Corneal Collagen Cross-linking

A Review of Clinical Applications

Abstract

Corneal collagen cross-linking (CXL) has been shown to slow down or stop the progression of keratoconus. In addition, CXL has been applied in cases of corneal ectasia. Recent reports of the use of CXL in cases of infectious keratitis have generated further interest in this treatment modality. This review discusses the principle, clinical uses, and complications associated with CXL.

Corneal collagen cross-linking (CXL) is used to increase the rigidity of the cornea and its structural integrity by a photopolymerization process that induces intrafibrillar and interfibrillar collagen cross-links. It has been used for the management of corneal ectasias, such as keratoconus and post–laser-assisted in situ keratomileusis (LASIK) ectasia. In this review, we discuss the principle and procedure of CXL, physiochemical changes in the cornea induced by CXL, clinical uses of CXL, and complications and contraindications of CXL.

Principle of Corneal Collagen Cross-linking

The current technique for CXL involves the use of riboflavin (vitamin B₂), which is exposed to a source of UV-A light with a wavelength of 370 nm. In the process of CXL, riboflavin acts as a photosensitizer and absorbs ultraviolet (UV) radiation. Free radicals are produced in the photosensitizing process, which catalyze a reaction that leads to the formation of covalent bonds between the collagen molecules and microfibrils.^1,2 With the standard irradiance of 3 mW/cm², apoptosis of keratocytes was present up to a depth of 300 μm.³

Surgical Procedure of Corneal Collagen Cross-linking

The treatment protocol for CXL is based on previous laboratory work to maximize the CXL effect while minimizing the damage to ocular tissue.^1,4,5 Nine millimeters of the central corneal epithelium is scraped off under topical anesthesia. Ultrasound-based pachymetry is often used to measure the baseline corneal thickness. Iso-osmolar riboflavin solution 0.1% in 20% dextran is applied on the corneal surface every 2 to 3 minutes for 30 minutes. The corneal pachymetry is then measured at various points to ensure that the thinnest point of the stroma is not less than 400 μm. The limbus of the cornea can be protected by placing a sponge ring to prevent UV light from passing beyond the outer periphery of the cornea, which thereby prevents any potential damage to the limbal stem cells.

The photoactivation of riboflavin within the cornea starts with the UV-A light illumination, which is calibrated at 365 nm to provide an irradiance of 3 mW/cm². This equals to a total dose of 5.4 J/cm². The light is usually located at a distance of 5 cm from the cornea. The light is kept in position for 30 minutes, and riboflavin solution is instilled every 3 minutes. Intraoperative corneal thickness can be measured with ultrasound pachymetry or optical coherence tomography.⁶ At the end of the irradiation process, a broad-spectrum antibiotic drop is instilled on the cornea, followed by the placement of a bandage contact lens. Topical corticosteroids can be used to reduce postoperative inflammation or corneal haze.⁷ The patient should be closely monitored in the postoperative period to ensure epithelial healing, after which the contact lens is removed.

Based on the Bunsen-Roscoe law of reciprocity, having a constant radiant exposure of 5.4 J/cm², accelerated CXL has been developed as an alternative treatment protocol with higher illumination intensity but shorter duration. The development of accelerated CXL has shortened the treatment duration tremendously. Biomechanical strength between the human corneas cross-linked with low-intensity (370 nm, 3 mW/cm² CXL for 30 minutes) and high-intensity (370 nm, 9 mW/cm² CXL for 10 minutes) protocols was found to be similar in in vitro experiments.⁸ There are several accelerated cross-linking systems available including the UV-X (IROC Innocross, Zurich, Switzerland), CCL-HE (Peschke Meditrade GmbH, Huenenberg, Switzerland), and KXL (Avedro Inc, Waltham, Mass). The Avedro KXL accelerated CXL system provides the shortest treatment time as it requires only a shorter riboflavin soaking time and UV-A illumination, typically up to 3 minutes. A few clinical studies also demonstrated the treatment efficacy and safety of these accelerated protocols,^9–12 although no general consensus on this form of treatment has been made.^13,14

Variations in Treatment Protocols of Corneal Collagen Cross-linking

Thin Cornea

It is not safe to perform the above standard procedure if the residual corneal thickness is less than 400 μm. In such cases, the de-epithelialized cornea can be swelled with hypo-osmolar 0.1% riboflavin for 30 minutes to increase the corneal thickness. Subsequently, the cornea is photoactivated using identical treatment parameters as described previously.

