Effects of age and disc area on optical coherence tomography measurements and analysis of correlations between optic nerve head and retinal nerve fibre layer

Authors

  • Sheng-Yao Hsu MD,

    Corresponding author
    • Department of Ophthalmology, Buddhist Tzu Chi General Hospital, Hualien, Taiwan and Institute of Medicine and Medical Sciences, College of Medicine, Tzu Chi University, Hualien, Taiwan
    Search for more papers by this author
  • Mei-Lan Ko MD,

    1. Department of Ophthalmology, National Taiwan University Hospital and Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsin Chu City, Taiwan
    Search for more papers by this author
    • These authors contributed equally to this work.
  • George Linn MD PhD,

    1. Department of Medicine, College of Medicine, Tzu Chi University and Director, Clinical Trial Center, Tzu Chi General Hospital, Hualien, Taiwan
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Ming-Shien Chang PhD,

    1. Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei, Taiwan
    Search for more papers by this author
  • Min-Muh Sheu MD,

    1. Department of Ophthalmology, Buddhist Tzu Chi General Hospital and Department of Ophthalmology and Visual Science, Tzu Chi University, Hualien, Taiwan
    Search for more papers by this author
  • Rong-Kung Tsai MD PhD

    1. Department of Ophthalmology, Buddhist Tzu Chi General Hospital and Department of Ophthalmology and Visual Science, Tzu Chi University, Hualien, Taiwan
    Search for more papers by this author

Corresponding author:

Dr Sheng-Yao Hsu

Department of Ophthalmology

Buddhist Tzu Chi General Hospital

No. 707, Section 3, Chung Yang Road

Hualien

TAIWAN

E-mail: wps59@yahoo.com.tw

Abstract

Background

The aim here was to investigate whether optic nerve head (ONH) parameters or retinal nerve fibre layer (RNFL) thickness correlate with age or disc area and whether the neuroretinal rim correlates with RNFL thickness.

Methods

This cross-sectional study enrolled 133 healthy subjects and analysed one randomly selected eye of each subject. All measurements of ONH parameters (including neuroretinal rim, disc and cup areas and cup-to-disc ratios) and RNFL thickness (global and quadrants) were taken by a single experienced operator using optical coherence tomography (OCT).

Results

Of the rim parameters analysed, average nerve width (the height of the nerve fibre bundle) was independent of age or disc area (p > 0.05). Disc area correlated positively with cup area (p < 0.05) but not with cup-to-disc ratios (p > 0.05). Of the RNFL thickness measurements analysed, temporal RNFL was independent of both age and disc area (p > 0.05). According to the analysis of the correlation between RNFL thickness and neuroretinal rim, global or non-temporal RNFL correlated positively with horizontal integrated rim width (p < 0.05, F > 4.000) and temporal RNFL was independent of all rim parameters (p > 0.05, F < 4.000).

Conclusion

Aging effect on neuroretinal rim loss or RNFL thickness change is non-uniform, and age is not a constant confounder when using OCT. The temporal RNFL is independent of age, disc area and neuroretinal rim.

Optical coherence tomography (OCT) enables the imaging of retinal layers and is reportedly effective for detecting abnormal retinal nerve fibre layer (RNFL) thickness in glaucomatous eyes.[1, 2] Some studies reported that the average global RNFL thickness is not significantly correlated with age[3, 4] but others indicate that age-related changes appear in either the global[5-8] or quadrantic[5-7] RNFL. In addition, some studies reported no significant correlations between RNFL thickness and disc area,[3, 4] while others indicated that RNFL thickness correlates positively with disc area.[8-10] These conflicting reports of correlations between RNFL thickness and age or between RNFL thickness and disc area indicate the need for further elucidation of these issues.

The OCT measurements can evaluate the optic nerve head (ONH) and analyse the neuroretinal rim and optic disc-cup structure (disc and cup areas and cup-to-disc ratios). Previous OCT studies have shown that some ONH parameters are age-related or correlate positively with disc area,[5, 11, 12] but some ONH parameters are not age-related or do not correlate with disc area.[11, 12] Therefore, the relationships between the ONH and age or between the ONH and disc area also need further elucidation.

Early optical coherence tomographers (OCT 2000, Humphrey Systems Inc, San Leandro, CA, USA) and the relatively more advanced model used for Stratus OCT (Model 3000, Carl Zeiss Meditec Inc, Dublin, CA, USA) all provide time-domain OCT.[13, 14] Although time-domain OCT has been partly superseded by the newer spectral domain technology, there have been studies that show that time-domain and spectral-domain OCT perform comparably in diagnosing glaucomatous optic nerve and RNFL damage.[15-18]

The correlations between neuroretinal rim and RNFL thickness and the relative importance of OCT measurements have not been thoroughly investigated. This study measured the ONH and RNFL using OCT in a population of healthy subjects to investigate whether ONH parameters or RNFL correlates with age or disc area and whether neuroretinal rim correlates with RNFL thickness. The objective was to investigate the effects of age and disc area on OCT measurements and to identify independent OCT parameters.

