Effects of Age, Sex, and Axial Length on the Three-Dimensional Profile of Normal Macular Layer Structures

Abstract

Purpose.: To identify sex-related differences and age-related changes in individual retinal layer thicknesses in a population of healthy eyes across the lifespan, using spectral domain optical coherence tomography (SD-OCT).

Methods.: In seven institutes in Japan, mean thicknesses of the retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor inner segment (IS), and photoreceptor outer segment (OS) were measured using SD-OCT with a new automated segmentation protocol in 256 healthy subjects.

Results.: Interoperator coefficients of variability for measurements of each layer ranged from 0.012 to 0.038. The RNFL, GCL, IPL, and INL were thinnest in the foveal area, whereas the OPL+ONL and OS were thickest in this area. Mean thicknesses of the INL and the OPL+ONL were significantly greater in men (P = 0.002 and 0.001, respectively). However, mean RNFL thickness was greater in women (P = 0.006). Thicknesses of the RNFL, GCL, IPL, INL, and IS correlated negatively with age. Thickness of the OPL+ONL was not correlated with age, and thickness of the OS correlated positively with age. Inner retinal (RNFL+GCL+IPL) thickness over the whole macula correlated negatively with age (P < 0.001), but outer retinal (OPL+ONL+IS+OS) thickness did not. Thicknesses of layers did not correlate with axial length.

Conclusions.: Macular layer thicknesses measured on SD-OCT images in healthy eyes showed significant variations by sex and age. These findings should inform macular layer thickness analyses in SD-OCT studies of retinal diseases and glaucoma.

Measurement of retinal thickness in the macula using optical coherence tomography (OCT) is an established method for diagnosing and monitoring macular edema and evaluating the efficacy of medical and surgical treatments for macular diseases. Recently, advances in OCT have allowed measurement of the thickness of specific retinal layers, which is useful for diagnosing and monitoring pathologic changes in various ocular diseases. ^{1 –10} Studies have found that the thickness of the photoreceptor layer, measured by ultra–high-resolution OCT, is correlated with the severity of photoreceptor loss associated with decreased visual acuity in retinitis pigmentosa. ^1,2 Studies using spectral-domain (SD) OCT showed that the thickness of the outer nuclear layer (ONL) in the fovea correlates with visual acuity in eyes with central serous chorioretinopathy, ^3,4 polypoidal choroidal vasculopathy, ⁵ and epiretinal membrane. ^6,7 Studies in which a layer segmentation algorithm was used with time-domain (TD) OCT and SD-OCT to automatically measure the combined thickness of the three innermost retinal layers (retinal nerve fiber layer [RNFL] + ganglion cell layer [GCL] + inner plexiform layer [IPL]), which are the most susceptible to glaucomatous damage, differed in the thickness of these layers between glaucomatous and healthy eyes. ^8,9 In a study using SD-OCT, GCL thickness was found to be reduced in patients with type 1 diabetes who had no or minimal diabetic retinopathy. ¹⁰ As these studies demonstrate, measuring the thickness of individual retinal layers is becoming a powerful tool for assessing and monitoring changes in the macula resulting from retinal diseases, glaucoma, and optic neuropathy. 

The usefulness of these technologies is predicated on an understanding of the normal anatomy of the many layers in the macula, each of which has its own normal 3D shape and may be affected in various ways by disease. However, to our knowledge, no normative database exists against which thickness maps of individual retinal layers in diseased eyes can be compared. 

Macular thickness measurements have been reported in healthy eyes by TD-OCT and SD-OCT ^{11 –28} and by three-dimensional raster scans. ^{11 –15} Earlier studies using TD-OCT ^{16 –22} or SD-OCT ^{23 –28} in healthy eyes reported changes in the macular profile in relation to factors such as age, sex, and axial length. It was only recently, however, that an algorithm was developed for SD-OCT that allows for measurement of the thickness of each macular layer in each of the nine areas of the macula. ²⁹ This development provides the unique opportunity to profile the 3D anatomy of individual macular layers. 

In this study, we used SD-OCT with the new automated retinal layer segmentation algorithm ²⁹ and a raster scanning protocol to measure the thickness of individual retinal layers in 256 healthy eyes of men and women in six age groups, with the objective of developing a database for assessing and monitoring intraretinal pathologies when using SD-OCT instruments. We evaluated mean measurements in nine macular areas/sectors on the basis of areas defined in the Early Treatment Diabetic Retinopathy Study (ETDRS) and evaluated the results according to subject age, sex, and axial length of eyes. 

Subjects and Methods

Subjects

In this prospective, cross-sectional, multicenter study, data were collected at seven clinical centers in Japan—Kyoto University (Kyoto, Japan), University of Tokyo (Tokyo, Japan), Gunma University (Maebashi, Japan), Osaka University (Suita, Japan), Niigata University (Niigata, Japan), Kanazawa University (Kanazawa, Japan), and Tajimi Municipal Hospital (Tajimi, Japan). The Institutional Review Board and Ethics Committee of each participating center approved the study, which adhered to the tenets of the Declaration of Helsinki, and written informed consent was obtained from each subject. 

