Collagen cross-linking: when and how? A review of the state of the art of the technique and new perspectives

Since the late 1990s corneal crosslinking (CXL) has been proposed as a new possibility to stop progression of keratoconus or secondary corneal ectasia, with the promising aim to prevent progressive visual loss due to the evolution of the pathology and to delay or avoid invasive surgical procedures such as corneal transplantation. The possibility of strengthening corneal tissue by means of a photochemical reaction of corneal collagen by the combined action of Riboflavin and ultraviolet A irradiation (UVA), radically modified the conservative management of progressive corneal ectasia. This is a review of the state of the art of CXL, reporting basic and clinical evidence. The paper describes basic principles, advantages and limits of different CXL techniques and possible future evolution of the procedure.


Introduction
Corneal ectasia is a progressive corneal thinning associated with alterations of stromal collagen matrix resulting in irregular protrusion of the cornea. Primary forms include keratoconus, pellucid marginal degeneration and keratoglobus, while secondary forms are mainly related to refractive surgery [1].
Many studies on keratoconus epidemiology from different countries reported an incidence of 1.3 to 22.3 per 100 000 and a prevalence of 0.4 to 86 cases per 100 000 [2].
The incidence of corneal ectasia after refractive surgery is still unknown, but it has been estimated to be 0.04-0.6 % after laser in situ keratomileusis(LASIK) [3][4][5].
Keratoconus generally starts during the second decade of life with a variable rate of progression of corneal steepening that continues until the fourth decade, when the corneal shape generally becomes stable [7].
A conservative approach in the management of keratectasia initially involves spectacles and subsequently, contact lenses.
However, surgical intervention can be necessary when patients are contact lens intolerant or cannot achieve adequate vision correction. In patients with contact lens intolerance or poor visual recovery with contact lens use, implantation of intracorneal rings (ICRS) may improve the regularity of the corneal curvature, improving contact lens fitting and visual rehabilitation [8].
Alternatively, in advanced stages or in presence of central corneal scarring, corneal transplantation may represent the only viable treatment option [1]. In these cases the preferred procedure is deep anterior lamellar keratoplasty (DALK) or, alternatively, penetrating keratoplasty (PK).
Until recently, in the "pre-CXL era", all of the treatment options for corneal ectatic diseases were aimed at only overcoming refractive limitations and not to impeding the underlying physiopathology [9].
Corneal crosslinking has grown from an interesting concept to its introduction in clinical practice in the late 1990s when it radically modified conservative management of progressive corneal ectasia with the possibility of strengthening corneal tissue [10].

Basic principles of corneal crosslinking
The primary aim of corneal crosslinking is to stop the progression of corneal ectasia. To obtain a strengthening of corneal tissue, the use of riboflavin is combined with ultraviolet A irradiation (UVA). Riboflavin plays the role of a photosensitizer in the photopolymerization process and when combined with UVA irradiation, increases the formation of intrafibrillar and interfibrillar carbonyl-based collagen covalent bonds through a molecular process that has still not been completely elucidated [1].
It was shown that during the early aerobic phase of the process of crosslinking, riboflavin molecules are excited to a single or triplet state and stromal proteins undergo a photosensitized oxidation via interaction with reactive oxygen species [11]. During the second anaerobic phase, when oxygen is depleted, corneal stroma interacts with reactive species of radical ions. This photochemical reaction results in an increase of corneal rigidity, of collagen fiber thickness and of resistance to enzymatic degradation, with consequent decrease of stromal swelling and permeability maximally, above all in the anterior stroma [12].

Basic research results
Currently, the photochemically induced effect of CXL in the cornea cannot be evidenced directly by staining methods or microscopic techniques. However, CXL induces several changes to collagen-containing tissue, from which indirect signs of the cross-linking effect can be deduced [9]. In fact, stress-strain measurements performed on human and porcine corneas documented an increased corneal rigidity after CXL treatment. The firming effect seems to be more evident in corneas with higher collagen content and in older tissue [13,14]. Moreover, it has been reported that porcine crosslinked corneas showed a reduced tendency to swelling and hydration when compared to untreated controls [15]. Ex vivo studies on corneas of humans and rabbits indicated an increase of collagen fibers thickness after CXL treatment [16,17]. Results of basic research studies showed that CXL procedure improve the corneal resistance to degradation processes mediated by pepsin, trypsin and collagenase with lengthening of the turnover time of the collagen [18].

