To compare the clinical effects and safety of transepithelial corneal cross-linking (CXL) to epithelium-off (epi-off) CXL in progressive keratoconus.
Randomized clinical trial (noninferiority).
Patients received either transepithelial CXL with Ricrolin TE (n = 35) or epi-off CXL with isotonic riboflavin (n = 26) in 1 academic treatment center, using a simple unrestricted randomization procedure. The main outcome measure was clinical stabilization of keratoconus after 1 year, defined as a maximal keratometry (Kmax) increase <1 diopter (D).
Average Kmax was stable at all visits in the transepithelial group, while after epi-off CXL a significant flattening of 1.2–1.5 D was demonstrated from the 3-month follow-up onwards. The trend over time in Kmax flattening was significantly different between the groups ( P = .022). Eight eyes (23%) in the transepithelial group showed a Kmax increase of >1 D after 1 year (range 1.3–5.4 D) vs none in the epi-off group ( P = .017). There was significant different trend in corrected distance visual acuity (CDVA), with a more favorable outcome in the transepithelial group ( P = .023). In the transepithelial group, no complications occurred and in the epi-off group, 4 eyes (15%) developed complications owing to healing problems (sterile infiltrate, herpes keratitis, central haze, and stromal scar).
This study showed that although transepithelial CXL was a safe procedure without epithelial healing problems, 23% of cases showed a continued keratoconus progression after 1 year. Therefore, at this time, we do not recommend replacing epi-off CXL by transepithelial CXL for treatment of progressive keratoconus.
Progressive keratoconic corneas can be stabilized and strengthened by corneal cross-linking (CXL). The standard technique of CXL was first applied in 1998 and consists of an epithelial removal, after which riboflavin eye drops and ultraviolet-A (UVA) light are applied. The rationale for the removal of the epithelium was described as allowing adequate penetration of riboflavin into the stromal tissue, where it absorbs the UVA light and produces the actual cross-linking between collagen fibrils in the corneal stroma.
The downside of epithelial removal is that it causes significant pain and discomfort during the first postoperative days, in addition to the 3%–8% chance of epithelial healing problems. To circumvent these downsides of epithelium removal, a transepithelial CXL technique was developed. Transepithelial CXL avoids the need for epithelial removal. Wollensak and associates investigated the biomechanical effect in rabbit eyes and estimated that transepithelial CXL with benzalkonium chloride would create one fifth of the corneal biomechanical rigidity compared with epithelium-off (epi-off) CXL in human eyes. Transepithelial CXL with the use of sodium ethylenediaminetetraacetic acid (EDTA) in riboflavin (Ricrolin TE) has been investigated in ex vivo rabbit eyes as well, showing minimal riboflavin uptake in the group with intact epithelium receiving Ricrolin TE solution.
The clinical effects of transepithelial CXL with Ricrolin have been reported in case series and nonrandomized comparative trials. Filippello and associates reported clinical outcomes after 18 months in 20 eyes treated by Ricrolin-assisted transepithelial CXL, compared with their untreated fellow eye. A significant improvement in visual acuity (0.35 to 0.24 logMAR) and decreased central keratometry values (steepest keratometry [Ksteep]: 51.0 to 48.1 diopter [D]) were seen in the transepithelial CXL eyes, not in the untreated group. A stromal demarcation line at 60 μm depth was measured, indicative of an effective treatment. Caporossi and associates performed Ricrolin-assisted transepithelial CXL in 26 eyes, age 11–26 years, and reported unchanged visual acuities, but significantly increased maximal keratometry (Kmax) values (48.6 D to 50.1 D), after 2 years of follow-up. Leccisotti and associates reported the 1-year results on transepithelial CXL with Ricrolin TE in 51 eyes with the untreated fellow eye serving as control and found some stabilizing effect in the transepithelial CXL group (Kmax changed from 54.3 D to 54.8 D, compared to 51.7 to 53.3 in the control group). A prospective case series by De Bernardo and associates in 36 eyes treated by Ricrolin-assisted transepithelial CXL showed an increased visual acuity and stable keratometry after 6 months of follow-up.
The natural course of keratoconus can be long lasting, with years of apparent stable keratometry readings after a period of latent progression. Furthermore, the clinical effects of epi-off CXL have been well described in randomized controlled trials with adequate follow-up. To address these 2 considerations and adequately describe the clinical effects of transepithelial CXL, a noninferiority randomized study design is mandatory.
