Measurement of the Tilt Angle of the Optic Disc Using Spectral-Domain Optical Coherence Tomography and Related Factors in Myopia

Abstract

Purpose: This study presents a novel, three-dimensional method for measuring the tilt angle of the tilted optic disc (TOD) using spectral-domain optical coherence tomography (SD-OCT) and investigates the correlation between ocular-related parameters and TOD.

Methods: We included the right eyes of 243 healthy young individuals, categorized by axial length. We measured the ovality index (OI) and dihedral angle (DA) using SD-OCT infrared ray fundus photographs and high-resolution cross-sectional images of the optic disc, respectively. The relationships between OI, DA, and ocular-related parameters were analyzed.

Results: Eyes in the longer axial length group exhibited a lower OI and a higher DA, along with thinner nasal and inferonasal circumpapillary retinal nerve fiber layer (cpRNFL) and thicker temporal and superotemporal cpRNFL. There was a significant relationship between DA and cpRNFL thickness. The new method utilizing DA to measure the tilt angle of TOD demonstrated high repeatability.

Conclusions: We propose a novel, three-dimensional, and quantitative method for evaluating the tilt degree of TOD. A higher degree of myopia indicated a greater tilt angle of the TOD, and a greater TOD suggested additional changes in cpRNFL thickness. These findings should be considered when interpreting increased susceptibility and early assessment of glaucoma in myopia.

Translational Relevance: DA could serve as a superior indicator for describing TOD morphology during eyeball elongation and evaluating its impact on related parameters of the optic disc and peripapillary structures in the myopic population.

Introduction

Morphology of the optic disc plays an important role in the diagnosis of ocular diseases. As a morphologic characteristic of the optic disc in myopia,¹ a tilted optic disc (TOD) is commonly considered a conformational alteration resulting from posterior scleral remodeling associated with axial elongation of the eyeball.² With the escalating global prevalence of myopia,³ TOD has become increasingly common and garnered significant attention in recent years. Effective TOD observation and measurement not only reveal the visible abnormal appearance of the optic disc, aiding in the evaluation of myopia progression, but also reflect asymmetrical histologic changes in both the optic disc and peripapillary structures with increasing axial length (AL).^4–6 Several studies have established a close relationship between the tilt degree of TOD and damage to peripapillary retinal nerve fiber layer (RNFL) thickness, ultimately leading to vision loss or visual field defects.^7–10 Given the crucial role of RNFL thickness in the diagnosis and prognosis of glaucoma, quantifying TOD contributes to identifying the clinical significance of myopia in glaucoma pathogenesis, particularly in revealing potential mechanisms for heightened susceptibility and early assessment of glaucomatous optic neuropathy in myopia. 

Previous studies have determined the presence of TOD through subjective estimation by examiners or optic disc stereophotographs.^11,12 In 1997, Jonas et al.¹³ introduced an elliptic parameter calculated as the ratio of minimal to maximal disc diameter. Subsequently, the optic disc ovality index (OI) emerged as a traditionally employed surrogate indicator of relevant tilt, aiming to objectively and qualitatively define TOD. However, inconsistencies in different investigations, such as controversial criteria for defining TOD incidence^14,15 and variations in clinical disc margin determination,¹⁶ may introduce errors and compromise the reliability of TOD evaluation. Importantly, the inherent limitation of OI as a two-dimensional optic disc-related morphologic parameter fails to highlight the actual three-dimensional and biometric features of the optic disc, limiting its broader clinical application. Consequently, additional indicators are required to enhance TOD measurement. 

The advancement of optical coherence tomography (OCT), facilitating the acquisition of high-resolution images of every layer of the choroid and retina and enabling the precise analysis of the anatomic construction of the optic disc and peridisc without contact operation,¹⁷ has considerably contributed to the improved assessment of TOD and its degree of tilt. In various studies focusing on the quantification of TOD using OCT,^5,7,8,18,19 different tilt angles have been defined and measured based on distinct configurational landmarks from high-resolution cross-sectional images of the optic disc. Among these, Bruch's membrane openings (BMOs) have been the most commonly used reference points due to their representation of critical optic disc anatomy and histologic implications regarding the impact of myopic-related axial extension on glaucoma occurrence and development.^20–22 However, a limitation of these established OCT-based approaches is that measuring the tilt angle of TOD horizontally or vertically only indicates the tilt degree in a specific direction, thereby leaving a gap in the comprehensive survey of disc stereo-structure. To address this limitation and achieve a dimensional transformation for accurate TOD evaluation, we analyzed spectral-domain OCT (SD-OCT) scan images in myopia, leveraging the obtained cross-sectional two-dimensional parameters to devise a novel method for measuring the three-dimensional tilt angle of TOD. This approach aims to provide additional beneficial information on the association between myopia and specific ocular diseases for clinical practice. The primary focus of this study is to explore the relationship between the novel tilt angle of TOD and RNFL thickness, contributing to the speculation of possible mechanisms explaining glaucoma diagnosis and prognosis in myopia. 

