Three-year outcomes of customized transepithelial corneal cross-linking with oxygen supplementation in keratoconus

Abstract

Background

Customized corneal cross-linking (CXL) has primarily been performed using the epithelium-off technique. Evidence on epithelium-preserving (transepithelial or epi-on) customized procedures remains limited, particularly with high oxygen concentrations (> 90%) beyond 1 year. We evaluated the 3-year clinical outcome and safety of customized transepithelial CXL (custom TE-CXL) with oxygen supplementation in patients with progressive keratoconus.

Methods

This retrospective study included 74 individuals (90 eyes) with progressive keratoconus who underwent custom TE-CXL with oxygen supplementation and were followed for at least 3 years. Outcomes included visual acuity, keratometric parameters, corneal thickness, and total corneal higher-order aberrations (HOAs). Examinations were performed at baseline and at 1, 2, and 3 years postoperatively. Longitudinal changes were analyzed using a linear mixed-effects model adjusted for baseline values.

Results

From 1 year onward, uncorrected visual acuity (UCVA) and maximum keratometry (Kmax) showed significant improvement compared with preoperative values. Least squares mean (LSMean) Kmax values decreased from 56.2 D at baseline to 54.9 D, 54.4 D, and 54.2 D at 1, 2, and 3 years, respectively (all P < 0.01). UCVA improved at 1 year and remained stable thereafter, whereas corrected distance visual acuity (CDVA) showed no significant changes. Total HOAs, including coma and spherical aberration, decreased progressively over time. Apical corneal thickness decreased gradually, and thinnest corneal thickness showed significant reductions from 1 year onward. No adverse events, such as infection or corneal haze, were observed during follow-up.

Conclusions

This descriptive, non-comparative analysis showed that custom TE-CXL with oxygen supplementation may be associated with stabilization of keratoconus over a 3-year period without observed adverse events. Further randomized controlled studies are warranted to evaluate its comparative effectiveness.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12886-026-05028-9.

Background

Keratoconus, characterized by thinning and forward protrusion of the cornea, is a non-inflammatory disease that typically develops in early adulthood [1]. As keratoconus progresses, corrected visual acuity declines and irregular corneal astigmatism may necessitate the use of rigid gas-permeable (RGP) contact lenses [1]. Historically, the management of keratoconus has been limited to contact lenses, including RGP or scleral lenses, and, in advanced cases, corneal transplantation [2]. The introduction of corneal cross-linking (CXL) into clinical practice in 2003 significantly changed the treatment landscape [3]. It is now considered the standard prophylactic treatment for slowing or halting keratoconus progression.

To optimize treatment efficiency and expand clinical applications, various CXL protocols have been developed, including the original Dresden protocol, accelerated CXL with reduced ultraviolet (UV) irradiation time and transepithelial CXL with an intact epithelium [2]. Among these, customized CXL (custom CXL) has emerged as an advanced technique featuring adjustment of UV irradiation energy based on corneal shape [4].This approach has been shown to result in greater corneal flattening and improved corneal regularity compared with standard, non-customized CXL [4–6]. Custom CXL also offers biomechanical advantages by targeting the structurally weaker cone region, an approach supported by the evidence that keratoconic eyes exhibit lower corneal biomechanical properties than normal eyes [3, 7, 8]. In a study using Brillouin microscopy, the corneal region near the apex was weaker in keratoconic eyes than in normal eyes, whereas the peripheral regions showed negligible differences [9]. By delivering higher UV energy to the cone apex, custom CXL induces a stronger collagen crosslinking effect in these weaker regions than standard CXL, making it a rational treatment option for keratoconus.

Oxygen concentration appears to be a key factor in the success of CXL as it is essential for the reaction that facilitates the cross-linking of corneal collagen fibrils [10]. Standard CXL has been reported to reduce oxygen concentration in tissues immediately after UV irradiation [11]. Ex vivo experiments have shown that CXL is less effective at low oxygen concentrations [12]. Based on these findings, a CXL system with oxygen supplementation was developed to enhance crosslinking efficacy by maintaining high oxygen concentrations [13].

To date, custom CXL has primarily been performed using the epithelium-off (epi-off) method [4–6, 14]. However, information on customized procedures for preserving the epithelium (transepithelial or epi-on) remains limited. Moreover, very few studies have reported the effectiveness of custom CXL with high oxygen concentrations (> 90%) for periods longer than 1 year [15].Therefore, we performed customized transepithelial CXL (custom TE-CXL) with oxygen supplementation and reported 3-year clinical outcomes and safety in a cohort of Japanese patients with progressive keratoconus who were followed for at least 3 years postoperatively.

