Scanning laser polarimetry of the retinal nerve fiber layer in the detection and monitoring of glaucoma

Introduction

Identifying retinal nerve fiber layer (RNFL) abnormalities and their progression is of vital importance for the diagnosis and monitoring of glaucoma. However, until recently, this task has been subjective, and descriptions of change have been primarily qualitative. Recently, techniques have been developed to objectively measure correlates of RNFL thickness in intact human eyes. One such technique, scanning laser polarimetry (SLP), has rapidly disseminated to the clinics of glaucoma specialists worldwide. Although this technique is relatively easy to use and provides quantitative measurements of the RNFL, it is not yet apparent how the large amount of information provided by this technology should be interpreted clinically. Further, research to determine whether this methodology is effective for monitoring glaucomatous change over time is in its early stages.

The modern scanning laser polarimeter is an advanced version of the prototype Fourier ellipsometer developed by Weinreb et al. a decade ago.1 The prototype ellipsometer employed a 514-nm laser source, while the current instrument employs a 780-nm source. SLP exploits the birefringent property of the human RNFL which originates from the parallel organization of the microtubules that support retinal ganglion cell axons.2,3 When light from a polarization modulated laser beam (780-nm diode), with its optic axis parallel to the surface of the birefringent RNFL, is focused on the retina by the optical media of the eye and double-passes the RNFL, the light is split into two beams with different polarization axes travelling at different veloci-

ties. This difference in velocities results in a phase shift, or relative retardation, of the existing beams, resulting in a change in polarization of the reflected light. The polarization axis of the beam is changed (retarded) by an amount dictated by the thickness of the RNFL.4,5 The thicker the birefringent structure, the greater the retardation of transmitted light. Therefore, this technique provides a direct measurement of light retardation that is reportedly linearly related to RNFL thickness in histological preparations of primate retinae.1,6 Retardation values are transformed into indirect measurements of RNFL thickness, based on this known linear relationship. A complete SLP scan measures retardation at 256 × 256 retinal positions, thus acquiring 65,536 data points in less than one second. Three individual scans are usually combined to create a composite mean.

The majority of scientific and clinical studies investigating SLP technology have used the Laser Diagnostic Technology (San Diego, CA) GDx or its compatible predecessor, the Nerve Fiber Analyzer (NFA) II. Recently, a more automated, compact device; the Access (Laser Diagnostic Technology), has been introduced. A number of studies have shown significant differences in GDx/ NFA II-measured RNFL thickness between healthy and glaucomatous eyes.7-13

For the clinician’s purposes, a summary GDx printout is available that provides RNFL thickness parameters obtained by the instrument. Information provided on the summary printout includes a reflectance map, a retardation map (also called a thickness map) illustrating RNFL thickness using color coding, a graph depicting RNFL thickness within a

measurement ellipse that circumscribes the optic nerve head, a table that provides quadrant specific deviation from normal thickness values (relative to a large normative database), several summary values (NFA parameters) with corresponding probability values compared to the normative database, and a neural network derived value (The Number) that indicates the likelihood of glaucoma in the examined eye (values for this parameter range from 0 or very unlikely glaucoma, to 100 or very likely glaucoma). Figure 1 shows a GDx printout from a healthy eye.

between operators. One reproducibility study using the GDx showed that consecutive RNFL thickness measurements obtained from healthy eyes differ by only about 8 µm, or about 10%.17 Values are slightly higher in glaucoma eyes. The relatively low variability observed between operators when imaging the same patients suggests that using different technicians for follow-up examinations is not likely to be a cause for concern. Nonetheless, although GDx images are relatively easy to obtain, adequate training of technicians and quality control assurance are necessary to obtain reliable data.

Reproducibility

For a new disease diagnostic/monitoring instrument to be effective, it must provide similar measurements over time in the same subjects. That is, its measurements must be reproducible. Reproducibility of measurements must be good both within and

Retinal nerve fiber layer thickness in healthy and glaucoma eyes

Briefly, in primary open-angle glaucoma, the retinal ganglion cell population is decreased, probably due to changes in neurotrophic mechanisms, effects of increased intraocular pressure (or increased sus-

Fig. 2. GDx measured average RNFL thickness in 134 healthy eyes. The horizontal line depicts the mean value of 66.9 µm (standard deviation = 11.1 µm). The range of values is 47-100 µm.

