Introduction

Corneal collagen cross-linking (CXL) constitutes a minimally invasive surgical intervention employed for the management of ectatic corneal disorders, such as keratoconus, pellucid marginal corneal degeneration and post-laser in situ keratomileusis (LASIK) corneal ectasia [1–4]. It has been previously demonstrated that in keratoconus, the number of diagonal links of collagen fibrils is significantly reduced [5]. These fibrils provide the cornea with mechanical stability. When they are lacking, the cornea gradually becomes destabilized due in part to thinning of the central and para-central areas, which in turn causes irregular astigmatism, myopia and reduction in visual acuity.

The CXL principle is based on the formation of chemical bonds (cross-links) among stromal collagen fibrils, thereby strengthening and stabilizing the diseased cornea. The use of riboflavin, also known as vitamin B2, in conjunction with ultraviolet-A (UV-A) irradiation facilitates the formation of cross-links between collagen fibrils in the corneal stroma, providing a stiffening effect capable of halting progression of the ectasia [1, 2].

Prior to the introduction of CXL, the possible treatment options for ectatic corneal disorders included spectacle correction, contact lenses, intrastromal corneal ring segment implantation [6] and, in advanced cases, lamellar or penetrating keratoplasty [7]. All these options had a single goal of symptomatic treatment and did not aim to stabilize the ectatic disorder per se. In contrast, CXL arrests the progression of the primary disorder, thereby addressing the pathophysiology of the disease rather than just its symptoms.

Original Surgical Technique (Dresden Protocol)

The term ‘cross-linking’ in the biological sciences is used to express the formation of chemical bridges following chemical reactions between proteins or other molecules. Usually, cross-links can be formed by chemical reactions that are initiated by heat, pressure, or radiation. The result of such reactions is the change in the biological molecules’ physical properties. Natural enzymatic cross-linking is part of the post-translational modification of collagen. During the aging process of the human body, both enzymatic and non-enzymatic cross-linking occur in various parts, such as the skin or the arteries. A key observation that resulted in the introduction of CXL for the management of keratoconus is the fact that diabetics often do not show progression of corneal ectatic disorders due to naturally occurring non-enzymatic cross-linking [1].

The standard CXL protocol was first described by Wollensak and colleagues [1] and is often referred to as the ‘Dresden protocol’. This treatment protocol constitutes the benchmark of the CXL procedure and has set the foundation for the evaluation of safety and efficacy of the technique. CXL is always conducted under sterile conditions in the operating room. After application of topical anesthesia, the central 8–9 mm of the epithelium is removed. It is now possible to perform mechanical removal of the epithelium with a blade (or more recently employing a rotating brush), removal with the use of alcohol (laser-assisted sub-epithelial keratectomy, LASEK), or removal with a laser (transepithelial CXL).

Riboflavin 0.1% solution is applied every 2–5 min for approximately 30 min to facilitate penetration of the corneal stroma, until the stroma is completely penetrated, as indicated by yellow flare in the anterior chamber. Different commercially available UV-A light sources can be used. The role of riboflavin in CXL is twofold. Not only does it work as a photo sensitizer for the induction of cross-links, but by acting as a selective filter, it also protects the underlying tissues from the harmful influence of UV-A. It has been shown by Wollensak and coworkers [8] that the cytotoxic irradiance level stands at 0.5 mW/cm2 for keratocytes after UV-A irradiation combined with the photosensitizer riboflavin, which is 10-times lower than the cytotoxic irradiance of 5 mW/cm2 after UV-A-irradiation alone.

Before treatment, the intended 3 mW/cm2 surface irradiance (5.4 J/cm2 surface dose) can be confirmed using a UV light meter. In a previous investigation, Wollensak [2] proposed a pre-operative corneal thickness of 400 μm as a minimum safety limit to avoid posterior corneal tissue damage during CXL. In rabbits, corneal endothelial toxicity was reached by irradiance of 0.36 mW/cm2, while this level of radiation exposure in human corneas reached a depth of less than 400 μm. Spoerl and coworkers [9] also reported that a safety threshold of 400 μm corneal pachymetry in the presence of riboflavin was necessary to limit UV-A irradiance to less than 1 J/cm2 at the level of the corneal endothelium, anterior chamber, lens and retina. Unquestionably, the presence of riboflavin enhances the safety profile. The cornea is exposed to the above stated level of UV-A energy for a total of 30 min. During treatment, riboflavin solution is applied every 2–5 min to ensure saturation of the tissue. After treatment, a bandage contact lens is applied until the epithelium is completely healed and is combined with the application of topical corticosteroids, antibiotic, and non-steroidal anti-inflammatory agents.

