Long-Term Outcomes of Corneal Collagen CrossLinking: Results of 6-Years Follow-Up

Background: Our aim was to evaluate the long-term effects of conventional epithelium-off corneal crosslinking performed on patients with progressive keratoconus. Methods: A retrospective analysis was performed on 34 eyes of 34 patients with progressive keratoconus underwent conventional cross-linking using Pentacam HR Scheimpug camera. Visual acuity, spherical equivalent, keratometry, pachymetry and corneal topographic indices were analysed preoperatively and at 1, 3, and 6 years after the surgery. Results: Statistically signicant decrease was detected at 1 year in spherical equivalent (p = 0.022), with no signicant changes afterward (p =0.616). Uncorrected and best corrected distance visual acuity improved signicantly in the rst postoperative year (p =0.017 and p= 0.003, respectively), and remained stable with no signicant changes at further visits (p = 0.203 and p = 0.336, respectively). Signicant decrease of central corneal thickness and thinnest corneal thickness was observed at 1 year (both p <0.001), with additional thinning of thinnest corneal pachymetry up to 3 years (p = 0.01). Maximum keratometry and mean keratometry showed signicant, continuous improvement between all consecutive visits over the whole follow-up period (all p-values <0.05). Signicant improvement was found in the following 5 topographic indices at 1 year: ISV (p = 0.001), IVA (p = 0.028), KI (p = 0.002), CKI (p <0.001 and R min (p <0.001), with further improvement up to 3 years in ISV (p = 0.007), in CKI (p = 0.019) and in R min (p = 0.015). R min showed more improvement up to the end of follow-up (p <0.001). Conclusions: Conventional epithelium-off corneal cross-linking is effective in halting the progression of keratoconus and in regularization of the anterior corneal surface over a long-term

R min (p = 0.015). R min showed more improvement up to the end of follow-up (p <0.001).
Conclusions: Conventional epithelium-off corneal cross-linking is effective in halting the progression of keratoconus and in regularization of the anterior corneal surface over a long-term follow-up period up to 6 years.

Background
Cross-linking (CXL) is a minimal-invasive surgical therapy of ectatic corneal disorders such as keratoconus or postrefractive keratectasia. Results of the rst human clinical study reported by  presented conventional CXL as a new treatment possibility to stop the progression of keratoconus [1]. Afterwards, CXL became the primary option for surgical management of keratoconus, radically reducing the number of more invasive interventions such as penetrating keratoplasty [2,3].
A number of research groups have demonstrated that CXL signi cantly improves corneal biomechanical properties [4][5][6]. Increased corneal stiffness and improved resistance to enzymatic degradation have been reported in several in vitro studies [7][8][9][10][11]. This is believed to be achieved by combined use of photosensitizer ribo avin and UV-A light that induces photochemical covalent bonds between collagen bers. Development of cross-links strengthens intermolecular cohesion and modify the underlying pathophysiology of keratoconus.
Successful of standard epithelim-off CXL in slowing down or halting the deterioriation of keratoconus have been con rmed by various clinical studies [12][13][14][15][16]. Recent data supports that CXL is effective in the management of early and advanced keratoconus as well [12]. It has been proven that CXL causes signi cant changes on both anterior and posterior corneal surface [17,18]. Flattening of the cornea, reduction of astigmatic power and wavefront aberrations have been reported [19,20]. Uncorrected and best corrected visual acuity also seemed to be improved compared to untreated keratoconic eyes [21,22].
Currently, rotating Scheimp ug photography is the most prevalent imaging technique in the diagnosis, staging and follow-up of patients with keratoconus. This technique allows rapid and noncontact measurement of the anterior segment parameters. Comprehensive assessment of both anterior and posterior corneal surface, thus, ectatic changes can be identi ed in early stages. Pentacam system provides evaluation of corneal topographic indices, which can re ect optical properties and pathological alterations in the curvature of the anterior shape.
However, several studies investigate corneal changes after CXL, most have reported results with a relatively short follow-up period within two years, thus, available data about long-term results is limited.
The aim of this study was to evaluate the long-term outcomes of conventional epi-off CXL for progressive keratoconus by using Scheimp ug topographic and tomographic parameters and analyzing clinical results.

