparison of the quantitative variables between two groups with non-normal and normal distribution was analyzed using Mann–Whitney test and independent t-test, respectively. Paired t-test/Wilcoxon signed rank test was used for comparison across follow-up. The confidence interval was 95%; therefore, P value <0.05 was considered statistically significant.

Results

A total of 20 eyes of 20 patients were enrolled in the study, including 12 male (60%) and 8 females (40%). The mean age of the patients was 20.6 ± 1.98 years. Ten patients were allotted Group 1 (Cases), and ten patients were allotted Group 2 (Controls). The baseline UDVA, CDVA, and MRSE were comparable in both groups [Table 1]. Similarly, keratometry, pachymetry, posterior elevation, and other tomographic values were comparable at baseline in both groups [Table 2]. Corneal endothelial cell density, hexagonality, and coefficient of variance were also comparable in both groups [Table 3].

Table 2.

Summary of comparison of visual and refractive outcomes between Group 1 (Cases) and Group 2 (Controls) before and after their designated procedure

Parameter		Group 1 (Cases)						Group 2 (Cases)						Intergroup variation (P)
		Baseline		6 months		P		Baseline		6 months		P		Baseline		6 months
UDVA (LogMAR units)		0.77 (0.17,1.17)		0.67 (0.17,1.176)		0.23		0.685 (0.3,1.44)		0.70 (0.17,1.3)		0.18		0.84		0.89
CDVA (LogMAR units)		0.17 (0,0.6)		0.235 (0,0.477)		0.8		0.235 (0.17,0.6)		0.3 (0,0.77)		0.23		0.09		0.51
MRSE (Diopters)		-3.75 (-9, -1)		-3.25 (-11, -1.625)		0.87		-2.82 (-13.3,0)		-3.31 (-13.75,-1.375)		0.44		0.25		0.62
Refractive Sphere (Diopters)		-2.75(-7, -0.75)		-2.375(-9,1)		0.50		-1.25 (-14.75,1.5)		-2.625(-13,-0.25)		0.50		0.08		0.57

Table 3.

Changes in corneal tomography and corneal biomechanics as measured using Schiempflug imaging (from baseline to 6 months postoperative period in both Group 1, Cases and Group 2, Controls)

Parameter		Group 1 (Cases)						Group 2 (Controls)						Intergroup comparison
		Baseline		6 months		P		Baseline		6 months		P		P at baseline		P at 6 months
Flat K (K1 in Diopters)		45.59±2.97		44.26±3.06		0.055		48.07±2.7		47.2±3.34		0.062		0.833		0.149
Steep K (K2 in Diopters)		50.26±4.6		49.18±5.03		0.012*		52.55±4.35		52.19±4.807		0.368		0.596		0.48
Steepest Keratometry (Kmax in Diopters)		56.01±6.07		54.32±6.0		0.037*		59.13±5.52		56.87±6.16		0.041*		0.754		0.52
Corneal Astigmatism (Diopters)		5.55 (0.7, 7.3)		5.3 (1.2, 8.6)		0.151		3.85 (0.3, 9.2)		4.65 (0.8, 9.4)		0.226		0.766		0.97
Corneal Optical Densitometry (Grayscale units)		16.7±1.2		28.43±7.06		0.005*		16.83±0.9		23.2±3.9		0.006*		0.924		0.12
Posterior Elevation (µm)		58 (9, 43)		67 (18,91)		0.059		72.5 (15,49)		72 (9, 100)		0.483		0.216		0.44
Thinnest Pachymetry (µm)		475.5±29.47		439.8±30.6		0.002*		458.9±27.18		437.6±35.6		0.008*		0.541		0.89
Integrated Radius (mm)		11.29±2.35		11±1.65		0.423		11.64±1.87		11.05±1.61		0.07		0.609		0.92
Stress Strain Index		0.776±0.1		0.831±0.168		0.282		0.757±0.189		0.785±0.12		0.63		0.803		0.83

P values with Asterix (*) are statistically significant

At 6 months postoperative period, the median UDVA was comparable between Group 1 (Cases) and Group 2 (Controls) (P = 0.18). Even within the groups, there was no difference between the UDVA at baseline and 6 months (Cases: P = 0.26, Controls: P = 0.33). A similar trend was seen with CDVA and MRSE between the groups and within the groups. The results are summarized in Table 1.

