Corneal epithelial remodeling induced by combined small incision lenticule extraction and accelerated corneal collagen crosslinking for myopia

Purpose To evaluate the changes of the corneal epithelial thickness (ET) profile induced by combined small incision lenticule extraction and accelerated corneal collagen crosslinking (SMILE-xtra) for myopia compared with the standard small incision lenticule extraction (SMILE). Setting Nuri Eye Hospital, 61, Dunsan-ro, Seo-gu, Daejeon, 35233, Korea. Design Retrospective cross-sectional study. Methods Thirty-one myopic eye undergoing SMILE-xtra and control group of 36 myopic eyes undergoing SMILE were retrospectively analyzed. Spectral-domain optical coherence tomography (CIRRUS™ HD-OCT 5000, ZEISS, Dublin, CA) was used to measure corneal ET of 17 zones within the central 7-mm zone at preoperative, postoperative 1 month, 3 months and 6 months. Postoperative ET alterations were analyzed for correlation with treatment parameters. Results There was no difference in preoperative mean age, postoperative MRSE, visual acuity, and ablation depth between the two groups, and there was a significant difference in preoperative central corneal thickness. Both groups showed the greatest increase in corneal ET in the paracentral area on the inferotemporal area, respectively, for 6 months. The preoperative MRSE and the ablation depth showed significant correlation with the postoperative epithelial thickening in mid-peripheral sectors in both groups, and significant negative correlations in paracentral sectors only in SMILE-xtra group. Conclusions It is significant as the first study to compare corneal epithelial remodeling between SMILE and SMILE with accelerated corneal collagen crosslinking. The SMILE-xtra with the relatively large corneal ablation did not show a significant difference in the pattern of corneal epithelial remodeling compared to the SMILE group.


Introduction
The corneal epithelium is an anterior layer of the cornea that contributes to maintain the integrity of the ocular surface and the refractive power of the eye.Corneal epithelium plays a dynamic role in establishing corneal regularity that changes its thickness against stromal irregularities after corneal refraction surgery, contributing to a smooth ocular surface and improving vision [1,2].
Corneal collagen cross-linking (CXL) utilizes photosensitizer (riboflavin) and ultraviolet-A light (UVA) to ultimately slow or prevent keratoconus progression by creating strong covalent bonds within the corneal stromal collagen strand, increasing stiffness.The simultaneous combination of prophylactic CXL and refractive cornea surgery improves the biomechanical stability of the post-operative cornea and potentially prevents iatrogenic ectasia [3].With the development of prophylactic CXL, the indications for SMILE (small incision lenticule extraction) could be expanded even in high myopia or patient with ectasia risk factor [3,4].

Patients
We retrospectively reviewed the medical records of patients who had undergone SMILE and SMILE-xtra at Nuri Eye Hospital from January 2019 to October 2021.Data was accessed after IRB approval in December 2022.The study protocol was approved by the Institutional Review Board (IRB number: KNUH-2022-11-013) at Kangwon national university hospital, Chuncheon-si, Korea, and the study was conducted in accordance with the tenets of the Declaration of Helsinki.This study was a retrospective study with a level below the minimum risk and consent waiver was granted by the Institutional Review Board.
After the ophthalmologist determined if they were suitable for surgery, they underwent SMILE surgery following informed consent.The inclusion criteria were (1) bilateral myopia or myopia with astigmatism, (2) age older than 18 years and less than 45 years, (3) stable refraction error for at least 1 year, that is, a change � of 0.50 diopters (D) in the spherical and cylindrical refraction (4) presence of myopia in manifest refraction spherical equivalent (MRSE) between − 1.00 D and − 8.00 D, (5) presence of astigmatism between 0.00 D and-4.00D, and (6) best preoperative corrected distance visual acuity of 0.8 (decimal value) or better in each eye.Soft contact lens wearers were instructed not to use them at least 15 days prior to surgery.The exclusion criteria were as follows: the use of hard contact lenses, a central corneal thickness of less than 480 μm, a calculated postoperative residual stromal bed of less than 250 μm, and the presence of other ocular pathologic conditions such as corneal dystrophy, keratoconus, corneal opacity, or a history of previous ocular surgery.
Based on pre-operative examination, risk factors for corneal ectasis were assessed using the scoring system proposed by Randleman et al. [15].Based on corneal topography, age, central corneal thickness, and expected postoperative remnant corneal stromal thickness, a score of 4 or more was classified as a high-risk group, a score of 3 was a moderate risk group, and a score of 2 or less was a low-risk group.Since the above classification was made based on LASIK surgery, the thickness of the corneal remnant matrix was calculated assuming that the cap thickness of SMILE was 120 μm.Eyes with moderate and high risk were selected for SMILE-xtra.

