Evaluation of Corneal Tomographic, Biomechanical and Pachymetric Characteristics in Patients With Keratoconus, Their First-Degree Relatives, and Normal Individuals

Abstract

Purpose:

This study assessed corneal tomographic, biomechanical, and pachymetry using 3 multimodal imaging (MMI) diagnostic capabilitiesto detect keratoconus in first-degree relatives (FDR) of patients with keratoconus and normal populations.

Methods:

This was a prospective cross-sectional study. A total of 118 eyes from 78 patients with keratoconus, 96 eyes from 54 FDR subjects, and 126 eyes from 63 healthy individuals were analyzed using the Oculus Pentacam HR, Oculus Corvis ST, and Cirrus OCTA 5000. Corneal tomography, biomechanics, and pachymetry were performed and compared between the 3 groups.

Results:

Notable disparities in corneal tomography, biomechanics, and pachymetry were observed between FDR and healthy subjects. FDR exhibited a thinner cornea and reduced corneal biomechanical strength compared with normal eyes. In distinguishing keratoconus in FDR, the diagnostic capabilities of B.Ele.Th (back elevation at thinnest), BAD (Belin/Ambrósio Enhanced Ectasia Display), and TBI (tomographic biomechanics index) proved to be highly effective with area under the curve (AUC) beyond 0.90. Stepwise logistic regression (SLR) model combination of corneal tomography and biomechanics of IHD, CBI, and TBI showed an excellent accuracy of AUC:0.999 for detecting keratoconus in FDR than using single MMI alone.

Conclusions:

FDR with keratoconus indeed have an increased likelihood of developing corneal ectasia, including keratoconus itself. MMI screening of asymptomatic relatives can facilitate early stage or subclinical keratoconus detection.

Diagnosis of keratoconus in the early stages is crucial to initiate early treatment and to avoid possible refractive surgery that could produce ectasias.¹ Imaging modalities such as corneal topography, tomography, corneal biomechanics, anterior segment optical coherence tomography (OCT), and confocal microscopy are believed to optimize the diagnosis of early keratoconus and forme fruste keratoconus.² Forme fruste keratoconus (FFKC) is defined as having normal topography, normal slit-lamp examination, and diagnosis of keratoconus in the fellow eye.^3,4 Diagnosing FFKC is often challenging because many patients lack the characteristic clinical manifestations, such as thinning of the cornea or cone-shaped bulges, typically associated with this condition.

Corneal tomography, including Scheimpflug and optical coherence tomography, is currently the most frequently used diagnostic test for detecting keratoconus (KC) in its early stages. Scheimpflug tomography, specifically Pentacam, is a highly reliable screening tool that provides quantitative parameters for the diagnosis of corneal ectasia.⁵ Corneal Visualization Scheimpflug Technology (Corvis ST) is a noninvasive method that can detect corneal biomechanical changes before morphological changes occur. In addition, corneal biomechanics have been used to detect early or suspected keratoconus, and it has been suggested that changes in corneal biomechanical properties may be the initial event in keratoconus.⁶ Furthermore, anterior segment optical coherence tomography (OCT)-based epithelial mapping is gaining increasing recognition as a valuable diagnostic tool for detecting the early stages of keratoconus.⁷ The utilization of multiple imaging techniques has significantly enhanced the early and accurate diagnosis of keratoconus, thus reducing the severity of the disease.⁸

First-degree family members have a high prevalence of keratoconus,^9,10 especially in Asian countries, such as Iran, Turkey, and India. The heritability of corneal thickness at the apex, pupil, and thinnest points was higher in the first-degree relatives (FDRs).¹¹ Furthermore, patients with an FDR who have keratoconus are at an elevated risk of developing corneal abnormalities,¹² suggesting that keratoconus has a complex etiology involving multiple genes. Genetic risk factors for keratoconus (KC) vary significantly across populations because of genetic diversity and environmental influences. The prevalence of keratoconus ranges from 2340 per 100,000 in Israel to 0.2 to 0.4 per 100,000 in Russia, likely reflecting differences in diagnostic criteria and genetic variations.¹³ Research suggests that 10%-15% of patients with keratoconus have relatives with the same condition, indicating a possible genetic link.¹⁴ The inheritance pattern of keratoconus is complex because of its heterogeneous nature, with various families potentially exhibiting different genetic markers associated with the disorder.¹⁵

