DETECTION OF GLAUCOMATOUS VISUAL FIELD LOSS USING MULTIFOCAL VISUAL EVOKED POTENTIAL

BRAD FORTUNE, KAI GOH, SHABAN DEMIREL, KARIN NOVITSKY, STEVEN L. MANSBERGER, CHRIS A. JOHNSON and GEORGE A. CIOFFI

Discoveries in Sight, Devers Eye Institute, Portland, OR, USA

Abstract

Purpose: To evaluate a new multifocal visual evoked potential (VEP) system for its ability to detect glaucomatous visual field loss in a clinical setting. Methods: Multifocal VEPs were recorded in both eyes of 36 patients with glaucoma and 35 control subjects using the AccuMap system (ObjectiVision Pty, Ltd, Sydney). Glaucoma diagnosis was based upon optic disc appearance (glaucomatous optic neuropathy in at least one eye) and visual field loss measured by standard automated perimetry (SAP; p < 0.05 for mean defect (MD) or pattern standard deviation (PSD); or glaucoma hemifield test (GHT) outside normal limits) confirmed in at least one eye. Controls had normal visual fields (by above criteria) and normal optic disc appearance in both eyes. Sensitivity and specificity levels were determined for various criteria using the AccuMap internal normative database. Results: SAP visual field MD ranged from 1.8 to –28.6 dB (mean, –8.4, SD 7.4). At a fixed specificity level of 90%, sensitivity for detection of an abnormal SAP visual field was 75 and 82%, based on the number of abnormal VEP points (max 58 possible per eye) below the p = 0.02 and p = 0.01 level, respectively. Using the AccuMap ‘severity index’ (a proprietary algorithm that uses weighted combinations of above, interocular asymmetry, and clustering factors), sensitivity was 78 at 90% specificity. Adding the requirement of a cluster of at least three abnormal sectors on the interocular asymmetry plot (p < 0.1% level) to the monocular amplitude criteria improved sensitivity to 92% for all glaucoma eyes and 97% for glaucoma subjects, with 90% specificity. Conclusions: The AccuMap multifocal VEP (mfVEP) system can be used to detect glaucomatous visual field loss with reasonably high sensitivity and specificity. Although control subjects commonly have isolated ‘defects’ for mfVEPs obtained using the AccuMap system, these defects are rarely clustered.

Introduction

Numerous studies have demonstrated that the visual evoked potential (VEP) can be used effectively for assessment of vision function in glaucoma.1-9 Yet clinical application of traditional VEP techniques10,11 remains hampered for several reasons including: the relatively low spatial resolution of conventional methods, dramatic differences in response amplitudes between normal individuals, and dominance of the conventional VEP response by contributions from foveal and inferior visual field areas. The wide range of normal

Address for correspondence: Brad Fortune, OD, PhD, Discoveries in Sight, 1225 NE Second Avenue, Portland, OR 97232, USA. Email: bfortune@discoveriesinsight.org

Perimetry Update 2002/2003, pp. 251–260

Proceedings of the XVth International Perimetric Society Meeting, Stratford-upon-Avon, England, June 26–29, 2002

edited by David B. Henson and Michael Wall

© 2004 Kugler Publications, The Hague, The Netherlands

VEP responses that complicates discrimination of glaucoma from normal is partly attributable to the fact that the anatomy of the striate cortex (primary visual cortex, V1) can vary greatly between individuals. Further, the relationship between the position and orientation of V1 to the external landmarks of the skull (i.e., the references for standard electrode placements) can also vary widely between individuals.27,28 For these and other reasons, conventional VEP responses, reflecting the sum of signals from the whole (typically large) stimulus area, can actually become larger when focal field loss is present in glaucoma (e.g., vertical hemifield loss) (see, for example, Fortune et al.21).