Fig. 3. GDx measured average RNFL thickness in 65 glaucomatous eyes (eyes with repeatable abnormal standard automated perimetry results) and 75 healthy eyes. Diamonds represent 95% confidence intervals. The mean value for glaucomatous eyes is 61.8 µm (standard deviation, 13.88 µm) and the mean value for healthy eyes is 68.1 µm (standard deviation, 12.6 µm). There is a statistically significant difference in RNFL thickness between groups, despite the large overlap in measurements (t (1,138) = -2.83; p = 0.005).

ceptibility to the effects of intraocular pressure), decreased ocular perfusion, or a combination of these factors. This decrease in the retinal ganglion cell population in glaucoma results in a thinner than normal RNFL in glaucoma eyes compared to healthy ones. However, within the healthy population, optic never fiber count varies by a factor of more than two (reported range: approximately 700,000-1.5 million fibers per eye).18-21 Because optic nerve fiber count is correlated with RNFL thickness, RNFL thickness in healthy eyes also varies dramatically. For instance, healthy global RNFL thickness measured using GDx ranges from

approximately 50-100 µm.22,23 Figure 2 shows an example of the range of GDx RNFL thickness measurements in a group of 134 healthy eyes.23

In glaucoma eyes, the reported average RNFL thickness across several studies ranges from approximately 61-77 µm. Some of this variation is likely due to the difference of glaucoma severity (and other factors) across studies, however, it is clear that many glaucoma eyes fall within the normal range. Figure 3 shows the distribution of GDx RNFL thickness measurements in 75 healthy and 65 glaucoma eyes, illustrating the substantial overlap in RNFL thickness between subjects and patients. This overlap hampers the ability to effectively discriminate between healthy and glaucomatous eyes based on this parameter alone. Because of this, other parameters that describe the relationships between superior and inferior thickness and temporal and nasal thickness (ratio and modulation parameters) are available.

Detecting glaucoma with GDx/NFA II

Using the standard clinical GDx printout, the clinician may gain some insight into the health of the RNFL and, therefore, in combination with other sources of information, the probability that the patient has glaucomatous damage.

The RNFL thickness map (or retardation map) provides information regarding the thickness of the RNFL across the available field of view. In healthy eyes, the RNFL is thicker in the inferior and superior quadrants than in the nasal and temporal ones, resulting in a sinusoidal ‘double hump’ thickness profile around the optic disc. Furthermore, the thickness in the inferior and superior quadrants is approximately equal and uniform (except in the cases of healthy eyes with split nerve fiber bundles).24 Therefore, any dramatic deviation from this pattern may be cause for concern. The RNFL thickness graphs provide similar thickness modulation information. Information comparing quadrant specific thickness values relative to a normative database is also informative, assuming the normative database includes accurate measurements from subjects with a wide variety of racial backgrounds, spanning all ages.

Figure 4 shows a GDx printout from a glaucomatous eye. In this case, the patient is a 90- year-old female glaucoma patient. The standard automated perimetry result closest to the GDx imaging date shows a superior arcuate defect extending from the blind spot to the nasal step region, and a less deep inferior arcuate defect. The Glaucoma Hemifield Test result is outside normal limits, and mean deviation is -10.5 dB (p < 0.5%). The GDx retardation map suggests decreased RNFL thickness overall, with a decrease in thickness amplitude around the disc. This information is mirrored by the RNFL thickness graph. The quadrant-spe-

cific deviation from normal values information suggests RNFL thinning in the superior and inferior quadrants. Several NFA parameters are outside normal limits, particularly those that describe the amplitude of the RNFL thickness profile (ratio and modulation parameters). The GDx ‘Number’ is within the range indicative of glaucoma.

Figure 5 shows an alternate GDx printout that compares the right and left eyes of the above patient in order to display possible asymmetries in RNFL thickness measurements and their deviations from normal. This type of analysis may be valuable because glaucoma often initially presents in one eye only, or is asymmetric in severity.

