Keratoconus is a progressive corneal disease in which an ongoing loss of stromal tissue leads to irregular astigmatism and reduced quality of vision.1 In recent years, corneal collagen crosslinking (CXL) has become an established treatment modality designed to increase the mechanical and biochemical strength of the stromal tissue.2 The effectiveness of CXL stems from its ability to stabilize keratoconus and its effects on corneal curvature.3 Specifically, CXL flattens the cone, which in turn increases uncorrected (UDVA) and corrected (CDVA) distance visual acuity.4–6 This flattening can persist for several years or longer.7 Factors that can potentially predict treatment outcomes following CXL have been studied extensively.5
Corneal CXL is considered to be a safe, effective, and predictable treatment for the prevention of keratoconus progression. With respect to safety, adverse events occur in a minority of cases, with only a small risk for severe keratitis.8 A transient demarcation line or subepithelial haze has been reported after CXL, although these are rarely observed 1 year after treatment. High visual acuity (CDVA >20/25) is not usually regarded as an exclusion criterion for performing corneal CXL.2 This fact, combined with the favorable safety profile and increased availability of CXL, has led to an increase in the number of patients with high visual acuity who receive this treatment. The archetypal corneal curvature in keratoconus contributes to an increase in higher-order aberrations (HOAs) and a subsequent decreased CDVA. Overall, visual acuity is reported to increase after CXL.5 However, from a clinical perspective, it is important to assess whether acceptable levels of HOAs can be retained after CXL.
We examined the relationship between visual acuity, manifest refraction, and changes in HOAs 1 year after CXL performed to treat keratoconus. We also assessed whether HOA subtypes contribute independently to visual acuity outcomes or manifest refraction using multivariable modeling.
Patients and methods
Dataset and Study Design
Data were derived from an ongoing prospective treatment cohort of patients who had CXL for the treatment of progressive keratoconus. All consecutive patients who were treated at the University Medical Center Utrecht, Utrecht, the Netherlands, from January 2010 through April 2013 with 1 year of follow-up were included. The study of HOAs in this treatment cohort was approved by the institution’s Ethics Review Board, and the requirement for informed consent was waived.
The following inclusion criteria were applied: a progression of maximum keratometry (K) of more than 1.00 diopter (D) within 6 to 12 months and corneal thickness (at the thinnest point) of greater than 400 μm. The exclusion criteria included corneal scarring, the presence of a concurrent infection, pregnancy, and lactation.
Treatment effects were assessed at the 1-year follow-up visit. The detailed data collection and surgical procedure have been reported7 and were adapted for this study.
An epithelium-off procedure was performed following the Dresden protocol.9 The epithelium was removed with a blunt spatula. After isotonic riboflavin was instilled for 30 minutes, the cornea was exposed to a 3 mW/cm2 ultraviolet (UV) light source (UV-X, Peschke Meditrade GmbH) equipped with a perpendicular emission plane and with a wavelength of 365 nm ± 10 (SD); the total exposure time was 30 minutes.
Measurements included UDVA, CDVA, corneal tomography measured with Scheimpflug tomography (Pentacam HR type 70900, Oculus Optikgeräte GmbH), endothelial cell count with noncontact specular microscopy (SP-3000P, Topcon Europe Medical B.V.), and automated tonometry (CT-80, Topcon Europe Medical B.V.). If the tomogram failed to reach the 90% quality criterion, it was repeated up to 3 times and the best scan was used for calculation of HOAs. The CDVA was measured using manifest refraction taken by the same optometrist (N.S.). The measurements were repeated 1, 3, 6, and 12 months after CXL. All patients were asked to stop using contact lenses 2 weeks before each evaluation.
