A Review of Corneal Collagen Cross-linking – Current Trends in Practice Applications

Objective: To review the literature on current applications of corneal Collagen Cross-Linking (CXL). Methods: A review of publications on corneal cross-linking was conducted. This included systemic reviews, randomized controlled clinical trials, cohort studies, case-controlled studies and case series. A summary of the publications is tabulated. Results: The original indication of riboflavin – Ultraviolet-A (UVA) induced corneal collagen cross-linking is to arrest the progression of keratoconus. Studies show that it is effective in arresting the progression of keratoconus and post-LASIK ectasia with the standard Dresden protocol (epithelium-off). There are also improvements in visual, keratometric and topographic measurements over time. Severe complications of cross-linking are rare. The epithelium-on techniques have less efficacy than the Dresden protocol. Accelerated protocols have variable results, with some studies reporting comparable outcomes to the Dresden protocol while other studies reporting less efficacious outcomes. Cross-linking combined with refractive procedures provide better visual outcome but long term studies are warranted. Cross-linking for the treatment of infective keratitis is a promising new treatment modality. Initial studies show that it is more effective for superficial rather than deep infections and for bacterial rather than fungal infections. Conclusions: Corneal cross-linking is a procedure with an expanding list of indications from the treatment of corneal ectasias to infective keratitis. While the standard Dresden protocol is established as the gold standard treatment for progressive keratoconus, the more recent protocols may require further refinements, investigative and long-term studies.


INTRODUCTION
Wollensak, Spoerl and Seiler reported the first clinical study on riboflavin -UVA induced corneal collagen crosslinking procedure for the treatment of progressive keratoconus in adults [1]. Since then, corneal cross-linking has been widely used for the treatment of progressive keratoconus as well as other conditions including post-LASIK ectasia. Corneal cross-linking is also used in conjunction with laser vision correction procedures and intrastromal ring procedures. More recently, it is used to treat infective keratitis. The purpose of this paper is to provide a review of the     The primary indications for corneal cross-linking are progressive keratoconus in adults and post-LASIK ectasia. Stabilisation of ectasia with up to five years follow-up was reported in 23 eyes that underwent the epithelium-off CXL technique with an average reduction in spherical equivalent refractive error of 1.0D and maximum keratometry (Kmax) of 2.0D [1]. Subsequently, other prospective case cohort studies showed similar results of stabilization of keratoconus and improvements in visual acuity and topography [10 -16]. Comparative studies using the fellow eye as control showed stabilization of the treated eye and continued progression in the fellow untreated eye [17].
Wittig-Silva et al performed the first randomized controlled trial and found significant flattening of the steepest keratometry and a trend towards better visual acuity with longterm follow-up showing continued flattening up to 4 years after treatment [3,4]. Long term studies with follow-up ranging from 4 to 7 years also show improvements in visual acuity and corneal topography [18 -24]. O'Brart et al. reported improvements in topographic and wavefront parameters evident at 1 year and which continue to improve after 7 years [24].  Epithelium-off CXL has also been shown to be effective in stabilizing post-LASIK ectasia with improvements in the visual acuity with more than 2 years follow-up [25 -27]. The US based prospective clinical trial for cross-linking showed improvement in visual acuity and maximum keratometry in patients with keratoconus and post-LASIK ectasia [28]. The keratoconus group had more corneal flattening than ectasia patients. A longer-term study by Richoz [29] with a mean follow-up of 25 months showed improved mean corrected visual acuity and mean Kmax.

PAEDIATRIC KERATOCONUS
Paediatric cases often present with keratoconus that progress more rapidly than adult onset keratoconus. One study showed that 88% of paediatric cases progress over a short period of time [33]. Hence it is not necessary to document progression and treatment of paediatric keratoconus is recommended at the time of presentation. Studies show that there is an initial favorable response with improvements in visual acuity, keratometry and spherical equivalent at one year followup [34,35]. In the long term, studies show that the keratoconus continue to progress despite the initial response [33]. Studies also show that the epithelium-off technique is to be preferred over transepithelial cross-linking as the latter technique show worsening keratometry values over time [36].

