Ranking of Optic Disc Variables for Detection of Glaucomatous Optic Nerve Damage

Abstract

purpose. To describe optic disc variables assessed by evaluation of clinical optic disc photographs and to compare sensitivity and specificity of these optic disc parameters in identifying patients with ocular hypertension who have nerve fiber layer defects and normal visual fields and patients with visual field defects.

methods. The study included 500 normal subjects, 132 patients with ocular hypertension with retinal nerve fiber layer defects and normal visual fields (preperimetric glaucoma), and 840 patients with glaucomatous visual field defects. Color stereo optic disc photographs were morphometrically evaluated.

results. Highest diagnostic power for the separation between the normal group and the preperimetric glaucoma group had the vertical cup-to-disc diameter ratio corrected for its dependence on the optic disc size, total neuroretinal rim area, rim-to-disc area ratio corrected for disc size, and cup-to-disc area ratio corrected for disc size. Diagnostic power was lower for rim area in the temporal inferior and temporal superior disc sector, cup area corrected for disc size, and horizontal cup-to-disc diameter ratio corrected for disc size. Less useful for the differentiation between the normal subjects and the preperimetric glaucoma group were size of zones alpha and beta of parapapillary chorioretinal atrophy, and ratios of neuroretinal rim width and rim area comparing various optic disc sectors with each other.

conclusions. In subjects with ocular hypertension with retinal nerve fiber layer defects and normal conventional achromatic visual fields, the vertical cup-to-disc diameter ratio corrected for optic disc size, total neuroretinal rim area, rim-to-disc area ratio, and cup-to-disc area ratio corrected for disc size are the most valuable optic disc variables for early detection of glaucomatous optic nerve damage. Correction for optic disc size is necessary for optic disc variables directly or indirectly derived from the optic cup. Parapapillary atrophy is less important in the early detection of glaucoma.

Agrowing number of histologic and clinical studies have convincingly shown that optic nerve damage in patients with glaucoma occurs and can be detected before conventional achromatic computerized perimetry uncovers early visual field defects. ^{1 2 3 4 5 6} Clinical investigations using morphologic techniques have shown that quite a number of optic nerve head variables, such as the neuroretinal rim, as a whole and measured separately in various disc sectors; the shape of the neuroretinal rim; and the presence and size of parapapillary atrophy, are abnormal in some individuals with ocular hypertension who have normal conventional visual field examinations. ^{7 8 9 10 11 12 13}  

The purpose of the present study was to describe optic disc variables assessed by evaluation of clinical optic disc photographs and to compare sensitivity and specificity of these optic disc parameters in identifying ocular hypertensive patients with nerve fiber layer defects and normal visual fields and patients with visual field defects. Because some of these optic disc variables can be estimated ophthalmoscopically, it may be important for the clinical diagnosis of glaucoma in a busy practice in which recently developed imaging techniques such as scanning laser tomography, video ophthalmography, or retinal nerve fiber layer polarimetry are not yet available. It may additionally show the value of optic nerve head variables, such as parapapillary atrophy, that cannot reliably be measured by confocal scanning laser tomography or polarimetry. ¹⁴  

Materials and Methods

Patient Characteristics

The study included 500 normal subjects, 132 patients with ocular hypertension with retinal nerve fiber layer defects and normal visual fields (preperimetric glaucoma group), and 840 patients with glaucomatous visual field defects (perimetric glaucoma group; Table 1 ). Only one randomly selected eye per patient and subject was taken for statistical analysis. All subjects included in the study were white. Highly myopic eyes with a myopic refractive error exceeding −8 D were excluded because of a markedly different appearance of the optic nerve head in normal subjects, as well as in glaucoma patients. ^{15 16} The preperimetric glaucoma group and the normal group did not have a statistically significant (P = 0.07) difference in age. The difference in age between the normal group and the perimetric glaucoma group was significant. Because the optic disc parameters were statistically (P > 0.05) independent of age, the study groups were not matched for age. The study design complied with the Declaration of Helsinki for research involving human subjects. Informed consent was obtained from all subjects included in the study. Institutional Review Board and Ethics Committee approval was not required for this study. 

Selection criteria for all patients in the preperimetric and the perimetric glaucoma groups included an open anterior chamber angle observed on gonioscopy; retinal nerve fiber layer defects; and, except for the patients with normal-pressure glaucoma, intraocular pressure higher than 21 mm Hg. Retinal nerve fiber layer defects were defined as localized defects or as diffuse loss. A localized defect was a wedge-shaped rather than a spindlelike defect, running toward or touching the optic disc border for not more than 60° of the optic disc circumference. A pseudodefect was defined as a spindlelike rather than a wedge-shaped narrow dark area in the retinal nerve fiber layer. Diffuse loss of retinal nerve fiber layer was signaled by reduced visibility of the retinal nerve fiber layer in relation to the age of the patient and background pigmentation of the fundus. For assessment of the retinal nerve fiber layer, black-and-white, wide-angle retinal nerve fiber layer photographs were available for 348 patients. They had been taken and evaluated as reported recently. ¹⁷ Instead of paper prints, diapositives were developed. The slides were projected with a magnification of 15 times after maximal defocusing of the projector. The area of the blurred image of the optic disc was covered and then the projector refocused and the retinal nerve fiber layer evaluated. For the remaining patients, 15° color fundus photographs were used. 

