Boehm MD, Nedrud C, Greenfield DS, Chen PP. Scanning Laser Polarimetry and Detection of Progression After Optic Disc Hemorrhage in Patients With Glaucoma. Arch Ophthalmol. 2003;121(2):189-194. doi:10.1001/archopht.121.2.189
To examine retinal nerve fiber layer changes with scanning laser polarimetry(SLP) in the eyes of patients with glaucoma and optic disc hemorrhage.
Automated perimetry and SLP were performed in 17 eyes of 17 patients identified prospectively with optic disc hemorrhage. Criteria for visual field progression were based on decreased sensitivity seen at 3 adjacent points on the total deviation plot. Progression on SLP images was defined as a 15% or more decrease in the average thickness of the affected quadrant (superior or inferior), a 25% or more reduction in the affected quadrant ratio, an increase in the nerve fiber analyzer number of 10 or more (GDx Nerve Fiber Analyzer; Laser Diagnostic Technologies), or any change on Serial Analysis of the SLP images.
Main Outcome Measures
Visual field progression and SLP image progression.
The mean follow-up was 31 months (minimum, 12 months). Of the 17 eyes, 10 (59%) had visual field progression. No significant change was seen on SLP images for either the total group or the group with visual field progression. Five eyes (29%) showed progression on SLP images, 3 of which also showed visual field progression. Ten eyes showed progression on SLP images that was not confirmed on subsequent imaging.
In eyes with visual field progression after optic disc hemorrhage, a significant change in the SLP image was not seen. Fluctuation of SLP results in patients with glaucoma necessitates confirmation of progression seen on SLP images.
GLAUCOMA IS frequently characterized by progressive changes in the optic nerve, such as enlargement of the optic nerve cup or notching of the optic nerve rim, or visual field loss. However, several researchers1- 5 have shown that defects in the peripapillary retinal nerve fiber layer (RNFL) can precede changes in optic nerve head appearance and visual field loss. Scanning laser polarimetry (SLP) uses the birefringent properties of the RNFL to estimate its thickness.6 Scanning laser polarimetry is performed using an analyzer (GDx Nerve Fiber Analyzer; Laser Diagnostic Technologies) that is composed of a near-infrared diode scanning laser ophthalmoscope with a polarization modulator, a corneal polarization compensator, and a polarization detection unit. The change detected in the polarization state of light as it passes through the birefringent RNFL correlates with the RNFL thickness.7 Although diagnosis of glaucoma using the nerve fiber analyzer without corneal compensation may be problematic, 8- 12 several studies13- 19 have shown that mean SLP values are significantly lower in groups of patients with glaucoma than in control subjects without glaucoma and that SLP values correlate significantly with visual field loss. Some researchers20,21 have reported the use of SLP in the detection of RNFL thickness reduction over time in patients with ischemic or traumatic optic neuropathy. Although the mechanism of damage in glaucoma differs from those optic neuropathies, longitudinal measurement with SLP could potentially provide objective evidence of glaucomatous progression at an earlier stage of damage than visual field testing, and could provide an alternative or adjunctive means of following up patients with glaucoma.
Optic disc hemorrhage is associated with localized defects in the optic disc and RNFL, and with progression of RNFL defects, visual field defects, and optic nerve damage from glaucoma.3,22- 31 We examined RNFL thickness in patients with glaucoma, using SLP prospectively after optic disc hemorrhage, to assess the ability of SLP to detect changes in the RNFL due to glaucoma.
Patients with an optic disc hemorrhage were prospectively identified from the outpatient clinics of glaucoma subspecialists at the University of Washington and the Bascom Palmer Eye Institute, University of Miami. Institutional review board approval was granted at each institution for this study, and all patients gave informed consent to participate. An optic disc hemorrhage was defined as an isolated hemorrhage located on the rim of the optic disc or within it. Exclusion criteria included a visual acuity of worse than 20/40, a media opacity that prevented SLP imaging, the inability to perform visual field testing or to cooperate with SLP imaging, or systemic or ocular disease other than glaucoma that might alter the results of visual field testing or SLP imaging. All subjects had been diagnosed as having glaucoma before disc hemorrhage, and had optic disc findings (vertical cup-disc ratio, ≥0.7; and/or rim thinning or notching) and/or visual field loss (at least 1 of the following: an abnormal glaucoma hemifield test result; pattern SD at P<.05; or 3 points on the total deviation plot at P<.05, with at least 1 point at P<.01) consistent with the diagnosis of glaucoma.32 In all patients, the optic disc hemorrhage was considered to be associated only with glaucoma. Patients with a disc hemorrhage were identified between May 1, 1998, and April 30, 1999.
