Diagnostic Capability of Optical Coherence Tomography in Evaluating the Degree of Glaucomatous Retinal Nerve Fiber Damage

Abstract

purpose. To evaluate the role and ability of optical coherence tomography (OCT) to detect differences in peripapillary retinal nerve fiber layer (RNFL) thickness between normal and glaucomatous eyes and also between different severities of glaucoma.

method. This cross-sectional observational study included 160 eyes of 160 healthy subjects and 134 eyes of 134 patients with primary open-angle glaucoma (POAG). Peripapillary RNFL thickness was measured on OCT using the fast RNFL thickness protocol. The RNFL thickness parameters used for evaluation included average RNFL thickness and inferior, superior, nasal, and temporal RNFL thickness. The glaucomatous eyes were subdivided into three subgroups on the basis of visual field defects and a fourth subgroup of eyes blinded by glaucoma. RNFL thickness parameters were compared among the normal eyes and the glaucoma subgroups. Correlation of global visual field indices with RNFL thickness parameters was also performed.

results. The average RNFL in control subjects, early glaucoma, moderate glaucoma, severe glaucoma, and blind glaucoma were 102.30 ± 10.34, 77.68 ± 15.7, 66.07 ± 15.5, 53.65 ± 14.2, and 44.93 ± 4.95 μm, respectively. There was a significant difference in all RNFL thickness parameters between normal and all glaucoma subgroups (P < 0.001). Average and inferior RNFL thicknesses showed the highest area under the receiver operating characteristic curve, with 0.905 and 0.862 for normal versus early glaucoma, 0.705 and 0.722 for early versus moderate glaucoma, 0.737 and 0.717 for moderate versus severe glaucoma, and 0.635 and 0.584 for severe versus blind glaucoma. Both mean deviation (MD) and corrected pattern standard deviation (CPSD) showed a significant correlation with all the RNFL thickness parameters in eyes with glaucoma (P < 0.001).

conclusions. RNFL thickness measured on OCT may serve as useful adjuncts in accurately and more objectively distinguishing normal from glaucomatous eyes, even in the early stages of glaucoma and may help to differentiate various severities of glaucoma. Average and inferior RNFL thicknesses are among the most efficient parameters for distinguishing such a differentiation. RNFL thicknesses in eyes blinded by glaucoma provide an estimate of the component of the RNFL thickness, which is not related to visual function.

Establishing the severity of glaucoma has largely been based on perimetric parameters, ¹ but there is a need for an objective method for assessment of the glaucomatous damage, especially when performing and following up standard automated perimetry may not be a feasible option as in the late stage of the disease. It has been shown that damage to the retinal nerve fiber layer (RNFL) precedes visual field loss. ² Quigley et al. ³ reported that up to 40% to 50% of the RNFL could be lost before visual field defects are detected by conventional perimetry. Thus, RNFL assessment has emerged as an important parameter for preperimetric diagnosis of glaucoma and may aid ophthalmologists in making an accurate and early diagnosis. Various imaging technologies have been introduced for an objective assessment of RNFL. ^{4 5 6 7}  

Optical coherence tomography (OCT) is a noncontact, noninvasive diagnostic technique that allows measurement of RNFL thickness by in vivo visualization of the retina and RNFL with good reproducibility. ^{7 8 9 10 11} OCT-determined RNFL thickness has been shown to be significantly reduced in patients with glaucoma, although with interindividual variability. ^{10 11 12 13 14}  

To the best of our knowledge, there are no reports of RNFL thickness in the different grades of glaucoma. We conducted a comparative study to evaluate RNFL thickness in normal eyes, eyes with differing grades of POAG, and eyes blinded by glaucoma. 

Materials and Methods

One hundred sixty healthy volunteers and 134 patients with POAG were included in this cross-sectional study. The study was approved by our institutional review board and complied with the tenets of Declaration of Helsinki. Informed consent was received. All subjects underwent applanation tonometry, gonioscopy, and fundus examination with a plus 90-D lens. Automated refraction (Retinomax 2 Autorefractor; Nikon Corp., Yokyo, Japan), axial length measurement (EchoScan US 3300; Nidek Corp., Gamagori, Japan), and automated visual field examination (model 745 Humphrey visual field analyzer [HFA], full-threshold program 30-2; Carl Zeiss Meditec, Inc., Dublin, CA) were performed. 