The multifocal technique has improved clinical VEP testing by enabling much finer spatial resolution without significantly increasing recording time.29,30 Analysis of individual field locations reduces some of the confounds inherent to the traditional large-field VEP. Moreover, the typical multifocal VEP (mfVEP) stimulus is spatially scaled with eccentricity, such that a larger extent of the visual field can be examined.16,29,31 Of course, inter-subject variability can still pose problems, and this was recognized by Baseler et al. during their initial studies on mfVEP.29 Recent improvements in recording and analysis methods, however, have led to rapid growth in the use of mfVEP for assessment of glaucoma and other optic nerve diseases.

Subsequent to their initial studies, Graham et al. developed a new instrument for acquisition and analysis of mfVEP data, incorporating several attractive features such as an electroencephalograph (EEG) based scaling algorithm, probablistic analyses of local monocular amplitudes and binocular asymmetry values, a fixation task and monitor, and online evaluation of signal improvement with each recorded segment.40,42 The results of their first study using this new system for ‘objective perimetry’ showed high sensitivity and specificity for detection of glaucomatous visual field loss.40 The purpose of the current study was to evaluate the ability of this new, commercially available system to detect glaucomatous visual field loss in a clinical setting.

Methods

Subjects

Seventy-one subjects participated in this study, including 36 patients with glaucoma and 35 control subjects. All procedures adhered to the tenets of the Declaration of Helsinki, and were approved by the institutional review board for the protection of human subjects. All subjects volunteered their written consent to participate after they were fully informed about the purpose of the study and its potential risks and benefits.

The diagnosis of glaucoma was based upon optic disc appearance (glaucomatous optic neuropathy (GON) in at least one eye) as determined by clinical examination and stereo disc photographs (the latter by consensus of two masked graders, both glaucoma specialists), as well as visual field loss measured by standard automated perimetry (SAP; Humphrey 24-2 SITA threshold VFs, Zeiss Humphrey, Dublin, CA) with at least one of the following criteria: p < 0.05 for mean defect (MD), pattern standard deviation (PSD); or glaucoma hemifield test (GHT) outside normal limits, confirmed in at least one eye. Control subjects had normal visual fields (by the above criteria) and normal optic disc appearance in both eyes according to clinical examination and consensus of the same two masked graders. All visual fields were required to be

reliable: fixation losses, false positives and false negatives <33%. None of the control subjects had a history of ocular disease, surgery, or trauma, nor systemic disease known to affect the visual system.

The mean age (± standard deviation) of the control and glaucoma subject groups was 53.2 ± 13.7 and 66.8 ± 12.1 years, respectively. The slight age difference between groups has little impact for this study because the EEG scaling algorithm of the AccuMap system substantially reduces inter-subject variability of mfVEP amplitudes normally attributed to age and gender.40,42 The average visual field findings for the control group were: MD = 0.56 ± 0.9 dB, PSD = 1.51 ± 0.3 dB; and for the glaucoma group, MD = –8.36 ± 7.4 dB, PSD = 7.56 ± 4.7 dB.

Multifocal VEP data acquisition

Multifocal VEPs were obtained using the AccuMap objective perimetry system (ObjectiVision Pty, Ltd, Sydney). Operational details of the system have been published previously.40,42 Briefly, patients were seated 30 cm in front of the stimulus monitor (21-inch high resolution monitor, model CM811, Hitachi, Ltd, Tokyo) which was positioned slightly above head level so that neck muscles would be relaxed during the recordings. Optimal refraction was determined for each eye and correction provided with trial lenses in the spectacle plane, pupils were not dilated. Gold-disc VEP electrodes (Grass Instrument Division, Astro-Med, Inc., Warwick, RI) were placed at four occipital scalp locations using the ‘occipital cross’ holder, such that when the cross was centered over the inion, the upper vertical midline electrode was positioned 3 cm above the inion, the lower vertical midline electrode 4.5 cm below the inion, and the right and left lateral electrodes 4 cm on either side of the inion, while an ear-clip served as ground. The skin at each site was prepared using a cotton-tipped wooden swab with Nuprep (D.O. Weaver & Co., Aurora, CO) and the electrode cups were filled with conductive gel (Dracard, London).

