Corneal Collagen Cross-Linking

From EyeWiki
Original article contributed by: Brad H. Feldman, M.D.
All contributors: Brad H. Feldman, M.D. and Erica Bernfeld M.D.
Assigned editor: Erica Bernfeld
Review: Assigned status Up to Date by Erica Bernfeld M.D. on January 16, 2015.

Surgical Therapy

Corneal collagen cross-linking is a technique which uses UV light and a photosensitizer to strengthen chemical bonds in the cornea. The goal of the treatment is to halt progressive and irregular changes in corneal shape known as ectasia. These ectatic changes are typically marked by corneal thinning and an increase in the anterior and/or posterior curvatures of the cornea, and often lead to high levels of myopia and astigmatism. The most common form of ectasia is keratoconus and less often ectasia is seen after laser vision correction such as LASIK.


Cross-linking of collagen refers to the ability of collagen fibrils to form strong chemical bonds with adjacent fibrils. In the cornea, collagen cross-linking occurs naturally with aging due to an oxidative deamination reaction that takes place within the end chains of the collagen. It has been hypothesized that this natural cross-linkage of collagen explains why keratoectasia (corneal ectasia) often progresses most rapidly in adolescence or early adulthood but tends to stabilize in patients after middle-age.

In addition to the cross-linking that occurs commonly with corneal maturation, there are several other pathways that can lead to crosslinkage. Glycation refers to a reaction seen predominantly in diabetics that can lead to the formation of additional bonds between collagen. In the pathway most relevant to this topic, oxidation has been shown to be able to trigger corneal crosslinkage through the release of oxygen free radicals. 

The bases for the currently employed corneal collagen cross-linking techniques were developed in Europe by researchers at the University of Dresden in the late 1990's.  UV light was used to induce collagen cross-linking in riboflavin soaked porcine and rabbit corneas via the oxidation pathway. The resultant corneas were shown to be stiffer and more resistant to enzymatic digestion. Investigation also proved that the treated corneas contained higher molecular weight polymers of collagen due to fibril crosslinking. Safety studies showed that the endothelium was not damaged by the treatment if proper UV irradiance was maintained and if the corneal thickness exceeded 400 microns.[1]

Human studies of UV-induced corneal cross-linking began in 2003 in Dresden, and early results were promising. The initial pilot study enrolled 16 patients with rapidly progressing keratoconus and all of the patients stopped progressing after treatment. Additionally, 70% had flattening of their steep anterior corneal curvatures (decreases in average and maximum keratometric values), and 65% had an improvement in visual acuity. There were no reported complications. [2]

Corneal collagen cross-linking using riboflavin and UV received FDA approval on April 18, 2016.

In late 2011, orphan drug status was awarded by the FDA to Avedro for its formulation of riboflavin ophthalmic solution to be used in conjunction with the company's particular UVA irradiation system. This now entitles the company to up to 7 years of exclusive marketing of corneal crosslinking in the United States for their riboflavin ophthalmic solution/KXL® system.

Patient Selection


The primary purpose of crosslinking is to halt the progression of ectasia. Likewise, the best candidate for this therapy is an individual with a progressive ectatic disease of the cornea. The most common indication is keratoconus. Other diseases that may be candidates include Pellucid Marginal Degeneration, Terrien Marginal Degeneration, and post-refractive surgery (such as LASIK or Radial Keratotomy) ectasia. There currently are no definitive criteria for progression, but parameters to consider are change in refraction (including astigmatism), uncorrected visual acuity, best corrected visual acuity, and corneal shape (topography and tomography).


  • Corneal thickness of less than 400 microns is a contraindication to the standard treatment protocol
  • Prior herpectic infection is a contraindication because it may result in viral reactivation
  • Concurrent infection
  • Severe corneal scarring or opacification
  • History of poor epithelial wound healing
  • Severe ocular surface disease (ex. dry eye)
  • Autoimmune disorders

Surgical Technique

The primary goal of the first stage of therapy is to allow riboflavin to diffuse into the cornea. While there are several variations on the techniques used to accomplish this, all entail either removing or weakening the epithelial barrier of the cornea. In all instances the patient is first given anesthetic drops. Some ophthalmologists will also give preoperative antibiotics. A lid speculum is placed. After disrupting the epithelium, drops of riboflavin 0.1% (vitamin B2) are given at intervals of 1-5 minutes for 15 - 30 minutes, or until riboflavin can been seen in the anterior chamber of the eye by use of the blue filter on slit lamp examination. 

After adequate riboflavin absorption, the patient is positioned with the UV light (typically 365-370um) at a small distance (1-5cm) from the corneal apex for 30 minutes.