Imaging using SLP (without variable corneal compensation, which was commercially unavailable during this study) was performed by trained experienced technicians at the 2 study sites with undilated pupils and under ambient lighting. A mean image was created from 3 images produced with the nerve fiber analyzer during each study visit. Each separate image was given a "pass" quality rating by the nerve fiber analyzer software, using the manufacturer's standard preset criteria. Automated static threshold perimetry was performed (Humphrey Field Analyzer II; Zeiss-Humphrey, Dublin, Calif). All Humphrey visual field results had good reliability, 32 and were either a central 24-2 or a 30-2 threshold test with a size III white stimulus. After identification, patients underwent SLP and visual field testing at approximately 6- to 12-month intervals. Patients were tested using 1 of 2 algorithms (standard or full Swedish Interactive Threshold Algorithm), but not both, with 1 exception.
The location of the disc hemorrhage on the optic nerve head was categorized as either superior or inferior. The superior quadrant was defined as the standard superior 120° on the nerve fiber analyzer topographic map, and the inferior quadrant was defined as the inferior 120°. All study disc hemorrhages fell within these 2 quadrants. (The standard temporal and nasal sectors are 50° and 70°, respectively.) The location of the disc hemorrhage was used to define the affected quadrant for study purposes, and the opposite quadrant was defined as the unaffected quadrant. For example, if a disc hemorrhage was observed at the 7-o'clock location of the optic disc, the inferior quadrant was the affected quadrant and the superior quadrant was the unaffected quadrant.
Specific nerve fiber analyzer parameters investigated included the following: superior and inferior averages, which are the average SLP thicknesses of the nerve fiber layer from those respective quadrants; superior and inferior ratios, which correspond to the ratio of the average of the 1500 thickest points in the respective quadrant divided by the average of the 1500 thickest points in the temporal quadrant; deviation of the quadrant from normal on the nerve fiber analyzer printout; and the number, which is a value assigned to each nerve fiber analyzer analysis, ranging from 0 (normal) to 100 (advanced glaucoma), and is derived from a proprietary algorithm. The nerve fiber analyzer also allows for different images to be compared using Serial Analysis, in which one image is designated a reference scan to which subsequent images can be compared. Serial Analysis was performed for each study eye, for which the initial SLP mean image was designated the reference scan. These parameters and the Serial Analysis were chosen because they were accessible to physicians using the nerve fiber analyzer in clinical practice and they were considered the most likely to show changes specifically related to the disc hemorrhage. Because eyes with a disc hemorrhage may show visual field progression that is not in concordance with the site of the hemorrhage, 28,29,31 SLP data were collected for the affected and the opposite, unaffected, quadrant.
Data were entered into a computer spreadsheet program (Statistical Product and Service Solutions 10.1 for Macintosh; SPSS Inc, Chicago, Ill), and statistical analysis was performed using the t test (2-tailed, paired and unpaired). Results are given as mean ± SD, where applicable.
Visual field progression during the study period was determined using criteria modified from previous researchers32- 34;3 contiguous points must have at least a 5-d B decrease from baseline on the total deviation plot, with 1 of these points having at least a 10-d B decrease. In addition, visual field progression was also dependent on the occurrence of progression in the proper hemifield, corresponding to the location of the disc hemorrhage in the opposite quadrant of the optic nerve.
Progression on SLP images was defined as a decrease of 15% or more in the average thickness of the affected quadrant, a decrease in the affected quadrant ratio of 25% or more, or an increase in the nerve fiber analyzer number of at least 10 during follow-up. These criteria were based on the findings of a longitudinal study of 27 eyes of 16 subjects without glaucoma or other ocular disease (visual acuity, 20/25 or better; intraocular pressure, ≤21mmHg; cup-disc ratio, ≤0.5; age, 45 ± 11 years; and refraction, −2.7 ± 2.7 diopters) performed at the University of Washington, in which fewer than 5% of healthy eyes showed changes of this magnitude for these parameters on subsequent evaluation and testing after 41.0 ± 4.6 months of follow-up (Table 1). The conditions and technicians used to obtain SLP images in the healthy subjects were identical to those used to obtain SLP images in the study population. These criteria are consistent with other published data16,35- 38 on the short-term reproducibility of SLP measurements in healthy patients and in patients with glaucoma.