Methods

Study population

This was a single-center, descriptive, non-comparative study conducted at the Nagoya Eye Clinic. We enrolled patients who underwent custom TE-CXL between October 2018 and April 2019 to for the management of keratoconus progression. Among them, only patients who could be followed up for more than 3 years postoperatively were included in the present study. As a result, we evaluated 90 eyes from 74 patients (mean age: 26.4 ± 10.1 years; 57 men, 17 women). Sixteen patients contributed both eyes, whereas 58 patients contributed a single eye (31 right eyes and 27 left eyes).The diagnostic criteria for keratoconus progression were defined as an increase in the mean keratometric (K) value by ≥ 1 D, steep K value by ≥ 1 D, or corneal astigmatism by ≥ 1 D within 2 years. Before examination, RGP contact lenses were discontinued for 2 weeks, and soft contact lenses for 1 week. The patients were examined preoperatively and at 1, 2, and 3 years after custom TE-CXL. At each follow-up visit, the examination included visual acuity assessment, endothelial cell count, corneal topography, and tomography.

The custom TE-CXL protocol

Anterior segment optical coherence tomography (AS-OCT) (CASIA2; TOMEY, Nagoya, Japan) was performed before surgery, and the UV irradiation pattern of the custom TE-CXL was determined based on these results, as previously described [3]. The irradiation pattern was determined using both the posterior elevation map and anterior keratometric map. Three concentric treatment zones were defined: the minimum diameter was set as 1 mm added to the shortest diameter of the area of abnormal posterior float (defined as > 20 μm elevation); the maximum diameter was determined to cover the steep anterior area identified on the keratometric map (≥ 5 μm step); and the intermediate diameter was set between these values. UV energy levels were 15, 10, and 7.2 J/cm² for the inner, intermediate, and outer zones, respectively. Irradiation was delivered at 30 mW/cm² in pulsed mode (1 s on/1 s off) via a Mosaic system (Glaukos). The customization algorithm is rule-based and can be applied at any center with AS-OCT capability for posterior elevation and anterior keratometric mapping, combined with a sector-programmable UV irradiation device.

To allow riboflavin penetration, the same concentration and protocol as standard transepithelial CXL were used. During UV irradiation, supplemental oxygen was delivered via a dedicated goggle system integrated with the Mosaic platform (Glaukos). Oxygen flow rate and delivery parameters are device-dependent and were set according to the manufacturer’s specifications for the Mosaic-integrated goggle system. The target oxygen concentration at the corneal surface (> 90%) was maintained throughout the irradiation period. Postoperatively, as with standard transepithelial CXL, topical antibiotics and fluorometholone (0.1%) were administered four times daily for two weeks.

Clinical outcome measures

Measurements were performed at 4 time points: preoperatively and at 1, 2, and 3 years postoperatively. Data on the following outcome measures were extracted from the patients’ medical records: UCVA, CDVA, pachymetry, and corneal tomography. Corneal tomography was performed using AS-OCT (CASIA2; TOMEY, Nagoya, Japan). Intraoperative and postoperative complications were also documented. Changes in keratoconus progression following custom TE-CXL were evaluated mainly based on the maximum K value (Kmax) measured by AS-OCT. The procedure was considered a failure if Kmax increased by ≥ 1 D and corneal thickness at the apex decreased by ≥ 2% at 3 years postoperatively compared with baseline. The corneal endothelial cell density was measured by a specular microscope.

Anterior-segment optical coherence tomography parameters

Central corneal thickness, corneal thickness at the thinnest point, and total higher-order aberrations (HOAs) within a 6 mm diameter zone including coma and spherical aberrations, were calculated. HOAs were calculated as previously described [16–18]. Briefly, a 3D model was reconstructed from anterior and posterior corneal elevation data. Anterior, posterior, and total corneal aberrations over 4 and 6 mm diameter zones were calculated by ray-tracing software (version 5.1). The refractive indices of the cornea and aqueous humor used for this purpose were 1.376 and 1.336, respectively. The wavefront aberration was decomposed using a Zernike polynomial; each Zernike coefficient was calculated. HOAs were shown as the root mean square of the Zernike coefficient from the 3rd to 8th order.

