Purpose To establish the normative ranges of macular ganglion cell layer (mGCL) and macular inner plexiform layer (mIPL) thickness using Spectralis spectral domain optical coherence tomography (SD-OCT) (Heidelberg Engineering, Inc., Heidelberg, Germany) in both Korean children and adults, and to determine factors associated with mGCL and mIPL thickness.
Methods We conducted a retrospective, observational study of 573 healthy subjects (5–70 years old) who underwent comprehensive ophthalmic examinations in a single institution. Each inner retinal layer thickness was measured using SD-OCT and automatic segmentation software. Cross-sectional analysis was used to evaluate the effect of gender, age and ocular parameters on mGCL and mIPL thickness. Normative ranges of mGCL and mIPL thickness according to age, gender and factors associated with mGCL and mIPL thickness were measured.
Results The mean mGCL and mIPL thickness were 40.6±2.8 and 33.8±2.0 µm, respectively. Determinants of inner sector mGCL thickness were circumpapillary retinal nerve fibre layer (cpRNFL) thickness (β=1.172, p<0.001), age (β=−0.019, p=0.021) and male gender (β=1.452, p<0.001). Determinants of inner sector mIPL thickness were cpRNFL (β=0.952, p<0.001) and male gender (β=1.163, p<0.001). The inner sector mGCL and mIPL thickness increased significantly with age in children (β=0.174, p=0.009 and β=0.115, p=0.013), and then decreased in adults (β=−0.070, p<0.001 and β=−0.024, p=0.032). In the case of outer sectors, mGCL and mIPL thickness were not significantly related to age and gender.
Conclusions This study ensured a normative range of the mGCL and mIPL thickness using Spectralis OCT. Gender, age and cpRNFL thickness significantly correlated with mGCL and mIPL thickness. This information should be considered in the interpretation of SD-OCT data.
- optic nerve
- diagnostic tests/investigation
Statistics from Altmetric.com
Optical coherence tomography (OCT) is a non-invasive cross-sectional imaging technique widely used to assess retinal and optic nerve diseases in the ophthalmology clinic.1 Since 2004, spectral domain OCT (SD-OCT) has entered clinical practice with increased axial resolutions of up to 1–5 µm and faster scanning speed compared with the earlier generation time domain OCT.2 By the use of eye-tracking systems, SD-OCT can provide stable image quality in more than 95% of children over 5 years of age with lower susceptibility to eye movement artefacts.3 4
More recent automatic segmentation software provided by Spectralis OCT (Heidelberg Engineering, Inc., Heidelberg, Germany) enables individual measurements of each layer of the retina.5 Reference values of individual retinal layer thickness measured by Spectralis OCT can be effectively used in the diagnosis and monitoring of retinal and optic nerve diseases.6 In addition, because each segmentation software identifies the outer retinal boundary at slightly different locations, reference values obtained from one system cannot directly be applied to data acquired from other devices.7 8 Although numerous literature have provided the normal reference data of ganglion cell-inner plexiform layer (GCIPL) measurements using Cirrus OCT (Carl Zeiss Meditec, Oberkochen, Germany),9–12 separate measurements of the macular ganglion cell layer (mGCL) and macular inner plexiform layer (mIPL) measured with Spectralis OCT are still lacking.13 14 To the best of our knowledge, the normal mGCL and mIPL thickness in both children and adults have not been reported in detail.6 In the present study, we measured the individual thickness of mGCL and mIPL using automatic segmentation of retinal layer boundaries in Spectralis OCT images of Korean subjects. We also identified the effects of age, gender and ocular parameters on mGCL and mIPL thickness measurements.
We performed a retrospective review of patients who visited the ophthalmology clinic of Seoul National University Bundang Hospital between January 2013 and November 2016. Patients between 5 and 70 years of age were included. Subjects were divided into two age groups: children (5–17 years) and adults (18–70 years).
