- Email: [email protected]
Differentiating patients with glaucoma from glaucoma suspects and normal subjects by nerve fiber layer assessment with scanning laser polarimetry1, 2
Differentiating Patients with Glaucoma from Glaucoma Suspects and Normal Subjects by Nerve Fiber Layer Assessment with Scanning Laser Polarimetry Neil T. Choplin, MD,1 Diane C. Lundy, MD,1 Andreas W. Dreher, PhD2 Purpose: A study was conducted to determine normative data for nerve fiber layer measurements as obtained by scanning laser polarimetry with the Laser Diagnostic Technologies Nerve Fiber Analyzer II, identify factors affecting the measurements, and identify parameters capable of differentiating normal subjects from patients with glaucoma and patients suspected of having glaucoma because of ocular hypertension or because of a large cup-to-disc ratio (GS-disc). Design: A case series. Participants: Four hundred normal subjects, 35 patients with ocular hypertension, 42 patients with glaucoma, and 17 glaucoma suspects based on optic disc appearance participated. Methods: Nerve fiber layer thickness assessments were determined in normal subjects (with normalappearing optic nerves and normal visual fields). The results were compared to measurements from samples of age-matched patients with ocular hypertension (with normal visual fields), patients suspected of having glaucoma based on enlarged cup-to-disc ratios, and patients with open-angle glaucoma who had visual field loss. Results: The majority of the parameters derived from the measurements showed no significant relationship to age, although some parameters tended to decrease with increasing age. Multiple parameters showed statistically significant differences between normal subjects and patients with glaucoma. In particular, the intraellipse sector variability, an indirect measure of the shape of the nerve fiber layer in an ellipse surrounding the nerve head, showed statistically significant differences between normal subjects and patients with glaucoma as well as between glaucoma suspects and normal subjects. Similar results were seen with the superior maxima, the average thickness assessment value of the 1500 thickest points in the superior bundle. Conclusions: Assessments of nerve fiber layer thickness as determined by scanning laser polarimetry can differentiate patients with glaucoma from normal subjects and may identify otherwise undetected damage in glaucoma suspects. Ophthalmology 1998;105:2068 –2076
Originally received: October 10, 1997. Revision accepted: June 16, 1998. Manuscript no. 97722. 1 Departments of Ophthalmology and Clinical Research, Naval Medical Center, San Diego, California. 2 Laser Diagnostic Technologies, San Diego, California. Presented in part at the Association for Research in Vision and Ophthalmology annual meeting, Ft. Lauderdale, Florida, May 1997. Supported by Navy Clinical Investigation Program #S-93-LH000-082. Drs. Choplin and Lundy have no proprietary interest in any product discussed in this presentation, and this presentation is not intended to be a commercial endorsement of any product by the Navy or the United States government. Dr. Dreher is the President and Chief Executive Officer of Laser Diagnostic Technologies and has a proprietary interest in the Nerve Fiber Analyzer. The views contained in this presentation are those of the authors and are not to be construed as being the official opinions of the Department of the Navy, the Department of Defense, or the United States government. Address correspondence and reprint requests to Neil T. Choplin, MD, Clinical Research Department, Naval Medical Center San Diego, 34800 Bob Wilson Drive, San Diego, CA 92134-5000.
2068
Glaucoma is a disease of the retinal nerve fiber layer characterized by progressive loss of axons, presumably as a result of elevated intraocular pressure, some destructive process induced by elevated intraocular pressure, or some other initiating factor. Functional loss in glaucoma, usually manifested by visual field loss, occurs when sufficient nerve fibers have been lost either focally, resulting in a nerve fiber bundle defect, or diffusely, causing generalized loss of sensitivity. It has been estimated that 40% to 50% of nerve fibers may be lost before functional loss occurs. Therefore, standard perimetric techniques may not be sensitive enough to detect early disease or detect subtle progression. The Nerve Fiber Analyzer II (NFA II; Laser Diagnostic Technologies, San Diego, California) is a scanning laser polarimeter capable of assessing the thickness of the retinal nerve fiber layer in a grid centered on the optic nerve head.1 The machine actually measures the phase shift of a polarized laser light traversing the nerve fiber layer (retardation), with the amount of retardation directly proportional to the
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects thickness of the tissue through which it has passed. Although the machine does not directly measure nerve fiber layer thickness, the measurements obtained with the instrument have been shown to correlate with histopathologic measurements of nerve fiber layer thickness in a primate model.2 Prior studies using the NFA I have shown differences in nerve fiber layer measurements between normal subjects and patients with ocular hypertension,3 but normative values for nerve fiber layer measurements as determined by the NFA II were not well established, and factors influencing nerve fiber layer measurements are not well known. A study was conducted to determine values for nerve fiber layer assessment in a normal population as measured by the NFA II, determine factors that may affect the thickness assessments in normal subjects, and determine which parameters, if any, could differentiate patients with glaucoma from the normal subjects or from those suspected of having glaucoma (based on ocular hypertension or a suspicious optic disc but normal visual fields).
Subjects and Methods Normative Data Collection Normal subjects were recruited from the Naval Medical Center San Diego (Drs. Choplin and Lundy), the New York Eye and Ear Infirmary (Dr. Robert Ritch), the Wilmer Eye Institute (Dr. Harry Quigley), the Glaucoma Associates of Denver (Dr. Randy Craven), the University of Washington (Dr. Richard Mills), and the Rotterdam Eye Hospital (Dr. Hans Lemij). Inclusion criteria included age of at least 18 years (no upper limit), healthy volunteer requiring no ocular medications, no history of any ocular disease except refractive error or strabismus, best-corrected visual acuity of 20/40 or better in both eyes, no family history of ocular hypertension or glaucoma, intraocular pressure of 21 mmHg or less in both eyes, and refractive error less than 5 diopters sphere and/or 2 diopters cylinder with no previous refractive surgery. Potential subjects underwent a comprehensive eye examination consisting of refraction and measurement of best-corrected visual acuity, pupillary examination, slit-lamp examination and gonioscopy, and a preliminary fundus examination. If preliminary inclusion criteria were met, the subjects then underwent visual field testing using the full-threshold 24 –2 examination of the Humphrey Field Analyzer, a dilated fundus examination, optic disc photography, and axial length measurements by A-scan ultrasonography. Patients were included in the study if the visual fields and optic nerves were judged to be normal, and no exclusion criteria were identified. Criteria for normal visual fields included acceptable reliability (false-positive and false-negative responses ⬍ 33%, fixation losses ⬍ 20%), no points showing depressions below the 10th percentile compared to age-corrected normal (unless judged to be caused by artifact), and no visual field indices below the 10th percentile (mean deviation [MD], a measure of height of the visual field, short-term fluctuation [SF], a measure of intratest variability, and corrected pattern standard deviation [CPSD], a measure of irregularity in the shape of the visual field suggestive of the presence of scotomata). Exclusion criteria included unreliable visual fields, any nonartifactual visual field defect, abnormal optic discs, cup-to-disc (C/D) ratio greater than 0.5 or asymmetry between C/D ratio of the two eyes greater than 0.2, cataract resulting in visual acuity less than 20/40, or any other concomitant ocular
pathology. Age-related macular degeneration was not a criterion for exclusion, provided the best-corrected visual acuity was 20/40 or better. After informed consent was obtained from subjects meeting inclusion criteria, nerve fiber layer assessments were obtained using the NFA II before pupillary dilation. All measurements of normal subjects at Naval Medical Center San Diego were performed by an experienced examiner from Laser Diagnostic Technologies using the current hardware version of the NFA II; measurements at other centers were similarly performed by an experienced examiner. Each eye was measured three times using a 15° by 15° grid and a mean image created from the three images for each eye. The standard deviation of the mean of the three measurements was 8 m or less for inclusion.
Glaucoma Suspects and Glaucoma Patients A representative sample of patients with glaucoma and patients suspected of having glaucoma was selected at random from the measurements obtained previously on patients from Naval Medical Center San Diego. All available medical records from patients who had undergone NFA II measurements were reviewed by one of the authors (NTC or DCL) to verify the diagnosis. Only patients in whom the diagnosis was evident from available medical records were included in the study. Three distinct populations of patients were identified from the medical record review: 1. Patients with glaucoma (including open-angle, chronic-angle closure requiring ongoing treatment, pigment dispersion, exfoliation syndrome, low-tension glaucoma) had characteristic visual field defects of glaucoma or optic nerve changes or both. 2. Patients with ocular hypertension (OHTN) had untreated intraocular pressures greater than 22 mmHg with normalappearing optic nerves and visual fields. 3. Glaucoma suspects based on the appearance of the disc (GS-disc) who had enlarged C/D ratios of 0.6 or greater in one or both eyes or C/D asymmetry of 0.3 or greater but normal visual fields and IOP less than 22 mmHg. A mean image was created for each eye from the previously obtained NFA II measurements using three scans obtained during the same session.