Methods

This study conformed to the principles of the Declaration of Helsinki, and all procedures were approved by the Ethics Committee of the First Affiliated Hospital of China Medical University. The purpose of this study was communicated to all participants, and informed consent was obtained. 

A comprehensive medical record focusing on the history of ocular diseases and a routine ophthalmic examination including automated refraction (spherical equivalent [SE]), uncorrected and best-corrected logarithm of minimum angle of resolution (logMAR) visual acuity, intraocular pressure (IOP), detailed slit-lamp biomicroscopy of the ocular anterior segment, and ophthalmoscopy of the fundus examinations were conducted on volunteers recruited between May 2020 and May 2021 (mainly from China Medical University). A total of 243 participants were included in this study who met the following inclusion criteria: individuals with no diagnostic systemic disease aged 20 to 30 years; exclusion of any ocular diseases, such as neuro-ophthalmic diseases, uveitis, and diabetic retinopathy, on slit-lamp and fundus microscope examination; best-corrected logMAR visual acuity ≥0.096; IOP ≤21 mm Hg; normal anterior chamber angle; no history of ophthalmic surgery; and no family history of glaucoma in first-degree relatives. 

The right eye was selected for subsequent ophthalmologic examinations performed by a single technician engaged in specialized work for more than 5 years. AL was measured using the IOLMaster 500 (Carl Zeiss Meditec, Dublin, CA, USA), and the average of five consecutive measurements was used as the final AL for each participant. The central corneal thickness (CCT), anterior chamber depth (ACD), and anterior chamber angle (ACA) were measured using a computer-assisted corneal topographic analysis system (Pentacam; Oculus, Wetzlar, Germany). Biometric features of the optic disc and peripapillary structures were described with SD-OCT (Hra2 Spectralis2; Heidelberg Engineering GmbH; Heidelberg, Germany) for further data analysis. 

All participants were divided into three groups according to their AL based on the axial myopia categories proposed by Rezapour et al.²³: no axial myopia (group 1, AL ≤ 24.0 mm), mild axial myopia (group 2, 24.0 mm < AL ≤ 26.0 mm), and high axial myopia (group 3, AL > 26.0 mm). This approach enables the assessment of morphologic alterations of the optic disc in association with AL and helps mitigate the impact of refractive changes in eyes resulting from cataract surgery or refractive procedures. 

Image Acquisition

Optic nerve head (ONH) radial circle (ONHRC) scans from the Spectralis Glaucoma Module Premier Edition (version 6.10; Heidelberg Engineering Inc., Heidelberg, Germany), and single-line horizontal and vertical scans were acquired for all participants. To ensure good imaging quality, all participants were asked to sit quietly and fix a reference light with their heads in a straight position when the instrument started scanning. Images of low quality or poor centering, determined by the distance between the BMO centers of the ONHRC scans and the center of the image, were either repeated or manually corrected by another experienced ophthalmic technician. 

The ONHRC scan comprises 24 high-resolution ONH radial scans and 3 RNFL circle scans. The thickness of circumpapillary RNFL (cpRNFL) was measured according to the 3.5 mm diameter circle obtained from the three circle scans at approximately 3.5 mm, 4.1 mm, and 4.7 mm diameters using instrument software. By comparing the eyes of the participants to a reference database of normal eyes, a sectorial pie chart of cpRNFL thickness with relevant changes was obtained, including average, nasal, superonasal, inferonasal, temporal, superotemporal, and inferotemporal cpRNFL thickness parameters (Fig. 1). It is noteworthy that nearly all pertinent studies that focused on quantifying the tilt angle of the optic disc through OCT have reported the use of high-resolution cross-sectional images of the optic disc for measurement. In line with the original intention to maintain this methodologic commonality, single-line B-scans along the reference line (connecting the central point of the BMOs and macular fovea) and its vertical line were selected as the high-resolution cross-sectional images of the optic disc in our proposed novel method (Figs. 2A, 2C). 

Figure 1.

