Effects of corneal crosslinking on corneal shape stabilization after orthokeratology

Orthokeratology (Ortho-K) works to reshape cornea and is the only non-surgical way to enable vision without corrective aids. However, its effect is only temporary, and successful stabilization requires ongoing Ortho-K wear to maintain the reshaping effect. Corneal crosslinking (CXL) is a commonly-used technique in clinical practice to stabilize corneal shape in keratoconic eyes. However, whether or not CXL can stabilize corneal shape after Ortho-K in normal cornea has not been reported. Therefore, this proof-of-concept study using 2 rhesus monkeys aimed to determine the efficacy of the combined procedure. One monkey wore Ortho-K bilaterally for 24 hours, and the other from 6 pm to 8 am for 7 days. The left eyes of both monkeys underwent CXL after Ortho-K while the contralateral eye served as control. Results showed a gradual regression of corneal shape in all eyes with or without CXL. However, eyes underwent CXL regressed more slowly than the control eyes. The control eyes and the CXL treatment eye in the 7-day Ortho-K monkey regressed completely at last, while the CXL treatment eye in the 24 h Ortho-K monkey maintained a corneal flattening of −1.48 D 27 days after procedure. These findings suggest CXL can slow the regression of Ortho-K for a short duration, but cannot sustain its effect according to the current protocol.


Results
Slit lamp microscopy. Given the fact that the shapes of both eyes in our monkeys were not identical, optimal lenses were selected for each eye with only slight differences in fitting curve parameters. The Ortho-K lenses were well-fitted and well-tolerated by both monkeys. No epithelial fluorescein staining was observed after the removal of Ortho-K lenses in either eye of both monkeys. Corneal epithelial defects in the CXL treatment eyes were healed by the fifth day post CXL procedure. Mild to moderate corneal edema or stromal haze gradually improved after corneal epithelial healing and completely recovered 10 days following the procedure.
Corneal topography and refractive power change. The corneal tangential maps and corresponding subtractive maps of the 2 monkeys are presented in Figs. 1 and 2. Their corresponding value of change in mean keratometry within the reshape zone was shown in Fig. 3. In Monkey A that wore Ortho-K lens for 24 h, despite the small central reshape zone due to this short duration of lens wear, the right eye (control eye) achieved a mean keratometry change of −1.13 D within an irregular shape zone of 3.84 mm 2 (length × width, 3.53 × 1.86 mm), versus −2.16 D within a nearly-circular zone of 3.16 mm 2 (length × width, 2.00 × 2.04 mm) in the left eye (treatment eye). Corneal flattening was more obvious in Monkey B with overnight Ortho-K lens wear for 7 days, with a mean keratometry change of −3.48 D achieved in the right eye (control eye) within a round reshape zone of 9.27 mm 2 , versus −2.45 D within a round reshape zone of 6.87 mm 2 in the left eye (treatment eye).
In Monkey A, regressions of corneal shape were observed in both CXL treatment and control eyes. By day 8, the control eye regressed to its original shape (mean keratometry change of −0.07 D compared to baseline in the reshape zone), while the central corneal area in the CXL treatment eye remained flat (mean change of −1.01 D in the reshape zone). At day 27, a change of −1.48 D of corneal flattening still remained in the treatment eye relative to its baseline status.
In Monkey B, the control eye had completely regressed to baseline level (mean change of −0.17 D in the reshape zone) by day 6, whereas the effect of central flattening in CXL treatment eye still remained (mean change of −1.28 D in the reshape zone) till day 14 when it returned to baseline shape (mean change of 0.21 D in the reshape zone).
The mean keratometry of reshape zone and the difference between follow-up and baseline for two monkeys are presented in detail in Fig. 3. After Ortho-K wear an abrupt decrease in corneal curvature of reshape zone occurred, followed by a gradual regression in all eyes. Eight days post CXL procedure in Monkey A, the mean keratometry of the control eye returned to baseline level, while that of CXL treatment eye decreased. In Monkey   Figure 1. Tangential and subtractive maps of the control and crosslinking (CXL) eye from monkey A wearing Ortho-K lens for 24 h. Picture A-F 2 are profiles for the control eye while picture a-g 2 are profiles for the CXL eye. Picture (A),(a) are the baseline tangential maps of the control and treatment eye, respectively. Picture B 1 and b 1 are tangential maps after wearing Ortho-K lens for 24 h. Picture C 1 -F 1 show corneal shape at 5 days, 8 days, 13 days and 19 days after lens removal while c 1 -g 1 show corneal shape at 5 days, 8 days, 13 days, 19 days and 27 days after CXL. Picture B 2 -F 2 and b 2 -g 2 are corresponding tangential subtractive maps compared to their baseline (B 2 was generated from B 1 minus A and b 2 was generated from b 1 minus a, etc.), representing curvature change at each visit relative to original shape. Warm colors (red) indicate corneal steepening while cold colors (blue) indicate flattening in subtractive maps. Central corneal flattening was achieved in both eyes after wearing Ortho-K lenses (B 2 and b 2 ), followed by gradual regression (C 2 -F 2 and c 2 -g 2 ). The control eye regressed completely by 8 days (D 2 ), while the CXL eye maintained a −1.48 D of corneal flattening at 27 days post-operatively (g 2 ).