Patients included in the present study were born at term (≥37 weeks’ gestational age) with normal birth weight (≥2500 g). They also had best corrected visual acuities (BCVAs) of at least 20/30 in children under 7 years of age,3 15–17 and at least 20/20 in adults and children of 7 years old and above. Only patients with no other ocular abnormalities except for low refractive error and dry eye syndrome were included. All subjects had an intraocular pressure (IOP) of 21 mm Hg or less and a normal appearing optic disc, which was defined as the absence of glaucomatous optic neuropathy, pallor or swelling of the optic disc. When glaucomatous optic neuropathy was suspected, the diagnosis was confirmed by visual field tests using standard automated perimetry (24-2 Swedish interactive thresholding algorithm) and Humphrey Field Analyzer II 750 (Carl Zeiss Meditec, Inc., Dublin, CA).
Exclusion criteria were high refractive error, defined as spherical equivalent (SE) refractive errors exceeding ±6.00 dioptres (D) or astigmatism exceeding 3.00 D, and ocular conditions, such as strabismus, amblyopia or any retinal or optic disc anomaly as determined by mydriatic fundus examination. Subjects with a history of ocular surgery (eg, refractive surgery, cataract extraction, glaucoma surgery), intraocular diseases (eg, optic disc abnormalities such as optic disc drusen, optic disc oedema, or optic atrophy and retinal diseases such as retinal vessel occlusion, age-related macular degeneration or diabetic retinopathy), or neurologic diseases that involve the anterior or posterior visual pathway were also excluded.
Patients underwent a complete ophthalmic examination including visual acuity, refraction, slit-lamp biomicroscopy, non-contact tonometry (ICT-800; KOWA, Tokyo, Japan) or rebound tonometry (ICare PRO; ICare, Helsinki, Finland) and dilated stereoscopic examination of the optic disc. If the IOP measured by the non-contact tonometry or rebound tonometry was above 21 mm Hg, a conﬁrmation test was performed with Goldmann applanation tonometry (Haag-Streit, Bern, Switzerland). Fundus photography (EOS D60 digital camera; Canon, Utsunomiya-shi, Japan) and OCT scanning of the macular and peripapillary areas were also performed using Spectralis OCT.
All images were obtained by a single, well-trained technician using Spectralis OCT (HEYEX software V.6.0), as described previously.1 Two sets of SD-OCT images were obtained on the same day: (1) macular scan to measure mGCL and mIPL thickness and (2) optic disc scan to measure circumpapillary RNFL (cpRNFL). For cpRNFL thickness measurements, a scan circle of approximately 12° in diameter was positioned at the centre of the optic disc while the eye-tracking system was activated. The diameter of scan circle in millimetres depended on the axial length. The values of corneal curvature and axial length were entered into the Spectralis OCT system before the scan so as to adjust for magnification error. The Spectralis OCT software calculates the average cpRNFL thickness and mean RNFL thickness in each of the following six sectors relative to the fovea–disc axis: nasal superior (90°–135°), nasal (135°–225°), nasal inferior (225°–270°), temporal inferior (270°–315°), temporal (315°–45°) and temporal superior (45°–90°) (online supplementary figure 1).
Spectralis OCT provided optic nerve head images using the enhanced depth imaging technique. The imaging was performed using a 10°×15° rectangle covering the optic disc. This rectangle was scanned for approximately 70 sections, which were 30–34 µm apart (the slicing distance is determined automatically by the machine). Each section had an average of 42 frames, which provided the best trade-off between the image quality and patient cooperation.18
The macular scans were performed in the 30° perifoveal area using a 30°×25° OCT volume scan. Sixty-one B-scan sections parallel to the fovea–disc axis were obtained, and frame averaging algorithms of nine OCT frames were used in each section. Images were obtained only when the quality score was >15. The new software for the Spectralis OCT device automatically segments boundaries of the 10 retinal layers and provides measurements of individual retinal layer thickness. In this study, the macular total retinal thickness, mGCL, mIPL and cpRNFL thickness were investigated for analysis. The accuracy of segmentation of each retinal layer (cpRNFL, mGCL and mIPL) and adequate centration on the fovea were reviewed independently by masked observers (YJY and HKY). Only images that were considered adequate by both observers were included in the analysis. For retinal thickness maps, three circular lines representing 1, 3 and 6 mm scan diameters (Early Treatment Diabetic Retinopathy Study; ETDRS macula) were obtained. The average of all points within the inner 1 mm diameter circle was defined as the central subfield thickness. As the mGCL and mIPL in the central subfield are very thin and difficult to delineate, we did not include values obtained from the 1 mm diameter central subfield area for analysis. The intermediate ring was divided into the inner superior, inner nasal, inner inferior and inner temporal subfields, while the outer ring was divided into the outer superior, outer nasal, outer inferior and outer temporal subfields (online supplementary figure 1).
