Scanning Laser Polarimetry of Edematous and Atrophic Optic Nerve Heads

Objective  To determine if scanning laser polarimetry (SLP) measures form birefringence of the retinal nerve fiber layer (RNFL).

Methods  Consecutive patients with either acute unilateral disc edema or chronic diffuse unilateral disc atrophy underwent SLP using the GDx Nerve Fiber Analyzer(Laser Diagnostic Technologies Inc, San Diego, Calif). The former group had peripapillary RNFL edema, presumably with no change in form birefringence elements, while the latter had optic nerve atrophy, presumably with a loss of birefringence elements in the RNFL. A subset of patients with acute unilateral disc edema who subsequently developed disc atrophy had repeated SLP at 6 months. Intereye and intertest comparisons of 6 SLP parameters representative of RNFL thickness were performed using the paired t test.

Results  In the acute unilateral disc edema group (n = 28), none of the SLP parameters were significantly increased in affected vs fellow eyes. In the chronic unilateral disc atrophy group (n = 30), all SLP parameters were significantly decreased in affected vs fellow eyes (P<.001). Patients with disc edema who had a follow-up SLP demonstrated significant declines of all parameters in the affected eyes (P<.007) but no change in SLP parameters in unaffected eyes.

Conclusion  Scanning laser polarimetry measures form birefringence properties of the RNFL, but not necessarily the RNFL thickness.

SCANNING LASER polarimetry (SLP) is believed to assess form birefringence properties of the ganglion cell axons of the retinal nerve fiber layer (RNFL). A medium consisting of parallel cylindrical rods with diameters considerably smaller than the wavelength of the light (ie, similar to the axons of the RNFL) behaves like a form birefringent medium.¹ A birefringent medium changes the state of polarization of light that passes through it. The amount of polarization, which correlates with the "retardation" of reflected light, is proportional to the thickness of the birefringent medium. Hence, by measuring the retardation of the light beam that passes through such a medium, one can determine the medium's thickness. Weinreb et al² showed a correlation between retardation of reflected light as determined by Fourier ellipsometry and histopathologic measurement of the RNFL thickness in the nonhuman primate eye. Within the RNFL, axonal cell membranes, microtubules, neurofilaments, and mitochondria may all contribute to birefringence. In vitro experimentation confirms that microtubules of retinal ganglion cell axons dominate RNFL birefringence.³

The utility of SLP in glaucoma is well documented. Eyes with glaucoma demonstrate a decrease in retardation of reflected light in the superior and inferior peripapillary regions compared with normal eyes.^4-6 Fifty-eight percent of patients with ocular hypertension had an abnormal RNFL parameter compared with normal eyes, ⁷ which suggests that SLP may be useful in identifying those at risk for glaucoma. Cases of"suspect" glaucoma (ie, those patients who have suspicious-appearing optic nerve heads or ocular hypertension) can be differentiated from normal subjects by assessing the variability of RNFL thickness in an ellipse around the optic nerve.⁸ Scanning laser polarimetry parameters correlate with the mean defect on Humphrey Visual Field Analyzer (Humphrey Systems, San Diego, Calif) perimetry, ⁹ although the correlation may not be evident until some threshold amount of RNFL dropout occurs.¹⁰ Scanning laser polarimetry has limitations in discriminating between patients with various degrees of glaucoma and those with normal optic nerves.^11-13 Yet, SLP improves the predictive power to detect glaucoma and may ultimately have a role in population-based screening for the disease.¹⁴ Finally, longitudinal analysis reveals that SLP can detect progressive RNFL thinning in glaucoma.¹⁵

Patients with neuro-ophthalmic disorders present an opportunity to clinically test the assertion that SLP measures form birefringence of the RNFL as a surrogate of RNFL thickness. This assertion would logically hold if there were no elements other than form birefringence elements that contribute to RNFL thickness. For instance, histopathologic studies confirm that eyes with disc edema have increased thickness of the peripapillary RNFL.^16,17 The increased thickness is caused by intracytoplasmic swelling of ganglion cell axons and not by an increase in microtubules or neurofilaments that contribute to RNFL birefringence. Hence, we hypothesize that SLP measurements would not increase in patients with disc edema despite the increased thickness of the RNFL. On the other hand, eyes with optic atrophy have fewer ganglion cell axons¹⁸ whose microtubules and other structural proteins contribute to RNFL birefringence. Hence, we hypothesized that SLP measurements would be lower in patients with clinically evident optic nerve head atrophy. This study tests both hypotheses.