Nerve Fiber Layer Data From each mean image, an ellipse denoting the edge of the optic disc was placed by an experienced operator. An ellipse measuring 1.75 disc diameters in size was determined by the defined disc edge. Nerve fiber layer thickness assessments and various parameters, defined in Table 1, were determined for the whole image, the image quadrants (temporal, 50° wide; superior, 120° wide; nasal, 70° wide; and inferior, 120° wide), the entire ellipse, the ellipse quadrants corresponding to the image quadrants, and the ellipse sectors (numbered 1–32) defined by dividing the ellipse into 32 equal sections of 11.25°, each beginning at the temporal horizontal midline (sector 1) and extending clockwise for right eyes and counterclockwise for left eyes. For labeling purposes, the sectors were distributed evenly into four quadrants of eight sectors each, with sectors 1 through 4 and 29 through 32 corresponding to the temporal region, 5 through 12 superior, 13 through 20 nasal, and 21 through 28 inferior. The measured and calculated parameters included integrals for the area under the plot of thickness assessment around the ellipse by location (part of the standard machine printout), representing nerve fiber volume, average thickness assessment along the ellipse,
2069
Ophthalmology Volume 105, Number 11, November 1998 Table 1. Definitions of Nerve Fiber Layer Analysis Parameters Parameter Group Integrals
Averages
Maximas
Medians
Modulation Ratios
IESV
Parameter Name Polar integral Superior integral Temporal integral Inferior integral Nasal integral Polar average Superior average Temporal average Inferior average Nasal average Superior maxima Inferior maxima Temporal median Nasal median Average without blood vessels Ellipse modulation Maximum modulation S/I ratio S/N ratio S/T ratio I/N ratio I/T ratio Intra-ellipse sector variability
Definition The polar integral represents the total area under the nerve fiber layer curve along the ellipse surrounding the optic nerve; the others are the area measurements under the respective quadrants. The integrals also represent the volume of the nerve fiber layer.
The polar average is the average of all of the nerve fiber layer thickness measurements along the ellipse (10 pixels wide) surrounding the optic nerve, expressed in microns; the others are for the respective quadrants of the ellipse.
The maxima are the averages of the 1500 thickest measurements within the respective bundles. By selecting the thickest pixels, areas overlying blood vessels are automatically excluded. The median (50th percentile) thickness measurement within the respective image quadrants. The average thickness measurement for the entire image after application of a blood vessel removal algorithm. The difference between the thickest and thinnest measurements within the ellipse. The difference between the thickest and thinnest measurements within the image. The ratio of the average thickness measurements between the quadrants: S ⫽ superior; I ⫽ inferior; N ⫽ nasal; T ⫽ temporal.
After dividing the ellipse into 32 equal sectors and determining the average thickness in each sector, the spread of the measurements from the mean thickness of the 32 sectors is the intra-ellipse sector variability. This is actually the standard deviation of the mean sector thicknesses.
“maximas” representing the average of the 1500 thickest pixels in the superior and inferior image quadrants, medians for the nasal and temporal quadrants of the entire image, the average thickness assessment for the entire image after removal of areas overlying blood vessels, modulation parameters for the entire image and the ellipse representing the difference between the thickest and the thinnest pixels, and ratios between thickness assessments in different quadrants. The mean thickness assessment value was calculated for each ellipse sector, and the mean and standard deviation of the sector means were then determined for all sectors. The standard deviation of the mean of the 32 sectors was termed intra-ellipse sector variability (IESV).
Silver Springs, MD), which was used for further analysis. Analysis of variance (ANOVA) was used to determine differences among the four groups, and one-way ANOVA using Bonferroni correction for multiple comparisons was used to determine differences between normal subjects and those in each of the other three groups for each parameter as well as the thickness assessments in each of the ellipse sectors. The level of statistical significance was 0.05.
Results Characteristics of the Study Groups
Statistical Analysis Data from all images from the selected subset of the normal subjects and all patients in the other three groups, as well as patient demographics, were entered into a spreadsheet using Microsoft Excel (Microsoft Corp, Redmond, WA). The data from the normal subjects were analyzed using Microsoft Excel, looking for relationships between nerve fiber layer parameters and age, race, axial length, refractive error, and symmetry between the two eyes. Regression analysis and Pearson correlation coefficients were calculated between all parameters and age in the normal subjects, looking for factors that might influence the nerve fiber layer thickness assessments with age. For the non-normal groups, one eye was chosen for analysis: the eye showing the most glaucoma damage was selected from each patient in the glaucoma group and one eye chosen at random from each of the OHTN and GS-disc patients. The data from both eyes in the normal group were averaged together for purposes of comparison to the non-normal groups. The data were imported into GB-STAT for Windows Version 6.0 (Dynamic Microsystems, Inc,
2070
The normative group consisted of more than 400 subjects, from whom 235 subjects were chosen at random so that the mean age of the normal group matched that of the other groups. The glaucoma group had 42 patients, the OHTN group had 35 patients, and the GS-disc group had 17 patients. Table 2 summarizes the demographics of all of the study groups. There was no difference between the mean ages of the four groups. Of the three visual field indices (mean defect, short-term fluctuation, and corrected pattern standard deviation), only the glaucoma group patients showed a statistically significant difference from the normal subjects in all three parameters, while neither of the other groups showed a statistically significant difference from normal in any visual field parameter. The patients with glaucoma therefore were the only ones to have visual field loss, consistent with the definitions of the group. The C/D ratios were statistically significantly larger in all groups compared to those of normal subjects (C/D ⫽ 0.5 was the maximum allowable in the normative group), but the C/D ratios were statistically significantly larger in the GS-disc and glaucoma groups compared to the OHTN group (P ⫽ 0.00 for both).
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects Table 2. Patient Demographics
Normals Ocular hypertension with normal visual fields (OHTN) (n ⫽ 35) Glaucoma suspects based upon disc appearance (GS-disc) (n ⫽ 17) Glaucoma (GL) (n ⫽ 42) Differences between groups, mean ⫾ standard error of the mean and level of significance (NS ⫽ not significant with P ⬎ 0.05) OHTN to normal
Age (yrs)
Mean Defect (dB)
Short-term Fluctuation (dB)
Corrected Pattern Standard Deviation (dB)
62.6 ⫾ 12.1 (n ⫽ 235)
⫺0.78 ⫾ 2.24 (n ⫽ 220)
1.52 ⫾ 0.58 (n ⫽ 220)
1.57 ⫾ 1.32 (n ⫽ 220)
0.33 ⫾ 0.14 (n ⫽ 182)
64.0 ⫾ 13.6
⫺1.26 ⫾ 1.93
1.42 ⫾ 0.53
1.93 ⫾ 1.47
0.44 ⫾ 0.17
57.6 ⫾ 10.0 63.6 ⫾ 14.0
⫺1.45 ⫾ 1.94 ⫺7.45 ⫾ 6.36
1.57 ⫾ 0.47 2.35 ⫾ 0.86
2.15 ⫾ 1.31 6.49 ⫾ 3.86
0.61 ⫾ 0.17 0.71 ⫾ 0.16
All groups* P ⫽ 0.000 (ANOVA) 0.10 ⫾ 0.10 NS ⫺0.05 ⫾ 0.14 NS ⫺0.83 ⫾ 0.11* P ⬍ 0.01
All groups* P ⫽ 0.000 (ANOVA) ⫺0.36 ⫾ 0.24 NS ⫺0.58 ⫾ 0.33 NS ⫺4.92 ⫾ 0.32* P ⬍ 0.01
All groups* P ⫽ 0.000 (ANOVA) ⫺0.11 ⫾ 0.03* P ⬍ 0.01 ⫺0.28 ⫾ 0.04* P ⬍ 0.01 ⫺0.38 ⫾ 0.02* P ⬍ 0.01
All groups NS (ANOVA) —
GS-disc to normal
—
GL to normal
—
All groups* P ⫽ 0.000 (ANOVA) 0.48 ⫾ 0.15 NS 0.67 ⫾ 0.56 NS 6.67 ⫾ 0.54* P ⬍ 0.01
Cup-to-Disc Ratio
* Significant difference using Bonferroni correction for multiple comparisons.