Self-reported ophthalmologically healthy men and women 20 years of age and older were recruited for the study to ensure an even distribution of subjects by sex in six age groups (20–29, 30–39, 40–49, 50–59, 60–69, and 70+ years). Ocular examinations at the first visit included autorefractometry/keratometry, uncorrected and best-corrected visual acuity (BCVA) measurements using a 5-m Landolt chart, axial length measurement using an optical biometer (IOL Master; Carl Zeiss Meditec, Dublin, CA), slit-lamp examination, intraocular pressure measurement using Goldmann applanation tonometry, funduscopy, and visual field testing using a field analyzer (Humphrey Field Analyzer [HFA]; Carl Zeiss Meditec) with 24–2 Swedish Interactive Threshold Algorithm Standard Strategy. 

Exclusion criteria were contraindication to dilation, BCVA worse than 20/25, refractive error >5.0 or <−6.0 diopters, intraocular pressure ≥22 mm Hg to exclude glaucoma and ocular hypertension (high intraocular pressure may influence retinal thickness ³⁰ ), unreliable HFA results (fixation loss or >33% false-positive or false-negative results), abnormal HFA findings suggesting glaucoma according to the criteria of Anderson and Patella, ³¹ abnormal visual field loss consistent with ocular disease, history of intraocular surgery, evidence of vitreoretinal disease, and diabetes mellitus or other systemic disease that might affect the eye. 

Eyes were scanned by experienced operators using 3D OCT (OCT-1000; Topcon, Tokyo, Japan) after pupil dilation with 1% tropicamide. Color fundus photographs were taken immediately after SD-OCT scanning. 

Three glaucoma experts (MA, HA, AI) and two macular experts (NY, SK) examined all the color fundus photographs and reached consensus on whether an eye had evidence of glaucomatous optic neuropathy or another optic nerve abnormalities or whether it had evidence of retinal disease, respectively; such eyes were excluded. 

SD-OCT Measurements

SD-OCT examinations were performed according to the following scan protocol selected by the study group. Three-dimensional imaging data were acquired using a raster scan protocol of 512 × 128 (horizontal × vertical) axial scans per image (total, 65,536 axial scans/image). Each 3D raster scan covered a 6 × 6-mm area centered at the fixation point in the posterior pole and required ∼2.4 seconds for completion. Accurate scan lengths were obtained by correcting the magnification effect in each eye using the SD-OCT equipment manufacturer's formula (modified Littman's method ³² ) based on that eye's refractive error, corneal radius, and axial length. Only high-quality images with a Q-factor score >65 were used. 

To assess the interobserver reproducibility of the macular thickness measurements, the same macular scan protocol was performed separately by two operators on the same day on 46 subjects (21 men, 25 women; age, 43.0 ± 15.4 years). 

Measurements of Individual Retinal Layer Thickness by Macular Sector

A new algorithm based on a customized edge detector and edge refinement feature was recently developed ²⁹ to automatically calculate mean macular layer thicknesses of individual retinal layers (Fig. 1, Supplementary Movie S1) in nine separate areas (based on ETDRS sectors) within the macula (Fig. 2). All SD-OCT B-scan images were sent to the Kyoto University OCT Reading Center at the Kyoto University Graduate School of Medicine, where the novel multilayer segmentation algorithm was used to calculate the thickness of each retinal layer—RNFL, GCL, IPL, inner nuclear layer (INL), outer plexiform layer (OPL) + ONL, photoreceptor inner segment (IS), and photoreceptor outer segment (OS) (Fig. 1, Supplementary Movie S1). 

Figure 1.

 

Delineation of retinal layers on a 3D SD-OCT image. Thickness was measured between yellow lines for the RNFL, GCL, IPL, INL, OPL+ONL, IS (distance between the external limiting membrane and the inner border of the highly reflective line representing the inner and outer segment junction line), and OS (distance between the inner border of the inner and outer segment junction line and the inner border of the RPE).

Figure 2.

 

Delineation of the nine macular sectors, according to the ETDRS, within which we measured macular layer thickness. Green box: the 6 × 6-mm area for 3D raster scanning.

The 3D B-scan images were visually inspected and excluded if there was any evidence of algorithm failure, such as inaccurately drawn lines marking the boundaries between retinal layers. Each B-scan was checked by an investigator (SO), and only images with appropriate segmentation lines were included in this study. The segmentation algorithm allows for manual corrections, and minor errors were corrected manually. 

