Secondary Logo

Journal Logo


Epithelial remodeling after corneal crosslinking using higher fluence and accelerated treatment time

Haberman, Ilyse D. MD; Lang, Paul Z. BA; Broncano, Alvaro Fidalgo MD, FEBO; Kim, Sang Woo MD; Hafezi, Farhad MD, PhD; Randleman, Bradley J. MD*

Author Information
Journal of Cataract & Refractive Surgery: March 2018 - Volume 44 - Issue 3 - p 306-312
doi: 10.1016/j.jcrs.2017.12.021
  • Free


Corneal crosslinking (CXL) has recently been approved by the U.S. Food and Drug Administration for the treatment of progressive keratoconus. Corneal crosslinking has been shown to increase biomechanical stability and improve or stabilize eyes with progressive keratoconus and postoperative ectasia based on reduction in anterior curvature.1–5 However, the mechanism by which this curvature change occurs is still not fully understood. Epithelial remodeling might be responsible in part for these changes after treatment.

The epithelium has the capacity for significant remodeling in an attempt to regularize irregular curvatures.6,7 Reinstein et al.8 reported significant differences in epithelial thickness profiles in normal and keratoconic corneas using very-high-frequency digital ultrasound (US). Our group reported significant differences in regional epithelial thickness profiles between normal, keratoconus, and post-laser in situ keratomileusis (LASIK) ectasia eyes using spectral-domain ocular coherence tomography (SD-OCT).9 We also reported early changes in regional epithelial thickness, with improved thickness regularity over a 3-month period after CXL using the standard crosslinking protocol in a pilot study using SD-OCT with manual epithelial thickness measurements.10 Reinstein et al.11 also found changes in epithelial thickness after CXL over time using very-high-frequency digital US. We are not aware of any longitudinal epithelial thickness studies after isolated CXL using an accelerated protocol.

The purpose of this study was to evaluate changes in regional epithelial thickness over 1 year in patients with keratoconus having a higher fluence accelerated CXL protocol using SD-OCT with automated epithelial mapping software, and to correlate these changes with focal changes in anterior corneal curvature.


Patients included in this analysis were enrolled and consented at Emory University as part of a prospective multicenter randomized placebo-controlled study of the KXL System with VibeX riboflavin ophthalmic solution (Avedro, Inc.) for CXL in eyes with keratoconus ( identifier: NCT01643226A) and approved by the Western and Emory University Institutional Review Boards. The study adhered to the tenets of the Declaration of Helsinki. The patients included in this subanalysis had treatment between March 2013 and May 2015.

Standard inclusion and exclusion criteria for CXL have been reported.4 Briefly, inclusion criteria were patients aged 12 years or older with topographic evidence of mild, moderate, or severe keratoconus, a maximum corneal curvature of 47.00 diopters (D) or higher, corrected distance visual acuity (CDVA) of 80 letters or fewer on the Early Treatment of Diabetic Retinopathy Study chart, and removal of contact lenses for at least 1 week before the initial examination and a follow-up examination to confirm stability. Exclusion criteria included sensitivity or known allergy to the use of the test article(s) or their components, patients who were pregnant or nursing, eyes classified as either normal, atypical normal, or keratoconus suspect on the severity grading scheme, eyes with a history of previous corneal or ocular surgery, history of herpes simplex or other possible corneal scarring disorders, corneal pachymetry less than 375 μm before epithelial debridement at the thinnest point, or eyes with a maximum corneal curvature outside the central 5.0 mm zone. Corneal thickness and corneal curvature measurements including maximum corneal curvature were determined using Scheimpflug imaging (Pentacam HR, Oculus Optikgeräte GmbH).

Corneal Crosslinking Protocol

The specific treatment protocol for this study was as follows: After 9.0 mm blunt epithelial debridement at the slitlamp, corneas were pretreated with 5 drops of riboflavin ophthalmic solution 0.12% (VibeX) every 2 minutes for 20 minutes. Using the KXL system, ultraviolet-A (UVA) irradiation was applied at 30 mW/cm2 for 4 minutes (fluence of 7.2 J/cm2). The protocol is herein referred to as accelerated CXL. After treatment, a bandage contact lens was applied and antibiotic drops were used for 1 week. No steroid or nonsteroidal drops were used after treatment.