Methods

This observational cross-sectional study analysed 133 healthy subjects with the average age of 38.4 ± 19.6 years. The study was approved by the Review Board Ethics Committee at Tzu Chi General Hospital and was performed in accordance with Declaration of Helsinki principles. In each subject, one eye was randomly selected for OCT measurements (Stratus OCT, Model 3000, version 4.0, Carl Zeiss Meditec Inc). Our preliminary data revealed good intra-observer and inter-observer reproducibility before and after pupillary dilation.[19] In the current study, all measurements without pupillary dilation were taken in a single day by a single experienced operator.

Subjects with spherical equivalent between -1.00 dioptre (D) and +1.00 D were enrolled and each subject had a Snellen visual acuity better than 6/7.5 and no dry eye syndrome or history of systemic disease or ocular surgery. The intraocular pressures were lower than 21 mmHg by Goldmann applanation tonometry (Goldmann AT 900 / 870, Haag-Streit, Switzerland). In each subject, fundoscopic examinations of the disc in each eye without pupillary dilation were performed using a direct ophthalmoscope (Welch Allyn Inc, Skaneateles Falls, NY, USA) and a Canon CR-DGi non-mydriatic retinal camera (Canon Inc, Tochigiken, Japan). None of our subjects had a cup-to-disc ratio between-eye asymmetry exceeding 0.2 or abnormal disc appearance such as tilt, haemorrhage, notch, pallor, neuroretinal rim defect or peripapillary atrophy. Normal visual field examinations in each subject were confirmed by both frequency doubling technology (FDT, Zeiss-Humphrey, Welch Allyn, Dublin, CA, USA), screening C-20 program and static automated 24-2 SITA standard strategy (Humphrey Field Analyzer; Carl Zeiss Meditec Inc).

The OCT examinations were performed five minutes after applying one drop of lubricant to the eye and were performed in a dark room. The images with signal strength of at least seven were considered acceptable. Fast ONH scan protocol consisted of six cross-sectional radial scans, each consisting of 128 A-scans. The edge of the ONH was determined by projecting a line between the retinal pigment epithelium and the choriocapillaris.[18, 20] The structures below and above the line were defined as the optic cup and the neuroretinal rim, respectively. The ONH analysis included the following measurements: rim area (vertical cross section in mm[2]), average nerve width (the height of the nerve fibre bundle in mm),[20] rim length (mm), vertical integrated rim area (volume in mm3), horizontal integrated rim width (circular area in mm2), disc area (mm2), cup area (mm2), rim area (mm2), cup/disc area ratio, cup/disc horizontal ratio and cup/disc vertical ratio. The fast RNFL scan protocol included three consecutive 3.4 mm diameter circular scans, each consisting of 256 A-scans. Mean RNFL thicknesses were calculated for a global circle and four quadrants. The RNFL analysis also included the following thickness measurements of RNFL: global area, superior quadrant, inferior quadrant, temporal quadrant and nasal quadrant.

Statistical analyses were performed using SPSS software (SPSS 15.0, SPSS Institute Inc, Chicago, IL, USA) and MedCalc (version 11.6.1.0, MedCalc Software, Mariakerke, Belgium). Pearson correlation method was used to analyse the correlations of ONH or RNFL thickness with age or disc area. Stepwise multiple regression analysis was used to identify neuroretinal rim affecting RNFL thickness. Deming regression analysis was adopted to reveal the correlation between RNFL thickness and neuroretinal rim. The coefficient (r) was used to estimate the degree of correlation. A p-value lower than 0.05 or an F-value greater than 4.000 was considered statistically significant. For multiple comparisons, the alpha levels were adjusted by the Bonferroni modified test.

Results

Figure 1 shows the age distribution of individuals. The age range was 10 to 77 years with the mean and standard deviation of 38.4 ± 19.6 years. Mean spherical equivalent was 0.017 ± 0.871 D (range, -1.00 D to 0.75 D). Table 1 shows the ONH and RNFL measurements, including neuroretinal rim, disc and cup areas, cup-to-disc ratios and RNFL thickness.

Figure 1.