On each B-scan included in the study, the mean thickness of each retinal layer in each of the nine macular sectors was calculated. First, the central fovea on each image was manually registered, if necessary, to coincide with the central circle on the ETDRS segmentation diagram (Fig. 2). Because inner retinal layers are nearly absent in the fovea, only the outer retinal layer thicknesses were analyzed in this 1-mm diameter area in the center of the fovea. The mean thickness of each retinal layer was measured in four sectors (superior, inferior, nasal, and temporal) of the “inner ring” 1 to 3 mm from the center of the fovea and the “outer ring” 3 to 6 mm from the center of the fovea (Fig. 2). 

Statistical Analysis

The effects of age, sex, and axial length on macular layer thickness were analyzed using the Pearson product moment correlation coefficient. For evaluating sex-related differences and comparing retinal thicknesses among sectors, a t-test was used. To evaluate interobserver differences, coefficient of variation (CV) and intraclass correlation coefficient (ICC) were used. Statistical analyses were performed using a statistics software program (SPSS17; SPSS Inc., Chicago, IL). P < 0.05 was considered statistically significant. 

Results

Four hundred sixty-four subjects were initially enrolled in the study, but 189 were excluded, and acceptable OCT images were not obtained in 19 subjects. Thus, 256 subjects (130 men [51%], 126 women [49%]) ranging in age from 20 to 77 years (mean, 50.5 years) were included in the analysis. When both eyes qualified for the study, one eye was randomly chosen for inclusion. As shown in Table 1, men and women were equally represented in each age group. Differences between them in mean age and axial length were not significant (P = 0.556 and P = 0.080, respectively). 

Table 1.

 

Characteristics of 256 Subjects/Healthy Eyes

Age Group (y)	No.	Men/Women (ratio)	Mean Refractive Error (diopters)	Mean Axial Length (mm)	Mean IOP (mm Hg)
20–29	40	19/21 (0.9)	−1.5 ± 1.6	24.1 ± 0.9	13.8 ± 2.0
30–39	40	21/19 (1.0)	−1.3 ± 1.1	24.0 ± 0.9	14.3 ± 2.5
40–49	36	20/16 (1.3)	−1.1 ± 1.1	23.9 ± 0.8	14.5 ± 2.3
50–59	47	21/26 (0.8)	−0.7 ± 1.5	23.7 ± 1.0	13.9 ± 2.6
60–69	57	29/28 (1.0)	0.3 ± 1.1	23.4 ± 0.8	14.7 ± 2.3
70+	36	20/16 (1.3)	0.7 ± 1.6	23.2 ± 0.9	13.9 ± 2.4
Total	256	130/126 (1.1)	−0.5 ± 1.6	23.7 ± 0.9	14.2 ± 2.4

 Mean indicates mean ± 1 SD. When both eyes qualified for the study, one eye was randomly selected for inclusion.

Table 2 shows the reproducibility of measurements of retinal layer thickness. For measurements over the entire macular area, CVs ranged from 0.012 to 0.038; for measurements of various layers, ICCs ranged from 0.931 to 0.988. 

Table 2.

 

Reproducibility of Measurements of Normal Macular Layer Thickness Obtained Using 3D-OCT with a Multilayer Segmentation Algorithm

Macular Area	CV				ICC
	Center	Inner Ring	Outer Ring	Whole Macula	Center	Inner Ring	Outer Ring	Whole Macula
RNFL	N/A	0.030	0.035	0.031	N/A	0.949	0.961	0.959
GCL	N/A	0.009	0.020	0.012	N/A	0.973	0.980	0.988
IPL	N/A	0.014	0.024	0.016	N/A	0.973	0.987	0.987
INL	N/A	0.011	0.017	0.012	N/A	0.981	0.984	0.987
OPL+ONL	0.023	0.013	0.013	0.015	0.992	0.984	0.978	0.982
IS	0.025	0.027	0.035	0.029	0.986	0.866	0.939	0.931
OS	0.060	0.044	0.074	0.038	0.972	0.945	0.943	0.945

 N/A, not applicable; center, central fovea (1 mm); inner ring, 1–3 mm from the central fovea; outer ring, 3–6 mm from the central fovea.

Mean retinal layer thicknesses for all subjects are shown in Figure 3 and Table 3. Thickness maps (Fig. 3) show vertical symmetry for all retinal layers. In the central fovea, the RNFL, GCL, IPL, and INL became so thin that their thicknesses were nonmeasurable, and the OPL+ONL and OS were at their thickest. The IS displayed nearly uniform thickness in all areas. Areas of maximal thickness of the GCL, IPL, and INL formed circular or “C” patterns in the parafovea. In the inner ring, mean thicknesses of the GCL and INL in the nasal quadrants were significantly greater than in the temporal quadrants (Table 3; P < 0.001 for both). 

Figure 3.