Epithelial Thickness Measurements

Spectral-domain OCT (RT-Vue-100, Optovue, Inc.) running on report software version A6, 8, 0, 27, with a scan rate of 26 000 axial scans per second, axial resolution of 5 μm full-width-half-maximum, and transverse resolution of 15 μm, was used to assess regional corneal epithelial thickness preoperatively and 1, 3, 6, and 12 months after CXL, in the center 6.0 mm zone (centered on the pupil) using the automated machine software. The SD-OCT system reports sector values that are an average of multiple values within each sector, thereby increasing the reproducibility of measurements to approximately 1 μm12 and has been found highly correlated with very high frequency digital US measurements.13,14

All tests were performed by the same experienced ophthalmic technician, who also served as the clinical trial coordinator.

Data Collection and Analysis

The AS-OCT analysis report provided the pachymetry and epithelial statistics. In the central 5.0 mm zone, the report analyzed the mean epithelial thickness in the superior and inferior regions, the minimum and maximum thickness, the standard deviation (SD) across points, and the thickness range (minimum to maximum thickness). The epithelial map of the 6.0 mm central corneal was divided into 17 regions, including a 2.0 mm central section, 8 sections divided between the 2.0 mm and 5.0 mm zone, and 8 sections within the 5.0 mm and 6.0 mm ring (Figure 1).

Figure 1.
Figure 1.:
Representative total and epithelial thickness map (I = inferior; IN = inferonasal; IT = inferotemporal; N = nasal; S = superior; SN = superonasal; ST = superotemporal; T = temporal).

The Wilcoxon signed-rank test to compare matched groups was used to evaluate differences in regional epithelial thickness, epithelial SD, and epithelial minimum and maximum values in the same eyes across timepoints. Because of the pilot study nature of this evaluation, individual P values were not adjusted; thus, P values less than 0.05 were considered statistically significant.


Twenty-seven eyes from 20 patients who had accelerated CXL were evaluated with epithelial mapping preoperatively (n = 27) and at 1 month (n = 24), 3 months (n = 23), 6 months (n = 25), and 12 months (n = 27) postoperatively. Table 1 shows the patient demographics. Table 2 shows the mean epithelial thickness measurements at each timepoint; Table 3 shows the mean intraindividual epithelial thickness changes. The mean epithelial thickness varied across the central 6.0 mm by 15.2 μm (range 46.6 to 61.8 μm), with the inner inferotemporal sector exhibiting the thinnest mean epithelial thickness.

Table 1
Table 1:
Patient demographics.
Table 2
Table 2:
Mean regional epithelial thickness measurements after accelerated CXL.
Table 3
Table 3:
Mean intraindividual epithelial thickness changes after accelerated CXL.

One month after CXL, the mean epithelium was significantly thinner in the outer nasal region by a mean of 3.9 μm (P = .003), inner nasal by 3.0 μm (P = .04), outer inferonasal by 4.4 μm (P = .007), outer inferior by 3.6 μm (P = .01), outer inferotemporal by 1.5 μm (P = .04), and outer superotemporal by 3.3 μm (P = .01). The 5.0 mm minimum epithelial thickness was significantly thicker by a mean of 3.1 μm (P = .01). The mean epithelial thickness across all points ± SD were not significantly different from preoperative values.

Compared with pre-CXL values, at 3 months the mean epithelium was significantly thinner in the outer nasal region by 2.9 μm (P = .01) and outer inferonasal region by 3.6 μm (P = .002). The epithelium was significantly thicker in the center by 1.9 μm (P = .009), inner inferotemporal region by 1.5 μm (P = .04), and the inner temporal region by 3.2 μm (P = .009). The 5.0 μm minimum epithelial thickness was significantly thicker by a mean of 2.7 μm (P = .001), and the thickness range was significantly less (P = .03). The mean epithelial thickness across all points ± SD were not significantly different from preoperative values.

Compared with pre-CXL values, at 6 months the mean epithelium was significantly thinner in the inner superonasal region by 2.2 μm (P = .02), outer nasal by 3.4 μm (P = 0), inner nasal by 2.7 μm (P = .002), outer inferonasal 2.9 μm (P = .001), inner inferonasal by 1.0 μm (P = .02), outer inferior by 2.1 μm (P = .002), outer inferotemporal by 2.7 μm (P = .002), outer temporal by 1.3 μm (P = .04), and outer superotemporal region by 1.8 μm (P = .04). The 5.0 mm minimum epithelial thickness was significantly thicker by 1.7 μm (P = .03), the 5.0 mm maximum epithelial thickness was significantly thinner by 2.8 μm (P = .01), and the epithelial thickness range was significantly less (P = .002). The mean epithelial thickness across all points was 1.4 μm less (P = .007) and the epithelial SD was significantly less by 1.2 μm (P = .001) than the preoperative values.