The objective of this study was to explore corneal tomography, biomechanical characteristics, and anterior segment pachymetry in first-degree relatives of patients with keratoconus using 3 distinct imaging devices. Earlier research has demonstrated notable increases in tomographic indices and irregularity,¹⁰ and substantially thinner corneas,¹² among family members of individuals with keratoconus. Evaluations of corneal biomechanics in FDR revealed decreased measurements for corneal hysteresis (CH) and corneal resistance factor (CRF),¹⁶ and reduced corneal stiffness.¹⁷ This highlights the need for additional research to confirm the effectiveness of these 3 multimodal imaging techniques in accurately diagnosing keratoconus among first-degree relatives and healthy individuals. In this study, we aimed to determine which of these imaging tools is more effective and accurate in detecting keratoconus in first-degree relatives. In addition, we devised a diagnostic model of these 3 multimodal imaging for detecting keratoconus in FDR using binary logistic regression model. Investigating corneal tomography, biomechanical, and pachymetry features in first-degree relatives of patients with keratoconus can help enhancing ectasia detection, and ensuring safe surgical practices in refractive surgery.

MATERIAL AND METHODS

Study Design and Population

This retrospective cross-sectional study included 78 patients with keratoconus and their first-degree relatives (n = 54) who visited the Ophthalmology Clinic at the Hospital Canselor Tuanku Muhriz, National University of Malaysia (UKM) in Kuala Lumpur, Malaysia, between December 2021 and December 2023. A control group of 63 healthy volunteers matched for age and sex was recruited. Informed consent was obtained from all participants in accordance with the Declaration of Helsinki, and the study protocol was approved by the Ethics Committee of the National University of Malaysia (JEP-2021-654). The diagnosis of keratoconus was based on abnormal corneal topography findings including anterior curvature indices, posterior elevation values, thinnest pachymetry, skewed asymmetric bowtie, and ≥1 of the following clinical findings: conical protrusion, Munson sign, Vogt striae, or Fleischer ring, as observed through slit-lamp examination. Eyes with late keratoconus changes such as corneal scars or hydrops were excluded because they did not pose any diagnostic challenges and had anomalous pachymetry findings. Participants with corneal inflammation, glaucoma, corneal or ocular surgery, eye trauma, or systemic diseases affecting the eyes including intracorneal ring segments and corneal collagen cross-linking were excluded from the study. The authors declare that there are no conflicts of interest with the manufacturers of Oculus Pentacam, Oculus Corvis ST, or Cirrus 5000 Spectral Domain SD OCT.

Ocular Examinations

The following parameters were assessed using Pentacam for corneal evaluation: keratometry in the flat/steep meridian (K1/K2), maximum keratometry (Kmax), thinnest point pachymetry (TP), central corneal thickness (CCT), front elevation at the thinnest point (F.Ele.Th), back elevation at the thinnest point (B.Ele.Th), maximum Ambrósio relational thickness (ARTmax), Belin/Ambrósio Enhanced Ectasia Display (BAD), index of surface variance (ISV), index of vertical asymmetry (IVA), keratoconus index (KI), inferior–superior index (I–S), index of height asymmetry (IHA), index of height decentration (IHD), minimum radius of curvature (Rmin), and inferior–superior astigmatism index (KISA). The parameters collected by Corvis ST included the time until the first or second applanation (A1T/A2T), velocity of the corneal apex during the first or second applanation (A1V/A2V), length of the first or second applanation (A1L/A2L), corneal deflection amplitude at the time of the highest corneal concavity (HCDA), peak distance at the highest concavity (HCPD), central concave curvature at the highest concavity (HCR), maximum deformation amplitude (DA), central corneal thickness (CCT), deformation amplitude ratio max 2 mm (DA ratio), integrated radius (IR), Ambrósio relational horizontal thickness (ARTh), stiffness parameter at A1 (SPA1), Corneal biomechanical index (CBI), and tomographic and biomechanical indices (TBI). Cirrus anterior segment OCT measures the pachymetry thickness map (2–5) mm for Min, average (Avg), maximal (Max), superior–inferior (S–I), superonasal–inferotemporal (SNIT), and epithelial thickness map (2–5) mm for Min, average (Avg), maximum (Max), superior–inferior (S–I), and superonasal–inferotemporal (SNIT).

Statistical Analysis

Statistical analyses were performed using IBM SPSS (version 27.0) and MedCalc (version 22.030). Continuous data are presented as mean and standard deviation. The Shapiro–Wilk test was used to test the normality of the corneal parameters. If the data followed a normal distribution, the mean ± SD was used for the basic statistical description; otherwise, median and quartile ranges were used. Independent T test was used to compare the mean values of continuous variables if the data were normally distributed, and a post-hoc Bonferroni test was conducted. If the data were not normally distributed, the Mann–Whitney test was conducted, followed by post-hoc Dunn test. All tests were two-tailed, and P < 0.05 was considered significant.