 

SPECTRALIS OCT peripapillary RNFL printout report (3.5 mm diameter RNFL circle scan). (A) Grid showing the average cpRNFL thickness in six zones around the nerve. (B) Green represents normal cpRNFL thickness (P > 0.05). (C) Yellow represents a critical cpRNFL thickness on one side (0.01 ≤ P < 0.05). (D) Red represents a side of cpRNFL thickness that was apparently thin (P < 0.01).

Figure 2.

 

Measurement of the two-dimensional tilt angles of a tilted optic disc for the definition and calculation of the DA. The single-line scan pattern of SD-OCT was used to obtain high-resolution cross-sectional B-scan images along the reference line connecting the center of the macular fovea and the optic disc (A) and its vertical line (C). The lines (line EG and eg) connecting the inner tips of the Bruch's membrane on each side of the optic disc were drawn as the optic disc plane. The horizontal imaginary lines (line EF and ef) were set as the macular fovea–optic disc center line on behalf of the coronal plane of the posterior pole of the eyeball. (B) The tilt angle α was defined as the angle between the optic disc plane (line EG) and the coronal plane (line EF). (D) The tilt angle γ was defined as the angle between the optic disc plane (line eg) and the coronal plane (line ef) in the vertical direction.

The magnification effect in each eye was corrected using the Littmann–Bennett method,²⁴ which determines the adjusted analytical area for each fundus image and the true size of fundus parameters when entering the AL data into an established formula. 

Determination of Ovality Ratio and Dihedral Angle

The magnification-corrected OCT infrared ray fundus photographs and single-line B-scan images were exported to ImageJ version 1.47 (ImageJ Software, National Institutes of Health, Bethesda, MD, USA). For measuring the OI, the longest and shortest lines crossing the center of the optic disc and terminating at the BMO were manually demarcated on the infrared ray fundus photograph. These lines represented the maximum and minimum diameters of the optic disc, respectively, and OI was calculated as the ratio of the minimum to the maximum optic disc diameter (Fig. 3). 

Figure 3.

 

Measurement of the OI. SD-OCT system automatically detected the edge of BMO as the boundary of the optic disc and then the center of the optic disc. The BMO image (red dotted circles) was regarded as the infrared ray (IR) fundus photographs from SD-OCT. The lengths of the lines passing through the center of the optic disc and ending at the BMO were measured. The longest line (yellow line) and the shortest line (green line) were defined as the maximum and minimum diameter of the optic disc, respectively. OI was calculated as the ratio of minimum to maximum optic disc diameter.

For measuring the tilt degree of the TOD, we marked two lines on each B-scan image to depict the position lines of two defined planes. Instead of using the clinically defined disc margin from a fundus photograph, we conducted custom calculations according to the BMO to emphasize the tilted configuration of the optic disc.²⁵ We drew the lines connecting the inner tips of the Bruch's membrane (BM) on each side of the optic disc as the optic disc position lines (lines EG and eg), the anatomic characteristics of which imply shearing deformations of the ONH and peripapillary tissue, verified by histologic examinations. Simultaneously, horizontal imaginary lines (lines EF and ef) projected from the single-line scanned reference lines and their vertical lines represented the coronal position lines of the eyeball posterior pole. 

A single position line can generally represent a plane. Following established OCT-based measurements, the horizontal tilt angle (angle α) was defined as that between the optic disc plane (line EG) and the coronal plane (line EF) in the B-scan images along the reference line (Fig. 2B). We defined the vertical tilt angle (angle γ) as that between the optic disc plane (line eg) and the coronal plane (line ef) in the B-scan images perpendicular to the reference line (Fig. 2D). Recognizing the stereo-structure of the optic disc, we found that reproducing the positional information of the four abovementioned lines in the spatial coordinate system based on the single-line scan reference line and its vertical line could improve the tilted parameter dimensions. Therefore, considering only two-dimensional tilt angle data are obtained, the angle between the two defined planes that is determined by the two intersecting position lines could be further calculated. Eventually, the lifting angle of the inclined plane relative to the horizontal plane was also obtained. Supported by a mathematical model of the dihedral angle (DA), we essentially realized this idea and obtained a novel tilt parameter, DA, via a series of mathematical transformations. As depicted in the schematic diagram (Fig. 4), DA (angle θ) is defined as the tilt angle between the two defined planes, considered the approximate fitting planes of the optic disc and coronal planes of the eyeball posterior pole. Therefore, DA contributes to the effective evaluation of the tilt degree of the three-dimensional plane of TOD. 

Figure 4.