The stimulus (see Fig. 1) consists of 58 sectors, each with eight black checks (< 1 cd/m2) and eight white checks (150 cd/m2), providing a Michelson contrast of ~ 99% and a mean luminance of 75 cd/m2. The size of the individual stimulus sectors, and correspondingly, their check sizes, are approximately scaled with eccentricity according to a cortical magnification factor. The stimulus subtended 26 degrees (radius), in all meridia except nasally where it extends to 32 degrees. Contrast reversals are determined by a pseudo-random binary sequence (using a spread spectrum technique) whereby each sequence has a total of 4096 steps that requires ~55 sec to complete. Reversals at each stimulus sector follow a different sequence and each recording segment at a given sector also follows a unique sequence.

Monocular recordings for both eyes of each subject were obtained (right eye first). Each recording consisted of ten segments (thus ~10 minutes total per eye). The operator allowed subjects to rest between segments as necessary. Subjects maintained fixation at the center of the stimulus display and pressed a response button each time the number 3 appeared during a continuous random presentation of four numerals (3, 6, 8, or 9). This task helps to control fixation as well as to reduce alpha rhythm interference. The system operator also monitored fixation visually as well as the quality of the online signals during each recording segment. If fixation was poor, or if a segment contained obvious artifact (e.g., alpha or muscle interference), the operator stopped

Fig. 1. AccuMap multifocal VEP stimulus display (as viewed for right eye testing).

and re-recorded the segment. Signals in four channels were recorded with a Grass amplifier (Model 15 Neurodata, Astro-Med, Inc., Warwick, RI), at 502 Hz sampling rate, using 100,000 times amplification, and band-pass filtering between 3 and 30 Hz.

Multifocal VEP data analyses

The AccuMap system calculates the maximum peak-to-trough amplitude from each of the four channels within the response epoch 60-180 msec for each of the 58 stimulus locations in each eye. The final response array is made up of the ‘best’ VEP at each location. That is, the channel chosen at each location is determined by the VEP with the largest amplitude from among the eight channels possible (four channels x two eyes). The amplitude at each location in each eye is then compared with the internal normative database (currently consisting of 100 subjects) to determine probability levels of normality (a Statpac-like analysis). Locations with a p value of less than 0.05 are considered abnormal. Similarly, the inter-ocular amplitude difference at each location (less the two nasal step locations for each eye) can be compared with the normative database and assigned a probability value. The system also calculates a ‘severity index’ (SI) for each eye based on the number of abnormal points and their severity (i.e., the p values for monocular amplitude data) with added weighting for field location and clustering. The SI also takes into account any significant asymmetry within an eye and between the two eyes. The proprietary algorithm used to calculate the SI is described in general terms elsewhere and is not easily reverse engineered. The AccuMap system flags an SI score between 30 and 39 as borderline (3% < p < 5%); and an SI score greater than 39 as abnormal (p < 3%). An SI score less than 30 is considered normal.

Fig. 2. Monocular amplitude results: number of abnormal points in each eye (at the p < 2% level) versus SAP visual field mean defect. Glaucoma group, filled triangles; controls, open circles. Dashed line is the 95th percentile for the control group.

Fig. 3. Monocular amplitude results: number of abnormal points in each eye (at the p < 1% level) versus SAP visual field mean defect. Glaucoma group, filled triangles; controls, open circles. Dashed line is the 95th percentile for the control group.

Results

Figure 2 plots the number of responses per eye that are abnormal at the p < 2% level against SAP mean defect. Although there is a strong relationship between the number of abnormal points and MD for the glaucoma group (filled triangles), there is also modest overlap with the control group, even for MD values as high as –10 dB. Figure 3 plots the number of responses per eye that are abnormal at the p < 1% level against SAP mean defect. As expected, there are many fewer sectors in each eye with abnormal responses using this more conservative criterion. Yet there is still overlap between the glaucoma and control groups, with several glaucoma eyes having a ‘normal’ num-

Fig. 4. AccuMap SI versus visual field mean defect. Dashed line is the 95th percentile for the control group.

ber of points at the 1% level (i.e., zero or one) despite moderate disease severity as determined via examination of their SAP MD values.