Transepithelial Corneal Collagen Cross-linking

Riboflavin does not readily penetrate the intact epithelium. Different techniques have been developed to increase the absorption of riboflavin into the stroma, including the use of eye drops containing preservatives, such as benzalkonium chloride preoperatively, to break the epithelial tight junctions or the creation of superficial epithelial trauma without complete epithelial debridement.¹⁵ Animal eye studies have not supported the use of incomplete manual epithelial removal.¹⁶ Wollensak and Iomdina¹⁷ showed that transepithelial CXL had one fifth of the CXL effect compared with conventional CXL with complete epithelial removal.

Chan and colleagues¹⁸ reported the first study describing CXL without epithelial debridement. In this study, the authors used intracorneal ring segments with or without CXL to treat keratoconus. After the placement of an inferior intracorneal ring, the cornea was soaked with riboflavin with diluted carboxymethylcellulose for 5 minutes followed by UV-A for 30 minutes. Improvements in manifest cylinder, average keratometry, and the steepest keratometry in the group with CXL were found.

Filippello et al¹⁹ used a 0.1% aqueous riboflavin involving trometamol and sodium ethylenediaminetetraacetic acid (EDTA) to break down epithelial intercellular junctions. In this study, anterior segment optical coherence tomography showed a dense line at 100 μm, whereas the demarcation line was seen at a depth of 320 to 340 μm in conventional CXL. This may suggest a more superficial effect of transepithelial CXL.

Mastropasqua et al²⁰ investigated the differences in riboflavin concentration in the anterior, intermediate, and posterior stroma after 3 different CXL imbibition techniques of 0.1% riboflavin, including the standard epithelium-off, epithelium-on, and iontophoresis-assisted administration. In this study, the authors found that CXL transepithelial iontophoresis imbibition produced greater and deeper riboflavin saturation with respect to conventional epithelium-on CXL while maintaining the advantages of avoiding epithelial removal, but did not reach concentrations obtained with standard epithelium-off CXL.

Physiochemical Changes of Cornea Induced by Corneal Collagen Cross-linking

Corneal collagen cross-linking has been shown to increase the Young modulus, resistance to bending, and shape retention.²¹ It also increased collagen fiber diameter and resistance to enzymatic digestion.^22,23 In addition, it was observed that CXL increased corneal surface temperature and shrinking temperature.^24,25

Clinical Uses of Corneal Collagen Cross-linking

Keratoconus

Keratoconus is a progressive, ectatic corneal disease that leads to irregular astigmatism due to weakening of the stromal collagen layers and subsequent stromal thinning.²⁶ Irregular astigmatism is the primary cause of vision loss in early and intermediate stages of the disease, whereas corneal scarring may compromise visual acuity in advanced stages.²⁷ The biomechanical strength of the cornea is reduced in cases of keratoconus. This may be due to the reduction of the number of intralamellar and interlamellar cross-links compared with normal controls.^28,29 Keratoconus spares corneas with an increased number of natural cross-links, such as in the elderly,³⁰ smokers,^31,32 and patients with diabetes.^33,34

Corneal collagen cross-linking was proposed as a treatment modality to stabilize weak corneas and prevent the progression of keratoconus, thereby preventing visual loss and significantly reducing the number of patients requiring surgical treatment. In a previous study, 23 eyes of 22 patients with moderate to severe progressive keratoconus were treated with CXL.¹ The length of follow-up was between 4 months and 3 years. The authors reported that 70% of cases had a reduction in maximal keratometry by 2.01 diopters (D) and in refractive error by 1.14 D. There was no change in corneal transparency, intraocular pressure, lens transparency, or corneal endothelial cell count. The progression of keratoconus was halted in all cases. The same author published a long-term review with a 5-year follow-up period.⁴ This study included 60 eyes of 48 patients with progressive keratoconus and showed an average reduction of maximum keratometry of 2.87 D and an improvement of visual acuity by 1.14 Snellen lines.