 

A schematic diagram of the DA (angle θ) between the optic disc plane and the coronal plane of the eyeball. The BMO lines in the transverse (OM, blue line) and longitudinal (ON, green line) sections represented the optic disc plane (yellow plane). O´M (blue dotted line) and O´N (green dotted line) represented the coronal plane (horizontal plane) in the transverse and longitudinal sections, respectively. The definitions of tilt angle α and tilt angle γ are presented in Figure 2.

The formula for calculating the DA (angle θ) was as follows:  

\begin{eqnarray*} && \theta = {\cos ^{ - 1}}\left| {\frac{{\cos \left( {\frac{\alpha }{{180^\circ }} \times \pi } \right) * \cos \left( {\frac{\gamma }{{180^\circ }} \times \pi } \right)}}{{\sqrt {\cos {{\left( {\frac{\alpha }{{180^\circ }} \times \pi } \right)}^2} + \cos {{\left( {\frac{\gamma }{{180^\circ }} \times \pi } \right)}^2} \times \sin {{\left( {\frac{\alpha }{{180^\circ }} \times \pi } \right)}^2}} }}} \right| \\ && \div \pi \times 180^\circ \end{eqnarray*}

 

The OI and DA measurements were performed separately by two experienced clinicians without any information about each other. All measurements were repeated twice, and OCT images were randomly arranged for each measurement. The second measurement was started 2 weeks after the first measurement was completed. 

Measurement Reproducibility

As is commonly known, in OCT B-scans, the BMO is defined as the innermost termination of the BM or the BM-retinal pigment epithelium (RPE) complex, identified as the hyperreflective signal immediately above the choroidal layer. Its innermost termination is marked by a pixel location exhibiting an abrupt drop in signal intensity of the hyperreflective stripe. However, it is crucial to acknowledge that accurate BMO segmentation may be influenced by several factors,²⁶ with one of the most prominent being the presence of parapapillary atrophy (PPA), particularly β PPA, where atrophy of the RPE, choriocapillaris, and medium-sized choroidal vessels renders the interface between the BM and the choroidal layer indistinct. Given that BMO determination is pivotal for measuring the DA and OI, it is important to note that BMO positions were selected through a semiautomated OCT-based method, and the variability in BMO position selection should be considered. Initially, BMO points were automatically detected during the ONHRC scan process using Spectralis Glaucoma Module Premier Edition (GMPE) software. Subsequently, the BMO of 48 meridians in the 24 B-scans of each eye were independently examined and manually corrected by two trained observers. Additionally, BMO was semiautomatically traced using ImageJ in single-line B-scan images by two masked observers. This tracing included the macular fovea–optic disc center radial scan and its vertical scan within the 24 B-scans. Therefore, the repeatability of the novel measurement using DA and the traditional measurement using OI was required, which was assessed using the intraclass correlation coefficient (ICC) and Bland–Altman analyses, based on the data of 60 participants selected from all 243 participants with simple random sampling. Because the orders of magnitude and units between OI and DA were different, a z-score calculation was first performed in this study by applying the mean (µ) and SD (δ) values of the population data to define the computational formula (OI/DA-µ)/δ and transform the measured data into standard z-scores, so as to enable the comparability of OI and DA. The intraobserver correlation coefficients were calculated using a one-way random-effects model, and the interobserver correlation coefficients were calculated using a two-way random-effects model. The ICC estimates of agreement were categorized as poor (0.01–0.39), fair (0.40–0.59), good (0.60–0.74), and excellent (0.75–1.00).²⁷ However, in clinical applications, an ICC value >0.9 indicates good reliability and repeatability.²⁸ Bland–Altman plots were generated to evaluate the agreement between the measurements of the two observers or the two measurements of each observer, which could demonstrate the difference in reproducibility between DA and OI for measuring the tilt degree of TOD in the inter- or intraobserver comparison. The limits of agreement (LOAs) were defined as the mean ± 1.96 × SD of the difference between the measured and remeasured values in inter- or intraobserver comparisons. 

Statistical Analyses

Statistical analyses were conducted using IBM SPSS Statistics for Windows (version 25.0; IBM Corp., Armonk, NY, USA) and MedCalc version 15.6. (MedCalc Software, Ostend, Belgium). One-way analysis of variance (ANOVA) and Welch's test were employed to compare intergroup differences in the measured data. Multiple comparisons between groups were performed using the least significant difference (LSD) t-test. Furthermore, the relationships between OI, DA, and ocular-related parameters were analyzed using Pearson's correlation analysis. We conducted univariate and multivariate (adjusted for SE and AL) linear regression analyses to identify the association between the TOD and cpRNFL thickness. The dependent variables included average, nasal, superonasal, inferonasal, temporal, superotemporal, and inferotemporal cpRNFL thickness, respectively. OI and DA were considered independent variables. In all analyses, P values < 0.05 were considered statistically significant. 