Statistical analysis was performed using SPSS V.21.0 software (IBM). Although data were acquired from both eyes, only data from one randomly selected eye per participant were entered in the analysis. Comparisons of macular thickness between age groups were performed using the unpaired t-test with Bonferroni adjustment, and results were considered statistically significant when two-sided p values were less than 0.00625 (0.05/8 Bonferroni adjustment). Associations between mGCL and mIPL thickness and the following seven factors were assessed: gender, age, laterality, IOP, astigmatism, SE refractive errors and average cpRNFL thickness.9 19–22 Univariate linear regression analyses were performed, followed by multivariate linear regression models for factors with an association at a p value of 0.1 or less in the univariate analysis. To determine how the factors accounted for variability, we converted the coefficient of determination into percentage. The cut-off for statistical significance was set at p<0.05. Except where stated otherwise, the data are presented as mean±standard deviation (SD) values.
Demographics and ocular characteristics
Five hundred and seventy-three subjects were included in the present study. The mean age of the study population was 26.1±19.5 years, and 323 (56.2%) were female. After randomisation, 282 right eyes and 293 left eyes were included. The mean SE refractive error was −0.73±2.12 D (range: −6.00 to +5.00 D) and the mean axial length was 23.7±1.3 mm (range: 20.2–27.3 mm). The ages ranged from 5 to 70 years and were divided into two age groups: the younger group consisted of subjects ranging from 5 to 17 years of age (n=276; 115 male (42%) and 161 female (58%)), the older group consisted of subjects ranging from 18 to 70 years of age (n=297; 135 male (45%) and 162 female (55%)). Within each age group, the sample size was equally distributed among all ages by decade. In subgroup analysis based on age, BCVAs, IOP, SE refractive errors and axial length were not significantly different between children over 10 years of age (aged 10–17 years) and adults (aged 18–70 years) (all p>0.05).
cpRNFL and macular thickness measurements
Mean cpRNFL and total macular thickness followed a normal distribution (online supplementary figure 2). There was no difference in scan quality between age groups. The cpRNFL thickness was greatest at the temporal-superior and temporal-inner sectors, and thinnest at the nasal and temporal sectors. The average macular thickness of the eight ETDRS subfields is shown in online supplementary figure 3. Of the total subjects, the average macular thickness was 313.0±12.3 µm (range: 277.6–371.1 µm). The inner superior sector showed significant difference between children and adults (p=0.009 after Bonferroni correction), which was thinner in children. Table 1 shows the average and 95% confidence intervals (CIs) of the mGCL and mIPL thickness in each of the eight ETDRS sectors of total subjects and subgroups divided by age and gender. The average mGCL and mIPL thickness of total subjects were 40.6±2.8 µm (range: 31.4–54.0 µm) and 33.8±2.0 µm (range: 28.4–42.0 µm), respectively. Among the four ETDRS inner sectors of the mGCL, the inner temporal sector was the thinnest (all p<0.05, one-way analysis of variance (ANOVA) and all p<0.01, post hoc analysis by Tukey). Regarding the four ETDRS outer sectors of the mGCL, the outer nasal sector was the thickest among all subgroups (all p<0.05, one-way ANOVA and all p<0.01, post hoc analysis by Tukey).