Discussion
Our study demonstrates that the current standard CXL technique may slow the regression of corneal shape molded by Ortho-K in normal corneas in rhesus monkeys, however it appears that CXL cannot stabilize the Ortho-K effect longer than one month.
The combined therapy of Ortho-K and CXL has already been applied in some previous studies, but not for myopia treatment. In clinical experiments for keratoconus patients wearing Ortho-K lenses, only temporary effects of corneal flattening were observed following CXL 10,11 . It is difficult to draw direct comparisons of Ortho-K efficacy in keratoconic and non-keratoconic eyes due to the differences in stromal structure. Our study was supported by the contralateral eye as a control, and it showed that the standard CXL procedure failed to stabilize the corneal shape molded by Ortho-K in normal cornea. While residual corneal flattening in the treated eye with 24 h of Ortho-K lens wear was observed, the associated refraction changes warrants further validation in more samples.
We hypothesize that the failure of CXL to stabilize the corneal reshape effect by Ortho-K may be attributable to several possible reasons. Firstly, although it is yet conclusive that the effect of Ortho-K is mainly attributable to the corneal epithelial re-distribution [12][13][14] , other studies suggested the changes of corneal posterior surface Tangential and subtractive maps of the control and crosslinking (CXL) eye from monkey B wearing Ortho-K lens for 7 day. Picture A-E 2 are profiles for the control eye while picture a-e 2 are profiles for the CXL eye. Picture (A),(a) are the baseline tangential maps of the control and treatment eye, respectively. Picture B 1 and b 1 are tangential maps after wearing Ortho-K lens for 7 days. Picture C 1 -E 1 show corneal shape at 6 days, 14 days and 25 days after lens removal while c 1 -e 1 show corneal shape at 6 days, 14 days and 25 days after CXL. Picture B 2 -E 2 and b 2 -e 2 are corresponding tangential subtractive maps compared to their baseline (B 2 was generated from B 1 minus A and b 2 was generated from b 1 minus a, etc.), representing curvature change at each visit relative to original shape. Warm colors (red) indicate corneal steepening while cold colors (blue) indicate flattening in subtractive maps. Central corneal flattening was achieved in both eyes after wearing Ortho-K lenses (B 2 and b 2 ), followed by gradual regression (C 2 -E 2 and c 2 -e 2 ). The control eye regressed completely by 6 days (C 2 ), while the CXL eye regressed completely by 14 days post-operatively (d 2 ).
Another possible explanation for this negative result is related to the intensity of crosslinking. In our experiment, we followed the most commonly used protocol of CXL for the treatment of keratoconus 19 . However, the effect of CXL is associated with total absorbed radiant exposure, concentration distribution, and treatment time 20,21 . Though the Dresden protocol is currently the most effective way to conduct UV-A mediated CXL, issues regarding efficacy, such as pulsing versus continuous wave light, accelerated versus conventional CXL, are still in debate. Besides, other novel CXL techniques including riboflavin concentration-controlled method which is more theoretically efficient than the Dresden protocol 22 , the rose bengal-green light CXL which was proved to have a sgreater density of CXL than UV-A CXL 23 , and the novel femtosecond laser CXL to correct vision 24 , have been investigated and developed. In light of this, the intensity of CXL in our study may fall short of adequacy, but with the potential to be improved to create long-term corneal changes. However, it must be noted that methods to strengthen CXL, such as using higher energy doses, may damage the underlying structures 25,26 , hence the appropriate trade-off between efficacy and safety of crosslinking when combined with Ortho-K requires further evaluation.
