Effects of fixation instability on multifocal VEP (mfVEP) responses in amblyopes

Abstract

The cortical activity of subjects with compromised central vision (e.g., amblyopes) is thought to be much weaker for stimulation of the affected eye than in the fellow eye. Because these subjects are known to exhibit considerable difficulties in keeping steady fixation, we investigated the effects of anomalous fixation on their multifocal visual-evoked potential (mfVEP) responses using a dual Purkinje image (dPi) eye tracker. Our results show that mfVEP responses to stimulation of the central 5 degrees were depressed in the affected eye compared to those in the normal eye and the magnitude of response reductions was proportional to the degree of visual acuity loss in amblyopic subjects. Fixation was far less stable while viewing with the affected eye than with the fellow eye, some exhibiting jerk nystagmus and/or saccadic oscillations. Normal subjects with artificially imposed nystagmus showed similar reductions of VEP responses. The relative magnitudes of the deficits in mfVEP responses were tightly correlated with the degree of fixation instability. These results suggest that the interpretation of anomalous neural or perceptual processing in amblyopic subjects must take the effects of unsteady fixation during measurements into consideration in order to reveal the true nature and extent of sensory neural deficits.

Introduction

Amblyopia is a developmental loss of visual sensitivity resulting from experiencing binocularly discordant images early in life. The most common visual deficit in amblyopia is a loss of contrast sensitivity for mid to high spatial-frequency stimuli in the central vision of the affected eye, along with reduced optotype visual acuity (Hess & Howell, 1977; Levi & Carkeet, 1993; Levi & Harwerth, 1977). Amblyopic subjects also exhibit “crowding,” reduced positional acuity, and deficits in “higher-order” visual functions such as contour integration, motion integration, and second-order detection (Chandna, Pennefather, Kovacs, & Norcia, 2001; Hess & Baker, 1984; Hess & Holliday, 1992; Hess & Howell, 1977; Hess, Wang, Demanins, Wilkinson, & Wilson, 1999; Hess & Watt, 1990; Levi & Harwerth, 1982; Levi & Klein, 1982; Levi & Klein, 1985, 1986; Levi & Polat, 1996; Polat, Ma-Naim, Belkin, & Sagi, 2004; Wong & Levi, 2005). 

Because the visual brain of infants exhibits robust plasticity shortly after birth, the developing functional connections in visual cortex are unstable and readily disrupted by experiencing binocularly discordant vision. However, despite many interesting theories of amblyopia and numerous neurophysiological investigations with animal models, it is still not clearly understood how and where in the visual brain the functional connections are altered, resulting in amblyopia (Chino, Bi, & Zhang, 2003; Kiorpes, 2006; Kiorpes & McKee, 1999; Kiorpes & Movshon, 2003). The most common explanation has been that a reduced proportion of the primary visual cortex (V1) neurons are effectively driven by the amblyopic eye, in particular those V1 neurons tuned to higher spatial frequencies, i.e., “undersampling” (Crewther & Crewther, 1990; Kiorpes, Kiper, O'Keefe, Cavanaugh, & Movshon, 1998; Levi & Klein, 1986). However, in primates the previously reported loss of V1 neurons tuned to high spatial frequencies appears to be of marginal magnitude, and to occur in a very limited number of amblyopic monkeys (Kiorpes et al., 1998). Additionally, dramatic ocular dominance shifts in V1, common to monocular form-deprivation amblyopia, are largely absent or smaller in strabismic or anisometropic amblyopes (Hubel & Wiesel, 1977; Kiorpes et al., 1998). As a result, an emerging view is that the major neural limits imposed on the developing primate visual system by experiencing early strabismus or anisometropia and resulting in amblyopia may reside beyond V1 (Chino et al., 2003; Kiorpes, 2006; Kiorpes et al., 1998; Kiorpes & McKee, 1999; Kiorpes & Movshon, 2003; Levi, 2006). 