Wittig-Silva and colleagues³⁵ reported the results of CXL in 66 eyes of 49 patients with documented progression of keratoconus. Interim analysis of treated eyes showed a flattening of the steepest simulated keratometry value by 0.74 D at 3 months, 0.92 D at 6 months, and 1.45 D at 12 months. A trend toward an improvement in the best spectacle-corrected visual acuity was observed. In the control eyes, the mean maximum keratometry steepened by 0.60 D after 3 months, by 0.60 D after 6 months, and by 1.28 D after 12 months. Best spectacle-corrected visual acuity decreased by logMAR 0.003 over 3 months, 0.056 over 6 months, and 0.12 over 12 months.

Goldich and colleagues³⁶ reported their results among patients with progressive keratoconus who underwent CXL. After 2 years of follow-up, the authors found that these patients had stable uncorrected visual acuity, improved best-corrected visual acuity, and reduced keratometry. In this study, the corneal pachymetry, endothelial cell density, and foveal thickness were unchanged.

O’Brart and colleagues³⁷ published a randomized, fellow-eye controlled trial assessing the effect of CXL on the progression of keratoconus. In this study, compared with untreated eyes, there was an improvement in simulated keratometry, simulated astigmatism, cone apex power, and wavefront measurements (root mean square, coma, and pentafoil) after an 18-month follow-up period. The authors reported that none of the treated eyes progressed, whereas 3 of 22 untreated eyes showed progression.

Vinciguerra et al³⁸ published their 12-month data from a prospective nonrandomized clinical study on CXL in advanced keratoconus. The authors assessed 28 eyes that underwent CXL at 1-year follow-up. The study found an improvement in visual acuity and a decrease in minimal and maximal keratometry readings in the treated group compared with deterioration in the contralateral untreated eye. There was also a reduction of total and corneal higher-order aberrations in the eyes treated with CXL. No significant change in endothelial cell count after CXL was found in this study.

Caporossi et al³⁹ reported a 3.6-line increase in uncorrected visual acuity, a 1.66-line improvement in best spectacle-corrected visual acuity, a mean reduction in maximum keratometry of 2.1 D, and a 2.5-D reduction in manifest refraction spherical equivalent at 3 months after CXL in a series of 10 eyes in 10 patients with progressive keratoconus after CXL treatment. Later, the same authors published the results of an open, prospective, nonrandomized, phase 2 clinical trial of CXL on keratoconus.⁴⁰ In this study, 363 eyes with progressive keratoconus were treated with a standard CXL protocol. Of these, 48 eyes were followed up between 48 and 60 months. The authors reported stability or improvement of keratoconus in 44 cases (92%), with a mean reduction of average keratometry readings of 2 D and a significant improvement in visual acuity and higher-order aberrations. It was found that the mean best spectacle-corrected visual acuity improved by 1.9 Snellen lines, and the mean uncorrected visual acuity improved by 2.7 Snellen lines in the treated group. In contrast, the fellow untreated eye showed progression of keratoconus in more than 65% of the cases within 24 months. No significant change in endothelial cell counts and no adverse events were recorded.

Hafezi and colleagues⁴¹ published their study of CXL in patients with progressive keratoconus, where the corneas in question were thinner than the required 400 μm after epithelial removal. In this study, 20 patients with progressive keratoconus and keratectasia after refractive laser treatment were treated with a modified version of the standard protocol, where hypo-osmolar riboflavin was used to swell thin corneal stroma to greater than 400 μm before the application of UV-A light. The corneal thickness of all the patients after epithelial removal was at least 320 μm and was swollen to greater than 400 μm in each case before UV-A light application. All patients were followed up at 6 months, and the progression of ectasia was halted in all patients. Stabilization of keratometry was noted in 12 patients, whereas regression was noted in 8 patients. There were no changes in corneal endothelium or corneal clarity or adverse effects noted. The authors concluded that this modified CXL treatment for thin corneas appeared to be safe and effective up to 6 months. However, a longer follow-up is required to determine if the behavior of these thin corneas is comparable to corneas with normal thickness undergoing CXL.