 

Mean thickness maps for each retinal layer in 256 ophthalmologically healthy subjects. The color spectrum to the right of each image shows the thickness range for the layer measured in that image. Right: nasal; left: temporal. (A) RNFL, (B) GCL, (C) IPL, (D) INL, (E) OPL+ONL, (F) IS, and (G) OS.

Table 3.

 

Mean Retinal Layer Thickness Measurements in Macular EDTRS Sectors

Retinal Layer	Center	Inner Ring				Outer Ring				Whole Macula
		Upper	Lower	Nasal	Temporal	Upper	Lower	Nasal	Temporal
RNFL	N/A	25.0 ± 3.3	25.8 ± 3.0	21.9 ± 3.4	17.7 ± 2.7	37.9 ± 5.4	39.1 ± 6.0	48.5 ± 7.6	19.4 ± 2.5	32.2 ± 3.7
GCL	N/A	59.3 ± 6.9	58.7 ± 6.5	59.9 ± 7.4	54.6 ± 7.1	36.6 ± 3.1	34.6 ± 2.8	40.7 ± 4.0	39.1 ± 4.0	42.2 ± 3.0
IPL	N/A	36.9 ± 3.4	37.4 ± 3.5	35.8 ± 3.5	35.4 ± 3.6	30.9 ± 3.1	29.0 ± 2.7	32.2 ± 3.3	34.7 ± 3.2	32.6 ± 2.3
INL	N/A	39.3 ± 3.9	39.7 ± 3.9	41.1 ± 4.1	37.1 ± 4.1	31.4 ± 2.6	30.6 ± 2.5	33.5 ± 2.9	32.6 ± 2.9	33.4 ± 2.1
OPL+ONL	103.9 ± 10.6	91.8 ± 8.5	88.3 ± 8.3	94.1 ± 8.8	93.1 ± 9.1	74.0 ± 7.3	66.5 ± 6.3	74.3 ± 7.4	72.2 ± 7.6	77.6 ± 5.4
IS	26.9 ± 2.3	23.1 ± 3.3	22.8 ± 2.5	23.7 ± 2.6	24.1 ± 3.2	23.0 ± 3.5	22.1 ± 2.5	22.1 ± 2.8	23.8 ± 4.4	23.1 ± 2.5
OS	39.8 ± 8.2	39.1 ± 7.2	38.5 ± 6.9	37.7 ± 7.8	37.7 ± 7.5	38.2 ± 5.3	37.1 ± 4.2	36.5 ± 5.3	37.9 ± 5.5	37.8 ± 4.8

 Values are mean ± SD (μm). N/A, not applicable; center, central fovea (1 mm); inner ring, 1–3 mm from the central fovea; outer ring, 3–6 mm from the central fovea.

Mean thicknesses of the INL and OPL+ONL were significantly greater in men than in women (P = 0.002 and P = 0.001, respectively; Fig. 4, Table 4). However, in women, the mean RNFL thickness was greater (P = 0.006), especially in the peripheral macula (outer ring) (P = 0.001) (Fig. 4, Table 4). 

Figure 4.

 

Mean thickness maps for men (left) and women (right), calculated based on all subjects (n = 130 men, n = 126 women), for (A) RNFL, (B) INL, and (C) OPL+ONL. The color spectrum to the right of each image shows the thickness range for the layer measured in that image. Right: nasal; left: temporal.

Table 4.

 

Sex Differences in Mean Macular Layer Thicknesses

Macular Layer	Total (n = 256)	Men (n = 130)	Women (n = 126)	P *
RNFL
Whole	32.2 ± 3.7	31.6 ± 3.7	32.8 ± 3.7	0.006†
Inner ring	22.7 ± 2.3	22.7 ± 2.4	22.7 ± 2.3	0.875
Outer ring	36.3 ± 4.5	35.4 ± 4.4	37.2 ± 4.4	0.001†
GCL
Whole	42.2 ± 3.0	42.2 ± 3.2	42.2 ± 2.8	0.994
Inner ring	58.3 ± 5.6	58.4 ± 6.1	58.2 ± 5.1	0.785
Outer ring	37.9 ± 2.5	37.9 ± 2.5	37.9 ± 2.4	0.989
IPL
Whole	32.6 ± 2.3	32.8 ± 2.4	32.5 ± 2.2	0.349
Inner ring	36.5 ± 2.4	36.8 ± 2.6	36.2 ± 2.3	0.032†
Outer ring	31.8 ± 2.4	31.9 ± 2.5	31.7 ± 2.4	0.615
INL
Whole	33.4 ± 2.1	33.8 ± 2.0	33.0 ± 2.1	0.002†
Inner ring	39.4 ± 3.0	40.2 ± 3.0	38.7 ± 2.7	<0.001†
Outer ring	32.2 ± 2.0	32.4 ± 1.9	31.9 ± 2.1	0.034†
OPL+ONL
Whole	77.6 ± 5.4	78.6 ± 5.4	76.4 ± 5.2	0.001†
Center	103.9 ± 10.6	106.0 ± 9.0	102.4 ± 7.5	0.001†
Inner ring	92.2 ± 6.3	93.6 ± 6.6	90.6 ± 5.7	<0.001†
Outer ring	72.1 ± 5.4	73.0 ± 5.3	70.9 ± 5.3	0.002†
IS
Whole	23.1 ± 2.5	22.9 ± 1.9	23.2 ± 3.0	0.368
Center	26.9 ± 2.3	27.0 ± 1.6	27.0 ± 1.7	0.933
Inner ring	23.5 ± 2.3	23.3 ± 1.6	23.7 ± 2.9	0.233
Outer ring	22.8 ± 2.8	22.7 ± 2.2	22.9 ± 3.2	0.441
OS
Whole	37.8 ± 4.8	38.0 ± 4.6	37.6 ± 5.1	0.548
Center	39.8 ± 8.2	39.6 ± 7.4	40.2 ± 8.2	0.536
Inner ring	38.4 ± 6.7	38.3 ± 6.5	38.5 ± 7.0	0.814
Outer ring	37.5 ± 4.4	37.8 ± 4.1	37.2 ± 4.7	0.301