The perimetric glaucoma group was divided into 562 patients with primary open-angle glaucoma, 128 patients with secondary open-angle glaucoma due to pseudoexfoliation syndrome (pseudoexfoliative glaucoma), 64 patients with secondary open-angle glaucoma due to primary melanin dispersion syndrome (pigmentary glaucoma), and 86 patients with normal-pressure glaucoma. Primary open-angle glaucoma was characterized by elevated intraocular pressure measurements higher than 21 mm Hg with no evident reason for it. Pseudoexfoliation syndrome was defined by the presence of a dandrufflike material on the lens zonules and the lens surface, especially in its center and periphery, separated by an intermediate clear zone, a secondary melanin dispersion with translucent defects in the parapupillary region of the iris pigment epithelium, often a hyperpigmentation of the anterior chamber angle, and a decreased facility to dilate the pupil. The eyes with pigmentary glaucoma were characterized by radial translucent defects in the periphery of the iris pigment epithelium and hyperpigmentation in the posterior and anterior chamber, including pigmentation of the ligamentum hyaloideocapsulare, Krukenberg’s spindle, melanin granules on the iris surface, pronounced hyperpigmentation of the anterior chamber angle, and pigment dispersion on mydriasis. Other reasons for pigment dispersion such as intraocular surgery or herpetic iritis had been excluded. Criteria for the diagnosis of normal-pressure glaucoma were maximal intraocular pressure readings equal to or less than 21 mm Hg in at least two 24-hour pressure profiles obtained by slit lamp applanation tonometry and containing measurements at 5 PM, 9 PM, midnight, 7 AM, and noon. Ophthalmoscopy, medical history, and neuroradiologic, neurologic, and medical examinations did not reveal any other reason, such as intrasellar or suprasellar tumors, retinal vessel occlusions, optic disc drusen, or nonarteritic anterior ischemic optic neuropathy, for optic nerve damage than glaucoma. 

Visual Field Testing

A glaucomatous visual field defect was defined as a G1 field (Octopus; Interzeag, Schlieren, Switzerland) with a mean visual field defect of more than 2 dB and location of the visual field defects typical for glaucoma, such as at least three adjacent test points with a deviation of 5 dB or more and with one test point with a deviation more than 10 dB lower, at least two adjacent test points with a deviation of 10 dB or more, and at least three adjacent test points with a deviation of 5 dB or more abutting the nasal horizontal meridian. The rate of false-positive or false-negative answers was less than 15%. Other reasons for visual field loss had been ruled out. 

Photography and Planimetry

For all eyes, 15° color stereo optic disc transparencies had been taken using a telecentric fundus camera (Carl Zeiss, Oberkochen, Germany). The study included all readable photographs that had been taken between the end of 1995 and the beginning of 1998 and that showed a sufficient photographic quality to be evaluated. This accounted for more than 85% of all photographs taken and for more than 90% of the ocular hypertensive subjects. Mixed together with photographs of more than 500 patients with other forms of glaucoma or optic nerve diseases, the disc transparencies were projected in a scale of 1 to 15. The outlines of the optic cup, optic disc, peripapillary scleral ring, and alpha zone and beta zone of parapapillary atrophy were plotted on paper and morphometrically analyzed. The border of the optic disc was identical with the inner side of the peripapillary scleral ring. The optic cup was defined on the basis of contour and not of pallor. Correspondingly, in some eyes the area with cupping was larger than the area with pallor. Parapapillary atrophy was differentiated into two zones. The alpha zone was located in the periphery of the parapapillary region and was characterized by irregular hypopigmentation and hyperpigmentation with a round border to the retina on its outer side. The beta zone was located between the alpha zone and the peripapillary scleral ring. It was whitish and showed visible sclera and visible large choroidal vessels. ¹⁸ For all measurements, the intrapapillary and parapapillary region was divided into four sectors. The temporal inferior sector and the temporal superior sector were right angled, and their bisector was tilted 13° temporal to the vertical optic disc axis. The temporal horizontal sector (64°) and the nasal sector (116°) covered the remaining areas. To obtain the measurements in millimeters or square millimeters, the ocular and photographic magnification was corrected according to the Littmann method, by using the anterior corneal curvature and the refractive error. ¹⁹ The planimetry of the optic disc photographs was performed by three examiners (KIP, WMB, and JBJ), each of whom had performed more than 500 examinations before the study was started. 

Statistical Analysis

Using these optic disc measurements, we additionally calculated the horizontal and vertical cup-to-disc diameter ratio, the ratio of the horizontal to vertical cup-to-disc diameter ratio, the cup-to-disc area ratio, the rim-to-disc area ratio, and the inferior-to-temporal, inferior-to-nasal, superior-to-temporal, and superior-to-nasal neuroretinal rim width ratios. Because some of the optic disc variables such as the area of the neuroretinal rim and optic cup depend on the size of the optic disc, ^{20 21 22} we tested all optic disc variables used in the study for a dependence on the optic disc size in the normal control group. We then calculated predicted and corrected values for the preperimetric glaucoma group and the perimetric glaucoma group (Tables 2 3) . The predicted values were determined using the linear regression line as determined in the normal group and taking the optic disc area of each eye of the preperimetric glaucoma group and perimetric glaucoma group as the independent variable. The corrected values were calculated as the ratio of measured value divided by the predicted value times the mean of the measured value in the normal study group (Table 3) . 