We identified a consecutive series of patients with either unilateral optic nerve head edema (group 1, n = 28) or atrophy (group 2, n = 30) in the neuro-ophthalmology service. Only 5 patients declined to participate. Most patients in groups 1 and 2 had either optic neuritis or nonarteritic anterior ischemic optic neuropathy (NAION). The following criteria were used by an experienced neuro-ophthalmologist (J.F.R.) to distinguish between these 2 diagnoses for patients who had acute unilateral disc edema. A diagnosis of NAION was made on clinical grounds alone using the following criteria as guidelines: acute or subacute monocular visual loss, age 45 years or older, absence of pain, presence of disc edema in the acute phase of visual loss, and no evidence of giant cell arteritis or multiple sclerosis.¹⁹ Nonarteritic anterior ischemic optic neuropathy was occasionally diagnosed in patients younger than 45 years if there was compelling evidence supporting the diagnosis and no substantive evidence to the contrary. A diagnosis of optic neuritis was made on clinical grounds alone using the following criteria as guidelines: acute or subacute monocular visual loss, age 20 to 45 years, retrobulbar pain or pain with eye movements, or history of multiple sclerosis.^20,21 While some overlap can exist between NAION and optic neuritis, ²² female sex and a central scotoma biased the diagnosis toward optic neuritis, and male sex and an altitudinal field defect biased the diagnosis toward NAION.²³ The fellow eye of study patients was felt to have no prior evidence of optic nerve disease on the basis of the following criteria: no history of visual disturbance, intact color vision, and no evidence of optic atrophy. Patients with evidence of glaucoma, ocular hypertension, bilateral nonglaucomatous optic nerve disease, retrobulbar optic neuropathy, optic nerve head drusen, elevated intracranial pressure, or known afferent visual pathway lesions were excluded from this study. The human studies committee at the Massachusetts Eye and Ear Infirmary (Boston) approved the protocol, and all patients signed informed consent prior to participation.

Each patient had a complete neuro-ophthalmic examination, including assessment of visual acuity, color vision (Ishihara pseudoisochromatic plates), pupillary reaction, slitlamp examination, applanation tonometry, Goldmann(Haag-Streit, Bern, Switzerland) or Humphrey 30-2 visual field testing, and dilated funduscopy.

All patients had RNFL analysis performed by the same person (N.J.C.) in both eyes using the GDx Nerve Fiber Analyzer (Laser Diagnostic Technologies Inc, San Diego, Calif), an SLP with an integrated polarization modulator, a fixed corneal compensator, and a polarization detection unit. The technique of SLP has been described in detail elsewhere.^4,24 In summary, the SLP provides an assessment of the RNFL using a near-infrared diode laser (780-nm wavelength) as a light source. The laser light reflecting back from the RNFL is split into 2 parallel reflected rays because of a change in the state of polarization of the light. The amount of phase shifting of the reflected light is referred to as the "retardation, " which is believed to correlate with the density of form birefringence elements in the RNFL. A total of 65 536 retinal points (256 × 256 pixels) in a 15° field centered on the optic nerve are measured. External fixation light was used to center the optic nerve head. We used the patient's hand-to-eye proprioception to externally fixate eyes with poor vision. Three images from each eye were obtained. Based on an operator-dependent outline of the margin of the optic disc, an ellipse is generated at a concentric location 1.75 times the diameter of the disc. Scanning laser polarimetry internal software version 1.0.16 (Laser Diagnostic Technologies Inc) was used to calculate the amount of retardation. Measurement of RNFL thickness was calculated by the software-generated conversion of degrees of retardation into micrometers of RNFL thickness.

