Techniques for Corneal Collagen Crosslinking: Epi-off vs Epi-on

From EyeWiki
Assigned editor:
Assigned status Up to Date
 by Augustine Hong, MD on October 12, 2020.


Corneal Collagen Crosslinking (CXL) is a minimally invasive technique primarily used to stop ectasia (keratoconus or post-LASIK). The technique combines riboflavin and UV light to strengthen chemical bonds in the cornea, thereby stiffening a structurally weak and unstable cornea. The two main approaches to performing corneal collagen cross-linking are "Epi-off" and "Epi-on", with "epi" referring to the corneal epithelium.

The standard epi-off surgical approach involves removing 7-9 mm of epithelium with subsequent application of a riboflavin solution at regular interval (2-5 minutes) for 30 minutes, followed by 30 minutes of UVA irradiation with continued concurrent intermittent administration of the riboflavin solution. The patient is then given topical antibiotics and corticosteroids until re-epithelialization occurs, and corticosteroids for a period after re-epithelialization[1]

There is no standard epi-on surgical approach, but in general the technique leaves the corneal epithelium intact and achieves penetration of riboflavin in a variety of manners such as the addition of chemical agents meant to loosen the epithelial barrier, iontophoresis, or partial disruption of the epithelium[2]. This is followed by a similar time period of UVA irradiation with continued intermittent administration of the riboflavin solution during irradiation.

Purported advantages of Epi-off

  • Better riboflavin and UV absorption
  • Deeper, more effective therapy
  • More corneal stiffening, flattening and lower risk of post-treatment ectasia progression

Purported advantages of Epi-on

  • Faster return to baseline vision and return into contact lenses
  • Shorter duration of discomfort
  • Lower risk for corneal infections and haze

Surgical Technique, Epi-off

Inclusion/exclusion criteria

CXL is indicated for disorders of progressive ectasia. These include keratoconus, iatrogenic post-laser surgery ectasia, and more recently pellucid marginal degeneration. Inclusion criteria varies amongst studies but in general it consists of a measurable form of evidence of progressive ectasia documented over a specified time course. There are many different measurements used to document corneal ectasia, but generally changes in corneal topography such as increases in Kmax or increases in astigmatism and refractive error are used. Exclusion criteria includes conditions that could predispose the eyes to further complications such as current infection, history of ocular herpes zoster or herpes simplex, autoimmune disorders, history of poor epithelial healing, and severe or central corneal scarring. Traditionally a corneal thickness of less than 400 microns after epithelial removal was considered a contraindication to CXL, but reports using hypo-osmolar riboflavin solutions to induce corneal swelling during treatment suggest that an overly thin cornea may no longer be a barrier to treatment, as long as the cornea can be brought to an thickness prior to UV exposure[3].

Epithelial removal techniques

Several techniques exist to remove the corneal epithelium in epi-off CXL. Mechanically removing the epithelium with a scalpel or rotating brush has classically been used in the Dresden protocol. Other investigators have pre-treated the epithelium with ethyl alcohol followed by mechanically rubbing the epithelium with a cellulose sponge. Phototherapeutic Keratectomy (PTK) laser scrape has also been reported as a method for removing epithelium for the purposes of CXL. For PTK/laser scrape, an excismer laser is used at a constant depth of 50 microns in a central corneal area of 6.5 mm, after which the debridement area is extended to 9 mm mechanically with a scalpel or rotating brush[4].

Riboflavin solutions used

The initial Riboflavin solution used was 0.1% Riboflavin in 20% dextran solution. This solution is still commonly used but many clinical studies now use an isotonic solution suspended in Hydroxypropyl methylcellulose (HPMC) instead of dextran. For corneas that were previously thought to be too thin for CXL treatment hypo-osmolar 0.1% riboflavin solution can be utilized to induce corneal swelling to at least 400 microns before UVA irradiation[3].