Results

Dependence on Optic Disc Size in the Normal Group

To test the variables for dependency on the size of the optic disc, we correlated all optic disc variables with optic disc area and calculated the equations of linear regressions in the normal study group (Table 2) . The regression lines were steepest, the ratio of the ascent of the regression line to the value of the intercept was highest, and the correlation coefficients were largest for horizontal cup-to-disc diameter ratio, vertical cup-to-disc diameter ratio, optic cup area, and cup-to-disc area ratio. The regression lines were flattest; the ratio of the steepness of the regression line to the value of the intercept was lowest; and the correlation coefficients were lowest for neuroretinal rim area, measured in the whole optic disc and determined separately in the four optic disc sectors; neuroretinal rim width; neuroretinal rim width ratios; and area of alpha zone and beta zone of parapapillary atrophy, measured as a whole and determined separately in the four sectors. 

Diagnostic Power

Highest diagnostic power, expressed as sensitivity and specificity, for the separation between the normal group and the ocular hypertensive group with nerve fiber layer defects and normal visual fields had the vertical cup-to-disc diameter ratio corrected for its dependence on the optic disc size, total neuroretinal rim area, rim-to-disc area ratio corrected for optic disc size, and the cup-to-disc area ratio corrected for optic disc size (Table 4 ; Figs. 1 2 and 3 ). These variables were followed by rim area in the temporal inferior disc sector, rim area in the temporal superior disc sector, the optic cup area corrected for its dependence on the optic disc size, and the horizontal cup-to-disc diameter ratio corrected for optic disc size. Less useful for the differentiation between the normal subjects and the ocular hypertensive individuals with nerve fiber layer defects were size of alpha zone and beta zone of parapapillary chorioretinal atrophy, neuroretinal rim width ratios, and neuroretinal rim area ratios. 

When the raw data were compared with the corrected variables, correction for a dependence on optic disc size increased the diagnostic power for those variables that were directly or indirectly connected with the size of the optic cup. These variables were the vertical cup-to-disc diameter ratio (Fig. 2) , horizontal cup-to-disc diameter ratio, cup area, and cup-to-disc area ratio. In contrast, the direct neuroretinal rim measurements such as neuroretinal rim area, measured as a whole and separately in the four optic disc sectors, did not markedly improve in diagnostic power when the corrected data instead of the raw data were used (Table 4 ; Fig. 1 ). The variable rim-to-disc area ratio was significantly dependent on the optic disc area, and its diagnostic power increased when the corrected data instead of the raw data were used (Table 4 ; Fig. 3 ). 

When the normal group was compared with the glaucoma group with visual field defects, total neuroretinal rim area (uncorrected) had the highest diagnostic power, followed by the vertical cup-to-disc diameter ratio corrected for its dependence on the optic disc size, rim-to-disc area ratio corrected for optic disc size, optic cup area corrected for optic disc size, and the horizontal cup-to-disc diameter ratio corrected for disc size (Table 4) . Diagnostic power was lower for area of the beta zone of parapapillary atrophy and neuroretinal rim width measurements. Least useful were the variables, area of alpha zone of parapapillary atrophy and neuroretinal rim area ratios. 

For all optic disc variables examined in the study, pronounced overlap between the normal control group and the two study groups was found (Figs. 1 2 3) . The differences were more marked when comparing the normal group with the glaucoma group with visual field defects than when comparing the normal group with the ocular hypertensive group with nerve fiber layer abnormalities and normal visual fields (Table 3) . Correspondingly, diagnostic power was higher for the separation of the perimetric glaucoma group from the normal group than for the separation of the preperimetric glaucoma group and the normal group (Table 4) . 

Discussion

Glaucoma has traditionally been defined by the triad of increased intraocular pressure, optic disc changes, and visual field defects. Histologic studies performed by Quigley et al. ¹ have shown, however, that there can be a significant loss of ganglion cells before evidence of functional loss on conventional achromatic visual field testing. For this reason attention has been focused on alternative, more sensitive ways of detecting early ganglion cell damage than is possible with white-on-white perimetry. In several studies on eyes with elevated intraocular pressure and normal visual fields, abnormal results in various psychophysical and electrophysical examinations were reported. ^{23 24 25 26} As with the newer psychophysical and electrophysiological techniques, it has been shown that abnormalities in the appearance of the optic disc may precede visual field defects. ^{2 3 4 5 6 7 8 9 10 11 12 13} These abnormalities include an unusually small area of the neuroretinal rim, an abnormal shape of the rim, high cup-to-disc ratios, unusually extensive parapapillary chorioretinal atrophy, and the presence of splinter-shaped hemorrhages at the optic disc border. Purpose of the present study was to evaluate the diagnostic power of these optic nerve head variables in the differentiation between normal eyes and eyes with increased intraocular pressure, abnormal retinal nerve fiber layer, and normal visual fields. 

