Corneal changes in adult patients with progressive keratoconus after accelerated corneal cross-linking: prospective 6-month interim study results

INTRODUCTION

The cornea is an avascular and transparent structure of the eye. It consists of five separate layers: epithelium, Bowman’s layer, stroma, Descemet’s membrane and endo- thelium. Each of its layers has a specific function and any change in their structure results in consequential corneal dis- orders [1].
Thinning of the cornea is a characteristic feature of ectatic disorders. The most common one is keratoconus (KC) [2]. A wide range of histopathological studies have revealed which abnormalities are usually found in the KC-affected corneal tissues. These include the enlargement of basal epithelial cells, their irregular arrangement and a significantly lowered cell density in comparison to normal corneas [3-5]. The ruptures in Bowman’s layer [6] and morphological folds of Descemet’s membrane [7] were also described as a characteristic feature of KC-affected corneas. The most apparent changes are found in the corneal stroma which is known to have a significantly decreased number of anterior lamellae [8, 9] and stromal keratocytes [4, 6, 10-12]. Findings in the corneal endothelium are controversial. While some studies suggest an increase in endothelial cell density [13], others show a significant decrease in moderate-to-severe KC cases [7, 10].
The corneal cross-linking (CXL) procedure was introduced to halt the progression of KC [14]. It improves the distribution of the collagen fibrils, increases their diameter and interfibrillar spacing in KC-affected corneas. Furthermore, the mean pro-teoglycan area becomes smaller after CXL is performed [15]. However, this procedure carries a risk of the development of corneal haze or scarring [16, 17].
Corneal changes after CXL can be evaluated using corneal imaging tools. Pentacam®HR (Oculus Inc., Wetzlar, Germany) is a noninvasive optical system that uses the Scheimpflug principle to assess the anterior segment of the eye. It provides com-plete information about the corneal thickness, topography and transparency [18, 19]. Specular microscopy (NIDEK, USA; Specular Microscope CEM-530) is another non-invasive corneal imaging tool that allows one to perform a quantitative analy-sis of corneal endothelium [20].
The purpose of this study was to objectively evaluate the impact of accelerated CXL (A-CXL)-induced changes on corneal transparency and endothelium.