Nerve Fiber Layer Data in the Normal Group Analysis was conducted only on the subgroup of 235 normal subjects, not the entire group. There was no statistically significant Table 3. Correlations of Nerve Fiber Layer Parameters with Age in Normals Parameter Inferior integral Temporal median Temporal average (along ellipse) Temporal maximum Temporal integral Temporal average (entire quadrant) Temporal minimum Inferior average (along ellipse) Polar integral Inferior maximum Nasal minimum I/N ratio Polar average Nasal median Average without blood vessels Nasal average (along ellipse) Nasal average (entire quadrant) Nasal maximum Nasal integral Superior integral Superior maximum I/T ratio S/N ratio Superior average (along ellipse) Maximum modulation S/I ratio S/T ratio Intra-ellipse sector variability Ellipse modulation
Correlation Coefficient
P
0.17 0.15 0.14 0.11 0.11
0.01 0.02 0.03 0.09 0.09
0.11 0.10 0.09 0.07 0.07 0.04 0.03 0.02 0.01 0.01 0.01 ⫺0.01 ⫺0.01 ⫺0.02 ⫺0.03 ⫺0.07 ⫺0.10 ⫺0.11 ⫺0.13 ⫺0.16 ⫺0.19 ⫺0.19 ⫺0.11 ⫺0.24
0.10 0.14 0.21 0.23 0.29 0.56 0.59 0.84 0.88 0.86 0.85 0.82 0.83 0.69 0.60 0.20 0.11 0.07 0.04 0.01 0.00 0.00 0.00 0.00
I ⫽ inferior; S ⫽ superior; N ⫽ nasal; T ⫽ temporal.
correlation between nerve fiber layer parameters and axial length or refractive error. Table 3 summarizes the relationship between all of the nerve fiber layer parameters (defined in Table 1) and age, rank-ordered by correlation coefficient. The inferior integral, temporal median, and temporal average along the ellipse showed a statistically significant tendency to increase with age, while the superior average along the ellipse, the maximum modulation, the S/I ratio, the S/T ratio, the intra-ellipse sector variability, and the ellipse modulation showed a statistically significant tendency to decrease with age. None of the other parameters showed a statistically significant correlation with age.
Nerve Fiber Layer Parameters in the Non-normal Groups Table 4 lists the means and standard deviations for the NFA II parameters for each of the four groups, and Table 5 lists the probability values for the observed differences between the normal group and each of the other three groups based on one-way ANOVA with Bonferroni correction for multiple comparisons. The probability values are indicated in bold if the difference was statistically significant, and the value was higher in the normal subjects and in italics if the difference was statistically significant and the value was lower in normal subjects. Almost all parameters, with the exception of temporal values, showed a statistically significant difference between patients with glaucoma and normal subjects. Multiple parameters also showed statistically significant differences between the other two groups and normal subjects. Interestingly, the nasal values and temporal average were statistically significantly thicker in the ocular hypertension group compared with those of normal subjects, and the nasal median was thicker in patients with glaucoma compared with that of normal subjects. Although almost all of the thickness assessments in the GS-disc group were not statistically significantly different from those of normal, the superior values, the maximum modulation, the IESV, and all of the superior ratios were statistically significantly different from normal, indicating that perhaps in patients with large cup-to-disc ratios, the nerve fibers are distributed
2071
Ophthalmology Volume 105, Number 11, November 1998 Table 4. Nerve Fiber Layer Parameter Values: Mean ⫾ SD
Parameter
Normals (n ⴝ 235)
Glaucoma (n ⴝ 42)
Ocular Hypertension (n ⴝ 35)
Polar integral Superior integral Temporal integral Inferior integral Nasal integral Polar average Superior average Temporal average Inferior average Nasal average Superior maxima Inferior maxima Temporal median Nasal median Average without blood vessels Ellipse modulation Maximum modulation S/I ratio S/N ratio S/T ratio I/N ratio I/T ratio Intra-ellipse sector variability
0.568 ⫾ 0.107 0.220 ⫾ 0.42 0.050 ⫾ 0.016 0.221 ⫾ 0.046 0.077 ⫾ 0.019 68.42 ⫾ 11.78 78.80 ⫾ 14.20 44.25 ⫾ 13.49 78.04 ⫾ 14.56 49.17 ⫾ 10.63 93.28 ⫾ 15.98 90.79 ⫾ 16.92 42.32 ⫾ 12.45 44.54 ⫾ 10.91 81.81 ⫾ 13.50 98.77 ⫾ 15.70 59.03 ⫾ 11.86 1.04 ⫾ 0.16 2.16 ⫾ 0.40 2.33 ⫾ 0.59 2.09 ⫾ 0.35 2.25 ⫾ 0.51 30.32 ⫾ 5.40
0.497 ⫾ 0.107 0.174 ⫾ 0.041 0.051 ⫾ 0.0136 0.191 ⫾ 0.048 0.081 ⫾ 0.020 61.71 ⫾ 12.66 64.54 ⫾ 15.35 46.67 ⫾ 12.81 69.59 ⫾ 15.29 53.08 ⫾ 11.70 73.54 ⫾ 18.45 80.08 ⫾ 18.64 43.12 ⫾ 12.06 49.91 ⫾ 12.19 77.42 ⫾ 14.40 74.95 ⫾ 21.19 42.78 ⫾ 16.07 0.93 ⫾ 0.18 1.51 ⫾ 0.38 1.78 ⫾ 0.49 1.64 ⫾ 0.37 1.93 ⫾ 0.49 20.32 ⫾ 6.33
0.566 ⫾ 0.133 0.204 ⫾ 0.055 0.053 ⫾ 0.015 0.222 ⫾ 0.051 0.087 ⫾ 0.025 72.37 ⫾ 14.73 77.19 ⫾ 18.55 51.13 ⫾ 13.94 83.16 ⫾ 16.25 58.75 ⫾ 14.15 87.54 ⫾ 19.71 93.91 ⫾ 17.88 48.24 ⫾ 14.19 53.30 ⫾ 13.66 86.30 ⫾ 17.05 93.17 ⫾ 22.76 52.62 ⫾ 12.64 0.94 ⫾ 0.14 1.67 ⫾ 0.32 1.89 ⫾ 0.45 1.80 ⫾ ⫺.31 2.04 ⫾ 0.47 26.96 ⫾ 6.61
Glaucoma Suspects Based upon Disc Appearance (n ⴝ 17) 0.550 ⫾ 0.116 0.193 ⫾ 0.043 0.054 ⫾ 0.014 0.219 ⫾ 0.054 0.084 ⫾ 0.021 65.20 ⫾ 11.74 67.78 ⫾ 13.27 47.70 ⫾ 10.46 76.17 ⫾ 16.81 52.81 ⫾ 10.89 77.42 ⫾ 15.27 86.78 ⫾ 18.68 44.77 ⫾ 10.05 47.40 ⫾ 10.89 79.19 ⫾ 12.87 89.88 ⫾ 20.62 48.51 ⫾ 13.88 0.91 ⫾ 0.17 1.66 ⫾ 0.27 1.80 ⫾ 0.50 1.87 ⫾ 0.40 2.00 ⫾ 0.50 25.71 ⫾ 6.52
I ⫽ inferior; S ⫽ superior; N ⫽ nasal; T ⫽ temporal; SD ⫽ standard deviation.
differently around the nerve head than in normal subjects. The average thickness assessment across the image after removal of the blood vessels (average without blood vessels) showed a statistically significant difference between patients with ocular hypertension and patients with glaucoma (higher in OHTN), but it did not differentiate normal from glaucoma. (Note that there is no parameter indicating the average thickness for the entire image without blood vessel removal.) In this study, the ellipse measuring 1.75 disc diameters in size and centered on the optic nerve head was divided into 32 equal sectors and the average nerve fiber layer thickness assessment value determined within each sector. Figure 1 shows the average thickness assessment value per sector plotted against location along the ellipse for each of the groups. The typical “double-hump” pattern of distribution of thickness around the ellipse is preserved in each group, but the peaks are lowered considerably in the glaucoma group, indicative of loss of superior and inferior fibers. Indeed, sectors 5 through 13 (superior) and sectors 21 through 27 (inferior) for the glaucoma group all show differences that are statistically significantly lower than those of normal with P values of less than 0.01. None of these sectors showed statistically significantly lower values from normal in the OHTN group (P ⬍ 0.05), although there were sectors in the GS-disc group with lower values than those of the normal subjects: sectors 6 to 8 (P ⬍ 0.01) and sectors 9 and 10 (P ⬍ 0.05). These differences in the GS-disc group were only in superior sectors. The thinnest values for the nasal quadrant were observed in the normal group, and the difference from the normal in both the OHTN and the glaucoma groups (but not the GS-disc group) was statistically significant. Figure 2 summarizes the probability values for all of the observed differences between the groups. Figure 3 shows the mean intra-ellipse sector variability and the 95% confidence intervals for each group, and Table 6 summarizes the differences between the groups. The OHTN and GS-disc
2072
groups appear to be similar to each other, but this parameter differentiates the two suspect groups and the patients with glaucoma from the normal subjects.