Raiskup-Wolf et al⁴² reported the 7-year results of CXL in keratoconus. In this study, the authors found a decrease in maximum keratometry of 2.7 D at 1 year, 2.2 D at 2 years, and 4.8 D at 3 years. It was reported that visual acuity improved by 1 line per year in 54% of patients in the first 3 years. In addition, 2 patients had continued progression and had to undergo subsequent CXL procedures.

Corneal collagen cross-linking has also been performed in pediatric populations (up to 18 years of age) with progressive keratoconus.^43,44 Caporossi et al⁴³ reported a statistically significant improvement in uncorrected visual acuity and best spectacle-corrected visual acuity 48 months after CXL in patients (≤18 years) with progressive keratoconus. Vinciguerra and colleagues⁴⁴ also reported significant improvements in spherical equivalent (1.57 D), simulated keratometry in the flat meridian (from 46.32 to 45.30 D), and mean average corneal power (from 49.69 to 48.90 D) for patients younger than 18 years who had CXL for documented progressive keratoconus with a 2-year follow-up.

Corneal Collagen Cross-linking in Refractive Surgery–Induced Keratectasia

Several studies have assessed the use of CXL in post–refractive surgery corneal ectasia. Hafezi et al⁴⁵ performed CXL in 10 patients (1 eye per patient) with post-LASIK keratectasia. In this study, the authors found that CXL was able to arrest and partially reverse the progression of LASIK-induced iatrogenic keratectasia over a postoperative follow-up of up to 25 months. In a retrospective, interventional case series, Yildirim et al⁴⁶ reported the long-term results of combined same-day intrastromal corneal ring segment (ICRS) placement and CXL for postoperative LASIK ectasia. In this study, 16 eyes of 14 patients with postoperative ectasia after LASIK were treated with femtosecond laser-assisted ICRS implantation followed by same-day CXL. The patients were followed up between 36 and 62 months. The uncorrected distance visual acuity improved from logMAR 1.18 ± 0.42 to 0.44 ± 0.22 (P < 0.001), and the corrected distance visual acuity improved from logMAR 0.46 ± 0.26 to 0.21 ± 0.14 (P < 0.001). No serious complications were found in this study.

Vinciguerra et al⁴⁷ studied the outcomes of CXL in 13 eyes after refractive surgery. An improvement in best spectacle-corrected visual acuity by logMAR 0.1 and stabilization of keratometry after CXL were found. It was reported that corneal response was less marked compared with keratoconus.

Combined Corneal Collagen Cross-linking and Refractive Surgery

Combined CXL and photorefractive keratectomy (PRK) is performed to offer functional vision for patients with keratoconus. Kymionis and colleagues⁴⁸ performed customized topography-guided PRK immediately followed by CXL on 14 eyes of 12 patients with progressive keratoconus. In this study, the range of follow-up was between 3 and 16 months. The mean steepest keratometry was reduced from 48.20 ± 3.40 D preoperatively to 45.13 ± 1.80 D at last follow-up. Preoperatively, the mean spherical equivalent refraction was −3.03 ± 3.23 D, and defocus was 4.67 ± 3.29 D. At the last follow-up, the mean spherical equivalent refraction and defocus were statistically significantly reduced to −1.29 ± 2.05 D and 3.04 ± 2.53 D, respectively. Preoperative mean uncorrected visual acuity was logMAR 0.99 ± 0.81, and best spectacle-corrected visual acuity was 0.21 ± 0.19, which improved postoperatively to 0.16 ± 0.15 and 0.11 ± 0.15, respectively. Kanellopoulos⁴⁹ reported the efficacy of combined PRK and CXL among 325 eyes with keratoconus. In this study, the mean uncorrected visual acuity and mean best-corrected visual acuity were improved. However, the authors suggested caution because the long-term effects and safety of removing the Bowman layer with laser ablation in this group of patients have not been established. Coskunseven and colleagues⁵⁰ performed a randomized trial over a 6-month period comparing CXL followed by intracorneal ring segment implantation and CXL performed after ring segment implantation. In this study, the mean duration between treatments was 7 months. The authors reported that while uncorrected vision, visual acuity, mean spherical equivalent, cylinder, and mean keratometry values improved in both groups, the overall effect was greater when CXL was performed after ring segment implantation.