 Values are mean ± SD (μm). Center, central fovea (1 mm); inner ring, 1–3 mm from the central fovea; outer ring, 3–6 mm from the central fovea.

*   t-test.

†  Statistically significant.

As Figure 5 shows, thicknesses of the RNFL, GCL, IPL, INL, and IS decreased with increasing age. The thickness of the OPL+ONL in the peripheral macula did not vary with age; however, foveal thickness increased with increasing age, as did OS thickness. 

Figure 5.

 

Mean thickness maps of 80 subjects 20 to 39 years of age (left), 83 subjects 40 to 59 years of age (middle), and 93 subjects 60 years and older (right) for (A) RNFL, (B) GCL, (C) IPL, (D) INL, (E) OPL+ONL, (F) IS, and (G) OS. The color spectrum to the right of each image shows the thickness range for the layer measured in that image. Right: nasal; left: temporal.

Table 5 shows that after adjusting for axial length, the thicknesses of the RNFL, GCL, IPL, INL, and IS correlated negatively with age. Age-related losses in the thicknesses of the RNFL, GCL, IPL, INL, and IS over the whole macula were −0.05, −0.07, −0.05, −0.03, and −0.05 μm/year, respectively. OS thickness correlated positively with age, as did the thickness of the OPL+ONL in the central foveal sector. Supplementary Table S1 shows that after adjusting for axial length, the thicknesses of the GCL, IPL, INL, and IS correlated negatively with age in men and women, whereas RNFL thickness correlated negatively and OS thickness correlated positively with age only in women. 

Table 5.

 

Correlations of Age with Thickness of Macular Layers before (P *) and after (P †) Adjusting for Axial Length

Macular Layer	Regression	R	P *	P † (adjusted for axial length)
RNFL
Whole	34.7 − 0.05*age	−0.223	<0.001	0.001
Inner ring	24.5 − 0.04*age	−0.253	<0.001	<0.001
Outer ring	39.0 − 0.05*age	−0.196	0.001	0.002
GCL
Whole	45.6 − 0.07*age	−0.383	<0.001	<0.001
Inner ring	65.4 − 0.14*age	−0.414	<0.001	<0.001
Outer ring	40.3 − 0.05*age	−0.327	<0.001	<0.001
IPL
Whole	35.4 − 0.05*age	−0.403	<0.001	<0.001
Inner ring	38.4 − 0.04*age	−0.261	<0.001	<0.001
Outer ring	34.9 − 0.06*age	−0.422	<0.001	<0.001
INL
Whole	35.1 − 0.03*age	−0.261	<0.001	<0.001
Inner ring	40.2 − 0.01*age	−0.087	0.160	0.060
Outer ring	34.2 − 0.04*age	−0.335	<0.001	<0.001
OPL+ONL
Whole	78.2 − 0.01*age	−0.041	0.509	0.485
Center	100.2 + 0.08*age	0.159	0.010	0.005
Inner ring	91.3 − 0.02*age	−0.045	0.470	0.308
Outer ring	73.5 − 0.03*age	−0.096	0.123	0.128
IS
Whole	25.4 − 0.05*age	−0.306	<0.001	<0.001
Center	29.3 − 0.05*age	−0.445	<0.001	<0.001
Inner ring	26.1 − 0.05*age	−0.370	<0.001	<0.001
Outer ring	25.1 − 0.04*age	−0.270	<0.001	<0.001
OS
Whole	35.3 + 0.05*age	0.167	0.007	0.002
Center	35.8 + 0.08*age	0.176	0.004	0.002
Inner ring	34.8 + 0.07*age	0.173	0.005	0.002
Outer ring	35.5 + 0.04*age	0.152	0.013	0.004

 Center, central fovea (1 mm); inner ring, 1–3 mm from the central fovea; outer ring, 3–6 mm from the central fovea.

