Article Text

Download PDFPDF

Clinical science
Correlation between optical coherence tomography and glaucomatous optic nerve head damage in children
  1. M A El-Dairi,
  2. S Holgado,
  3. S G Asrani,
  4. L B Enyedi,
  5. S F Freedman
  1. Duke University Eye Center, Durham, North Carolina, USA
  1. Correspondence to Dr, S F Freedman, Duke University Eye Center, 2308 Erwin Rd, Durham, NC 27710, PO Box: 3802, USA; freed003{at}


Aim: To compare analysis of macular and nerve fibre layer thickness by optical coherence tomography (OCT) with optic nerve head (ONH) morphology based on stereophotography.

Design: Prospective observational case–control series.

Methods: Normal and glaucomatous eyes of children (age 4–17 years) were scanned using Stratus OCT (Carl Zeiss Meditec, Dublin, California, USA). Fast macular and retinal nerve fibre layer (RNFL) thickness map were performed on 372 eyes of 222 children. ONH stereophotographs were taken and evaluated by two masked observers using a grading system of 0 to 5 based on both cupping ratio and morphology. OCT3 analyses were compared across ONH grades for different areas around the macula and the peripapillary RNFL.

Results: Analysis included OCT values and ONH grading for 139 eyes of 139 children. There was a negative correlation between ONH grade and both macular thickness and RNFL thickness in all areas measured. There was a difference in the correlation identified for black versus white children.

Conclusion: OCT measurements of RNFL and macular thickness declined with increasing grade of glaucomatous damage seen on stereophotographs in black and white children. Further study will help quantify the value of OCT in the diagnosis and management of paediatric glaucoma.

Statistics from

In children, objective evaluation of glaucomatous damage to the optic nerve and visual fields is challenging. Reliable visual fields are difficult to obtain in many children with glaucoma, enhancing the importance of optic nerve head (ONH) evaluation. Optical coherence tomography (OCT) measurements of the peripapillary and macular nerve fibre layers have been correlated with the severity of glaucoma in adults.12

OCT, which is relatively easy to perform in cooperative children with clear media, has been used in eyes of normal children,3456 but published reports of the use of OCT in paediatric glaucoma are sparse.78 Comparing OCT analysis of the peripapillary retinal nerve fibre layer (RNFL) and macula in children, Hess et al found a significant difference between normal and glaucomatous eyes of children.7

Attempts have been made to correlate glaucomatous ONH damage (using a grading system on stereophotographs) and glaucomatous visual field loss in adults.9 We chose to evaluate a possible correlation between glaucomatous ONH damage as seen on ONH stereophotographs and macular and RNFL analysis by OCT in children.

Subjects and methods

Children (n = 222) aged 4–17 years were enrolled in this prospective observational case series, conducted from September 2003 to August 2006. The children who were recruited had known or suspected glaucoma, and were presenting for routine ophthalmic evaluation.

Data gathered on each subject included age, sex, self-reported race, family history of ocular disease, and ocular and medical past history. All subjects had a full eye examination including measurements of cycloplegic refraction and intraocular pressure (IOP), OCT scanning of the macula and peripapillary region, and ONH stereophotographs. Children with known paediatric glaucoma had IOP >22 mmHg at diagnosis, together with clinical evidence of glaucoma such as enlarged corneal diameter, Haab’s striae, glaucomatous ONH appearance and glaucomatous visual fields (when available). Eyes with enlarged optic nerve cups but with normal IOP (⩽21 mmHg) and no other clinical evidence of glaucoma were considered as glaucoma suspects (by virtue of optic nerve cupping), as were those eyes with IOP >21 mmHg but with normal optic nerve appearance and no other evidence of glaucoma.

Both the OCT and photographs were performed after pupil dilation at the end of the examination. Subjects with prematurity, systemic disorders, Sturge–Weber syndrome, aphakia or pseudophakia, any media opacities or any non-glaucomatous ocular pathology other than strabismus were excluded. All subjects had a spherical equivalent <6 D and astigmatism <3 D.

OCT scanning

Study subjects were scanned using Stratus OCT (Carl Zeiss Meditec, Dublin, California, USA), for both “fast macular thickness map” (fig 1a) and “fast RNFL map” (fig 2a) protocols using an internal fixation target. The scans were performed as described by Hess et al.7 Only scans with good quality (signal strength better than 5, without misalignment or movement artefacts) were included.