Sensitivity for detecting glaucoma with GDx/ NFA II

Despite the large and overlapping range of RNFL thickness measurements in both healthy and glaucoma eyes, some studies have reported relatively good sensitivities for detecting eyes with glaucomatous damage, using information available to the

clinician. Choplin and Lundy reported that experienced observers provided with masked GDx printouts were able to successfully identify 83% of 42 known glaucoma patients.25 In a similar study, also using clinical printouts, Sanchez-Galeana et al. reported that three experienced observers identified between 72% and 82% of 39 known glaucoma patients.26 Other studies have reported better sensitivities for discriminating glaucoma eyes from healthy ones using individual parameters or combinations of parameters in statistical models.10,12,27,28 However, in some cases, reported sensitivities have been worse.7,13 Differences in reported sensitivities across studies are probably affected by study methodology, glaucoma severity in the different study populations, and possibly by image quality affected by calibration of the instrument and operator skill. Moreover, sensitivity for detecting glaucoma cannot always be compared across studies because reported sensitivities may be dependent upon different specificity cut-offs for individual continuous variable GDx parameters or combinations thereof.

Other sources of birefringence in the human eye

For laser light from the GDx to measure RNFL thickness, it must first pass through two structures other than the RNFL that might provide birefringent information that could affect measurements: the cornea and the lens. To address the possible artifactual contribution of these structures to RNFL thickness measurements, the GDx incorporates a corneal polarization compensator that assumes an axis of corneal polarization of 15° nasally downward. Recently, Greenfield et al. determined that there is a large variation in the corneal polarization axis in normal eyes.29 Because of this, it is likely that polarization artifacts from the cornea influence GDx RNFL measurements in a substantial number

of eyes in which corneal polarization compensation is not complete, thus making the measurements inaccurate.

Evidence suggests that birefringence of the lens has little or no effect on GDx measurements.30,31

Addressing corneal polarization effects

A recent prototype instrument designed for research incorporates a variable corneal compensator with the GDx to address the effect of the considerable range of corneal polarization axes within the population. Using this device, RNFL thickness is measured at the macula where no ‘double-hump’ thickness profile is expected. The variable compensator

is adjusted until a flat thickness profile is obtained, and the resulting corrected polarization axis value is used in measuring RNFL thickness in the peripapillary region. Zhou and Weinreb found that individualized anterior segment compensation could be achieved so that the measured birefringence largely reflects the RNFL birefringence.32 Recent studies using this device have shown an improvement in discrimination between healthy and glaucomatous eyes, using variable compensation compared to fixed axis compensation for some but not all parameters.33,34 These results suggest that using patient-specific polarization axis compensation will likely improve the glaucoma diagnosis ability of the GDx.

Following glaucoma with GDx/NFA II

A recent study by Greenfield and Knighton showed that the axis of corneal polarization in healthy eyes is probably stable over time. This information suggests that, even if baseline measurements are inaccurate because of corneal contributions to RNFL thickness measurements, changes in RNFL

measurements over time are not likely to be effected by changes in corneal polarization axis and that, although an inaccurate baseline measurement may exist, changes in RNFL thickness over time in glaucoma eyes can probably be attributed to changes in the disease state. Therefore, it appears that SLP technology, even with fixed corneal polarization axis compensation, may be promising for glaucoma monitoring and follow-up, although this claim has not yet been sufficiently tested.

For the monitoring of glaucomatous progression, a serial examination report printout is available to the clinician. This printout includes color-coded deviation from baseline thickness values for each follow-up examination. Significant thickness changes in serial images are determined by comparison to the variability in the three scans that make up the composite baseline image. This printout also includes retardation images for each examination, quadrant-specific deviation from normal values, select parameter values, and graphs depicting RNFL thickness within the measurement ellipse for baseline and follow-up images. Figure 6 shows a serial analysis printout depicting progressive RNFL thinning over the course of 18 months.