Indications for CXL
Not every cornea with keratoconus needs to undergo crosslinking. The main aim of CXL is to stop the progression of corneal ectasia, consequently the best candidates for this treatment are patients suffering from primary or post refractive surgery ectasia with documented progression of the disease. Although the criteria to classify ectasia as progressive have not been defined, changes in refraction, uncorrected visual acuity (UCVA), best-corrected visual acuity (BCVA), and topographical parameters are to be included. To date, published clinical studies indicated that CXL was used in cases of progression over a well-defined time period. In many reports, progression was defined as an increase in Kmax of 1 diopter (D) in 1 year, or a change in either myopia and/or astigmatism ≥3 D in 6 months, a mean central K-reading change ≥1.5 D observed in three consecutive topographies in 6 months, or a mean central corneal thickness decrease ≥5 % in three consecutive tomographies in the previous 6 months. Contraindications to undergoing standard CXL treatment are the presence of corneal thickness of less than 400 microns, prior herpetic infection, severe cornea scarring or opacification, history of poor epithelial wound healing, severe ocular surface disease, history of immune disorders, and pregnancy/breast-feeding [19][20][21].

Standard procedure and clinical results
The standard Dresden protocol, as initially described by Wollensask et al. includes initial epithelial removal, the application of 0.1 % riboflavin solution for 30 min followed by 30 min of UVA irradiation with a wavelength of 370 nm and power of 3 mW/cm 2 (5.4 J/cm 2 ) [22].
A list of publications reporting the clinical results of traditional CXL procedures is shown in Table 1  .
In the last few years, several prospective and retrospective studies with a considerable follow-up period documented the effectiveness of the standard procedure in halting the progression of primary and secondary corneal ectasia, and in many cases, with an improvement of visual performance and topographical indexes.
Most of the reports about clinical outcomes of standard epi-off CXL are prospective or retrospective case series. In the follow-up after treatment, the main parameters evaluated are the maximal keratometry (K max ) and the best corrected visual acuity (BCVA). The follow-up periods ranged between one and six years. All authors reported stabilization or flattening of corneal keratometry and stabilization or improvement of visual acuity after standard epi-off procedure.
The small number of randomized controlled trials may affect the interpretation of these results. However, the results reported by Wittig-Silva et al. (2015) of 100 eyes with a three-year follow-up constitute an important milestone that confirms the effectiveness of epi-off technique in stabilizing keratoconus progression [35].
Limits and complications of standard procedure Treatment failure Treatment failure that occurs in 8.1-33.3 % of the cases is usually defined as continued progression with an increase in maximum K readings of 1.0 D over the preoperative value [66]. Poli et al. recently reported a failure rate of 11 % during a follow-up period of six years. Keratoconus worsening was considered if patients presented an increase of more than 0.1 in logMAR uncorrected and best corrected visual acuity and/or an increase of keratometric values by more than 0.75 D during the follow-up [24].
After standard CXL procedure, corneal haze is a relatively common complication reported by 10-90 % of patients. However, to date the etiology and the natural course of clinical corneal haze after epi-off procedure has not been clearly defined [67,68]. In vivo confocal microscopy showed an increased stromal reflectivity associated to edema and keratocyte activation mainly evident 3-6 months after treatment, while in the late postoperative period, anterior and intermediate stromal layers showed a reduction of cellular density and fibrosis of extracellular matrix [69].
Several cases of infective keratitis following CXL treatment have been described including bacterial, protozoal, herpetic, and fungal keratitis [70].
The rare serious adverse events following traditional CXL that have been reported included diffuse lamellar keratitis at LASIK interface, corneal melting and persistent corneal edema due to endothelial failure [71][72][73].

Introduction of epi-on technique
The diffusion process of riboflavin in the stroma is limited by corneal epithelial tight junctions [74,75], but epithelial debridement is considered the cause of the most important complications after CXL treatment such as intraoperative and postoperative pain, infective keratitis and abnormal wound-healing response [76,77]. Riboflavin penetration through the epithelium can be increased by different strategies such as changing the physicochemical properties of the riboflavin molecule by adding chemical enhancers in the riboflavin formulation [78] or performing a mechanical disruption of corneal epithelium [79]. An in vivo confocal microscopy study reported that by increasing the duration of riboflavin application up to two hours, the obtained depth of CXL effect is similar to that achieved with standard epi-off technique [80].