*   P using Pearson's correlation coefficient.

†   P using partial correlation coefficient.

Figure 6 and Table 6 show thicknesses of the combined inner retinal layers (RNFL, GCL, and IPL), known as the ganglion cell complex (GCC), versus the outer retinal layers (ORL; consisting of the OPL+ONL, IS, and OS). After adjusting for axial length, GCC thickness over the whole macula correlated negatively with age (P < 0.001), but ORL thickness did not (Table 6). 

Figure 6.

 

Mean thickness maps in 80 subjects 20 to 39 years of age (left), 83 subjects 40 to 59 years of age (middle), and 93 subjects 60 years of age and older (right) for the inner retinal sectors. (A) GCC, consisting of the RNFL + GCL + IPL. (B) Outer retinal layers (ORL), consisting of the OPL + IS + OS. The color spectrum to the right of each image shows the thickness range for the layer measured in that image. Right: nasal; left: temporal.

Table 6.

 

Correlation between Age and Inner or Outer Retinal Layer Thickness before (P *) and after (P †) Adjusting for Axial Length

Macular Layer	Regression	R	P *	P † (adjusted for axial length)
GCC
Whole	115.8 − 0.17*age	−0.387	<0.001	<0.001
Inner ring	128.4 − 0.21*age	−0.412	<0.001	<0.001
Outer ring	114.3 − 0.15*age	−0.351	<0.001	<0.001
ORL
Whole	139.0 − 0.01*age	0.023	0.705	0.915
Center	165.3 + 0.12*age	0.167	0.007	0.002
Inner ring	152.25 + 0.04*age	0.062	0.317	0.151
Outer ring	134.17 − 0.04*age	−0.077	0.215	0.362

 Whole, whole macula; inner ring, 1–3 mm from the central fovea; outer ring, 3–6 mm from the central fovea.

*   P using Pearson's correlation coefficient.

†   P using partial correlation coefficient.

Layer thickness was not significantly correlated with axial length after adjusting for age (Supplementary Table S2). 

Discussion

Several studies of normal macular layer thickness have used TD-OCT measurements ^{33 –35} ; however, the acquisition of accurate 3D macular measurement data using TD-OCT is limited by the imaging speed and axial resolution of these instruments. Using TD-OCT, mean macular layer thickness is calculated based on only six radial linear scans of 6 mm each, requiring interpolation to fill in the nonimaged areas. SD-OCT instruments allow 43× to 125× faster image acquisition and provide approximately double the axial resolution (∼3–7 μm) and a higher signal-to-noise ratio than the Stratus OCT. These features provide images on which measuring fine details of macular structures is possible, specifically the thicknesses of individual retinal layers. Loduca et al. ³⁶ recently reported the use of Spectralis (Heidelberg Engineering, Heidelberg, Germany), a type of SD-OCT instrument, to map the thicknesses of retinal layers on 19 horizontal B scans in 15 healthy eyes. We used a different type of SD-OCT instrument, together with 3D raster scanning, to measure retinal layer thickness in a large population of sex-matched, age-stratified groups. 

The clinical usefulness of any imaging system depends on the reproducibility of the measurements it is used to obtain. In this study, interoperator CVs ranged from 0.012 to 0.038, and ICCs ranged from 0.931 to 0.988. In a study of 43 healthy eyes using the same SD-OCT system with a multilayer segmentation algorithm, intervisit CVs ranged from 0.021 to 0.044 (Hangai M, et al. IOVS 2010;51:ARVO E-Abstract 221). These results indicate that the SD-OCT system with the multilayer segmentation algorithm permits reliable measurement of the thickness of each retinal layer. 

Mean Retinal Layer Thickness

The mean values we obtained for retinal layer thickness were similar to those obtained in other studies of healthy eyes. ^{36 –38} In our study, the mean RNFL thickness averaged over the whole macula was 32.2 ± 3.7 μm, compared with the values reported by Loduca et al. (36 ± 7 μm) ³⁶ and Kotera et al. (29 ± 4 μm), ³⁷ who also used SD-OCT. Similar small differences can be found among RNFL thicknesses measured in studies that used TD-OCT. ⁸  

Furthermore, the thicknesses of the GCL, IPL, and INL in our study (Table 3) were comparable to measurements of GCL, IPL, and INL thicknesses averaged over all ETDRS segments, excluding the central fovea in the TD-OCT study. ⁸  

Using TD-OCT, Ishikawa et al. ³³ found a mean of 51 ± 4 μm for OPL thickness and 94 ± 7 μm for mean outer retinal complex (ONL+IS+OS) thickness; the total thickness of these layers (145 μm) was slightly greater than what we obtained (138.5 ± 7.8 μm). We believe this can be accounted for by the differences in study populations and imaging systems. 