Figure 4 shows the AccuMap severity index (SI) plotted against mean defect. Again, there is a strong relationship between this global parameter of mfVEP normality and the commonly used SAP visual field global parameter. Note though, that the SI value decreases for larger MD values. This result is similar to what one might expect for a plot of SAP PSD versus MD. As the damage becomes more severe, the asymmetry within an eye and more importantly for the mfVEP – between the two eyes – decreases, resulting in fewer asymmetric sectors and a smaller SI score. However, there is still substantial overlap with the control group, even for a few eyes with moderate damage. Interestingly, several of the control eyes have an SI score well above the normality cut-off used by the AccuMap (> 39). This is because there were a relatively high number of monocular responses judged as ‘abnormal’ in many of the control eyes (data for sectors with p values < 5% not shown, but see Figures 2 and 3 for examples of controls with numerous abnormal sectors at the p < 2% and p < 1% level, respectively). However, abnormal sectors were rarely clustered in control eyes. This will be addressed further below.

The 95th percentile of the SI score for the control group in this study was 50.2 (see dashed line in Fig. 4). Using this criteria, the sensitivity to detect eyes of glaucomatous subjects (i.e., including fellow eyes with GON but normal SAP VFs) was 61%, while the sensitivity to detect glaucomatous VF loss was 72%. Using a slightly more relaxed criteria with a fixed specificity of 90%, the sensitivity to detect glaucomatous eyes was 67%, while the sensitivity to detect glaucomatous VF loss was 78%. Figure 5 shows receiver operating characteristic (ROC) curves based on the SI score. The area under the ROC labeled ‘glaucoma, all eyes’ was 0.85 with an overall correct classification of 76%. The area under the ROC labeled ‘glaucoma eyes, abnormal SAP VF’ was 0.93 with an overall correct classification of 85%.

Although there were several control eyes with numerous abnormal sectors (see Figs. 2 and 3), it was rare to find clusters of abnormal sectors in control eyes. In fact, there were no control eyes with a cluster of three or more contiguous abnormal sectors

Fig. 5. ROC curves based on the AccuMap SI score. Sensitivity is plotted against the false alarm rate (one-specificity) for two different discriminations: the lower curve shows results for controls versus all eyes with glaucoma (12 fellow eyes with GON, but normal SAP visual fields); while the upper curve shows the results for controls versus only the glaucoma eyes with abnormal SAP visual fields.

at the p < 1% level. Therefore, adopting a clustering requirement raises specificity for a given sensitivity level. At a fixed specificity of 90%, using the number of abnormal points at either the p < 2% or the p < 1% level, and adding the requirement of a cluster of at least three abnormal sectors on the interocular asymmetry plot (p < 0.1% level) improved sensitivity to 92% for all glaucoma eyes and to 97% for glaucoma subjects.

Discussion

These results generally confirm the findings of one previous study which found that the AccuMap System offers reasonably high sensitivity and specificity for ‘objective’ detection of glaucomatous VF loss.40 However, sensitivity and specificity values in this study were slightly lower than those reported by Goldberg et al.40 The differences may be due to the fact that the control group in this study was independently evaluated using the internal normative database of the system. In contrast, the control group in the Goldberg et al. study appears to be the same group of subjects used to develop the system’s normative database. Thus, in studies using an independent control group, it is unlikely that discriminatory performance could match or exceed that of the Goldberg et al. study.