Assessment of Corneal Optical Aberrations
Corneal optical aberrations were calculated using the Pentacam Scheimpflug imaging software program based on the central 6.0 mm as determined by the corneal apex of anterior and posterior elevation maps obtained using Scheimpflug imaging. The software program reports corneal optical aberrations for the anterior and posterior surfaces as well as for the total cornea. Total cornea optical aberrations were chosen as the outcome parameter. The Scheimpflug tomography software then subdivides this outcome into the following 2 composite values: total corneal lower-order aberrations (LOAs) and total corneal HOAs. Normalized coefficients were used, expressed in microns of wavefront error (root mean square [RMS]), and labeled with International Organization for Standardization (ISO) standardized double-index Zernike symbols.10 The HOAs were reported with their Zernike weight coefficient because the polynomial coefficient is considered invariant. Total corneal HOAs were calculated based on the 3rd- to 8th-order aberrations. The following HOA subtypes were reported in detail: horizontal coma Z(3,1), vertical coma Z(3,−1), horizontal trefoil Z(3,3), vertical trefoil Z(3,−3), and spherical aberration Z(4,0).
Visual acuity was converted to logMAR notation. The UDVA and CDVA were used as outcome parameters. A 2-tailed paired-samples Student t test was used to determine the significance between HOAs at baseline and HOAs 1 year after CXL. In cases with missing data at the 1-year follow-up, the 6-month follow-up data were entered, if available (ie, the last measurement was carried forward). The baseline characteristics of the cases lost to follow-up were compared with all other cases in the cohort. Linearity of the baseline data and outcome measurements was determined visually in a scatterplot, normality was tested based on skewness, and kurtosis was based on a cutoff value of 3.29 (P < .001). Mutual correlations between the different HOA subsets were calculated.
Univariate analyses with changes in UDVA and CDVA as dependent variables were performed for all baseline parameters to aid in identifying potential confounders for the relationship between changes in HOAs and changes in UDVA and CDVA. The following factors were determined to be potential confounders: visual acuity at baseline and the LOAs defocus Z(0,2), horizontal astigmatism Z(2,2), and vertical astigmatism Z(2,−2). These factors were entered into the multivariable analysis. This analysis was performed using generalized estimating equations to correct for patients in whom both eyes were included in the dataset. Data collection and analyses were performed using SPSS software (version 21.0, International Business Machines Corp.). Patients who developed postoperative scarring, haze, or both were excluded from the HOA analysis because these might reflect a pathophysiologic mechanism other than a change in corneal curvature that affected visual acuity.
One hundred eighty-seven eyes of 162 patients were treated consecutively. Five of 187 eyes (2.6%; 3.1% of patients) were excluded from analysis because they were lost to follow-up, and 8 eyes (4.3%; 4.9% of patients) had the last follow-up measurement carried forward. The baseline characteristics of these 13 patients did not differ significantly from those in the main group; however, only patients with an affected right eye were lost to follow-up. Table 1 shows the patients’ baseline characteristics.
By the 1-year follow-up visit, the maximum K value had decreased or was unchanged in 164 of 187 eyes. In 16 eyes, the keratoconus progressed by more than 1.00 D, with a mean increase in maximum K of 2.6 ± 2.0 D (range 1.0 to 9.40 D). The improvement in UDVA and CDVA from baseline to the 1-year follow-up was statistically significantly (P = .002 and P < .001, respectively) (Table 1). At 1 year, the logMAR UDVA was better (mean improvement 0.25 ± 0.29) in 92 (49%) of 187 eyes, remained stable (within ±0.03) in 41 eyes (22%), and was worse (mean decrease 0.32 ± 0.24) in 54 eyes (29%). The logMAR CDVA was better (mean improvement 0.21 ± 0.23) in 120 (64%) of 187 eyes, remained stable (within ±0.03) in 32 eyes (17%), and was worse (mean decrease 0.20 ± 0.22) 35 eyes (19%).