For the Serial Analysis, each subsequent mean image was compared with the reference scan and examined for evidence of change (preset minimum of 20 µm of change) in the RNFL thickness at the site of the optic disc hemorrhage (within ±0.5 clock hours). Such change is indicated on the Serial Analysis printout as an area of different color, corresponding to the magnitude of the RNFL thickness change from the reference scan. If negative change occurred (ie, thinning), this was considered indicative of progression of RNFL loss. Subsequent nerve fiber analyzer scans were examined for similar change in the same area.
Deviation from normal in the affected and unaffected quadrant, which is provided on the nerve fiber analyzer printout and is based on a normative database, was analyzed, but no criteria for progression were based on this variable.
Seventeen eyes with an optic disc hemorrhage of 17 patients (15 women; all white) with glaucoma were recruited for this study. The patients' age at recruitment was 71.2 ± 7.1 years. The initial characteristics of the group studied are listed in Table 2. Glaucoma diagnoses included primary open-angle glaucoma (n = 8), normal-tension glaucoma (n = 6), pseudoexfoliative glaucoma (n = 2), and pigmentary glaucoma (n = 1). Six eyes were pseudophakic, and no eyes underwent cataract surgery during follow-up.39 The study disc hemorrhage was observed in 8 right and 9 left eyes. There were 5 disc hemorrhages located superiorly and 12 inferiorly. The time between the observation of the optic disc hemorrhage and initial SLP and visual field testing was 0.4± 1.0 and 2.6 ± 2.6 months, respectively. The follow-up (minimum, 12 months) was 30.5 ± 8.0 months for visual field testing and 30.7± 9.0 months for SLP. Table 2 and Table 3 show changes in visual field and SLP variables for the whole group of 17 eyes. No significant differences were found, except for mean deviation (P = .001).
Of the 17 eyes, 10 (59%) had visual field progression by modified Anderson criteria32 during the study period, which was confirmed with subsequent testing in 9 eyes (1 eye did not undergo further testing). In 1 other eye, the criteria for progression were fulfilled, but the visual field reverted to a nonprogressed state on subsequent testing. The time to progression was 17.2 ± 9.5 months. Eyes with progression had a significantly larger initial pattern SD (8.5 ± 3.3 d B) compared with those without progression (4.5 ± 2.9 d B) (P = .02); these eyes also had significant worsening of the mean deviation(P = .005) when compared with eyes without progression(Table 4). No significant difference(P>.05) was found between the groups with and without visual field progression in any of the SLP parameters studied (Table 4), patient age, glaucoma diagnosis, follow-up time, or time to initial visual field or SLP testing (data not shown).
Some parameters showed a decrease in the average measurement in the affected quadrant in the eyes with visual field progression compared with those without progression, including change in the average thickness of the affected quadrant (−4.0 ± 7.0 vs −1.6 ± 5.3 µm; P = .45), but these differences were not statistically significant. Within the group with visual field progression, no significant difference was found for the change in value of any of the SLP parameters studied (Table 5). Of 10 eyes with visual field progression, 3 (30%) also had progression on SLP images(confirmed on subsequent imaging in all 3 patients).
Of the 17 eyes, 5 (29%) showed progression on SLP images after disc hemorrhage, in 4 by an increase of the nerve fiber analyzer number by 10 or more, and in 3 by other criteria (reduction of the affected quadrant ratio by ≥25% in 2 eyes and reduction of mean quadrant thickness by 15% in 1 eye). Some eyes showed progression by more than 1 criterion. The time to progression on SLP images was 16.4 ± 9.4 months. All 5 eyes had progression confirmed on subsequent imaging. In 10 other eyes, progression on SLP images was noted at some point during follow-up, but reverted on subsequent testing. In 9 eyes, this was by increase of the nerve fiber analyzer number by 10 or more; in 3, by 15% or more decrease in the average thickness of the affected quadrant; and in 2, by 25% or more decrease in the affected quadrant ratio parameter(3 eyes showed variability of >1 criterion during follow-up).