Following irradiation, antibiotic drops are given and a bandage contact lens is typically placed. The patient is given antibiotic drops to use postoperatively 3-4 times daily.

In the video shown, anesthetic drops are given, then the speculum is placed and the epithelium is removed. Next, drops of riboflavin are administered, followed by UV exposure.


The observational evidence for the role of corneal crosslinking has been strong. This data is also supported by several well designed randomized controlled clinical trials The most consistent finding of observational and randomized controlled studies has been that corneal crosslinking induces a slight decrease in keratometry values that tends to be maintained over at least a year. This is an important finding, as in progressive keratoconus keratometry typically rises over time and is a marker of disease progression.

While the goal of therapy is to either halt or reverse a progressive condition (keratoconus or ectasia) the various studies have not all clearly defined "progression". In fact, many studies have either failed to define this starting point of enrollment (eyes with "progressive" disease) or have defined it in a way that may not be acceptable to the ophthalmology community. For each study listed below, the definition of progression is emphasized due to its importance in our understanding of the resultant data.

C.G. Carus University Hospital, Dresden, Germany Study [3]

The strength of this study is its large sample size at 1 year. The weaknesses are its poor definition of the disease being treated (a major flaw) and poor sample size after 1 year.

  • Enrolled 480 eyes of 272 patients

Definition of progression:  ≥1D change in keratometry value over 1 year,or, Need for a new contact lens fit ≥1 in 2 years,or,"Patient reports of decreasing visual acuity"

  • 241 eyes with ≥6 months data post-CXL, 33 eyes with ≥3 years data post-CXL
  • Significant improvement in BCVA at 1 year (-0.08 logMAR BCVA) and 3 yrs (-0.15 logMAR BCVA)
  • Significant decrease in mean keratometry in 1st year (-2.68D)
  • 53% of eyes with ≥1 lines improvement BCVA 1st year; Another 20% stable in 1st year
  • 87% of eyes were stable or improved at 3 yrs (However, very low numbers were included in this analysis so conclusions must be drawn carefully)

Siena Eye Cross Study [4]

The strength of this study is also its sample size at 1 year. Similar to the Dresden study, its interpretation and application to a wider set of patients is limited by a poorly defined patient population. There is also a small sample size at 4 years.

  • Enrolled 363 eyes with progressive keratoconus

Definition of progression:  Only states that it was defined "clinically and instrumentally within 6 months"

  • 44 eyes with ≥48 months of data post-CXL
  • Significant improvement in manifest spherical equivalent at 1 year (+1.62D) and 4yrs (+1.87D)
  • Significant reduction of mean keratometry values by 1 year (-1.96D) and 4 years (-2.26D)
  • No significant change in pachymetry
  • No significant change in UCVA/BCVA
  • No significant change in cylinder

Australian Study [5]

This ongoing study has the best published study design and definition of progression to date. The researchers are looking at patients with clearly defined progressive keratoconus and will follow them for 5 years. The three-year data was published in April 2014 and is described below. [6]

Definition of progression (all over 12 months): Increase in cylinder on manifest refraction by ≥1D; or Increase in steepest keratometry value (on Sim K or Manual) by ≥1D; or Decrease in back optic zone radius of best-fitting contact lens by >0.1mm


  • Eligible eyes randomized independently to either cross-linking or control group
  • Primary outcome measure: Maximum simulated keratometry value (Kmax)
  • Secondary outcome measures: Uncorrected visual acuity (UCVA), Best-spectacle corrected visual acuity (BSCVA); spherical and cylindrical error on manifest refraction, spherical equivalent, minimum simulated keratometry value (Kmin), corneal thickness at thinnest point; endothelial cell density; and intraocular pressure
  • Assessments were performed at 3,6,12,24, and 36 months.
  • Treatment: Dresden Protocol (Epi-off): Riboflavin 0.1% drops applied (after epithelial removal with a 57 Beaver blade) every 1-3 minutes x 15 minutes and continued every 1-3 minutes as needed over the 30 minute UV exposure period. UV-X device delivered UV-A 370nm at 3.0mW/cm2 through a 9mm aperture at a distance of 50mm from the corneal apex. 
  • Control eyes did not receive sham. At 6 months compassionate treatment with CXL was allowed in control eyes, but this then excluded further patient inclusion in the study. Therefore, final results are comparing only treated vs untreated eyes. 
  • Recruitment ended in 2009 with 50 control eyes and 50 treatment eyes. 