In human amblyopes, mfVEP responses to central stimulation of the amblyopic eye were reported to be substantially depressed (Zhang & Zhao, 2005). These deficits in amblyopes were interpreted as evidence for anomalous alterations in V1 because mfVEP signals in normal humans are thought to originate primarily from the cortical areas adjacent to the calcarine fissure, i.e., V1 (Fortune & Hood, 2003; Hood et al., 2006; Slotnick, Klein, Carney, Sutter, & Dastmalchi, 1999; Zhang & Hood, 2004). The mfVEP technique employs an array of independently flickering target patches, and the central field is tested with targets near the acuity limit. For this reason, reliable testing with the mfVEP technique requires accurate fixation, and signals are much reduced when subjects fixate away from the center (Menz, Sutter, & Menz, 2004). Although amblyopes are known to exhibit considerable difficulties in keeping steady fixation (Bedell & Flom, 1985; Bedell, Yap, & Flom, 1990; Ciuffreda, Kenyon, & Stark, 1979; Schor & Flom, 1975; Schor & Hallmark, 1978; Schor & Westall, 1984; Westall & Aslin, 1984), there has not been any systematic, quantitative analysis of their fixation patterns in those amblyopic subjects who served in mfVEP experiments. 

To investigate how much of the cortical deficits revealed by mfVEP recording experiments might be explained by instability of fixation during the recording session, we compared mfVEP response amplitudes to fixation instability measurements in strabismic or anisometropic amblyopes while viewing the visual stimuli used for the mfVEP recordings. For comparison, we measured multifocal electro-retinogram (mfERG) responses in these amblyopic subjects because ERG amplitudes are likely to be influenced by fixation instability but not by neural deficits in retina (Arden, Vaegan, Hogg, Powell, & Carter, 1980; Arden & Wooding, 1985; Delint, Weissenbruch, Berendschot, & Norren, 1998; Hendrickson et al., 1987; Hess, Baker, Verhoeve, Keesey, & France, 1985; Kiorpes, 2006; Kiorpes & McKee, 1999). Finally, the sole effects of unsteady fixation on mfVEP and mfERG responses were determined for normal control subjects while simulating unsteady fixation with a “nystagmus simulator” (Bedell, 2000). We found that unstable fixation in amblyopia and in simulated nystagmus greatly reduced both mfVEP and mfERG responses of the affected eye, particularly to stimulation within the central five degrees. These response reductions were well correlated with the magnitude of fixation instability of individual subjects. While our results do not specifically rule out V1 as a major locus of processing deficits in amblyopia, they point to an important factor that is often ignored in studies of the neural basis of amblyopia or similar spatial vision deficits. 

Methods

Subjects

Six amblyopes (four strabismic and two anisometropic) who received comprehensive examinations of their visual capacities served as subjects ( Table 1). Two young adults served as normal subjects. Our experimental procedures followed the tenets of the Declaration of Helsinki and informed consent was obtained from the subjects after explanation of the nature and possible consequences of the study. The experimental methods were approved by the IRB of the University of Houston. 

Table 1

 

List of subject type, visual acuity for each eye and interocular difference of visual acuity.

Subject	Type	Visual acuity (logMar)		Interocular difference	Notes
		Amblyopic eye	Fellow eye
1	Anisometropia	0.06	−0.12	0.18
2	Esotropia	0.06	−0.07	0.13	Intermitten, 4 prism diopter
3	Esotropia	0.06	−0.12	0.18	12 prism diopter
4	Esotropia	0.48	0	0.48	2 prism diopter
5	Anisometropia	0.24	0	0.24
6	Esotropia	0.48	0	0.48	10 prism diopter
7	Normal	0	0	0	0.10 at 2 Hz, 0.48 at 4 Hz
8	Normal	0	0	0	0.12 at 2 Hz, 0.46 at 4 Hz