Interestingly, some of the control subjects had several abnormal sector responses within an eye. This undermined discriminatory power for early glaucoma and ‘preperimetric’ GON when simple criteria such as the number of abnormal responses were used. However, specificity was greatly improved when a clustering requirement was added. Clustering criteria have been found to improve specificity in other mfVEP

References

1.Cappin JM, Nissim S: Visual evoked responses in the assessment of field defects in glaucoma. Arch Ophthalmol 93:9-18, 1975

2.Towle VL, Moskowitz A, Sokol S, Schwartz B: The visual evoked potential in glaucoma and ocular hypertension: effects of check size, field size, and stimulation rate. Invest Ophthalmol Vis Sci 24:175183, 1983

3.Howe JW, Mitchell KW: Visual evoked cortical potential to paracentral retinal stimulation in chronic glaucoma, ocular hypertension, and an age-matched group of normals. Doc Ophthalmol 63:37-44, 19862

4.Howe JW, Mitchell KW: Electrophysiologically determined contrast sensitivity in patients with ocular hypertension and chronic glaucoma. Doc Ophthalmol 80:31-41, 1992

5.Marx MS, Bodis-Wollner I, Lustgarten JS, Podos SM: Electrophysiological evidence that early glaucoma affects foveal vision. Doc Ophthalmol 67:281-301, 1987

6.Korth M, Koca M: Clinical electrophysiology relevant for early glaucoma diagnosis. Curr Opin Ophthalmol 4:22-28, 1993

7.Korth M, Nguyen NX, Junemann A, Martus P, Jonas JB: VEP Test of the blue-sensitive pathway in glaucoma. Invest Ophthalmol Vis Sci 35:2599-2610, 1994

8.Korth M: The value of electrophysiological testing in glaucomatous diseases. J Glaucoma 6:331343, 1997

9.Porciatti V, Di Bartolo E, Nardi N, Fiorentini A: Responses to chromatic and luminance contrast in glaucoma: a psychophysical and electrophysiological study. Vision Res 37:1975-1987, 1997

10.Harding GF, Odom JV, Spileers W, Spekreijse H: Standard for visual evoked potentials 1995. The International Society for Clinical Electrophysiology of Vision. Vision Res 36:3567-3572, 1996

11.Celesia GG, Brigell MG: Recommended standards for pattern electroretinograms and visual evoked potentials. The International Federation of Clinical Neurophysiology. Electroencephalogr Clin Neurophysiol Suppl 52:53-67, 1999

12.Rietveld WJ, Tordoir WE, Hagenouw JR, Van Dongen KJ: Contribution of foveo-parafoveal quadrants to the visual evoked response. Acta Physiol Pharmacol Neerl 13:340-347, 1965

13.Jeffreys DA: Striate and extra-striate origins of pattern-related visual evoked potential (VEP) components. J Physiol 211:Suppl:29P+, 1970

14.Spekreijse H, Van der Twell LH, Zuidema T: Contrast evoked responses in man. Vision Res 13:15771601, 1973

15.Michael WF, Halliday AM: Differences between the occipital distribution of upper and lower field pattern-evoked responses in man. Brain Res 32:311-324, 1971

16.Meredith JT, Celesia GG: Pattern-reversal visual evoked potentials and retinal eccentricity. Electroencephalogr Clin Neurophysiol 53:243-253, 1982

17.Srebro R: Localization of visually evoked cortical activity in humans. J Physiol 360:233-246, 1985

18.Ossenblok P, Spekreijse H: The extrastriate generators of the EP to checkerboard onset: a source localization approach. Electroencephalogr Clin Neurophysiol 80:181-193, 1991

19.Brusa A, Mortimer C, Jones SJ: Clinical evaluation of VEPs to interleaved checkerboard reversal stimulation of central, hemiand peripheral fields. Electroencephalogr Clin Neurophysiol 96:485494, 1995

20.Regan D: Human Brain Electrophysiology: Evoked Potentials and Evoked Magnetic Fields in Science and Medicine. New York, NY: Elsevier 1989

21.Fortune B, Hood DC, Johnson CA: Comparison of conventional and multifocal VEPs. (ARVO Abstract No. 2126). Invest Ophthalmol Vis Sci 43: 2002

22.Fortune B, Hood DC: Conventional pattern-reversal VEPs are not equivalent to summed multifocal VEPs. Invest Ophthalmol Vis Sci 2002 (under review)