EPITHELIUM-ON (TRANSEPITHELIAL) CORNEAL CROSS-LINKING (TABLES 3A and 3B)
The standard Dresden protocol (epithelium-off) corneal cross-linking is associated with significant postoperative pain and visual recovery is gradual. There are also risks of infection and corneal scarring. Hence, epithelium-on corneal cross-linking was introduced to reduce the issues associated with the standard protocol. However, riboflavin is a hydrophilic molecule making penetration through the intact hydrophobic corneal epithelium difficult. In order to improve epithelial permeability to riboflavin, additives such as benzalkonium chloride, topical aneasthetic, tris-hydroxymethyl aminomethane (trometamol), sodium ethylenediaminetetraacetic acid (EDTA) are included in the riboflavin. Other techniques include increased riboflavin concentration and iontophoresis.
The outcomes for transepithelial cross-linking using riboflavin with trometamol/EDTA are mixed. Some studies report improvement in visual acuity and keratometry measurements [37 -39] while other studies report worsening of keratometry measurements [36,40]. Also the demarcation line was noted to be shallower in the transepithelial group [37].
Soeters et al, 2015 [42] Randomised clinical trial/ 3mW/cm 2  Kmax decreased in the epi-off group but increased in epi-on group.
-    The majority of randomized clinical trials comparing the standard and transepithelial technique report better outcomes with the standard technique (reduced Kmax), whilst the transepithelial technique had worsening of keratometry measurements (Kmax) [41,42].

Current Trends in Corneal Cross-Linking
Transepithelial iontophoretic cross-linking involves the application of a small (1mA) negative charge to enhance riboflavin absorption. Clinical studies showed better results than transepithelial cross-linking alone with improved visual acuity and stability of refraction and topography [43,44]. Iontophoretic transepithelial cross-linking has also been used in the treatment of paediatric keratoconus using the accelerated protocol with favorable outcome (improved visual acuity and stability of refraction and topography) [45,46].

ACCELERATED CROSS-LINKING (TABLES 4A and 4B)
The Bunsen-Roscoe Law of Reciprocity states that the photochemical biological effect of ultraviolet light is proportional to the total energy dose delivered, regardless of the applied irradiance and time [49]. In the context of UVA cross-linking, for the same energy dose delivered, one could shorten the duration of treatment by applying a higher irradiance power. Laboratory studies show that this law could be applied for corneal cross-linking. Wernli et al treated ex vivo porcine eyes with CXL using a total of 5.4J/cm 2 delivered in a range of irradiances from 3 to 90mW/cm 2 . Significant stiffening was observed in eyes treated with irradiances from 3 to 45mW/cm 2 [50].
(  Kanellopoulos reported the first clinical study (randomized prospective contralateral eye study) on accelerated cross-linking using 7mW/cm 2 irradiation 15 minute protocol (5.4J/cm 2 ) and Dresden protocol [51]. Stabilisation of keratoconus was achieved in both groups with a flattening of steep keratometry observed in both groups with no change in the endothelial cell density. Subsequent studies employed higher irradiances and shorter duration times: Gatzioufas et al reported preliminary results using 18mW/cm 2 for 5min at 5.4J/cm 2 with no complications [52], Tomita reported on a comparative study on accelerated CXL(30mW/cm 2 for 3 minutes at 5.4J/cm 2 ) and CXL(Dresden protocol). No statistical differences were found between the two groups for uncorrected distance visual acuity or corrected distance acuity and significant flattening was observed in average keratometry in (Table  contd.....

Current Trends in Corneal Cross-Linking
The Open Ophthalmology Journal, 2018, Volume 12 199 both groups at 1 year follow-up [53]. Tomita also compared the demarcation line between CXL and accelerated CXL(30mW/cm 2 for 3 minutes at 5.4J/cm 2 ) and described a mean demarcation line depth of 294.38 +/-60.57 um in the accelerated group and 380.78 +/-54.99 um in the CXL group. The difference was not statistically significant. Kymionis reported a greater depth mean demarcation line (350.75 +/-49.34um) in the Dresden CXL than the accelerated CXL (9mW/cm 2 for 10 minutes) with mean demarcation line at 288.46 +/-42.37um [54]. However, when the protocol was changed to 9mW/cm 2 for 14 minutes, there was no difference in corneal stromal demarcation line depth between Dresden CXL and accelerated CXL [55].  year follow-up [56]. He reported that the accelerated CXL (9mW/cm 2 for 10 minutes and 18mW/cm 2 for 5 minutes) had similar outcomes to standard CXL but the accelerated CXL using 30mW/cm 2 for 3 minutes was not as efficacious.
The reduced efficacy of the 30mW/cm 2 treatment is postulated to be due to the depletion of oxygen in these high fluence treatments and pulsed treatments were introduced in an effort to replenish oxygen in the cornea during high fluence treatments. Mazotta et al [57] reported a greater reduction of keratometry in pulsed compared to continuous treatment The treatment protocol of accelerated CXL is still in evolution due to the variability of the outcomes reported. Further long term studies are needed to confirm the comparability of accelerated CXL to CXL. (Table  contd.....