Discussion This study examined characteristics of the normal nerve fiber layer as determined by scanning laser polarimetry (or nerve fiber analysis) as performed by the NFA II and compared the results to age-matched patients with glaucoma or those suspected of having glaucoma based on elevated intraocular pressure (OHTN) or a suspicious-appearing disc (GS-disc). It also attempted to identify the relationship between aging and changes in the nerve fiber layer. Although some parameters showed weak correlation with age that were statistically significant, none would be considered clinically significant. Prior studies have shown similar correlation between nerve fiber layer thickness assessments and age.4 However, the normal population analyzed in this study is actually a subset of all of the normative data collected. Failure to show a negative correlation between age and thickness values as demonstrated by Chi and coworkers4 may reflect that most patients younger than 40 years of age were not included in analysis to age match the normal group to the other groups. In this study, scanning laser polarimetry with the NFA II was capable of distinguishing normal subjects from those with glaucoma by a number of parameters. Weinreb et al5 showed that nerve fiber layer thickness assessments as determined by scanning laser polarimetry were correlated with
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects Table 5. Statistical Significance of Observed Differences P* for Difference between Normal and:
Parameter
P for ANOVA (all groups)
Glaucoma
Ocular Hypertension
Glaucoma Suspects Based upon Disc Appearance
Polar integral Superior integral Temporal integral Inferior integral Nasal integral Polar average Superior average Temporal average Inferior average Nasal average Superior maxima Inferior maxima Temporal median Nasal median Average without blood vessels Ellipse modulation Maximum modulation S/I ratio S/N ratio S/T ratio I/N ratio I/T ratio Intra-ellipse sector variability
0.002 0.000 0.504 0.003 0.028 0.001 0.000 0.029 0.001 0.000 0.000 0.001 0.070 0.001 0.043† 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.001‡ 0.000‡ 1.000 0.001‡ 1.000 0.007‡ 0.000‡ 1.000 0.005‡ 0.228 0.000‡ 0.002‡ 1.000 0.031§ 0.372 0.000‡ 0.000‡ 0.000‡ 0.000‡ 0.000‡ 0.000‡ 0.001‡ 0.000‡
1.000 0.252 1.000 1.000 0.04§ 0.456 1.000 0.028§ 0.358 0.000§ 0.352 1.000 0.056 0.000§ 0.467 0.478 0.033‡ 0.003‡ 0.000‡ 0.000‡ 0.000‡ 0.135 0.008‡
1.000 0.082 1.000 1.000 1.000 1.000 0.020‡ 1.000 1.000 1.000 0.001‡ 1.000 1.000 1.000 1.000 0.270 0.006‡ 0.009‡ 0.000‡ 0.001‡ 0.079 0.259 0.009‡
I ⫽ inferior; S ⫽ superior; N ⫽ nasal; T ⫽ temporal. * Derived from one way analysis of variance (ANOVA) with Bonferroni correction for multiple comparisons. † Statistically significantly higher in ocular hypertension compared with glaucoma but each group no different from normals. ‡ Significant difference with normal value higher. § Significant difference with normal value lower.
visual field loss in patients with glaucoma. Compared to visual field testing, nerve fiber layer analysis is faster, easier for the patient, completely objective, and has less variability because of patient factors that would be considered subjective. Therefore, if it can be shown that nerve fiber layer analysis can identify patients with glaucoma with a high degree of certainty and identify patients with normal visual fields as having evidence of structural damage to their nerve fiber layers, it will prove to be a useful tool in the diagnosis and management of glaucoma. Identifying patients with early damage offers the theoretic advantage of beginning therapy earlier in the hopes of preventing permanent visual loss. Some of the parameters studied appear to differentiate between not only normal subjects and those with glaucoma but between normal subjects and glaucoma suspects as well. For example, as illustrated in Figure 3, the IESV, an indirect measure of the shape of the nerve fiber layer along the ellipse (analogous to pattern standard deviation in visual field testing), clearly separates normal subjects from those with glaucoma, with the two glaucoma suspect groups in between. Because the normal nerve fiber layer is thickest in the superior and inferior bundles and thinner nasally and temporally, there will be a larger spread of measurements around the mean thickness estimates along the ellipse com-
pared to a group that has lost fibers from the superior or inferior bundles (or both), such as patients with glaucoma. The IESV would therefore decrease as tissue is lost, since there would be less variability in the measurements. Indeed, of all the parameters examined in this study, the IESV was the only one to differentiate not only normal subjects from the other groups but patients with glaucoma from the other groups as well (one-way ANOVA with Bonferroni correction for multiple comparisons). Even though the range of normal values is fairly large and there is some overlap in thickness values found in normal subjects and the other groups (based on the means and standard deviations), parameters that examine the shape of the distribution of thickness assessments around the optic nerve head, such as the IESV, may prove more valuable in differentiating between normal subjects and patients with glaucoma. One thickness assessment that may prove useful in differentiating the groups is the superior maxima, as illustrated in Figure 4. This parameter, which is the average of the 1500 thickest points in the superior bundle, shows a clear delineation between normal subjects and glaucoma patients, with no overlap between the 95% confidence intervals. Parameters such as this one may prove to be valuable in screening populations at risk for glaucoma. For example, a superior maxima value of less than 80 may be highly
2073
Ophthalmology Volume 105, Number 11, November 1998
Figure 1. Average thickness per sector around ellipse.
unlikely to be found in normal subjects, and a value greater than 90 may be highly unlikely to be found in patients with glaucoma. This significance or predictive ability of values between 80 and 90 is not known at this time. Further studies are needed to determine the usefulness of NFA II assessments for screening.
The group of patients studied who were suspected of having glaucoma based on the appearance of their discs (GS-disc), although small, raises some interesting questions. The intraocular pressure and visual fields of the subjects in this group were normal, but many parameters of their nerve fiber layers were outside of the normal
Figure 2. Significance of observed differences between groups for ellipse sector.
2074
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects
Figure 3. Intra-ellipse sector variability, mean, and 95% confidence interval.
range. For example, the superior integral (a measure of the nerve fiber layer volume), the average thickness assessment in the superior quadrant, and the superior maxima (the average of the 1500 points with the highest assessment values in the superior bundle), the maximum modulation (difference between highest and lowest values), and the IESV (another measure of the difference between highest and lowest) all showed statistically significant decreases from the normal values. Does this mean that this group of patients has lost fibers from the superior bundle (like “low-tension glaucoma”), or are their nerve fiber layers anatomically different (i.e., the fibers are distributed differently around the nerve head) from normal? The argument for a different distribution of Table 6. Intra-ellipse Sector Variability Group
Value*
Normal OHTN
30.32 ⫾ 5.40 26.96 ⫾ 6.61
GS-disc
25.71 ⫾ 6.52
Glaucoma
20.32 ⫾ 6.33
Differences† From From From From From From
normal: 3.36 ⫾ 1.01 GS-disc: 1.25 ⫾ 1.95 glaucoma: 6.64 ⫾ 1.48 normal: 4.61 ⫾ 1.38 glaucoma: 5.39 ⫾ 1.84 normal 10.00 ⫾ 0.93
* Microns, mean ⫾ standard deviation. † Mean ⫾ standard error of the mean.
P value 0.00 0.52 0.00 0.00 0.01 0.00
fibers around the optic nerve is partially supported by the fact that the polar integral and the polar average were not statistically significantly different from normal in this group, suggesting that the total nerve fiber layer in this group is not different from that in normal subjects. Certainly, these patients bear careful follow-up over time to look for changes, since otherwise healthy individuals with physiologic cupping may represent a variant of normal. Remember, the assessments of nerve fiber layer thickness as determined by the NFA II have not yet been shown to correlate with actual known nerve fiber layer thickness measurements as determined from anatomic study. Thickness assessments obtained with this instrument are calculated based on the measurement of retardation, or phase shift of the incident polarized laser light due to the birefringent property of the nerve fiber layer. The reported thickness assessments as expressed in micrometers are based on studies in two cynomolgus monkeys, in which 1° of retardation corresponded to 7.4 m of thickness.2 The actual relationship between retardation and nerve fiber layer thickness in humans is not known but for clinical purposes probably is not important. This study establishes the differences between different groups of clinical subjects using the NFA II and shows that normal subjects measure differently than patients with glaucoma. With further study, it may become possible to select a parameter or number of parameters that
2075
Ophthalmology Volume 105, Number 11, November 1998
Figure 4. Superior maxima (average of 1500 thickest points in superior bundle), mean, and 95% confidence interval.
will identify patients with glaucoma or those at risk of having glaucoma develop with a high degree of specificity and sensitivity. If this proves to be the case, then not knowing whether the assessment is equal to the precise anatomic thickness is probably of little significance. In conclusion, the following items can be determined: 1. Multiple parameters appear capable of differentiating normal subjects from patients with glaucoma, and many patients currently considered to be glaucoma suspects (with normal visual fields) have intermediate values between the normal subjects and the patients with glaucoma. It is not known at this time whether these parameter values are predictive of further progression to glaucoma. 2. Some parameters, such as the superior maxima and the intra-ellipse sector variability, seem to differentiate between normal, glaucoma, and the two glaucoma suspect groups (OHTN and GS-disc) and may prove useful as a screening tool. 3. Future studies of nerve fiber layer analysis are required to determine the usefulness of these parameters for glaucoma screening in the general population, the predictability of the development of glaucoma in patients with abnormal parameter values and normal visual fields, the applicability of NFA II assessments to other optic neuropathies, and the differentiation of
2076
glaucoma from other optic neuropathies or diseases of the retinal nerve fiber layer. Acknowledgments. The authors thank the following Nerve Fiber Analyzer users who contributed to the normative database that was used in this study: E. Randy Craven, MD, Hans Lemij, MD, Richard Mills, MD, Robert Ritch, MD, and Harry Quigley, MD.