Corneal collagen cross-linking has been combined with both PRK and ICRS insertion as a 3-step procedure.⁵¹ In a prospective case series of 16 eyes with progressive keratoconus, all patients underwent topography-guided transepithelial PRK after intracorneal ring segment implantation, followed by CXL treatment. This study showed significant improvement in visual acuity, refraction, and keratometry within 6 months after completion of the 3-step procedure.

Apart from treating post-LASIK ectasia, CXL has been used prophylactically to prevent this rare complication.⁵² The additional cross-linking has been proposed to induce early stabilization of the cornea after LASIK and improve the refractive and keratometric predictability in highly myopic eyes.^53,54 However, the long-term needs and efficacy of ectasia prophylaxis with CXL are still lacking.

Corneal Collagen Cross-linking and Pellucid Marginal Degeneration

A few studies have evaluated the use of CXL in patients with pellucid marginal degeneration (PMD). Spadea⁵⁵ performed CXL in a patient with an inferiorly decentered treatment zone and found improvement in corrected distance visual acuity at 3 months and stabilization thereafter up to 12 months. Stojanovic et al⁵⁶ reported improvement in visual, refractive, and topographic outcomes in 5 eyes with PMD that underwent topography-guided surface ablation followed by CXL after 1 year. In addition, Kymionis et al⁵⁷ published a case report describing the successful outcome of intrastromal rings followed 12 months later by combined PRK and CXL in 1 patient with PMD.

Corneal Collagen Cross-linking and Infectious Keratitis

Photoactivation of riboflavin may be used to treat corneal infections because of its antimicrobial effects. Riboflavin has an affinity for nucleic acid, and its absorption of UV-A light can lead to the oxidation of guanine bases, which prevents the replication of the viral and bacterial genome.⁵⁸ The antimicrobial efficacy of the combination of riboflavin and UV-A light against common bacterial pathogens that cause infectious keratitis has been reported in vitro.^59–61

In a prospective study by Price et al,⁶² 40 patients with infectious keratitis were treated with topical antimicrobial agents together with CXL. The authors concluded that CXL worked best in keratitis cases without deep corneal involvement as its effects diminished with increasing depth of the corneal infiltrates. Makdoumi et al⁶³ excluded patients who had received antibiotics before presentation and investigated the use of CXL as the primary and sole treatment for patients with bacterial keratitis. Sixteen patients were recruited in total. After CXL, 15 patients had complete re-epithelialization of the infective corneal ulcer without additional intervention. Fourteen of the 16 patients did not require antibiotics at all throughout the course of their disease. The mean time from CXL to complete corneal healing was 7.1 days. None of these 16 patients experienced major complications, and keratoplasty was not required in any of them.

A prospective randomized controlled trial comparing standard antimicrobial treatment with standard antimicrobial treatment with corneal CXL was reported in 40 patients.⁶⁴ The mean duration to complete epithelialization of the corneal ulcer was 39.76 ± 18.22 days in the CXL group and 46.05 ± 27.44 days in the control group, although this difference did not reach a level of statistical significance.

In a retrospective comparative study analyzing the effectiveness of adjuvant CXL in 20 patients with fungal keratitis, resolution of infection was seen in 90% of the cases, although the addition of CXL did not affect the time to resolution or final visual outcome compared with eyes receiving medical therapy only.⁶⁵ Li et al⁶⁶ also applied CXL to 8 cases of keratomycosis resistant to topical antifungal treatment. Complete healing was noted in all cases within 3 to 8 days, and 6 cases had improved vision with no corneal transplantation required.

It is important to know that there is still no consensus on the treatment protocol of CXL in cases of infectious keratitis. The role of CXL in infectious keratitis remains unclear despite the encouraging results in some case reports and series.⁶⁷

Corneal Collagen Cross-linking and Cornea Edema

Corneal collagen cross-linking has been used for the treatment of corneal edema. This may be supported by changes in the hydration behavior of the porcine cornea after CXL⁶⁸ and stromal compaction after CXL.⁶⁹ Ehlers and Hjortdal⁷⁰ found a reduction in corneal thickness in 10 of 11 eyes treated with CXL, with the majority experiencing some improvement in vision. In addition, Wollensak and colleagues⁷¹ found thinning in 3 eyes with corneal edema after CXL. Because of lack of evidence, more studies to assess CXL for corneal edema are indicated.