SMILE procedure
SMILE was performed using a VisuMax 500-kHz femtosecond laser (Carl Zeiss Meditec AG, Jena, Germany).The pulses of the laser were applied with a pulse energy of approximately 120 nJ.The spot distance of each laser spot was 4.0 μm.A 2 mm incision was made on the 145m eridian, and the upper and lower edges of the lenticule were delineated so that the tissue planes were clearly defined.The upper interface was separated, and the lower layer was dissected.When the two layers were separated, the lenticle was removed from the cornea.The diameter of the cap was 7.5mm and the optical zone diameter was 6.5mm.The intended cap thickness was 120 μm.CXL was performed as follows in SMILE-xtra.After lenticule extraction, Vibex Rapid ™ (Avedro, Inc., Waltham, MA, USA) containing 0.25% saline-diluted riboflavin mixed with a balanced salt solution was injected into the intrastromal pocket [16].The corneal stromal bed was soaked with the solution for 60 seconds, followed by cleaning the interface with a saline solution.The surface was irradiated with 45 mW/cm 2 ultraviolet light of 375 nm using the KXL System1 (Avedro, MA, USA) for 75 seconds with a total energy of 3.4 J/cm2 and a diameter area treatment of 9.00 mm.[16] (Table 1).Postoperative medications included topical Levofloxacin 0.5% (Cravit1 ophthalmic solution 0.5%; Santen Pharmaceutical Co., Ltd., Osaka, Japan) 4 times for 7 days, Loteprednol etabonate 0.5% (Lotepro1, Hanlim, Korea) in tapering dosages for 4 weeks, and artificial tear 4-6 times for 4 weeks or more.

Measurement of corneal epithelial thickness
The examination was performed by an one expert optometrist, and it included uncorrected and corrected distance visual acuity (UDVA and CDVA, decimal value), manifest refraction with and without cycloplegia by the fogging method of refraction.Astigmatism was assessed by the Jackson cross-cylinder technique.Corneal pachymetry, keratometry, and tomography patterns were measured with the Pentacam1 AXL (Oculus Optikgera ¨te GmbH, Wetzlar, Germany).The epithelial thickness (ET) and retinal optical coherence tomography were measured with spectral-domain Cirrus 5000 HD-OCT (Carl Zeiss Meditec, Germany).Corneal ET profiles of the central 7 mm zone were acquired preoperatively and at all follow-up measurements.Since the ET measurement in this study included the pre-cornel tear film, it was measured 1 minutes after instillation of a drop of artificial tear eyedrop (0.5% carboxymethylcellulose, Refresh, Allergan) before measurement [17].In addition, topical eye drops were not permitted for a period of 2 hours prior to scanning to minimize their effect on tear film thickness.All OCT examinations were scheduled between the hours of 0900 and 1300 and were conducted by an trained optometrist to reduce the influence of diurnal and operator-related changes in ET measurements [18].Two consecutive acquisitions were made for each patient to ensure the validity of the data, and the average value was used for the analysis.All OCT images were exported and processed using the Cirrus HD-OCT visualization software that provides an average automated ET of three ring-shaped concentric zones centered on the center of the cornea.The epithelial thickness maps were divided into a total of 17 sectors: a central 2 mm diameter zone, eight paracentral sectors within an annulus between the 2-5 mm diameter ring area, and eight mid-peripheral sectors within an annulus between the 5-7 mm diameter ring area (Fig 1).