References 1. Dreher AW, Reiter K, Weinreb RN. Spatially resolved birefringence of the retinal nerve-fiber layer assessed with a retinal laser ellipsometer. Applied Optics 1992;31:3730 –5. 2. Weinreb RN, Dreher AW, Coleman A, et al. Histopathologic validation of Fourier-ellipsometry measurements of retinal nerve fiber layer thickness. Arch Ophthalmol 1990;108:557– 60. 3. Tjon-Fo-Sang MJ, de Vries J, Lemij HG. Measurement by nerve fiber analyzer of retinal nerve fiber layer thickness in normal subjects and patients with ocular hypertension. Am J Ophthalmol 1996;122:220 –7. 4. Chi Q, Tomiat G, Inazumi K, et al. Evaluation of the effect of aging on the retinal nerve fiber layer thickness using scanning laser polarimetry. J Glaucoma 1995;6:406 –13. 5. Weinreb RN, Shakiba S, Sample PA, et al. Association between quantitative nerve fiber layer measurement and visual field loss in glaucoma. Am J Ophthalmol 1995;120:732– 8.
Originally received: October 10, 1997. Revision accepted: June 16, 1998. Manuscript no. 97722. 1 Departments of Ophthalmology and Clinical Research, Naval Medical Center, San Diego, California. 2 Laser Diagnostic Technologies, San Diego, California. Presented in part at the Association for Research in Vision and Ophthalmology annual meeting, Ft. Lauderdale, Florida, May 1997. Supported by Navy Clinical Investigation Program #S-93-LH000-082. Drs. Choplin and Lundy have no proprietary interest in any product discussed in this presentation, and this presentation is not intended to be a commercial endorsement of any product by the Navy or the United States government. Dr. Dreher is the President and Chief Executive Officer of Laser Diagnostic Technologies and has a proprietary interest in the Nerve Fiber Analyzer. The views contained in this presentation are those of the authors and are not to be construed as being the official opinions of the Department of the Navy, the Department of Defense, or the United States government. Address correspondence and reprint requests to Neil T. Choplin, MD, Clinical Research Department, Naval Medical Center San Diego, 34800 Bob Wilson Drive, San Diego, CA 92134-5000.
2068
Glaucoma is a disease of the retinal nerve fiber layer characterized by progressive loss of axons, presumably as a result of elevated intraocular pressure, some destructive process induced by elevated intraocular pressure, or some other initiating factor. Functional loss in glaucoma, usually manifested by visual field loss, occurs when sufficient nerve fibers have been lost either focally, resulting in a nerve fiber bundle defect, or diffusely, causing generalized loss of sensitivity. It has been estimated that 40% to 50% of nerve fibers may be lost before functional loss occurs. Therefore, standard perimetric techniques may not be sensitive enough to detect early disease or detect subtle progression. The Nerve Fiber Analyzer II (NFA II; Laser Diagnostic Technologies, San Diego, California) is a scanning laser polarimeter capable of assessing the thickness of the retinal nerve fiber layer in a grid centered on the optic nerve head.1 The machine actually measures the phase shift of a polarized laser light traversing the nerve fiber layer (retardation), with the amount of retardation directly proportional to the
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects thickness of the tissue through which it has passed. Although the machine does not directly measure nerve fiber layer thickness, the measurements obtained with the instrument have been shown to correlate with histopathologic measurements of nerve fiber layer thickness in a primate model.2 Prior studies using the NFA I have shown differences in nerve fiber layer measurements between normal subjects and patients with ocular hypertension,3 but normative values for nerve fiber layer measurements as determined by the NFA II were not well established, and factors influencing nerve fiber layer measurements are not well known. A study was conducted to determine values for nerve fiber layer assessment in a normal population as measured by the NFA II, determine factors that may affect the thickness assessments in normal subjects, and determine which parameters, if any, could differentiate patients with glaucoma from the normal subjects or from those suspected of having glaucoma (based on ocular hypertension or a suspicious optic disc but normal visual fields).
Subjects and Methods Normative Data Collection Normal subjects were recruited from the Naval Medical Center San Diego (Drs. Choplin and Lundy), the New York Eye and Ear Infirmary (Dr. Robert Ritch), the Wilmer Eye Institute (Dr. Harry Quigley), the Glaucoma Associates of Denver (Dr. Randy Craven), the University of Washington (Dr. Richard Mills), and the Rotterdam Eye Hospital (Dr. Hans Lemij). Inclusion criteria included age of at least 18 years (no upper limit), healthy volunteer requiring no ocular medications, no history of any ocular disease except refractive error or strabismus, best-corrected visual acuity of 20/40 or better in both eyes, no family history of ocular hypertension or glaucoma, intraocular pressure of 21 mmHg or less in both eyes, and refractive error less than 5 diopters sphere and/or 2 diopters cylinder with no previous refractive surgery. Potential subjects underwent a comprehensive eye examination consisting of refraction and measurement of best-corrected visual acuity, pupillary examination, slit-lamp examination and gonioscopy, and a preliminary fundus examination. If preliminary inclusion criteria were met, the subjects then underwent visual field testing using the full-threshold 24 –2 examination of the Humphrey Field Analyzer, a dilated fundus examination, optic disc photography, and axial length measurements by A-scan ultrasonography. Patients were included in the study if the visual fields and optic nerves were judged to be normal, and no exclusion criteria were identified. Criteria for normal visual fields included acceptable reliability (false-positive and false-negative responses ⬍ 33%, fixation losses ⬍ 20%), no points showing depressions below the 10th percentile compared to age-corrected normal (unless judged to be caused by artifact), and no visual field indices below the 10th percentile (mean deviation [MD], a measure of height of the visual field, short-term fluctuation [SF], a measure of intratest variability, and corrected pattern standard deviation [CPSD], a measure of irregularity in the shape of the visual field suggestive of the presence of scotomata). Exclusion criteria included unreliable visual fields, any nonartifactual visual field defect, abnormal optic discs, cup-to-disc (C/D) ratio greater than 0.5 or asymmetry between C/D ratio of the two eyes greater than 0.2, cataract resulting in visual acuity less than 20/40, or any other concomitant ocular
pathology. Age-related macular degeneration was not a criterion for exclusion, provided the best-corrected visual acuity was 20/40 or better. After informed consent was obtained from subjects meeting inclusion criteria, nerve fiber layer assessments were obtained using the NFA II before pupillary dilation. All measurements of normal subjects at Naval Medical Center San Diego were performed by an experienced examiner from Laser Diagnostic Technologies using the current hardware version of the NFA II; measurements at other centers were similarly performed by an experienced examiner. Each eye was measured three times using a 15° by 15° grid and a mean image created from the three images for each eye. The standard deviation of the mean of the three measurements was 8 m or less for inclusion.
Glaucoma Suspects and Glaucoma Patients A representative sample of patients with glaucoma and patients suspected of having glaucoma was selected at random from the measurements obtained previously on patients from Naval Medical Center San Diego. All available medical records from patients who had undergone NFA II measurements were reviewed by one of the authors (NTC or DCL) to verify the diagnosis. Only patients in whom the diagnosis was evident from available medical records were included in the study. Three distinct populations of patients were identified from the medical record review: 1. Patients with glaucoma (including open-angle, chronic-angle closure requiring ongoing treatment, pigment dispersion, exfoliation syndrome, low-tension glaucoma) had characteristic visual field defects of glaucoma or optic nerve changes or both. 2. Patients with ocular hypertension (OHTN) had untreated intraocular pressures greater than 22 mmHg with normalappearing optic nerves and visual fields. 3. Glaucoma suspects based on the appearance of the disc (GS-disc) who had enlarged C/D ratios of 0.6 or greater in one or both eyes or C/D asymmetry of 0.3 or greater but normal visual fields and IOP less than 22 mmHg. A mean image was created for each eye from the previously obtained NFA II measurements using three scans obtained during the same session.
Nerve Fiber Layer Data From each mean image, an ellipse denoting the edge of the optic disc was placed by an experienced operator. An ellipse measuring 1.75 disc diameters in size was determined by the defined disc edge. Nerve fiber layer thickness assessments and various parameters, defined in Table 1, were determined for the whole image, the image quadrants (temporal, 50° wide; superior, 120° wide; nasal, 70° wide; and inferior, 120° wide), the entire ellipse, the ellipse quadrants corresponding to the image quadrants, and the ellipse sectors (numbered 1–32) defined by dividing the ellipse into 32 equal sections of 11.25°, each beginning at the temporal horizontal midline (sector 1) and extending clockwise for right eyes and counterclockwise for left eyes. For labeling purposes, the sectors were distributed evenly into four quadrants of eight sectors each, with sectors 1 through 4 and 29 through 32 corresponding to the temporal region, 5 through 12 superior, 13 through 20 nasal, and 21 through 28 inferior. The measured and calculated parameters included integrals for the area under the plot of thickness assessment around the ellipse by location (part of the standard machine printout), representing nerve fiber volume, average thickness assessment along the ellipse,
2069
Ophthalmology Volume 105, Number 11, November 1998 Table 1. Definitions of Nerve Fiber Layer Analysis Parameters Parameter Group Integrals
Averages
Maximas
Medians
Modulation Ratios
IESV
Parameter Name Polar integral Superior integral Temporal integral Inferior integral Nasal integral Polar average Superior average Temporal average Inferior average Nasal average Superior maxima Inferior maxima Temporal median Nasal median Average without blood vessels Ellipse modulation Maximum modulation S/I ratio S/N ratio S/T ratio I/N ratio I/T ratio Intra-ellipse sector variability
Definition The polar integral represents the total area under the nerve fiber layer curve along the ellipse surrounding the optic nerve; the others are the area measurements under the respective quadrants. The integrals also represent the volume of the nerve fiber layer.