Correlations and determinants of mGCL thickness
Table 2 shows the univariate and multivariate linear regression analyses between associated factors and inner and outer sector mGCL thickness. Multivariate analysis revealed that male gender, age and average cpRNFL thickness were significantly associated with inner sector mGCL thickness (β=1.452, p<0.001, β=−0.019, p=0.021, and β=1.172, p<0.001, respectively) (table 2, figure 1A,B). These factors accounted for 11.0% of inner sector mGCL thickness variability. Regarding the average outer sector mGCL thickness, SE refractive errors and average cpRNFL thickness were significantly associated with outer sector mGCL thickness (β=−0.167, p=0.007 and β=2.269, p<0.001, respectively) (table 2, figure 1C). These two factors explained approximately 31.2% of the variability in outer sector mGCL thickness.
By subgroup analysis, a positive correlation was found between inner sector mGCL thickness and age in children, while a negative correlation was found in adults. The inner sector mGCL thickness increased by 0.174 µm/year in children (table 3, figure 1A). However, this correlation was reversed in adults as the inner sector mGCL thickness tended to decrease by 0.070 µm/year (table 3, figure 1A).
Correlations and determinants of mIPL thickness
Table 4 shows the univariate and multivariate analyses between associated factors and inner and outer sector mIPL thickness using linear regression analyses. In multivariate analyses, significant determinants of average inner sector mIPL thickness were male gender (β=1.163, p<0.001) and cpRNFL thickness (β=0.952, p<0.001) (table 4, figure 1E). These two factors accounted for 15.0% of inner sector mIPL thickness variability. Regarding the outer sector mIPL thickness, cpRNFL thickness was the only significant predictor which explains approximately 38.5% of the variability in outer sector mIPL thickness (β=1.781, p<0.001) (table 4, figure 1F).
After dividing subjects by age, the inner sector mIPL thickness increased significantly at 0.115 µm/year in children (table 5, figure 1D). However, the correlation was reversed in adults as the inner sector mGCL tended to decrease by 0.024 µm/year (table 5, figure 1D).
Recently, SD-OCT has played a critical role in early identification of diseases of the retina and optic nerve, providing new structural parameters including cpRNFL thickness and individual retinal layer thickness in the macula. Currently, Spectralis OCT only shows a thickness distribution map of retinal layers and does not provide any data about abnormality of individual layer thickness. The normative database for GCIPL thickness in Cirrus OCT consists of 284 healthy subjects of different ethnicities with a mean age of 46.5 years (range: 18–84 years).23 However, the normative database of mGCL/mIPL thickness measured by Spectralis OCT has not been available yet, and only few studies have reported normal data in which the number of subjects was not sufficiently large.24–26
In this study, we evaluated the normative distribution of cpRNFL, mGCL and mIPL thickness in a relatively large number of Korean subjects using the Spectralis OCT and automatic segmentation software to provide normal reference ranges of mGCL and mIPL. To the best of our knowledge, this is the first report of average mGCL and mIPL thickness measured separately by automatic segmentation of Spectralis OCT in normal subjects including both children and adults.
The three main findings in regard to factors affecting mGCL and mIPL thickness are as follows. First, the effects of ageing on inner retinal thickness has been reported in older adults, but only a few information is available for children and young adults.6 In our study, the inner sector mGCL and mIPL thickness significantly increased with older age in children, but age-related changes tend to be opposite in adults. Second, gender-related difference was identified in both inner sector mGCL and mIPL, which was thicker in males. Finally, the inner and outer sector mGCL and mIPL thickness were positively correlated with average cpRNFL thickness.
Owing to the development of powerful segmentation algorithms, several studies demonstrated that mGCL+mIPL thickness measured by SD-OCT appeared to be a better biomarker of early structural injury in acute optic nerve diseases compared with cpRNFL thickness,27 28 and is also useful in the differentiation of chronic optic neuropathy.29–31 Previous studies also revealed the clinical usefulness of separating mGCL and mIPL.32 33 Isolated mGCL thickness appeared to have the best diagnostic performance of glaucoma compared with mIPL, macular RNFL and cpRNFL.32 33 Based on the histological findings that IPL is composed of retinal ganglion cell dendrites, amacrine cells and bipolar cells,34 we speculated that in conditions of retinal ganglion cell loss, the thinning of mIPL was less obvious due to abundant processes of bipolar and glycinergic amacrine and gamma-aminobutyric acidergic cells. Using the results of this study as reference values, Spectralis OCT could be more effectively used as a diagnostic tool in retinal and optic nerve diseases.