Figure 1

Fast macular thickness map protocol. (A) Fast macular map protocol, right eye. This is a sketch of the optical coherence tomography (OCT) scan area and labelling for the right eye, using the fast macular thickness map. This map is formed from scanning with six equally spaced intersecting radial scans through the centre of the fovea. Each radial scan has a length of 6.0 mm. The results were reported as average thickness values within each quadrant (superior, temporal, inferior and nasal) in each of two concentric circles outside the foveal central circle of 1 mm diameter. The outer circle diameters were 3 mm (inner ring) and 6 mm (outer ring). (B) Macular volume plotted against optic nerve head grade. (C) Average macular thickness for both the outer and inner rings (y-axis), plotted against optic nerve head grade (x-axis).

Figure 2

Fast retinal nerve fibre layer map protocol. (A) Fast retinal nerve fibre layer (RNFL) map protocol, right eye. This is a sketch of the optical coherence tomography (OCT) scan area and labelling for the right eye, using the fast RNFL map protocol. In this protocol, the nerve fibre layer thickness is measured with the OCT in the peripapillary region using six concentric circular scans. The smallest circle has a diameter of 2.9 mm, while the largest circle has a diameter of 6.8 mm. We averaged all values across each quadrant as well as across the inner and outer rings. (B) Average RNFL values for both the outer and inner rings (y-axis) plotted against the optic nerve head (OHN) grade (x-axis).

ONH stereophotographs and grading system

ONH stereophotographs were obtained using the 3Dx Simultaneous Stereo Retinal Camera (NIDEK Co., Ltd, Gamagori, Aichi, Japan) or the Zeiss Fundus Camera FF450 Plus (Carl Zeiss Meditec, Jena, Germany). Photographs were evaluated by two masked observers (both fellowship-level trained in glaucoma), using a grading system of 0–5 based on both cupping size and morphology as described by Jonas et al10 (see fig 3).

Figure 3

Diagram (left panel) and photographic examples (right panel) of optic nerve head grading scale used for comparison with optical coherence tomography evaluation of both the macula and peripapillary retinal nerve fibre layer. Descriptions of the grades of optic nerve head cupping are as follows: grade 0, optic nerve considered normal, with cup-to-disc ratio <0.6; grade 1, increased cup-to-disc ratio >0.6 but ⩽0.8, with intact rim; grade 2, presence of rim sloping or notching in the infero- or superotemporal quadrant, with or without vessel exposure (white arrowhead); grade 3 presence of rim sloping or notching in two quadrants; grade 4, extensive narrowing of the temporal rim; grade 5, total optic nerve cupping.

Statistical analysis

Data in this study were analysed using JMP 6.0 (SAS Institute, Cary, North Carolina, USA). A Shapiro–Wilk test for normality was performed for each variable in each ONH grade. A generalised linear model (GLM) was used to determine the effect of age, race, refractive error and glaucoma grade on each OCT variable. The average ONH stereophotographic grade given by the two masked readers was used for analysis after the kappa value was calculated to assess the agreement between the two readings.

Linear regression analysis and ANOVA were used to study the variation of the RNFL and macular thicknesses in quadrant areas in relation to the ONH grade. p Values were considered statistically significant at <0.05.

An unpaired t test was used to compare the mean OCT values of black versus white subjects across each ONH grade. Regression analysis and ANOVA were done for all subjects, then for white and black subjects independently. Bonferroni adjustment of p values was used for multiple comparisons.

An unpaired t test was used to compare means of OCT values among ONH grades 0 and 1 for the entire population then for black white subjects independently. Similarly, an equivalence test was also performed for the entire population then for black and white subjects independently. The practical difference threshold (chosen from published OCT reproducibility data11) was specified at 2 µm for macular thickness in each quadrant, 0.1 mm3 for macular volume and 3 µm for RNFL thickness in each quadrant.

Normal data were included for the right eye from subjects with a bilaterally normal examination and from the normal eye of those with uniocular non-glaucomatous pathology. For subjects with glaucoma, data were included from the eye with worse glaucoma if bilateral disease was present, and from the right eye if the severity of glaucoma was the same in both eyes.


Good quality scans were obtained on 372 eyes of 206 subjects. Ninety-one eyes of 49 subjects in the present study were included in the previously published work by Hess et al as either normal subjects or eyes with known glaucoma.7 Of the 372 OCT scans obtained, 136 scans of 53 subjects were excluded for the ocular reasons mentioned above, and 30 OCT scans of 16 subjects were excluded on second review for quality. While 264 eyes (139 subjects) had usable OCT scans of the macula, 237 eyes (124 subjects) had scans of both the peripapillary RNFL and the macula. For each ONH grade, the OCT values tested were normally distributed (data not shown).