The best variables to separate between the normal subjects and the individuals with ocular hypertension who have nerve fiber layer defects were the vertical cup-to-disc diameter ratio corrected for optic disc size, the total neuroretinal rim area, the rim-to-disc area ratio corrected for optic disc size, and the cup-to-disc area ratio corrected for disc size (Tables 3 4 ; Fig. 1 ). Interestingly, the vertical cup-to-disc diameter ratio corrected for disc size was one of the best variables. It confirms early studies by Armaly et al. ^{27 28} who have reported that the vertical and the horizontal cup-to-disc diameter ratios are useful for quantification of glaucomatous optic nerve damage and for the early detection of glaucoma. Later, the cup-to-disc diameter ratios partially lost their importance in the clinical diagnosis of glaucoma because it became apparent that the cup-to-disc diameter ratios depend on the disc size. ^{20 21 22} A high cup-to-disc diameter ratio can be normal, if the optic disc is large, ²⁹ and a low cup-to-disc diameter ratio can be glaucomatous if the optic disc is small. ³⁰ The results of the present study thus lead to a renaissance of the cup-to-disc diameter ratios in the clinical diagnosis of glaucoma if the dependence of the cup-to-disc diameter ratios on the disc size is taken into account. It confirms a previous study by Garway–Heath et al. ³¹ in which the vertical cup-to-disc diameter ratio in relation to the optic disc size was found to be useful clinically, especially to assist in identifying small glaucomatous optic discs. The clinical value of the vertical cup-to-disc diameter ratio corrected for disc size is further emphasized by the fact that the cup-to-disc diameter ratios and the disc size can be determined by a slit lamp examination, whereas neuroretinal rim area and the rim-to-disc area ratio as the two other important optic disc variables (Table 4) have to be determined by time-consuming and sophisticated techniques such as morphometric evaluation of fundus photographs or confocal scanning laser tomographic imaging of the optic nerve head. It means for the setting of a busy glaucoma practice, in which confocal tomographic laser scanning systems have not been introduced, that the optic disc size can be measured ophthalmoscopically using an ophthalmoscopic lens and a slit lamp, with which the length of the beam can be adjusted to the diameter of the optic disc. ³² After correction for the measured disc size, the vertical cup-to-disc diameter ratio is one of the most important variables to describe the status of the optic nerve in glaucomatous eyes. For the clinical description of an optic nerve, it may thus be acceptable to give the vertical cup-to-disc diameter ratio in combination with the estimated disc size. 

The reason that the corrected vertical cup-to-disc ratio was one of the best variables for the early detection of glaucomatous optic nerve damage may be the pattern of glaucomatous loss of neuroretinal rim. Rim loss usually starts predominantly in the inferior and superior optic disc regions. ^{3 9 33} It leads to a vertical elongation of the optic cup. ^{3 9 33} It also explains why the vertical cup-to-disc diameter ratio was superior to the horizontal cup-to-disc diameter ratio in differentiating the preperimetric glaucoma patients from the normal control individuals. The pattern of early glaucomatous rim loss, however, should be dealt with cautiously. It may be that because the retinal areas represented by the arcuate nerve fiber layer bundles are sampled more densely by the visual field test than other parts of the visual field, the chances of getting adjacent depressed points is increased. Early axonal loss elsewhere may therefore go unrecognized and may not lead to a diagnosis of glaucoma, because clusters of depressed field points are absent. This possible limitation, however, accounts for glaucomatous eyes with early visual field loss and it does not account for preperimetric glaucomatous eyes that were the main target of the present study. 

The findings of the present study may, with limitations, be transferred to optic disc measurements performed by confocal laser scanning tomography. They account especially for the vertical cup-to-disc diameter ratio, because rim width measurements expressed in relative terms are almost identical when planimetric measurements of optic disc photographs are compared with confocal scanning laser tomographic measurements of the optic nerve head. ³⁴ Measurements of the horizontal cup-to-disc diameter ratio, however, are different between the two methods, because the scanning laser tomograph considers all parts of the central retinal vessel trunk lying above the reference level to belong to the neuroretinal rim, although no optic nerve fibers may be present. Because of this difference between the two methods, the horizontal cup-to-disc diameter ratio is lower when measured by a confocal scanning laser tomograph than when determined by clinical judgment or by planimetry of optic disc photographs. In the clinical ophthalmoscopic estimation of the cup-to-disc diameter ratio, the marked interobserver variability must be considered, which, even for glaucoma experts, can be remarkably high. ^{35 36 37} Furthermore, it must be emphasized that the determination of the cup-disc-diameter ratio is only a part of the whole optic nerve head examination. 

Parapapillary atrophy was one of the worst of the tested variables for the separation of the normal subjects and the preperimetric glaucoma patients (Tables 3 4) . It confirms previous studies in which the differences between normal subjects and patients with early glaucomatous optic nerve damage were less marked for parapapillary atrophy than for neuroretinal rim area and cup-to-disc diameter ratios. ³⁸ It shows that parapapillary atrophy is helpful for the early diagnosis of glaucomatous optic nerve atrophy in some patients; however, one of its main advantages is the differentiation between the various types of chronic open-angle glaucoma. ³⁹  

The findings of the present study agree with the results of other investigations. The role of a vertical elongation of the optic cup resulting in an increased vertical cup-to-disc diameter ratio has already been shown by Armaly, ²⁷ Armaly and Saydegh, ²⁸ Pederson and Anderson, ³ Caprioli et al., ⁵ Tuulonen et al., ⁷ Tezel et al., ¹⁰ Garway–Heath et al., ³¹ and others. The importance of an abnormally small neuroretinal rim area for the early glaucoma diagnosis has been emphasized by Balazsi et al., ⁴ Caprioli et al., ⁵ Tuulonen et al., ⁷ Zeyen and Caprioli, ⁸ Tezel et al., ¹⁰ and Bathija et al. ¹³ The parapapillary chorioretinal atrophy in its role for the early detection of glaucoma has been highligted by Tezel et al., ^{11 12} and others. 