The polar average is the average of all of the nerve fiber layer thickness measurements along the ellipse (10 pixels wide) surrounding the optic nerve, expressed in microns; the others are for the respective quadrants of the ellipse.
The maxima are the averages of the 1500 thickest measurements within the respective bundles. By selecting the thickest pixels, areas overlying blood vessels are automatically excluded. The median (50th percentile) thickness measurement within the respective image quadrants. The average thickness measurement for the entire image after application of a blood vessel removal algorithm. The difference between the thickest and thinnest measurements within the ellipse. The difference between the thickest and thinnest measurements within the image. The ratio of the average thickness measurements between the quadrants: S ⫽ superior; I ⫽ inferior; N ⫽ nasal; T ⫽ temporal.
After dividing the ellipse into 32 equal sectors and determining the average thickness in each sector, the spread of the measurements from the mean thickness of the 32 sectors is the intra-ellipse sector variability. This is actually the standard deviation of the mean sector thicknesses.
“maximas” representing the average of the 1500 thickest pixels in the superior and inferior image quadrants, medians for the nasal and temporal quadrants of the entire image, the average thickness assessment for the entire image after removal of areas overlying blood vessels, modulation parameters for the entire image and the ellipse representing the difference between the thickest and the thinnest pixels, and ratios between thickness assessments in different quadrants. The mean thickness assessment value was calculated for each ellipse sector, and the mean and standard deviation of the sector means were then determined for all sectors. The standard deviation of the mean of the 32 sectors was termed intra-ellipse sector variability (IESV).
Silver Springs, MD), which was used for further analysis. Analysis of variance (ANOVA) was used to determine differences among the four groups, and one-way ANOVA using Bonferroni correction for multiple comparisons was used to determine differences between normal subjects and those in each of the other three groups for each parameter as well as the thickness assessments in each of the ellipse sectors. The level of statistical significance was 0.05.
Results Characteristics of the Study Groups
Statistical Analysis Data from all images from the selected subset of the normal subjects and all patients in the other three groups, as well as patient demographics, were entered into a spreadsheet using Microsoft Excel (Microsoft Corp, Redmond, WA). The data from the normal subjects were analyzed using Microsoft Excel, looking for relationships between nerve fiber layer parameters and age, race, axial length, refractive error, and symmetry between the two eyes. Regression analysis and Pearson correlation coefficients were calculated between all parameters and age in the normal subjects, looking for factors that might influence the nerve fiber layer thickness assessments with age. For the non-normal groups, one eye was chosen for analysis: the eye showing the most glaucoma damage was selected from each patient in the glaucoma group and one eye chosen at random from each of the OHTN and GS-disc patients. The data from both eyes in the normal group were averaged together for purposes of comparison to the non-normal groups. The data were imported into GB-STAT for Windows Version 6.0 (Dynamic Microsystems, Inc,
2070
The normative group consisted of more than 400 subjects, from whom 235 subjects were chosen at random so that the mean age of the normal group matched that of the other groups. The glaucoma group had 42 patients, the OHTN group had 35 patients, and the GS-disc group had 17 patients. Table 2 summarizes the demographics of all of the study groups. There was no difference between the mean ages of the four groups. Of the three visual field indices (mean defect, short-term fluctuation, and corrected pattern standard deviation), only the glaucoma group patients showed a statistically significant difference from the normal subjects in all three parameters, while neither of the other groups showed a statistically significant difference from normal in any visual field parameter. The patients with glaucoma therefore were the only ones to have visual field loss, consistent with the definitions of the group. The C/D ratios were statistically significantly larger in all groups compared to those of normal subjects (C/D ⫽ 0.5 was the maximum allowable in the normative group), but the C/D ratios were statistically significantly larger in the GS-disc and glaucoma groups compared to the OHTN group (P ⫽ 0.00 for both).
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects Table 2. Patient Demographics
Normals Ocular hypertension with normal visual fields (OHTN) (n ⫽ 35) Glaucoma suspects based upon disc appearance (GS-disc) (n ⫽ 17) Glaucoma (GL) (n ⫽ 42) Differences between groups, mean ⫾ standard error of the mean and level of significance (NS ⫽ not significant with P ⬎ 0.05) OHTN to normal
Age (yrs)
Mean Defect (dB)
Short-term Fluctuation (dB)
Corrected Pattern Standard Deviation (dB)
62.6 ⫾ 12.1 (n ⫽ 235)
⫺0.78 ⫾ 2.24 (n ⫽ 220)
1.52 ⫾ 0.58 (n ⫽ 220)
1.57 ⫾ 1.32 (n ⫽ 220)
0.33 ⫾ 0.14 (n ⫽ 182)
64.0 ⫾ 13.6
⫺1.26 ⫾ 1.93
1.42 ⫾ 0.53
1.93 ⫾ 1.47
0.44 ⫾ 0.17
57.6 ⫾ 10.0 63.6 ⫾ 14.0
⫺1.45 ⫾ 1.94 ⫺7.45 ⫾ 6.36
1.57 ⫾ 0.47 2.35 ⫾ 0.86
2.15 ⫾ 1.31 6.49 ⫾ 3.86
0.61 ⫾ 0.17 0.71 ⫾ 0.16
All groups* P ⫽ 0.000 (ANOVA) 0.10 ⫾ 0.10 NS ⫺0.05 ⫾ 0.14 NS ⫺0.83 ⫾ 0.11* P ⬍ 0.01
All groups* P ⫽ 0.000 (ANOVA) ⫺0.36 ⫾ 0.24 NS ⫺0.58 ⫾ 0.33 NS ⫺4.92 ⫾ 0.32* P ⬍ 0.01
All groups* P ⫽ 0.000 (ANOVA) ⫺0.11 ⫾ 0.03* P ⬍ 0.01 ⫺0.28 ⫾ 0.04* P ⬍ 0.01 ⫺0.38 ⫾ 0.02* P ⬍ 0.01
All groups NS (ANOVA) —
GS-disc to normal
—
GL to normal
—
All groups* P ⫽ 0.000 (ANOVA) 0.48 ⫾ 0.15 NS 0.67 ⫾ 0.56 NS 6.67 ⫾ 0.54* P ⬍ 0.01
Cup-to-Disc Ratio
* Significant difference using Bonferroni correction for multiple comparisons.
Nerve Fiber Layer Data in the Normal Group Analysis was conducted only on the subgroup of 235 normal subjects, not the entire group. There was no statistically significant Table 3. Correlations of Nerve Fiber Layer Parameters with Age in Normals Parameter Inferior integral Temporal median Temporal average (along ellipse) Temporal maximum Temporal integral Temporal average (entire quadrant) Temporal minimum Inferior average (along ellipse) Polar integral Inferior maximum Nasal minimum I/N ratio Polar average Nasal median Average without blood vessels Nasal average (along ellipse) Nasal average (entire quadrant) Nasal maximum Nasal integral Superior integral Superior maximum I/T ratio S/N ratio Superior average (along ellipse) Maximum modulation S/I ratio S/T ratio Intra-ellipse sector variability Ellipse modulation
Correlation Coefficient
P
0.17 0.15 0.14 0.11 0.11
0.01 0.02 0.03 0.09 0.09
0.11 0.10 0.09 0.07 0.07 0.04 0.03 0.02 0.01 0.01 0.01 ⫺0.01 ⫺0.01 ⫺0.02 ⫺0.03 ⫺0.07 ⫺0.10 ⫺0.11 ⫺0.13 ⫺0.16 ⫺0.19 ⫺0.19 ⫺0.11 ⫺0.24
0.10 0.14 0.21 0.23 0.29 0.56 0.59 0.84 0.88 0.86 0.85 0.82 0.83 0.69 0.60 0.20 0.11 0.07 0.04 0.01 0.00 0.00 0.00 0.00
I ⫽ inferior; S ⫽ superior; N ⫽ nasal; T ⫽ temporal.
correlation between nerve fiber layer parameters and axial length or refractive error. Table 3 summarizes the relationship between all of the nerve fiber layer parameters (defined in Table 1) and age, rank-ordered by correlation coefficient. The inferior integral, temporal median, and temporal average along the ellipse showed a statistically significant tendency to increase with age, while the superior average along the ellipse, the maximum modulation, the S/I ratio, the S/T ratio, the intra-ellipse sector variability, and the ellipse modulation showed a statistically significant tendency to decrease with age. None of the other parameters showed a statistically significant correlation with age.