30-minute vs accelerated protocols

The most extensively studied approach is the Dresden protocol which entails application of Riboflavin every 3 minutes for 30 minutes, followed by 30 minutes of UVA irradiation. Some researchers have attempted an accelerated protocol wherein UVA illumination intensity is increased while UVA irradiation time is decreased. The increase in illumination intensity, and decrease in irradiation time varies in different accelerated protocols but in general, a constant radiant exposure of 5.4J/cm^2 is maintained. Current data comparing standard vs accelerated CXL is limited. Some studies demonstrate equivalent efficacy and safety[5][6] [7] in clinical parameters, while others demonstrate that the accelerated protocol produces less topographical flattening and stiffening than the Dresden protocol. [7][8] [9]

Postoperative course, drops, expected recovery, restrictions

Following the procedure, patients are given daily topical antibiotics and corticosteroids for 1-3 weeks with frequent follow up. A bandage contact lens is usually placed, though timing of removal of bandage lens differs. For patients with corneas too steep for a contact lens, delayed healing, or those who failed to re-epithelialize as expected, corneal self-retained amniotic membrane may be placed. Patients can expect post-operative pain within the first 3 days[10] as well as blurry vision for 1-2 weeks following the procedure. Patients should not use contact lens in the post-operative period until steroid use is discontinued, which is typically after 3 weeks[11].

Reported Outcome data, Epi-off

In 2016, Meiri Z, Keren S, Rosenblatt A, et al.[12] published a systematic review and analysis of corneal collagen crosslinking specifically for keratoconus. They performed a search and included all “population-based prospective and retrospective studies on the outcome of CXL in patients with KCN, from peer-reviewed journals in all languages from January 2003 to April 2014." The analysis covered patients of all ages with progressive KCN, clear corneas without scars or opacity, pre-CXL data, and a minimum of 1 month of post-CXL data (averages, SD, and the number of eyes per outcome tested). Excluded were cases with corneal infection, dry eye syndrome, pregnancy, lactation, systemic collagen disease, and a history of ocular surgery in the study eye. Only cases performed according to the standard Dresden protocol or modifications were included.

Results for Standard epi-off protocol (52 studies) are discussed below.

Primary outcomes

  • Mean difference between pre-CXL and post-CXL UDVA of -0.1 to -0.15 logMAR at 12 months (P=0008). Longer term follow up studies were fewer but show continued improvement in UDVA.
  • Topography measurements: Reduction in Kmax of 1 D at 12 and 24 months follow up, not statistically significant.

Secondary Outcomes

  • Topography: Kavg, Kcyl, and steepest K reduced after 12 months but not earlier or later, not statistically significant.
  • Refractive: Spherical equivalent was more positive 12 and 24 months after CXL. This change was not statistically significant
  • Corneal: Central Corneal Thickness (CCT) measurements varied depending on method used and were not statistically significant.
  • Intraocular Pressure: The IOPg was studied in 2 groups with no reported changes. The IOPcc [Corneal compensated intraocular pressure] was studied in 3 groups and was increased 1 month after treatment, but not on later measurements. The IOP-GAT [Goldmann applanation tonometry] was studied in 5 groups and was increased 12 months after.


Evangelista C, and Hatch K[13] did a review of complications following epi-off CXL and reported the following:

  • Pain: 42.7% of patients reported intense pain on the day of the surgery with that number decreasing significantly with each day post-op, reaching zero by post op day 3 [10]
  • Keratitis: Both infectious and sterile keratitis have been reported with organisms including gram positive and gram negative bacteria, as well as fungi, acanthamoeba and herpes simplex virus
  • Persistent Epithelial Defect/Corneal Melt: 9 out of 21 cases in a study had delayed epithelial healing and did not heal until post op day 9[14]. Patients with prolonged time course of epithelial defect may be at risk for corneal melt which has been documented in case reports[15][16].
  • Corneal Opacity: Transient corneal haze as well as a more permanent stromal scarring have been reported.
  • Corneal Edema: A retrospective study reported a 2.9% rate of corneal edema on day one post-op[17].
  • Endothelial Damage: Endothelial damage is a risk for patients undergoing UVA irradiation with corneas thinner than 400 microns. For those patients, risk is mitigated by inducing corneal swelling to at least 400 microns before treatment.