Results

The right eyes of 243 individuals (84 men and 159 women) with an average age 24.66 ± 1.73 years were analyzed in this study. The median AL and SE was 25.15 ± 1.29 mm and −3.90 ± 2.92 D, respectively. For the traditional method using OI to measure the tilt degree of TOD, the average OI value was 0.84 ± 0.09. For the new method, the average DA value was 11.45° ± 3.89°. 

There were 44 eyes in the no axial myopia group (group 1), 139 eyes in the mild axial myopia group (group 2), and 60 eyes in the high axial myopia group (group 3) (Table 1). The average SE of the three myopia groups increased (all P < 0.05). Compared to group 1 and group 2, the average DA and ACD of group 3 were significantly higher (13.48° ± 3.62° and 3.42 ± 0.28 mm, all P < 0.001). Group 3 had lower average OI (0.82 ± 0.09, P < 0.05) and higher average ACA (39.19° ± 5.01°, P < 0.05) than group 1. There was no significant difference in the average age (P = 0.446) and CCT (P = 0.108) between the three groups. 

Table 1.

 

Demographics of Enrolled Participants Among the Three Groups in This Study

 

Table 2 summarizes the characteristics of overall average and six sectoral cpRNFL thicknesses among the three myopia groups. The average nasal (P < 0.001), inferonasal (P < 0.001), temporal (P < 0.001), superotemporal (P = 0.003), and inferotemporal (P = 0.019) cpRNFL thickness showed significant differences among the three groups. The average nasal and inferonasal cpRNFL thickness (55.45 ± 16.22 µm and 92.10 ± 18.18 µm, all P < 0.05) were significantly thinner, while the average temporal cpRNFL thickness (106.65 ± 24.57 µm, P < 0.05) was significantly thicker in group 3 compared to group 1 and group 2. Group 3 had thicker average superotemporal cpRNFL thickness (164.48 ± 25.32 µm, P < 0.05) than group 1 and group 2. The superonasal (P = 0.569) and overall average cpRNFL thickness (P = 0.065) were not significantly different between the three groups, but group 3 had thinner overall average cpRNFL thickness (103.38 ± 12.43 µm, P < 0.05) than group 1. 

Table 2.

 

Characteristics of Overall Average and Six Sectoral cpRNFL Thicknesses by Myopia Group

 

The relationship between OI, DA, and ocular-related parameters of the total participants was assessed in Table 3. There was a significant negative correlation between OI and DA (R = −0.317, P < 0.001). OI was significantly negatively correlated with AL (R = −0.181, P = 0.005), and DA was significantly positively correlated with AL (R = 0.442, P < 0.001). Similarly, OI was significantly positively correlated with SE (R = 0.236, P < 0.001), while DA was significantly negatively correlated with SE (R = −0.457, P < 0.001). The nasal cpRNFL thickness was significantly correlated with OI (R = 0.283, P < 0.001) and DA (R = −0.512, P < 0.001). The superonasal and inferonasal cpRNFL thickness were significantly negatively correlated with DA (R = −0.297 and R = −0.294, all P < 0.001). The temporal, superotemporal, and inferotemporal cpRNFL thickness were significantly negatively correlated with OI (R = −0.290, P < 0.001; R = −0.220, P = 0.001; R = −0.252, P < 0.001) and were significantly positively correlated with DA (R = 0.491, P < 0.001; R = 0.223, P < 0.001; R = 0.189, P = 0.003). Additionally, there was no significant correlation between the overall average cpRNFL thickness and OI (P = 0.236), as well as DA (P = 0.273). 

Table 3.

 

Relationship between OI, DA, and Ocular-Related Parameters of the Total Participants in This Study

 

Table 4.

 

Association between DA, OI, and cpRNFL Thickness of the Three Myopia Groups and All Study Participants, While Controlling for AL and SE

 