In this study, we did not directly address the possibility that corneal birefringence could confound our findings. At the time of this study, an SLP with a variable corneal compensator was not available. To indirectly address this issue, we obtained a follow-up study on 10 consecutive patients in group 1 who were observed to have developed optic atrophy at examinations scheduled for medical reasons at least 6 months after their initial study. All patients agreed to the second test. Since corneal birefringence does not change over time, ²⁵ we removed the potentially confounding effect of corneal birefringence by performing more than 1 test in the same eye of the same patients.

We analyzed 6 of the SLP parameters that provide absolute measurements of the RNFL retardation. Superior maximum represents the average of the 1500 thickest pixels in the superior peripapillary quadrant. Inferior maximum is the average of the 1500 thickest pixels in the inferior peripapillary quadrant. Average thickness is calculated from all 65 536 pixels used to create an image. The ellipse average denotes the average thickness of the nerve fiber layer in a 10-pixel-wide band with a diameter that is 1.75 times the diameter of the optic nerve. Superior average and inferior average are the average thickness of the nerve fiber layer along the superior and inferior portion of the ellipse surrounding the optic nerve. These parameters are automatically generated by the SLP software and presented on the symmetry analysis printout. A comparison of these parameters between affected and unaffected eyes in each group of patients was performed with the paired t test. The t test was used for an intertest comparison in affected and unaffected eyes of the follow-up group. Intereye differences(fellow eye subtracted from affected eye) between subgroups of patients with acute unilateral disc edema and chronic unilateral disc atrophy were analyzed with a 2-way analysis of variance. Since 6 SLP parameters were assessed, a Bonferroni approach was used to establish P≤.008 for statistically significant differences between groups. Correlation between baseline vision and SLP parameters was performed with linear regression analysis.

The demographic characteristics as well as the patients' visual acuity at initial examination are presented in Table 1. We subdivided patients with acute unilateral optic nerve head edema (group 1) and those with chronic unilateral optic nerve head atrophy(group 2) into inflammatory (subgroup a) and ischemic (subgroup b) subgroups. Group 1a included patients with optic neuritis (n = 8) and neuroretinitis(n = 2). Group 1b contained patients with nonarteritic (n = 16) and arteritic(n = 2) anterior ischemic optic neuropathy. The range of time from the onset of symptoms to the initial examination for group 1 was 1 day to 5 weeks. Group 2a contained patients with optic neuritis (n = 10), whereas group 2b included patients with nonarteritic (n = 19) and arteritic (n = 1) anterior ischemic optic neuropathy. The range of the time from the onset of symptoms to the initial examination for group 2 was 4 weeks to 27 months. The range of vision for all subgroups was similar. There was no correlation between baseline vision and SLP parameters in affected and unaffected eyes for any group (data not shown).

Figure 1 shows a typical example of SLP findings from a patient with optic neuritis (group 1a). There is clinically evident disc edema in the right eye; yet, the retardation in the ellipse around the optic nerve is higher in the fellow eye, which incidentally demonstrates physiologic cupping. Table 2 presents the mean SLP parameters for all patients. In group 1, none of the SLP parameters were increased in eyes with acute unilateral disc edema compared with contralateral fellow eyes (Table 2). Four of 6 RNFL measurements were decreased in eyes with unilateral disc edema compared with contralateral unaffected eyes, but the differences were not significant after adjusting for multiple comparisons (.02≤P>.008). In group 2, all SLP parameters were significantly reduced in eyes with unilateral chronic optic atrophy compared with the fellow unaffected eyes (P<.001).