Initially, Group 1 (Cases) showed a flattening in the flat meridian (K₁) at 1 week (P = 0.04) and 1 month (P = 0.03) as compared to Group 2 (Controls). However, the final difference at 6 months in the flat meridian (K₁) between both groups was not statistically significant (P = 0.149).

The comparison of effect of the procedures on tomographic parameters within and among both groups is summarized in Table 3.

Unlike Group 2 (Controls), Group 1 (Cases) exhibited statistically significant flattening in the steep meridian (K₂) at 6 months postoperative period (P = 0.012). However, the difference between the final K₂ of both groups was not statistically significant (P = 0.48).

A statistically significant reduction in the steepest keratometry (Kmax) was noted in both groups, respectively. However, the difference between the Kmax reduction of both groups was not statistically significant (P = 0.52).

At 6 months, the difference between both groups with respect to median corneal astigmatism changed was not statistically significant (P = 0.97).

There was a statistically significant reduction in corneal thickness from baseline to 6 months post-op in both groups, respectively (Thinnest pachymetry Group 1, Cases: 475.5 ± 29.47 µm to 439.8 ± 30.6 µm, P = 0.002; Group 2, Controls: 458.9 ± 27.18 µm to 437.6 ± 35.6 µm, P = 0.008). Group 1 (Cases) exhibited higher mean pachymetry than Group 2 (Controls) at 1 week (P = 0.04). However, this difference resolved over time as the thinnest pachymetry was comparable at 6 months in both groups (P = 0.89) [Fig. 1].

Corneal haze was measured by corneal optical densitometry (COD) on Pentacam^TM. COD peaked at 1 week in both groups, being significantly higher in Group 2 (Controls) than in Group 1 (Cases) (P = 0.002). However, it was comparable between both groups at 1 (P = 0.100), 3 (P = 0.051), and 6 months (P = 0.123) postoperative period. Corneal haze decreased over time but was significantly higher than baseline in both the groups at 6 months [Fig. 2].

Figure 2.

Comparison of corneal optical densitometry in both Group 1 (Cases) and Group 2 (Controls) at baseline, 1 week, 1 month, 3 months, and 6 months

The baseline length of the first applanated segment (AL1) was comparable between both the groups (Group 1, Cases = 1.99 ± 0.2 mm; Group 2, Controls = 1.99 ± 0.2 mm; P = 0.52). The AL1 values increased in Group 1 (Cases) at 1 week of follow-up by a factor of 16.2% (P = 0.0261). However, there was no statistically significant difference between both groups at 1 week (Group 1, Cases = 2.27 ± 0.33mm; Group 2, Controls = 1.9 ± 0.3mm; P = 0.0592). Both the groups continued to be comparable at 6 months of follow-up. The velocity of first applanation (AV1) showed opposite trends with comparable baselines (Group 1, Cases = 0.177 ± 0.03 m/s; Group 2, Controls = 0.181 ± 0.02 m/s; P = 0.76) and a statistically significant decrease of 8% in the case arm at 1 week of follow-up (Group 1, Cases = 1.71 ± 0.31 m/s; Group 2, Controls = 1.59 ± 0.3 m/s; P = 0.03). AV1 values were comparable between the groups at all subsequent follow-ups.