Subject demographics

The mean age of all subjects was 10.0 (SD 3.1) (range 2–17) years, and did not differ significantly among ONH grade groups. Girls (n = 68, 49%) and boys (n = 71, 51%) were equally represented in this study and were of similar mean age (9.5 (SD 3.0) vs. 10.4 (SD 3.3) years, respectively). Racial composition of subjects included white (70 subjects, 50%), black (40 subjects, 29%) and other (19 subjects, 13.5%, of mixed race, Hispanic, Asian, middle Eastern, or Indian origin) (see table 1).

Table 1

Average macular thickness measured by optical coherene tomography (fast macular thickness map protocol) in paediatric subjects, grouped by optic nerve head (grades 0 to 4) and race

Thirty-five (24%) subjects were diagnosed with glaucoma—primary infantile glaucoma in 12, juvenile open angle glaucoma in 22, and congenital-onset, unilateral neurofibromatosis-related glaucoma in the remaining child. Fifty-eight subjects (43%) were given the diagnosis of glaucoma suspects—49 by virtue of increased ONH cup with normal IOP and nine due to elevated IOP with a non-glaucomatous ONH appearance. The remaining 46 subjects (33%) were considered normal.

There were no eyes classified as ONH grade 5 because of the subjects’ inability to fix at the OCT internal fixation target (data not shown, 14 eyes of 12 subjects). Agreement between the two observer readings of the ONH stereophotographs was 87%.

Effects of refractive error and age

Only race and ONH grade had a statistically significant effect on OCT variables studied. The effects of age, sex and spherical equivalent on the OCT values tested were not statistically significant (data not shown).

Fast macular thickness map protocol

OCT measurement of macular thickness in each quadrant (fig 1A) correlated negatively with increasing ONH grade (slope (µm/grade) = −14 to −16, p<0.0001 for all). Similarly, macular volume decreased by about 0.22 mm3 with every one grade of increase of ONH grade (p<0.0001, fig 1B). These results did not change when subanalysis was performed separately for either white or black subjects. OCT measurement of the average outer macular ring thickness decreased more with increasing ONH grade than did the average inner macular ring thickness (slope (µm/grade ) = −34 vs. −16; p<0.0001) (fig 1C) (see also table 1).

Comparing macular thickness values measured by OCT among grades 0 and 1 did not reveal a statistical difference between the two groups. For white subjects, there was a trend towards reduced macular thickness in each quadrant; however, that difference was not statistically significant. By contrast, among black subjects, there was statistical equivalence between grades 0 and 1.

Fast RNFL map protocol

We compared the values of the fast RNFL protocol in each quadrant (superior, temporal, inferior and nasal) across the values of the ONH grade. The RNFL thickness showed a significant decrease with increasing ONH grade in three quadrants (superior, inferior and nasal; p<0.001). The slope of decrease in RNFL with increasing grade of ONH was steepest for the superior quadrant (slope (µm/grade) = −36). Repeating the analysis for white subjects only yielded similar results. In black subjects, however, only the superior and inferior quadrants showed a significant decrease with increasing ONH grade, with p<0.0001. In addition, in these two quadrants, the slope of RNFL decrease with increasing ONH grade was steeper in black than white subjects (slope (µm/grade) for black vs. white: for superior RNFL = −52 vs. −30, for inferior RNFL = −49 vs. −26).

Similar analysis performed for the inner and the outer rings of the RNFL map protocol showed a statistically significant decrease with increasing ONH grade (fig 2b). The average RNFL thickness of the inner ring had a steeper negative slope than that of the outer ring (slope (µm/grade) = −62 and −33 respectively, p<0.0001 for both) (fig 2b). This result was similar when analysis was done for white subjects, and was reversed for black subjects (table 2).

Table 2

Average peripapillary nerve fibre layer thickness measured by optical coherence tomography (fast RNFL map protocol) in paediatric subjects, grouped by optic nerve head (grades 0 to 4) and by race

Comparing RNFL thickness in each quadrant measured by OCT among ONH grades 0 and 1 did not reveal a statistical difference between the two groups. For white subjects, there was a trend towards thinner RNFL in each quadrant; however, that difference was not statistically significant. In contrast, among black subjects, there was either a trend towards thicker RNFL among subjects with ONH grade 1 compared with grade 0 or there was statistical equivalence between the two grades.