Iontophoresis
A novel approach to enhance riboflavin penetration is based on iontophoresis, a non-invasive system aimed to enhance the delivery of charged molecules into tissues using a small electric current [90]. Riboflavin, in the formulation used for iontophoresis, is negatively charged. It has been shown that an iontophoresis imbibition lasting five minutes achieves a sufficient riboflavin concentration in the corneal stroma for CXL treatment, with the advantage of shortening the imbibition time while preserving epithelial integrity [9].
Numerous ex vivo studies confirmed the effectiveness of iontophoresis imbibition in obtaining an adequate riboflavin concentration into the stroma and the induction of important biomolecular and structural modifications of corneal tissue [90][91][92]. Ex vivo biomechanical studies on rabbit and human cadaveric corneas showed that transepithelial crosslinking with iontophoresis imbibition induced an increase of the biomechanical resistance of human cornea comparable to that obtained with the standard crosslinking procedure [93,94].
Preliminary clinical results of iontophoresis assisted corneal CXL are promising. The technique halts keratoconus progression without significant complications (Table 3) [95-97] however, longer follow-up and studies with larger patient populations are needed to assess the real effectiveness of this technique.

Accelerated corneal crosslinking
Accelerated CXL was introduced in clinical practice in order to shorten the time required for a CXL procedure. This technique is based on the Bunsen-Roscoe law of photochemical reciprocity. That is, the same photochemical effect can be achieved with reducing the irradiation interval provided that the total energy level is kept constant by a corresponding increase in irradiation intensity [1]. Currently, commercially available ultrafast devices can achieve an irradiance intensity of 43 mW/ cm 2 . Using this setting, a total treatment time of two minutes is required to achieve a standard Dresden protocol energy dose of 3.4 J or a radiant exposure of 5.4 J/cm 2 [1]. Several recent in vivo studies using different protocols showed the procedure to be safe and effective in stopping ectasia progression (Table 4) [98][99][100][101][102][103][104][105].
Comparative studies of the effectiveness of the different CXL procedures are described in Table 5 [106][107][108][109][110][111][112][113][114][115]. Surgical protocols reported are very different and unlikely comparable. Also, the follow-up periods are very limited. Therefore, it is very difficult to deduce reliable conclusions. It seems likely that transepithelial CXL, although associated with a lower complication rate, has a lower therapeutic effect than standard CXL and would be ideal for patients with thin corneas, uncooperative individuals, or those with uncertain documented progression. Iontophoresis assisted CXL is a promised technique that could obtain clinical effects similar to those obtainable with a standard technique while maintaining the advantages of epithelium preservation. However, while the basic research results are evident, clinical outcomes are still poor. Accelerated CXL seems to represent a valid strategy to shorten the long treatment time, however the extreme variability of the protocols proposed has not been supported by adequate safety assessment. Well-designed randomized controlled trials comparing traditional CXL and all the alternative procedures are required in order to establish which is the ideal protocol for obtaining the best clinical outcomes and complication profile.

Combined treatments CXL and photorefractive keratectomy
Keratoconus has always been considered a contraindication for PRK. However, during the last few years, the idea of performing PRK in patients with stable keratoconus has been proposed. Consequently, the possibility of combining CXL and PRK was introduced in clinical practice [116][117][118][119][120]. Several clinical reports demonstrated stability in corneas that had undergone a combination of CXL and PRK, either sequentially or combined. Patients experienced improvement in spherical equivalent (SE), defocus equivalent, uncorrected and best corrected visual acuity, high order aberrations and Kmax with stabilization of keratoconus progression during a follow up period of 12-24 months [121][122][123][124].
The timing of the ablation treatment and CXL as well as the interval between the two procedures has become topics of discussion. It was reported that patients who underwent both PRK and CXL procedures in the same day obtained better clinical and topographical results with a lower rate of corneal haze, compared to patients treated sequentially [125]. This may be related to the unpredictable refractive outcomes when excimer ablation is performed on cross-linked tissue [126]. However, performing both procedures concurrently in the same day may cause an irregular healing process with the formation of persistent stromal haze, probably related to keratocyte activation, which permanently affects visual performance [127,128].