Compared with pre-CXL values, at 12 months the mean epithelium was significantly thinner in the outer nasal region by 3.2 μm (P = .001), inner nasal by 3.2 μm (P = .007), outer inferonasal by 3.4 μm (P = .001), inner inferonasal by 1.4 μm (P = .01), outer inferior by 1.8 μm (P = .007), outer inferotemporal by 1.8 μm (P = .008), and outer temporal by 1.1 μm (P = .01). The epithelial SD was significantly less by 0.9 μm (P = .009). The mean epithelial thickness across all points was 1.5 μm less (P = .02) than preoperative values. There were no statistically significant changes in the thinner regions from the 6- to 12-month timepoint.

Figure 2 shows the mean relative differences before and 12 months after accelerated CXL. Although on average there were consistent changes, with thinning in the preoperatively thicker regions and reduced SD of epithelial thickness differences, there was significant variability in remodeling between eyes (Figure 3).

Figure 2.
Figure 2.:
Mean regional epithelial thickness in microns (A) before crosslinking and (B) 12 months after crosslinking. The regions that are statistically significantly thinner after crosslinking are shown in (C) (P < .05).
Figure 3.
Figure 3.:
Representative epithelial thickness maps before and after accelerated CXL. A: The typical response, with no change in thickness centrally or in the inferior inner or outer areas but with significant thinning nasally and inferonasally. Overall thickness deviation has decreased. B: A less typical response, with thickening centrally, inferiorly, and superiorly but with minimal changes nasally. Overall deviation has increased (I = inferior; IN = inferonasal; IT = inferotemporal; N = nasal; S = superior; SN = superonasal; ST = superotemporal; T = temporal).

Overall, the correlation between focal epithelial thickness change and focal anterior curvature change (expressed as percentage change for both for comparison) was poor (r = −0.035, range −0.78 to 0.65). Some eyes had a positive correlation, with thinner epithelium resulting in flatter curvature, whereas some eyes had opposite correlations. Figure 4 shows examples of eyes with good and poor correlations.

Figure 4.
Figure 4.:
Representative epithelial thickness difference map (left) and anterior curvature difference map (right) after accelerated CXL. A: Good correlation, with changes in epithelial thickness mirroring changes in corneal curvature. B: Poor correlation, with areas of maximal epithelial thinning corresponding to minimal changes in curvature, with the area of maximal flattening corresponding to epithelial steepening.


The results of this study indicate that significant corneal epithelial remodeling occurred after accelerated CXL using a higher fluence. Furthermore, this remodeling was relatively regionally predictable, with minimal change in the thinnest preoperative areas (central and mid-inferotemporal region) and thinning in the inferonasal region resulting in reduced thickness variability across the central 6.0 mm. However, there was minimal overall correlation between focal changes in epithelial thickness and corneal curvature.

There was epithelial thinning in most corneal regions at 1 month, followed by epithelial remodeling between 1 and 3 months, with focal thinning in the center, nasal outer, and inferonasal outer regions from baseline, and thickening in the temporal inner and inferotemporal outer regions. There were continued but less significant changes from 6 to 12 months after accelerated CXL. By the 6- and 12- month timepoints, there was significant epithelial thinning from baseline in the inferior, temporal, and nasal zones, whereas there was no significant change in the epithelial thickness in the superior cornea. This is consistent with published findings that keratoconic eyes have thicker epithelium over ectatic areas of stroma, as a hypertrophic response to abrupt changes in stromal thickness.8,12,15 It would follow that the areas of significant epithelial change after CXL would be those most affected by the treatment; as such, the superior epithelium might remain relatively less affected.

By 6 months after accelerated CXL, the epithelial thickness range and the SD across points were significantly less, indicating a more uniform thickness distribution in these eyes. This regularization of the corneal epithelium could contribute to some improvement in CDVA after crosslinking, as has been reported.4 When comparing the significantly thinner sectors, there was no significant difference between the 6-month and 12-month timepoints, indicating that the remodeling might plateau sometime between 3 and 6 months.