In this randomized controlled study, we investigated the clinical effects and safety of transepithelial CXL with Ricrolin compared to epi-off CXL in progressive cases of keratoconus and tested the hypothesis that transepithelial CXL is equally effective.
Materials and Methods
Study Group and Protocol
This noninferiority randomized clinical trial included patients diagnosed with progressive keratoconus who were found eligible for a CXL procedure at a tertiary academic center (University Medical Center Utrecht, The Netherlands), from May 30, 2011 through September 4, 2013, with a follow-up of 1 year. The study was prospectively approved by the University Medical Center Utrecht Ethics Review Board (REF number NL29961) and registered at ClinicalTrials.gov (identification number NCT02349165). All procedures complied with the Declaration of Helsinki and local laws regarding research on human subjects. Written informed consent was obtained from all patients prior to their participation.
Inclusion criteria were age ≥18 years, a clear central cornea, and a documented progression as defined by an increase in Kmax, Ksteep, mean keratometry, and/or topographic cylinder value by ≥0.5 D over the previous 6–12 months. Exclusion criteria were a minimal pachymetry of less than 400 μm prior to UVA irradiation, pregnancy or breastfeeding, and a history of previous ocular infection.
Keratoconus diagnosis and study eligibility were determined by 1 corneal specialist (N.T.). Progression of keratoconus was documented by minimally 2 topography measurements in all patients. Patients were randomized using a simple unrestricted randomization procedure to either transepithelial CXL or epi-off CXL.
Measurements and Devices
Patients were examined at baseline and at 1, 3, 6, and 12 months post CXL. Manifest refraction, visual acuity, Goldmann applanation tonometry, slit-lamp examination, and Scheimpflug topography (Pentacam HR; Oculus, Wetzlar, Germany) measurements were performed at each follow-up. Endothelial cell density (Topcon SP3000P microscope; Topcon, Tokyo, Japan) was measured at baseline and at the 6- and 12-month follow-up. Demarcation line depth was measured at the 1-, 3-, and 6-month follow-up using high-resolution corneal imaging (Visante Optical Coherence Tomography; Carl Zeiss, Jena, Germany). All contact lens wearers were instructed to discontinue contact lens wear at least 1 week (for scleral and soft contact lenses) or 2 weeks (for hybrid and rigid permeable lenses) prior to all evaluations.
During CXL, pachymetry measurements were performed with a handheld ultrasound (US) pachymeter (Handy Pachymeter, SP-3000; Tomey, Nagoya, Japan). The CXL device was used at a working distance of 5 cm with an irradiance of 3 mW/cm 2 (UV-X; Peschke Meditrade, Hünenberg, Switzerland). Before every treatment session, a calibration was performed to confirm the correct UVA emission level. Throughout the whole study, the same devices and time points were applied.
In the transepithelial CXL group, local anesthetic eye drops (oxybuprocaine 0.4% and tetracaine 1%) were applied 3 times during 5 minutes, and Ricrolin TE solution (consisting of riboflavin 0.1% eye drops with Dextran T500 15 mg and EDTA; SOOFT Italia) were instilled every 2 minutes for 15 minutes. Next, an eyelid speculum was placed and a silicone ring was positioned between the eyelids; the ring was filled with Ricrolin TE and used to retain a Ricrolin “pool” on the cornea. After 15 minutes, the silicone ring was removed, the cornea was rinsed with balanced salt solution, and pachymetry was performed. UVA irradiation was performed during 30 minutes, while Ricrolin TE solution was reapplied to the cornea every 5 minutes.
The epi-off CXL technique was performed following the Dresden protocol, adjusted with the avoidance of the eyelid speculum during riboflavin instillation. Epithelial removal (9-mm) was performed using a blunt knife. After pachymetry measurements, isotonic riboflavin 0.1% solution with 20% Dextran (Medio Cross) was applied every 3 minutes for 30 minutes, with no eye lid speculum in place. When pachymetry was <400 μm, hypo-osmolar riboflavin was additionally applied every 20 seconds for 5 minutes and repeated up to 2 times until the required pachymetry value of ≥400 μm was achieved. With an eyelid speculum in place, UVA irradiation was performed during 30 minutes, during which isotonic riboflavin drops were given every 5 minutes.