Another result of the study is that correcting for optic disc size did not markedly increase the diagnostic power of all optic disc variables tested. Correcting for optic disc size was useful and increased the diagnostic power of the optic cup area, cup-to-disc area ratio, vertical and horizontal cup-to-disc diameter ratio, and rim-to-disc area ratio (Table 4 ; Figs. 2 3 ). It did not pronouncedly change the sensitivity and specificity of the neuroretinal rim area measured as a whole and separated into the four disc sectors (Table 4 ; Fig. 1 ). Correspondingly, in the normal study group, the regression lines of the correlations between the variables and optic disc size were steepest, the ratio of the ascent of the regression line to the value of the intercept was highest, and the correlation coefficients were largest for vertical and horizontal cup-to-disc diameter ratio, optic cup area, and cup-to-disc area ratio. The regression lines were flattest, the ratio of the steepness of the regression line to the value of the intercept was lowest, and the correlation coefficients were lowest for neuroretinal rim area measured in the whole optic disc and determined separately in the four optic disc sectors, neuroretinal rim width, and neuroretinal rim width ratios (Fig. 2) . This agrees with a previous study on glaucoma in which the neuroretinal rim area was correlated with the mean visual field loss and in which the correlation coefficients did not significantly increase after correction of the neuroretinal rim area measurements for the individual optic disc size. ⁴⁰ The reason may be that with increasing optic disc size, the optic cup enlarges much more than the rim in normal eyes. ²² Correspondingly, the regression line of the correlation between optic disc area and cup area is by a factor of 0.70 to 0.33, or 2.1 steeper than the regression line of the correlation between disc area and neuroretinal rim area. ²² This may explain why correction for optic disc size increases the diagnostic power more for optic disc variables that are directly or indirectly derived from the optic cup, such as cup area, cup-to-disc diameter ratios, and the cup-to-disc area ratio, than for optic disc variables that mainly depend on the neuroretinal rim area. In this discussion, it must be considered that the slope of the regression line of rim area against disc area was somewhat shallower in the present study than that found in previous investigations, ^{21 22} in some of which other techniques had been used. To cite an example, in confocal laser scanning laser tomography, parts of the central retinal vessel trunk are incorporated into the neuroretinal rim area measurements, and the rim area measurements are therefore relatively larger than in the planimetric assessment of the optic disc photographs. This may be one reason that in confocal laser tomographic studies the slope of the regression line of rim area against disc area is steeper than in the present study. 

Another reason that the corrected rim was no better than the actual rim area in discriminating between normal and glaucomatous eyes may be that it is the absolute number of neurons that is important for visual function, not the number in relation to the predicted number. 

Seemingly in contrast to the finding that the corrected rim values were no better than the measured rim area data in discriminating between normal and glaucomatous eyes, the variable rim-to-disc area ratio increased in diagnostic power when corrected for disc size (Table 3 ; Fig. 3 ). The reason may be that the variable rim-to-disc area ratio can be described as:EQUATION 1  

\[\mathrm{Rim\ area/disc\ area\ ratio{=}\ (disc\ area\ -\ cup\ area)/disc\ area}\]

or  

\[\mathrm{(Disc\ area\ -\ cup\ area)/disc\ area{=}\ 1\ -\ cup\ area/disc\ area}\]

Equation 2 shows that cup area has replaced rim area in the variable rim-to-disc area ratio, and that disc area has an important position in the equation. Correspondingly, in the normal study group, the rim-to-disc area ratio was highly significantly (P < 0.001) correlated with the optic disc area, with a relatively high correlation coefficient (R = 0.643; Table 2 ). 

For all variables examined in the study, we found a pronounced interindividual variability in the normal group (Table 3) . Similar results have been obtained in other studies using planimetry of optic disc photographs ⁴¹ and for optic disc variables measured by confocal scanning laser tomography. ^{13 14 42 43 44 45 46} This marked interindividual variability is typical for many quantitative biologic variables, including body height and weight, and may be the reason that the normal group and the preperimetric glaucoma group showed a pronounced overlap in the quantitative optic disc variables. Correspondingly, sensitivity and specificity of these variables were relatively low in differentiating between the normal eyes and the eyes with preperimetric glaucoma (Tables 3 4 ; Figs. 1 2 3 ). 

There are limitations of the present study. It can be argued that the patients in the ocular hypertensive study subgroup with retinal nerve fiber layer defects and normal achromatic visual fields were highly selected and that they were not a representative sample of an unselected population group with ocular hypertension. This may have increased the difference between the normal control group and the ocular hypertensive group. The purpose of the study was, however, to compare the various optic disc variables in their diagnostic power to predict observable glaucomatous abnormalities of the retinal nerve fiber layer in eyes without visual field loss. It was not the goal of the study to examine how often a certain disc variable shows an abnormal result in an unselected group of subjects with ocular hypertension. Using nerve fiber layer loss as a morphologic selection criterion may have influenced the results of this morphologic study on optic disc variables. We did not use a perimetric parameter as a nonmorphologic selection criterion, because achromatic perimetric defects can be detectable later than morphologic alterations. ^{7 8 9 10 11 12 13} Blue-on-yellow visual field data were not available. ²⁵ Furthermore, using a perimetric parameter as selection criterion would have rendered impossible the formation of the preperimetric glaucoma study group. It can also be argued that the normal group and the perimetric glaucoma group differed significantly in age. The optic disc parameters were, however, statistically (P > 0.05) independent of age in the present study. This agrees with the results of the Rotterdam Study ⁴⁷ and other investigations ^{48 49} in which size and shape of the optic disc, optic cup, neuroretinal rim, and alpha and beta zones of parapapillary atrophy were not correlated with age. 

Future studies may investigate the question of which optic disc variables are the best to detect glaucomatous optic nerve damage in those patients in whom the assessment of the retinal nerve fiber layer is not conclusive and in whom there is doubt about the diagnosis. Future studies using confocal scanning laser tomographic techniques may evaluate whether the vertical cup-to-disc diameter ratio corrected for disc size is as useful in the computerized optic disc analysis as it is in the clinical assessment of the optic nerve head for the early detection of glaucomatous optic nerve damage. 