Nerve Fiber Layer Parameters in the Non-normal Groups Table 4 lists the means and standard deviations for the NFA II parameters for each of the four groups, and Table 5 lists the probability values for the observed differences between the normal group and each of the other three groups based on one-way ANOVA with Bonferroni correction for multiple comparisons. The probability values are indicated in bold if the difference was statistically significant, and the value was higher in the normal subjects and in italics if the difference was statistically significant and the value was lower in normal subjects. Almost all parameters, with the exception of temporal values, showed a statistically significant difference between patients with glaucoma and normal subjects. Multiple parameters also showed statistically significant differences between the other two groups and normal subjects. Interestingly, the nasal values and temporal average were statistically significantly thicker in the ocular hypertension group compared with those of normal subjects, and the nasal median was thicker in patients with glaucoma compared with that of normal subjects. Although almost all of the thickness assessments in the GS-disc group were not statistically significantly different from those of normal, the superior values, the maximum modulation, the IESV, and all of the superior ratios were statistically significantly different from normal, indicating that perhaps in patients with large cup-to-disc ratios, the nerve fibers are distributed
2071
Ophthalmology Volume 105, Number 11, November 1998 Table 4. Nerve Fiber Layer Parameter Values: Mean ⫾ SD
Parameter
Normals (n ⴝ 235)
Glaucoma (n ⴝ 42)
Ocular Hypertension (n ⴝ 35)
Polar integral Superior integral Temporal integral Inferior integral Nasal integral Polar average Superior average Temporal average Inferior average Nasal average Superior maxima Inferior maxima Temporal median Nasal median Average without blood vessels Ellipse modulation Maximum modulation S/I ratio S/N ratio S/T ratio I/N ratio I/T ratio Intra-ellipse sector variability
0.568 ⫾ 0.107 0.220 ⫾ 0.42 0.050 ⫾ 0.016 0.221 ⫾ 0.046 0.077 ⫾ 0.019 68.42 ⫾ 11.78 78.80 ⫾ 14.20 44.25 ⫾ 13.49 78.04 ⫾ 14.56 49.17 ⫾ 10.63 93.28 ⫾ 15.98 90.79 ⫾ 16.92 42.32 ⫾ 12.45 44.54 ⫾ 10.91 81.81 ⫾ 13.50 98.77 ⫾ 15.70 59.03 ⫾ 11.86 1.04 ⫾ 0.16 2.16 ⫾ 0.40 2.33 ⫾ 0.59 2.09 ⫾ 0.35 2.25 ⫾ 0.51 30.32 ⫾ 5.40
0.497 ⫾ 0.107 0.174 ⫾ 0.041 0.051 ⫾ 0.0136 0.191 ⫾ 0.048 0.081 ⫾ 0.020 61.71 ⫾ 12.66 64.54 ⫾ 15.35 46.67 ⫾ 12.81 69.59 ⫾ 15.29 53.08 ⫾ 11.70 73.54 ⫾ 18.45 80.08 ⫾ 18.64 43.12 ⫾ 12.06 49.91 ⫾ 12.19 77.42 ⫾ 14.40 74.95 ⫾ 21.19 42.78 ⫾ 16.07 0.93 ⫾ 0.18 1.51 ⫾ 0.38 1.78 ⫾ 0.49 1.64 ⫾ 0.37 1.93 ⫾ 0.49 20.32 ⫾ 6.33
0.566 ⫾ 0.133 0.204 ⫾ 0.055 0.053 ⫾ 0.015 0.222 ⫾ 0.051 0.087 ⫾ 0.025 72.37 ⫾ 14.73 77.19 ⫾ 18.55 51.13 ⫾ 13.94 83.16 ⫾ 16.25 58.75 ⫾ 14.15 87.54 ⫾ 19.71 93.91 ⫾ 17.88 48.24 ⫾ 14.19 53.30 ⫾ 13.66 86.30 ⫾ 17.05 93.17 ⫾ 22.76 52.62 ⫾ 12.64 0.94 ⫾ 0.14 1.67 ⫾ 0.32 1.89 ⫾ 0.45 1.80 ⫾ ⫺.31 2.04 ⫾ 0.47 26.96 ⫾ 6.61
Glaucoma Suspects Based upon Disc Appearance (n ⴝ 17) 0.550 ⫾ 0.116 0.193 ⫾ 0.043 0.054 ⫾ 0.014 0.219 ⫾ 0.054 0.084 ⫾ 0.021 65.20 ⫾ 11.74 67.78 ⫾ 13.27 47.70 ⫾ 10.46 76.17 ⫾ 16.81 52.81 ⫾ 10.89 77.42 ⫾ 15.27 86.78 ⫾ 18.68 44.77 ⫾ 10.05 47.40 ⫾ 10.89 79.19 ⫾ 12.87 89.88 ⫾ 20.62 48.51 ⫾ 13.88 0.91 ⫾ 0.17 1.66 ⫾ 0.27 1.80 ⫾ 0.50 1.87 ⫾ 0.40 2.00 ⫾ 0.50 25.71 ⫾ 6.52
I ⫽ inferior; S ⫽ superior; N ⫽ nasal; T ⫽ temporal; SD ⫽ standard deviation.
differently around the nerve head than in normal subjects. The average thickness assessment across the image after removal of the blood vessels (average without blood vessels) showed a statistically significant difference between patients with ocular hypertension and patients with glaucoma (higher in OHTN), but it did not differentiate normal from glaucoma. (Note that there is no parameter indicating the average thickness for the entire image without blood vessel removal.) In this study, the ellipse measuring 1.75 disc diameters in size and centered on the optic nerve head was divided into 32 equal sectors and the average nerve fiber layer thickness assessment value determined within each sector. Figure 1 shows the average thickness assessment value per sector plotted against location along the ellipse for each of the groups. The typical “double-hump” pattern of distribution of thickness around the ellipse is preserved in each group, but the peaks are lowered considerably in the glaucoma group, indicative of loss of superior and inferior fibers. Indeed, sectors 5 through 13 (superior) and sectors 21 through 27 (inferior) for the glaucoma group all show differences that are statistically significantly lower than those of normal with P values of less than 0.01. None of these sectors showed statistically significantly lower values from normal in the OHTN group (P ⬍ 0.05), although there were sectors in the GS-disc group with lower values than those of the normal subjects: sectors 6 to 8 (P ⬍ 0.01) and sectors 9 and 10 (P ⬍ 0.05). These differences in the GS-disc group were only in superior sectors. The thinnest values for the nasal quadrant were observed in the normal group, and the difference from the normal in both the OHTN and the glaucoma groups (but not the GS-disc group) was statistically significant. Figure 2 summarizes the probability values for all of the observed differences between the groups. Figure 3 shows the mean intra-ellipse sector variability and the 95% confidence intervals for each group, and Table 6 summarizes the differences between the groups. The OHTN and GS-disc
2072
groups appear to be similar to each other, but this parameter differentiates the two suspect groups and the patients with glaucoma from the normal subjects.
Discussion This study examined characteristics of the normal nerve fiber layer as determined by scanning laser polarimetry (or nerve fiber analysis) as performed by the NFA II and compared the results to age-matched patients with glaucoma or those suspected of having glaucoma based on elevated intraocular pressure (OHTN) or a suspicious-appearing disc (GS-disc). It also attempted to identify the relationship between aging and changes in the nerve fiber layer. Although some parameters showed weak correlation with age that were statistically significant, none would be considered clinically significant. Prior studies have shown similar correlation between nerve fiber layer thickness assessments and age.4 However, the normal population analyzed in this study is actually a subset of all of the normative data collected. Failure to show a negative correlation between age and thickness values as demonstrated by Chi and coworkers4 may reflect that most patients younger than 40 years of age were not included in analysis to age match the normal group to the other groups. In this study, scanning laser polarimetry with the NFA II was capable of distinguishing normal subjects from those with glaucoma by a number of parameters. Weinreb et al5 showed that nerve fiber layer thickness assessments as determined by scanning laser polarimetry were correlated with
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects Table 5. Statistical Significance of Observed Differences P* for Difference between Normal and:
Parameter
P for ANOVA (all groups)
Glaucoma
Ocular Hypertension
Glaucoma Suspects Based upon Disc Appearance
Polar integral Superior integral Temporal integral Inferior integral Nasal integral Polar average Superior average Temporal average Inferior average Nasal average Superior maxima Inferior maxima Temporal median Nasal median Average without blood vessels Ellipse modulation Maximum modulation S/I ratio S/N ratio S/T ratio I/N ratio I/T ratio Intra-ellipse sector variability
0.002 0.000 0.504 0.003 0.028 0.001 0.000 0.029 0.001 0.000 0.000 0.001 0.070 0.001 0.043† 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.001‡ 0.000‡ 1.000 0.001‡ 1.000 0.007‡ 0.000‡ 1.000 0.005‡ 0.228 0.000‡ 0.002‡ 1.000 0.031§ 0.372 0.000‡ 0.000‡ 0.000‡ 0.000‡ 0.000‡ 0.000‡ 0.001‡ 0.000‡
1.000 0.252 1.000 1.000 0.04§ 0.456 1.000 0.028§ 0.358 0.000§ 0.352 1.000 0.056 0.000§ 0.467 0.478 0.033‡ 0.003‡ 0.000‡ 0.000‡ 0.000‡ 0.135 0.008‡
1.000 0.082 1.000 1.000 1.000 1.000 0.020‡ 1.000 1.000 1.000 0.001‡ 1.000 1.000 1.000 1.000 0.270 0.006‡ 0.009‡ 0.000‡ 0.001‡ 0.079 0.259 0.009‡
I ⫽ inferior; S ⫽ superior; N ⫽ nasal; T ⫽ temporal. * Derived from one way analysis of variance (ANOVA) with Bonferroni correction for multiple comparisons. † Statistically significantly higher in ocular hypertension compared with glaucoma but each group no different from normals. ‡ Significant difference with normal value higher. § Significant difference with normal value lower.