In both groups, the post-CXL medication consisted of antibiotic eye drops (Vigamox, 5 mg/mL; Alcon Nederland BV) and preservative-free artificial tears (Duratears Free, 2%; Alcon Nederland BV) and were used for 4 weeks, while nonsteroidal anti-inflammatory drops (Nevanac 0.1%; Alcon Nederland BV) were used during the first week. Starting 1 week after CXL, topical steroids (fluorometholone 0.1% drops; Allergan Nederland BV) were applied twice a day for 2 weeks. In the epi-off group only, oral pain medications (Tramadol 50 mg, 1–2 per day; diclofenac 25 mg, 1–2 per a day) were prescribed on the treatment day and the day after. A bandage lens (Purevision; Bausch & Lomb) was placed in the epi-off group and was removed after 1 week if the epithelial healing was complete.
Statistical Analysis and Power Calculation
Baseline measurements between the treatment groups were compared using an independent-samples t test. Primary outcome was predefined in the study protocol as clinical stabilization of keratoconus 1 year after CXL, defined as a Kmax increase of no more than 1 D over the preoperative Kmax value. Fisher exact test (2-tailed) was used to determine the relation between treatment and stabilization.
Decimal visual acuity was converted to the logarithm of the minimal angle of resolution (logMAR).
We analyzed all outcome measures at all follow-up visits using a linear mixed model with a generalized estimating equations correction. The outcomes over time were corrected for baseline values. Normality and homoscedasticity of the residuals were tested visually, and in a Q-Q plot and scatterplot, respectively. A P value <.05 was considered statistically significant. Data were recorded as mean ± standard deviation. All tests were performed in SPSS version 20.0 for Windows.
Power of this study was calculated based on a noninferiority design, which was determined by the expected average Kmax change after treatment (Raiskup-Wolf and associates: −1.46 D ) minus the acceptable average Kmax change after treatment (Koller and associates: Kmax + 1 D ). The standard deviation reported by Raiskup-Wolf and associates was 3.76. Using alpha 0.05, beta 0.2, and a noninferiority margin of −2.46, we calculated a sample size of 29 for each group.
Of the 105 patients eligible for this study, 61 patients were willing to participate and provided informed consent. This study included 61 eyes from 61 patients (47 male and 14 female) with progressive keratoconus, who were randomly assigned to either epi-off (n = 26) or transepithelial CXL (n = 35). One eye in the epi-off group received hypo-osmolar riboflavin, since the corneal thickness was <400 μm after 30 minutes of isotonic riboflavin instillation.
Four patients (6%), 2 in each group, did not complete the 1-year follow-up; 2 patients were lost to follow-up owing to a move abroad, 1 patient scheduled the follow-up visits in another hospital closer by, and 1 patient was retreated by epi-off CXL after 10 months (see the “treatment failure” section for details).
Both groups were comparable at baseline, apart from a lower spherical equivalent and logMAR uncorrected distance visual acuity (UDVA) in the transepithelial CXL group. Mean keratoconus progression before treatment was not significantly different between the groups. Baseline characteristics are listed in Table 1 . All variables, except for age, were normally distributed.
|Baseline Parameter||Transepithelial CXL||Epithelium-off CXL|
|Median age,y (range)||24 (18–48)||24 (18–44)|
|Spherical equivalent, D (mean ± SD)||−1.5 ± 2.5||−3.0 ± 3.0|
|Uncorrected distance visual acuity, logMAR (mean ± SD)||0.8 ± 0.5||1.1 ± 0.6|
|Corrected distance visual acuity, logMAR (mean ± SD)||0.3 ± 0.3||0.3 ± 0.3|
|Pachymetry thinnest point, μm (mean ± SD)||457 ± 27||467 ± 29|
|Maximal keratometry, D (mean ± SD)||56.4 ± 5.0||57.8 ± 7.1|
|Intraocular pressure, mm Hg (mean ± SD)||10 ± 2||11 ± 3|
|Endothelium, cells/mm 2 (mean ± SD)||2627 ± 363||2764 ± 252|
Table 2 shows the outcomes at all follow-up time points in the transepithelial CXL and the epi-off group.