MATERIAL AND METHODS

Patients and selection
We performed a prospective, longitudinal, non-randomized study. We included all patients with progressive KC (PKC) who underwent the A-CXL procedure in the period from 5 August 2020 to 22 March 2021 performed by 1 surgeon (U.R.) in the Department of Ophthalmology, Hospital of Lithuanian University of Health Sciences Kauno Klinikos (Kaunas, Lithuania) and healthy age-, sex- and eye-matched myopic controls. The study protocol was reviewed and approved by Kaunas Regional Biomedical Research Ethics Committee (No. BE-2-63, 23/04/2020). Inclusion criteria for the PKC group were age above 18 years and being diagnosed with PKC. A diagnosis of PKC was defined as an increase of the maximum K-value (Kmax) by at least 1 diopter (D), decrease of the thinnest corneal thickness (TCT) by at least 30 μm, an increase of cylinder by at least 1 D, an increase of manifest refraction spherical equivalent (MRSE) by at least 0.50 D or more in over a 12-month observation period [21]. The control group consisted of healthy myopic age-, sex- and eye-matched study participants. Informed written consent was obtained from all eligible study participants after a detailed explanation of the ongoing study in accordance with the Declaration of Helsinki. Exclusion criteria for both groups were history of ocular trauma, current systemic or ocular dis-ease and the refusal to sign the informed written consent. Patients in the PKC group were not included if there were any contraindications to perform the A-CXL procedure. Preoperative ophthalmologic examination All study participants and controls underwent a standard ophthalmic examination in an outpatient clinic performed by the same ophthalmologist (U.R.) in the Department of Ophthalmology, Hospital of Lithuanian University of Health Sciences Kauno Klinikos (Kaunas, Lithuania), including uncorrected distance visual acuity (UDVA) and best corrected visual acuity (BCVA) evaluation, IOP measurement (iCare TA01i, Icare Finland Oy, Vantaa, Finland), slit-lamp examination, Pentacam® analysis (corneal topography and densitometry), and specular microscopy for the evaluation of endothelium. The assessment of UDVA and BCVA was performed using logMAR units 4 meters from the participant under photopic conditions.
Surgical procedure
Only patients with PKC were scheduled to undergo a 10-minute “epithelium-off” A-CXL [22]. All procedures were per-formed in a single center (Department of Ophthalmology, Hospital of Lithuanian University of Health Sciences Kauno Klinikos) by the same surgeon (U.R.) under local anesthesia. First a lid speculum was inserted and the central 9 mm zone of corneal epithelium was debrided. Then an isotonic solution of 0.1% riboflavin (MEDIOCROSS® M, Avedro, Waltham, MS, USA) every 2 minutes for 30 minutes was instilled. The central corneal thickness of the treated eye was then measured with an ultrasonic pachymeter (Tomey Bio and Pachymeter AL-4000, Nuremberg, Germany) to ensure it has a minimum of 400 μm prior to UV irradiation. UVA light (375 nm) (UV-X™ 2000 Crosslinking System, Avedro, Waltham, MS, USA) was used to irradiate the cornea for 10 minutes at 9 mW/cm2 irradiance (5.4 J/cm2 total energy). During the irradiation, the riboflavin solution was administered every 2 minutes. At the end of each procedure, one drop of levofloxacin was admin-istered and a soft bandage contact lens was applied to the treated eye.
Postoperative care and follow-up ophthalmologic examinations
After each procedure patients with PKC received standard care with levofloxacin 0.5% (4 times daily for 2 weeks) and dexamethasone 0.1% (twice daily for the first 5 days and then tapering it down by 1 drop weekly) drops. Patients were also advised to use preservative-free artificial tears and vitamin C supplements. Soft bandage contact lenses were removed after full epithelialization of the cornea (5 days after the procedure). Subsequent postoperative outpatient follow-up visits were scheduled 1 (35 ±5 days), 3 (94 ±4 days) and 6 (189 ±20 days) months after A-CXL. Data about patients’ UDVA, BCVA, IOP, corneal topographic and densitometry values were recorded. Corneal densitometry (CD) was assessed at the anterior (120 μm front), central, posterior (60 μm rear) and total layers and their central 2 ( 0-2) mm, radial 2-6 ( 2-6) mm, radial 6-10 ( 6-10) mm and radial 10-12 ( 10-12) mm zones. Its values are expressed in grayscale units (GSUs), which define backward light scatter and range from 0 (maximum transparency) to 100 (completely opaque). Specular microscopy was also performed for each study participant to evaluate postoperative changes in endothelial cell density (ECD), the coefficient of variation (CV) and the number of hexagonal cells (6A) at each follow-up visit.
Statistical analysis
The SPSS Statistics (IBM) software platform (version 27.0) was used for the statistical analysis of the gathered data. De-scriptive statistics (mean, standard deviation [SD] and percent) were used to summarize the baseline data. The Kolmogo-rov-Smirnov Z test was used to assess the normal distribution. Comparisons of UDVA, BCVA, IOP, TCT, and K-values were performed against baseline at each follow-up point using non-parametric Mann-Whitney U test or Wilcoxon’s signed-rank tests, as appropriate. Repeated measures analyses of variance (RM-ANOVA) with Bonferroni adjusted post-hoc comparisons were performed to evaluate CD, as well as values of ECD, CV and 6A preoperatively and 1, 3 and 6 months after A-CXL. Multi-factor ANOVA with Bonferroni adjusted post-hoc comparisons was chosen to perform the assessment of the difference be-tween the CD and ECD, CV and 6A values of the baseline, 6-month postoperative PKC group and control group. For the assessment of the correlation of CD values with UDVA, BCVA, TCT, K-values, ECD, CV and 6A Pearson correlation coeffi-cients (r) were calculated. P-values less than 0.05 were considered statistically significant.