On the nerve fiber analyzer's Serial Analysis, 3 eyes showed localized decreases of at least 20 µm in RNFL thickness during follow-up at the location of the disc hemorrhage or within a half clock hour of it. One of these eyes showed changes on a single mean image that were not confirmed by later SLP imaging. Both of the other 2 eyes showed focal thinning on Serial Analysis and an increase in the nerve fiber analyzer number at 13 months after disc hemorrhage. These changes were confirmed with subsequent imaging. Both eyes also had visual field progression, noted at approximately the same time as SLP imaging (15 months).
In most patients, observation of the disc hemorrhage resulted in initiation of additional treatment to lower the intraocular pressure. Four eyes underwent trabeculectomy during follow-up. In none of these eyes was an improvement noted postoperatively in the SLP variables studied, 40 although 1 eye had an improvement in visual field indexes after surgery.
In this study, 17 eyes with glaucoma were followed up for an average of 31 months after optic disc hemorrhage to assess quantitative changes in the RNFL, based on SLP imaging. Although significant change was seen in visual field mean deviation for the group of 10 eyes with visual field progression(the criteria for which did not include changes in visual field indexes), no SLP parameter investigated demonstrated a statistically significant change in the whole group analysis, in the comparison of groups with and without visual field progression, or in the analysis of the group with visual field progression alone. Of 10 eyes with visual field progression, 3 showed progression on SLP images using the criteria selected.
One other study has measured SLP variables during a comparable follow-up in patients with glaucoma. Poinoosawmy et al41 compared RNFL thickness in 75 eyes of patients with normal-pressure glaucoma with that of 35 healthy control subjects for 2 years, and found a significant difference in decrease in RNFL thickness when comparing the median values (8% vs 2.4%) using the Wilcoxon rank sum test. However, no attempt was made to discern individuals who had shown progression on visual fields from those who had not. In the present study, an analysis of the average thickness in the affected quadrant in the 10 eyes with visual field progression showed a mean decrease of 4.0 µm (−5.9%), compared with a mean decrease of 1.6 µm(−1.8%) in the 7 eyes without visual field progression, but this difference was not significant. Average thickness in the affected and unaffected quadrant for all 17 eyes had mean decreases of 3.0 and 1.3 µm (−4.2% and−1.4%), respectively.
Perhaps more important than the group data are the individual eye data, in which SLP did not show changes after optic disc hemorrhage while standard visual field testing showed progression more frequently, using the criteria selected. We investigated all the parameters included on the standard nerve fiber analyzer printout that might show change related to the disc hemorrhage. Admittedly, the effect of a disc hemorrhage may be only on a small specific area of the peripapillary RNFL, and the relatively large size of the quadrant studied (120°) may have masked the small area of affected RNFL damaged after the disc hemorrhage. Parameters such as the quadrant ratios use the thickest points in the quadrant, which may not be where the disc hemorrhage occurs. In addition, SLP is relatively insensitive to focal defects in the visual field.42,43 For these reasons, we also studied Serial Analysis results, which would be expected to be more sensitive to localized changes in the RNFL. However, Serial Analysis detected only 2 confirmed cases of progression, both of which were noted at essentially the same time on visual field testing.
The variable most commonly observed to exhibit change in eyes with SLP progression was the nerve fiber analyzer number, but high variability prevented reliable determination of progression. Confirmation of SLP changes at a subsequent imaging session during the follow-up of patients with glaucoma seems to be of utmost importance. The variability of SLP values was generally greater for this population of elderly patients with glaucoma than for eyes of younger healthy subjects (from which we derived the criteria for progression on SLP images), and higher than the 95% limits of agreement that other researchers37 have reported with the nerve fiber analyzer for healthy eyes or eyes with glaucoma. Other studies35,36,38 have found less variability in small selected populations examined expressly to determine variability with SLP.