Healthy subjects included persons with a refractive error of less than 4 D, attendants of patients who were not blood relatives, and those without any evidence of glaucoma or family history of glaucoma. Normal contralateral eyes of patients with cataracts or unilateral macular holes were also included. Subjects with intraocular pressure of ≥21 mmHg, vertical asymmetry of cup-to-disc ratio (>0.2) between the two eyes, high cup-to-disc ratio (>0.6), disc hemorrhages, disc pallor, and localized RNFL defects were excluded. Mean deviation (MD) and corrected pattern standard deviation (CPSD) of all the eyes on HFA had to be within the 95% confidence interval, with normal results in the glaucoma hemifield test. 

Glaucomatous eyes were those that exhibited an IOP of >22 mm Hg on more than two occasions and glaucomatous optic neuropathy with glaucomatous visual field loss consistent with optic nerve damage. Glaucomatous visual field loss was taken as the consistent presence of a cluster of three or more nonedge points on the pattern deviation plot with a probability of occurring in fewer than 5% of the normal population (P < 5%), with one of these points having the probability of occurring in <1% of the normal population (P < 1%), CPSD with P < 5%; or a glaucoma hemifield test result outside normal limits. Only patients who had more than two reliable consistent visual field results were included. 

Patients with any intraocular disease, intraocular surgery, ocular trauma, and secondary glaucoma were excluded. The eyes with consistently unreliable visual fields (defined as false-negative >33%, false-positive >33%, and fixation losses >20%) were excluded from the study. Patients with a possible neurologic field loss were also excluded. 

The categorization into three subgroups of glaucoma was performed by Hodapp’s classification based on the MD index of visual fields. ¹ Early glaucoma was defined by a visual field loss with an MD ≤ −6 dB, moderate glaucoma with an MD between −6 dB and −12 dB, and severe glaucomatous loss with an MD worse than −12 dB. Eyes that were blind and had no light perception due to glaucoma were included in the last subgroup of blind glaucomatous eyes. 

OCT measurements were performed on the Stratus OCT 3 (software version 4; Carl Zeiss Meditec, Inc.). The basic principle and technical characteristics of this OCT system have been described previously. ^{7 8} Each eye was dilated with tropicamide 1% before the images were recorded, and scans were performed with a minimum pupillary diameter of 5 mm. The internal fixation target was used owing to its higher reproducibility. The fast retinal nerve fiber layer thickness protocol was used. This protocol provides better reproducibility than a single scan. ^{10 11} It consists of three circular scans, each 3.46 mm in diameter, centered on the optic disc. This diameter has been shown to be optimal and reproducible for RNFL thickness analysis. ^{10 11} One author (PS) performed all the image acquisitions. 

In end-stage absolute glaucomatous eyes with no light perception, an external fixation target was used for the fellow eye. 

Mean RNFL thickness was calculated with the inbuilt RNFL thickness average analysis protocol. Retinal thickness was measured with the location of the vitreoretinal interface and the retinal pigment epithelium defining the inner and outer boundaries of the retina, respectively. These landmarks are seen as sharp edges with high reflectivity. The boundaries of the RNFL were defined by first determining the thickness of the neurosensory retina. The location of the posterior boundary of the RNFL was determined by evaluating each A-scan for a threshold chosen to be 15 dB greater than the filtered maximum reflectivity of the adjacent retina. Various machine-generated parameters were used for evaluation of RNFL thickness, including RNFL average thickness over the entire cylindrical section and average RNFL thickness in each quadrant (superior, nasal, temporal, and inferior). 

The analysis was performed using commercial statistical software package (SAS Institute, Inc., Cary, NC). One eye of each patient was randomly selected for analysis. One-way ANOVA was used to compare the different parameters among normal eyes and different glaucoma subgroups with a post hoc test (Bonferroni), when applicable. The associations of MD, CPSD, and RNFL parameters were evaluated by Pearson’s correlation coefficient. P <0.05 was considered statistically significant. 