CROSS-LINKING COMBINED WITH REFRACTIVE PROCEDURES FOR THE TREATMENT OF CORNEAL ECTASIA (CXL PLUS)
Although CXL is effective in stabilizing keratoconus, in many cases, patients are unable to achieve functional vision after CXL and still require rigid contact lens wear. Hence refractive treatments in combination with CXL (CXL plus) have been introduced to provide patients with better visual acuity.

Photorefractive Keratectomy (PRK) and CXL
Kanellopoulos and Binder reported on the first case of topography-guided PRK performed 1 year after CXL for the treatment of keratoconus showing improvement in visual acuity [67]. Subsequently Kanellopuolos reported that simultaneous treatment (PRK followed by CXL) is more effective than sequential treatment (CXL followed 6 months later by PRK) [68] in the visual rehabilitation of keratoconus. Other studies also confirmed the safety and efficacy of simultaneous topography guided PRK and CXL [69 -77]. Some studies advocate the use of mitomycin C 0.02% after laser ablation while others do not.

Transepithelial Phototherapeutic Keratectomy (PTK) and CXL
The removal of corneal epithelium in the CXL procedure is replaced with PTK which not only removes epithelium but also regularises the anterior corneal surface [78]. Kymionis et al, in a comparative study showed that epithelial removal using PTK during CXL (Cretan protocol) results in better visual and refractive outcomes than mechanical removal of epithelium [79]. Transepithelial PTK uses the patient's epithelium as a masking agent. At the apex of the cone, epithelium and the anterior stromal surface is removed resulting in a more regularized anterior corneal surface [78].

CXL and Intrastromal Corneal Ring Segment (ICRS) Implantation
Intrastromal Corneal Ring Segment (ICRS) implantation is currently a treatment option for keratoconus and post-LASIK ectasia [80 -82]. However, it does not prevent keratoconus progression and in young patients with progressive keratoconus, CXL may be performed in addition to ICRS to add biomechanical stability. Chan et al reported that ICRS (Intacs) with CXL resulted in better keratoconus improvement than Intacs insertion alone [83]. Coskunseven reported that ICRS implantation followed by CXL resulted in greater keratoconus improvement than CXL followed by ICRS [84]. El Awady reported that CXL has an additive effect after Keraring implantation (Mediphacos, Belo Horizonte, Brazil) [85]. Studies on simultaneous transepithelial ICRS -CXL report that CXL has an additive effect on ICRS [86,87]. Lam et al. reported a case of post-LASIK ectasia treated with femtosecond laser-assisted ICRS implantation followed by CXL resulting in stabilization of ectasia and improvement in vision [88].

CXL and Phakic Intraocular Lens Implantation
Several case series report on the safety and efficacy of CXL followed by toric Visian ICL [89 -91]. Similar results were obtained with Artiflex lens implantation 6 months after CXL and toric iris-claw lens implantation (Artiflex: Ophtec BV) [92]

CUSTOMISED CROSS-LINKING
Kanellopoulos first reported on a case of customized high fluence toric application of transepithelial cross-linking which resulted in a reduction in corneal astigmatism(0.8D) and improvement in the uncorrected visual acuity from 20/40 to 20/25 at 6 months followup [93].
Roy and Dupps demonstrated using three dimensional finite element analysis model that there is differential biomechanical weakening in the area of the cone. They concluded that there is greater efficacy of smaller diameter cone-centric treatments for the reduction of corneal curvature [94]. The Mosaic delivery system (KXL II, Avedro Inc.,Waltham, MA, USA) offers customised cross-linking (photorefractive intrastromal cross-linking-PiXL). Initial studies on customised cross-liking report greater corneal regularisation and reduction in maximum keratometry than conventional cross-linking [95 -97].
Customised cross-linking has recently been used to correct low degrees of refractive error in a patient without keratoconus. Kanellopoulos first described the preliminary results for low myopic correction [98]. Lim et al reported on the results of PiXL for the treatment of low myopia in a cohort of 14 eyes with a 1 year followup [99]. High fluence UV-A irradiation ranging from 10-15 J/cm was delivered over a 4.5mm central zone. A mean reduction of 0.72 +/-0.43D was noted at 1 year followup. Kanellopoulos reported on the results of PiXL for hyperopia, with a mean correction of +0.85D [100].

