Abstract
Purpose:
The multicenter studies reviewed were designed to evaluate the safety and efficacy of corneal collagen crosslinking (CXL) for the treatment of progressive keratoconus and corneal ectasia after laser refractive surgery. The results of these studies led to approval by the United States Food and Drug Agency for both conditions in 2016. This paper reviews these studies, as well as single-center substudies investigating other aspects of crosslinking outcomes.
Methods:
As part of prospective, randomized, controlled clinical trials, the treatment group received standard CXL, and the sham control group received only riboflavin ophthalmic solution. The primary efficacy criterion was maximum keratometry (Kmax) 1 year after CXL. Secondary outcomes were corrected distance visual acuity (CDVA) and uncorrected distance visual acuity (UDVA). Safety and adverse events were analyzed. In single-center substudies, corneal topography, ocular aberrations, corneal haze measurements, corneal thickness, corneal biomechanics, subjective visual function, and outcomes predictors were also investigated. This paper presents a general review of the design and outcomes of crosslinking in these studies.
Results:
In the crosslinking treatment group, Kmax flattened by 1.6 diopters (D) and 0.7 D in eyes with keratoconus and ectasia, respectively. In both studies, there was continued progression in the control group. The CDVA improved by an average of 5.7 logMAR letters (LL) in the keratoconus treatment group and by 5.0 LL in the ectasia group. In both studies, corneal haze was the most frequently reported crosslinking-related adverse finding. This was most prominent at 1 month and generally returned to baseline between 3 and 12 months. In general, corneal topography, ocular aberrations, and subjective visual function improved after crosslinking.
Conclusions:
In the US multicenter trials, CXL was shown to be safe and effective in stabilizing Kmax, CDVA, and UDVA in eyes with progressive keratoconus or corneal ectasia.
Translational Relevance:
Corneal crosslinking was originally developed in the laboratory at the University of Dresden in the late 1990s. The combination of ultraviolet-A light and riboflavin was found to be the most effective of a number of different modalities tested to increase the biomechanical strength of the cornea. The clinical study design for the US multicenter clinical trials of crosslinking demonstrated the safety and effectiveness of this technique for treatment of progressive keratoconus and corneal ectasia, bringing this important advancement to patients in the United States.
Patients were enrolled as part of two separate multicenter prospective, randomized, sham-controlled clinical trials—one to treat progressive keratoconus and the other to treat corneal ectasia after refractive surgery—conducted in support of an FDA New Drug Application (NDA no. 203324) for a corneal crosslinking (iLink; Glaukos, San Clemente, CA) with Photrexa Viscous (0.146% riboflavin ophthalmic solution with 20% dextran), Photrexa (0.146% riboflavin ophthalmic solution), and the KXL System (Glaukos). Randomization was computer generated, and both the patient and the investigator were aware of the randomly assigned group.
For both studies, inclusion criteria included patients 14 years of age or older, axial topography pattern consistent with keratoconus, maximum keratometry (Kmax) on corneal topography (Pentacam; Oculus, Wetzlar, Germany) ≥ 47.0 diopters (D), an inferior–superior difference greater than 1.5 D on topography mapping, corrected distance visual acuity (CDVA) worse than 20/20, and corneal thickness as measured on the Pentacam of ≥300 µm. In the keratoconus study, progressive keratoconus was defined as one or more of the following changes over a period of 24 months: an increase of 1.0 D or greater in the steepest keratometry measurement, an increase of 1.0 D or greater in manifest cylinder, or an increase of 0.5 D or greater in manifest refraction spherical equivalent. In the ectasia study, inclusion criteria required an axial topography pattern consistent with corneal ectasia (including relative inferior steepening with inferior–superior difference). Exclusion criteria included patients with a history of corneal surgery (except for the previous LASIK or photorefractive keratectomy [PRK] procedure), corneal pachymetry < 300 µm, and history of corneal disease that would interfere with healing after the procedure, such as chemical injury or delayed epithelial healing in the past. Patients pregnant or lactating during the study were excluded.
Patients were initially randomized into a treatment or control group. The treatment group received standard ultraviolet-A (UVA)–riboflavin CXL treatment, performed according to the methodology described by Wollensak et al.
11 Initially, a topical anesthetic agent was administered, and the central 9.0-mm epithelium was removed by mechanical debridement. Riboflavin ophthalmic solution in 20% dextran was then administered topically every 2 minutes for 30 minutes. Riboflavin absorption throughout the corneal stroma and anterior chamber flare was confirmed by slit-lamp examination.
Ultrasound pachymetry was performed, and if the cornea was thinner than 400 µm then hypotonic riboflavin ophthalmic solution without dextran was administered, one drop every 10 seconds for 2-minute sessions, after which ultrasound pachymetry was performed to ascertain that the stroma had swollen to 400 µm or greater. This was repeated in 2-minute sessions until adequate corneal thickness was obtained. The cornea was aligned and exposed to UVA 365-nm light for 30 minutes at an irradiance of 3.0 mW/cm2.