Standard SD-OCT images acquired along the optical axis typically do not show axons of the photoreceptor nuclei or Henle's fiber layer (HFL). Recently, Lujan et al. ³⁹ and Otani et al. ⁴⁰ were able to identify the HFL by varying where the SD-OCT beam was directed through the pupil. With this maneuver, they were able to view reflected light from the HFL when the beam was directed eccentrically, toward the side of the fovea opposite the side of beam entry. When they included the HFL in measurements of ONL thickness, these measurements increased by 52%. ³⁹ These results demonstrate that accurate measurement of ONL thickness is challenging. Thus, in the study we report here, we did not evaluate ONL thickness alone; instead we measured the combined thickness of the OPL and ONL. 

Macular Sector Differences in Retinal Layer Thickness

We found that the GCL and INL were thinner in the temporal sector than in the nasal sector within the inner ring (1–3 mm from the center of the fovea) but not within the outer ring (3–6 mm from the foveal center). The circular or “C”-like shapes of the GCL, IPL, and INL in the parafovea were similar to the findings of a histologic study wherein the spatial distribution of ganglion cells was quantified in whole mounts of healthy human retinas. ⁴¹ Within the central fovea, ganglion cell density was the highest in a horizontally oriented ellipse 0.4 to 2.0 mm from the foveal center. The number of ganglion cells was lower in the temporal sector than in the nasal sector within 0.5 to 2 mm of the foveal center, but the difference was minimal within 2 to 3 mm of the foveal center, which may explain the thinner GCL and INL in the temporal sector than in the nasal sector within the inner ring (0.5–1.5 mm from the center of the fovea) but not within the outer ring (1.5–3 mm from the foveal center). We also found that peak INL thickness occurred in the same area as peak GCL thickness; thus, we speculate that the densities of bipolar cells, amacrine cells, horizontal cells, and Müller cells might be high in these areas. 

Numerous studies have found that total retinal thickness is greater in the nasal quadrant than in the temporal quadrant and greater in the ETDRS inner ring than in the outer ring. ^{11,13 –15,17,19 –28} Thickness maps for each retinal layer in the present study clearly show that the RNFL is vertically but not horizontally symmetrical. This asymmetry of the RNFL probably accounts for the finding that total macular thickness in the outer ring is less symmetrical than in the inner ring. 

Differences due to Sex

Several studies have reported sex-related differences in total macular thickness and in thicknesses in some sectors. ^{13,15,19,20,25,28} For example, mean central foveal sector thickness was found to be significantly greater in men than in women, and mean retinal thicknesses in all quadrants of the inner ring and in the temporal quadrant of the outer ring were found to be significantly greater in men than in women. ²⁸  

This study is the first to identify sex-related differences in the thicknesses of individual retinal layers. We found that the INL and OPL+ONL were significantly thicker in men, whereas the macular RNFL was thicker in women. In the central foveal sector, ETDRS inner ring, and temporal quadrant of the outer ring, where the RNFL is relatively thin, the sex-related difference in INL or OPL+ONL thickness may be responsible for the observed sex-related difference in total retinal thickness. 

Changes with Age

OCT measurements have demonstrated a negative correlation between age and total retinal thickness, ^16,21,22,25 and one study found that in the fovea, mean thickness was not correlated with age, but in 5 of 8 extrafoveal sectors, thickness correlated negatively with age. ²⁸ Using SD-OCT (HD-OCT), Song et al. ²⁵ studied 198 eyes and reported that average extrafoveal retinal thicknesses decreased as mean subject age increased, although no significant correlation was found between mean central foveal sector thickness and age. 

In this study, the thicknesses of the RNFL, GCL, IPL, INL, and IS correlated negatively with age. However, the thickness of the OPL+ONL was not correlated with age, and the thickness of the OS correlated positively with age. 

Of the three primary retinal layers in the central fovea, with increasing age, the OPL+ONL did not change in thickness, the OS thickened, and the IS thinned. The varying effects of aging on the layers of the fovea indicated why previous studies found no overall statistically significant correlations between aging and foveal retinal layer thickness. ^22,25,28 Similarly, our findings support previous studies showing that total retinal thickness in extrafoveal regions is negatively correlated with age. ^{16,21,22,25,28} Among the retinal layers we examined, only the thickness of the OS was positively correlated with age in the inner and outer ring. 

Aging has been shown to be associated with loss of neurons in the inner retina. ^{42 –46} Histologic studies showed age-related losses of 0.3% to 0.6% of retinal neurons per year, ^42,43 whereas age-related thinning of the RNFL occurs at a lower rate of ∼0.2% per year. ⁴⁵ Most OCT studies report similar findings, including decreases in the mean circumpapillary or peripapillary RNFL thickness with age. ^{16,22,44 –46} We found a linear decrease in macular RNFL and GCL thickness with age, with negative slopes of −0.05 and −0.07 μm/year (0.2% and 0.2%/year), respectively. We also found age-related thinning of other inner retinal layers (IPL and INL), suggesting that aging is associated with loss of other neurons or glial cells in the INL. 