visual field loss in patients with glaucoma. Compared to visual field testing, nerve fiber layer analysis is faster, easier for the patient, completely objective, and has less variability because of patient factors that would be considered subjective. Therefore, if it can be shown that nerve fiber layer analysis can identify patients with glaucoma with a high degree of certainty and identify patients with normal visual fields as having evidence of structural damage to their nerve fiber layers, it will prove to be a useful tool in the diagnosis and management of glaucoma. Identifying patients with early damage offers the theoretic advantage of beginning therapy earlier in the hopes of preventing permanent visual loss. Some of the parameters studied appear to differentiate between not only normal subjects and those with glaucoma but between normal subjects and glaucoma suspects as well. For example, as illustrated in Figure 3, the IESV, an indirect measure of the shape of the nerve fiber layer along the ellipse (analogous to pattern standard deviation in visual field testing), clearly separates normal subjects from those with glaucoma, with the two glaucoma suspect groups in between. Because the normal nerve fiber layer is thickest in the superior and inferior bundles and thinner nasally and temporally, there will be a larger spread of measurements around the mean thickness estimates along the ellipse com-
pared to a group that has lost fibers from the superior or inferior bundles (or both), such as patients with glaucoma. The IESV would therefore decrease as tissue is lost, since there would be less variability in the measurements. Indeed, of all the parameters examined in this study, the IESV was the only one to differentiate not only normal subjects from the other groups but patients with glaucoma from the other groups as well (one-way ANOVA with Bonferroni correction for multiple comparisons). Even though the range of normal values is fairly large and there is some overlap in thickness values found in normal subjects and the other groups (based on the means and standard deviations), parameters that examine the shape of the distribution of thickness assessments around the optic nerve head, such as the IESV, may prove more valuable in differentiating between normal subjects and patients with glaucoma. One thickness assessment that may prove useful in differentiating the groups is the superior maxima, as illustrated in Figure 4. This parameter, which is the average of the 1500 thickest points in the superior bundle, shows a clear delineation between normal subjects and glaucoma patients, with no overlap between the 95% confidence intervals. Parameters such as this one may prove to be valuable in screening populations at risk for glaucoma. For example, a superior maxima value of less than 80 may be highly
2073
Ophthalmology Volume 105, Number 11, November 1998
Figure 1. Average thickness per sector around ellipse.
unlikely to be found in normal subjects, and a value greater than 90 may be highly unlikely to be found in patients with glaucoma. This significance or predictive ability of values between 80 and 90 is not known at this time. Further studies are needed to determine the usefulness of NFA II assessments for screening.
The group of patients studied who were suspected of having glaucoma based on the appearance of their discs (GS-disc), although small, raises some interesting questions. The intraocular pressure and visual fields of the subjects in this group were normal, but many parameters of their nerve fiber layers were outside of the normal
Figure 2. Significance of observed differences between groups for ellipse sector.
2074
Choplin et al 䡠 NFA in Normal Subjects and Glaucoma Suspects
Figure 3. Intra-ellipse sector variability, mean, and 95% confidence interval.
range. For example, the superior integral (a measure of the nerve fiber layer volume), the average thickness assessment in the superior quadrant, and the superior maxima (the average of the 1500 points with the highest assessment values in the superior bundle), the maximum modulation (difference between highest and lowest values), and the IESV (another measure of the difference between highest and lowest) all showed statistically significant decreases from the normal values. Does this mean that this group of patients has lost fibers from the superior bundle (like “low-tension glaucoma”), or are their nerve fiber layers anatomically different (i.e., the fibers are distributed differently around the nerve head) from normal? The argument for a different distribution of Table 6. Intra-ellipse Sector Variability Group
Value*
Normal OHTN
30.32 ⫾ 5.40 26.96 ⫾ 6.61
GS-disc
25.71 ⫾ 6.52
Glaucoma
20.32 ⫾ 6.33
Differences† From From From From From From
normal: 3.36 ⫾ 1.01 GS-disc: 1.25 ⫾ 1.95 glaucoma: 6.64 ⫾ 1.48 normal: 4.61 ⫾ 1.38 glaucoma: 5.39 ⫾ 1.84 normal 10.00 ⫾ 0.93
* Microns, mean ⫾ standard deviation. † Mean ⫾ standard error of the mean.
P value 0.00 0.52 0.00 0.00 0.01 0.00
fibers around the optic nerve is partially supported by the fact that the polar integral and the polar average were not statistically significantly different from normal in this group, suggesting that the total nerve fiber layer in this group is not different from that in normal subjects. Certainly, these patients bear careful follow-up over time to look for changes, since otherwise healthy individuals with physiologic cupping may represent a variant of normal. Remember, the assessments of nerve fiber layer thickness as determined by the NFA II have not yet been shown to correlate with actual known nerve fiber layer thickness measurements as determined from anatomic study. Thickness assessments obtained with this instrument are calculated based on the measurement of retardation, or phase shift of the incident polarized laser light due to the birefringent property of the nerve fiber layer. The reported thickness assessments as expressed in micrometers are based on studies in two cynomolgus monkeys, in which 1° of retardation corresponded to 7.4 m of thickness.2 The actual relationship between retardation and nerve fiber layer thickness in humans is not known but for clinical purposes probably is not important. This study establishes the differences between different groups of clinical subjects using the NFA II and shows that normal subjects measure differently than patients with glaucoma. With further study, it may become possible to select a parameter or number of parameters that
2075
Ophthalmology Volume 105, Number 11, November 1998
Figure 4. Superior maxima (average of 1500 thickest points in superior bundle), mean, and 95% confidence interval.
will identify patients with glaucoma or those at risk of having glaucoma develop with a high degree of specificity and sensitivity. If this proves to be the case, then not knowing whether the assessment is equal to the precise anatomic thickness is probably of little significance. In conclusion, the following items can be determined: 1. Multiple parameters appear capable of differentiating normal subjects from patients with glaucoma, and many patients currently considered to be glaucoma suspects (with normal visual fields) have intermediate values between the normal subjects and the patients with glaucoma. It is not known at this time whether these parameter values are predictive of further progression to glaucoma. 2. Some parameters, such as the superior maxima and the intra-ellipse sector variability, seem to differentiate between normal, glaucoma, and the two glaucoma suspect groups (OHTN and GS-disc) and may prove useful as a screening tool. 3. Future studies of nerve fiber layer analysis are required to determine the usefulness of these parameters for glaucoma screening in the general population, the predictability of the development of glaucoma in patients with abnormal parameter values and normal visual fields, the applicability of NFA II assessments to other optic neuropathies, and the differentiation of
2076
glaucoma from other optic neuropathies or diseases of the retinal nerve fiber layer. Acknowledgments. The authors thank the following Nerve Fiber Analyzer users who contributed to the normative database that was used in this study: E. Randy Craven, MD, Hans Lemij, MD, Richard Mills, MD, Robert Ritch, MD, and Harry Quigley, MD.
References 1. Dreher AW, Reiter K, Weinreb RN. Spatially resolved birefringence of the retinal nerve-fiber layer assessed with a retinal laser ellipsometer. Applied Optics 1992;31:3730 –5. 2. Weinreb RN, Dreher AW, Coleman A, et al. Histopathologic validation of Fourier-ellipsometry measurements of retinal nerve fiber layer thickness. Arch Ophthalmol 1990;108:557– 60. 3. Tjon-Fo-Sang MJ, de Vries J, Lemij HG. Measurement by nerve fiber analyzer of retinal nerve fiber layer thickness in normal subjects and patients with ocular hypertension. Am J Ophthalmol 1996;122:220 –7. 4. Chi Q, Tomiat G, Inazumi K, et al. Evaluation of the effect of aging on the retinal nerve fiber layer thickness using scanning laser polarimetry. J Glaucoma 1995;6:406 –13. 5. Weinreb RN, Shakiba S, Sample PA, et al. Association between quantitative nerve fiber layer measurement and visual field loss in glaucoma. Am J Ophthalmol 1995;120:732– 8.