Statistical analysis
The sample size was determined to have a power of at least 80% based on existing published results using MedCalc software (version 22.013, MedCalc Software Ltd., Ostende, Belgium) [8][9][10][11][12][13][14].Data were analyzed using SPSS, version 24.0 (IBM Corporation, Armonk, NY, USA), for Windows (Corporation, Redmond, WA, USA).The Shapiro-Wilk test was used to evaluate the normality of numerical data.The student's t-test was used if variables were normally distributed, whereas the Mann-Whitney U test was used if one or more variables were not normally distributed for analysis of the difference between the two groups.Continuous variables were compared using a two-way repeated measures ANOVA with the post hoc Bonferroni test, with follow-up time as a within-subjects factor (preoperative, 1 month, 3 months, and 6 months), and CXL treatment as a between-subjects factor (SMILE and SMILE-xtra).Spearman's coefficient was used to determine the association between the ET changes of different zones and treatment parameters.Statistical significance was set at p < 0.05.

Results
In this study, we enrolled a total of 67 eyes (36 eyes of SMILE, 31 eyes of SMILE-xtra) of 67 patients, and only one eye from each participant was randomly selected for the study.There was a significant difference of preoperative central corneal thickness, residual stromal thickness (RST) between two groups even though the mean age, preoperative manifest refraction spherical equivalent (MRSE), ablation depth, axial length and optical zone diameter (6.5 mm) were matched between the two groups (Table 2).There was no significant difference in the mean postoperative refractive error at 6 months (SMILE: -0.09 ± 0.11 D, SMILE-xtra: -0.14 ± 0.37 D, p = 0.812).The achieved spherical equivalent refraction versus attempted spherical equivalent refraction is presented in Fig 2 .(R 2 = 0.995 in SMILE group, R 2 = 0.995 in SMILE-xtra group).
Table 3 showed the changes in corneal ET in 17 sectors.In all sectors, it showed an increasing pattern up to 3 months, and showed a stabilizing pattern between 3 and 6 months.There was a significant difference only in sector 16 (p = 0.029) at the first month by each follow-up period, and no significant difference was shown in other sectors.There was no statistically significant difference of the corneal ET in each area according to the follow-up time between the two groups in all sectors (two way-RM ANOVA, p> 0.05 in all sectors).The central and paracentral annular zone showed significant epithelial increase at each follow-up period up to 3 months in the SMILE group.However, in the SMILE-xtra group, the overall tendency of the increasing ET was similar to that of the SMILE group, but there was no statistical significance in the central area in the first 1 month, and significant ET increase was seen between 1 and 3 months.There was no statistical significance in any zone between 3 and 6 months.
Table 4 shows the correlation between ET change and age, preoperative MRSE, ablation depth, RST.There was no correlation between age and the postoperative ET change for 6 months in all annular zones.The preoperative MRSE showed significant negative correlation with the postoperative epithelial thickening in central 2 mm and mid-peripheral sectors in both groups, and significant negative correlations in paracentral sectors only in SMILE-xtra group.The ablation depth showed significant positive correlation with the postoperative epithelial thickening in mid-peripheral sectors in both groups, and significant positive correlations in paracentral sectors only in SMILE-xtra group.RST showed a significant negative correlation with corneal epithelial thickening in all areas only in the SMILE-xtra group.