The cylinder value obtained using manifest refraction increased significantly (mean increase 0.62 D) (P < .001), whereas the corneal astigmatism obtained using tomography remained virtually stable (mean −0.06 D) (P = .493). The endothelial cell density (ECD) was unchanged from baseline; the mean ECD at the 1-year follow-up was 2526 ± 366 cells/mm2, with no apparent clinical signs of endothelial dysfunction. At the 1-year follow-up, 16 eyes had a slight, albeit persistent, haze. The baseline characteristics in this subgroup did not differ significantly from those in the main group with the exception of CDVA, which was worse in the eyes with haze (0.52 logMAR versus 0.31 logMAR) (P = .026). In 14 of 16 eyes with persistent haze, preexisting striae were noted. The mean CDVA at follow-up was also significantly worse in this subgroup (0.37 logMAR versus 0.17 logMAR) (P = .011). These 16 eyes were excluded from further HOA analysis. No patient in the cohort developed infectious keratitis.
Change in Higher-Order Aberrations
Table 2 shows the absolute values for optical aberrations at baseline and 1 year after CXL as well as the percentage change. Total LOAs significantly decreased after CXL (P < .001). However, total HOAs did not (P = .272), although the HOA subtype of spherical aberration did significantly decrease (P < .001). The effect size of this decrease was relatively small. Vertical coma HOAs contributed the most to the total corneal HOAs; however, this subtype did not change significantly after treatment. Univariate confounder analysis of CDVA identified baseline spherical refraction (P = .037, B = 0.12), maximum K (P = .004, B −0.009), baseline logMAR UDVA (P = .034, B = −0.102), and baseline logMAR CDVA (P < .000, B = 0.748) as being significantly associated with the dependent variable. Based on the effect size, only pretreatment CDVA was considered a relevant confounder. For UDVA, baseline logMAR UDVA was the only parameter significantly associated (P = .003, B = 0.257) and considered relevant. An analysis of mutual correlations for each HOA subtype showed a significant correlation for horizontal coma Z(1,3) and vertical coma Z(3,−1) (P = .009, ρ = −0.204), horizontal trefoil Z(3,3) and vertical trefoil Z(3,−3) (P = .006, ρ = 0.213), and vertical coma and vertical trefoil (P = .001, ρ = 0.264).
Table 3 shows the results of the multivariable analysis of CDVA and UDVA. The calculated effects of the potential confounders (visual acuity and LOAs at baseline) and the HOA subtypes are given for both determinants. No independent relationship between any HOA variable and change in CDVA was observed. The putative confounder CDVA at baseline was strongly related to the change in CDVA. An independent effect of the change in horizontal coma was observed on the change in UDVA (P = .003, B = −0.475), and again UDVA at baseline was strongly related to this change.
The principal aim of this study was to report on HOAs 1 year after corneal CXL performed to treat keratoconus and to determine whether variations in HOAs are independently associated with a change in CDVA. On average, with the exception of spherical HOAs, the HOAs were largely unchanged after treatment. Multivariable analysis found no independent effect of any HOA subtype on the change in CDVA after CXL. However, changes in horizontal coma were significantly and strongly associated with the postoperative change in UDVA. Strikingly, the measured corneal astigmatism did not change (mean 4.12 D versus 4.06 D); however, the manifest refraction increased and became more in agreement with the topographical cylinder (mean −3.15 D versus −3.77 D) (P < .001).
A major strength of this prospective study is the inclusion of a relatively large treatment cohort (187 eyes of 162 patients), with few cases lost to follow-up (approximately 3% of patients). The intervention was standardized in accordance with current protocols and did not change throughout the course of study. All patients had epithelium-off CXL with nonaccelerated UVA irradiation, and all refractions were measured by an optometrist experienced in keratoconus care. Moreover, the treatment outcomes (ie, improvement in keratometry, UDVA, and CDVA) are consistent with those in the recent published literature.7,11,12 Furthermore, we focused on the HOA subtypes that are most relevant to clinical practice (ie, coma, trefoil, and spherical aberration), and the effect of more complex forms of optical aberrations was assessed via the compound HOA variable. The Pentacam Scheimpflug tomography software program calculates the total corneal HOAs based on anterior and posterior elevation maps. We therefore chose to measure these composite HOAs because individual anterior and posterior outcomes are less relevant from a patient-oriented perspective.