COMBINED LASER IN-SITU KERATOMILEUSIS (LASIK) AND ACCELERATED CORNEAL CROSS-LINKING
LASIK, with the creation of a corneal flap and ablation of corneal tissue weakens the biomechanical strength of the cornea and in susceptible eyes, may predispose to post-LASIK ectasia. In order to strengthen the cornea, accelerated corneal cross-linking is performed simultaneously after the LASIK procedure. Studies have reported that combined laser in-situ Keratomileusis (LASIK) and accelerated corneal cross-linking may confer additional benefits of early refractive and keratometric stability after LASIK, improving the predictability of refractive outcomes in patients. The indications are high myopia corrections, hyperopic corrections, patients with lower residual stromal bed thickness and patients with thin corneas.
LASIK in patients with high myopia has a higher incidence of refractive regression [101,102]. Hence the use of simultaneous accelerated CXL and LASIK to stabilize the patient's refraction may be useful particularly in patients with high myopia. In a prospective study comparing 73 LASIK Xtra eyes and 82 LASIK only eyes, Kanellopoulos et al found that 90.4% of LASIK Xtra eyes had UDVA of 20/20 or better as compared to 85.4% of LASIK only eyes at postoperative month 12 (p = 0.042) [103]. Similar findings were also shown in another prospective study comparing LASIK Xtra in one eye and LASIK only in the fellow eye over a 12-month period [104].
Kanellopoulos et al also reported that corneal keratometry measurements were stable for LASIK Xtra eyes and slightly regressing in LASIK only eyes (p = 0.039) [103]. Subsequently, Kanellopoulos also reported a statistically significant reduction in regression in a 2 year analysis of LASIK-CXL for high myopia compared to the LASIK only group [105]. LASIK and accelerated cross-linking for hyperopia also showed better refractive stability and less regression than LASIK only [106]. Another study by Kanellopoulos found significantly less epithelial thickening in the LASIK and accelerated CXL group compared to the LASIK only group. This could possibly explain the differences in the refractive stability between the 2 groups [107]. LASIK and accelerated CXL has been shown to be comparable to LASIK only in terms of safety, as evidenced by similar loss of corrected visual acuity in both groups [103,105].
Studies on LASIK and accelerated cross-linking report the use of different levels of UV irradiance, energy levels and illumination times [104,105,108,109]. The energy levels vary from 1.8J/cm 2 to 5.4J/cm 2 . It is postulated energy settings may be lower (1.8J/cm 2 ) than conventional cross-linking treatment for keratoconus (5.4J/cm 2 ) since eyes undergoing LASIK and accelerated cross-linking are normal eyes.
One of the goals of performing LASIK with accelerated CXL is reducing the risk of post-LASIK ectasia. A review of the literature of eyes that had undergone LASIK and accelerated cross-linking with at least 2 years follow-up showed no report of post-LASIK ectasia supporting the claim that LASIK with accelerated CXL may prevent post-LASIK ectasia [110]. However, post-LASIK ectasia has been shown to develop as long as 5 to 10 years postoperatively. Hence these reports are not sufficient to make this conclusion and further long term studies are warranted.

CROSS-LINKING FOR THE TREATMENT OF INFECTIVE KERATITIS (PACK-CXL) (TABLE 5)
Infectious keratitis is a serious, sight-threatening condition that can result from bacterial, viral, fungal or protozoal infection. Standard treatment for infectious keratitis involves both systemic and topical antimicrobial therapy. However, the effectiveness of this treatment depends on microbial sensitivity to the drug as well as severity of the disease process.
Infections not responding to antimicrobial therapy may require therapeutic keratoplasty (lamellar or penetrating). Corneal Collagen cross-Linking (CXL), or Photo Activated Chromophore for Keratitis (PACK-CXL) has been investigated as a possible alternative treatment for infectious keratitis. CXL treatment stiffens the corneal stroma through the effect of photo-activated riboflavin on collagen fibers. This makes the cornea more resistant to enzymatic degradation by microbes, thus reducing the progression of corneal melting [111,112]. Also, in CXL, riboflavin enters an excited state and reacts with ambient oxygen to create Reactive Oxygen Species (ROS). These ROS cause cell death by damaging intracellular components. Microorganisms are also killed by ROS damage to microbe DNA and cytoplasmic membrane, resulting in leakage of cellular contents and inactivation of enzymes and membrane transport systems [113]. Iseli et al first reported the use of PACK-CXL in infectious keratitis in 2008 [114]. Progression of corneal melting was successfully halted, with emergency keratoplasty not required in any of the cases. Subsequently, other case series report that PACK-CXL is effective in treating infectious keratitis caused by different organisms [115 -124]. CXL is contraindicated in eyes with previous herpes simplex [125]. It has been shown to be more effective for superficial rather than deep infections [125 -127] and for bacterial rather than fungal infections [125]. Makdoumi et al. is the first to report treating bacterial keratitis with only PACK-CXL and no antibiotics [120]. Results were successful and all eyes responded to the treatment, with only 2 eyes requiring additional antibiotics and 1 eye requiring an amniotic membrane transplant.
However, comparative clinical trials show that PACK-CXL with antimicrobial treatment had similar results as the control group (only antimicrobial treatment) in terms of healing time and corrected visual acuity [128 -130]. The PACK-CXL group had a bigger corneal ulceration width and length [128] and a higher risk of perforation was noted for deep fungal keratitis [130].