RESULTS

Patients’ demographics and baseline data of corneal parameters
The baseline data of study participants’ demographic and corneal parameters are shown in Table I. All patients completed a 6-month postoperative follow-up. The mean values of BCVA, IOP and TCT in the PKC group were significantly lower than in the control group. The mean preoperative corneal K-values (Ksteep, Kflat and Kmax) were statistically significantly higher in the PKC group compared with controls. Visual acuity and intraocular pressure Six-month follow-up postoperative results revealed a statistically significant improvement of both UDVA and BCVA com-pared with the preoperative data. In the PKC group, 7 out of 12 (58%) patients’ UDVA improved, for 3 out of 12 (25%) it remained stable, and for 2 patients (17%) it decreased by at least one line. Furthermore, 7 out of 12 (58%) patients’ BCVA improved, for 4 out of 12 (33%) it remained stable and for 1 patient (9%) it decreased by at least one line.
At the 6th postoperative month, in the PKC group the mean value of IOP was non-significantly lower (9.59 ±2.74 mmHg) compared with the mean preoperative (9.68 ±1.88 mmHg) results (p = 1.0).
Corneal topography and thickness
Figure 1 shows changes in the mean values of main keratometry readings in the PKC group after A-CXL. Kmax was statisti-cally significantly lower at the 3rd (p = 0.023) and 6th (p = 0.003) postoperative month compared to the baseline data, where-as Ksteep and Kflat values were statistically significantly higher after 1st (p = 0.008 and p = 0.002) postoperative month but be-came statistically significantly lower at the 6th (p = 0.009 and p = 0.034) postoperative month compared with the baseline results.
TCT in the PKC group was statistically significantly lower at the 6th postoperative month (437.25 ±28.38 μm) compared to the preoperative (454.42 ±30.39 μm) data (p = 0.004).
Corneal densitometry
The analysis of the mean CD values in the PKC group revealed that they were statistically significantly higher in the  0-2,  2-6 zones of the anterior, central and total corneal layers after each follow-up visit compared with the baseline results (Table II). They were statistically significantly higher in the  6-10 zone of anterior, central, posterior and total corneal layers only after the 1st postoperative month (all p < 0.05). In the posterior corneal layer, a significant increase of the mean CD val-ues was observed in the  0-2 zone but only at the 3rd postoperative month. At the 6th postoperative month, in the same corneal layer mean CD values were statistically significantly higher only in the  2-6 zone. After an entire follow-up period, we also found a statistically significant decrease of the mean CD values in the  2-6 and  6-10 zones of the anterior (p = 0.019 and p < 0.001), central (p = 0.003 and p = 0.003) and total (p = 0.008 and p < 0.001) corneal layers compared with the values at the 1st postoperative month.
The mean CD values of myopic corneas were statistically significantly lower only in the  0-2 and  2-6 zones of the anterior, central and total layers compared with PKC-affected corneas 6 months after A-CXL procedure (Figure 2). In the posterior corneal layer, the mean CD values of myopes were statistically significantly lower only in the  2-6 zone com-pared with baseline (p = 0.004) and 6-month post-A-CXL values (p < 0.001).
Endothelium
In the PKC group, at the 6th postoperative month there were no statistically significantly changes in ECD, CV and 6A values compared with the preoperative results (Table III). No significant difference was found after comparing mean ECD values be-fore the procedure, 6 months after A-CXL and controls (all p > 0.05). Nonetheless, there was a statistically significant differ-ence when comparing the 6-month postoperative mean values of CV and 6A with controls (p = 0.042 and p = 0.025, respec-tively).
Interrelationship between corneal densitometry, visual acuity and other corneal parameters
No significant association between CD values of anterior, central and total layers and UDVA, BCVA, TCT, K-values, ECD, CV and 6A was found at the 6th postoperative month (all p > 0.05).