Mean SLP parameters for the inflammatory and ischemic subgroups of patients with acute unilateral disc edema and chronic unilateral disc atrophy are presented in Table 3 and Table 4. All 6 SLP measurements in group 1a (eyes with acute inflammatory optic nerve edema) and group 1b (eyes with acute ischemic optic nerve edema) were reduced compared with fellow eyes, but none of these differences was statistically significant after adjusting for multiple comparisons. (Table 3). Furthermore, comparing the intereye differences in SLP parameters between group 1a and group 1b revealed no significant trends (P>.5; data not shown). A comparison of SLP parameters of affected eyes compared with fellow eyes in group 2a (optic atrophy secondary to prior optic neuritis) was not statistically significant (Table 4). In contrast, all SLP parameters of the affected eyes of group 2b (optic atrophy secondary to prior ischemic optic neuropathy) were significantly reduced compared with unaffected eyes(Table 4). Furthermore, the intereye differences (fellow eye value subtracted from affected eye value) in superior average, superior maximum, inferior maximum, and inferior average between groups 2a and 2b were statistically significant (P≤.005; data not shown).

The prospectively followed group consisted of patients with optic neuritis(n = 4) and nonarteritic (n = 5) and arteritic (n = 1) anterior ischemic optic neuropathy who developed optic atrophy at 6 months. The average age for this group was 52 years (range, 27-73 years). The visual acuity of this group ranged from 20/25 to no light perception at initial examination and 20/20 to no light perception at the 6-month follow-up. At baseline (acutely), there were no differences in any of the SLP parameters between affected and unaffected eyes(P>.10). There was a statistically significant reduction in all SLP parameters in affected eyes during follow-up (Table 5). In contrast, all SLP parameters in the unaffected eyes were not significantly changed at follow-up. Furthermore, the intereye differences in superior average and superior maximum (SLP value at baseline subtracted from SLP value at follow-up) between affected and unaffected eyes were statistically significant (P<.008; data not shown).

Retinal nerve fiber layer measurements are recognized as useful tools in assessing glaucomatous optic neuropathy. A variety of nonglaucomatous disorders that affect the optic nerve can also produce RNFL changes. The 2 most frequent nonglaucomatous optic neuropathies are optic neuritis and NAION. We undertook this study of GDx Nerve Fiber Analyzer RNFL measurements to test 2 hypotheses:(1) that SLP measurements would not increase in patients with disc edema despite the increased thickness of the RNFL and (2) that SLP measurements would be lower in patients with clinically evident optic nerve head atrophy. Our study confirmed both hypotheses.

Scanning laser polarimetry measurements of RNFL thickness did not increase when the RNFL was swollen. Therefore, for patients with optic nerve edema, measurements of the form birefringent properties of the RNFL do not substitute for RNFL thickness.^2,26 With regard to the second hypothesis, we found that all SLP measurements were significantly lower in patients with optic atrophy, which is consistent with the known loss of axons that is present in atrophic nerves.

These findings are consistent with histologic and scientific evidence. Optic nerve and RNFL edema is not known to be associated with an increase in form birefringence elements. Neither histopathologic nor ultrastructural observations of other investigators have found an increase in neurofilaments or microtubules (structures felt to confer birefringence) in cases of human or experimental optic nerve edema. Rather, optic disc swelling results from obstruction of axonal transport in the region of the lamina cribrosa and is associated with axonal swelling and an accumulation of mitochondrial aggregates.^16,17,27,28 Axonal swelling, therefore, would not be expected to increase form birefringence of the RNFL. Theoretically, mitochondrial aggregation could confer form birefringence, but there is no experimental evidence to support this notion. A second assumption is that optic atrophy, a morphologic description that results from a variety of insults to the optic nerve, is associated with decreased thickness of the RNFL. There is strong histopathologic evidence to support this assertion.^29,30

In subgroup 2a (optic atrophy secondary to optic neuritis), the SLP parameters did not decrease significantly in the affected eye compared with the fellow eye even though the optic nerve appeared pale (Table 4). Perhaps the small number of eyes in this subgroup reduced our power to find a statistically significant decrease in SLP parameters. However, the statistically insignificant reduction of SLP parameters could be explained by the generally good visual prognosis enjoyed by patients with optic neuritis. Vision remains poor in only 20% of cases, presumably because of a loss of ganglion cell axons.³¹ By comparison, a larger series of patients with optic neuritis showed a high percentage of patients with abnormal SLP, although the study design and objectives were dissimilar to ours.³²