Receiver operating characteristic (ROC) curves were used to determine the discriminatory power of each parameter from the 3 multimodal imaging devices for detecting keratoconus in FDR and normal individuals. To determine the diagnostic efficiency of a biomarker, the area under the curve (AUC), sensitivity, and specificity were determined. The Youden index was used to select the optimal cutoff value for ROC analysis. The De Long test was used to assess the differences between the AUCs of the various parameters. It compares the differences between paired area under the curve (AUC) values and their standard errors to calculate the P-value. If the P-value was below the chosen significance level, the difference in the AUC values was considered statistically significant. In addition, a predictive model for identifying keratoconus in FDR was developed using binary logistic regression analysis, incorporating data from multiple imaging modalities.

Variables with an area under the ROC curve (AUC) greater than 0.9 will be further analyzed using stepwise binary logistic regression (Wald) to determine the most effective model for diagnosing keratoconus in first-degree relatives and normal individuals. The stepwise logistic regression analysis was performed in 4 different scenarios: (1) using only parameters from the Oculus Pentacam HR, (2) using only parameters from the Oculus Corvis ST, (3) using only parameters from the anterior segment OCT, and (4) using combined parameters from all 3 multimodal imaging techniques. The predictor variables comprised tomographic (Oculus Pentacam HR), biomechanical (Oculus Corvis ST), and pachymetric (Cirrus OCTA) parameters. The Hosmer–Lemeshow test was employed to assess the goodness-of-fit of the logistic regression models. Initially, all variables were included in the model and then removed based on their significance. The new model was compared with the previous one; if no significant differences were observed, the simplest model with the lowest AIC value, minimal collinearity (below 2), and optimal goodness of fit was selected.

RESULTS

Among the 54 FDR subjects, 7 individuals (12.96%) were diagnosed with keratoconus through corneal topography and slit-lamp examination. Of the 7 FDRs with keratoconus, 4 were siblings and 3 were parents. In addition, 2 subjects had bilateral keratoconus, whereas 5 had asymmetric fruste forme keratoconus. Seven individuals with keratoconus were excluded from further analyses.

Comparison of Corneal Tomography, Biomechanics, and Pachymetry

Differences were observed in all Oculus Pentacam HR, Oculus Corvis ST, and Anterior Segment Cirrus 5000 OCT variables between the keratoconus, first-degree relative, and normal groups (P < 0.001). Statistically significant differences (P > 0.05) were observed between first-degree relatives and normal subjects in the IVA, KISA, CCT, TP, ARTmax, and BAD_D. Only the highest concavity peak distance (HCPD) parameter showed no significant difference between keratoconus and FDR. Furthermore, differences in corneal biomechanics were noted between FDR and healthy individuals for CCT, SPA1, DA, A1V, A2V, A1T, and HCDA. Nonetheless, when using the Cirrus OCT 5000, only MIN, AVG, and MAX of the pachymetry thickness map (2–5 mm) and AVG of the epithelial thickness map (2–5 mm) exhibited notable variations between the first-degree relatives and the control group. A summary of these findings is presented in Table 1. Compared with healthy individuals, patients with FDR of keratoconus exhibit reduced corneal thickness and diminished corneal biomechanical strength.

TABLE 1.

Pairwise Comparison of Oculus Pentacam HR, Oculus Corvis ST, and Cirrus HD Anterior Segment OCT Variables in Keratoconus, First-Degree Relatives, and Normal (P < 0.05)

Parameters	KCN versus FDR P	KCN versus Normal P	FDR versus Normal P
Oculus Pentacam HR
IVA	<0.001*	<0.001*	0.0025*
KISA	<0.001*	<0.001*	0.0005*
CCT	<0.001*	<0.001*	0.0005*
TP	<0.001*	<0.001*	0.0011*
ARTmax	<0.001*	<0.001*	0.0020*
BAD	<0.001*	<0.001*	<0.001*
Oculus Corvis ST
CCT	<0.001*	<0.001*	<0.001*
SPA1	<0.001*	<0.001*	0.0003*
DA	<0.001*	<0.001*	0.0108*
A1V	<0.001*	<0.001*	0.001*
A2V	<0.001*	<0.001*	0.002*
A1T	<0.001*	<0.001*	<0.001*
HCDA	<0.001*	<0.001*	0.004*
Cirrus anterior segment OCT (pachymetry)
Pachymetry thickness map (2-5mm) MIN	<0.001*	<0.001*	0.005*
Pachymetry thickness map (2-5mm) AVG	<0.001*	<0.001*	<0.001*
Pachymetry thickness map (2-5mm) MAX	0.003*	<0.001*	<0.001*
Epithelial thickness map (2-5mm) AVG	1.00	0.047*	<0.001*

Significant value, P < 0.05, 2 tailed test.