23.Brindley GS: The variability of the human striate cortex. J Physiol 225:1P-3P, 1972

24.Stensaas SS, Eddington DK, Dobelle WH: The topography and variability of the primary visual cortex in man. J Neurosurg 40:747-755, 1974

25.Rademacher J, Caviness VS Jr, Steinmetz H, Galaburda AM: Topographical variation of the human primary cortices: implications for neuroimaging, brain mapping, and neurobiology. Cereb Cortex 3:313-329, 1993

26.Aine CJ, Supek S, George JS, Ranken D, Lewine J, Sanders J, Best E, Tiee W, Flynn ER, Wood CC: Retinotopic organization of human visual cortex: departures from the classical model. Cereb Cortex 6:354-361, 1996

27.Steinmetz H, Furst G, Meyer BU: Craniocerebral topography within the International 10-20 System. Electroencephalogr Clin Neurophysiol 72:499-506, 1989

28.Hood DC, Zhang X: Multifocal ERG and VEP responses and visual fields: comparing disease-re- lated changes. Doc Ophthalmol 100:115-137, 2000

29.Baseler HA, Sutter EE, Klein SA, Carney T: The topography of visual evoked response properties across the visual field. Electroencephalogr Clin Neurophysiol 90:65-81, 1994

30.Slotnick SD, Klein SA, Carney T, Sutter E, Dastmalchi S: Using multi-stimulus VEP source localization to obtain a retinotopic map of human primary visual cortex. Clin Neurophysiol 110:1793-1800, 1999

31.Slotnick SD, Klein SA, Carney T, Sutter EE: Electrophysiological estimate of human cortical magnification. Clin Neurophysiol 112:1349-1356, 2001

32.Klistorner AI, Graham SL, Grigg JR, Billson FA: Multifocal topographic visual evoked potential: improving objective detection of local visual dield defects. Invest Ophthalmol Vis Sci 39:937-950, 1998

33.Graham SL, Klistorner A, Grigg JR, Billson FA: Objective perimetry in glaucoma: recent advances with multifocal stimuli. Surv Ophthalmol 43(Suppl 1):S199-209, 1999

34.Hood DC, Odel JG, Zhang X: Tracking the recovery of local optic nerve function after optic neuritis: a multifocal VEP study. Invest Ophthalmol Vis Sci 41:4032-4038, 2000

35.Hood DC, Zhang X, Greenstein VC, Kangovi S, Odel JG, Liebmann JM, Ritch R: An interocular comparison of the multifocal VEP: a possible technique for detecting local damage to the optic nerve. Invest Ophthalmol Vis Sci 41:1580-1587, 2000

36.Graham SL, Klistorner AI, Grigg JR, Billson FA: Objective VEP perimetry in flaucoma: asymmetry analysis to identify early deficits. J Glaucoma 9:10-19, 2000

37.Klistorner A, Graham SL: Objective perimetry in glaucoma. Ophthalmology 107:2283-2299, 2000

38.Betsuin Y, Mashima Y, Ohde H, Inoue R, Oguchi Y: Clinical application of the multifocal VEPs. Curr Eye Res 22:54-63, 2001

39.Hasegawa S, Abe H: Mapping of glaucomatous visual field defects by multifocal VEPs. Invest Ophthalmol Vis Sci 42:3341-3348, 2001

40.Goldberg I, Graham SL, Klistorner AI: Multifocal objective perimetry in the detection of glaucomatous field loss. Am J Ophthalmol 133:29-39, 2002

41.Hood DC, Zhang X, Winn BJ: Detecting glaucomatous damage with the mfVEP: how can a monocular test work? J Glaucoma 12(1):3-15, 2003

42.Klistorner AI, Graham SL: Electroencephalogram-based scaling of multifocal visual evoked potentials: effect on intersubject amplitude variability. Invest Ophthalmol Vis Sci 42:2145-2152, 2001