Discussion
Corneal laser refractive surgery alters the anterior corneal contour that induces corneal epithelial remodeling to compensate for the change in corneal stroma.This compensatory change has been also described in myopic [19] and hyperopic [20] laser in situ keratomileusis (LASIK), photorefractive keratectomy (PRK) [21] before SMILE.In particular, the corneal  epithelial hyperplasia after corneal laser refractive surgery is related to myopic regression, so it is important to know the pattern of epithelial changes in predicting clinical outcomes [22,23].
The corneal epithelial remodeling induced by SMILE has been reported in previous OCT based studies (Table 5).In this study, we used the Cirrus HD-OCT 5000 evaluate corneal epithelial remodeling of 7 mm circular zone over 6 months after SMILE and SMILE-xtra.This device has been demonstrated to produce excellent repeatability and reproducibility in normal [24] and keratoconus [25].This study is meaningful that it is the first study to evaluate the change of the corneal epithelium in SMILE combined with corneal collagen cross-linking using this device.
Prophylactic CXL with corneal refractive surgery has been shown to enhance biomechanical strength and stability and prevent the risk of postoperative corneal ectasis.Previous studies [26,27] have shown cytotoxic effects of riboflavin-UVA on corneal cells, including limbal epithelial stem cells and delayed corneal epithelial healing in patients following CXL [28].However, compensatory corneal epithelial changes were also found in the SMILE-xtra group, and it was found that the combination of accelerated CXL did not significantly affect corneal epithelial remodeling in this study.Rather, epithelial changes were more prominent in the SMILE-xtra group with a relatively large ablation depth.Unlike PRK or conventional CXL, corneal de-epithelization is not performed during SMILE-xtra surgery and the effect of riboflavin-UVA in the femtosecond laser created stromal pocket is relatively limited.Therefore, there is less possibility of affecting the corneal limbal epithelial stem cells, and it is thought that it will act more on the stromal side to induce effective collagen cross-linking [29].
In this study, the mean ET of the central 2 mm increased by 4.15 μm of SMILE group and 3.90 μm of SMILE-xtra group in six months postoperatively, which was relatively small compared to previous studies [9][10][11].The discrepancy may be due to the different measuring instruments [30] and the different preoperative SE and race [31] of the participants.In the paracentral zone, our results demonstrated that epithelial hyperplasia is more pronounced in the temporal zone than in the nasal zone.This asymmetric profile is similar to previous reports [9,11,13,14].Ivarsen and Hjortdal [32] reported an abrupt stromal change with increased corneal ET corresponding to the edge of the removed lenticule at the astigmatism axis.The ablation depth is significantly different between preoperative flat and steep axis in eyes of high astigmatism, with a higher ablation depth applied to flat axis [33].Hence, the discrepancy between the two meridional ablation depths may lead to different amounts of corneal epithelial hyperplasia, and more significant corneal epithelial hyperplasia may occur at the preoperative flat axis.The epithelial hyperplasia pattern response to corneal refractive surgery may be partially due to the mechanical influence of the upper eyelid tarsal during blinking.The Asian eyelid has a narrower palpebral opening, a larger eyelid volume and therefore a higher eyelid pressure.Previous research has suggested that there is a relationship between eye tension and changes in corneal topography, including induced astigmatism [34].The corneal epithelial hyperplasia is more likely in the inferior portion of the cornea, which is easily in contact with the tear film, and can cause the vertical asymmetry of ET [35].It is not clear why the corneal epithelial thickening at mid-peripheral area is greater than the central area changes after SMILE.For the above reasons, it is thought that a specific pattern occurs in the modeling of the corneal epithelium after SMILE and SMILE-xtra.
In this study, there was no significant correlation between age and corneal epithelial hyperplasia.This is a different result from previous study [9] and is considered to be a selection bias due to a relatively smaller age distribution (19 to 35 years) than previous study (25 to 45 years) and a different refraction error.The central and mid-peripheral 5-7 mm annular area showed a significant negative correlation with MRSE similar to previous studies [11,14].Ablation depth was relatively large in the SMILE-xtra group, but not statistically significant.Corneal epithelial hyperplasia showed a statistically significant correlation with ablation depth, especially in the paracentral areas (2-5mm and 5-7mm areas).The SMILE group with a relatively thick enough RST compared to the SMILE-xtra group showed no significant correlation with corneal epithelial hyperplasia, and the SMILE-xtra group with a relatively thin RST showed a significant negative correlation.It can be seen that the corneal epithelial hyperplasia is a compensatory growth in the ablated cornea, and more significant corneal epithelial hyperplasia is shown with a more corneal stromal ablation depth.
This study had some limitations.First, it had a retrospective design and a smaller sample size.Therefore, there might have been a selection bias in retrospectively comparing the SMILE-xtra group with from the SMILE group.Since corneal collagen cross-linking was additionally performed when the patient had a risk factor for corneal ectasia, there may be selection bias in the choice of surgical method.However, we tried to minimize the bias by comparing the two groups with no statistically significant difference in factors such as preoperative MRSE, corneal ablation, optic zone diameter, and cap thickness.In addition, the effect on epithelial thickness change was further analyzed by conducting Spearman's correlation analysis separately in each group.The second limitation is that the follow-up period of 6 months was not long enough to observe any refractive regression or to conclude that corneal epithelial remodeling had finished.Corneal epithelial hyperplasia after high myopic correction is known as a myopic regression factor [10].In this study, there was no difference in visual acuity correction at 6 months between the two groups, but further research is needed to see if myopic regression continues to be suppressed in the SMILE-xtra group, which had a relatively more stromal ablation.
In conclusion, the SMILE-xtra which performed in the relatively high-risk group of corneal ectasia based on preoperative evaluation and the relatively large corneal ablation did not show a significant difference in the pattern of corneal epithelial remodeling compared to the SMILE RTVue-XR OCT (Optovue, Inc., Fremont, CA) -9 mm area -The greatest epithelial thickness is observed in the paracentral zone (9.75%), followed by the central (8.79%,), mid-peripheral (8.12%), and peripheral zones(and 0.98%).
-For the midperipheral and peripheral zones, the average epithelial thickness of the flat meridian show the strongest thickening trends compared with the steep meridian.64 eyes 14   28.14 ± 6.38 6 months −5.76 ± 2.01 (−1.25 to−9.88) RTVue-XR OCT (Optovue, Inc., Fremont, CA) -9 mm area -The average epithelial thickness of the temporal section is significantly thicker than nasal section in paracentral (P < 0.001) and mid-peripheral zones (P = 0.049).
-The average epithelial thickness of superior sections is significantly thinner than that of inferior sections in paracentral, mid-peripheral, and peripheral zones (All: P < 0.001) https://doi.org/10.1371/journal.pone.0294121.t005 group during the 6-month follow-up period.Rather, there was no significant difference in corneal stroma ablation depth, but corresponding corneal epithelial hyperplasia could be observed in SMILE-xtra with a thin RST.This study is meaningful as the first study to look at changes in the corneal epithelium in the SMILE combined with CXL using SD-OCT.Long-term studies are needed to investigate the relationship between the degree of corneal epithelial hyperplasia and myopic regression in SMILE-xtra.In addition, a comparative study with other corneal laser refractive surgeries performed concurrently with CXL is considered necessary.