A1V, applanation at first velocity; A2V, applanation at second velocity; A1T, applanation at first time; AVg, average; BAD, Belin/Ambrósio enhanced ectasia display deviation value; DA; deformation amplitude maximum at 2 mm; F.Ele.Th, front elevation at the thinnest location; KCN, keratoconus.

Receiver Operating Characteristics (ROC) Curve Analysis of Oculus Pentacam HR, Oculus Corvis ST and Cirrus 5000 OCT to Differentiate Keratoconus in First-Degree Relatives and Normal

Sixteen Oculus Pentacam HR parameters exhibited significant higher discriminatory power to detect keratoconus in FDR and normal subjects, with an area under curve (AUC) above 0.80. B.Ele.Th also exhibited sufficient strength to differentiate keratoconus among FDR, with an AUC of 0.990 (cutoff value, >11; sensitivity, 97.46; specificity, 94.79), followed by BAD_D with an AUC of 0.989 (cutoff value, >2.64; sensitivity, 100; specificity, 98.92). On the other hand, in the keratoconus versus normal group, the BAD_D showed the perfect discriminate power with an AUC of 1.00 (cutoff value, >2.10; sensitivity, 100; and specificity, 100), followed by F.Ele.Th with an AUC of 0.997 (cutoff value: >5, sensitivity: 94.92, and specificity: 96.83). The overall data are presented in (Table 2).

TABLE 2.

Pentacam Parameters of AUC (>0.90) and Sensitivity/Specificity Analyses for the Highest Youden Index Determined Cutoff in Keratoconus versus First-Degree Relatives and Keratoconus versus Normal

Parameters	AUC, P, (95% CI)	Cutoff (≥)	Youden Index J	Sensitivity (%)	Specificity (%)
Keratoconus versus first-degree relatives
ISV	0.987 (<0.001) 0.961–0.998	>41	0.8951	95.76	93.75
IVA	0.972 (<0.001) 0.940–0.990	>0.37	0.8266	84.75	97.92
IHD	0.986 (<0.001) 0.960–0.997	>0.031	0.9075	94.92	95.83
KI index	0.962 (<0.001) 0.927–0.984	>1.06	0.8971	94.92	94.79
IS	0.935 (<0.001) 0.893–0.964	>1.57	0.8436	86.44	97.92
KISA	0.941(<0.001) 0.900–0.968	>71.676	0.8416	87.79	96.87
Rmin	0.967 (<0.001) 0.933–0.987	≤6.99	0.9114	93.22	97.92
K2	0.925 (<0.001) 0.881–0.957	>46.7	0.7680	83.05	93.75
Kmax	0.983 (<0.001) 0.955–0.996	>48.1	0.9283	94.92	97.92
TP	0.932 (<0.001) 0.890–0.962	≤501	0.7355	88.14	85.42
F.Ele.Th	0.974 (<0.001) 0.942–0.991	>5	0.8762	94.92	92.71
B.Ele.Th	0.990 (<0.001) 0.966–0.999	>11	0.9225	97.46	94.79
ARTmax	0.971 (<0.001) 0.938–0.989	≤316	0.8912	97.46	91.67
BAD-D	0.989(ρ<0.0001) 0.964–0.998	>2.64	0.9892	100	98.92
Keratoconus vs normal
ISV	0.996 (<0.001) 0.978–1.00	>35	0.9582	96.61	99.21
IVA	0.991 (<0.001) 0.969–0.999	>0.21	0.9264	96.61	96.03
IHD	0.987 (<0.001) 0.964–0.997	>0.021	0.9434	98.31	96.03
KI	0.968 (<0.001) 0.937–0.986	>1.06	0.9412	94.92	99.21
IS	0.929 (<0.001) 0.889–0.958	>1.37	0.8819	88.98	99.21
KISA	0.962 (<0.001) 0.930–0.982	>32.03	0.8914	91.53	97.62
Rmin	0.960 (<0.001) 0.928–0.981	≤6.99	0.9005	93.22	96.83
K2	0.921 (<0.001) 0.880–0.952	>46.7	0.7511	83.05	92.06
Kmax	0.980 (<0.001) 0.954–0.994	>48.2	0.9174	94.92	96.83
CCT	0.944 (<0.001) 0.907–0.969	≤517	0.7549	88.98	86.51
TP	0.967 (<0.001) 0.936–0.986	≤508	0.8364	92.37	91.27
F.Ele.Th	0.997 (<0.001) 0.950–0.992	>5	0.9174	94.92	96.83
B.Ele.Th	0.994 (<0.001) 0.974–1.00	>14	0.9497	95.76	99.21
ARTmax	0.983 (<0.001) 0.958–0.995	≤316	0.9270	97.46	95.24
BAD-D	1.00 (ρ < 0.0001)) 0.984–1.00	1.00	>2.10	100	100