Although we cannot extrapolate our results to different protocols, the findings in this study agree with those published by Kanellopoulous and Asimellis,16 which used AS-OCT to evaluate the corneal epithelium in eyes having the Athens protocol. They also found that crosslinked eyes displayed thinner epithelium as well as decreased variability in epithelial thickness in a period of up to 1 year. That protocol is significantly different, consisting of excimer laser debridement of 50 μm of epithelium, partial topography-guided excimer ablation, followed by high-fluence UVA radiation (10 mW/cm3) for a total period of 10 minutes.16

The findings in our study also agree with those previously published by our group.10 on epithelial changes after standard CXL with 3 mW/cm3 UVA irradiation for 30 minutes. That study also found significant thinning in certain regions and decreased SD across the measured points, indicating a more uniform thickness profile after crosslinking with the standard protocol. Thickness measurements were also performed with AS-OCT, but were measured manually, somewhat limiting comparison across these studies.10 With the automated epithelial mapping software now available, authors should be better able to reproducibly assess regional changes in epithelial thickness. These changes need to be measured over a longer period to fully assess the role of epithelial remodeling in CXL. It will be important to study regional epithelial changes after crosslinking across different treatment protocols, as well as compare visual outcomes, to optimize treatment for these patients. These findings might be even more impactful for combination ablative and crosslinking treatments where differential remodeling could affect visual outcomes.

Although the overall values for epithelial remodeling were small, 5 μm represents at least 10% or more of the total epithelial thickness, especially in keratoconic eyes that usually have thinner epithelium before treatment, and changes in epithelial thickness directly affect the anterior curvature and the regularity of the surface of the cornea. Changes in the epithelium can effectively change the dioptric power of the cornea as well as decrease higher-order aberrations (HOAs) by making the surface more regular. It has been shown that the mean power change after removal of epithelium is at least 1.0 D and that there is also a change in the Q value by 0.2215; therefore, even small changes in the epithelium can contribute to refractive and asphericity values.

Flattening of maximum corneal curvature has been seen in most CXL cases and has been used as a metric of success when evaluating crosslinking efficacy. The specific mechanism resulting in this flattening remains unclear and some have postulated that changes in anterior curvature might be primarily caused by changes in epithelial thickness. In our study, we found no such correlation. Although some eyes did have concordant epithelial thinning in areas of flattening, others had the opposite response. The correlation seemed unpredictable. A larger study sample might shed further light on this; however, based on our findings, we do not believe change in anterior corneal curvature can be explained solely or primarily by epithelial healing after CXL.

This study was limited by a relatively small number of patients having the treatment protocol with epithelial thickness measurements obtained at all timepoints. This epithelial mapping software has just recently become commercially available, and with the ability to obtain automated measurements and better access to crosslinking, it will be important to look at the results from a larger population of patients over longer periods of time. In addition, the software only gave epithelial data measurements from the central 6.0 mm of the cornea. This limits the ability to evaluate overall remodeling of keratoconic eyes outside of the central 6.0 mm, especially in those patients with displaced cones. Outside the U.S., mapping is now available to 9.0 mm; this will further elucidate epithelial changes in the peripheral cornea. Additional correlation between epithelial thickness changes and change in front surface corneal topography in addition to HOAs might better demonstrate clinical relevance of these findings.

In conclusion, the corneal epithelium becomes thinner, especially inferonasally, and more regular across the central 6.0 mm over the first 6 months after accelerated CXL and stabilizes by 1 year. There was minimal correlation between these changes and focal changes in corneal curvature. The most significant remodeling takes place over the nasal, temporal, and inferior aspects of the cornea. Epithelial changes can be easily monitored over time with AS-OCT. As the epithelium contributes significantly to the total refractive power of the cornea, further studies are required to better understand the epithelial remodeling process and optimize treatment protocols.


  • There are significant differences in regional epithelial thickness profiles between normal, keratoconus, and post-LASIK ectasia eyes.
  • Epithelial thickness changes after CXL using the standard protocol and after combined CXL and excimer laser ablation protocols, although it has not been reported after accelerated CXL alone.


  • Significant epithelial remodeling and regularization occurred after accelerated CXL for keratoconus and were easily measured with anterior segment SD-OCT.
  • Changes in epithelial thickness patterns did not correlate well with changes in corneal curvature, with wide variability in correlations between patients.