  • Three year study reports on 46 treated eyes and 49 control eyes. Out of the 49 control eyes 12 underwent CXL and 5 had corneal transplantation. Five treated and 4 control withdrew for personal reasons.The results are not described by an intention to treat analysis (ITT), so the data after drop-out or crossover on the study patients is not included in the reported results. 
  • Primary outcome results: Significant difference in Kmax at all time points.
    1. Treated: Average Kmax flattening was -1.03 +/-0.19D. 6/46 eyes (13%) flattened by ≥ 2.0 D. 1 eye steepened by ≥ 2.0 D.   
    2. Control: Average Kmax steepening was +1.75 +/- 0.38 D. No eyes flattened by ≥ 2.0 D.  19/49 eyes (39%) steepened by ≥ 2.0 D.
    3. A negative correlation reported between baseline Kmax and change in Kmax at 36 months. Greatest improvement with eyes having a baseline Kmax ≥ 54.0 D in treatment group. 
    4. A negative correlation reported between patient's age at enrollment and change in Kmax in control group.

  • Secondary outcome results:
    1. UCVA: Improved in treatment group compared to baseline at 12, 24, 36 months (P < 0.001). Worsened in control group compared to baseline at 36 months (P<0.001).                 -
    2. BSCVA: Improved in treatment group compated to basline at 12, 24, and 36 months (P<0.007). No significant change in control group compared to baseline at 36 months. No significant difference between treated and control at any time point.       
    3. Manifest spherical refraction: No significant difference at any time point.
    4. Manifest cylindrical error: No significant change from baseline in treatment group.
    5. Corneal thickness at thinnest point on ultrasound: No significant change in treatment group at any time point. Decreased in control group at 36 months (p=0.029).
    6. Corneal thickness at thinnest point on Orbscan: Treatment group showed significant decrease most marked at 3 months of -93.00 +/- 7.98 microns (p<0.001). This reversed over the follow-up period of 36 months to -19.52 +/- 5.06 microns.Control group showed progressive decrease at 12, 24, 36 months (p<0.001).
    7. Intraocular pressure: No significant change using Tonopen in either group. Using Goldmann, significant decrease at 36 months in both groups, but no significant difference between groups. 

  • Adverse Events:
    1. Keratitis and corneal edema: 1 case. Authors attributed to premature resumption of RGP wear. Did not adversely affect outcome but did cause scar.
    2. Keratitis and iritis: 1 case. Started two days after treatment and presumed to be microbial keratitis. Resolved on ofloxacin and fluorometholone acetate 0.1%. Culture negative. 
    3. Peripheral corneal neovascularization: 1 case. Noted at 36 months and attributed to acne rosacea and not CXL. 
    4. Haze: All patients had some degree of haze and this resolved with time


The authors of this study declared in their three year report the following pertinent statments :

  • "The findings of this study suggest that CXL should continue to be considered as a treatment option for patients with progressive keratoconus"
  • "Despite the growing body of literature and continuing efforts to optimize the treatment protocol, there remains a lack of randomized controlled studies with longer-term follow-up to support the widespread clinical use of CXL for keratoconus"   

US FDA Phase III trials


The unique strengths of the completed and reported Avedro studies, NCT00674661 and NCT00647699, were an actual sham control group. However, this was also a major design shortcoming which limited the interpretation of the study data because the sham patients were allowed to seek treatment as early as 3 months into the study (see below). This time point of crossover was changed at the request of FDA. The definition of progressive disease was not as rigorous as in the Australian study but was more clearly defined than most other reported RCCTs.

11 U.S. sites

  • Keratoconus: 204 eyes enrolled
  • Ectasia: 178 eyes enrolled

A single investigator, Dr. Peter Hersh, reported his results and published in 2011, but the full trial group results have not since been reported. Here is what Dr. Hersh reported:[7]

  • Enrolled 112 eyes with progressive ectasia (77 eyes with keratoconus, 35 eyes with post-LASIK ectasia)

Definition of progression (all over 24 months--twice the amount of time as study above): Decrease sphere by ≥0.5D*, or, Decrease cylinder by ≥1D, Increase steepest keratometry value by ≥1D

Note: While ≥0.5D change in sphere over 2 years was considered disease progression, baseline refractions for patients were considered "stable" if they were within 0.75D on two consecutive measurements. Also, this study enrolled patients starting at age 14, complicating the interpretation of disease progression vs progressive myopia in cases of small changes in refraction (just >0.5D), especially in adolescents. 