Efficacy, Safety and Clinical Outcomes

Several clinical studies of CXL have now been conducted in Europe and the USA, all of which provide information on the efficacy and safety of the procedure in the short, medium and long term. Even though CXL has become common practice in Europe, in the USA, the US Food and Drug Administration (FDA) has not yet approved this treatment modality. Nevertheless, there are two ongoing clinical trials, the results of which may lead to FDA approval. This will positively impact availability and the cost of treatment options, and also liability issues in the USA.

The first study in human eyes was conducted in 2003 by Wollensak and coworkers and included 23 cases [1]. This study included follow-up data for up to 4 years and demonstrated topographic stability and improvement of the mean keratometric (K) values in approximately 70% of treated patients. Furthermore, 65% of treated patients also showed a small improvement in visual acuity [1].

Caporossi and coworkers [10] presented preliminary results on CXL, including ten cases with 6 months follow-up. Refractive results demonstrated a reduction of about 2.5 diopters (D) in the mean spherical equivalent, topographically confirmed by the reduction in mean K values. A second study by the same research group detected stability of the corneal ectatic disorder in 44 cases after a minimum of 48 months of follow-up [11]. Corneal symmetry improvement was seen in 85% of patients.

In a subsequent comparative study, Coscunseven and coworkers [12] confirmed the initial findings reported by Wollensak and coworkers [1]. Following CXL, this group detected a mean decrease in spherical equivalent refraction of 1.03 ± 2.22 D and an increase in uncorrected distance visual acuity (UDVA) and corrected distance visual acuity (CDVA) of 0.06 ± 0.05 and 0.10 ± 0.14 D, respectively, for the group treated. In contrast, they documented progression for all tested parameters in the eyes that were not treated.

Agrawal [13] presented his results in a series of Indian eyes showing that 1 year after CXL treatment, 54% of eyes gained at least one line of CDVA. The K value of the apex decreased by a mean of 2.73 D in 66% of eyes and remained stable (within ±0.50 D) in 22% of eyes. Similar 1-year follow-up results confirming the efficacy of the procedure were reported by Asri and coworkers [14] in a series of 142 eyes with a greater than 2-D difference in K readings in 21.3% of cases and stability in another 68.8%. Similar statistically significant improvements in all tested parameters after 12 months have been reported for two other studies [15, 16]. Vinciguerra and coworkers [17] detected similar gains in keratometric and refractive findings but also showed that corneal and total wavefront aberrations were reduced 1 year after CXL treatment in their series of 28 eyes. Koller and coworkers [18, 19] made an interesting observation during the 12-month follow-up period of their patients treated for keratoconus. They observed corneal flattening [18] with regularization of the corneal shape [19], as captured by means of Scheimpflug imaging. Thus, the authors concluded that the CXL ‘effect’ causes a progressive topographic improvement throughout the follow-up period.

In another study, Goldich and coworkers [20] observed a significant improvement in CDVA (0.21 ± 0.1 to 0.14 ± 0.1; p = 0.002) and stability in UDVA (0.62 ± 0.5 to 0.81 ± 0.49; p = 0.475). There was a significant decrease in the steepest-meridian keratometry (53.9 ± 5.9 to 51.5 ± 5.4 D, p = 0.001) recorded 24 months after CXL in eyes with keratoconus. Similar long-term, successful results (with 3-year follow-up) have been published by Raiskup-Wolf and coworkers [21], who conducted a study in 241 keratoconus cases. Only two patients needed a second CXL treatment because of apparent progression of the ectasia.

Kymionis and coworkers [22] established a significant increase in intraocular pressure (IOP) measurements by Goldman applanation tonometry (GAT) at 6 months (from 9.95 ± 3.01 to 11.40 ± 2.89 mmHg) and then at 12 months (from 9.95 ± 3.01 to 11.35 ± 3.38 mmHg) following CXL (both p < 0.001). The authors attributed this pressure rise to the increased corneal rigidity and stiffness, which came about due to the formation of cross-links in the corneal stroma of the treated cases. Gkika and coworkers [23] evaluated IOP with three different tonometers—GAT, Pascal dynamic contour tonometer (PDCT) and ocular response analyzer (ORA) tonometer—before and after CXL, and concluded that PDCT had greater accuracy in keratoconus patients before and after CXL. The same research group tried to assess corneal hysteresis and corneal resistance factor in keratoconic eyes before and after CXL, proving that CXL exerts a non-significant impact on ORA measurements [24].

Several studies in the literature have investigated the use of CXL in post-LASIK corneal ectasia with up to 25-month follow-up [3, 25, 26]. These studies have demonstrated no progression of ectasia in conjunction with visual and topographic improvement.

