Introduction

Keratoconus is a corneal ectatic disorder primarily affecting individuals between 10 and 30 years of age. Progression of the disease leads to increased corneal irregular astigmatism and a decline in spectacle-corrected visual acuity.^1–3 Corneal crosslinking (CXL), developed by Seiler et al, has been evaluated for its effectiveness in preventing the progression of keratoconus. Recently, the adoption of CXL has expanded as evidence of its safety and efficacy has grown.^4,5 Consequently, early detection of keratoconus has become increasingly important.

Numerous studies have highlighted the utility of corneal topography and tomography in the early diagnosis of keratoconus, and these techniques are now widely applied in clinical practice.^6–9 Recent advances in examination devices have enabled direct assessment of corneal biomechanical properties, providing insights into the structural vulnerability of keratoconus.^10,11 Methods that integrate assessments of corneal biomechanical properties with corneal shape have demonstrated enhanced diagnostic utility.^12–14

The corneal epithelium undergoes compensatory remodeling of its thickness in response to alterations in the anterior corneal surface. This adaptive mechanism may obscure early-stage anterior corneal protrusion in keratoconus.⁴ Efforts to detect keratoconus at its earliest stages have focused on identifying these epithelial thickness changes. Very high-frequency digital ultrasound (VHFDU) and anterior segment optical coherence tomography (AS-OCT) have been used to measure corneal epithelial thickness.¹⁵ However, VHFDU is a water immersion system that requires considerable skill, whereas noninvasive AS-OCT has gained broader acceptance in clinical settings due to its ease of use.^5,16–19

OCT captures backscattered light from the boundary surfaces where the refractive index changes within the cornea. In normal corneal tissue, each layer maintains a relatively uniform refractive index, allowing boundary surfaces to appear distinct on OCT images due to differences in refractive indices.^6,7 The interface between the corneal epithelium and the corneal stroma produces less backscatter compared to the anterior and posterior surfaces of the cornea, necessitating high-resolution OCT systems to accurately identify the corneal epithelium.

Although no direct comparative studies have been conducted between CASIA2 and other anterior segment OCT devices, its key advantage lies in the broad range of clinically useful analyses it offers. CASIA2 uses a 1310 nm swept-source and high-speed scanning to provide high-quality 3D imaging of the anterior segment. It enables preoperative assessment of total corneal astigmatism, higher-order aberrations, topography, and lens geometry including tilt and decentration. Recent advancements have allowed CASIA2 to segment the posterior surface of the corneal epithelium and measure corneal epithelial thickness. However, reports on the reproducibility of corneal epithelial thickness measurements or the evaluation of keratoconus using CASIA2 remain limited.

This study aimed to assess the reproducibility of corneal epithelial thickness measurements in normal and keratoconus eyes using CASIA2. Additionally, the study investigated the utility of corneal epithelial thickness as a parameter for diagnosing keratoconus.

Materials and Methods

Patients and Healthy Controls

This retrospective study was conducted at the Nagoya Eye Clinic. The inclusion criteria for keratoconus patients were as follows: keratoconus was diagnosed by a corneal specialist using slit-lamp microscopy, corneal topography, and corneal tomography. The severity of keratoconus was classified as stage 1 or 2 according to the Amsler-Krumeich classification. Eyes with diseases other than keratoconus or with corneal opacities were excluded from the study.For the normal control group, the inclusion criteria were no history of ocular surgery, no ocular diseases other than refractive error, and no abnormal findings on anterior segment optical coherence tomography (AS-OCT, CASIA2, TOMEY, Nagoya, Japan) or slit-lamp microscopy.

According to the inclusion and exclusion criteria, 16 eyes from 16 patients (mean age 29.2 ± 13.1 years) with keratoconus and 11 eyes from 11 healthy controls (mean age, 24.2 ± 5.1 years) were included for the reproducibility analysis. To evaluate corneal thickness and assess keratoconus screening, an additional 40 eyes from 40 keratoconus patients (mean age, 25.5 ± 8.0 years) and 44 eyes from 44 healthy controls (mean age, 24.2 ± 5.1 years) were included.

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Nagoya Eye Clinic (2023–61). The ethics committee approved the use of an opt-out method, and the requirement for informed consent was waived due to the retrospective nature of the study.

Evaluation of Corneal Epithelial Thickness

Corneal epithelial thickness was analyzed using the AS-OCT CASIA2 (TOMEY). Measurements were performed thrice on the same day by a certified and experienced orthoptist, and the average of the three measurements was recorded. The analysis encompassed a 7-mm diameter area of the cornea, divided into three concentric zones: 0–2 mm, 2–5 mm, and 5–7 mm from the center. Each zone was further subdivided into eight 45-degree sections, yielding a total of 24 segments. The mean and standard deviation of corneal epithelial thickness in each segment were analyzed.

Corneal Epithelial Thickness Ratio Calculation Method

The corneal epithelial thickness ratio was defined as the ratio of the average corneal epithelial thicknesses of the upper and lower corneas (upper corneal thickness/lower corneal thickness). This ratio was calculated within the zones of 0–2 mm, 2–5 mm, 5–7 mm, and 0–5 mm. Comparisons of the corneal epithelial thickness ratio were conducted between the keratoconus and control groups.