Statistical Analysis

Descriptive statistics were used to summarize demographic and clinical characteristics, with continuous variables expressed as mean ± standard deviation. A linear mixed-effects model was used to assess changes in key outcome measures over time. The model included time points as a fixed effect and patient ID as a random effect to account for inter-eye correlation, assuming a compound-symmetry covariance structure. Restricted maximum likelihood estimation was applied with the baseline values of each parameter included as covariates in the fixed effects. Least squares means (LSMeans) and their 95% confidence intervals (CIs), estimated from the model, were reported for each time point. Dunnett’s adjustment was applied for multiple comparisons, testing outcome measures at each postoperative time point (1, 2, or 3 years) against the baseline (Time 0). Statistical significance was set at p < 0.05. All analyses were performed using SAS Studio Release 3.81 (Enterprise Edition; SAS Institute Inc., Cary, NC, USA).

Results

Table 1 presents the baseline characteristics and selected outcome measures. The average age ± standard deviation of the 74 patients was 26.4 ± 10.1 years, ranging from 12 to 61 years. The mean UCVA (logMAR), preoperative Kmax, and thinnest corneal thickness was 0.71 ± 0.51, 56.1 ± 7.2 D, and 447.0 ± 42.1 μm, respectively. Regarding a 6-mm pupil, the mean total corneal HOA was 3.0 ± 1.8 μm (range, 0.5–10.0 μm). Among its components, the mean coma aberration was 2.5 ± 1.7 μm (range, 0.3–9.5 μm); the mean spherical aberration was − 0.4 μm (range, − 2.5 to 1.2 μm).

Table 1.

Baseline characteristics of the study participants (N = 74) before the customized corneal-crosslinking

transepithelial corneal cross-linking
Variable	Mean	SD	Min	Max
Age	26.4	10.1	12	61
Visual acuity
UCVA (log MAR)	0.71	0.51	-0.18	2.0
CDVA (logMAR)	0.11	0.27	-0.18	1.15
ECD (cells/mm2)	2731.3	272.7	2227.0	3497.0
Keratometric Index (D)
Kmax	56.1	7.2	45.8	78.2
Ks	50.1	5.1	43.4	73.1
Kf	46.6	3.9	38.9	63.9
CYL	3.5	2.3	0.2	9.8
Pachymetry (µm)
Apex	472.5	42.3	331.0	553.0
Thinnest	447.0	42.1	281.0	521.0
Total HOA (6 mm, µm)	3.0	1.8	0.5	10.0
Spherical aberration (µm)	-0.4	0.8	-2.5	1.2
Coma aberration (µm)	2.5	1.7	0.3	9.5

Abbreviations: UCVA: Uncorrected Visual Acuity; CDVA: Corrected Distance Visual Acuity

LogMAR: Logarithm of the Minimum Angle of Resolution

ECD: Endothelial cell density; Kmax: maximal K reading; Ks: K steep; Kf: K flat; CYL: Cylinder

D: diopter; HOA: high order aberration

Ninety eyes from 74 patients were included in the analysis. The linear mixed model (LMM) demonstrated a good fit for repeated measurements in the same individuals. Overall, the LMM results showed that UCVA, corneal topography, HOAs, and other key parameters significantly improved postoperatively over time, as evaluated by LSMeans adjusted for baseline values. A significant reduction in Kmax was found over time after custom TE-CXL (Fig. 1). The LSMean at baseline (Time 0) and subsequent time points (1, 2, and 3 years postoperatively) were 56.2 D, 54.9 D, 54.4 D, and 54.2 D, respectively. Compared with baseline, the mean differences in Kmax were − 1.3 D at 1 year (P < 0.01), -1.8 D at 2 years (P < 0.01), and − 2.0 D at 3 years (P < 0.01) postoperatively. The differences were statistically significant after correction for multiple comparisons at each year postoperatively.

Fig. 1.

Longitudinal changes in maximum keratometry (Kmax) following custom transepithelial corneal cross-linking (TE-CXL). The x-axis shows time in years (0 = baseline; 1–3 = follow-up), and the y-axis shows Kmax in diopters (D). Data are presented as least squares (LS) means, with shaded areas indicating 95% confidence intervals (CIs). Comparisons were made versus baseline (Time 0) using Dunnett-adjusted pairwise tests. *P < 0.05; **P < 0.01

UCVA (logMAR) showed significant improvement over time (Fig. 2). The LSMean UCVA at baseline (Time 0) and at subsequent time points (1, 2, and 3 years postoperatively) were 0.67, 0.51, 0.53, and 0.49, respectively. Compared with baseline, the mean differences in UCVA were − 0.16 at 1 year, -0.14 at 2 years, and − 0.16 at 3 years, with significant improvement observed at each year postoperatively. In contrast, CDVA did not show significant improvement at any time point (Fig. 3).