The variability found in this study made determination of progression on SLP images difficult. Systematic sources of error in imaging, such as inconsistent alignment, were unlikely to occur in our study, because of the standard conditions used during imaging (trained experienced technicians, undilated pupils, and ambient light). Others have also found higher variability for some derived SLP parameters (eg, the nerve fiber analyzer number) in patients with glaucoma compared with healthy patients37 or a trend toward higher variability in older patients with glaucoma compared with younger healthy patients.44 Changing the criteria for progression on SLP images, to allow for greater sensitivity in the detection of progression seen on visual field testing, would only be at the expense of specificity because of higher rates of reversion on subsequent testing. Further investigation may better define which nerve fiber analyzer parameters, or combination of parameters, are most useful to quantitatively demonstrate significant change in the RNFL over time. The evaluation of SLP measurements with other forms of analysis may provide greater sensitivity for localized changes.45
Two eyes showed progression on the nerve fiber analyzer but not on visual field testing. Some researchers25 have reported that visual field defect progression was concurrent with observation of a disc hemorrhage, while others3,46 have noted an average latency of many years. However, 2 studies28,31 with criteria similar to those used in the present study for visual field progression have reported average latencies of between 16 and 20 months, which would indicate that the mean follow-up after disc hemorrhage in the present study (31 months) is sufficient to document visual field progression associated with optic disc hemorrhage. In addition, the proportion of eyes with visual field progression in the present study (10 [59%] of 17) is consistent with that reported in previous studies25,27- 29,31 during the same period. Only further longitudinal follow-up will determine if those eyes with progression on SLP images alone will subsequently show progression of the visual field.
It is possible that disc hemorrhages may have occurred in areas of preexisting RNFL loss and, therefore, visual field changes seen may have been delayed and unrelated to the disc hemorrhage. Further investigation may be needed to explore whether changes in the RNFL may occur before the disc hemorrhage, which was not addressed in this study. However, several researchers3,27,28 have shown the RNFL to be the initial site of visible damage after optic disc hemorrhage, followed by changes in the visual field or optic nerve appearance, which would argue that any changes in the RNFL associated with the disc hemorrhage should have been seen during follow-up among the eyes that showed visual field progression. One prospective study47 showed that 52 (81%) of 64 disc hemorrhages in patients with normal-tension glaucoma occurred either at the border of an existing RNFL defect (33%), in the healthy RNFL adjacent to an existing RNFL defect (28%), or in an apparently uninvolved area of the RNFL (20%), and that only 17% occurred within an existing RNFL defect. A later study30 confirmed these findings, and found no difference between those with normal-tension glaucoma and those with primary open-angle glaucoma. These findings would indicate that, although changes in the RNFL may occur before disc hemorrhage, most hemorrhages occur in relatively unaffected areas of the RNFL and should theoretically result in new and potentially detectable changes to the RNFL. In this study, no attempt was made to document, by ophthalmoscopy or photographs, the presence of preexisting RNFL defects at the site of disc hemorrhages.
Eyes in this study were not imaged with variable corneal compensation, which has been shown to improve the performance of SLP in the diagnosis of glaucoma.48 The effect of variable corneal compensation on the longitudinal follow-up of patients with glaucoma remains to be seen.
This study has several shortcomings, including the relatively small number of patients enrolled, the relatively short follow-up in relation to the duration of disease, and the nonstandardized method and timing of visual field testing and SLP imaging. No control group of patients with glaucoma without disc hemorrhage was included in our study design. Nevertheless, in this study of SLP after optic disc hemorrhage in patients with glaucoma, 10 of 17 eyes showed progression on visual field testing after an average of 31 months of follow-up; 3 of these eyes showed progression on SLP images using the criteria selected. No significant change was seen in the SLP parameters studied for the whole group, nor was significant change seen in the group of eyes with visual field progression, either within the group or compared with the group without visual field progression. High variability was seen in the SLP results in this group of patients with glaucoma, which made determination of progression difficult and necessitates confirmation of progression on SLP images, just as confirmation of progression is necessary in visual field testing. Scanning laser polarimetry is a relatively young technology, and may prove to be useful for monitoring RNFL changes over time in patients with glaucoma, but further investigation is necessary to identify or develop the parameters that may be best suited to this purpose.
Corresponding author: Philip P. Chen, MD, Department of Ophthalmology, University of Washington, Campus Box 356485, 1959 NE Pacific St, Seattle, WA 98195-6485 (e-mail: firstname.lastname@example.org).
Submitted for publication February 6, 2002; final revision received October 7, 2002; accepted October 23, 2002.
This study was supported in part by departmental grants from Research to Prevent Blindness, Inc, New York, NY; and by grant R01-EY08684 from the National Institutes of Health, Bethesda, Md (Dr Greenfield).
Reprints not available from the authors.