|Parameter||Group||1 Month||3 Months||6 Months||12 Months||P Value a|
|ΔKmax (D)||Transepithelial||−0.1 ± 1.1||0.0 ± 1.0||−0.1 ± 1.2||0.3 ± 1.8||.022∗|
|Epithelium-off||0.3 ± 1.1||−1.2 ± 2.0||−1.4 ± 2.0||−1.5 ± 2.0|
|ΔCDVA (logMAR)||Transepithelial||−0.05 ± 0.24||−0.10 ± 0.21||−0.12 ± 0.22||−0.14 ± 0.21||.023∗|
|Epithelium-off||0.09 ± 0.18||−0.04 ± 0.18||−0.09 ± 0.23||−0.07 ± 0.21|
|ΔUDVA (logMAR)||Transepithelial||−0.06 ± 0.25||−0.08 ± 0.29||−0.02 ± 0.31||−0.06 ± 0.37||.591|
|Epithelium-off||−0.10 ± 0.36||−0.18 ± 0.31||−0.16 ± 0.35||−0.15 ± 0.43|
|ΔSE (D)||Transepithelial||0.4 ± 1.1||0.3 ± 1.1||0.3 ± 1.6||0.3 ± 1.6||.436|
|Epithelium-off||0.6 ± 1.4||0.5 ± 1.6||0.9 ± 1.8||0.4 ± 3.0|
|ΔCorneal thickness (μm) b||Transepithelial||0 ± 7||2 ± 9||−3 ± 8||0 ± 12||<.001∗|
|Epithelium-off||−18 ± 10||−14 ± 15||−9 ± 11||−4 ± 8|
a P value from generalized estimating equations corrected for baseline; ∗ = statistically significant.
Transepithelial CXL showed less potent effects on keratoconus stabilization and regression compared to epi-off CXL; in the transepithelial CXL group, Kmax remained virtually stable at all follow-up visits, while in the epi-off group Kmax demonstrated flattening from 3 months post treatment onward ( Figure 1 ). The trend over time in Kmax flattening was significantly different between both groups ( P = .022).
The steep and flat central keratometry values (Ksteep and Kflat) increased slightly over time in the transepithelial CXL group and decreased slightly in the epi-off group ( Supplemental Table , available at AJO.com ).
Primary Outcome: Treatment Failure, as Predefined in the Study Protocol
In the transepithelial CXL group, 8 of the 35 eyes (23%) showed continued progression of the disease (range 1.3–5.4 D). One eye showed a 4.7 D increase in Kmax after 10 months and was retreated by epi-off CXL; 7 other eyes showed a Kmax increase after 1 year, of which currently 4 eyes are retreated by epi-off CXL ( Table 3 ). In the epi-off group, all eyes demonstrated clinical stabilization after 1 year. This difference in clinical stabilization between the 2 treatments was statistically significant ( P = .016).
|Patient#||Maximal Keratometry (Kmax) Increase||Time After Initial Treatment||Result After Retreatment|
|1||4.7 diopter||10 months||Kmax decreased 1.6 diopter after 1 year|
|2||1.8 diopter||27 months||Kmax decreased 1.1 diopter after 1 year|
|3||2.9 diopter||15 months||Kmax decreased 0.2 diopter after 1 year|
|4||5.4 diopter||13 months||Kmax decreased 0.4 diopter after 1 month|
|5||4.6 diopter||33 months||No data available after retreatment|
The number of patients with a continued progression was considered too small for subgroup analysis to detect predictors for the transepithelial CXL outcome. The baseline characteristics of the eyes that presented with continued progression after transepithelial CXL compared to the transepithelial CXL group in general, or the total study population, were shown in Table 4 .
|Group||Kmax (D)||Range||CDVA (logMAR)||Range||CCThin (μm)||Range|
|Transepithelial CXL entire group||56.4||46.2–68.1||0.30||−0.08 to 1.00||457||410–516|
|Transepithelial CXL stable/regression||56.4||46.2–68.1||0.32||−0.08 to 1.00||456||410–516|
|Transepithelial CXL progression||55.8||50.7–59.6||0.21||0.00 to 0.52||460||424–495|
|Epithelium-off CXL entire group||57.8||47.2–73.8||0.26||−0.08 to 1.00||467||412–546|
Visual Acuity and Refraction
There was a statistically significant different trend in corrected distance visual acuity (CDVA) between both groups, with a more favorable outcome in the transepithelial CXL group ( P = .023). Figure 2 shows the largest difference in CDVA at the 1-month follow-up. When analyzing the data without the 1-month results, there is no significant difference between the 2 groups ( P = .088). No difference in the trend over time in uncorrected visual outcomes was observed between the groups ( P = .591).