During UVA exposure, administration of the riboflavin/dextran solution was continued every 2 minutes. Postoperatively, antibiotic and corticosteroid drops were administered, a soft contact lens bandage was placed, and the eye was reexamined at the slit-lamp. The contact lens was removed after the epithelial defect had closed. Antibiotics and corticosteroid drops were continued four times daily for 1 week and 2 weeks, respectively. Patients had complete examinations at 1, 3, 6, and 12 months postoperatively.
The control group received the riboflavin ophthalmic solution in 20% dextran alone. In this group, the epithelium was not removed. Riboflavin was administered topically every 2 minutes for 30 minutes. Next, the cornea was exposed to a sham treatment in which the UVA light was not turned on, during which time riboflavin was administered topically every 2 minutes for an additional 30 minutes. Per the study protocols, the patients were allowed to cross over and receive full CXL treatment after the 3-month follow-up examination.
Randomization was generated by the sponsor and allocated to each study site in a numbered sequence of envelopes containing subject assignment. The primary efficacy endpoint was the difference between the CXL group and the control group. Because the control group was eligible to receive treatment after the 3-month visit, those eyes that subsequently received treatment were considered lost to follow-up.
As a result of this crossover of control eyes to the treatment group, only two control eyes in each of the studies were available for examination at 12 months. Therefore, a last observation carried forward (LOCF) method was used to impute missing data for the 12 month analysis of the control group. In the LOCF analysis, efficacy data prior to crossover were carried forward to month 12, the study endpoint. The significance of the individual outcome changes from preoperative to 1-year postoperative were presented as P values (P < 0.05).
The progressive keratoconus study included 205 eyes of 205 patients (102 in the treatment group, 103 in the control group). The ectasia study included 179 eyes of 179 patients (91 in the treatment group, 88 in the control group).
In the ectasia treatment and control groups, 166 eyes had previous LASIK, eight eyes had previous LASIK with a PRK enhancement, and five eyes had undergone previous PRK alone. The mean ages of patients enrolled in the keratoconus study and the ectasia study were 31.1 and 43.5 years, respectively.
In the keratoconus treatment group, preoperative uncorrected distance visual acuity (UDVA) and corrected distance visual acuity (CDVA) were 11.9 ± 12.2 logMar letters (LL) and 33.2 ± 13.4 LL, respectively. In the keratoconus control group, preoperative UDVA and CDVA were 8.2 ± 11.0 LL and 32.8 ± 13.6 LL, respectively. In the treatment group, UDVA improved by 4.4 LL and CDVA improved by 5.7 LL 1 year after treatment. These changes were all statistically significant (P < 0.05). In the control group, there was a gain of 2.6 LL of UDVA and 2.2 LL of CDVA; however, these changes failed to reach statistical significance. Regarding these changes in CDVA 1 year after CXL, there was a statistically significant difference between the treatment and control group (P < 0.01). However, regarding changes in UDVA, there was no statistically significant difference between the treatment and control groups.
In the ectasia treatment group, preoperative UDVA and CDVA were 14.4 ± 13.5 LL and 37.0 ± 13.0 LL, respectively. In the ectasia treatment group, UDVA improved by 4.5 LL (
P < 0.001); in the control group, there was a decrease of 0.1 LL of UDVA (
P > 0.05). CDVA improved by 5.0 LL in the treatment group (
P < 0.001) and improved by 0.3 LL in the control group (
P > 0.05). These differences between the treatment and control groups were statistically significant (
P < 0.001). Data stratified to individual eyes are shown in
Figure 2.
Traditionally, patients with keratoconus and ectasia after laser refractive surgery have been treated with rigid gas-permeable contact lenses and, in more severe cases, penetrating keratoplasty.
26–29 Corneal crosslinking, which, uniquely, arrests the progression of these diseases, has been one of the most important advances in their treatment.
23–25,30 Early international studies showed efficacy in reducing the progression of keratoconus and also showed that crosslinking can have beneficial visual and topographic outcomes.
31–34
Corneal crosslinking was approved in 2016 by the FDA as a drug and device combination (Glaukos) for the treatment of progressive keratoconus and corneal ectasia after laser refractive surgery. The two prospective randomized clinical controlled trials that formed the clinical basis for FDA approval were two of the largest studies performed on this treatment and compared a treatment group with a sham control group.
24,25
With regard to study analysis, because the control group was eligible to receive treatment after the 3-month visit, many control eyes did not have data available subsequent to that visit. To account for this, a LOCF analysis was used to impute missing data for the 12-month analysis of treatment versus control. The justification for this methodology relies on the fact that keratoconus and ectasia are progressive conditions without spontaneous remission or improvement. Thus, untreated eyes would be expected to progress or, at best, remain stable per the natural history of the disease. Therefore, an LOCF model would seem to be a conservative methodology to compare the efficacy of CXL to control; the control data are imputed going forward as no further change, whereas disease progression, in fact, would be expected in the setting of progressive keratoconus and ectasia. Such progression would increase the difference between treatment and control as compared with the LOCF model. A further limitation of the control group is that the epithelium was not removed in these eyes; therefore, any contribution of de-epithelialization to outcomes could not be assessed.