On the other hand, several features of our study and analysis may have affected the results. First, we used the Pentacam Scheimpflug tomography software program, which calculates/expands optical aberrations, rather than using an aberrometer, which measures optical aberrations. A wavefront device was not used in this study, and we were unable to determine whole-eye HOAs. Furthermore, internal optical aberrations can potentially compensate for aberrations that are attributable to the anterior segment; however, a previous study13 reported that these internal optical aberrations are relatively unchanged after corneal treatment. Our study design could be considered suitable to detect changes in corneal HOAs after treatment rather than measuring whole-eye HOAs. The Pentacam Scheimpflug tomographer is considered a reliable instrument to assess corneal shape with good repeatability and reproducibility,14–17 although recent studies debate its reproducibility with regard to HOA assessment. A second consideration is that we excluded from our analysis cases with an apparent corneal haze. Corneal haze can, at least in principle, affect optical aberrations without changing the corneal curvature (or the resulting elevation maps). Although the Pentacam Scheimpflug device can perform densitometry measurements, these measurements are not used to calculate corneal HOAs.18 Corneal haze might have influenced the edge-detection software; however, this likely had little effect because all Scheimpflug images used in this study were of sufficient quality.
Previous reports of post-CXL HOAs point toward a general decrease in ocular HOAs. For example, Greenstein et al.19 found a significant decrease in corneal coma HOAs based on anterior and posterior elevation maps. The authors also found no significant correlation between HOAs and the change in visual acuity, although their analysis was based on 31 keratoconus eyes only. In 2009, Vinciguerra et al.20 reported a significant decrease in total ocular HOAs, coma HOAs, and spherical aberrations in 28 eyes. In a more recent study with a larger cohort (n = 92),6 the same group reported a decrease in total HOAs and coma HOAs, but not in spherical aberrations. They did not, however, examine the correlation between HOAs and treatment outcomes. The authors used absolute values to calculate the change in HOAs, thus accounting for shifts from negative HOAs to positive HOAs. Here, we chose to report the outcomes as they were supplied by the Scheimpflug tomographer. Analyses were performed based on absolute values and did not materially alter our findings (data not shown). Ghanem et al.21 reported 12-month and 24-month follow-up measurements in 42 eyes. Both coma and trefoil showed a solid decline, possibly resembling continued corneal flattening, although spherical aberration did not change materially from baseline. No correlations were found between changes in individual corneal aberrations and visual acuity after CXL. Baumeister et al.22 found no significant change in HOAs at the 6-month follow-up visit (n = 20). This finding is more consistent with our finding that, with the exception of spherical aberration, no relevant change in corneal HOAs was observed. We used ISO standard double-indexed Zernike polynomials in an effort to present our findings unambiguously.
Previous experimental research23,24 showed that the individual Zernike polynomials have a different impact on visual function; spherical aberration RMS error contributes more than coma, which in turn contributes more than trefoil. Our results do not agree with those findings because horizontal coma had the strongest relationship with changes in UDVA in our multivariable analysis. Naturally, keratoconus eyes have a different distribution of HOAs than healthy eyes, and decentered cones in particular might induce high amounts of coma.
The inconsistency in our data on changes in astigmatism obtained using manifest refraction and corneal tomography deserves attention. On average, manifest cylinder measurements increased whereas topographic-derived corneal astigmatism did not. This effect could partly be attributable to the inability to correct for HOAs using spectacles. A wrong amount of astigmatic correction can be measured when the cylinder axis is placed on top of the coma because the patient perceives a slight improvement. We hypothesize that increased visual acuity leads to improved quality of manifest refraction, whereas the better perception of coma partly translates to a higher manifest refraction. The discrepancy of the independent effect of horizontal coma in UDVA versus CDVA might reflect this. Without spectacle correction, horizontal coma is a strong independent factor in visual acuity; however, after a manifest refraction, this effect diminishes (on average). We hypothesize that horizontal coma is coincidentally corrected by increasing the cylinder power, meaning that it lost its independent effect on visual acuity. On the other hand, one can debate whether the Pentacam Scheimpflug device is the best tool to detect these subtleties in corneal tomography.
