Thickness of the Macula, Retinal Nerve Fiber Layer, and Ganglion Cell Layer in the Epiretinal Membrane: The Repeatability Study of Optical Coherence Tomography

Abstract

Purpose: To analyze the repeatability of measurements of the thicknesses of the macula, retinal nerve fiber layer (RNFL), and ganglion cell inner plexiform layer (GCIPL) using spectral-domain optical coherence tomography (SD-OCT) in the epiretinal membrane (ERM).

Methods: The prospective study analyzed patients who visited our retinal clinic from June 2013 to January 2014. An experienced examiner measured the thicknesses twice using macular cube 512 × 128 and optic disc cube 200 × 200 scans. The repeatability of the thicknesses of the macula, RNFL, and GCIPL were compared using the intraclass correlation coefficient (ICC) of two groups based on the central macular thickness (group A, ≤450 μm; group B, >450 μm).

Results: A total of 88 patients were analyzed. The average thicknesses of the central macula, RNFL, and GCIPL were 256.5, 96.6, and 84.4 μm, respectively, in the normal fellow eye and 412.3, 94.6, and 56.7 μm in the affected eye. The ICCs of the central macula, RNFL, and GCIPL were 0.995, 0.994, and 0.996, respectively, for the normal fellow eye and 0.991, 0.973, and 0.881 for the affected eye. The average thicknesses of the central macula, RNFL, and GCIPL in group A were 360.9, 93.5, and 63.4 μm, respectively, and the ICCs were 0.997, 0.987, and 0.995. The thicknesses in group B were 489.5, 96.2, and 46.6 μm, respectively, and the ICCs were 0.910, 0.942, and 0.603, significantly lower repeatability compared with group A (P < 0.05).

Conclusions: The macular contour change with the ERM results in low repeatability and tendency to be measured thinner in GCIPL thickness using SD-OCT. This can be explained by the unstable gaze of the patient due to decreased visual acuity and autosegmentation error following changes in the macula.

An idiopathic epiretinal membrane (ERM) is a disease that induces retinal tangential traction via the proliferation of retinal glia, pigment epithelial cells, and myofibroblasts between the internal limiting membrane and hyaloid membrane and resulting in decreased visual acuity or metamorphopsia. It is a relatively common disease, found in 20% of patients older than 75 years.^1–3 

The development of spectral-domain optical coherence tomography (SD-OCT) and management of retinal diseases based on the high-resolution images obtained with the faster scanning has made it possible to examine each retinal layer in detail, with increased scanning density and decreased noise with averaging and autosegmentation.⁴ Many SD-OCT studies of the ERM have reported that visual prognosis is influenced by the photoreceptor inner/outer segment junction and cone outer segment tip.^5–9 Moreover, it was reported recently that postoperative vision is related to the thickness of the inner retinal layer and changes in the ganglion cell layer.^10,11 

Spectral-domain OCT enables repeatable, reproducible measurement of the thickness of the macula and retinal nerve fiber layer (RNFL) around the optic disc.^12,13 However, the measurement of the ganglion cell inner plexiform layer (GCIPL) thickness by autosegmentation can be affected by macular contour changes, such as macular edema or atrophy.^14,15 

Therefore, this study analyzed the repeatability of measurements of the thicknesses of the central macula, RNFL, and GCIPL in ERM under the assumption that morphology could influence the repeatability of the measurement. 

Methods

This prospective cohort study was approved by the institutional review board of Chungnam National University Hospital. All participants gave informed consent and the study adhered to the tenets of the Declaration of Helsinki. 

Subjects

The study prospectively targeted patients diagnosed with an idiopathic ERM from June 2013 to January 2014. A detailed history, uncorrected visual acuity, autorefraction, best-corrected visual acuity (BCVA), IOP measurement using noncontact tonometry, slit lamp microscopy, fundoscopy, and OCT were obtained in all participants. 

Based on the central macular thickness (CMT) of the affected eyes, the patients were divided into two groups: group A had a CMT less than or equal to 450 μm and group B had a CMT greater than 450 μm.^5,7 

Patients with previously diagnosed ophthalmic diseases, such as optic nerve diseases, glaucoma, and other retinal diseases except for the idiopathic epiretinal membrane, with a history of ophthalmic surgery or with an OCT signal intensity less than five were excluded. Also, patients with myopia of greater than 6.0 diopters or ocular axial length greater than 25.0 mm were excluded from this study. 

Optical Coherence Tomography Measurement

A expert performed two macular cube 512 × 218 and optic disc cube 200 × 200 scans twice consecutively at 5-minute intervals using a Cirrus HD OCT (Carl Zeiss Meditec, Dublin, CA, USA). 