Fig. 2.

Longitudinal changes in uncorrected visual acuity (UCVA) following transepithelial corneal cross-linking (TE-CXL). Data are presented as least squares (LS) means with 95% confidence intervals (CIs). Comparisons were made versus baseline (Time 0) using Dunnett-adjusted pairwise tests. *P < 0.05; **P < 0.01

Fig. 3.

Longitudinal changes in corrected distance visual acuity (CDVA) following transepithelial corneal cross-linking (TE-CXL). Data are presented as least squares (LS) means with 95% confidence intervals (CIs). Comparisons were made versus baseline (Time 0) using Dunnett-adjusted pairwise tests. *P < 0.05; **P < 0.01

The total corneal HOA for the 6-mm pupil improved postoperatively (Fig. 4). The LSMean values at baseline and subsequent time points were 2.97, 2.59, 2.62, and 2.57, respectively. The mean differences from baseline were − 0.38 at 1 year, -0.35 at 2 years, and − 0.40 at 3 years, suggesting a gradual improvement in corneal optical quality. Similarly, as a major component of HOA, coma aberration showed significant improvement over time following custom TE-CXL. The LSMean values at baseline (time 0) and at subsequent time points (1, 2, and 3 years postoperatively) were 2.50, 2.15, 2.13, and 2.11, respectively. Compared with baseline, the mean differences were − 0.35 at 1 year, -0.37 at 2 years, and -0.39 at 3 years, indicating a progressive reduction over time. Spherical aberrations, another key component of HOA, improved significantly at 2 and 3 years postoperatively.

Fig. 4.

Longitudinal changes in corneal higher-order aberrations (HOA, 6 mm), coma, and spherical aberration (SA) following transepithelial corneal cross-linking (TE-CXL). (A) HOA. (B) Coma. (C) SA Data are presented as least squares (LS) means with 95% confidence intervals (CIs). Comparisons were made versus baseline (Time 0) using Dunnett-adjusted pairwise tests. *P < 0.05; **P < 0.01

Corneal thickness at the apex showed a gradual decrease over time, reaching statistical significance at 2 years (P < 0.05) and 3 years (P < 0.01), whereas no significant change was observed at 1 year (Fig. 5). In contrast, statistically significant reductions in thinnest corneal thickness were observed from 1 to 3 years postoperatively (P < 0.01).

Fig. 5.

Longitudinal changes in the pachymetry index following transepithelial corneal cross-linking (TE-CXL). (A) Apex. (B) Thinnest point. Data are presented as least squares (LS) means with 95% confidence intervals (CIs). Comparisons were made versus baseline (Time 0) using Dunnett-adjusted pairwise tests. *P < 0.05; **P < 0.01

During the 3-year follow-up period, no complications affecting visual function were observed. Of the 90 treated eyes, Kmax data were missing in 9 eyes (10.0%) at baseline, 5 eyes (5.6%) at 1 year, 8 eyes (8.9%) at 2 years, and 18 eyes (20.0%) at 3 years. Among the 64 eyes with both baseline and 3-year Kmax data available, 6 eyes (9.4%) met the predefined failure criterion.

Discussion

In a cohort of 74 individuals with progressive keratoconus who underwent custom TE-CXL, UCVA, Kmax, and total HOAs (coma and spherical aberrations) for a 6-mm pupil showed statistically significant postoperative changes compared with corresponding preoperative values from 1 year onward. These parameters remained generally stable through the 3-year follow-up period. Given the study design, our study could not demonstrate superiority over existing CXL approaches, but instead provided supportive longitudinal observational data regarding postoperative changes following custom TE-CXL with oxygen supplementation.

Increasing evidence supports the effectiveness of CXL in stabilizing keratoconus progression [19, 20]. However, the advantage of transepithelial CXL over epi-off CXL remains controversial. A 2021 Cochrane systematic review and meta-analysis included seven randomized controlled trials (RCTs) comparing transepithelial CXL with epi-off CXL in patients with progressive keratoconus [21].The review concluded that it remains uncertain whether transepithelial CXL is more effective than epi-off CXL in halting disease progression or improving visual acuity outcomes, mainly due to the low quality of the included RCTs [21].For example, few RCTs reported long-term follow-up beyond 3 years, and there was inconsistency in both the measurement and reporting of clinical outcomes across studies.