The topography showed clear profile of the changes in corneal shape from re-modelling to regression, but the refraction changes were not completely in line with the topographic measurements. Although measurement showed that a flattening of −1.13 D to −2.16 D was achieved in the reshape area in Monkey A, a lack of hyperopic shift was observed. One possible explanation is that the flattening area in monkey A was relatively small. Optometrists were easily misled by the surrounding areas where no curvature change occurred. Furthermore, measurement errors of retinoscopy should be noted for the discrepancy between topography and refraction measurement. Although the effect of CXL treatment faded over time, corneal flattening of −1.48 D remained in the CXL treated eye in Monkey A. One of the confounding factor is the potential effect on corneal flattening solely by CXL itself, since previous studies have reported mild corneal flattening and myopia reduction in keratoconus patients after CXL treatment using the current protocol 5,27,28 , though most of which was observed in steeper cornea 29 . Elling et al. 30 have performed CXL alone in myopic patients with normal corneas and observed the mean keratometry changed from 42.64 D at baseline to 42.11 D at 1 month post-op. However, the energy they used was 2-3 times higher than that in our study (10-15 J/cm 2 vs. 5.4 J/cm 2 ). Therefore, we do not think the effect of corneal flattening caused by CXL alone could be excluded, but it was of much less significance in the current protocol.
Interestingly, residual corneal flattening was only observed in Monkey A. It is conceivable that different Ortho-K modalities may influence the effect of CXL. Although high oxygen transmission materials are developed and applied in Ortho-K lenses, there is evidence that ocular surface is under additional hypoxic stress during overnight Ortho-K lens wear 31 . Therefore, we speculate that long-term wear of Ortho-K lenses may lead to more severe hypoxia and corneal swelling response, resulting in the diminished effect of CXL. Moreover, biomechanical changes with increasing duration of lens wear might also be attributable to the discrepancy in CXL effect 32 . Additional data is required to further elucidate the mechanistic hypotheses for this finding.
While the current study showed that CXL failed to sustain the corneal shape molded by Ortho-K lens, it was able to retard its regression. Therefore, reducing the frequency of lens wear while prolonging the effects of Ortho-K, especially for those with low compliance, is of substantial patient benefit. However, a more suitable protocol to achieve a better outcome and less complication is needed before its application in clinical practice.
The strengths of this study include the design of contralateral eye as a control, the use of customized Ortho-K lenses and a standardized protocol for CXL, and thorough quality control of the clinical measurements. A number of limitations must also be considered. Firstly, only two rhesus monkeys were included and assigned to different Ortho-K modalities, which may have limited the generalizability of our findings. However, given the similarities of structure, shape and size in monkey and human corneas, monkeys are the most relevant animal model available to investigate the combined treatment effects of Ortho-K and CXL. We observed consistent results in both monkeys with the application of various durations of Ortho-K wear, which would therefore provide strong evidence to support our conclusion. Another possible limitation was the short duration of follow-up that may have failed to capture the long-term effect of CXL, as some studies have observed the effects of CXL treatment on keratoconus eyes years after the procedure 33 . Although the latency and duration of CXL treatment effect remained unknown, our experiment confirmed that effect of CXL on corneal shape molded by Ortho-K faded within a month, hence treatment effect was unlikely even if follow-up duration was extended further.

Conclusions
In conclusion, standard CXL procedure failed to stabilize the corneal shape molded by short-term Ortho-K wear in rhesus monkeys. However, whether or not this combined therapy could achieve satisfactory outcome under other circumstances, including different Ortho-K modalities or CXL protocols, needs to be validated in further study. New research to develop a novel cross-linking technique for the stabilization of Ortho-K is ongoing. www.nature.com/scientificreports www.nature.com/scientificreports/ Methods Experimental design. This was a proof-of-concept study designed to establish the efficacy of CXL in stabilizing corneal shape following Ortho-K wear. Two male rhesus monkeys (Macaca mulatta) aged 2-year-old were used as subjects in these experiments. The monkeys were fitted with Ortho-K lenses bilaterally with different modalities to establish a transient or stable remolding status: Monkey A wore the Ortho-K lens continuously for 24 hours, while Monkey B wore the Ortho-K lens for 14 hours per day (6:00 PM to 8:00 AM) for 7 consecutive days to simulate habitual Ortho-K lens use in humans. This setting aims to test the efficay of CXL to stabilize the corneal shape at any time with any status. Ambient lighting was minimized and shade cloths were used on the cages in order to maximize adherence to the protocol and simulate the conditions most human users of Ortho K would experience.