Complications of Corneal Collagen Cross-linking

With a surge in the number of CXL procedures performed, there is an increase in the number of adverse events reported. The complications include stromal edema,^40,72–75 corneal haze,^40,76,77 sterile infiltrates,^78–82 endothelial irregularity or damage,^45,72,75 microbial keratitis,^83–85 corneal melting and perforation,^86,87 recurrent corneal erosion syndrome,⁸⁸ herpes simplex virus keratouveitis,⁸⁹ and endothelial damage.^45,75 Sterile corneal infiltrates are one of the most commonly seen complications.^78–81 The pathogenesis of sterile infiltrates is not known but may be the result of an altered immune response to antigens or due to phototoxic effects from CXL itself.⁹⁰ These infiltrates can be seen in up to 7.6% of cases. Resolution with an increase in topical steroid therapy has been reported.⁸²

Microbial keratitis has been reported with different organisms, including Pseudomonas,⁸³ microsporidia,⁹¹Acanthamoeba,⁸⁴ and Fusarium.⁸⁵ The presence of an epithelial defect, contact lens use, and the instillation of topical steroids may predispose to the development of microbial keratitis. In addition, CXL may also change the response of the cornea to injury and infection.⁹²

Treatment failure with a loss of 2 Snellen lines was reported in up to 3.5% of cases.^73,82 Koller and colleagues⁸² followed 117 eyes during the first postoperative year in a prospective trial. In this study, the risk factors for complications included age older than 35 years and corrected visual acuity better than 20/25. It was reported that 7.6% of eyes continued to have an increase in maximum keratometry reading of 1 D or more during the study period, which indicated the failure of CXL. It was reported that there was a significant effect of preoperative maximum keratometry of 58 D or more and female sex on increased failure rate.⁷³

Greenstein and colleagues⁷ studied the natural history of corneal haze after CXL. In this study, the authors measured the corneal densitometry and its correlation with slit lamp–detected haze. Maximum haze was reported at 1 month, which plateaued for 3 months, followed by a decrease until 1 year. Haze returned to baseline in the ectasia subgroup compared with the keratoconus subgroup, in which haze remained at 1 year but was not a predictor of clinical outcomes. Raiskup et al⁷⁷ reported that permanent stromal haze was found in 8.6% of eyes after CXL. Advanced keratoconus was associated with risk of haze with decreased postoperative visual acuity. In another study, it was reported that posterior stromal haze was found in 46.4% of eyes treated with combined PRK and CXL.⁹³ However, this was not associated with a decrease in corrected distance visual acuity.

Contraindications of Corneal Collagen Cross-linking

Corneal collagen cross-linking requires a minimum corneal thickness of 400 μm after the removal of the epithelium. Individuals with thin corneas (<400 μm) may not be appropriate candidates for the CXL procedure because of possible endothelial cell damage.^1,5,94,95 Corneal collagen cross-linking seems to be a safe treatment with a complication rate of approximately 1% if the patient is younger than 35 years of age and the best-corrected visual acuity is worse than 20/25.⁸² Also, the efficacy of CXL would likely increase if the treatment was limited to eyes with a maximum keratometry of less than 58.0 D. The goal of CXL is to stabilize or halt the progression of keratoconus with a small chance of visual improvement. Patients who have poor best-corrected visual acuity associated with extensive corneal scarring should not undergo CXL. Patients with prior incisional refractive surgery (radial keratotomy or astigmatic keratotomy) might not be suitable for CXL. It was reported that alterations within the corneal stroma after CXL, particularly the contraction of the collagen lamellae, can cause the keratotomy incisions to rupture.⁹⁶ Other contraindications include pregnancy, nursing, or systemic collagen vascular diseases because the safety and effects of CXL procedures in these populations have not been sufficiently assessed.

CONCLUSIONS

In the past decade, CXL has been investigated for its applications in ophthalmology. The results reported in most of the studies investigating its safety and efficacy in the management of keratoconus are promising. Although CXL is being used for other indications such as keratitis and corneal edema, proper clinical trials are awaited to support its efficacy for these indications.