Determining the true effects of CXL requires separating many interrelated variables.5 The continuous flattening of the cone is a structural parameter that can affect HOAs, and the possible migration of the cone apex can result in reduced cone eccentricity.7,25 Changes in corneal collagen fibril composition and/or the development of corneal haze can exert effects on contrast sensitivity and HOAs.18 We therefore used a structured approach to identify potential confounders regarding the role of measured HOAs on changes in visual acuity, and we assessed the independent contribution of each HOA subtype on the treatment outcomes. Future studies should perform wavefront analyses in addition to Scheimpflug tomographer–calculated aberrations to better discern the effect of CXL on whole-eye HOAs.
In conclusion, on average, HOAs were essentially unchanged 1 year corneal CXL to treat progressive keratoconus when assessed using Scheimpflug imaging. Only changes in horizontal coma had a strong and independent effect on uncorrected visual acuity.
What Was Known
- Higher-order aberrations play an important role in the diminished quality of vision of keratoconus patients.
- The advent of CXL created a paradigm shift in the treatment of progressive keratoconus. A mild regression of K readings is observed after CXL treatment.
What This Paper Adds
- Outcomes in this large cohort (n = 187) counter previous observations of HOA changes after CXL. Contrary to clinical intuition, HOAs did not appear to change substantially 1 year after CXL.
- An independent effect of HOA changes on UDVA was identified for horizontal coma.
1. Rabinowitz YS. Keratoconus. Surv Ophthalmol. 42, 1998, p. 297-319, Available at: http://www.keratoconus.com/resources/Major+Review-Keratoconus.pdf
. Accessed May 5, 2016.
2. National Institute for Health and Care Excellence. Photochemical corneal collagen cross-linkage using riboflavin and ultraviolet A for keratoconus and keratectasia; NICE interventional procedure guidance. 2013, NICE, London, UK, p. 1-6, Available at: https://www.nice.org.uk/guidance/ipg466
. Accessed May 5, 2016.
3. Wollensak G, Spoerl E, Seiler T. Riboflavin/ultraviolet-A–induced collagen crosslinking for the treatment of keratoconus. Am J Ophthalmol. 135, 2003, p. 620-627, Available at: http://iogen.fi/files/2011/10/Wollensak_et_al_Keratoconus_2003.pdf
. Accessed May 5, 2016.
4. Greenstein SA, Fry KL, Hersh PS. Corneal topography indices after corneal collagen crosslinking for keratoconus and corneal ectasia: one-year results. J Cataract Refract Surg
5. Wisse RPL, Godefrooij DA, Soeters N, Imhof SM, van der Lelij A. A multivariate analysis and statistical model for predicting visual acuity and keratometry one year after cross-linking for keratoconus. Am J Ophthalmol
6. Vinciguerra R, Romano MR, Camesasca FI, Azzolini C, Trazza S, Morenghi E, Vinciguerra P. Corneal cross-linking as a treatment for keratoconus; four-year morphologic and clinical outcomes with respect to patient age. Ophthalmology
7. Wittig-Silva C, Chan E, Islam FMA, Wu T, Whiting M, Snibson GR. A randomized, controlled trial of corneal collagen cross-linking in progressive keratoconus; three-year results. Ophthalmology. 121, 2014, p. 812-821, Available at: http://www.aaojournal.org/article/S0161-6420(13)01004-X/pdf
. Accessed May 5, 2016.
8. Seiler TG, Schmidinger G, Fischinger I, Koller T, Seiler T. Komplikationen der Vernetzung der Hornhaut. Complications of corneal cross-linking, Ophthalmologe 2013;110:639-644.
9. Spoerl E, Mrochen M, Sliney D, Trokel S, Seiler T. Safety of UVA–riboflavin cross-linking of the cornea. Cornea
10. International Organization for Standardization (ISO)., 2008. Ophthalmic Optics and Instruments – Reporting Aberrations of the Human Eye, ISO, Geneva, Switzerland, (ISO 24157:2008).