Ocular structures other than the RNFL, such as the lens and especially the cornea, exhibit birefringence.³³ In most eyes, the axis of corneal birefringence is 15° inferior and nasal. The SLP unit with fixed corneal compensator neutralizes the effect of corneal birefringence in this orientation. Greenfield et al³⁴ demonstrated that while the mean corneal birefringence axis is downward and nasal, there is considerable variability, which can alter the apparent RNFL thickness measured by SLP. In our study, no additional compensation for corneal birefringence was used other than the standard built-in compensator device. We relied on intereye comparisons in our analysis, which is suboptimal because up to 20% of patients will have an intereye difference in corneal birefringence of at least 20%.³⁴ It is possible that intereye differences in corneal birefringence could have confounded results in groups 1 and 2.

There are 2 reasons, however, to believe that the corneal birefringence did not significantly alter our results. First, we included group 2 (unilateral optic atrophy group) with the expectation that we should obtain lower retardation in affected eyes compared with contralateral eyes even if corneal birefringence confounded the data. We found a significant decrease in the retardation of all 6 parameters in the affected atrophic eyes compared with the unaffected eyes (Table 2). Second, Greenfield and Knighton²⁵ reported that corneal polarization axis measurements remained stable for at least 1 year. This finding is important in the longitudinal evaluation of RNFL thickness by SLP. In our follow-up group, there was no significant difference in the sequential RNFL measurements in the unaffected eyes, whereas the affected eyes that had disc edema and later developed atrophy showed a significant decrease in RNFL thickness (Table 5). This finding is consistent with Colen et al³⁵ who found loss of RNFL birefringence over time in a patient with NAION. The decrease in RNFL thickness found in our follow-up group cannot be attributed to unilateral change in corneal birefringence over time. Rather, the decrease must result from RNFL atrophy.

There are other caveats with respect to our work. While the RNFL was clinically thickened in patients with acute unilateral disc edema (Figure 1) and thinned in patients with chronic unilateral disc atrophy, we did not objectively determine RNFL thickness changes with alternative methods, such as retinal tomography. Second, members of the follow-up group represent a subset of patients in group 1. However, the participation in the prospective phase of the study was determined solely by our belief that additional neuro-ophthalmic care was warranted and by the patient's readiness to return. In future studies, it would be preferable to uniformly follow-up all patients with inflammatory and ischemic optic nerve disease to determine if the SLP parameters predict visual outcome. Such studies might contribute to the understanding of the mechanism of optic nerve damage in these disorders. Finally, we cannot rule out the possibility that patients in group 1 had subacute attacks of optic neuropathy prior to entering our study or that the fellow eyes of study patients were subject to subclinical optic neuropathy.

In conclusion, our study of patients with nonglaucomatous optic nerve disease provides support for the notion that SLP assesses form birefringence of ganglion cells axons. The conclusion is primarily based on the pervasive decrease in SLP parameters in patients with optic nerve atrophy. A second significant finding is that SLP is not a surrogate for RNFL thickness in eyes with optic nerve swelling (clearly, in glaucoma, measuring form birefringence is a surrogate for NFL thickness). Scanning laser polarimetry parameters may be decreased with optic nerve swelling, but they do not increase despite the clinically evident swelling of the RNFL. This finding is not inconsistent with the belief that SLP measures form birefringence of structural elements within the RNFL, which would not be expected to substantially increase with optic nerve head edema. Scanning laser polarimetry is a useful tool in the assessment of the status of retinal ganglion cell axons, especially in cases of RNFL atrophy.

Corresponding author and reprints: Louis R. Pasquale, MD, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114 (e-mail: lpasquale@partners.org).

Submitted for publication August 15, 2002; final revision received December 9, 2002; accepted December 19, 2002.

This study was presented in part at the 2001 meeting of the Association for Research in Vision and Ophthalmology, Ft Lauderdale, Fla, April 29, 2001.

Drs Banks and Robe-Collignon are both first authors.