The radius of highest concavity (HCR) parameters in both the arms was comparable at baseline (Group 1, Cases = 5.983 ± 1.06 mm; Group 2, Controls = 5.756 ± 0.706 mm; P = 0.60). However, at 1 week, a statistically significant increase of 16.5% (P = 0.016) was noted in Group 1, Cases (6.94 ± 0.95 mm) as opposed to the control arm (5.79 ± 0.65 mm). However, the HCR parameters remained comparable in both groups at all subsequent follow-ups.

There was a statistically significant decrease from baseline of the integrated radius (IR) in both the groups, respectively, at 1 week of follow-up (P = 0.029). However, IR was comparable to baseline at 6 months follow-up both within the groups and was between both groups. The comparison of effect of the procedures on the biomechanical strength within and among both groups is summarized in Table 3.

The depth of demarcation line was comparable in both groups at 6 months (Group 1, Cases = 269.6 ± 26.85 µm; Group 2, Controls: 258.875 ± 32.9 µm; P = 0.19).

The endothelial cell density was comparable in both groups at baseline with a mean count of 3,137 ± 348 cells/mm² in Group 1 (Cases) and 3,147.4 ± 258 cells/mm² in Group 2 (Controls) (P = 0.90). At 6 months, this changed to 3,020.4 ± 681 cells/mm² in Group 1 (Cases) (P = 0.52) and 3,145.75 ± 344 cells/mm² in Group 2 (Controls) (P = 0.552). There was no statistically significant change in the endothelial cell density after the assigned procedures in either of the groups. The final endothelial cell count at 6 months was also comparable between both groups (P = 0.24). Similarly, hexagonality also remained comparable at baseline and 6 months within both groups. No difference was noted between the groups at 6 months (P = 0.761). A significant increase in CoV was noted among Group 1 (Cases) at 6 months (Baseline = 26.1 ± 4.08, 6 months-29.7 ± 5.29; P value = 0.018); however, this remained comparable at 6 months in Group 2 (Controls). Nevertheless, the final CoV was comparable in between both groups.

Discussion

The role of oxygen in collagen cross-linking is well established. However, as per the current evidence in the literature, it still remains uncertain as to how local supplementation of oxygen effects the outcomes of accelerated CXL. The results of our pilot study indicate that despite the significant flattening in steepest keratometry seen in both group at 6 months, there was no statistically significant intergroup difference in steepest keratometry. Further, UDVA, CDVA, and MRSE did not significantly change after the procedures in both groups. This may be attributed to the fact that as opposed to conventional regimes, efficacy of accelerated corneal crosslinking protocols affects visual outcomes better at longer follow-ups.[7] Although the local stromal concentrations have not been measured in in vivo studies, systemic oxygen supplementation has been shown to improve outcomes in accelerated collagen cross-linking at the 6-month follow-up.[8] Further, the studies carried out in porcine models evaluating several accelerated corneal cross-linking protocols supplemented with increased oxygen concentration suggest an increased biomechanical stability of the cornea.[3,9,10,11] However, even with a similar apparatus design as used by Richoz et al.,[3] we believe the study differs in the quintessential substance of the sample corneas that have been assessed. Since human corneas are much more biomechanically stiffer than the porcine corneas,[12] we believe the rates of oxygen permeation through a stiffer, de-epithelialized human cornea may be much lower than through a de-epithelialized porcine cornea under normal baric conditions.

As suggested by Faramarzi et al.,[8] it is possible that the dissolved oxygen from the systemic circulation may influence the corneal stromal environment. The partial pressures of the dissolved oxygen, which when depleted during accelerated cross-linking, are replaced by diffusion of dissolved oxygen from the local circulation from both the aqueous and the limbal circulation in graded variation.[13,14] This may facilitate why systemic oxygen supplementation in human eyes, rather than direct local oxygenation, may affect better outcomes at accelerated corneal cross-linking. Further, local oxygen supplementation, if used in accelerated corneal collagen cross-linking protocols, may require hyperbaric supplemental conditions to effect better oxygen permeability into the tissues and increased biomechanical corneal strength at subsequent postoperative follow-ups. However, hyperbaric oxygen supplementation may further increase the risk of transient endothelial dysfunction, as noted in our study.