Linear regression analyses were conducted separately for each myopia group, as well as for the total participants, including the three myopia groups, in order to compare the association between TOD and the cpRNFL thickness (Table 4). On univariate and multivariate analyses adjusted for AL and SE in group 1, only DA was significantly associated with the temporal and inferotemporal cpRNFL thickness (the standardized regression coefficient β = 0.370, P = 0.013; β = 0.317, P = 0.036). In group 2, the nasal and superonasal cpRNFL thickness decreased (i.e., became more thinner) (β = −0.424, P < 0.001 and β = 0.378, P < 0.001; β = −0.401, P < 0.001 and β = 0.172, P = 0.042) and the temporal, superotemporal, and inferotemporal cpRNFL thickness increased (β = 0.515, P < 0.001 and β = −0.345, P < 0.001; β = 0.210, P = 0.013 and β = −0.194, P = 0.022; β = 0.227, P = 0.007 and β = −0.312, P < 0.001) with larger DA and small OI. While controlling for AL, DA and OI were significantly associated with the nasal and temporal cpRNFL thickness (β = −0.330, P = 0.001 and β = 0.271, P < 0.001; β = 0.448, P = 0.008 and β = −0.205, P < 0.001). In group 3, both the univariate and multivariate analyses adjusted for AL and SE showed DA was significantly associated with the nasal and inferotemporal cpRNFL thickness (β = −0.516, β = −0.530, and β = −0.531, all P < 0.001; β = 0.298, P = 0.021 and β = 0.274, P = 0.027 and β = 0.271, P = 0.040). OI was significantly associated with the superotemporal cpRNFL thickness (β = −0.284, P = 0.028; β = −0.271, P = 0.046) on the univariate and multivariate analyses adjusted for AL. 

As for the linear regression analysis of all study participants, the univariate regression analyses of all study participants showed the results that align with the correlation analyses presented in Table 3 (Fig. 5). Further multivariate analyses were performed, adjusting for AL and SE individually, on all study participants (Table 4). On multivariate analyses adjusted for AL, we found DA to be significantly associated with nasal, superonasal, inferonasal, temporal, and inferotemporal cpRNFL thickness with β values of −0.355, −0.315, −0.131, 0.316, and 0.234, respectively. OI was significantly associated with the nasal, temporal, superotemporal, and inferotemporal cpRNFL thickness (β = 0.123, P = 0.029; β = −0.141, P = 0.015; β = −1.161, P = 0.014; β = −0.226, P < 0.001). As for the multivariate analyses adjusted for SE, the results were consistent with the above findings when controlling for AL, with the exception that OI was not associated with nasal cpRNFL thickness (P = 0.050). 

Figure 5.

 

Scatterplots showing the correlation between the cpRNFL thickness and the DA. β indicates the standardized regression coefficient. The 95% confidence interval (95% CI) is based on the unstandardized regression coefficient. (A) The correlation between nasal cpRNFL thickness and DA (β = –0.512, r² = 0.262, P < 0.001). (B) The correlation between superonasal cpRNFL thickness and DA (β = –0.297, r² = 0.088, P < 0.001). (C) The correlation between inferonasal cpRNFL thickness and DA (β = –0.294, r² = 0.087, P < 0.001). (D) The correlation between temporal cpRNFL thickness and DA (β = 0.491, r² = 0.241, P < 0.001). (E) The correlation between superotemporal cpRNFL thickness and DA (β = 0.233, r² = 0.050, P < 0.001). (F) The correlation between inferotemporal cpRNFL thickness and DA (β = 0.189, r² = 0.036, P = 0.003). (G) The correlation between overall average cpRNFL thickness and DA (β = –0.071, r² = 0.005, P = 0.273).

Regarding the repeatability of the traditional method measured by OI and the new method measured by DA, the following results of Table 5 were based on the z-scores of OI and DA. The intra- and interobserver repeatability of the OI and DA measurements were determined. 

Table 5.

 

Repeatability of the Method Measured by OI and DA

 

As Bland and Altman²⁹ pointed out, the points in the scatterplot must lie along a straight line to achieve perfect correlation. The correlation shown by the Bland–Altman plots depends on the range of the measurement results. If this is small, the correlation is usually greater than if it is large. This study reported that the inter- and intraobserver 95% LOA of the OI was –0.97 to 0.97 (P < 0.001) and −1.2 to 1.2 (P < 0.001), respectively, while the inter- and intraobserver 95% LOA of the DA was −0.34 to 0.34 (P < 0.001) and −0.27 to 0.27 (P < 0.001), respectively (Table 5, Fig. 6). The results of the above analyses showed that DA had a larger ICC value and a smaller range of the Bland–Altman plots than OI. This means that the repeatability of the method measured by DA was superior to the repeatability of the method measured by OI; hence, DA would be a more significant parameter for the evaluation of the tilt degree of TOD in clinical practice. 

Figure 6.

 

Bland–Altman plots of the tilt degree of titled optic disc measured with (A) OI by the same observer (intraobserver), (B) DA by the same observer (intraobserver), (C) OI by different observers (interobserver), and (D) DA by different observers (interobserver).