At the time of the baseline image, this female patient was 68 years old. The standard automated perimetry result closest to the imaging date shows a superior arcuate defect extending from the blind spot to the nasal step region. The Glaucoma Hemifield Test result is outside normal limits, and the mean deviation is -7.2 dB (p < 0.5%). The serial retardation images show thinning (decreased brightness) in the superior, nasal, and inferior quadrants compared to baseline. This information is reflected in the deviation from normal figures, deviation from reference figures (shown as significant superior temporal thinning in the first follow-up image, and significant superior, nasal, and inferior nasal thinning in the second follow-up image), parameter values, and RNFL thickness graphs.

It is possible that eyes which have undergone corneal laser refractive procedures, such as laser assisted in situ keratomelusis (LASIK), will provide a challenge to accurate follow-up using GDx, because of the potential change in the polarization axis of the lens as a result of these procedure. In these cases, it may be necessary to change the baseline image for serial analysis to a post-treatment examination date. Some studies have shown a significant decrease in SLP measured RNFL thickness after LASIK.36-38 However, one study showed no effect of excimer laser photorefractive keratectomy (PRK) on GDx parameters.39 In none of these studies was the effect of the procedures on the axis or magnitude of corneal polarization investigated directly. The issue of possible corneal polarization changes following laser refractive procedures might be avoidable when using a variable corneal compensation device. In this case, changes in corneal polarization axis and magnitude could be controlled for prior to image acquisition.

Population-based glaucoma screening using GDx/NFA II

The majority of GDx-related studies are conducted in glaucoma clinics, which are often referral practices that may not accurately portray the demographic and disease characteristics of the general population. Moreover, referral patients may be more cooperative. For these reasons, a new glaucoma screening technique should be tested in a real-world environment. Two population-based studies have addressed the success of GDx as a screening tool. Vitale et al. reported a sensitivity and specificity of 69% and 82%, respectively, using the GDx/NFA II neural network ‘Number’ in a population of 280 healthy eyes and in 98 eyes with definite or probable glaucoma in the Baltimore Eye Survey FollowUp study.40 Using the same parameter (GDx ‘Number’), Yamada et al.41 reported an optimal sensitivity and specificity (determined from receiver operating characteristic (ROC) curves) of 68% and 90%, respectively, in 122 healthy and 22

glaucomatous eyes of attendees at a two-day public glaucoma screening program. These results are similar to those obtained in clinical settings.

Conclusions

SLP is a promising technique for assessing RNFL thickness in vivo by measuring polarization changes in laser light caused by the birefringent structure of the RNFL. Measurements using the GDx are reproducible, and differences in RNFL thickness between healthy and glaucomatous eyes are well substantiated. The GDx provides informative summary information which allows experienced clinicians to discriminate between healthy and glaucomatous eyes with reasonable accuracy. Furthermore, serial analysis information may allow the clinician to detect changes in nerve fiber layer thickness over time in diseased eyes. However, some GDx measurements can be inaccurate due to inadequate compensation of corneal polarization in some eyes. Moreover, some corneal laser refractive procedures may make comparisons between preand post-procedure measurements uninformative. Individually compensating for corneal birefringence will likely enhance the performance of this technology.

References

1.Weinreb RN, Dreher AW, Coleman A, Quigley H, Shaw B, Reiter K: Histopathologic validation of Fourierellipsometry measurements of retinal nerve fiber layer thickness. Arch Ophthalmol 108:557-560, 1990

2.Dreher AW, Reiter K: Scanning laser polarimetry of the retina nerve fiber layer. SPIE Proceedings 1746:34-38, 1992

3.Knighton RW, Huang X, Zhou Q: Microtubule contribution to the reflectance of the retinal nerve fiber layer. Invest Ophthalmol Vis Sci 39:189-193, 1998

4.Knighton RW, Jacobson SG, Kemp CM: The spectral reflectance of the nerve fiber layer of the macaque retina. Invest Ophthalmol Vis Sci 30:2392-2402, 1989

5.Knighton RW, Huang XR: Directional and spectral reflectance of the rat retinal nerve fiber layer. Invest Ophthalmol Vis Sci 40:639-647, 1999

