Impact of Rates of Change of Lamina Cribrosa and Optic Nerve Head Surface Depths on Visual Field Progression in Glaucoma

Abstract

Purpose: To investigate the impact of the rates of change of anterior lamina cribrosa surface depth (ALCSD) and optic nerve head surface depth (ONHSD) on visual field (VF) progression in glaucoma.

Methods: One hundred forty-six eyes of 95 glaucoma patients had optical coherence tomography ONH imaging and VF testing at approximately 4-month intervals for greater than or equal to 5 years. Anterior lamina cribrosa surface depth and ONHSD were measured with reference to (1) Bruch's membrane opening (BMO), and (2) choroid-sclera interface (CSI). The rates of change of ALCSD and ONHSD of individual eyes were measured with linear regression analysis. The hazard ratios (HRs) of the rates of change of ALCSD/ONHSD for prediction of VF progression as per Early Manifest Glaucoma Trial criteria were determined by joint longitudinal and survival models.

Results: Using the BMO reference, 23.3% and 28.1% of eyes showed a significant positive trend (posterior displacement), whereas 29.5% and 24.0% showed a significant negative trend (anterior displacement) of ALCSD and ONHSD, respectively. Using the CSI reference, the proportions with a significant negative trend decreased to 11.6% and 14.4%, respectively; and the proportions with a significant positive trend increased to 37.7% and 38.4%, respectively. The HRs of VF progression were 1.06 and 1.11 for each micrometer per year increase in the rates of change of ALCSD_BMO and ONHSD_BMO, respectively; and 1.07 and 1.09 for each micrometer per year increase in the rates of change of ALCSD_CSI and ONHSD_CSI, respectively.

Conclusions: Identifying fast progressors of posterior ALCS/ONHS displacement is relevant to the management of glaucoma patients as they have a higher risk of VF progression.

Optic nerve head (ONH) deformation is a defining feature of glaucoma. While confocal scanning laser ophthalmoscopy (CSLO) and optical coherence tomography (OCT) have been widely adopted for measurement of the neuroretinal rim,^1–6 parameters like anterior lamina cribrosa surface depth (ALCSD) and ONH surface depth (ONHSD) are rarely examined in clinical practice. Studies reporting long-term, progressive LC and ONH surface deformation in glaucoma patients are sparse. We previously investigated the longitudinal changes of ALCSD and ONHSD in glaucoma patients with reference to Bruch's membrane opening (BMO) using an event analysis.⁷ However, event analysis cannot reveal how fast an individual eye is progressing. The rates of change of ALCSD and ONHSD over the course of glaucoma progression have not been reported. The significance of measurement of LC and ONH deformation in the management of glaucoma patients remains unclear. 

Studies in nonhuman primates suggest deformation of the LC and the ONH surface to be detectable prior to axonal damage and functional loss of the optic nerve in experimental glaucoma.^8,9 We hypothesize that progressive posterior displacement of the LC and the ONH surface is indicative of an increased stress to the optic nerve and that individuals with a faster rate of posterior displacement of the anterior LC surface and the ONH surface entail a higher risk of development of visual field (VF) progression. In this prospective study, we imaged the ONH with OCT and measured the rates of change of ALCSD and ONHSD with reference to (1) the Bruch's membrane opening (BMO), and (2) the choroid-sclera interface (CSI) in 146 eyes of 95 glaucoma patients who were monitored at approximately 4-month intervals for greater than or equal to 5 years with an objective of investigating the impact of the rates of change of ALCSD and ONHSD on the risk of development of VF progression. 