The macular cube 512 × 218 scans were divided into the central circle, inner ring, and outer ring using 1-, 3-, and 6-mm diameters, respectively, based on the CMT using a retinal map analysis system. The scan was divided into the nine Early Treatment Diabetic Retinopathy Study subfields: the central macular subfield and the inner and outer superior, temporal, inferior, and nasal subfields. 

The GCIPL thickness was calculated using a ganglion cell analysis (GCA) algorithm in Cirrus HD-OCT. The GCA algorithm measures the GCIPL thickness by detecting the outer boundary of the RNFL in the macula and the outer membrane of the inner plexiform layer using three-dimensional data from the macular cube. We determined the average and minimum GCIPL thicknesses, and the values for six sectors (the superior, superotemporal, superonasal, inferior, inferotemporal, and inferonasal areas). 

The RNFL thickness was analyzed for four sectors (superior, temporal, inferior, and nasal) and as the average thickness from an optic disc cube 200 × 200 scan (Fig. 1). 

Figure 1

 

An example of measurements of the right eye. (A) Macular cube 512 × 128 scan, (B) optic disc cube 200 × 200 scan, and (C) GCIPL analysis. OS, outer superior; OT, outer temporal; OI, outer inferior; ON, outer nasal; IS, inner superior; IT, inner temporal; II, inner inferior; IN, inner nasal; S, superior; ST, superotemporal; SN, superonasal; I, inferior; IN, inferonasal; and IT, inferotemporal.

Statistical Analysis

The intraclass correlation coefficient (ICC), coefficient of variation (COV), and test–retest variability (TRV) were calculated to determine the repeatability of consecutively measured thicknesses of the central macula, RNFL, and GCIPL. Paired t-test and ANOVA were conducted to examine differences between the two groups and level of statistical significance was set at P < 0.05. 

Results

Demographics

The study enrolled 102 patients: 54 in group A and 48 in group B. After excluding 14 patients in group B due to an error in measuring GCIPL thickness, 88 patients (42 males, 46 females) were analyzed. There were no differences in age, sex, laterality, or refractions; however, the BCVA converted to logMAR was significantly lower in group B (0.2 ± 0.2 vs. 0.07 ± 0.1, P = 0.001; Table 1). 

Table 1

 

Demographics of Patients

 

Optical Coherence Tomography Measurement

We defined measurement error as the case in which the number of the GCIPL thickness was not indicated in the map due to severe macular elevation. We excluded measurement error from the analysis. There were 14 patients (29.2%) with measurement errors, all were in group B (Fig. 2). 

Figure 2

 

Autosegmented GCIPL. (A) Normal autosegmentation in the normal contralateral eye. (B) Segmentation errors in the affected eye. Measured GCIPL thickness is thinner than the real GCIPL thickness in epiretinal membrane patients.

The average thicknesses of the central macula, RNFL, and GCIPL were 256.5, 96.6, and 84.4 μm, respectively, in the normal fellow eye and 412.3, 94.6, and 56.7 μm in the affected eye. Analyzing the two groups separately, the average thicknesses of the central macula, RNFL, and GCIPL were 360.9, 93.5, and 63.4 μm, respectively, in group A and 489.5, 96.2, and 46.6 μm in group B. The average GCIPL was statistically significantly thinner in the affected eye compared with the normal fellow eye and the GCIPL was thinner in group B compared with group A (P = 0.001) (Fig. 3). 

Figure 3

 

Average GCIPL thickness for the normal group and groups A and B. The average GCIPL thickness of the affected side was thinner than that of the normal side and group B had a thinner average GCIPL thickness compared with group A.

Repeatability of OCT Measurement

For all patients, the ICCs of the central macula, RNFL, and GCIPL were 0.995, 0.994, and 0.996, respectively, for the normal side and 0.991, 0.973, and 0.881 for the affected eye, indicating high repeatability. However, analyzing the affected eye by group, although the ICCs of the central macula, RNFL, and GCIPL for group A were 0.997, 0.987, and 0.995, respectively, indicating high repeatability, the values for group B were 0.910, 0.942, and 0.603, indicating high repeatability for the central macula and RNFL, but low repeatability for the GCIPL. The other markers of repeatability (i.e., COV and TRV) gave similar results (Tables 2–4). 

Table 2

 

Comparison of the OCT Measurements in Group A

 

Table 3

 

Comparison of OCT Measurements in Group B

 

Table 4

 

Comparison of the OCT Measurements in the Normal Contralateral Eye

 

When the CMT is greater than 450 μm, the difference in the GCIPL measurements increases significantly in a scatterplot of the differences between two values measured consecutively compared with the differences in the central macula and RNFL (Fig. 4). 