6.Morgan JE, Waldock A, Jeffery G, Cowey A: Retinal nerve fibre layer polarimetry: histological and clinical comparison. Br J Ophthalmol 82:684-690, 1998

7.Bowd C, Zangwill LM, Berry CC et al: Detecting early glaucoma by assessment of retinal nerve fiber layer thickness and visual function. Invest Ophthalmol Vis Sci 42:19932003, 2001

8.Choplin NT, Lundy DC, Dreher AW: Differentiating patients with glaucoma from glaucoma suspects and normal subjects by nerve fiber layer assessment with scanning laser polarimetry. Ophthalmology 105:2068-2076, 1998

9.Hoh ST, Greenfield DS, Mistlberger A, Liebmann JM, Ishikawa H, Ritch R: Optical coherence tomography and scanning laser polarimetry in normal, ocular hypertensive, and glaucomatous eyes. Am J Ophthalmol 129:129-135, 2000

10.Horn FK, Jonas JB, Martus P, Mardin CY, Budde WM: Polarimetric measurement of retinal nerve fiber layer thick-

ness in glaucoma diagnosis. J Glaucoma 8:353-362, 1999

11.Lauande-Pimentel R, Carvalho RA, Oliveira HC, Goncalves DC, Silva LM, Costa VP: Discrimination between normal and glaucomatous eyes with visual field and scanning laser polarimetry measurements. Br J Ophthalmol 85:586591, 2001

12.Weinreb RN, Zangwill L, Berry CC, Bathija R, Sample PA: Detection of glaucoma with scanning laser polarimetry. Arch Ophthalmol 116:1583-1589, 1998

13.Zangwill LM, Bowd C, Berry CC et al: Discriminating between normal and glaucomatous eyes using the Heidelberg Retina Tomograph, GDx Nerve Fiber Analyzer, and Optical Coherence Tomograph. Arch Ophthalmol 119:985993, 2001

14.Hoh ST, Ishikawa H, Greenfield DS, Liebmann JM, Chew SJ, Ritch R: Peripapillary nerve fiber layer thickness measurement reproducibility using scanning laser polarimetry. J Glaucoma 7:12-15, 1998

15.Kook MS, Sung K, Park RH, Kim KR, Kim ST, Kang W: Reproducibility of scanning laser polarimetry (GDx) of peripapillary retinal nerve fiber layer thickness in normal subjects. Graefe’s Arch Clin Exp Ophthalmol 239:118121, 2001

16.Waldock A, Potts MJ, Sparrow JM, Karwatowski WS: Clinical evaluation of scanning laser polarimetry: I. Intraoperator reproducibility and design of a blood vessel removal algorithm. Br J Ophthalmol 82:252-259, 1998

17.Colen TP, Tjon-Fo-Sang MJ, Mulder PG, Lemij HG: Reproducibility of measurements with the nerve fiber analyzer (NfA/GDx). J Glaucoma 9:363-370, 2000

18.Jonas JB, Muller-Bergh JA, Schlotzer-Schrehardt UM, Naumann GO: Histomorphometry of the human optic nerve. Invest Ophthalmol Vis Sci 31:736-744, 1990

19.Jonas JB, Schmidt AM, Muller-Bergh JA, SchlotzerSchrehardt UM, Naumann GO: Human optic nerve fiber count and optic disc size. Invest Ophthalmol Vis Sci 33:20122018, 1992

20.Jonas JB, Schmidt AM, Muller-Bergh JA, Naumann GO: Optic nerve fiber count and diameter of the retrobulbar optic nerve in normal and glaucomatous eyes. Graefe’s Arch Clin Exp Ophthalmol 233:421-424, 1995

21.Mikelberg FS, Drance SM, Schulzer M, Yidegiligne HM, Weis MM: The normal human optic nerve: axon count and axon diameter distribution. Ophthalmology 96:1325-1328, 1989

22.Poinoosawmy D, Tan JC, Bunce C, Membrey LW, Hitchings RA: Longitudinal nerve fibre layer thickness change in normal-pressure glaucoma. Graefe’s Arch Clin Exp Ophthalmol 238:965-969, 2000