The mGCL and mIPL thickness were thicker in the parafoveal area (1.0–3.0 mm), and subsequently showed a slight decrease in the outer macular area. This is consistent with the normal anatomic macular contour.19 The mean mGCL+mIPL thickness in our study was 74.3±4.7 µm which is comparable to other studies using the Spectralis OCT (73.2 µm35), but thinner than the reported values using the Cirrus OCT (82.1 µm9–82.6 µm36). This discrepancy may have resulted from variable performances of automated segmentation software provided by different SD-OCT devices.19 Therefore, reference values obtained from one system cannot directly be applied to data acquired from other devices.7 8 There is only one previous study that measured the mGCL thickness in children, and the average values of each sector were significantly thinner than that of our study.37 This discrepancy can be explained by the fact that 15% (11/75) of subjects in the previous study had vision loss, while our study only included patients with normal visual acuity.37
Ethnic differences in macular thickness have been described.38–41 The mean central macular thickness of children in our study was approximately 14.0 µm thinner than that reported in non-Hispanic white children using Spectralis OCT.3 With respect to mGCL and mIPL thickness, the mean values were similar to that of other ethnicities, but the total macular thickness was different among ethnicities.24 35,24 Meanwhile, considerable variation of cpRNFL thickness among ethnicities has been reported; it has been found to be thinner in Caucasians as compared with Hispanics, Chinese and African-Americans.23 42 43 Even within the same ethnic group, recent studies show that inner retinal thickness parameters (cpRNFL and GCIPL) differ among Asian populations including Chinese, Filipino and Vietnamese descendants.25 Therefore, the normative range of mGCL and mIPL thickness in the Korean population may differ from other Asian ethnicities. However, comparisons among ethnicities are beyond the scope of the present study.
In the present study, age had a significant effect on inner sector mGCL and mIPL thickness. Subgroup analysis revealed a positive correlation between age and inner sector mGCL and mIPL thickness in children (5–17 years). From our results, we can presume that the inner sector mGCL thickness would increase by 0.174 µm per year in children aged 5–17 years. Conversely, a significant decrease with age was observed in the inner sector mGCL and mIPL thickness of adults (18–70 years). Previous reports using Spectralis or Cirrus OCT in adults demonstrated a negative correlation between age and mGCL thickness (Spectralis; −0.103 µm per year),19 or age and GCIPL thickness (Cirrus; −0.32 µm per year)44 which is consistent with the present study. Histological studies also support our results as the GCL and their axons (RNFL) are vulnerable to loss during ageing.42 45
Regarding GCIPL measurements in the paediatric population, a few studies have identified that GCIPL is positively correlated with age,46 47 whereas other studies have demonstrated no relationship.36 48 Histological evidence of the present study is limited because previous studies on the morphological development of the human fovea have focused on the prenatal period.49 The linear relationship between age and thickness measurements is hypothetical and should be verified through a longitudinal study.
In the present study, images were obtained only when the quality score was >15, a threshold for acceptable quality suggested by instrument manufacturers.50 A previous study demonstrated that when Spectralis images are obtained within the range of manufacturer-specified signal strength above 15, the variability of repeated measurements is likely to be clinically irrelevant.50
Our study also showed significant gender differences in the inner sector mGCL and mIPL thickness. The inner sector mGCL and mIPL were significantly thicker in males compared with females. There have been numerous studies reporting gender differences in the macular inner retinal layer thickness.51–55 Mwanza et al9 reported that macular GCIPL was significantly thicker in males compared with females. Ooto et al56 found significant thickness differences of the macular inner retinal layers (RNFL, IPL, inner nuclear layer) according to gender. Previous studies have shown that the fovea and extrafoveal region of the retina is approximately 10 µm thicker in males than that of females, which is consistent with our results.57–59 Therefore, gender differences should be accounted for interpretation of SD-OCT measurements.