Discussion

The primary objective of the first question in this study was to optimize the measurement of TOD, particularly focusing on enhancing parameter dimension during an era when the evaluation of optic disc tilt predominantly relies on OCT technology rather than subjective observations based on photography or fundus examinations. By adopting a unique design principle that involves dividing the TOD oblique plane and constructing a mathematical model based on two distinct yet interconnected tilt angles measured through high-resolution cross-sectional SD-OCT images of the optic disc, we introduced a novel and three-dimensional method for objectively calculating the tilt angle of TOD in the myopic population. It is essential to emphasize that the definition of the DA parameter originated from the initial definition of TOD, referring to the elevation of one side of the optic disc surface along vertical or horizontal axes. The establishment of the DA formula not only retained the two-dimensional tilt angle parameters obtained from prior OCT-based approaches but also accurately restored the planar construction of the disc. These methodologic innovations and enhancements enabled us to dimensionally transform TOD assessment, providing beneficial information to reflect the tilt degree of the optic disc plane relative to the coronal plane of the posterior pole of the eyeball. Additionally, our study revealed a more tilted optic disc tended to be associated with a longer AL. Considering the connection between deformations of the optic disc and peripapillary region and the scleral extension on the posterior eyeball segment during the axial growth of myopic eyes, our novel TOD measurement contributes to further exploring potential pathologic changes caused by myopic-related axial extension. It also helps clarify the essential role of myopia in the occurrence and development of certain ocular diseases. 

In general, TOD is diagnosed when one margin of the optic disc is raised above the opposite margin by subjectively observing the fundus stereophotographs. As a relatively objective two-dimensional parameter and qualitative indicator of TOD among various studies, our study employed the ratio of the minimum to maximum optic disc diameter as OI, and the average ratio was approximately 0.8, which was inconsistent with some reported results.^15,30,31 In terms of the validity of OI as an index of actual tilt degree, we found that OI was significantly correlated with DA. This showed that, to some extent, the method using OI has almost become the traditional standard for the diagnosis of TOD, which is consistent with the conclusions of most previous studies.^14,20,32 Meanwhile, our results demonstrated comparably higher intraobserver repeatability as well as fairly excellent interobserver repeatability of the DA we first put forward (ICC > 0.9), which convincingly signifies the clinical application value of the new assessment for TOD. The range of the Bland–Altman plots between the DA and OI indicated less agreement with the traditional measurement for the tilt angle and emphasized the need for a quantitative and standardized method for the measurement of tilt degree of TOD. 

It is widely recognized that the rapid, noninvasive, and high-resolution OCT technology significantly enhances the ability to reconstruct the optic disc and peripapillary histologic structure, comparable to pathologic biopsy. This technological advancement greatly facilitates the description of tilt angles in TOD. However, due to essential distinctions in the definitions of tilt angles across several studies on TOD quantification, it may be impractical to indiscriminately compare and discuss the variations in these tilt angle values obtained through diverse methods, including the novel method described in our study. For instance, we demonstrated in our study that the tilt angle of TOD in young Chinese individuals with myopia ranged from 1.50° to 18.14°. In contrast, another study focusing on three-dimensional tilt parameters reported that the mean BMO tilt angle in the high myopia axial group was 3.4° (3.1°–3.8°).²⁰ When considering the most commonly defined two-dimensional tilt angles, a concise summary of their related methods revealed more value in understanding methodologic similarities and differences, particularly in the selection of the two defined planes. Hosseini et al.¹⁸ reported a method dependent on the clinical disc margin to divide the corresponding position of the optic disc in a typical manner. Yoon et al.⁵ and Marsh-Tootle et al.³³ used the optic disc canal plane to measure the optic disc tilt angle relative to the scleral canal or the scleral canal tilt angle. Furthermore, consistent selections of the BMO plane have been made to characterize the actual and imperceptible changes of the optic disc.^18,33–36 The concept of BMO was first proposed in 2012³⁷ using SD-OCT, revealing that BMO, through which retinal ganglion cell axons passed to exit the eye, represented an actual anatomic border of the optic disc and referenced a consistent anatomic structure within or between eyes. On the other hand, several studies have revealed that the AL and tilt degree of TOD are positively correlated,^{5,7,18,20,30,38} which is consistent with our findings. This suggests that TOD may often be accompanied by other features that are found more frequently in cases of myopia and the TOD tilt angle could indirectly reflect the scleral strain of the eyeball elongation. Measuring the tilt angle of BMO plane is valuable for estimating the forces leading to the enlargement of the BMO, understanding the development of BM defects, and comprehending the mechanisms underlying myopic degeneration. 