In comparison between keratoconus and FDR, the Tomographic Biomechanical Index (TBI) was identified as the most effective discrimination parameter, achieving an area under the curve (AUC) of 0.975 (cutoff value: >0.74, sensitivity: 97.0, specificity: 96.43). The Corneal Biomechanical Index (CBI) followed closely with an AUC of 0.974 (cutoff value, >0.62; sensitivity, 91.0; specificity, 95.26). In contrast, when comparing keratoconus with normal subjects, TBI exhibited superior discriminative power, with an AUC of 0.996 (cutoff value, >0.82; sensitivity, 97.0; specificity, 100.0), whereas CBI showed an AUC of 0.973 (cutoff value: >0.63, sensitivity: 91.0, specificity: 96.80). A comprehensive overview of these findings is presented in (Table 3).

TABLE 3.

Oculus Corvis ST Parameters of AUC (>0.90) and Sensitivity/Specificity Analyses for the Highest Youden Index Determined Cutoff in Keratoconus versus First-Degree Relatives and Keratoconus versus Normal

Parameters	AUC, P, (95% CI)	Youden Index J	Cutoff (≥)	Sensitivity (%)	Specificity (%)
Keratoconus versus first-degree relatives
CBI	0.974 (ρ < 0.0001) 0.940–0.992	0.8624	>0.62	91.00	95.24
TBI	0.975 (ρ < 0.0001) 0.941–0.992	0.9343	>0.74	97.00	96.43
CCT	0.857 (ρ < 0.0001) 0.798–0.905	0.6390	≤512	80.00	78.60
ARTh	0.935 (ρ < 0.0001) 0.889–0.966	0.7533	≤433.7	92.00	83.33
IR	0.917 (ρ < 0.0001) 0.868–0.953	0.7029	>9.8	81.00	89.29
Keratoconus vs normal
CBI	0.973 (ρ < 0.0001) 0.943–0.990	0.8730	>0.63	91.00	96.80
TBI	0.996 (ρ < 0.0001) 0.977–1.00	0.97	>0.82	97.00	100
CCT	0.940 (ρ < 0.0001) 0.910–0.969	0.7603	≤518.0	80.00	96.03
ARTh	0.934 (ρ < 0.0001) 0.894–0.963	0.7375	≤433.7	92.0	81.75
SP-A1	0.902 (ρ < 0.0001) 0.856–0.937	0.6878	≤95.2	91.00	77.78
DA	0.881 (ρ < 0.0001) 0.832–0.920	0.7086	>5.0	78.00	92.86
IR	0.935 (ρ < 0.0001) 0.894–0.963	0.7465	>9.8	81.00	93.65

The pachymetry thickness map (2–5 mm) parameters of Min, Avg, and SI, along with the epithelial thickness map parameter (2–5 mm) of Min, effectively differentiated keratoconus from the corneas of first-degree relatives, demonstrating an excellent AUC value exceeding 0.80. Furthermore, all pachymetry thickness map parameters (2–5 mm) and the Min parameter from the epithelial thickness map (2–5 mm) successfully distinguished keratoconus from normal corneas, yielding an AUC value above 0.80. The results are summarized in (Table 4).

TABLE 4.

Anterior Segment Pachymetry Parameters of AUC (>0.80) and Sensitivity/Specificity Analyses for the Highest Youden Index Determined Cutoff in Keratoconus versus First-Degree Relatives and Keratoconus versus Normal

Parameters	AUC, P, (95% CI)	Youden Index J	Cutoff (≥)	Sensitivity (%)	Specificity (%)
Keratoconus versus first-degree relatives
Pachymetry thickness map (2–5)
Min	0.906, <0.001, (0.847–0.948)	0.7260	≤491	100	72.60
Avg	0.824, <0.001, (0.753–0.881)	0.5577	≤515	76.32	79.45
SI	0.808, <0.001, (0.736–0.868)	0.5582	>31	75.00	80.82
Epithelial thickness map (2–5)
Min	0.802, <0.001, (0.729–0.863)	0.5598	≤37	71.05	84.93
Keratoconus vs normal
Pachymetry thickness map (2–5)
Min	0.974, <0.001, (0.937–0.992)	0.8858	≤491	97.47	91.4
Avg	0.917, <0.001, (0.865–0.954)	0.7212	≤523	81.01	91.4
Max	0.833, <0.001, (0.769–0.886)	0.5502	≤570	68.35	86.67
SI	0.802, <0.001, (0.734–0.859)	0.5236	>46	55.70	96.67
SNIT	0.800, <0.001, (0.732–0.858)	0.5363	>50	56.96	96.67
Epithelial thickness map
Min	0.842, <0.001, (0.778–0.893)	0.5864	≤39	79.75	78.89