Surgical Technique, Epi-on

Inclusion/exclusion criteria

Similar to epi-off CXL, epi-on CXL is indicated in the treatment of disorders of progressive ectasia. These include keratoconus, iatrogenic post-laser surgery ectasia, and more recently pellucid marginal degeneration. Inclusion criteria varies amongst studies but similar to epi-off CXL, it consists of a measurable form of evidence of progressive ectasia documented over a specified time course. There are many different measurements used to document corneal ectasia, but generally changes in corneal topography such as increases in Kmax or increases in astigmatism refractive error are used. Exclusion criteria includes conditions that could predispose the eyes to further complications such as current infection, history of ocular herpes zoster or herpes simplex, autoimmune disorders, history of poor epithelial healing, and severe corneal scarring. A difference in exclusion criteria between epi-off and and epi-on CXL is that due to the fact that epi-on patients have their epithelium in tact during the irradiation, the corneal thickness requirement of 400 microns includes the epithelium thickness in that measurement. Patients who may have been excluded due to a corneal thickness thinner than 400 microns only after epithelium removal, can still be considered for epi-on CXL.

Epithelial disruption or absorption techniques

There are several different ways researchers have attempted to maintain the epithelium intact while still allowing Riboflavin to be absorbed into the stroma. These include various chemical additives to the riboflavin solution (Trometamol, Sodium ethylenediaminetetraacetic acid, benzalkonium chloride, Sodium Chloride, Vitamin E), application of topical anesthetics, increased concentration of riboflavin, increased time of application of the solution, iontophoresis, and partial epithelial disruption[11][18]. Iontophoresis employs the use of an electrical gradient to drive charged molecules across epithelium. Riboflavin’s negative charge and hydrophilic nature make it an ideal candidate to use iontophoresis to move it across the lipophilic corneal epithelium[19].

Riboflavin solutions

Riboflavin solution used is a variable that investigators have tried to manipulate to increase riboflavin absorption through an intact epithelium. As such, the various riboflavin solutions that have been used is wide ranging and includes many additives in an attempt to increase transepithelial absorption. Examples of the many different solutions used includes but is not limited to: hypotonic riboflavin solution 0.1% without dextran[20], hypotonic riboflavin solution 0.5% without dextran[21], riboflavin 0.1%, dextran 15.0%, trometamol, and EDTA[22], riboflavin 0.15%, dextran and vitamin E-TPGS 500 mg/100 mL[18], 0.25% riboflavin with BAC, EDTA, Tris, and 0.45% phosphate buffer saline[23]. There is currently no in vivo comparing different riboflavin solutions and additives in epi-on CXL to each other and there is no current preferred method.

30-minute vs accelerated protocols

Data for accelerated epi-on CXL is limited but shows promising results. Several studies have shown it to be effective in halting progression of ectasia but they have all differed in protocol in regards to time and irradiation and only one has compared it directly to another form of CXL. For example one study used a UVA irradiance of 9 mW/cm^2 for 10 minutes[24] while another study used a UVA irradiance of 45 mW/cm^2 for 2 minutes and 40 seconds[25]. The only study comparing accelerated to conventional used an irradiance of 6 mW/cm^2 for 15 minutes in the accelerated group and showed accelerated epi-on CXL to be similarly effective to non-accelerated epi-off CXL[23].

Postoperative course, drops, expected recovery, restrictions, bandage contact lenses

Some of the purported advantages of epi-on CXL is an improved post-operative course with less pain and quicker recovery. Similar to epi-off CXL patients are generally given daily topical antibiotics and corticosteriods for 1-2 weeks following the procedure with frequent follow up. A bandage contact lens may or may not be placed following the procedure with many comparison studies using a contact bandage lens in the epi-off group and none in the epi-on group[26] [27]. In a study comparing epi-on to epi-off CXL, the epi-off group was prescribed toradol and diclofenac for two days while the epi-on group was not, suggesting increased need for pain control in the epi-off group[26].