Methods
This retrospective study was conducted on a group of 34 eyes from 34 patients with progressive keratoconus underwent conventional collagen cross-linking. All patients were examined and treated at the Department of Ophthalmology, Semmelweis University. The protocol of the study was approved by the Semmelweis University Regional and Institutional Committee of Sciences and Research Ethics, and was performed in accordance with the Declaration of Helsinki. Patients with corneal thickness below 400 µm, preoperative corneal scarring, prior keratitis, history of any eye injury were excluded.
Conventional epithelium-off corneal CXL was performed by the same surgeon (Z.Zs.N.) following the standard Dresden protocol [1]. The surgical treatment was performed under sterile conditions in the operating room. Under topical anesthesia with oxybuprocaine eye drops (Benoxi®, Unimed Pharma, Bratislava, Slovakia), the epithelial layer in a central diameter of 8 mm was manually removed. Then, 0.1% ribo avin droplets (Medio-Haus Medizinprodukte GmbH, Rostock, Germany) were instilled in every 2 minutes for 30 minutes before irradiation. The central cornea was exposed to UV-A light with a wavelength of 370 ± 5 nm (CSO Vega CMB X Linker, CSO Scandicci, Firenze, Italy) at an irradiance of 3 mW/cm 2 intensity for the next 30 minutes. During irradiation, instillation of ribo avin in every 2 minutes was continued. At the end of the treatment, topical antibiotic drops were administered and a bandage was applied. On the rst postoperative day, the bandage was removed. Antibiotic eyedrops (5 mg/ml levo oxacin) were prescribed to be instilled ve times per day for one week postoperatively. After complete corneal reepithelization, topical corticosteroid (1 mg/ml uorometholone) was administered four times daily within the following one month.
All patients were examined preoperatively and postoperatively at 1, 3 and 6 years after CXL procedure.
Statistical analysis was performed using IBM® SPSS® Statistics for Windows, version 25.0 (IBM Corp., Armonk, N.Y., USA). Normality of the variables was assessed using Kolmogorov-Smirnov test. Differences were assessed by paired t-test for parametric values, and by Wilcoxon rank-sum test for nonparametric data. To present the tendencies in evolution of measured data, preoperative to 1 year, 1 year to 3 years and 3 years to 6 years values were compared. P value of < 0.05 was considered statistically signi cant.
In Table 1, pre-, and postoperative outcomes of visual acuity and spherical equivalent are shown. Regarding SE, statistically signi cant decrease was detected at 1-year follow-up (p = 0.022), with no signi cant changes between further subsequent examinations. The mean change in SE was + 1.12 ± 2.68 D 6 years after CXL treatment. Both UCDVA and BCDVA improved signi cantly at 1 year after the treatment compared to baseline (p = 0.017 and p = 0.003, respectively). After the rst year, UCDVA and BCDVA remained stable with no signi cant changes between consecutive visits (Table 1). Overall, during the study period, 58.8% of treated eyes had at least 1-line gain both in UCDVA and in BCDVA (Fig. 1). The preoperative and postoperative data (mean ± SD) of visual acuity (uncorrected (UCDVA) and best corrected distance visual acuity (BCDVA) in logMAR), spherical equivalent (SE; in diopters (D)), pachymetry (central corneal thickness (CCT) and thinnest corneal thickness (ThCT) in µm) and keratometry (maximum (K max ) and mean keratometry (K mean ) in diopters (D)) is shown. P < 0.05 was considered statistically signi cant.  Table 2). None of the treated corneas showed increased keratometry readings or deteriorating topographic indices in our cohort after CXL.