Kmax is a key outcome measure commonly used to assess the effectiveness of CXL across various protocols. Similar to previous reports of conventional epi-off CXL, a reduction in Kmax over time was observed following custom TE-CXL in this cohort. Previous long-term studies of conventional epi-off CXL have reported postoperative reductions in Kmax ranging from minimal change to approximately − 6 D [21–24]. The − 2.0 D postoperative reduction in Kmax observed in the present study was therefore broadly consistent with the existing literature [22–24]. A previous Japanese study using custom TE-CXL with oxygen supplementation, similar to our protocol, showed significant reductions in Kmax at 1 year postoperatively [25].Similarly, our results showed that these changes were observed at 1 year and were maintained at 2 and 3 years postoperatively, suggesting that these improvements may extend beyond the first year.

In addition to changes in Kmax, reductions in HOAs for a 6-mm pupil, including coma and spherical aberrations, were observed following custom TE-CXL. These parameters are commonly used to characterize optical quality and visual function. We observed gradual decreases in both aberrations over time, with the most pronounced reductions occurring in the second and third postoperative years compared with preoperative values. We expected both UCVA and CDVA to improve over time with a reduction in HOAs, particularly coma and spherical aberrations. However, only UCVA showed a significant improvement in this cohort. Two non-mutually exclusive explanations may account for this discrepancy. First, the magnitude of HOA reduction observed in our study may not have been sufficient to produce a measurable improvement in CDVA. In fact, the effect of CXL on CDVA has been inconsistent across studies, and a previous meta-analysis was unable to determine its effect because of substantial heterogeneity (I² = 70%) [21]. Second, CDVA may be relatively insensitive to subtle improvements in optical quality reflected by HOAs [26]. Previous studies have shown that keratoconus patients may exhibit increased HOAs and reduced contrast sensitivity despite preserved visual acuity [26], and that HOAs correlate more strongly with contrast sensitivity than with visual acuity [27]. Visual function in keratoconus is influenced by multiple factors, including irregular astigmatism and neural adaptation [28], which may not be fully captured by changes in HOAs alone. Given the importance of preserving or improving CDVA in the management of keratoconus, further follow-up is warranted to assess whether functional visual improvement emerges over a longer period. Moreover, despite statistically significant postoperative changes in Kmax and HOAs, the extent to which these changes translate into meaningful patient benefit remains uncertain, particularly because CDVA did not show significant improvement during follow-up. Given that patient-reported outcome measures, contrast sensitivity, and quality-of-vision metrics were not collected in this retrospective study, future prospective studies should incorporate such endpoints to better characterize the clinical significance of HOA reductions.

Whether oxygen supplementation improves the efficacy of CXL remains unclear. Favorable results have been observed in both our study and a previous Japanese study [25]. From a mechanistic perspective, CXL involves two photochemical pathways: the oxygen-independent Type I and the oxygen-dependent Type II reactions [29]. In Type II reactions, oxygen reacts with UV-activated riboflavin to produce singlet oxygen, which oxidizes stromal collagen and leads to cross-linking formation [11]. Ex vivo studies have underscored the critical role of oxygen in this process [12]. Additionally, a human study reported the effectiveness of photorefractive CXL performed under high-concentration oxygen [30], further supporting the mechanistic importance of oxygen in enhancing CXL efficacy. However, our study was limited by the absence of a comparison group, and therefore the contribution of oxygen supplementation to these outcomes cannot be determined and remains speculative. The efficacy of oxygen supplementation in CXL should be evaluated in prospective studies with appropriate comparison groups and control of confounding factors such as UV irradiation patterns.

The frequency of adverse outcomes associated with both epi-off and transepithelial CXL has been reported in previous studies, with transepithelial CXL showing a more favorable safety profile [25]. In our study, custom TE-CXL was not associated with adverse events such as infection or corneal haze over the 3-year follow-up period. These findings suggest that custom TE-CXL may offer safety advantages, as it preserves the corneal epithelium and may reduce the risks associated with epi-off CXL, including infection and prolonged healing times.