Fewer studies have examined the effects of aging on the outer retina. In one study, ²⁸ the mean thickness of the central fovea, which is composed primarily of outer retinal layers, was not correlated with age, whereas another study found a slight increase in thickness of the central foveal sector with increasing age. ²² In this study, the thickness of the outer retinal layers did not change in the inner ring and outer ring but was positively correlated with age in the central foveal sector. Taken together, these findings indicate that retinal thickness in the outer fovea does not decrease with age. This is supported by histologic findings. Curcio et al. ⁴⁷ counted cones and rods in 27 whole-mounted retinas from donors 27 to 90 years old with macroscopically healthy fundi and reported that changes in the cone density showed no consistent relationship with age, whereas rod density decreased by 30% from the youngest to the oldest specimen. Thus, foveal thickness may not decrease because of the following reasons: foveal cones remain stable with aging; subclinical vitreous traction has positive effects on the fovea; the OS thickens because of decreased retinal pigment epithelium phagocytosis with age. 

Effect of Axial Length

We found no correlation between total retinal thickness and age-adjusted axial length for any ETDRS sector. No previous study examined this correlation by sector, and results were contradictory in studies that examined total retinal thickness over the entire macula. For example, a study that included subjects with high myopia found a correlation between total macular thickness and axial length, ^18,19,25 but a study that excluded highly myopic eyes ²⁸ and studies that included highly myopic eyes ^48,49 found no significant correlation. Controversial results have also been reported for correlations between peripapillary RNFL thickness and axial length or refractive error. ^{45,46,50 –52} Several differences in study design could have contributed to these inconsistent results, such as inclusion of eyes with high myopia and adjustment for age or correct magnification. The relationship between retinal layer thickness and axial length should be investigated further, especially for subjects with highly myopic eyes. 

Study Limitations

Although several authors have measured OS thickness as the distance between the IS/OS and the RPE, ^5,36,53 OS thickness may be between the IS/OS and OS tip line. Thus, the OS thickness defined in the present study and the true OS thickness might differ. RPE and OS tip lines are difficult to identify independently in some subjects, and OS thickness might be underestimated in them. 

Another limitation of our study is that the 3D-OCT raster-scan protocol may not cover the entire ETDRS region in some eyes: the ETDRS sectors are encompassed in a 6-mm diameter circle, whereas the raster scan is a 6 × 6-mm square area. Thus, even if the central fovea and the central sector of the ETDRS layout match exactly, the corners of the outer ring may be omitted. Similar problems are encountered when obtaining measurements with the Stratus TD-OCT. 

A further limitation was that although we studied a large number of people evenly distributed by age and sex, all subjects were Japanese adults without high myopia; results could differ among other population groups or groups with high myopia. Specifically, foveal and extrafoveal thickness differences between different ethnic groups ²⁰ and between eyes with and without high myopia ¹⁹ have been reported. Thus, studies similar to ours on the relationships between macular layer thickness and age, sex, and axial length in other ethnic groups and among those with high myopia would add clarity. 

In summary, 3D macular layer dimensions in a large healthy population were profiled using an SD-OCT instrument. Our findings indicate that the effects of sex and age on macular dimensions vary by retinal layer; these subject variables must be considered while interpreting macular layer thickness data. Evaluation of the thickness of the ONL or photoreceptor layer may be used to predict visual prognosis in various macular diseases, and these thicknesses may also be used for monitoring the progression of retinitis pigmentosa. GCL thickness may be used to diagnose early-stage glaucoma and to evaluate the efficacy of medical and surgical treatments for glaucoma. Our findings provide basic information to facilitate macular layer thickness analysis using SD-OCT instruments for assessing various retinal diseases and glaucoma. 

Supplementary Materials

Table st1, PDF - Table st1, PDF 

Movie sv01, MOV - Movie sv01, MOV 

Footnotes

 Supported in part by Grant-in-Aid for Scientific Research 21796179 from the Japan Society for the Promotion of Science and by Topcon Inc. (Tokyo, Japan).

Footnotes

 Disclosure: S. Ooto, None; M. Hangai, Topcon Inc. (C); A. Tomidokoro, None; H. Saito, None; M. Araie, Topcon Inc. (C); T. Otani, None; S. Kishi, None; K. Matsushita, None; N. Maeda, Topcon Inc. (C); M. Shirakashi, None; H. Abe, None; S. Ohkubo, None; K. Sugiyama, None; A. Iwase, None; N. Yoshimura, Topcon Inc. (C)