Methods

Subjects

One hundred forty-six eyes of 95 glaucoma patients from the Optic Nerve Head Deformation in Glaucoma Study, a prospective study series investigating the longitudinal profiles of LC and ONH displacement in glaucoma, were analyzed. They were consecutively recruited and followed-up at approximately 4-month intervals from March 2008 to September 2015 at the University Eye Center, the Chinese University of Hong Kong. All patients had a complete ocular examination including measurement of visual acuity, IOP (dynamic contour tonometry), axial length (A-scan), and refraction. Color optic disc stereophotographs were taken with a fundus camera (TRC-50DX; Topcon, Tokyo, Japan). Eyes included in the study had visual acuity greater than or equal to 20/40 and had longitudinal follow-up greater than or equal to 5 years. Eyes were excluded if there was evidence of macular disease, neurologic disease, or refractive or retinal surgery. Glaucoma was diagnosed by characteristic optic disc/retinal nerve fiber layer (RNFL) changes (i.e., optic disc excavation/narrowed neuroretinal rim/RNFL defects) in color optic disc stereophotographs with corresponding VF abnormalities in at least one eye, independent of the level of IOP. At the 4-monthly follow-up visits, both eyes were examined under slit-lamp biomicroscopy with IOP measured with dynamic contour tonometry (DCT; the average of two diastolic IOP measurements with a quality score between 1 and 3 was recorded),^10–12 ONH imaged with OCT, and VF measured with standard automated white-on-white perimetry (described below). Patients were managed at the discretion of the attending ophthalmologists without taking consideration of OCT analysis of the anterior LC and the ONH surfaces. The study was conducted in accordance with the ethical standards stated in the 1964 Declaration of Helsinki and approved by the local research ethics committee with informed consent obtained. 

Optical Coherence Tomography Imaging

The ONH was imaged with the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). The scan protocol comprised 6 radial scan lines, each with 1024 A-scans equally spaced at 30° centered at the ONH with reference to the operator estimation of the clinically visible optic disc margin. Automatic image averaging was performed with 15 B-scans for each radial scan line. Follow-up scans were obtained at the same locations of the registered baseline visit using the eye-tracking function. All OCT volumes had a scan quality score of greater than or equal to 20. As the study was initiated before the availability of enhanced-depth imaging (EDI),¹³ only non-EDI baseline and follow-up scans with clear visibility of the anterior LC and the ONH surfaces were included in the analysis. After excluding 13 eyes of 13 glaucoma patients because of indiscernible anterior LC surface in at least one B-scan in the baseline and follow-up visits (i.e., all included eyes had 6 B-scans in each visit for analysis), 146 eyes of 95 glaucoma patients were analyzed. There were no significant differences in age, sex, IOP, axial length, central corneal thickness, average RNFL thickness, VF mean deviation (MD), and ONHSD measured at the baseline visit between the excluded and included eyes (Supplementary Table S1). 

Measurement of ALCSD and ONHSD

Optical coherence tomography B-scans (1:1 pixel in scale) were exported as image files after setting the baseline as reference. The ALCSD and ONHSD were measured with reference to the axial and transverse scaling factors (μm/pixel) of individual B-scans as reported in the Spectralis OCT operating system. Images were analyzed in a computer using a customized program developed in Matlab (R2010a; The MathWorks, Inc., Natick, MA, USA). Two anatomical landmarks (1) the Bruch's membrane opening (BMO) and (2) the choroid-sclera interface (CSI) were referenced for measurements of ALCSD and ONHSD (Fig. 1). ALCSD_BMO and ONHSD_BMO were measured after the BMO reference line (a line joining the BMO in a B-scan), the internal limiting membrane, and the anterior LC and the ONH surfaces were manually located (assisted by the piecewise cubic Hermite interpolation polynomial). ALCSD_CSI and ONHSD_CSI were measured with reference to the CSI reference line. The CSI reference line was a line joined by the CSI located between 1700 and 1800 μm from the BMO central axis point.¹⁴ The CSI positions were marked along the 1700- to 1800-μm distance from the BMO central axis point to minimize errors associated with focal fluctuations of the CSI.¹⁴ The ALCSD_BMO and ALCSD_CSI were defined as the perpendicular distances from the BMO and CSI reference lines, respectively, to the detectable anterior LC surface, which was visualized as the interface bordering the prelaminar tissue (moderate-intensity signal below the ONH surface) and the LC (high-intensity vertical striations; Fig. 1C).^5,7 ONHSD_BMO and ONHSD_CSI were defined as the perpendicular distances from the BMO and CSI reference lines, respectively, to the ONH surface (Fig. 1D). The ALCSD and ONHSD of an eye were measured as the averages of the six radial B-scans. To account for the fact that the radial scan pattern under-represented the peripheral compared with the central ONH in the measurement of ALCSD and ONHSD (the OCT sampling density was higher near the optic disc center than the peripheral), each of the delineated distances at the anterior LC and ONH surfaces was weighted in proportion to its distance from the center point between the pair of BMO points using the formulas:  

 

 

 

where ȳ represents the mean ALCSD/ONSHD of an eye, which is given by averaging the mean values of each of the 6 radial B-scans ȳ_i (for i = 1,2,…,6). ȳ_i represents the weighted average of the distance measure y_ij, which is the distance at point j in B-scan i with a weighting given by its distance from the BMO center point, d_ij.  