In the absence of reliable visual field data on many children with suspected or known glaucoma, ONH cup evaluation is essential for determining the presence and severity of glaucoma. Using the ONH grading scale that was adopted for use in the present study, Jonas et al showed that, in adults, damage from glaucoma correlated with visual field loss in an exponential manner, with the damage detectable on the ONH structure before the visual field loss.12 Without an accepted gold standard scale to grade glaucomatous ONH cupping in both adults and children, the Jonas scale provided a reasonable objective starting point. In this study in children, both the peripapillary RNFL and the macular thickness and volume showed a negative correlation with increasing severity of cupping. Thus, OCT measures may be an additional objective parameter in assessing children with cupping who are considered to be glaucoma suspects or have glaucoma.

RNFL thinning due to glaucoma has been shown to be detectable by OCT in adults.2 Many studies have used OCT to discriminate between adults who were normal or glaucoma suspects, and those with clinically diagnosed glaucoma.1131415161718192021 While most publications that compare OCT among adults with glaucoma use the fast RNFL 3.4 scan, the children in our study were scanned with the RNFL map protocol to evaluate their peripapillary RNFL. That is because during the data acquisition period of the present study, the fast RNFL map was performed according to a standard protocol rather than the RNFL thickness along the 3.4 mm circle. The fast RNFL map permitted distribution and measurement of RNFL thickness in the peripapillary area along six concentric rings for a distance of 6 mm from the edge of the optic nerve.

The fast macular map protocol has been used in adults with glaucoma with a moderate sensitivity and specificity for the detection of glaucoma2223 The correlation between OCT measurement of macular thickness and volume, and ONH stereophotographic grade in children is encouraging, since OCT macular scans of good quality are easier to perform in children than are the peripapillary RNFL protocols.

Another interesting finding from this study was the difference in the macular and RNFL thickness among black and white subjects within the same ONH grade, ie similar ONH shape and cup-to-disc ratio as that determined by two glaucoma-trained ophthalmologists. White children with large cup-to-disc ratios and intact optic nerve rim (ONH grade 1) had a trend towards having a reduced RNFL and macular thicknesses compared with those with smaller ONH cups (ONH grade 0). By contrast, black children with a large cup-to-disc ratio and no rim notching (ONH grade 1) had similar macular and RNFL thickness to those with smaller ONH cups (ONH grade 0). Furthermore, the rate of decline of the superior and inferior RNFL thicknesses with increasing glaucoma severity (reflected by increasing ONH grade) was higher in black than white children.

In the study by Hess et al mentioned above, significant differences were found in the OCT scans obtained from glaucoma patients compared with those of normal children, using both the RNFL map protocol and the fast macular map scan.7 No attempt was made in this early report to correlate OCT findings with severity of glaucoma or ONH damage. By contrast, the present study identified a negative correlation between macular thickness and peripapillary RNFL thickness by OCT measurement, and ONH glaucomatous damage as assessed by stereophotographic grade.

We are unaware of any prior publications on this subject in the medical literature (literature search protocol: Pubmed electronic search: Mesh terms included “pediatric”, “glaucoma” and “optical coherence tomography” from 1950 until September 2008).

The present study has several limitations. We did not calculate the sensitivity and specificity of OCT measurements for the detection of early glaucoma, because we classified eyes by ONH morphology (grade) rather than by clinical diagnosis. Categorisation of eyes on the basis of their clinical diagnosis as normal, glaucoma suspect, and early and advanced glaucoma might yield data for sensitivity and specificity calculations. We analysed OCT scans performed on each eye at one visit only. It would be interesting to know how the OCT scan values change with time and axial length growth in normal children, both black and white, as well as those with glaucoma.

In an attempt to avoid OCT changes unrelated to ONH damage from glaucoma, we intentionally excluded eyes with high refractive errors24 and those who were aphakic or pseudophakic; we therefore necessarily excluded several paediatric glaucoma patients with known ONH glaucomatous damage from the present study. In addition, eyes with severe glaucomatous ONH damage (grade 5) were also excluded due to poor fixation or nystagmus, which limits the OCT scan quality.

In conclusion, OCT is an easily acquired, non-invasive measurement in children. OCT scans of the peripapillary RNFL and macular thickness correlate with glaucomatous ONH cupping in children. Interesting differences in those correlations between black and white children have been identified. OCT measures may prove valuable in the evaluation and management of children with known or suspected glaucoma.



  • Funding Research to Prevent Blindness unrestricted departmental grant.

  • Competing interests None declared.

  • Ethics approval The study was compliant with the Health Insurance Portability and Accountability Act (HIPAA). The study was approved by the Duke University Medical Center’s Institutional Review Board.

  • Patient consent Obtained.

Request Permissions

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.