Reported Outcome data, Epi-on

Meiri Z, Keren S, Rosenblatt A, et al.[12] included the results of epi-on studies in their systematic review and analysis of corneal collagen crosslinking specifically for keratoconus with the following results:

Primary Outcomes

  • UDVA was improved by 0.22 to 0.28 logMAR 3 to 12 months after CXL. UDVA changes were statistically significant while CDVA changes were not.
  • Reduction in Kmax was not statistically significant.

Secondary Outcomes

  • Reductions in Kmin and SteepestK were not statistically significant. Kavg was unchanged at 3 to 12 months follow up.
  • There were no changes noted for the sphere, cylinder, and SE up to 12 months after CXL.
  • Changes in endothelial cell density were not statistically significant.


The purported advantages of epi-on CXL are centered on fewer and less severe complications than epi-off CXL. The fewer number of reported complications may be due to overall fewer performed epi-on CXL procedures. Some complications have still been reported including:

  • Transient hyperemia and foreign body sensation with resolution at 24 hours post-op
  • Transient post-operative photophobia[28]
  • Acute actinic keratitis with diffuse punctate epitheliopathy reported in an accelerated Epi-on CXL trial[29].



Comparative studies between epi-off and epi-on CXL have consistently shown epi-on CXL to be associated with less patient discomfort and pain. On a 1-5 pain scale, those treated with epi-on CXL reported an average of 2, while those treated with epi-off CXL reported an average of 4 (P=.0035) in a comparative study involving 70 patients[27]. While comparing epi-off CXl to epi-on CXL using iontophoresis, F. Cifariello et al[30] used a Ocular Surface Disease questionnaire to assess patients’ discomfort level and found those undergoing epi-on CXL to be significantly lower.


A systematic review and meta analysis including 8 comparative studies (5 randomized controlled trials and 3 non-randomized comparative trials) found no statistical significance in regards to UDVA and CDVA in epi-on vs epi-off CXL at one year follow-up[31].

Conversely in a longer term study of pediatric patients, those undergoing epi-on iontophoresis CXL showed no statistical improvement in corrected distance visual acuity from baseline at 3 years follow-up, while those in the epi-off CXL did show a statistically significant improvement in corrected distance visual acuity[32].

K max, refractive changes (sphere, cylinder)

The systematic review and meta analysis including eight comparative studies mentioned earlier showed that reduction in mean K was statistically greater with epi-off CXL when compared to the epi-on CXL with a standardized mean difference of .28 in K (95% CI, .03-.53; P=.03). When the epi-on CXL groups were separated into subgroups of those using chemical additives and those using iontophoresis the differences were more apparent. Epi-on CXL using chemical additives to enhance Riboflavin absorption was comparable in regards to reductions of mean K with epi-off CXL, but a statistically significantly greater reduction in mean K was seen in epi-off CXL when compared to epi-on CXL using iontophoresis. This difference was reported as a standardized mean reduction of K of 0.43; 95% CI, 0.10–0.75; P= 0.01[31].

Haze, scars, infections

Most comparative literature is focused on visual outcomes between the two procedures with little data specifically comparing adverse events. A review including 51 articles (45 epi-off, 6 epi-on) reported that haze, scarring and microbial infections were only reported in the epi-off CXL studies, with none reported in the epi-on CXL studies[2]. A sterile infiltrate rate as high as 7.6% has been reported in a prospective epi-off CXL study[33] while no such data exists for epi-on CXL.

Limitations of data/areas for continued study

There are currently a limited number of randomized controlled trials comparing epi-off CXL to epi-on CXL. Due to the wide variety of protocols and approaches to epi-on CXL, drawing conclusions about non-inferiority is difficult because there is currently not a general agreed upon epi-on CXL approach. In addition, there are very few articles that have follow up longer than one year after CXL treatment.