Although we found no significant changes in endothelial cell density following custom TE-CXL, the impact of UV exposure might vary depending on the distribution of corneal thickness. One remaining concern about custom TE-CXL is the safety of using higher-dose UV irradiation, as our protocol applied up to 15 J/cm², twice the dose typically used in standard CXL, directly to the thinnest area of the cornea. A minimum corneal thickness of ≥ 400 μm (excluding the epithelium) is recommended for safely performing epi-off CXL [31]. However, in custom TE-CXL, some UV light is absorbed by the epithelium and residual riboflavin remains in the corneal epithelium [32].This might help reduce damage to the corneal endothelium. Previous studies have shown a shallower demarcation line with epi-on compared with epi-off CXL [33], suggesting reduced penetration. Nevertheless, the safe upper limit of total UV energy for custom TE-CXL remains undetermined. Further research is needed to establish the safety thresholds for custom TE-CXL protocols.

In terms of clinical applicability, custom TE-CXL may be considered for patients with progressive keratoconus at early-to-moderate stages (e.g., Kmax < ~ 62 D and sufficient corneal thickness), based on the observed improvement in key outcomes in our study and existing literature [34, 35]. However, these observations should be interpreted cautiously, as they are derived from a single-arm analysis and subgroup analyses were not pre-specified. Further studies in well-characterized cohorts are needed to refine patient selection for custom TE-CXL.

Our study has certain limitations. First, the major limitation is the absence of a control group, which precludes direct comparison with conventional transepithelial CXL. As a result, the observed changes in corneal shape and visual acuity cannot be attributed specifically to custom TE-CXL. Without a comparison group, it is not possible to determine whether the magnitude of Kmax reduction differs from that expected with other established CXL approaches or from the natural course of the disease. Although we applied a linear mixed-effects model with adjustment for baseline variables to account for preoperative differences and to improve the robustness of longitudinal assessments, this approach does not address the lack of a control group or permit causal inference. Accordingly, the findings should be interpreted as descriptive, and RCTs are required to establish comparative effectiveness and causality. Second, although the sample size was comparable to previous single-center studies, the relatively modest number of patients may limit the generalizability of the findings. Third, including only patients with a minimum of 3 years of follow-up could have introduced survivorship bias, as patients with unfavorable outcomes or complications were more likely to discontinue follow-up, potentially leading to an overestimation of treatment effectiveness and limiting generalizability to broader keratoconus populations. Fourth, loss to follow-up for certain variables may be a concern, particularly at later follow-up time points. Mixed-effects modeling was applied to account for incomplete repeated measurements under the assumption of missing at random. Moreover, we conducted a sensitivity analysis comparing baseline characteristics between eyes with and without available 3-year Kmax data and found no evidence of systematic baseline differences in clinical variables, except for apex corneal thickness. Nonetheless, the possibility of informative dropout cannot be entirely excluded, which may have biased toward more favorable postoperative findings. Overall, these follow-up-related biases may have contributed to overestimation of favorable long-term postoperative outcomes. Finally, although out study indicates that custom TE-CXL was associated with stabilization of keratoconus progression over a 3-year period, ensuring its long-term effectiveness and safety remains crucial. To this end, we are continuing to follow up our study patients for at least 10 years to evaluate the long-term outcomes following custom TE-CXL.

Conclusions

In conclusion, this descriptive, non-comparative study showed that custom TE-CXL with oxygen supplementation may be associated with stabilization of keratoconus over a 3-year period without observed adverse events. Further randomized controlled studies are warranted to evaluate the comparative effectiveness and clinical relevance of custom TE-CXL with oxygen supplementation.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AS-OCT

Anterior segment optical coherence tomography

CDVA

Corrected distance visual acuity

CXL

Corneal Cross-Linking

Custom TE-CXL

Customized transepithelial CXL

HOA

Higher-order aberrations

Kmax

Maximum K value

LSMean

Least squares mean

LMM

Linear mixed model

RCT

Randomized controlled trial

RGP

Rigid gas-permeable

UCVA

Uncorrected visual acuity

UV

Ultraviolet

Author contributions

Concept and design: TK; Data collection: TK and TN; Analysis and interpretation: YL, AH, and TK; Manuscript preparation: YL and TK. All authors read and approved the final manuscript.

Funding

Not applicable.

Data availability

The datasets used in this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board and Ethics Committee of Nagoya Eye Clinic (UMIN ID: 000033372). This study adhered to the tenets of the Declaration of Helsinki. Since this study was conducted retrospectively, the opt-out method was approved by the Institutional Review Board and the need for obtaining written informed consent was waived.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.