In both US pivotal multicenter trials, CXL was found to stabilize keratoconus and corneal ectasia 1 year after treatment. Notwithstanding the differences of 2.6 D and 1.3 D between treatment and control at 12 months in the keratoconus and ectasia groups, respectively, as found using the LOCF model, looking at the treatment group alone, for which complete 1-year data are available, there was a 1.6-D average improvement in Kmax with keratoconus and a 0.7-D improvement in eyes with ectasia.
There was a mild, statistically significant, improvement of CDVA and UDVA in both treatment groups. There was also an improvement of many subjective visual outcomes; however, the influencing factors for these visual acuity improvements remain unclear. HOAs are a significant cause of visual impairment in keratoconus and ectasia; therefore, improvement of these aberrations might be expected to predict improvement in vision after CXL. In a single-center analysis
18 of HOAs, despite a significant improvement in total ocular and anterior corneal aberrations, there were no correlations between these aberrations and the improvements in UDVA or CDVA after CXL. Furthermore, despite improvement in four of seven Pentacam indices (ISV, IVA, KI, and R
min),
19 there were also no meaningful correlations with UDVA and CDVA.
In general, there appeared to be a less robust response to CXL in eyes treated for corneal ectasia after refractive surgery when compared with eyes with KC. At baseline, ectatic corneas were not as steep as those in the keratoconus trial. Additionally, the location of maximum keratometry tends to be more peripheral in eyes with ectasia. Studies have reported a more robust CXL response in eyes with a steeper and more central maximum keratometry, suggesting that the topographic attributes of the ectatic cornea may mitigate the CXL response.
21,35 Furthermore, biomechanical differences caused by the LASIK flap; possible differences in the riboflavin diffusion rate in a post-LASIK cornea, especially at the flap interface; and intrinsic pathophysiologic differences between keratoconus and ectasia may all contribute to the different CXL outcomes.
In addition to preoperative disease as a possible predictor of CXL outcomes, patients with worse CDVA (>20/40) and higher keratometry readings (K
max ≥ 55.0 D) in general were found to more likely to have improvement after CXL.
21 Notwithstanding the generally good outcomes, ophthalmologists should counsel patients about the risk of a possible loss of visual acuity postoperatively, particularly in eyes with a preoperative CDVA better than 20/40.
The clinical time course after crosslinking is similar for both keratoconus and cornea ectasia. There was a significant worsening of vision and steepening of the cone at 1 month postoperatively, little change at 3 months, and improvement thereafter. These postoperative outcomes appear to be related to both stromal and epithelial remodeling over time after crosslinking. Furthermore, this time course appears to be consistent with the postoperative thinning and crosslinking associated with corneal haze changes over time.
With regard to the safety of crosslinking, early in the postoperative course patients are subject to the typical complications of the epithelial wound healing process. In the US multicenter trial for CXL in keratoconus patients, 22.5% of patients had a remaining epithelial defect, and, in the trial for ectasia, 26% of patients had a remaining epithelial defect 1 week after their procedure. After the initial epithelial wound healing, crosslinking-associated cornea haze was reported in over 50% of treated eyes.
Crosslinking-associated corneal haze differs in clinical character from haze after other procedures, such as excimer laser photorefractive keratectomy. The former is a dust-like change in the corneal stroma or a mid-stromal demarcation line,
36 whereas the latter has a more reticulated appearance. Corneal haze has been confirmed using confocal microscopy and can be objectively quantified using Scheimpflug densitometry.
37,38 Similar to the time course of clinical outcomes after crosslinking, there appears to be an increase in haze that peaks at 1 month and plateaus between 1 and 3 months. Between 3 and 6 months, the cornea begins to clear and continues to return toward baseline at 1 year.
Corneal thinning also occurs early in the CXL postoperative course.
3–6,14–17 Postoperatively, similar to the time course of crosslinking-associated corneal haze and crosslinking clinical outcomes, the cornea appears to thin at 1 month and at 3 months and to re-thicken between 3 and 12 months. The physiology of this initial thinning and subsequent rethickening is not entirely clear; however, epithelial remodeling is likely an early factor in corneal thickness changes. Furthermore, anatomic and structural changes of corneal collagen fibrils, such as compression of collagen fibrils,
10,23 changes in corneal hydration
24 and edema,
25,26 keratocyte apoptosis,
13,27,28 changes in glycosaminoglycans,
29 and other processes, might be implicated in the distinct clinical time course after CXL.
Corneal crosslinking is an essential treatment to stabilize and even improve the visual acuity and topography of patients with keratoconus and post-refractive surgery ectasia. This was confirmed in the prospective randomized controlled US pivotal multicenter trials for patients with keratoconus and corneal ectasia after refractive surgery, which led to FDA approval. International long-term studies have reported continued stability in a majority of patients 7 to 10 years after CXL.
39–43 Further follow-up is required to determine the long-term stability of patients treated in the US multicenter trials. Such studies are ongoing.
Supported in part by an unrestricted grant to the Department of Ophthalmology, Rutgers - New Jersey Medical School, Newark, New Jersey, from Research to Prevent Blindness, New York, New York.
Disclosure: S.A. Greenstein, None; P.S. Hersh, Medical monitor of the FDA CXL trials (Avedro, MA)