Age distribution of subjects

Table 1. Optic nerve head and peripapillary RNFL thickness measurements
ParametersMean ± SD95% CIRange
  1. RNFL: retinal nerve fibre layer, SD: standard deviation, CI: confidence interval
Optic nerve head   
Rim area (vertical cross section)0.2122 ± 0.09570.185–0.2620.030–0.635
Average nerve width0.351 ± 0.0320.33–0.370.21–0.45
Rim length1.438 ± 0.3001.37–1.510.47–2.19
Vertical integrated rim area0.4224 ± 0.19940.384–0.4610.119–1.093
Horizontal integrated rim width1.7333 ± 0.21851.691–1.7751.169–2.398
Rim area1.9418 ± 0.39111.867–2.0171.101–3.159
Disc area2.7234 ± 0.40572.646–2.8011.812–3.956
Cup area0.7816 ± 0.33680.717–0.8460.067–1.510
Cup/disc area ratio0.2847 ± 0.11350.263–0.3060.035–0.491
Cup/disc horizontal ratio0.5708 ± 0.13210.545–0.5960.183–0.841
Cup/disc vertical ratio0.4780 ± 0.11400.456–0.5000.185–0.703
RNFL   
Global RNFL107.785 ± 9.486105.49–110.0873.91–139.53
Superior RNFL133.324 ± 16.908129.23–137.4291.00–179.00
Inferior RNFL140.603 ± 16.888136.52–144.6997.00–188.00
Temporal RNFL77.427 ± 12.90875.71–81.9644.00–114.00
Nasal RNFL78.838 ± 16.37873.46–81.3938.00–132.00

Table 2 shows that age correlated negatively with rim parameters, including vertical integrated rim area, horizontal integrated rim width and rim area (p < 0.05) but did not significantly correlate with rim area (vertical cross section), average nerve width or rim length (p > 0.05). Age revealed no correlation with disc or cup areas or with cup-to-disc ratios (p > 0.05). Age correlated negatively with global and superior RNFL (p < 0.05) but did not significantly correlate with inferior, temporal or nasal RNFL (p > 0.05). Disc area correlated positively with five rim parameters, including rim area (vertical cross section), rim length, vertical integrated rim area, horizontal integrated rim width and rim area (p < 0.05) but did not significantly correlate with average nerve width (p > 0.05). Disc area correlated positively with cup area (p < 0.05) but did not correlate with cup-to-disc ratios (p > 0.05). Disc area correlated positively with global, superior, inferior and nasal RNFL (p < 0.05), but did not significantly correlate with temporal RNFL (p > 0.05). Therefore, average nerve width and temporal RNFL are independent of both age and disc area.

Table 2. Correlations (shown as p-values and coefficients) between optical coherence tomographic measurements and age and with disc area
 AgeDisc area
  1. RNFL: retinal nerve fibre layer; * Pearson correlation method; p < 0.05
Optic nerve head  
Rim area (vertical cross section)0.236 (-0.048)0.001* (0.253)
Average nerve width0.448 (-0.030)0.257 (-0.094)
Rim length0.293 (-0.063)<0.001* (0.324)
Vertical integrated rim area0.001* (-0.358)<0.001* (0.307)
Horizontal integrated rim width<0.001* (-0.464)<0.001* (0.544)
Rim area0.003* (-0.316)<0.001* (0.793)
Disc area0.182 (-0.111) 
Cup area0.486 (-0.004)<0.001* (0.631)
Cup/disc area ratio0.361 (0.044)0.233 (0.063)
Cup/disc horizontal ratio0.240 (0.087)0.128 (0.097)
Cup/disc vertical ratio0.392 (-0.034)0.296 (0.046)
RNFL  
Global RNFL0.026* (-0.353)<0.001* (0.464)
Superior RNFL0.004* (-0.371)<0.001* (0.431)
Inferior RNFL0.240 (0.132)<0.001* (0.380)
Temporal RNFL0.455 (-0.078)0.743 (0.034)
Nasal RNFL0.242 (0.121)<0.001* (0.290)

In further analyses, in which RNFL thickness was used as a dependent variable and neuroretinal rim was set as an independent variable, stepwise multiple regression analysis and Bonferroni modified test indicated that global (p < 0.001, F = 58.706), superior (p < 0.001, F = 43.245), inferior (p < 0.001, F = 30.552) and nasal (p < 0.001, F = 10.880) RNFL were significant with horizontal integrated rim width. Temporal RNFL was independent of all rim parameters (p > 0.05, F < 4.000). Deming regression analysis further indicated the correlations between RNFL (global and non-temporal quadrants) and horizontal integrated rim width, and the results are shown in Table 3.