Results

One hundred sixty eyes of 160 normal subjects and 134 eyes of 134 patients with glaucoma were included in the study. Their ages ranged from 20 to 70 years, with 67 women and 93 men. Among the patients with glaucoma, 61 eyes had early, 31 eyes had moderate, and 25 eyes had severe glaucomatous visual field defects. The remaining 17 eyes had end-stage glaucoma with no light perception. 

The RNFL parameters computed in the control and glaucoma groups are presented in Table 1 . The average RNFL in control eyes and in eyes with early, moderate, severe, or blind glaucoma were 102.30 ± 10.34, 77.68 ± 15.7, 66.07 ± 15.5, 53.65 ± 14.2, and 44.93 ± 4.95 μm, respectively. All RNFL parameters were significantly different between control eyes and eyes in the various glaucoma groups, on post hoc (Bonferroni) analysis (P < 0.001). On evaluating the percentage loss of the RNFL between normal eyes and those having an early glaucomatous visual field defect, we observed a 24.9% decrease in the mean RNFL thickness. The percentage decrease in the mean average RNFL thickness among other glaucoma groups is given in Table 2 . The absolute eyes showed a 56.79% decrease in the mean average RNFL thickness when compared with the control group. Seventeen eyes that were blind due to glaucoma showed an obvious decrease in reflectivity in the zone corresponding to the RNFL on OCT. 

The correlation between the global indices on perimetry, MD, and CPSD, with various RNFL parameters in the all glaucomatous eyes was evaluated using Pearson’s correlation coefficient (Table 3) . The MD on Humphrey field analysis was found to have a significant positive correlation with average, superior, inferior, and temporal RNFL thickness. CPSD showed a significant negative correlation with average, superior, inferior, temporal, and nasal RNFL thickness (P < 0.001). 

Receiver operator characteristic (ROC) curves were drawn for RNFL parameters (global and the four quadrants; Table 4 ; Fig. 1 ). The area under the ROC (AROC) curve (160 normal vs. 61 eyes with early glaucoma) for average RNFL was 0.905. With a cutoff value of 91.41 μm, there was 89.4% sensitivity and 80.3% specificity. For inferior RNFL thickness, the AROC curve was 0.862, and a cutoff of 118.50 μm gave 77.2% sensitivity and 77% specificity. The AROC curve for superior RNFL was 0.856 and a cutoff of 112.5 μm gave sensitivity of 85.6% and specificity of 80.3%. The AROC curve for nasal and temporal RNFL was 0.808 and 0.704. The AROC curves for differentiating early from moderate glaucoma were 0.705 and 0.722; moderate from severe, 0.737 and 0.717; and severe from blind, 0.635 and 0.684, for average and inferior RNFL, respectively. 

The box plots for average RNFL thickness (in micrometers) showed a considerable overlap among the normal subjects and various glaucoma subgroups (Fig. 2) . 

Discussion

An attempt has been made to grade glaucoma on the basis of visual field changes detected on standard white-on-white perimetry. ¹ However, evaluation of peripapillary RNFL and optic disc changes constitutes an important part of glaucoma diagnosis. OCT offers a good and objective method of RNFL evaluation, with high reproducibility. ^{8 9 10 11} OCT-derived RNFL thickness, though different from actual histologic measurement, has been shown to possess good correlation with histologic measurement. ¹² OCT measurements have been shown to correlate well with RNFL thickness, as measured by scanning laser polarimetry (SLP, GDx-VCC). In addition, OCT measurements are uninfluenced by refractive error and corneal birefringence. ^{13 14 15 16 17}  

We observed a significant difference in the RNFL thickness parameters among normal eyes and all grades of glaucoma. Previous studies have shown that Stratus OCT-generated RNFL thickness is reliable for differentiating eyes with early glaucoma from normal eyes. ^{13 14 15 16 17 18 19 20 21} However, due to strict definitions of normal individuals and patients with glaucoma, as reflected by the inclusion and exclusion criteria application, results in day-to-day clinical practice may not be accurate in all individuals. Chen et al. ¹⁹ showed that average RNFL was the best parameter for differentiating early glaucoma from normal eyes with ROC curve area of 0.793, Kanamori et al. ¹³ showed an AROC of 0.93 with inferior RNFL as the best parameter for such a differentiation. We found that the average RNFL thickness followed by the inferior RNFL thickness had the highest power to discriminate between early glaucoma and normal eyes, with an area under the ROC of 0.905 and 0.862, respectively. 