Investigation of the Usefulness of Keratoconus Diagnosis

A logistic regression analysis was performed to evaluate the diagnostic utility for keratoconus. The dependent variable was a dummy variable, coded as 1 for keratoconus and 0 for normal. Independent variables included the mean corneal epithelial thickness, the standard deviation of corneal epithelial thickness, and the corneal epithelial thickness ratio.

Statistical Analysis

Comparisons of the corneal epithelial thickness, standard deviation of corneal epithelial thickness, and corneal epithelial thickness ratio between the keratoconus and control groups were conducted using the Mann–Whitney U-test. All statistical analyses were performed using SPSS (VER.29), with a significance level set at less than 5%.

Results

For the reproducibility analysis, the keratoconus group included 12 males and 4 females, with a mean keratometry of 49.4 ± 3.8 D. The healthy control group included 9 males and 2 females, with a mean keratometry of 43.2 ± 1.2 D. For the corneal thickness and keratoconus screening evaluations, the keratoconus group comprised 34 males and 6 females, with a mean keratometry of 45.11 ± 1.60 D. The healthy control group comprised 41 males and 3 females, with a mean keratometry of 42.77 ± 1.24 D. Table 1 summarizes the demographic characteristics of the study participants.

Table 1 Demographic Information of Patients with Keratoconus and Healthy Controls

Comparison of Reproducibility of Corneal Epithelial Thickness in Keratoconus and Normal Eyes Implant

Figure 1 shows the mean intraclass correlation coefficients (ICCs) for corneal epithelial thickness measurements in keratoconus and normal eyes. For the keratoconus group, the mean ICCs were 0.962, 0.961, and 0.970 for the 0–2 mm, 2–5 mm, and 5–7 mm zones, respectively. For the normal control group, the mean ICCs were 0.977, 0.985, and 0.985, respectively.

Figure 1 Mean intraclass correlation coefficient (ICC) of corneal thickness measurements for keratoconus and normal eyes. The keratoconus group exhibited slightly lower ICC values compared to the normal group; however, both groups had ICC values above 0.8 in all measured areas.

Mean Corneal Epithelial Thickness

Table 2 presents the mean corneal epithelial thickness results. In the nasal (N), inferonasal (IN), inferior (I), inferotemporal (IT), and temporal (T) areas, the keratoconus group exhibited significantly lower thickness values compared to the control group (p<0.05). In the superior (S) and superotemporal (ST) areas, the keratoconus group showed significantly higher thickness values than the control group (p<0.05).

Table 2 Comparison of Mean Corneal Epithelial Thickness Between the Keratoconus and Control Groups

Standard Deviation of Corneal Epithelial Thickness

The standard deviation of corneal epithelial thickness is shown in Table 3. In all regions, the keratoconus group showed significantly higher values than the control group (p<0.05 across all areas).

Table 3 Comparison of the Standard Deviation of Corneal Epithelial Thickness Between the Keratoconus and Control Groups

Corneal Epithelial Thickness Ratio

Table 4 shows the mean corneal epithelial thickness ratios for the keratoconus and control groups. In the keratoconus group, the mean ratios (95% confidence interval) were 1.05 (1.04–1.06) for the 0–2 mm zone, 1.07 (1.05–1.09) for the 2–5 mm, and 1.02 (1.00–1.02) for the 5–7 mm zone. For the control group, the corresponding values were 0.99 (0.99–1.00), 0.99 (0.98–0.99), and 0.98 (0.98–0.99), respectively. The keratoconus group exhibited significantly higher ratios across all zones compared to the control group (p<0.001).

Table 4 Comparison of the Mean Corneal Epithelial Thickness Ratio Between the Keratoconus and Control Groups

Investigation of the Usefulness of Keratoconus Diagnosis

Logistic regression analysis identified the standard deviation of corneal epithelial thickness in 2–5 mm and 0–5 mm ranges as explanatory variables. The area under the receiver operating characteristic curve (AUROC) was 0.955, with a cut-off value of 0.494, and the sensitivity and specificity were 92.1% and 100%, respectively (Figure 2).

Figure 2 Receiver operating characteristic (ROC) analysis illustrating the sensitivity and specificity of keratoconus screening using corneal epithelial thickness. Abbreviation: AUROC, area under the ROC curve.

Discussion

The CASIA2 AS-OCT system has gained widespread use in evaluating anterior segment structures.^20–22 This study employed CASIA2 to evaluate the reproducibility of corneal epithelial thickness measurement in keratoconus eyes and compared corneal epithelial thickness between keratoconus and normal eyes. Additionally, the study explored the usefulness of corneal epithelial thickness parameters for diagnosing keratoconus. To our knowledge, this represents the first study evaluating keratoconus eyes using corneal epithelial thickness measurements obtained with CASIA2.