Figure 1

 

Measurements of ALCSD and ONHSD. An OCT B-scan image of a glaucomatous eye (A) before and (B) after manual detection of the BMO (pink dots), the CSI between 1700 and 1800 μm from the BMO central axis point (red dots), the internal limiting membrane/ONH surface (green), and the anterior LC surface (orange). (C) The ALCSD represents the perpendicular distances from the BMO reference line (red; left panel) or CSI reference line (blue; right panel) to the detectable anterior LC surface. The BMO reference line is a line joined by the BMO. The CSI reference line was a line joined by the CSI located between 1700 and 1800 μm from the BMO central axis point. (D) The ONHSD represents the perpendicular distances from the BMO reference line (left panel) or the CSI reference line (right panel) to the ONH surface (the black lines are for illustrative purpose and do not represent all the lines the software measured).

Visual Field Examination and Progression Analysis

Standard white-on-white automated perimetry (SITA standard 24-2) was performed using the Humphrey Field Analyzer II-i (Carl Zeiss Meditec, Dublin, CA, USA). Included VF tests had fixation losses and false positive less than or equal to 20%. A VF defect had greater than or equal to three nonedge contiguous locations with P < 0.05 (≥1 location with P < 0.01) on the same side of horizontal meridian in the pattern deviation plot confirmed with greater than or equal to two consecutive examinations. Visual field progression was analyzed using the Early Manifest Glaucoma Trial (EMGT) criteria.¹⁵ Visual field progression was defined when greater than or equal to three locations had a significant change compared with two baseline examinations (separated by ∼4 months) for at least three consecutive visits (i.e., “likely progression” reported in the Guided Progression Analysis). 

Statistical Analysis

Statistical analyses were performed with Stata (ver. 14.0; StataCorp., College Station, TX, USA). The rates of change of ALCSD and ONHSD of individual eyes were calculated with linear regression analysis between ALCSD/ONHSD and time. Multilevel linear mixed-modeling was applied to identify factors associated with the rates of change of ALCSD and ONHSD. Time (follow-up duration at each follow-up visit), baseline age, baseline axial length, baseline central corneal thickness, baseline ALCSD/ONHSD, IOP during follow-up, and the interactions between the baseline covariates and time were included in the linear mixed-models as fixed effects with random intercepts and random slopes allowing different baseline ALCSD/ONHSD values and different rates of change of ALCSD/ONHSD for individual eyes. Normality assumption of the random effects in the linear mixed-models (i.e., rates of change of ALCSD/ONHSD of individual eyes) was checked (Supplementary Fig. S1). The HRs of the rates of change of ALSCD/ONHSD and other biometric variables for prediction of VF progression were computed with joint longitudinal and survival modeling with ALCSD/ONHSD measurements nested in eyes. The application of joint longitudinal and survival modeling was initially described by Faucett and Thomas.¹⁶ It has two components: (1) a longitudinal submodel and (2) a survival submodel. In this study, the trajectory function of ALCSD/ONHSD and the covariates were fitted using linear mixed-modeling in the longitudinal submodel. The estimates of the coefficients in the linear mixed-model and the first derivative of the trajectory function of ALCSD/ONHSD (i.e., the rates of change of ALCSD/ONHSD) were then carried forward to a proportional hazards survival model specified by a Weibull distribution in the survival submodel. Anterior lamina cribrosa surface depth and ONHSD were separately analyzed in the joint models. Model fitting and conditional VF survival predictions for individual eyes were performed using the Stata codes developed by Crowther.¹⁷ P value less than 0.05 was considered statistically significant. 

Results

Table 1 shows the demographics of the 146 eyes of 95 glaucoma patients. The mean follow-up duration was 6.5 years (range, 5.0–7.4 years) and the mean number of follow-up visits for each patient was 17.3 (range, 8–21 visits). A total of 15,144 B-scans (6 B-scans for each eye × 146 eyes × 17.2877 visits) were segmented and analyzed. At the baseline, 49.3% of eyes had VF MD greater than or equal to −6 dB, 22.6% had VF MD between −6 and −12 dB, and 28.1% had VF MD less than or equal to −12 dB. Fourteen eyes (9.6%) showed “likely” VF progression during the study follow-up. 