Discussion
Over the past decade, treatment of corneal ectatic diseases has fundamentally changed due to the introduction of corneal cross-linking. A number of former studies presented the bene cial effects of CXL [20,24,25]. Based on these results, treating of keratoconus with standard epithelium-off cross-linking seems to be safe and effective surgical technique. However, most of the available research provides results within two or three years after treatment. Existing literure of longer outcomes of conventional CXL is still limited. The goal of this study was to assess visual outcomes and corneal changes 6 years after conventional corneal cross-linking in patients with progressive keratoconus.
In the present study, enrolled patients were followed by a relatively long period of time (6 years), allowing for a potentially valuable analysis of the outcomes of conventional cross-linking using UV-A light at 3 mW/cm 2 irradiace intensity with 370 ± 5 nm wavelength performed according to standard Dresden protocol.
In our study, SE, UCDVA and BCDVA improved signi cantly in the rst postoperative year and remained stabile without any further signi cant changes between the subsequent visits up to 6 years. Outcomes of our analysis con rmed the results of several long-term studies, which assumed that CXL improves visual acuity [26][27][28]. Although in these studies, values were compared between different timepoints and baseline examination, thus, it was not revealed whether there was any further signi cant change between consecutive visits. Based on detailed statistical analyis between each visits, both Shaheen et al. and O'Brart et al. stated that signi cant visual and refractive improvement can occur even after the rst postoperative year after CXL [29,30]. Imrovement in SE and visual acuity is a consequence of more regularized corneal surface, which is indicated by reduced keratometry and improved topographic indices in our analysis. However, further continous signi cant improvement was found in some of these latter parameters in our data, results suggest that the extent of changes in keratometry and topographic indices did not generate further improvement in refraction and visual acuity after the rst postoperative year in our cohort. Thus, according to our ndigs, nal visual and refractive ourcomes can be achieved until the end of the rst postoperative year. In some recent studies, authors found deterioration in a few percent of involved eyes [29,31]. Similarly, 2.9% of treated eyes had more than two-line-loss 6 years after CXL in our series. However, severe corneal scarring was not detected in any patients, cellular changes in the corneal stroma may induce increased scattering and changes in transparency, which might be related to visual loss. Although, the greater percent of treated eyes showed stabilization or improvement of visual acuity, the possibility of visual deterioration after treatment con rms that corneal cross-linking should only be performed on patients with documented progression of keratoconus as recommended in the literature for this treatment [32]. Further studies are warranted to determine the expectable changes in visual acuity and nd those patients who have increased risk for visual loss in long-term after CXL.
In our series, both central and thinnest corneal pachymetry showed the greatest reduction in the rst postoperative year. Measured at thinnest point of the cornea, pachymetry showed further thinning up to 3 years. Cellular changes after CXL, i.e. more compacted structure of collagen brils have an impact on pachymetry and may result in corneal thinning [33]. Meanwhile, ndings about the evolution of corneal thickness after cross-linking are controversial. Greenstein et al. found that corneal thickness decreases after treatment, but later it recovers to preoperative values [33]. In contrast, Vinciguerra et al. found signi cant decrease in corneal thickness after CXL, but no recovering tendency [21]. According to our results, pachymetry did not change after 3 years, suggesting that the structure of the corneal stroma is stabilized thereafter.
In agreement with the results of previous studies [24,30,34], signi cant decrease in keratometry readings was observed after the treatment in our analysis. Flattening effect of CXL was most pronounced in the rst postoperative year, although it was found to be continued even after the third postoperative year up to 6 year. Nicula et al. observed progressive decrease in maximum and average keratometry up to 10 years postoperatively [27]. In an other previous long-term study, similar tendency in improvement of keratometry readings was also obseved with signi cant decrease 1 year postoperatively, maintained up to 7 years [29]. The authors suggested that long-term topographic changes are not only due to CXL, but also increasing age-related corneal stiffness can modify corneal reshaping [29]. So far, the processes in uencing changes of the corneal structure after conventional CXL in long term are unclear, but it has been con rmed by our results as well that corneal curvature changes can occur even several years after therapy. This change in curvature should be taking into consideration even long time after CXL in patients who require contact lens tting. In the current series, none of the eyes had increasing keratometry values over follow-up period, con rming the long-term stabilization effect of CXL. and IHA during the same follow-up period [13]. According to the study of Toprak et al., ISV, CKI and R min improved after conventional cross-linking [36]. In spite of the different ndings, improvement of these indices mirrors that the corneal front surface become more optically reagular after CXL treatment. In addition, in this study, further improvement of ISV, CKI and R min was revealed up to 3 years. Moreover, the latter had additional improvement up to the end of the follow-up period, which is in consistent with our ndings of continued decrease of keratometry in long-term. Meanwhile, no signi cant changes were observed in the remained two indices, i.e. IHD and IHA, suggesting postoperative stabilization of these values. Topographic indices are elevated in keratoconic eyes (except for R min , which showes decreased values in ectatic disorders) [35]. Improvement of ISV means decrease in deviation of corneal radii from the median value, which represents a more regularized corneal surface. Improvement of IVA indicates more symmetrical shape of anterior corneal surface with respect to horizontal meridian. Decrease of KI and CKI is in strong relationship with decrease of keratometry, thus, represents the regularization of keratoconic topography and improvement of the asymmetry between central and peripheral parts. Finally, increase of R min is related to attening of the maximum steepness of the cone. Overall, comprehensive analysis of topographic indices suggests an optically more regular and symmetric corneal surface with attening of the cone. To the best of our knowledge, our study is the rst in the literature that presents data of topographic indices measured by Pentacam with the long follow-up period.
In the present study, we assessed refractive and visual outcomes, keratometry, pachymetry and topographic indices after conventional corneal cross-linking. However, our study has some main limitations. A prospective study with larger number of involved patients could provide more accurate results regarding data. Furthermore, detailed long-term evaluation also of the features of the posterior corneal surface and corneal densitometry would specify corneal changes after CXL more precisely. In addition, estimation of potential long-term visual outcomes and corneal attening based on preoperative parameters would provide useful information for clinicians in predicting postoperative outcomes, therefore further studies are warranted.

Conclusion
In brief, according to our ndings, it can be stated that conventional epithelium-off corneal cross-linking is able to prevent the progression of keratoconus, and regularization of the anterior corneal surface can be achieved by the treatment in long-term. Both UDVA and BCDVA, keratometry and topographic indices of ISV, IVA, KI, CKI and R min showed improving tendency after the treatment. Corneal reshaping is most pronounced in the rst postoperative year and the process may continue further up to 6 years postoperatively.

Declarations
Ethics approval and consent to participate This study was carried out with the approval of the Semmelweis University Regional and Institutional Committee of Sciences and Research Ethics, and followed the tenets of Declaration of Helsinki. This was a retrospective study, hence, informed consent was not required.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Figure 3 Changes in topographic indices. Figure 3 shows the evolution of ISV (index of surface variance), IVA (index of vertical asimmetry), KI (keratoconus index), CKI (central keratoconus index) and Rmin (radius of minimum curvature). Signi cant improvement was found all of these 5 indices in the rst postoperative

Competing interests
year. Improvement in ISV, CKI and Rmin was found up to 3 years. Rmin showed further increase 6 years after CXL.