CXL and intracorneal rings
Studies reported that CXL halts keratoconus progression, but the overall results in terms of visual rehabilitation were insufficient. On the other hand, intracorneal ring segments (ICRS) produced rapid and substantial improvements of visual parameters but do not stop  progression. Theoretically, a combination of these two procedures can produce better results [1]. Several studies reported concordant results confirming that combining CXL and ICRS implantation improved uncorrected and best corrected visual acuity, refraction, and keratometry during variable follow-up periods (7-12 months) [129][130][131]. One study reported no difference in topographical or visual outcome after ICRS or ICRS combined with CXL. Thus, the real effect of the combined treatment to the keratoconus progression is still unclear [132].
Moreover, it was reported that after one or both ring explantations, the refractive effects may be stable or reversible while the topographic changes seems to be maintained [133]. Therefore, while collagen crosslinking can be performed before, in conjunction with, or after ICRS implantation, the ideal method for combining these two treatments is still undefined [1].

Alternative uses of corneal crosslinking Infections
Crosslinking has an antimicrobial effect inherent to UV light interacting with riboflavin as the chromophore. In fact, UV irradiation is used as an antimicrobial procedure for disinfecting water, surfaces and air. It damages both the DNA and RNA of pathogens including bacteria and viruses, and renders them inactive [134]. Additionally, the photoactivated riboflavin seems to produce an antimicrobial effect. In fact, the use of riboflavin as a photosensitizer to inactivate pathogens in plasma, platelet, and red cell products has been described [135].
Due to its nucleic acid specificity and its limited tendency toward indiscriminate oxidation, riboflavin was hypothesized as a photosensitizer for the inactivation of pathogens in infective keratitis. It was reported that riboflavin activated by UVA showed an antimicrobial effect on agar plates inoculated with Pseudomonas aeruginosa, Staphylococcus aureus, Staphylococcus epidermidis, Streptococcus pneumoniae, and Candida albicans. The inhibition of microbial growth was significantly higher in plates treated with UVA activated riboflavin than in those treated with UVA light alone. However, riboflavin alone did not show any significant bactericidal effect [136].
The first reported use of CXL in infective keratitis was in 2008, when Iseli at al. reported healing 4 out of 5 cases of mycobacterial and fungal corneal melting unresponsive to conventional therapy, treated with the standard Dresden protocol [137]. In 2013, Alio et al. in a systematic review and meta-analysis reported similar results [138]. In 2014, Said et al. reported a large prospective clinical trial on infective keratitis comparing 21 eyes treated with CXL in addition to antimicrobial therapy in 19 eyes that received only antimicrobial therapy. They did not find a significant difference between both groups in terms of healing time and final visual acuity. Three patients treated with antimicrobial therapy alone experienced corneal perforation and one an infection relapse while no significant complications occurred in CXL group. The authors conclude that CXL could serve as a valuable adjuvant therapy and may reduce or avoid severe complications preventing the need for emergency keratoplasty [139].

Pseudophakic bullous keratopathy
In case of corneal edema due to endothelial failure, it has been shown that CXL effect increases corneal resistance to swelling processes. In fact CXL increases the interfiber collagen connections and it's difficult for stromal fluid to separate collagen lamellae and create a potential space for edema accumulation . Therefore, the use of corneal CXL was proposed as an alternative approach for the management of pseudophakic bullous keratopathy (PBK) with the aim to reduce ocular discomfort, improve visual acuity, and delay the need for keratoplasty [140].
Clinical studies evaluating the effectiveness of corneal CXL in the treatment of PBK reported a significant improvement in corneal transparency, corneal thickness, and ocular pain one month postoperatively. However, CXL did not seem to have a long-lasting effect over six months in decreasing pain and maintaining corneal transparency [141,142].

Conclusions
At the light of this review we can conclude that there is still much to understand about the real modification of corneal collagen structure after the photochemical CXL reaction. Moreover the constant aim of basic and clinical research today is to identify the best strategies and combination of imbibition and irradiation that can lead to the better clinical efficacy together with the maximum safety of the treatment.

Competing interests
The author declares that he has no competing interests.