Comparison of Area Under the Receiver Operating Characteristic (AUROC) Curve of Multimodal Imaging to Detect Keratoconus in First-Degree Relatives and in Normal

Figures 1 and 2 illustrate the pairwise comparison of area under the receiver operating characteristic (AUROC) curves for the highest 6 parameters of Oculus Pentacam, Oculus Corvis ST parameters, and Cirrus OCT 5000 to distinguish keratoconus eyes in FDR and normal eyes groups. These comparisons were used to detect keratoconus-suspected corneas in FDR and normal individuals. The AUROC curves for all 6 Oculus Pentacam parameters demonstrated their efficacy as reliable indicators for identifying keratoconus eyes in FDR. Similarly, the parameters BAD_D, ISV, IHD, Kmax, B.Ele.Th, and F.Ele.Th proved to be effective indicators for distinguishing keratoconus eyes in normal subjects. For the Oculus Corvis ST, the AUCs of CBI and TBI were statistically different from those of ArtH, IR, HCR, and SPA1 in diagnosing keratoconus among first-degree relatives. The AUCs of CBI and TBI were also significantly different from those of CCT, ArtH, SPA1, and IR in the diagnosis of KC. In diagnosing keratoconus among first-degree relatives, the Oculus Corvis ST showed that the areas under the curve (AUCs) for the Corneal Biomechanical Index (CBI) and Tomographic Biomechanical Index (TBI) were significantly different from those of ARTh, IR, HCR, and SPA1. The overall results are presented in (see Supplemental Table, Supplemental Digital Content 1, http://links.lww.com/ICO/B938).

FIGURE 1.

Area under the receiver operating characteristic (AUROC) curves comparison in distinguishing keratoconus eyes in first-degree relatives. A, AUROC curves of Oculus Pentacam in distinguishing keratoconus eyes in first-degree relatives. B, AUROC curves of Oculus Corvis ST in distinguishing keratoconus eyes in first-degree relatives. C, AUROC curves of Cirrus OCT pachymetry in distinguishing keratoconus eyes in first-degree relatives.

FIGURE 2.

Area under the receiver operating characteristic (AUROC) curves comparison in distinguishing keratoconus eyes in normal. A, AUROC curves of Oculus Pentacam in distinguishing keratoconus eyes in normal. B, AUROC curves of Oculus Corvis ST in distinguishing keratoconus eyes in normal. C, AUROC curves of Cirrus OCT pachymetry in distinguishing keratoconus eyes in normal.

Stepwise Logistic Regression of Keratoconus Predictive Model in First-Degree Relatives

A stepwise logistic regression analysis was conducted to develop a predictive equation for the occurrence rate of KC in FDR based on the above-mentioned tomographic, biomechanical, and pachymetric parameters from these 3 multimodal imaging modalities. ISV, IHD and F.Ele.Th, CBI, TBI, Min (pachy), Min (epi), & SI (pachy) were included as covariates in the logistic regression analysis. Stepwise prediction model of combined Pentacam, Corvis ST, & Cirrus Pachymetry including IHD, CBI, and TBI. The equation demonstrated a sensitivity, specificity, and AUC of 98.99%, 98.80%, and 0.999, respectively. This combined model showed higher accuracy than the combined model of Oculus Corvis ST and Cirrus OCT Pachymetry alone. However, the combined model of Oculus Pentacam, including ISV, IHD, and F.Ele.Th, demonstrated almost the same high accuracy, with a sensitivity of 99.00%, specificity of 98.80%, and AUC of 0.998. In summary, the selected indicators, IHD, CBI, TBI from combined Oculus Pentacam and Corvis ST, could improve accuracy in identifying FDR with KC when combined in the equation. Details are presented in (see Supplemental Digital Content 2, Table, http://links.lww.com/ICO/B939). The equation is as follows:

$$Logit\hspace{0.17em}(p/1-p)=-11.56+7.10\hspace{0.17em}\text{CBI}+7.65\hspace{0.17em}\text{TBI}+45.44\hspace{0.17em}\text{IHD}$$