In addition to evaluation of the discriminating power of OCT between various grades of glaucomatous visual field defects, we again found that inferior and average RNFL thickness produced the highest AROC curves. The AROC curves for early versus moderate glaucoma eyes were 0.722 and 0.705; for moderate versus severe, 0.717 and 0.737; and for severe versus blind end-stage glaucoma, 0.684 and 0.635 for inferior and average RNFL thickness, respectively. Thus, based on inferior and average RNFL thickness, assessment of various grades of glaucomatous visual field defects may be achieved. In a recent study comparing RNFL thickness and visual field loss between various glaucoma groups with scanning laser polarimetry (GDx Nerve Fiber Analyzer; Laser Diagnostics, San Diego, CA), there was a significant difference between normal eyes and eyes with early glaucoma, but the difference was not significant when eyes with early glaucoma were compared with those with moderate glaucoma. ²²  

On studying the correlation between the visual field indices (MD and CPSD) and the average RNFL thickness, we detected a significant positive correlation with MD and a significant negative correlation with CPSD on standard white-on-white perimetry. This observation is in agreement with previous studies. ^{23 24 25} In a study of 30 patients with glaucoma and 14 control subjects, Parisi et al. ²⁵ showed a highly significant linear regression (r = 0.663, P < 0.001) between overall NFL thickness and CPSD, the correlation with MD having been less significant (r = 0.393, P = 0.031). ²⁵ The difference in the significance of correlation compared with the present study could be due to the difference in sample size. Significant correlation of quadrantic or sectoral RNFL thicknesses with global indices in eyes with well-defined sectoral visual field defects may provide useful information. However, determining this correlation was not feasible in the present study, because of the limited number of patients with clear-cut sectoral defects. 

A healthy nerve fiber layer comprises ganglion cell axons, neuroglia, and the astrocytes. ²⁶ The glaucomatous process results in a loss of retinal ganglion cells. ^{2 26} In our present study, we observed a 25% decrease in eyes with early glaucoma. Quigley et al. ³ showed that up to 40% to 50% of optic nerve fibers could be lost in the absence of a visual field defect. A later study showed that a reduction of at least 25% to 35% in the retinal ganglion cell population is associated with statistical abnormalities in automated perimetry. ²⁷  

It was interesting that no-light-perception eyes blinded by glaucoma continued to show an RNFL thickness that was 43% of the normal average RNFL thickness. Findings in previous studies suggest that the glaucomatous process results in loss of RNFL, but there is no concurrent loss of glial tissue. ²⁶ Therefore, it appears that this 43% of residual RNFL thickness could be a measure of the supporting glial tissue. However, histologic data to support this finding are presently unavailable in the literature. As a corollary, the lost 57% could be considered the “functional” RNFL. Subtracting the nonfunctional RNFL thickness from all the subgroups of glaucoma leaves a residual “functional” RNFL of 32%, 21%, and 9% in eyes with early glaucoma, moderate glaucoma, or severe glaucoma, respectively. Thus, a 25% decrease in the average RNFL thickness from normal to early glaucoma hypothetically equates to an approximately 40% to 50% decrease in “functional” RNFL. And thereafter an approximate 10% decrease in total RNFL from early to moderate and moderate to severe could be equated to a further 20% to 25% decrease of in “functional” RNFL at each stage. 

Although the RNFL parameters among the various glaucoma subgroups were significantly different, we observed a significant overlap in the RNFL thicknesses among normal subjects and the various glaucoma subgroups. In addition, the RNFL thickness showed a racial variation. Thus, despite the objective precision, the measurements by OCT do not directly translate to diagnostic precision and application of OCT in assessing the exact degree of glaucomatous damage in an individual appears to be limited. Our study has certain shortcomings, being a cross-sectional study with a smaller sample size in the advanced and end-stage glaucoma group with blind eyes. 