In the practical aspect, age and gender-related changes in mGCL (up to 2.7 µm) and mIPL (up to 2.1 µm) are not negligible compared with the standard deviation of mGCL (2.8 µm) and mIPL (2.0 µm) thickness measurements in our study. Thus, in clinical research, adjustment for gender and age should be performed in the analysis of mGCL and mIPL thickness.
In terms of refractive error-related changes, only the outer sector mGCL was affected by myopia in our study. This finding is contradictory with the recent study which showed a thinner ganglion cell complex associated with myopia using the iVue SD-OCT (Optovue Inc., Fremont, CA).60 In a study performed with Cirrus OCT consisting of children aged 3–18 years, average, inferior and superior GCIPL thickness were significantly thinner with more myopic refractive errors.36 However, direct comparison is limited because there are differences in the range (−12.88 to +7.7536 vs −6.00 to +5.00 D in our study) and average value of SE refractive errors (−0.1±2.4,60 –3.5±3.636 vs −0.79±2.23 D in our study) among studies. Our study population included subjects with a limited range of refractive errors (+5.00 to −6.00 D); therefore, measurements obtained from subjects with higher refractive errors should be interpreted with caution.
There was a positive correlation between mGCL and cpRNFL thickness. Since cpRNFL consists of axons of the GCL, it is likely that a thinner GCL would lead to a thinner cpRNFL, and vice versa. This is also described in other studies where the GCIPL thickness correlated with cpRNFL thickness.9 10 19
Our study has a few limitations. First, as this was a cross-sectional study, certain cohort effects may have influenced the data across age groups, which is often found as an increase in myopia prevalence related to rapid industrialisation in East Asia.61 A larger cohort study is needed to confirm our findings; however, we believe that our study results are valid since average refractive errors were not significantly different between age groups and the results are similar to the previous studies on SD-OCT. Second, the present study was conducted in a tertiary hospital which is not sufficiently representative of the general population. In order to overcome this limitation, only patients with no other ocular abnormalities except for low refractive errors or dry eye syndrome confirmed by experienced ophthalmologists were included. By applying strict criteria, we sought to reduce concerns about the generalisability of hospital-based research. Third, there is lack of thorough evidence regarding the reliability of retinal layer thickness measurements determined by automated segmentation in SD-OCT.62 63 However, Tian et al demonstrated that automated segmentation using the Spectralis software could achieve accuracy close to the interobserver difference.62 In addition, automated segmentation of the Spectralis software showed excellent repeatability for all retinal layers except the outer plexiform layer (all intraclass correlation coefficients higher than 0.95 and coefficients of repeatability less than 2 µm).63 Besides, masked observers independently reviewed the accuracy of inner retinal layer segmentation (cpRNFL, mGCL and mIPL) and adequate centration on the fovea in our study. More than all, the standard deviation of all sectors was within 10% of the mean values suggesting an acceptable variability of thickness measurements. Finally, the patients were unequally distributed across all age groups. However, each age group in 10-year intervals included a sufficient number of subjects compared with the previous studies.21 43 51 52 64
In conclusion, the present hospital-based study describes normative ranges of the mGCL and mIPL thickness measured with Spectralis OCT in a relatively large number of Korean children and adults. Gender, age and cpRNFL thickness showed significant correlation with mGCL and mIPL thickness. Therefore, adjustment for gender and age is likely to be relevant in the analysis of mGCL and mIPL thickness in clinical research. Our data can be useful to researchers studying mGCL and mIPL thickness measured by Spectralis OCT.
The authors thank the Medical Research Collaborating Center at Seoul National University Bundang Hospital for help with statistical analyses.
YJY and J-MH contributed equally.
Contributors YJY: data collection, data analysis and interpretation, drafting the article. JMH: data analysis and interpretation, critical revision of the article. HKY: design of the work, critical revision and final approval of the article to be published.
Funding This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No 2017R1A2B4011450).
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The Institutional Review Board of Seoul National University Bundang Hospital approved the study, which was conducted in accordance with the tenets of the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.