11. Hersh PS, Greenstein SA, Fry KL. Corneal collagen crosslinking for keratoconus and corneal ectasia: one-year results. J Cataract Refract Surg. 37, 2011, p. 149-160, Available at: http://www.vision-institute.com/UserFiles/File/CXL%20results%20published2011.pdf
. Accessed May 5, 2016.
12. O’Brart DPS, Chan E, Samaras K, Patel P, Shah SP. A randomised, prospective study to investigate the efficacy of riboflavin/ultraviolet A (370 nm) corneal collagen cross-linkage to halt the progression of keratoconus. Br J Ophthalmol. 95, 2011, p. 1519-1524, Available at: http://bjo.bmj.com/content/95/11/1519.full.pdf?sid=a585cf27-23ec-4a1a-be48-cdf9597f5584
. Accessed May 5, 2016.
13. Lee JM, Lee DJ, Jung WJ, Park WC. Comparison between anterior corneal aberration and ocular aberration in laser refractive surgery. Korean J Ophthalmol. 22, 2008, p. 164-168, Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2629903/pdf/kjo-22-164.pdf
. Accessed May 5, 2016.
14. Emre S, Doganay S, Yologlu S. Evaluation of anterior segment parameters in keratoconic eyes measured with the Pentacam system. J Cataract Refract Surg
15. Miháltz K, Kovács I, Takács Á, Nagy ZZ. Evaluation of keratometric, pachymetric, and elevation parameters of keratoconic corneas with Pentacam. Cornea
16. De Sanctis U, Missolungi A, Mutani B, Richiardi L, Grignolo FM. Reproducibility and repeatability of central corneal thickness measurement in keratoconus using the rotating Scheimpflug camera and ultrasound pachymetry. Am J Ophthalmol
17. Bourges J-L, Alfonsi N, Laliberté J-F, Chagnon M, Renard G, Legeais J-M, Brunette I. Average 3-dimensional models for the comparison of Orbscan II and Pentacam pachymetry maps in normal corneas. Ophthalmology
18. Greenstein SA, Fry KL, Bhatt J, Hersh PS. Natural history of corneal haze after collagen crosslinking for keratoconus and corneal ectasia: Scheimpflug and biomicroscopic analysis. J Cataract Refract Surg. 36, 2010, p. 2105-2114, Available at: http://www.vision-institute.com/UserFiles/File/CXL%20Haze%20Published.pdf
. Accessed May 5, 2016.
19. Greenstein SA, Fry KL, Hersh MJ, Hersh PS. Higher-order aberrations after corneal collagen crosslinking for keratoconus and corneal ectasia. J Cataract Refract Surg
20. Vinciguerra P, Albè E, Trazza S, Rosetta P, Vinciguerra R, Seiler T, Epstein D. Refractive, topographic, tomographic, and aberrometric analysis of keratoconic eyes undergoing corneal cross-linking. Ophthalmology
21. Ghanem RC, Santhiago MR, Berti T, Netto MV, Ghanem VC. Topographic, corneal wavefront, and refractive outcomes 2 years after collagen crosslinking for progressive keratoconus. Cornea
22. Baumeister M, Klaproth OK, Gehmlich J, Bühren J, Kohnen T. Änderung des Wellenfrontfehlers der Hornhautvorderfläche nach Kollagenvernetzungsbehandlung (UV-Crosslinking) bei Keratokonus. Changes in corneal first-surface wavefront aberration after corneal collagen cross-linking in keratoconus, Klin Monatsbl Augenheilkd 2009;226:752-756.
23. Rocha KM, Vabre L, Harms F, Chateau N, Krueger RR. Effects of Zernike wavefront aberrations on visual acuity measured using electromagnetic adaptive optics technology. J Refract Surg
24. Applegate RA, Sarver EJ, Khemsara V. Are all aberrations equal? J Refract Surg
25. Tu KL, Aslanides IM. Orbscan II anterior elevation changes following corneal collagen cross-linking treatment for keratoconus. J Refract Surg