  • Randomized eyes with progressive disease to either treatment (49 keratoconus / 22 ectasia) or sham (28 keratoconus /13 ectasia)
  • The sham group was allowed to cross-over into treatment by three months, limiting the usefulness of this true control group almost entirely, since most of the improvements in the treated group were not seen until after 3 months. There will be no data for comparison at 1 year. In effect, these studies are only observational studies after the 3 month time point.
  • Also looked at a fellow eye "control" group, a group of eyes failing to meet the definitions of progressive disease. Since these eyes do not have progressive disease, they are dissimilar to treated eyes at baseline and can not serve as a control.
  • Significant improvement in maximum, average, and minimum keratometry values (+1.7D, 1.1D, 0.9D respectively) in the treatment group at 1 year
  • Significant improvement in UCVA and BCVA at 1 year in treatment group (-0.07 and -0.12). However, there was also significant improvement in the sham group UCVA at 3 months (the final measurement for this cohort) by -0.08. This demonstrates how the study design limits our interpretation of the data. Our only comparison time points show the untreated eyes improved more in the primary outcome, visual acuity, then the treated eyes.
  • No significant change in spherical equivalent or cylinder in the treatment group at 1 year
  • Sham group also had no change in CDVA, spherical equivalent, cylinder, or K values at its endpoint (3 months)


  • Temporary stromal edema (up to 70%), temporary haze (up to 100%), and permanent haze (up to 10%)
  • Corneal scarring and sterile infiltrates[8][9]
  • Infectious keratitis: Bacterial/protozoan/herpetic [10][11][12]
  • Diffuse lamellar keratitis (DLK) in a post-LASIK patient[13]

Additional Resources


  1. P T Ashwin, P J McDonnell. Collagen cross-linkage: a comprehensive review and directions for future research. Br J Ophthalmol 2010;94:965e970.
  2. Wollensak G, Spoerl E, Seiler T. Riboflavin/ultraviolet-a-induced collagen crosslinking for the treatment of keratoconus.Am J Ophthalmol. 2003 May;135(5):620-7.
  3. Raiskup-Wolf F, Hoyer A, Spoerl E, Pillunat LE. Collagen crosslinking with riboflavin and ultraviolet-A light in keratoconus: long-term results. J Cataract Refract Surg. 2008 May;34(5):796-801.
  4. Caporossi A, Mazzotta C, Baiocchi S, Caporossi T.Long-term results of riboflavin ultraviolet a corneal collagen cross-linking for keratoconus in Italy: the Siena eye cross study. Am J Ophthalmol. 2010 Apr;149(4):585-93. Epub 2010 Feb 6.
  5. Wittig-Silva, C; Whiting M, Lamoureux E, Lindsay RG, Sullivan LJ, Snibson GR. A Randomized Controlled Trial of Corneal Collagen Cross-linking in Progressive Keratoconus: Preliminary Results. Journal of Refractive Surgery. 2008 (24): S720 - S725.
  6. Wittig-Silva C et al. A Randomized, Controlled Trial of Corneal Collagen Cross-linking in Progressive Keratoconus: Three-Year Results. Ophthalmology. 2014. Volume 121 (4); 812-821.
  7. Hersh PS, Greenstein SA, Fry KL. Corneal collagen crosslinking for keratoconus and corneal ectasia: One-year results. J Cataract Refract Surgery. 2011 (37): 149-160
  8. Mazzotta C, Balestrazzi A, Baiocchi S, et al. Stromal haze after combined riboflavineUVA corneal collagen cross-linking in keratoconus: in vivo confocal microscopic evaluation. Clin Experiment Ophthalmol 2007;35:580e2.
  9. Koller T, Mrochen M, Seiler T. Complication and failure rates after corneal crosslinking. J Cataract Refract Surg. 2009 Aug;35(8):1358-62.
  10. Pollhammer M, Cursiefen C. Bacterial keratitis early after corneal crosslinking with riboflavin and ultraviolet-A. J Cataract Refract Surg 2009;35:588e9.
  11. Rama P, Di Matteo F, Matuska S, et al. Acanthamoeba keratitis with perforation after corneal crosslinking and bandage contact lens use. J Cataract Refract Surg 2009;35:788e91.
  12. Kymionis GD, Portaliou DM, Bouzoukis DI, et al. Herpetic keratitis with iritis after corneal crosslinking with riboflavin and ultraviolet A for keratoconus. J Cataract Refract Surg 2007;33:1982e4.
  13. Kymionis GD, Bouzoukis DI, Diakonis VF, et al. Diffuse lamellar keratitis after corneal crosslinking in a patient with post-laser in situ keratomileusis corneal ectasia. J Cataract Refract Surg 2007;33:2135e7.