Additional Resources

  • O’Brart, D. P. S. (2017). Corneal Collagen Crosslinking for Corneal Ectasias: A Review. European Journal of Ophthalmology, 27(3), 253–269.


  1. Wollensak G, Spoerl E, Seiler T. Riboflavin/ultraviolet-a-induced collagen crosslinking for the treatment of keratoconus. Am J Ophthalmol. 2003;135:620.
  2. 2.0 2.1 Shalchi Z, Wang X, Nanavaty MA. Safety and efficacy of epithelium removal and transepithelial corneal collagen crosslinking for keratoconus. Eye (Lond). 2015;29(1):15–29. doi:10.1038/eye.2014.230.
  3. 3.0 3.1 Hafezi F, Mrochen M, Iseli HP, et al. Collagen crosslinking with ultraviolet-A and hypoosmolar riboflavin solution in thin corneas. J. Cataract Refract. Surg. 35 (2009), pp. 621-624.
  4. Kapasi M, Baath J, Mintsioulis G, et al. Phototherapeutic keratectomy versus mechanical epithelial removal followed by corneal collagen crosslinking for keratoconus. Can J Ophthalmol. 2012;47:344-347.
  5. Tomita, M., Mita, M. & Huseynova, T. Accelerated versus conventional corneal collagen crosslinking. J Cataract Refract Surg 2014:40, 1013–1020, 10.1016/j.jcrs.2013.12.012.
  6. Hashemian H, Jabbarvand M, Khodaparast M, et al. Evaluation of corneal changes after conventional versus accelerated corneal cross-linking: a randomized controlled trial. J Refract Surg 30, 837–842.
  7. 7.0 7.1 Chow VW, Chan TC, Yu M, et al. One-year outcomes of conventional and accelerated collagen crosslinking in progressive keratoconus. Sci Rep. 2015;5:14425.
  8. Choi M, Kim J, Kweon Kim E et al.Comparison of the Conventional Dresden Protocol and Accelerated Protocol With Higher Ultraviolet Intensity in Corneal Collagen Cross-Linking for Keratoconus. Cornea. 2017 (36): 523-529.
  9. Wernli, J,Schumacher, S, Spoerl, E, et al. The efficacy of corneal cross-linking shows a sudden decrease with very high intensity UV light and short treatment time. Invest Ophthalmol Vis Sci. 2013:54, 1176–1180.
  10. 10.0 10.1 Ghanem VC, Ghanem RC, De Oliveira R. Postoperative pain after corneal collagen cross-linking. Cornea. 2013:32(1):20–24.
  11. 11.0 11.1 O’Brart, D. P. S. Corneal collagen crosslinking for corneal ectasias: a review. European Journal of Ophthalmology. 2017:27(3), 253–269.
  12. 12.0 12.1 Meiri Z, Keren S, Rosenblatt A, et al. Efficacy of corneal collagen cross-linking for the treatment of keratoconus: a systematic review and meta-analysis. Cornea. 2016:35(3)417–428.
  13. Evangelista C, Hatch K. Corneal Collagen Cross-Linking Complications. Seminars in Ophthalmology. 2018:33:1, 29-35
  14. Kanellopoulos, AJ. Long term results of a prospective randomized bilateral eye comparison trial of higher fluence, shorter duration ultraviolet A radiation, and riboflavin collagen cross linking for progressive keratoconus. Clin Ophthalmol. 2012;6:97–101.
  15. Mohamed-Noriega K, Butrón-Valdez K, Vazquez-Galvan J, et al. Corneal melting after collagen cross-linking for keratoconus in a thin cornea of a diabetic patient treated with topical nepafenac: A case report with a literature review. Case Rep Ophthalmol. 2016;7:119–124.
  16. Labiris G, Kaloghianni E, Koukoula S, et al. Corneal melting after collagen cross-linking for keratoconus: A case report. J Med Case Rep. 2011;5:152.