Table 1

 

Demographics and Biometric Variable Measured at the Baseline Visit

 

Rates of Change of ALCSD and ONHSD – Linear Regression Analysis

Using linear regression analysis to measure the rates of change of ALCSD_BMO and ONHSD_BMO in individual eyes, 23.3% (34 eyes) and 28.1% (41 eyes) eyes showed a significant (P < 0.05) positive trend, respectively (i.e., posterior LC/ONH surface displacement); whereas 29.5% (43 eyes) and 24.0% (35 eyes) showed a significant negative trend, respectively (i.e., anterior LC/ONH surface displacement; Fig. 2). For eyes with a significant positive trend, the rates of change of ALCSD_BMO and ONHSD_BMO ranged from 1.3 to 23.3 μm/year, and from 1.4 to 29.0 μm/year, respectively (Fig. 3A). For eyes with a significant negative trend, they ranged from −1.7 to −21.0μm/year, and from −1.3 to −12.8μm/year, respectively. 

Figure 2

 

Agreement between anterior ALCSD and ONHSD. Venn diagrams showing the number of eyes (the number of patients in parentheses) with significant (A) posterior and (B) anterior displacement of the anterior lamina cribrosa surface and the ONH surface (determined by linear regression analysis between ALCSD/ONHSD and time) measured with reference to the BMO (left panel) and the CSI (right panel).

Figure 3

 

Rates of change of ALCSD and the ONHSD. Frequency distribution plots of the rates of change of ALCSD (left panel) and ONHSD (right panel) in eyes with significant (P < 0.05) positive or negative trends of ALCSD/ONHSD measured with reference to (A) the BMO and (B) the CSI.

Using the CSI reference, the proportions of eyes with a significant positive trend of ALCSD_CSI and ONHSD_CSI increased to 37.7% (55 eyes) and 38.4% (56 eyes), respectively (P ≤ 0.010); and the proportion of eyes with a significant negative trend decreased to 11.6% (17 eyes) and 14.4% (21 eyes), respectively (P ≤ 0.001; Fig. 2). The rates of change of ALCSD_CSI and ONHSD_CSI ranged from 1.2 to 26.3 μm/year, and from 1.2 to 34.8 μm/year, respectively, for eyes with a significant positive trend; and from −1.8 to −26.3 μm/year, and from −1.3 to −17.6 μm/year, respectively, for eyes with a significant negative trend (Fig. 3B). 

Determinants of the Rates of Change of ALCSD and ONHSD – Linear Mixed Modeling

Baseline ALCSD/ONHSD, baseline age, and IOP during follow-up (P ≤ 0.015), but not baseline axial length, baseline CCT, and baseline VF MD (P ≥ 0.122), were associated with the rates of change of ALCSD and ONHSD measured with reference to the BMO (Table 2) and the CSI (Table 3). Intraocular pressure during follow-up was positively, whereas age and ALCSD/ONHSD at the baseline were negatively associated with the rates of change of ALCSD and ONHSD. 

Table 2

 

Rates of Change of ALCSD_BMO and ONHSD_BMO and Their Determining Factors Analyzed With Multivariable Linear Mixed-Models

 

Table 3

 

Rates of Change of ALCSD_CSI and ONHSD_CSI and Their Determining Factors Analyzed With Multivariable Linear Mixed-Models

 

Impact of Rates of Change of ALCSD and ONHSD on VF Progression – Joint Longitudinal and Survival Modeling

The rates of change of ALCSD and ONHSD were significantly associated with the risk of VF progression after controlling for baseline age, baseline ALCSD/ONHSD, baseline VF MD, and IOP during follow-up, regardless of the reference (BMO/CSI) for measurements of ALCSD/ONHSD (Tables 4, 5). The HR of VF progression were 1.06 (95% confidence interval [CI]: 1.01–1.12) and 1.11 (95% CI: 1.02–1.21) for each micrometer per year increase in the rates of change of ALCSD_BMO and ONHSD_BMO, respectively (Table 4); and 1.07 (95% CI: 1.02–1.14) and 1.09 (95% CI: 1.01–1.18) for each micrometer per year increase in the rates of change of ALCSD_CSI and ONHSD_CSI, respectively (Table 5). Baseline age, baseline ALCSD/ONHSD, baseline VF MD, and IOP during follow-up were not associated with VF progression (P ≥ 0.414). The rates of change of maximum ALCSD/maximum ONHSD (i.e., the mean of the maximum ALCSD/ONHSD measured from each of the 6 radial scans of each eye) measured with reference to the BMO/CSI were also predictive of VF progression (Supplementary Tables S2, S3). 