1.Hallahan KM, Rocha K, Sinha Roy A, Randleman JB, Stulting RD, Dupps WJ Jr. (2014). Effects of corneal cross-linking on ocular response analyzer waveform-derived variables in keratoconus and post-refractive surgery ectasia. Eye Contact Lens, 40, 339-344, Available at:
2.Dupps WJ Jr, Roberts CJ. Corneal biomechanics: a decade later [editorial]. J Cataract Refract Surg. 40. 2014. 857. Available at:
3.Kanellopoulos AJ, Loukas YL, Asimellis G. Cross-linking biomechanical effect in human corneas by same energy, different UV-A fluence: an enzymatic digestion comparative evaluation. Cornea. 2016;35:557-561.
4.Hersh PS, Stulting RD, Muller D, Durrie DS, Rajpal RK., on behalf of the United States Crosslinking Study Group. United States multicenter clinical trial of corneal collagen crosslinking for keratoconus treatment. Ophthalmology. 2017;124:1259-1270.
5.Randleman JB, Khandelwal SS, Hafezi F. (2015). Corneal cross-linking. Surv Ophthalmol, 60, 509-523, Available at:
6.Gauthier CA, Holden BA, Epstein D, Tengroth B, Fagerholm P, Hamberg-Nyström H. Factors affecting epithelial hyperplasia after photorefractive keratectomy. J Cataract Refract Surg. 1997;23:1042-1050.
7.Reinstein DZ, Archer T. Combined Artemis very high-frequency digital ultrasound-assisted transepithelial phototherapeutic keratectomy and wavefront-guided treatment following multiple corneal refractive procedures. J Cataract Refract Surg. 2006;32:1870-1876.
8.Reinstein DZ, Gobbe M, Archer TJ, Silverman RH, Coleman DJ. (2010). Epithelial, stromal, and total corneal thickness in keratoconus: three-dimensional display with Artemis very-high frequency digital ultrasound. J Refract Surg, 26, 259-271, Available at:
9.Rocha KM, Perez-Straziota CE, Stulting RD, Randleman JB. SD-OCT analysis of regional epithelial thickness profiles in keratoconus, postoperative corneal ectasia, and normal eyes. J Refract Surg 2013; 29:173–179; erratum, 234. Available at: Erratum available at: Accessed January 24, 2018.
10.Rocha KM, Perez-Straziota CE, Stulting RD, Randleman JB. Epithelial and stromal remodeling after corneal collagen cross-linking evaluated by spectral-domain OCT. J Refract Surg. 2014;30:122-127. errata, 171.
11.Reinstein DZ, Gobbe M, Archer TJ, Couch D. Epithelial thickness profile as a method to evaluate the effectiveness of collagen cross-linking treatment after corneal ectasia. J Refract Surg. 2011;27:356-363.
12.Li Y, Tan O, Brass R, Weiss JL, Huang D. (2012). Corneal epithelial thickness mapping by Fourier-domain optical coherence tomography in normal and keratoconic eyes. Ophthalmology, 119, 2425-2433, Available at:
13.Urs R, Lloyd HO, Reinstein DZ, Silverman RH. Comparison of very-high-frequency ultrasound and spectral-domain optical coherence tomography corneal and epithelial thickness maps. J Cataract Refract Surg. 2016;42:95-101.
14.Reinstein DZ, Yap TE, Archer TJ, Gobbe M, Silverman RH. (2015). Comparison of corneal epithelial thickness measurement between Fourier-domain OCT and very high-frequency digital ultrasound. J Refract Surg, 31, 438-445, Available at:
15.Gatinel D, Racine L, Hoang-Xuan T. Contribution of the corneal epithelium to anterior corneal topography in patients having myopic photorefractive keratectomy. J Cataract Refract Surg. 2007;33:1860-1865.
16.Kanellopoulos AJ, Asimellis G. Epithelial remodeling after partial topography-guided normalization and high-fluence short-duration crosslinking (Athens protocol): Results up to 1 year. J Cataract Refract Surg. 2014;40:1597-1602.


A.U.S. National Institutes of Health Clinical Trials. Safety and Efficacy Study of Corneal Collagen Cross-Linking in Eyes With Keratoconus. NCT01643226. Available at: Accessed January 24, 2018.

Disclosures:Dr. Hafezi is coinventor of the PCT/CH2012/000090 application (ultraviolet irradiation device) and chief scientific officer and shareholder of EMAGine AG. None of the other authors has a financial or proprietary interest in any material or method mentioned.

© 2018 by Lippincott Williams & Wilkins, Inc.