  17. Sharma A, Nottage JM, Mirchia K, et al. Persistent corneal edema after collagen cross-linking for keratoconus. Am J Ophthalmol. 2012;154(6):922–926
  18. 18.0 18.1 Caruso C, Ostacolo C, Epstein RL, et al. Transepithelial corneal cross- linking with vitamin E-enhanced riboflavin solution and abbreviated low-dose UV-A:24-month clinical outcomes. Cornea. 2016;35(2):145–150.
  19. Gore DM, O'Brart DP, French P, et al. A comparison of different corneal iontophoresis protocols for promoting transepithelial riboflavin penetration. Invest. Ophthalmol. Vis. Sci. 2015;56(13):7908-7914.
  20. Raiskup F, Pinelli R, Spoerl E. Riboflavin osmolar modification for transepithelial corneal cross-linking. Curr Eye Res. 2012:37, 234-238.
  21. Chen S, Chan T, Zhang J. Epithelium-on corneal collagen crosslinking for management of advanced keratoconus. Journal of Cataract & Refractive Surgery. 2016: 42, 5, 738-749.
  22. Caporossi A, Mazzotta C, Paradiso AL, et al. Corneal collagen crosslinking for progressive keratoconus: 24-month clinical results. J Cataract Refract Surg. 2013:39, pp. 1157-1163
  23. 23.0 23.1 Madeira C, Vasques A, Beato J, et al. Transepithelial accelerated versus conventional corneal collagen crosslinking in patients with keratoconus: a comparative study. Clin Ophthalmol. 2019;13:445–452. Published 2019 Mar 1. doi:10.2147
  24. Akbar B, Intisar-Ul-Haq R, Ishaq M, Arzoo S, Siddique K. Transepithelial corneal crosslinking in treatment of progressive keratoconus: 12 months’ clinical results. Pak J Med Sci. 2017;33(3):570–575.
  25. Kır MB, Türkyılmaz K, Öner V. Transepithelial high-intensity cross-linking for the treatment of progressive keratoconus: 2-year outcomes. Curr Eye Res. 2017;42(1):28–31.
  26. 26.0 26.1 Soeters N, Wisse RP, Godefrooij DA, Imhof SM, Tahzib NG. Transepithelial versus epithelium-off corneal cross-linking for the treatment of progressive keratoconus: A randomized controlled trial. Am J Ophthalmol. 2015;159(5):821–828.
  27. 27.0 27.1 Al Fayez MF, Alfayez S, Alfayez Y. Transepithelial versus epithelium-off corneal collagen cross-linking for progressive keratoconus: A prospective randomized controlled trial. Cornea 2015: 34, 10:S53-6.
  28. Filippello M, Stagni E, O’Brart D. Transepithelial corneal collagen crosslinking: Bilateral study. J Cataract Refract Surg. 2012;38(2):283–291.
  29. Mazzotta C, Hafezi F, Kymionis G, et al. In vivo confocal microscopy after corneal collagen crosslinking. Ocul Surf. 2015;13:298–314.
  30. Cifariello F, Minicucci M, Di Renzo F, et al. Epi-off versus epi-on corneal collagen cross-linking in keratoconus patients: A comparative study through 2-year follow-up J Ophthalmol. 2018; 2018: 4947983.
  31. 31.0 31.1 Wen D, Song B; Li Q et al. Comparison of Epithelium-Off Versus Transepithelial Corneal Collagen Cross-Linking for Keratoconus: A Systematic Review and Meta-Analysis. Cornea 2018. 37:8 1018-1024.
  32. Buzzonetti L, Petrocelli G, Valente P. Iontophoretic transepithelial collagen cross-linking versus epithelium-off collagen cross-linking in pediatric patients 3-year follow-up. Cornea. 2019:38 (7) 859-863.
  33. Koller T, Mrochen M, Seiler T. Complication and failure rates after corneal crosslinking. J Cataract Refract Surg. 2009;35:1358–1362.