Table 4

 

Hazard Ratios of the Rates of Change of ALCSD_BMO and ONHSD_BMO and the Covariates for Development of VF Progression Analyzed With Joint Longitudinal and Survival Models

 

Table 5

 

Hazard Ratios of the Rates of Change of ALCSD_CSI and ONHSD_CSI and the Covariates for Development of VF Progression Analyzed With Joint Longitudinal and Survival Models

 

Case examples illustrating the application of joint longitudinal and survival modeling to estimate VF survival probability from longitudinal measurements of ALCSD/ONHSD for an individual eye are shown in Figure 4. 

Figure 4

 

Case examples. Conditional VF survival prediction derived from the joint longitudinal and survival modeling in two glaucomatous eyes from two glaucoma patients; one with (A) and one without (B) significant positive trends of the ALCSD (left panel) and ONHSD (right panel) measured with reference to the BMO (upper panel) and the CSI (lower panel). Significantly lower VF survival probabilities were observed for the eye with progressive posterior displacement of the anterior LC and the ONH surfaces (A). “Likely” VF progression (EMGT criteria) was detected in Patient A during the study follow-up. The VF of Patient B showed no progression throughout the study period.

Discussion

In this 5-year prospective study, we reported the rates of change of ALCSD and ONHSD and investigated the determinants and consequences of progressive displacement of the anterior LC surface and the ONH surface in glaucoma. With an average of more than 17 longitudinal ALCSD/ONHSD measurements collected from 146 eyes of 95 glaucoma patients followed at regular intervals over 5 years, we showed that the rates of change of ALCSD and ONHSD varied widely in glaucoma patients (Fig. 3) and they were influenced by age, baseline ALCSD/ONHSD, and IOP during follow-up (Tables 2, 3). Further, we examined the impact of the rates of change of ALCSD and ONHSD on prediction of VF progression using joint longitudinal and survival models. The risk of development of VF progression increased by 6.4% (BMO-based measurements) to 7.4% (CSI-based measurements) for each micrometer per year increase in the rate of change of ALCSD, and by 10.9% (BMO-based measurements) and 9.3% (CSI-based measurements) for each micrometer per year increase in the rate of change of ONHSD after controlling for the covariates (Tables 4, 5). Our finding underscores the importance of monitoring the ALCSD and ONHSD for risk assessment of VF progression in glaucoma patients. 

Short-term changes of ALCSD and ONHSD have been investigated after transient IOP elevation with an ophthalmodynamometer¹⁸ or IOP reduction,^19–21 but studies investigating long-term, progressive deformation of the LC and the ONH surface in glaucoma are sparse. To our knowledge, our data are the first to measure the rates of change of ALCSD and ONHSD and examine their impact on VF progression in glaucoma patients. While 29.5% and 24.0% of eyes exhibited anterior displacement of the anterior LC and the ONH surfaces over 5 years, respectively, when the ALCSD and ONHSD were measured with reference to the BMO, the respective proportions significantly decreased to 11.6% and 14.4% when the ALCSD and ONHSD were measured with reference to the CSI. The differences in the proportions of eyes with significant anterior LC and ONH surface displacement between BMO- and CSI-based measurements could be attributed to age-related²² and/or glaucoma-related reduction of choroidal thickness.²³ Progressive choroidal thinning can engender an appearance of anterior displacement of the LC and the ONH surface with reference to the BMO, which may not be evident when the ALCSD and ONHSD were measured with reference to the CSI. Prospective studies are ongoing to measure the longitudinal changes of choroidal thickness in normal individuals and glaucoma patients and delineate the relative contribution of age- versus disease-related choroidal thinning to ALCSD and ONHSD measurements. 