 Supported by Deutsche Forschungsgemeinschaft Grant SFB 539 (JBJ, AB, KIP, and WMB).

 Submitted for publication June 23, 1999; revised October 21 and December 29, 1999; accepted January 5, 2000.

 Commercial relationships policy: N.

 

Table 1.

 

Composition of the Study Groups

	Normal Subjects	Preperimetric Glaucoma Group	Perimetric Glaucoma Group
n	500	132	840
Age (y)	47.2 ± 15.6	49.5 ± 11.8	61.8 ± 15.1
Range	6–87	17–69	8–94
Women/men	228/272	49/83	424/416
Right eye/left eye	240/260	59/73	418/422
Refractive error (D)	−0.43 ± 1.95	−0.82 ± 2.20	−0.76 ± 2.29
Range	−8.00 to+4.75	−7.5 to+3.50	−8.00 to +4.75
Optic disc area (mm2)	2.72 ± 0.57	2.68 ± 0.72	2.72 ± 0.63
Range	1.31–3.98	1.42–6.43	1.40–6.00
Mean visual field defect (dB)		0.29 ± 0.98	8.78 ± 7.69
Range		−1.7–1.9	2.1–28.0

 Data are means ± SD.

Table 2.

 

Equations of the Regression Lines of the Correlations Between the Optic Disc Variables and Optic Disc Area in the Normal Study Group

	Pearson’s Correlation Coefficient	P	Equation of the Linear Regression Line	Ratio Regression Line Steepness/Intercept
Neuroretinal rim area
Total	0.237	<0.001	Y = 0.146 DA+ 1.302	0.112
Inferior temporal	0.222	<0.001	Y = 0.047 DA+ 0.356	0.132
Superior temporal	0.167	<0.001	Y = 0.029 DA+ 0.351	0.083
Nasal	0.027	0.55 (NS)
Temporal horizontal	0.224	<0.001	Y = 0.061 DA+ 0.151	0.403
Neuroretinal rim area ratios
Inferior/temporal	0.012	0.789 (NS)
Superior/temporal	0.001	0.979 (NS)
Inferior/nasal	0.180	<0.001	Y = 0.227 DA+ 0.523	0.434
Superior/nasal	0.163	<0.001	Y = 0.182 DA+ 0.550	0.331
Rim/disc area ratio	0.643	<0.001	Y =−0.183 DA+ 1.144	0.160
Neuroretinal rim width
Inferior	0.229	<0.001	Y =−0.065 DA+ 0.651	0.100
Superior	0.282	<0.001	Y =−0.078 DA+ 0.660	0.118
Nasal	0.257	<0.001	Y = −0.069 DA+ 0.617	0.112
Temporal	0.281	<0.001	Y =−0.085 DA+ 0.553	0.154
Neuroretinal rim width ratios
Inferior/temporal	0.057	0.299 (NS)
Inferior/nasal	0.093	0.091 (NS)
Superior/temporal	0.047	0.392 (NS)
Superior/nasal	0.002	0.966 (NS)
Cup/disc diameter ratio
Horizontal	0.571	<0.001	Y = 0.187 DA+ 0.085	2.200
Vertical	0.425	<0.001	Y = 0.168 DA+ 0.083	2.024
Horizontal/vertical	0.159	0.001	Y =−0.058 DA+ 1.297	0.045
Optic cup
Area	0.819	<0.001	Y = 0.854 DA− 1.302	0.656
Cup/disc area ratio	0.643	<0.001	Y = 0.183 DA− 0.144	1.270
Parapapillary atrophy
Alpha zone
Total	0.112	0.013	Y = 0.113 DA+ 0.292	0.387
Temporal inferior	0.127	0.005	Y = 0.033 DA+ 0.049	0.673
Temporal superior	0.092	0.041	Y = 0.022 DA + 0.055	0.400
Nasal	0.052	0.245 (NS)
Temporal horizontal	0.102	0.024	Y = 0.039 DA+ 0.173	0.225
Beta zone
Total	0.014	0.759 (NS)
Temporal inferior	0.033	0.470 (NS)
Temporal superior	0.002	0.968 (NS)
Nasal	0.026	0.558 (NS)
Temporal horizontal	0.008	0.859 (NS)

  N = 500. DA, disc area; NS, not significant.

Table 3.

 