DISCUSSION

The prevalence of keratoconus among FDR in our study was quite similar to that in previous large-scale studies conducted in Iran,^10,18 which reported incidences of 13.33% and 12.3%, respectively. Corneal topography and tomography are typically distinct in patients with keratoconus compared with normal individuals, as demonstrated in earlier studies.^19,20 Although the corneas of FDR keratoconus family members appear clinically healthy, our research revealed differences in BAD_D, IVA, KISA, CCT, TP, and ARTmax compared with normal corneas. This finding aligns with a recent study, which demonstrated that approximately 33.3% of FDR exhibited abnormal corneal topography, suggesting the presence of a subclinical phase of KC. Significant differences between the parents of patients with keratoconus and healthy individuals were also found for K1, ARTmax, Final D, and TP. Furthermore, Awwad et al and Kaya et al reported that the central corneal thickness, thinnest pachymetry, and steepest keratometry measurements were significantly different between the FDR and normal groups.²¹ In addition, Shneor et al discovered that curvature abnormalities of the posterior and anterior corneas were significantly higher in first-degree relatives of individuals with KC than in the control group.¹² Although there were no substantial differences in anterior surface features, differences were observed in posterior surface elevation values.^22,23 These features could be valuable for identifying early ectasia in FDR.

Previous studies have demonstrated that the biomechanical parameters of the cornea in patients with keratoconus differ significantly from those of a normal cornea when using the Oculus Corvis ST device.^24,25 In our study, we observed significant variations in CCT, SPA1, DA, A1V, A2V, A1T, HCPD, and HCDA in FDR. Previously, Li et al¹⁷ reported similar findings, where CCT, SP-A1, ARTh, IR, DA ratio, CBI, TBI, and SSI were significantly different between the parents of patients with keratoconus and healthy individuals. In another study, the biomechanical indices of the proband's FDR showed irregular measurements of CBI, DA ratio, SP-A1, IR, bIOP, and TBI, indicating the early signs of keratoconus.²⁶ In most research studies, CBI and TBI are widely regarded as the most valuable parameters for identifying keratoconus.^27,28 Our findings suggest that first-degree relatives show decreased corneal biomechanical strength and reduced stiffness compared with the average population. For screening keratoconus at its earliest stages, clinical biomechanical evaluations have become crucial and should be applied to both first-degree relatives and individuals with high-risk factors for keratoconus.

Previous studies have revealed that the corneal thickness in patients with keratoconus, as assessed by optical coherence tomography (OCT), varies significantly from that of healthy individuals.²⁹ There is minimal variation in corneal pachymetry OCT parameters between first-degree relatives (FDR) and normal healthy group. This observation may be attributed to the fact that, in the initial stages of ectasia, corneal thinning has not yet commenced. At this stage, only the posterior segment of the cornea exhibits alterations. Consequently, in the early phases, the posterior corneal surface is affected, whereas the overall thickness remains unchanged. However, diagnostic tools such as the Oculus Pentacam and Oculus Corvis ST reveal distinct differences between family members and healthy individuals. Previous literature review revealed no studies on optical coherence tomography in first-degree relatives or family members with keratoconus. Nonetheless, anterior segment OCT, like other corneal imaging techniques, consistently and dependably measures corneal and epithelial thicknesses in both central and peripheral areas of the cornea in patients with keratoconus, facilitating precise and accurate assessment.³⁰

A strong heritability factor of keratoconus among their first-degree family members, where relatives of patients with keratoconus had an elevated risk of possessing a keratoconus trait, explained the possible cause of corneal abnormalities in first-degree relatives.³¹ Furthermore, previous genetic study had identified HMX1, SLC4A11, TGFBI, PIKFYVE, and ZEB1 variants in Chinese families to be associated with keratoconus.²⁶ Furthermore, research conducted by Levy et al³² revealed that 2 distinct corneal configurations, designated as J and Jinv, were found to be overrepresented among clinically unaffected family members of individuals diagnosed with familial keratoconus. Research strongly indicates that the development of keratoconus in FDR of affected individuals is influenced by both genetic susceptibility and environmental elements. Recognizing these predisposing factors is crucial for implementing early detection and treatment strategies within families that have a history of keratoconus. To minimize the risks associated with this condition, it is important to conduct routine examinations and be aware of potential environmental triggers.