DISCUSSION

The 60-minute period of corneal exposure during the standard CXL (S-CXL) procedure carries an increased risk of complications [23]. This led to the search for alternative treatment protocols which would produce equally effective and safe results but would allow surgeons to reduce the time spent in the operating theatre. Several A-CXL protocols have been proposed which follow the photochemical law of reciprocity – the same photochemical effect can be achieved with reduced illumination time and increased irradiation intensity [24]. Kobashi et al. [25] found that A-CXL produce equally effective and safe outcomes regarding Kmax, central corneal thickness, UDVA, corneal hysteresis, corneal resistance factor and endothelial cell density.
One of the most common early findings in CXL-treated corneas is transient corneal haze [26, 27]. Various hypotheses have previously been proposed to explain this phenomenon. Disarrangement in corneal morphology is considered as the most probable cause of corneal densitometry elevation [28-33]. Additionally, damage to keratocytes has been linked with CXL treatment and their transformation to myofibroblasts which are also thought to cause corneal stroma remodeling [34, 35]. Therefore, here we objectively evaluated the impact of CXL-induced changes on corneal transparency.
In this study, we used a 10-minute A-CXL protocol when a 10-minute irradiation at 9 mW/cm2 was performed while de-livering 5.4 J/cm2 of energy. A significant increase of the mean CD values was mainly found in the  0-2 mm and  2-6 mm zones of the anterior, central and total layers at all follow-up times. These findings are partly in line with other studies that evaluated changes in CD after CXL procedure performed following accelerated protocols. Alzahrani et al. [36] in their prospec-tive, cross-sectional study found a significant increase of mean CD values at the anterior and central layers of the corneal  0-2 mm and  2-6 mm zones in the adult group after a 3-month follow-up. However, after that point they started to de-crease and at the 6th postoperative month an increase found at that time point was not significant. Bohm et al. [37] per-formed a study where they evaluated changes in CD after performing a 4-minute UVA irradiation at 30 mW/cm2 using 5.4 J/cm2 of energy during A-CXL procedure. After a 3-month follow-up, they found an increase of CD values in all three cor-neal layers but it was significant only in the anterior (front 120 μm) portion. Interestingly, Shen et al. [38] found a significant decrease in CD values in central, posterior and total corneal layers after performing a 320-second the A-CXL at 45 mW/cm2 using 7.2 J/cm2 of energy. In contrast to the aforementioned studies, we found a significantly reduced corneal transparency in the 6-10 mm zone in all corneal layers but only after first postoperative month.
In our study, we also objectively evaluated A-CXL- induced endothelial changes after performing A-CXL. After a 6-month follow-up, we found that the mean of ECD in the PKC group decreased by 1.5%, 6A decreased by 4.8% and CV increased by 5.9%. Nonetheless, these changes were non-significant when comparing them with the preoperative results. After performing a 9-minute A-CXL at 10 mW/cm2 using 5.4 J/cm2 of energy, Badawi et al. [39] also found a non-significant decrease of 6A and increase of CV at the 6th postop-erative follow-up month. In contrast, a significant decrease of ECD was found. When comparing these results with controls, we found a significantly higher CV and lower 6A. Our findings are in accordance with a study by Cingu et al. [40] where they performed a 5-minute A-CXL at 18 mW/cm2 and found an increase of CV and decrease of 6A after comparing CXL-treated eyes with controls but they were non- significant.
There are two main limitations of this study. First, a follow-up period of 6 months was relatively short. Second, the sample size was small. Therefore, further research with larger sample size and longer follow-up is required.

CONCLUSIONS

CD values in the anterior and central corneal layers remain increased 6 months after a 10-minute A-CXL is performed, thus impacting the overall corneal transparency change. Furthermore, morphological modifications in corneal endothelial polymegethism and pleomorphism may occur following a 10-minute A-CXL procedure.

DISCLOSURE

The authors declare no conflict of interest.

References