Table 3. Deming regression analysis of correlations between RNFL thickness and neuroretinal rim
Dependent variableDeming regression analysis
  1. RNFL: retinal nerve fibre layer
Global RNFL= 20.38 + 47.86 × horizontal integrated rim width
Superior RNFL= -15.44 + 82.72 × horizontal integrated rim width
Inferior RNFL= -14.23 + 84.36 × horizontal integrated rim width
Nasal RNFL= -63.80 + 76.51 × horizontal integrated rim width

Discussion

A previous report has shown that vertical integrated rim area, horizontal integrated rim width and rim area correlate negatively with age.[12] The current analysis of six OCT rim parameters also showed that these three parameters (vertical integrated rim area, horizontal integrated rim width and rim area) correlated negatively with age; however, another three (rim area [vertical cross section], average nerve width and rim length) did not. A previous report indicated that global RNFL thickness obtained by OCT correlates negatively with age.[9] The neuroretinal rim is located between the disc and the cup margin, where retinal nerve fibres enter the ONH. Neuroretinal rim changes associated with advancing age should be associated with changes in RNFL thickness; however, some studies report that age-related changes in RNFL thickness vary by quadrants,[5-7] which is consistent with the current findings for age correlations with only three rim parameters.

Analysis of optic disc-cup structure in this study revealed no correlation with age. A previous study with refractive errors within 6.00 D found that cup area was positively correlated with age but disc area and cup-to-disc ratios were not.[5] Another report indicated that disc area does not correlate with age,[21] whereas cup area correlates positively with cup-to-disc ratio. Therefore, age should either correlate with both cup area and cup-to-disc ratio or it should correlate with neither. The disc is usually tilted in subjects with myopia and the optic disc-cup structure obtained by OCT in myopic subjects may be imperfect. Therefore, the current study included subjects with refractive error within 1.00 D and without tilted discs.

The current study of RNFL thickness revealed that, of the RNFL quadrants, only the superior quadrant correlates (negatively) with age, which is consistent with Vernon and colleagues' report.[3] Parikh and colleagues[6] and Feuer and colleagues[22] agreed that age-related RNFL loss is non-uniform and that the highest rate of thickness loss occurs in the superior quadrant. Harwerth, Wheat and Rangaswamy[23] and Harwerth and Wheat[24] further reported that RNFL loss may not be age-related; they found that non-neuronal RNFL thickness increases with age, whereas neuronal RNFL thickness decreases with age. Users of OCT usually need to allow for the aging effect on measurements. According to the current study, the associations between age and neuroretinal rim and between age and RNFL quadrants are non-uniform and the correlations between age and OCT measurements are not consistent; consequently, age should not be a constant confounder to the measurements when using OCT.

The RNFL thickness reportedly correlates negatively with distance from ONH, and the neuroretinal rim and RNFL thickness correlate positively with disc area.[8, 10-12, 25] The current study found that five rim parameters, cup area, global RNFL and the non-temporal RNFL quadrants correlated positively with disc area, which is also consistent with the literature.[8, 10-12] Further stepwise multiple regression analyses revealed that global RNFL and non-temporal RNFL quadrants correlate positively with horizontal integrated rim width; however, the current study found that temporal RNFL is independent of age, disc area and neuroretinal rim. In clinical practice, global measurements are typically used to calculate RNFL thickness. Some studies have reported that RNFL thickness measurements are non-uniform among the RNFL quadrants and are thinner in glaucomatous eyes,[19, 26] while others reported that the temporal RNFL is thicker and the non-temporal RNFL quadrants are thinner in myopic eyes.[4, 27] Accordingly, temporal RNFL is a reliable parameter for differential diagnosis between glaucoma and myopia.

A glaucomatous appearance of the ONH in pre-perimetric glaucoma could include the thinning of the neuroretinal rim and RNFL and an asymmetric cup-to-disc ratio of more than 0.2 or vertical cup-to-disc ratio of more than 0.6 with normal visual field.[28] Although the current study did not exclude subjects with cup-to-disc ratio of more than 0.6, patients with large cup-to-disc ratios and normal standard achromatic automated perimetry results may have glaucoma.[29] The possibility of enrolment of pre-perimetric glaucoma patients without neuroretinal rim loss or RNFL defect should be considered. This study analysed the effects of age and disc area on OCT measurements and the clarified correlations among OCT measurements are beneficial for clinical application when simultaneously using ONH and RNFL scan protocols.

In conclusion, OCT measurements indicate that age-related neuroretinal rim loss or RNFL thickness change is non-uniform and inconsistent, and age is not a constant confounder when using OCT; however, temporal RNFL is independent of age, disc area and neuroretinal rim.

Acknowledgement

The author is grateful to Paul Steve Lugue for assistance with reviewing and editing this manuscript.

Ancillary