OTHER APPLICATIONS OF CROSS-LINKING
Corneal cross-linking can be employed to prevent further progression in pellucid marginal degeneration(PMD) which is considered a variant of keratoconus. Several studies report on its safety and efficacy [133,134] [135]. Additionally Kymionis performed simultaneous photorefractive keratectomy and CXL in a both eyes of a patient with PMD resulting in significant improvement in the corneal topography measurements and visual acuity [136].
Corneal cross-linking has also been used as a treatment for pain relief in bullous keratopathy. Sharma et al reported on cross-linking treatment in 50 eyes with bullous keratopathy and concluded that the pain relief achieved was temporary with corneal bullae recurring in 44% of the cases. No long term improvement in visual acuity was seen [137]. Kozobolis et al reported on CXL as an adjunctive treatment for patients with combined bullous keratopathy and infective keratitis [138].
Mukherjee et al performed an animal model evaluation of cross-linking donor corneas for penetrating keratoplasty and concluded that it reduces intraoperative induced astigmatism and aberrations in an animal model [139]. Ting et al [140] conducted a randomised controlled trial to investigate whether donor corneas pre-treated with cross-linking reduced myopic refractive errors for keratoconic eyes after penetrating keratoplasty. At 3 years followup, they found significantly improved corrected visual acuity, reduced Kmax and keratometric astigmatism in the CXL treated group.
Crosslinked corneal tissue has been shown to have stiffer biomechanical properties and to be more resistant to degradation by collagenolytic enzymes. Robert et al reported on cross-linking of the Boston keratoprosthesis donor carrier to prevent corneal melting in a patient with post KPro corneal melt. The patient maintained his visual acuity and showed no evidence of corneal thinning or melt in the first postoperative year [141].

CROSS-LINKING COMPLICATIONS
Complications of corneal cross-linking include corneal haze, corneal scarring, infective keratitis, sterile infiltrates, delayed epithelial healing, failure of treatment, excessive corneal flattening with hyperopic shift and endothelial failure [142] Anterior corneal haze occurs frequently and usually appears 1-2 months after cross-linking. It is usually transient and clears by 6 to 12 months [142]. Permanent stromal scarring [143] may occur and the incidence has been reported to be as high as 8.6% in one series [144]. It may also be more prevalent in eyes receiving simultaneous PRK followed by CXL [145]. Infective keratitis after cross-linking is rare. Shetty et al reported an incidence of 0.0017% (4 out of 2350 patients) with all 4 cases treated with the epithelium-off technique [146]. Sterile infiltrates present in the early postoperative period (days to weeks) and usually resolve with topical steroid medication [147]. Other uncommon complications include corneal melting associated with atopic eye disease [148] and reactivation of herpetic keratitis [149] Cross-linking should be avoided in patients with previous herpetic eye disease and atopic eye disease should be controlled prior to CXL.
Long-term studies show that progression of keratoconus after cross-linking may occur in about 8% of cases [23,24,150]. Hence it is necessary to counsel patients preoperatively about the various potential side-effects and also about the failure rate of the procedure.
Corneal endothelial damage may occur if the safety limits regarding corneal thickness to prevent endothelial toxicity are not adhered to. Sharma et al reported a 1.4% incidence of persistent endothelial failure in 350 eyes treated with the standard epithelium-off protocol although the safety limit of corneal thicknesses of greater than 400um (epithelium-off) were adhered to [151]. This could be due to intraoperative stromal dehydration resulting in stromal thinning, lack of homogeneity and focusing/alignment issues of the UV devices.
Although limbal stem cell damage after CXL has been shown in cadaveric eyes [152], long term studies show no evidence of limbal dysfunction [18,19,21].

CONCLUSION
Corneal cross-linking is a unique procedure with an expanding list of indications from the treatment of corneal ectasia to infective keratitis. While the standard Dresden protocol is established as the gold standard treatment for progressive keratoconus, the more recent protocols may require further refinements, investigations and long-term studies. New indications and treatment protocols are also being developed and we look forward to these treatments in the future.