Fig 1 .
Fig 1. Example showing the measurements zones (sectors) with the spectral-domain Cirrus 5000 HD-OCT (Carl Zeiss Meditec, Germany).https://doi.org/10.1371/journal.pone.0294121.g001 Fig 3 shows the sequential changes of average ET in 17 sections before and after SMILE and SMILE-xtra.During the 3-month follow-up, there was an increase in ET in all 17 areas, especially in the inferotemporal sectors than in others.This tendency was more pronounced in the SMILE-xtra group.Between 3 and 6 months, the epithelial changes were insignificant and maintained in both groups.Both groups showed the greatest increase in corneal ET in the paracentral area on the inferotemporal area, and the central sector showed a change of 4.15 μm (SMILE) and 3.90 μm (SMILE-xtra), respectively, for 6 months (Fig 4).The longitudinal ET changes of annular zones are shown in Fig 5.

Table 5 . Previous studies of corneal epithelial remodeling after SMILE using OCT.
Mid-peripheral epithelial thickening is greater than the central changes following SMILE.-It is interesting to note that the increase in central epithelial thickness is smaller in the SMILE group despite the removal of a larger amount of stromal tissue compared with LASIK.-Epithelialthicknesschanges are less marked centrally but increased radially (centrifugally) toward the mid-periphery.increase in epithelial thickness was observed in the mid-periphery.-Alleyes show significant epithelial remodeling during the first 6 months.All remodeling was stable without significant changes after 6 months, and there was no return to preoperative epithelial thickness.