Our research revealed that Oculus Pentacam corneal topography parameters exhibit remarkable diagnostic capabilities for identifying keratoconus in healthy individuals, which has also been highlighted in previous studies.^5,25 Among FDR, the B.Ele.Th model demonstrated the highest predictive capacity, achieving an AUC of 0.990. Furthermore, the BAD_D index delivered the second-best predictive performance for identifying keratoconus, with an AUC of 0.989. A previous study revealed that the parameters of BAD_D, K1, and ARTmax were sufficient to differentiate between the parent group of keratoconus and the healthy parent group.¹⁷ In contrast, different reported results showing that ARTave had the highest diagnostic value in identifying patients with keratoconus among their parents, with an area under the curve (AUC) of 0.705.²³ Furthermore, BAD_D model exhibited the best predictive performance, to detect keratoconus in normal has been demonstrated in the previous study.^33,34

Our research showed that CBI, TBI, Arth, and IR can consistently distinguish keratoconus from an FDR's cornea, attaining an impressive AUC score of >0.90. TBI, CBI, SP-A1, ARTh, IR, DA, and ARTmax exhibited a higher degree of discriminatory power between the parental keratoconus and healthy paternal groups, showing statistically significant differences.¹⁷ In addition, CBI, TBI, Arth, CCT, SPA1, and IR parameters from the Oculus Corvis ST were able to accurately differentiate between keratoconus and a normal cornea, achieving an excellent AUC score.³⁵ Previous research has shown that epithelial thickness data from anterior segment OCT can enable earlier diagnosis of keratoconus.^25,36 In our study, the parameters of the pachymetry thickness map (Min, Avg, and S-I) and epithelial thickness map (Min) effectively differentiated between keratoconus and a first-degree relative cornea, as evidenced by an AUC score exceeding 0.80, with no prior studies for comparison. The Min parameter, which was derived from the pachymetry thickness map, demonstrated an optimal prediction model with an AUC of 0.974.³⁷ Several key indicators are crucial for assessing the risk of keratoconus in first-degree relatives, including the Belin/Ambrosio Ectasia Detection Display (BAD_D), the corneal biomechanical index (CBI), the tomographic biomechanical index (TBI), back elevation at the thinnest point (B.Ele.Th), thinnest pachymetry (TP), and minimum pachymetry.

In summary, for the Pentacam, the parameters B.Ele.Th and BAD-D exhibit the highest sensitivity in detecting keratoconus in FDR, whereas in normal individuals, BAD-D and F.Ele.Th are the most sensitive parameters for keratoconus detection. B.Ele.Th is particularly sensitive in identifying early ectasia changes in the cornea of FDR, indicating abnormal protrusion or ectasia on the posterior corneal surface, which is a hallmark of keratoconus progression.³⁸ Regarding the Corvis ST, both TBI and CBI are crucial in identifying keratoconus among FDR and normal individuals. However, ARTh plays a more significant role compared with CCT in detecting keratoconus in FDR as opposed to normal individuals. This finding supports the notion that FDR tend to have lower ARTh values, indicating a thinner and more deformable cornea.³⁹

This research uses 3 commonly employed imaging methods to develop the most precise predictive model for detecting keratoconus in FDR of keratoconus. The techniques include corneal topography using Oculus Pentacam HR, corneal biomechanics assessment with Oculus Corvis ST, and corneal pachymetry measurement via Anterior Segment OCTA Cirrus 5000. The combined SLR model of IHD, CBI, and TBI showed excellent accuracy of AUC: 0.999 to detect keratoconus in FDR. Earlier research proposed logistic regression model of combined CBI, TBI, and TP with demonstrated sensitivity of 71.6% and specificity of 75.0% to detect keratoconus in FDR.¹⁷ Our findings showed that predictive power of combined tomographic and biomechanical parameters surpassed that of either parameter set alone, exhibiting enhanced accuracy.³⁹ Advancements in multimodal imaging have significantly enhanced the accuracy of keratoconus detection and facilitated earlier intervention. By integrating various imaging techniques, a comprehensive view of corneal structure and function can be achieved, thereby improving early keratoconus detection.⁴⁰ This study sought to underscore the significance of notifying immediate family members about their susceptibility to ectasia in keratoconus. By examining asymptomatic relatives, it is possible to identify keratoconus in the early or subclinical stages. Comprehensive imaging using multiple modalities also plays a crucial role in evaluating first-degree relatives of patients with keratoconus before refractive surgery, helping to prevent postoperative ectasia. Detecting subtle signs of ectasia or compromised biomechanical integrity can aid in averting surgical complications.

ACNOWLEDGMENT

The authors acknowledge the Faculty of Medicine at the National University of Malaysia for funding this study.

Supplementary Material

cornea-45-541-s001.docx ^{(16.9KB, docx)}

cornea-45-541-s002.docx ^{(15.8KB, docx)}