There are now a number of published investigations that have evaluated the safety of the CXL technique [20, 27, 28]. The first study [20] evaluated the corneal endothelium by specular microscopy and the retina by comprehensive fundus examination and optical coherence tomography analysis. The investigators concluded that no morphologic abnormalities were detected after CXL, and that the endothelial cell density and foveal thickness remained unchanged [20]. A subsequent study [27] reported no changes in crystalline lens density and foveal thickness 12 months after CXL, while the third study [28] also confirmed the absence of retinal morphologic changes after CXL.

A recent study investigated CXL specifically in pediatric patients and reported encouraging preliminary results [29], but these observations must be confirmed in large controlled trials, and the technique must be applied with caution in children. Chatzis and Hafezi [30] documented visual, refractive and topographic stabilization and improvements after pediatric CXL similar to those reported for adult treatment outcomes over 2 years. Nevertheless, they did observe some keratometric progression at 3 years of follow-up. The findings suggest that pediatric CXL may not provide long-term stability comparable to adult treatment and these younger patients may require re-treatment, especially in a subset of those patients with persistent eye rubbing.

Corneal Collagen Cross-Linking in Thin Corneas (Under 400 μm)

It should be noted that there are many cases with keratoconus and post-LASIK ectasia who exhibit corneal stromal thickness less than 400 μm, and who achieve a satisfactory visual acuity by means of spectacles or contact lenses. In these cases, the current CXL treatment protocol prohibits the surgical procedure due to inadequate corneal thickness. Nevertheless, two groups have now proposed an alternative treatment protocol targeting thin corneas [31, 32]. They employ hypo-osmolar riboflavin solution with overall satisfactory results. Still, it is essential to bear in mind that since these ectatic corneas with less than 400 μm corneal thickness are outside the range of the Dresden protocol, the risk of the procedure is greater and a higher rate of complications may occur. For example, a significant postoperative decrease in endothelial cell density has been documented by Kymionis and coworkers in a few of these cases [33].

Complications

To date, few complications have been reported during and after CXL. Therefore, CXL is now generally considered a safe and effective surgical procedure. In some cases, stromal edema is detected immediately after CXL surgery, but this is transient and, fortunately, without clinical significance.

A case report of herpetic keratitis with iritis after CXL [34] has led to the belief that cross-linking can induce herpetic keratitis with inflammation in rare cases, even in patients with no history of herpetic disease. Another case of diffuse lamellar keratitis developing after CXL in a patient with post-LASIK ectasia has also been reported [35]. This case was successfully managed with intensive topical corticosteroids. Labiris and coworkers [36] published a case of acute inflammatory response after CXL resulting in corneal melting and descemetocele, which led to perforation.

Finally, a few cases of infectious keratitis post CXL have been reported in the literature [37–39]. These resulted in corneal ulceration and scarring. Another single case of ectasia progression despite CXL treatment in a pregnant woman has been published [40]. In this last case, one possible hypothesis could be that the high estrogen levels associated with pregnancy may have adversely affected the rigidity of the cornea resulting in failure of the CXL procedure [41].

New CXL Indications

A new promising line of indications for CXL in other types of corneal pathology is currently under investigation.

Infectious Keratitis

The treatment of microbial keratitis with the use of CXL has recently raised interest among the scientific community. To date, CXL has been shown clinically to be beneficial in cases of resilient pathogens, such as drug-resistant Streptococcus pneumoniae and Gram-negative Escherichia coli [42–44]. Martins and coworkers [43] have proven the in vitro antimicrobial efficacy of riboflavin/UV-A (365 nm) combination for bacterial and fungal isolates. In all published studies, there was a rapid decrease in pain and the corneal re-epithelialization process was accelerated following CXL. In a pertinent published case series of five patients with infectious keratitis and corneal melting, Iseli and coworkers [45] employed CXL surgery after topical and systemic antibiotic treatments had failed. Encouragingly in all cases, corneal melting ceased and emergency corneal transplantation became unnecessary [45].

It goes without saying that CXL should not be seen as the procedure of choice for infectious keratitis and should only be applied with caution as it may have toxic effects on these susceptible diseased corneas. Moreover, not all pathogens will respond positively to CXL treatment and this especially applies in the case of herpes simplex virus because the use of UV light may act as a stimulus for virus replication, exacerbating the infection and potentially leading to corneal perforation [46, 47].

Ulcerative Keratitis

CXL seems to have an anti-edematous effect on the cornea and, therefore, it has been successfully applied in cases of bullous keratopathy [48]. Two reports by Kozobolis and coworkers [49] and Ehlers and coworkers [50] have presented the results after CXL in patients with combined ulcerative keratitis and bullous keratopathy that was unresponsive to conventional treatment regimens. In both reports, the patients’ ulcer, visual acuity and corneal edema were significantly improved.