Notably, anterior displacement of the anterior LC and the ONH surfaces remained evident in 11.6% and 14.4% of eyes, respectively, even when the CSI reference was adopted. Because glaucoma patients followed-up in this study received IOP-lowering treatment at the discretion of the attending ophthalmologists (the attending ophthalmologists were blind to ALCSD/ONHSD measurements during the study follow-up), it is plausible to observe anterior LC and ONH surface displacement as a consequence of IOP reduction.^19–21 Alternatively, anterior LC and ONH surface displacement can be age-related independent of the levels of IOP. This is supported by the finding that baseline age was negatively related to the rates of change of ALCSD/ONHSD after controlling for IOP and other covariates in the linear mixed-models (Tables 2, 3). This is also in agreement with a previous study, which showed that for the same degree of glaucomatous damage (measured by VF and OCT RNFL thickness), the ALCSD was shallower in older than younger patients.²⁴ Using finite element modeling, it has been reported that IOP elevation induces circumferential stress on the sclera, resulting in scleral canal expansion, which may pull the LC taut and lead to anterior LC displacement.²⁵ Whether IOP elevation is connected to scleral canal expansion and anterior displacement of the LC and ONH surfaces in glaucoma patients would be an important question for further studies. 

The rates of change of ALCSD/ONHSD were highly variable (Fig. 3). We reasoned that eyes with a more rapid posterior deformation of the anterior LC and the ONH surfaces are more likely to develop VF progression. Using joint longitudinal and survival modeling, which computed the HRs of the rates of change of ALCSD/ONHSD for development of VF progression by incorporating subject-specific predictions of the unobserved values derived from the linear mixed-submodel into the survival submodel, we showed that the rates of change of ALCSD and ONHSD were predictive of VF progression independent of the reference (BMO/CSI) used for measurements of ALCSD/ONHSD (Tables 4, 5). Posterior displacement of the anterior LC and ONH surfaces may represent an early biomarker of axonal damage and VF loss in glaucoma patients, signifying the need of a more intensive treatment regimen and follow-up schedule. Whereas progressive reduction of RNFL thickness and neuroretinal rim area has also been shown to be predictive of VF progression,^26,27 RNFL and neuroretinal rim loss is irreversible. Progressive posterior displacement of the LC and the ONH surface can be observed prior to detectable RNFL thinning and neuroretinal rim loss.^8,9,28 Lowering the IOP would be relevant to lessen posterior deformation of the anterior LC and ONH surfaces (as supported by the longitudinal submodel), which would in turn lower the risk of VF progression (as supported by the survival submodel). 

Our study is limited in obtaining only six radial B-scans for each eye in each visit. Localized displacement of the anterior LC and the ONH surfaces might have been missed. Nevertheless, we believe data collected from an average of 17.3 visits for each eye over an average of 6.5 years with a total of 15144 B-scans would be sufficient to determine the risk of global rates of change of ALCSD/ONHSD on VF progression. Another limitation is that depth measurements of anterior LC and ONH surfaces are dictated by the location of reference landmark (i.e., BMO/CSI). Change in ALCSD/ONHSD may not necessarily signify deformation of the LC and ONH surfaces. The visibility of the anterior LC surface can be obscured by retinal vasculature, which would hamper the measurement of ALCSD. Yet, there was good to excellent agreement for progressive deformation of the anterior LC and the ONH surfaces (Fig. 2) and the rates of change of ALCSD and ONHSD were both predictive of VF progression. Interpretation of the study results is constrained by the methodology of change analysis and the test-retest variability inherent to SD-OCT and perimetry. The relatively high specificity of the EMGT criteria to define VF progression²⁹ may in part contribute to the temporal disparity in detecting ALCSD/ONHSD changes and VF progression. Finally, the implementation of the joint longitudinal and survival models is currently limited to two-level modeling. Multilevel modeling structures are still under development. 

To summarize, longitudinal ALCSD and ONHSD measurements are informative to predict VF progression in glaucoma patients. Given that the rates of change of ALCSD and ONHSD were subject to the influence of visit-to-visit variation of IOP and that faster rates of posterior displacement of the anterior LC and ONH surfaces were associated with a higher risk of VF progression, regular follow-up for measurement of ALCSD and ONHSD is clinically relevant to help determine target IOP and for risk assessment for glaucoma patients. 

Acknowledgments

Supported by grants from the Hong Kong Research Grant Council General Research Fund 14101215. 

Disclosure: Z. Wu, None; C. Lin, Carl Zeiss Meditec (F, R), Heidelberg Engineering (R); M. Crowther, None; H. Mak, None; M. Yu, None; C.K.-S. Leung, None