Evidence indicates that myopia can introduce challenges in understanding the true manifestation, progression, and prognosis of glaucomatous eyes.^5,8,22,39 Given that glaucoma is characterized by specific optic disc changes and RNFL thinning, understanding the relationship between TOD and RNFL thickness is crucial for interpreting the connection between myopia and glaucoma. RNFL defects resulting from TOD typically signify nerve fiber damage, leading to visual field (VF) defects, which could potentially explain the heightened susceptibility to glaucomatous optic neuropathy in myopia. Conversely, TOD-induced increase in cpRNFL thickness suggests that clinicians interpreting glaucomatous-related examination reports may encounter challenges due to additional changes in cpRNFL thickness caused by TOD. This is particularly relevant in eyes with normal-tension glaucoma and primary open-angle glaucoma, wherein early diagnosis primarily depends on the analysis of RNFL thickness and VF.^11,31,36,40 Myopic eyes have been reported to exhibit changes in retinal and cpRNFL thicknesses, including those with optic disc tilt.^7,41,42 Among the conflicting data reported in various studies, a consensus emerges that in myopic tilted optic disc groups, the temporal RNFL is thicker, and the nasal RNFL is thinner than in nontilted optic disc groups. Similarly, our results from cpRNFL thickness correlation analysis with OI and DA revealed a decrease in nasal cpRNFL thickness and an increase in temporal cpRNFL thickness with an increase in the TOD tilt degree. Studies suggest that mechanical stress with myopic axial extension may cause the optic disc to become temporally tilted and retinal dragging toward the temporal horizon,^7,9,41 which means the impact of AL on the thickening of the RNFL in the temporal quadrant and temporal dragging of the superior/inferior peak locations should be taken into consideration. The results of our study showed that there were significant differences in the cpRNFL thickness among the different AL groups (Table 2). These findings align with the preceding mentioned opinion. Even so, the multivariate regression analyses, which controlled for AL in each myopia group and the total study participants, revealed that DA was an independent factor associated with the cpRNFL thickness. This suggests that the tilt angle of TOD could explain the variance of the cpRNFL thickness. The linear regression analysis conducted in this study demonstrates an association between TOD and cpRNFL thickness, which is consistent with previous research^7,14,41 suggesting that TOD and its tilt degree have a notable impact on the characteristic of nasal and temporal cpRNFL thickness, similar to the effects of AL and SE. Additionally, both univariate and multivariate analyses, after adjusting for AL and SE, indicated that DA has a stronger influence on cpRNFL thickness compared to OI. 

However, it is important to acknowledge several limitations in the present study. First, our novel measurement lacks consideration of the directional difference in TOD. Based on the center of the optic disc, its subpoint, and defined reference lines, the DA solely represents a tilt angle between the central point of the BMOs and the fovea. Considering the asymmetrical forces around the myopic optic disc leading to the abnormal appearance of TOD,^43–45 determining the tilt direction would contribute to identifying the source of posterior scleral stress and analyzing the predominant tilt direction. Kim et al.² introduced the concept of the deepest point of the eyeball (DPE) and its quantification, providing information regarding the posterior scleral configuration that may help compensate for the deficiency in tilt direction. Therefore, further studies are warranted for the integration of DA and DPE in the evaluation of the comprehensive characteristics of myopic TOD. Second, using simple mathematical geometry to fit the optic disc plane by several BMO markings of the optic disc margin may result in a loss of structural information regarding the real optic disc plane. An advanced method, such as a scientific and integrated algorithm, is required to process all relevant markings, including BMO or other reference points, to obtain fitting planes and calculate the tilt angle between the two planes. 

In conclusion, we propose a novel, three-dimensional, and quantitative method for evaluating the tilt degree of TOD utilizing SD-OCT technology to procure high-resolution cross-sectional images of the optic discs and measuring two-dimensional tilt angles obtained from established OCT-based methods. This method effectively reflects the tilt angle by calculating the DA between the optic disc plane and the coronal plane of the eyeball. Our study suggests that DA could be a superior indicator for evaluating the morphologic characteristics of myopic optic discs as AL increases, providing a basis for further investigations into myopic-related pathologic changes. In understanding abnormal RNFL thickness, the influence of TOD and its tilt degree should be recognized in the field of myopic glaucoma. 

Acknowledgments

Disclosure: Y. Li, None; W. Jia, None; X. Liu, None; Y. Chen, None; H. Chen, None; G. Ren, None; S. Jiang, None; X. Ma, None