Recently, CXL has also been investigated for modifying donor tissue prior to keratoplasty [51] and as an adjunct to orthokeratology [52]. Finally, the use of CXL for prophylaxis in patients whose corneas are deemed to be at a high risk for developing corneal ectasia after LASIK surgery for myopia [53] has been proposed.

CXL Plus

The term ‘CXL plus’ was introduced in 2011 and refers to several combined procedures aimed at enhancing the success of CXL [54]. It is well documented that when performed on its own, the CXL procedure is not intended to improve vision. However, at our disposal, there are now additional interventions to the original CXL protocol that can improve visual acuity and thus optimize the final surgical outcome. To date, controlled clinical evidence exists for the use of several complementary steps to the CXL procedure:

  • Transepithelial phototherapeutic keratectomy (t-PTK) [55 – 57]. Kymionis and coworkers [56] proved that epithelial removal using t-PTK (Cretan protocol) during CXL resulted in better visual and refractive outcomes in comparison with mechanical epithelial debridement.

  • Topography-guided and other forms of photorefractive keratectomy (PRK) [58 – 71]. The use of topography-guided PRK–CXL in post-LASIK ectasias, such as the Athens protocol described by Kanellopoulos and coworkers [68], has been successfully applied.

  • Corneal implants, also known as intracorneal ring segments [72 – 83]

  • Phakic intraocular lens implantation [84 – 89]. Labiris and coworkers [90, 91] investigated the effect of keratoconus, CXL and CXL combined with topography-guided photorefractive keratectomy (t-PRK) on self-reported quality of life (QOL) by means of the 25-item National Eye Institute Visual Function Questionnaire (NEI-VFQ 25) and concluded that keratoconus has a significant impact on patients’ QOL, even in its early stages, with functional best-spectacle-corrected visual acuity. Moreover, CXL, and especially CXL combined with t-PRK, appeared to exert a beneficial impact on self-reported QOL.

CXL Extra

In an attempt to accelerate the time required for CXL treatment utilizing the Dresden standard protocol (usually 1 h of surgical time), investigators have explored two different research avenues: riboflavin application by iontophoresis aiming at rapid stromal saturation, and the use of high fluence irradiation of UV-A light [92, 93].

The Dresden protocol relies on the application of UV-A light (365 nm) at the intensity of 3 mW/cm2 for 30 min, delivering a total of 5.4 J/cm2 energy onto the cornea [1]. In accordance with the Bunsen–Roscoe photochemical law of reciprocity, if the intensity and time change while the total energy is maintained, the effects of any photochemical reaction (in the current context, the CXL procedure) are similar. This implies that the total energy delivered and amount of cross-linkage induced in a standard CXL session should be similar to irradiation at 9 mW/cm2 for 10 min, 15 mW/cm2 for 6 min, and 30 mW/cm2 for 3 min, with all ultimately delivering the same energy (5.4 J/cm2) [93]. These new treatment protocols are referred to as accelerated CXL or ‘CXL extra’.

The main concerns for accelerated CXL are its repercussions on the safety of the procedure, given that despite a similar total energy being applied in CXL extra, the intensity of irradiation is higher and may have a harmful effect on the corneal endothelium. Initial accounts of CXL extra, however, report results comparable to those obtained with the standard Dresden protocol [94–96]. Epithelial healing occurs uneventfully and there are no detectable alterations in endothelial cell density as documented by confocal microscopy. In contrast, Cingü and coworkers [97] reported transient corneal endothelial changes following accelerated CXL (18 mW/cm2 for 5 min) for the treatment of progressive keratoconus in a case–control study. In this investigation, which employed corneal specular microscopy, a decrease in endothelial cell density was observed postoperatively at 1 month, which returned to pre-operative values after 6 months of follow-up.

Overall, accelerated CXL protocols seem to be a promising alternative in minimizing the duration of the treatment and lessening patient discomfort. Future large, controlled studies are needed to confirm the immediate and long-term safety of the procedure.

Conclusion

CXL has marked a new, less invasive era in the management of corneal ectatic disorders. Since the first pilot studies over a decade ago, many modifications and several improvements to the original protocol have been successfully carried out. These steps maximize the cross-linking effect and by doing so, halt progression and postpone or even avoid the need for corneal transplantation, as well as improving functional vision in patients with ectasias. The CXL plus and CXL extra protocols may represent the future of this procedure, but more research is needed before these steps are widely adopted. The clinical utility of CXL has already been well demonstrated even though there is, to date, no adequate knowledge regarding long-term, unforeseen consequences. Future research will further elucidate and consolidate the place of CXL among the most innovative surgical therapies in ophthalmology.