Our study suggests that certain OCT-generated RNFL thickness measurements such as average and inferior RNFL thickness may provide valuable information regarding the degree of glaucomatous damage. In an individual, these measurements can complement the results of perimetry and clinical observation in providing more objective and quantitative information regarding the degree of glaucomatous damage. Our findings may also help to develop an algorithm for quantifying the “functional” RNFL thickness reserve in patients with glaucoma. However, further larger studies with a longitudinal follow-up are needed, to substantiate these findings. 

 Submitted for publication August 20, 2005; revised November 27, 2005; accepted March 14, 2006.

 Disclosure: R. Sihota, None; P. Sony, None; V. Gupta, None; T. Dada, None; R. Singh, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Peripapillary RNFL Thickness Measured on OCT in Normal Eyes and the Various Glaucoma Subgroups

RNFL Thickness*	Normal (n = 160)	Early Glaucoma (n = 61)	Moderate Glaucoma (n = 31)	Severe Glaucoma (n = 25)	Blind Glaucoma Eyes (n = 17)	P (ANOVA)
Average	102.3 ± 10.34	77.68 ± 15.7	66.07 ± 15.5	53.65 ± 14.2	44.93 ± 4.95	<0.001
Inferior	129.87 ± 17.02	95.90 ± 27.57	74.61 ± 24.35	58.84 ± 28.37	58.76 ± 20.22	<0.001
Superior	128.5 ± 17.02	96.1 ± 24.11	79.26 ± 28.36	62.00 ± 21.25	60.47 ± 13.72	<0.001
Nasal	83.72 ± 18.88	61.48 ± 17.54	55.19 ± 15.86	47.68 ± 14.76	50.29 ± 12.94	<0.001
Temporal	66.57 ± 12.73	57.15 ± 13.47	53.90 ± 14.02	43.56 ± 13.04	43.65 ± 8.72	<0.001

 Post hoc test (Bonferroni) was significant for all parameters in all glaucoma groups. Thickness data are in micrometers (mean ± SD).

Table 2.

 

Percentage RNFL Thickness Loss in Various Subgroups of Glaucoma Compared with Normal Eyes on OCT

Group	Average RNFL Thickness (μm)	RNFL Thickness (% of Normal Eyes)
Normal eyes	102.3 ± 10.34	100.0
Early glaucoma	77.68 ± 15.7	75.9
Moderate glaucoma	66.07 ± 15.5	64.58
Severe glaucoma	53.65 ± 14.2	52.44
Blind glaucoma*	44.93 ± 4.95	43.4

*  No light perception, and complete visual loss due to glaucoma.

Table 3.

 

Pearson’s Correlation Coefficient for MD and CPSD on Humphrey Visual Field Analysis, Full-Threshold Program 30-2) with RNFL Parameters on OCT

RNFL Thickness	MD					CPSD
		R	R 2	P	R	R 2	P
Average	0.626	0.391	<0.001	−0.625	0.390	<0.001
Inferior	0.609	0.370	<0.001	−0.648	0.419	<0.001
Superior	0.54	0.291	<0.001	−0.538	0.289	<0.001
Nasal	0.349	0.121	0.001	−0.417	0.173	<0.001
Temporal	0.449	0.201	<0.001	−0.411	0.168	<0.001

Table 4.

 

AROC Curve for Normal Eyes and the Glaucoma Subgroups

RNFL Thickness	Normal vs. Early Glaucoma			Early vs. Moderate Glaucoma			Moderate vs. Severe Glaucoma			Severe vs. Blind Glaucoma
		AROC	SE	AROC	SE	AROC	SE	AROC	SE
Average	0.905	0.05	0.705	0.57	0.737	0.070	0.635	0.089
Inferior	0.862	0.031	0.722	0.55	0.717	0.074	0.684	0.088
Superior	0.856	0.031	0.669	0.63	0.710	0.072	0.520	0.090
Nasal	0.808	0.032	0.590	0.62	0.672	0.070	0.520	0.089
Temporal	0.704	0.040	0.570	0.65	0.707	0.075	0.311	0.082

Figure 1.

 

ROC curve for RNFL thickness parameters among various subgroups.

Figure 2.

 

Distribution of average RNFL thickness among normal eyes and the four glaucoma subgroups: 1, normal eyes; 2, early glaucoma; 3, moderate glaucoma; 4, severe glaucoma; 5, blind glaucoma eyes.