23.Bowd C, Zangwill L, Blumenthal E et al: Imaging of the optic disc and retinal nerve fiber layer: the effects of age, optic disc area, refractive error, and gender. J Opt Soc Am A 19:197-207, 2001

24.Colen TP, Lemij HG: Prevalence of split nerve fiber layer bundles in healthy eyes imaged with scanning laser polarimetry. Ophthalmology 108:151-156, 2001

25.Choplin NT, Lundy DC: The sensitivity and specificity of scanning laser polarimetry in the detection of glaucoma in a clinical setting. Ophthalmology 108:899-904, 2001

26.Sanchez-Galeana C, Bowd C, Blumenthal EZ, Gokhale PA, Zangwill LM, Weinreb RN: Using optical imaging summary data to detect glaucoma. Ophthalmology 108:18121818, 2001

27.Trible JR, Schultz RO, Robinson JC, Rothe TL: Accuracy of scanning laser polarimetry in the diagnosis of glaucoma. Arch Ophthalmol 117:1298-1304, 1999

28.Essock EA, Sinai MJ, Fechtner RD, Srinivasan N, Bryant FD: Fourier analysis of nerve fiber layer measurements from scanning laser polarimetry in glaucoma: emphasizing shape characteristics of the ‘double-hump’ pattern. J Glaucoma 9:444-452, 2000

29.Greenfield DS, Knighton RW, Huang XR: Effect of corneal polarization axis on assessment of retinal nerve fiber layer thickness by scanning laser polarimetry. Am J Ophthalmol 129:715-722, 2000

30.Collur S, Carroll AM, Cameron BD: Human lens effect on in vivo scanning laser polarimetric measurements of retinal nerve fiber layer thickness. Ophthalmic Surg Lasers 31:126130, 2000

31.Park RJ, Chen PP, Karyampudi P, Mills RP, Harrison DA, Kim J: Effects of cataract extraction with intraocular lens placement on scanning laser polarimetry of the peripapillary nerve fiber layer. Am J Ophthalmol 132:507-511, 2001

32.Zhou Q, Weinreb RN: Individualized compensation of anterior segment birefringence during scanning laser polarimetry. Invest Ophthalmol Vis Sci 43:2221-2228, 2002

33.Weinreb RN, Bowd C, Greenfield DS, Zangwill LM: Measurement of the magnitude and axis of corneal polarization with scanning laser polarimetry. Arch Ophthalmol 120:901906, 2002

34.Weinreb RN, Bowd C, Zangwill L: Glaucoma detection using scanning laser polarimetry with variable corneal polarization compensation. Arch Ophthalmol. In press, 2002

35.Greenfield DS, Knighton RW: Stability of corneal polarization axis measurements for scanning laser polarimetry. Ophthalmology 108:1065-1069, 2001

36.Gurses-Ozden R, Pons ME, Barbieri C et al: Scanning laser polarimetry measurements after laser-assisted in situ keratomileusis. Am J Ophthalmol 129:461-464, 2000

37.Gurses-Ozden R, Liebmann JM, Schuffner D, Buxton DF, Soloway BD, Ritch R: Retinal nerve fiber layer thickness remains unchanged following laserassisted in situ keratomileusis (1). Am J Ophthalmol 132:512-516, 2001

38.Tsai YY, Lin JM: Effect of laser-assisted in situ keratomileusis on the retinal nerve fiber layer. Retina 20:342345, 2000

39.Choplin NT, Schallhorn SC: The effect of excimer laser photorefractive keratectomy for myopia on nerve fiber layer thickness measurements as determined by scanning laser polarimetry. Ophthalmology 106:1019-1023, 1999

40.Vitale S, Smith TD, Quigley T et al: Screening performance of functional and structural measurements of neural damage in open-angle glaucoma: a case-control study from the Baltimore Eye Survey. J Glaucoma 9:346-356, 2000

41.Yamada N, Tomita G, Yamamoto T, Kitazawa Y: Changes in the nerve fiber layer thickness following a reduction of intraocular pressure after trabeculectomy. J Glaucoma 9:371375, 2000