Optic Disc Variables in the Study Groups

	Normal Subjects	Preperimetric Glaucoma Group	P	Perimetric Glaucoma Group	P
Neuroretinal rim area (mm2)
Total optic disc region
Uncorrected data	1.699 ± 0.352	1.368 ± 0.348	<0.0001	0.993 ± 0.531	<0.0001
Corrected area	1.698 ± 0.338	1.370 ± 0.327	<0.0001	0.991 ± 0.523	<0.0001
Temporal inferior sector
Uncorrected data	0.466 ± 0.105	0.376 ± 0.111	<0.0001	0.252 ± 0.164	<0.0001
Corrected area	0.449 ± 0.057	0.362 ± 0.100	<0.0001	0.242 ± 0.156	<0.0001
Temporal superior sector
Uncorrected data	0.432 ± 0.103	0.335 ± 0.094	<0.0001	0.251 ± 0.144	<0.0001
Corrected area	0.432 ± 0.100	0.335 ± 0.092	<0.0001	0.251 ± 0.143	<0.0001
Nasal sector
Uncorrected data	0.506 ± 0.192	0.393 ± 0.189	<0.0001	0.351 ± 0.184	<0.0001
Corrected area	0.507 ± 0.192	0.393 ± 0.188	<0.0001	0.351 ± 0.183	<0.0001
Temporal sector
Uncorrected data	0.317 ± 0.156	0.280 ± 0.167	<0.0001	0.168 ± 0.134	<0.0001
Corrected area	0.317 ± 0.149	0.282 ± 0.160	<0.0001	0.168 ± 0.132	<0.0001
Rim/disc area ratio
Uncorrected data	0.647 ± 0.162	0.530 ± 0.149	<0.0001	0.377 ± 0.208	<0.0001
Corrected values	0.647 ± 0.128	0.484 ± 0.541	<0.0001	0.384 ± 0.216	<0.0001
Vertical cup/disc ratio	0.541 ± 0.226	0.657 ± 0.149	<0.0001	0.776 ± 0.250	<0.0001
Vertical cup/disc ratio corrected for disc area	0.540 ± 0.260	0.677 ± 0.173	<0.0001	0.794 ± 0.279	<0.0001
Horizontal cup/disc ratio	0.593 ± 0.187	0.681 ± 0.143	<0.0001	0.767 ± 0.157	<0.0001
Horizontal cup/disc ratio corrected for disc area	0.588 ± 0.180	0.699 ± 0.166	<0.0001	0.779 ± 0.190	<0.0001
Horizontal/vertical cup/disc ratio	1.137 ± 0.203	1.047 ± 0.093	<0.0001	1.016 ± 0.346	<0.0001
Corrected horizontal/vertical cup/disc ratio	1.137 ± 0.196	1.045 ± 0.099	<0.0001	1.016 ± 0.339	<0.0001
Uncorrected data	1.019 ± 0.595	1.316 ± 0.683	0.0033	1.733 ± 0.765	<0.0001
Corrected area	1.030 ± 0.887	1.546 ± 0.840	<0.0001	1.679 ± 6.085	<0.0001
Cup/disc area ratio	0.353 ± 0.162	0.470 ± 0.149	<0.001	0.623 ± 0.208	<0.001
Cup/disc area ratio corrected for disc area	0.351 ± 0.146	0.507 ± 0.187	<0.001	0.665 ± 0.286	<0.001
Neuroretinal rim width (mm)
Inferior	0.330 ± 0.164	0.254 ± 0.088	0.09 (NS)	0.166 ± 0.139	<0.0001
Superior	0.457 ± 0.150	0.344 ± 0.070	0.006	0.237 ± 0.164	<0.0001
Nasal	0.483 ± 0.153	0.386 ± 0.090	0.008	0.235 ± 0.167	<0.0001
Temporal	0.437 ± 0.145	0.336 ± 0.073	0.002	0.261 ± 0.139	<0.0001
Rim width ratios
Inferior/temporal	1.876 ± 4.143	1.714 ± 0.775	0.88 (NS)	1.670 ± 2.216	<0.0001
Inferior/nasal	1.143 ± 0.320	1.175 ± 0.264	0.41 (NS)	0.879 ± 0.507	<0.0001
Superior/temporal	1.757 ± 3.61	1.464 ± 0.434	0.72 (NS)	1.663 ± 2.339	0.0002
Superior/nasal	1.072 ± 0.253	1.060 ± 0.278	0.92 (NS)	0.956 ± 0.775	<0.0001
Rim area ratio
Inferior/temporal	1.886 ± 2.901	1.726 ± 0.816	0.99 (NS)	1.776 ± 1.862	0.0015
Superior/temporal	1.722 ± 2.201	1.571 ± 0.961	0.18 (NS)	1.891 ± 2.856	0.59 (NS)
Inferior/nasal	1.139 ± 0.717	1.231 ± 0.758	0.46 (NS)	0.773 ± 0.583	<0.0001
Superior/nasal	1.046 ± 0.639	1.113 ± 0.725	0.36 (NS)	0.880 ± 2.119	<0.0001
Parapapillary atrophy
Alpha zone
Total (mm2)	0.60 ± 0.57	0.86 ± 0.72	<0.001	0.77 ± 0.68	<0.001
Temporal inferior	0.14 ± 0.15	0.21 ± 0.20	<0.001	0.18 ± 0.20	0.001
Temporal superior	0.11 ± 0.13	0.19 ± 0.19	<0.001	0.16 ± 0.18	<0.001
Nasal	0.07 ± 0.21	0.12 ± 0.23	0.002	0.12 ± 0.22	<0.001
Temporal	0.28 ± 0.22	0.35 ± 0.23	0.001	0.32 ± 0.24	0.005
Beta zone
Total (mm2)	0.24 ± 0.65	0.22 ± 0.50	0.23 (NS)	0.77 ± 1.73	<0.0001
Temporal inferior	0.07 ± 0.19	0.07 ± 0.16	0.21 (NS)	0.22 ± 0.53	<0.0001
Temporal superior	0.05 ± 0.14	0.04 ± 0.13	0.72 (NS)	0.15 ± 0.40	<0.0001
Nasal	0.03 ± 0.17	0.01 ± 0.07	0.93 (NS)	0.16 ± 0.55	<0.0001
Temporal	0.10 ± 0.23	0.10 ± 0.19	0.37 (NS)	0.24 ± 0.44	<0.0001

  P, significance of difference between the group in the preceeding column and the normal group. The significant differences between the study groups remained significant after applying Bonferroni’s correction for performing multiple comparisons. Data are means ± SD. NS, not significant.

Table 4.