The statistically significant increase in the central corneal thickness, corneal optical densitometry,[15] the corroborating changes in dynamic corneal response (A1L, A1V, HCR), and biomechanical stiffness parameters (integrated radius) among Group 1 (Cases) at 1 week may be suggestive of subclinical corneal edema. We hypothesize a plausible transient endothelial dysfunction due to increased free-radical synthesis in a shorter period of time as dictated by the Roscoe Bunsen law of photodynamic reciprocity.[3] Endothelial dysfunction with the normal Dresden cross-linking protocols has also been occasionally reported.[16,17,18] The increased coefficient of variation at 1 month in Group 1 (Cases) in our study may signify a compensatory mechanism to overcome the effects of ROS-mediated endothelial dysfunction in the early postoperative period.[19] Further, the discrepancy in the time period of occurrence of hypothesized transient corneal edema and coefficient of variation may be considered a part of the normal endothelial cell response to injury as evidenced in rabbit models.[20,21] The delayed advent of stage 2 response to endothelial injury[20] at 1 month in Group 1 (Cases), evidenced by a statistically significant increase in coefficient of variation in the case arm as opposed to rabbit models (7 days), may be attributed in turn to a limited regenerative capacity of human endothelial stem cell reserve at the Schwalbe’s line in humans.[22] The occurrence of subclinical corneal edema in the oxygen supplemented Group 1 (Cases), and not Group 2 (Controls), during the cross-linking process further gives us better insights into the biochemical principles of the cross-linking process. Had the present understanding of early singlet oxygen-mediated type II photochemical reaction followed by a more toxic hydrogen peroxide and riboflavin triplet-mediated type 1 photochemical held true, the endothelial damage and, hence, the corneal edema should have occurred in Group 2 (Controls) as opposed to Group 1 (Cases).[23,24] We therefore believe the photochemical reactions of the cross-linking process are not consecutive, where type I follows a type II photochemical response, but involves a simultaneous activation of both the photochemical pathways in the processing of the tissue permeated oxygen.

Further, we believe the confines of the use of 400 microns corneal thickness as the criteria for the use of iso-osmolar and hypo-osmolar riboflavin solution in corneal collagen cross-linking warrant additional considerations of the radiation fluence and baricity conditions used during the cross-linking protocols.[25,26] Higher fluence radiations (accelerated cross-linking protocols), with or without increased baricity, generating increased oxygen free-radical species, might necessitate a higher stromal thickness reserve, thereby obviating the effects of any unwarranted endothelial toxicity of the increased free radicals produced during the process and thereby effecting safer postoperative outcomes.

While it is established that CXL is oxygen-dependent, increasing the local oxygen concentration beyond a certain required threshold may not necessarily improve the outcomes of the procedure. We believe that there must be other undiscovered factors that play a role in affecting the outcomes of CXL and these need to be explored further.

Our study is limited by a small sample size, lack of measurement of corneal stromal oxygen concentration, and a short follow-up period. Nevertheless, our findings provide valuable insights into the function of supplemental oxygen in accelerated CXL for progressive keratoconus. Larger human studies with a longer follow-up period are warranted to further investigate the long-term outcomes of CXL with local oxygen-supplementation in real world.

Conclusion

Accelerated CXL is a useful modality to manage progressive keratoconus. However, in short term, local oxygen supplementation does not seem to have any additional benefit over room-air oxygen. Larger studies with longer follow-up period are required to further investigate the same.

Author contributions

RS - concept, design, definition of intellectual content and manuscript review; MS - literature search, data acquisition, data analysis, statistical analysis, manuscript preparation; MT - literature search, manuscript editing and manuscript review; SSP - data acquisition, data analysis; TA, NS, JST - manuscript editing and manuscript review.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.