Figure 4

 

Scatterplot of the differences in two repeated measured values in epiretinal membrane patients.

Discussion

Idiopathic ERM is a proliferative disease of the retinal glia, pigment epithelial cells, and myofibroblasts on the internal limiting membrane that causes tangential traction, structural changes such as retinal wrinkles and distortion, and decreased visual acuity.^1–3 

Spectral-domain OCT has enabled detailed analysis of each retinal layer and contributed to studies of the influence of the retinal layers on visual function.⁴ Many SD-OCT studies of ERM have reported that the visual prognosis is influenced by the photoreceptor inner/outer segment junction and disruption of the cone outer segment tip.^5–9 

A recent study reported a relationship between changes in the inner retinal layer thickness and the visual prognosis with intact photoreceptors in the ERM. The retinal damage is caused by traction as the disease progresses and damage is related to an unfavorable prognosis.¹⁰ 

In addition, there is a significant decrease in GCIPL thickness in the affected eye after surgical removal of the ERM. This decreased thickness was significantly correlated with the poor postoperative visual outcomes in patients with intact photoreceptor layers.¹¹ 

Thickness measurements of the macular area and RNFL using SD-OCT are highly repeatable, easy, and precise.^16,17 Pinilla et al.¹⁸ reported that both Cirrus and Spectralis Fourier-domain OCT enabled highly repeatable thickness measurements of the macula. 

Recent studies have examined the GCIPL, and reported high repeatability of GCIPL thickness measurement using SD-OCT in glaucoma patients.^14,19,20 However, few studies have examined the repeatability of GCIPL thickness measurements depending on the change in the macula thickness in retinal disease. 

The GCIPL thickness measurement can be affected by retinal elevation or atrophy.¹⁴ Especially, an algorithmic error in retinal thickness measurement might be more frequent in diseases causing foveal contraction, such as an ERM. 

This study found high repeatability of the three measured values in the overall ERM patient group. However, the repeatability of the GCIPL thickness measurement decreased in the group with a central macular thickness exceeding 450 μm. It is thought that the severe deformation of the retinal structure by ERM causes autosegmentation error of the boundary of the GCIPL. In addition, 14 patients (29.2%) in the group with a CMT greater than 450 μm could not be analyzed due to the measurement errors. It is not certain that every excluded patient in group B could have low repeatability if their GCIPL thicknesses were measured correctly. However, the exclusion of these patients may have influenced the sample size of group B and results of the repeatability. 

Comparing the GCIPL thickness of the affected and normal fellow eyes, the GCIPL thickness on the affected side decreased to 49.8 μm versus 84.4 μm in the normal eye. In addition, comparing GCIPL thicknesses of the two groups based on a 450-μm cutoff for the CMT, the GCIPL was significantly thinner in the group with the thicker central macula. Koo et al.¹⁰ reported that as an ERM progresses, the retina becomes thicker, especially the retinal inner layer, including the nerve fiber, ganglion cell, and inner plexiform layers. Therefore, it is thought that changes in the macula contour result in erroneous thin segmentation of the GCIPL as the CMT increases, rather than an actual decrease in GCIPL thickness, as measured in this study. 

In the group with a CMT greater than 450 μm, we found that the repeatability of the GCIPL measurement was low, although the repeatabilities of the central macula and RNFLs were high. This can be explained by the possibility of frequent autosegmentation error following the more distorted configuration of the macula than the group with a CMT less than or equal to 450 μm. In addition, the visual acuity in the group with the thicker central macula was significantly lower than the group with a CMT less than or equal to 450 μm, and it is thought that unstable gaze contributed to the lower repeatability because of difficulty measuring the same macular area.^21,22 Also, we measured the thickness by SD-OCT and the reason why unstable gaze and autosegmentation error influenced only the repeatability of GCIPL retinal layer would be SD-OCT's resolution and its capability of autosegmentation; whenever a layer's thickness falls under a certain level (probably at approximately 40 μm), the measurements become unreliable. 

The limitation of our study is that the measurements were acquired with only Cirrus OCT and we did not use any follow-up tool for obtaining the two measurements. 

In conclusion, the macular contour change caused by an ERM results in low repeatability and tendency to be measured thinner in GCIPL thickness using SD-OCT. This can be explained by the unstable gaze of the patient due to decreased visual acuity and autosegmentation error following changes in the macula. Morphological changes of the macula should be considered when measuring GCIPL thickness in various ophthalmic diseases, such as glaucoma and neuro-ophthalmology. 

Acknowledgments

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sector. 

Disclosure: H.-J. Lee, None; M.-S. Kim, None; Y.-J. Jo, None; J.-Y. Kim, None