 

Sensitivity at a Prefixed Specificity of 80% and at a Prefixed Specificity of 95% to Separate Normal Eyes from Preperimetric Glaucomatous Eyes and to Separate Normal Eyes from Perimetric Glaucomatous Eyes

	Normal Group/Preperimetric Glaucoma Group			Normal Group/Perimetric Glaucoma Group
		Sensitivity at Specificity 80%	Sensitivity at Specificity 95%	Sensitivity at Specificity 80%	Sensitivity at Specificity 95%
Neuroretinal rim area
Total optic disc region
Uncorrected data	58.3	34.1	78.8	66.3
Corrected area	63.6	28.8	78.4	62.9
Temporal inferior sector
Uncorrected data	53.8	26.5	78.4	62.5
Corrected area	53.8	26.5	77.7	62.5
Temporal superior sector
Uncorrected data	56.1	21.2	74.4	52.3
Corrected area	58.3	25.0	79.9	53.4
Nasal sector
Uncorrected data	37.1	15.2	50.9	16.1
Corrected area	37.1	15.2	50.2	14.7
Temporal sector
Uncorrected data	39.4	19.7	62.9	44.7
Corrected area	45.5	15.2	65.2	43.1
Rim/disc area ratio	45.5	16.7	72.3	55.3
Corrected values	62.1	30.3	78.4	63.4
Rim area ratio
Inferior/temporal	31.8	9.1	20.8	11.6
Superior/temporal	31.8	11.4	21.0	6.9
Inferior/nasal	23.5	9.8	52.4	35.9
Superior/nasal	32.6	8.3	48.0	29.9
Neuroretinal rim width
Temporal	28.6	21.4	63.8	46.0
Superior	42.9	14.3	79.6	57.9
Inferior	57.1	14.3	80.5	62.5
Nasal	42.9	14.3	70.1	47.7
Rim width ratio
Inferior/temporal	21.4	14.3	38.0	18.0
Superior/temporal	21.4	7.1	31.5	18.9
Inferior/nasal	21.4	7.1	52.9	34.1
Superior/nasal	21.4	7.1	46.8	26.1
Horizontal cup/disc ratio	36.4	10.6	62.2	47.0
Horizontal cup/disc ratio corrected for optic disc area	51.5	25.8	74.8	58.1
Vertical cup/disc ratio	49.2	18.9	67.0	46.6
Vertical cup/disc ratio corrected for optic disc area	57.6	31.8	76.6	58.2
Horizontal/vertical cup/disc ratio	37.2	15.5	62.1	28.3
Corrected horizontal/vertical cup/disc ratio	47.3	14.0	64.6	23.2
Optic cup area
Uncorrected data	26.5	9.8	56.5	29.7
Corrected area	54.5	23.5	76.0	61.7
Cup/disc area ratio	44.7	17.3	72.0	57.0
Cup/disc area ratio corrected for optic disc area	57.6	32.6	76.3	61.7
Parapapillary atrophy
Alpha zone
Total	31.8	12.1	33.6	9.7
Temporal inferior sector	36.4	9.1	29.9	8.0
Temporal superior sector	31.1	19.0	31.6	13.8
Nasal sector	27.3	9.1	31.8	6.9
Temporal sector	31.8	8.0	28.9	8.9
Beta zone
Total	20.5	3.0	49.9	19.5
Temporal inferior sector	19.7	6.1	48.4	17.8
Temporal superior sector	22.0	2.3	49.2	16.5
Nasal sector	32.6	3.0	29.0	23.6
Temporal sector	22.7	3.0	45.4	11.4

Figure 1.

 

Receiver operating characteristic (ROC) curve for the optic disc variables, total neuroretinal rim area uncorrected for optic disc size and total neuroretinal rim area corrected for optic disc size, in the separation of 500 normal subjects and 132 patients with ocular hypertension with retinal nerve fiber layer defects and normal achromatic visual fields. Area under the ROC for total neuroretinal rim area uncorrected for optic disc size is 0.759 ± 0.024 (SE; 95% confidence interval: 0.711–0.807), and area measurement for the ROC of total neuroretinal rim area corrected for optic disc size is 0.767 ± 0.024 (95% confidence interval: 0.720–0.814). The almost identical areas under the two ROC curves with largely overlapping confidence intervals suggest that the two ROC curves are not significantly different.

Figure 2.

 

Receiver operating characteristic (ROC) curve for the optic disc variables, vertical cup-to-disc diameter ratio uncorrected for optic disc size and vertical cup-to-disc diameter ratio corrected for optic disc size, in the separation of 500 normal subjects and 132 patients with ocular hypertension with retinal nerve fiber layer defects and normal achromatic visual fields. Area under the ROC for vertical cup-to-disc diameter ratio uncorrected for optic disc size is 0.720 ± 0.025 (95% confidence interval: 0.673–0.768), and area under the ROC for the vertical cup-to-disc diameter ratio corrected for optic disc size is 0.759 ± 0.024 (95% confidence interval: 0.711–0.807).

Figure 3.

 

Receiver operating characteristic (ROC) curve for the optic disc variables, rim-to-disc area ratio uncorrected for optic disc size and rim-to-disc area ratio corrected for optic disc size, in the separation of 500 normal subjects and 132 patients with ocular hypertension with retinal nerve fiber layer defects and normal achromatic visual fields. Area under the ROC for rim-to-disc area ratio uncorrected for optic disc size is 0.707 ± 0.025 (95% confidence interval: 0.658–0.755), and area under the ROC for the variable rim-to-disc area ratio corrected for optic disc size is 0.753 ± 0.025 (95% confidence interval: 0.704–0.802).