Firstly, in this study, the reproducibility of three consecutive measurements in keratoconus eyes was comparable to that in normal eyes, suggesting sufficient reliability for clinical application. However, the ICC showed a slight tendency to be lower in the central and lower regions of keratoconus eyes. This discrepancy may stem from the pronounced fluctuation in epithelial thickness observed in keratoconus eyes. A minor shift in the center of measurement could lead to variability in the recorded values due to these fluctuations.

Compensatory changes in corneal epithelial thickness are a hallmark of keratoconus, reflecting the thinning and steepening of the cornea.^3,23 Consistent with previous reports, corneal epithelial thickness in keratoconus eyes was thinner at the inferior region and the inferior temporal region compared to normal eyes.²⁴ Additionally, another study reported a significantly larger standard deviation of corneal epithelial thickness in keratoconus eyes compared to normal eyes.²⁵ These results suggest that assessing corneal epithelial thickness may enable the detection of subtle changes on the anterior corneal surface associated with localized steepening. Such changes may be masked by epithelial remodeling, particularly in the early stages of keratoconus.

In this study, the standard deviation was employed as an index to assess intra-sectoral variations in corneal epithelial thickness. The findings demonstrated that the standard deviation of corneal epithelial thickness increased across all regions within the 0–2 mm, 2–5 mm, and 5–7 mm ranges. This likely reflects more pronounced local corneal stromal shape changes in keratoconus eyes. Previous studies using RTVue have shown heterogeneity in corneal epithelial thickness in patients with keratoconus,²⁶ and other reports using SD-OCT (Spectralis Heidelberg) have shown an increased standard deviation of corneal epithelial thickness in keratoconus compared to normal corneas.²⁷ The present results were consistent with previous reports.

In this study, the mean corneal epithelial thickness ratio in keratoconus was greater than that in normal corneas, with a value of 1.07 for the 2–5 mm region. A previous study reported that in keratoconus (mean K: 46.44 ± 4.13), the corneal epithelial thickness ratio increased with disease severity, as indicated by worsening logMAR visual acuity.⁸ In that report, the corneal epithelial thickness ratio was below 0.95 in normal eyes and exceeded 1.00 in severe keratoconus. Although direct comparison is limited due to differences in the measurement areas between the two studies, a similar trend of increased corneal epithelial thickness ratio in keratoconus compared to normal corneas was observed.

The logistic regression analysis in this study revealed that the standard deviation of corneal epithelial thickness and the corneal epithelial thickness ratio effectively distinguished keratoconus from normal corneas with high accuracy. A previous study analyzing corneas approximately 10 mm in diameter reported that the thickness of the corneal epithelium and Bowman’s layer successfully differentiated subclinical keratoconus from normal corneas, achieving an AUC of 0.984, with sensitivity and specificity values of 100% and 90.5%, respectively.²⁸ Another study analyzing corneas with an 8 mm diameter used the standard deviation of corneal epithelial thickness and the corneal epithelial thickness ratio to identify keratoconus, reporting an AUC of 0.92, sensitivity of 74.0%, and specificity of 96.7% for the corneal epithelial thickness ratio.²⁹ The AUROC in our study was 0.955, which was similar to that in previous studies, indicating that the corneal epithelial thickness parameters measured by CASIA2 provide robust utility in diagnosing keratoconus.

This study had several limitations. Although this study demonstrated high intra-operator reproducibility, inter-operator reproducibility was not evaluated. As the anterior segment OCT used in this study includes an automatic parameter measurement function, inter-operator reproducibility is presumed to be high; however, further investigation is warranted. The diagnostic accuracy for early suspected keratoconus or forme fruste keratoconus (FFK) remained unknown, as the analysis included only cases already diagnosed with keratoconus. Future research should focus on evaluating even earlier cases than those assessed in this study. Previous reports have indicated an association between corneal epithelial thickness and corneal stromal thickness.^10,30 One study suggested that incorporating changes in the thickness of Bowman’s layer could enhance the accuracy of diagnosing suspected FFK or keratoconus.¹¹ These findings indicate that comprehensive evaluations considering corneal epithelial thickness, corneal stromal thickness, and posterior corneal shape may improve screening accuracy.¹³

In laser corneal refractive surgery, thorough preoperative screening is essential to ensure the safety and stability of postoperative outcomes. One of the most critical aspects of this screening process is the identification and exclusion of keratoconus and other forms of corneal ectatic disorders, as undetected subclinical keratoconus is a major risk factor for the development of postoperative keratectasia—a potentially sight-threatening complication. Although conventional corneal topography and tomography are widely used to detect keratoconus, these modalities may not always identify early or subclinical forms of the disease with sufficient sensitivity. In this context, evaluating the corneal epithelial thickness profile, which can reflect subtle biomechanical or morphological changes in the cornea, may offer additional diagnostic value. The use of anterior segment OCT to assess epithelial thickness has the potential to enhance the screening protocol by providing complementary information to traditional imaging techniques. Future studies should explore whether integrating epithelial thickness measurements with topographic and tomographic indices can lead to improved sensitivity and specificity in detecting early keratoconus, thereby contributing to safer refractive surgical practices.

Conclusion

In conclusion, corneal epithelial thickness measurements using CASIA2 demonstrated high reproducibility and may aid in the accurate diagnosis of keratoconus.