Following contact lens removal, the left eye of each monkey was arbitrarily chosen to undergo CXL, while the fellow eye served as the control. Slit-lamp microscopy, corneal topography (Medmont E300 Corneal Topographer; Medmont Pty Ltd., Victoria, Australia) and retinoscopy refraction measurements (streak retinoscope; Lifecare Medical Equipments Co., Ltd., Zhejiang, China) were performed after the recovery of corneal epithelium in treated eyes and repeated at 1-week interval for a duration of 1 month. Differences between the treated and control eyes were observed.
The use of the animals in this study and protocols were approved by the Animal Ethics Committee of Sun Yat-sen University. The study was conducted in compliance with the Association for Research in Vision and Ophthalmology statement for the use of animals in ophthalmic and vision research.
Ortho-K lenses and lens fitting. Spherical reverse-geometry gas-permeable rigid contact lenses (E&E Optics Ltd, Hong Kong), with a refractive power of −6.00 diopters (D), an overall diameter of 10.6 mm, and an optic zone diameter of 6 mm, were used in this study. The radius of lens central curvature was chosen based on the corneal curvature map captured by a corneal topographer. Additional assessment of lens fitting was conducted with fluorescein staining under slit lamp biomicroscope by an Ortho-K expert (YX). Given that monkeys are more active than human and the lenses tend to drop out easily, mobility and stabilization of the lens were the primary consideration when selecting the lenses, hence those with optimal fluorescein patterns were selected as the final fitting lenses.
Corneal cross-linking. Standard CXL was conducted according to the Dresden Protocol 19 . The cornea underwent epithelial debridement with toothed forceps and sterile gauze, followed by 20 minutes' soaking with 0.1% riboflavin (Avedro Inc., Waltham, MA, USA). Following this, the cornea was rinsed with a balanced salt solution prior to the application of UV-A irradiation. The cornea was exposed to UV-A light of 3 mW/cm 2 intensity, emitted by a KXL I UV-A source (Avedro Inc., Waltham, MA, USA) for 30 minutes, receiving a total energy dose of 5.4 J/cm 2 . In addition, drops of riboflavin were added every 5 minutes during the exposure period to ensure sufficient supply to cornea. Antibiotic ointment (Tobradex ® , Alcon, Fort Worth, USA) was administered prophylactically to the cornea directly after CXL treatment to prevent infection.
Corneal topography. Changes in the corneal shape were monitored with the Medmont E300 Corneal Topographer (Medmont Pty Ltd., Victoria, Australia). Only pictures with a systematic comprehensive score exceeding 85 points were considered acceptable. Tangential maps and subtractive maps were used to display changes in corneal shape, while the mean keratometry of the reshape zone was provided as a quantitative measurement.
Of note, the tear films of the monkey break up easily, especially in the condition of anesthesia, thus affecting the quality of topography images. Secondly, the fixation of the monkey was not stable and hard to modulate, making the comparisons between measurements unreliable. To minimize the variability of topographic measurements mentioned above, we used saline solution and artificial tears to wet the cornea and keep the integrity of the tear films before measurements. We also tried to adjust the head position of the monkeys to capture the images only when the fixation was at/near the center. In addition, we analyzed the curvature changes in central reshape zone where the measurements were most accurate and precise.
The calculation of mean keratometry of the reshape zone was performed as follows (See Supplementary  Materials). Firstly, the reshape zone of Ortho-K was identified using the tangential subtractive maps before and after Ortho-K lens wear. In subtractive maps, the reshape zone is located in the central circular zone and surrounded by the ring of mid-peripheral corneal steepening. According to a previous study 34 , it can be effectively determined using a custom step size setting of 0.1 D in Medmont Studio 6 software (Medmont Pty Ltd., Victoria, Australia). The shape of the reshape zone was portrayed and copied in each tangential map with Adobe Photoshop software. Maps with painted reshape zones were then imported into Image-Pro Plus software (IPP; produced by Media Cybernetics Corporation, USA) and the reshape zone was set as the region of interest (ROI). Using the "counting tool" in IPP, the area ratio (a i ) of each pseudo color (i) within the ROI was estimated separately and their corresponding keratometric reading (p i ) was verified with the color scale. The mean keratometry of ROI (P) was calculated as: Refraction. Refraction was performed following the instillation of tropicamide drops (1%) in each eye, using a streak retinoscope. Pupillary dilation and the absence of a pupillary light reflex was considered as successful