mfVEP procedures

The mfVEP recording was made with the VERIS 5.1 system (Electro-Diagnostic Imaging, San Mateo, California, USA). The stimulus was a 46.2° diameter dartboard pattern consisted of 60 sectors. Each with 16 checks (8 white and 8 black) scaled for cortical magnification and had a mean luminance of 76 cd/m ² and Michelson contrast of 95% ( Figure 1A). The black and white checks in each sector reversed contrast according to a pseudorandom m-sequence at a frame rate of 75 Hz (Sutter, 1991). Subjects viewed through their natural pupils with appropriate refractive corrections in place and were instructed to maintain fixation at the stimulus center (marked as an “x”). Responses in three signal channels, all referenced to the inion with active electrodes placed at 4 cm above the inion, and 1 cm up and 4 cm lateral to the inion on either side, were recorded in real time and responses in three additional channels were derived from offline analyses (Figure 1B) (Hood & Greenstein, 2003). Data collection began when stimulus presentation started, and no special period of adaptation to the stimulus display was given. Two runs were completed for each eye in an R–L–L–R sequence always beginning with the right (R) eye. The “non-tested” eye was patched. All analyses were based on the second-order kernel best channel (i.e., the channel with the highest signal-to-noise ratio) responses using customized software by Hood and his colleagues (Fortune, Zhang, Hood, Demirel, & Johnson, 2004; Hood & Greenstein, 2003). For each subject, the interocular response amplitude at each sector was represented as the ratio of the root mean square (RMS) amplitude of a signal window (45–150 ms) between the two eyes and compared to the values from a normal database (100 normal subjects from Devers Eye Institute, Portland, OH). Based on the statistics, each sector was classified into: no difference between the two eyes (black symbols); one eye different from the other at p < 0.01 level (e.g., saturated colors in Figures 5D–5F) or p < 0.05 level (e.g., desaturated color in Figures 5D–5F). 

Figure 1

 

Visual stimulus and representative responses of mfVEP. (A) Dartboard pattern (60 sectors) scaled for cortical magnification and flickering in m-sequence. (B) Examples of mfVEP response traces from the left (red) and right eyes (blue) of a normal subject. The visual field was divided into three eccentricity rings.

mfERG procedures

Following pupil dilation (1% tropicamide and 2.5% phenylephrine hydrochloride), mfERGs were recorded differentially between the two eyes using Dawson–Trick–Litzkow (DTL) electrodes with the forehead served as the ground (Dawson, Trick, & Litzkow, 1979). Each DTL fiber was anchored with a dab of petroleum jelly near the inner canthus and electrically connected by clip leads at the outer canthus. One of the standard stimuli of the VERIS 5.1 software was used (Sutter & Tran, 1992), which consisted of 103 white (127.8 cd/m²) and black (2.9 cd/m²) hexagons scaled for eccentricity. The display extended 46° horizontally and 42° vertically at a viewing distance of 36 cm (Figure 2A). At a frame rate of 75 Hz, each hexagon reversed in contrast following a pseudorandom m sequence (Sutter, 1991). The m-sequence chosen had 2¹⁵–1 steps and took about 7 minutes to complete. One 7-minute run was recorded for each eye beginning with the right eye, and each run was broken into 32 segments to allow for blinking. During recording, subjects were optically corrected for the viewing distance and were instructed to maintain fixation at the “X” of the stimulus center. Pupil positions were monitored with a CCD camera. Signals were amplified, sampled at 1200 Hz, and band-passed filtered at between 10 and 300 Hz. First-order kernel responses were derived with VERIS software and averaged across the 5 retinal eccentricity rings (Figures 2B and 2C). 

Figure 2

 

Visual stimulus for mfERG and representative responses. (A) Pattern stimulus with 103 hexagons was used for visual stimulation. The entire visual stimulus covers up to 25 degrees of visual field and was divided into 5 eccentricity rings. Ring 1: Red. Ring 2: Orange. Ring 3: Dark Blue. Ring 4: Light Blue. Ring 5: Gray. (B) Examples of the first-order responses of a normal subject to stimuli presented for the 5 rings in Panel A. (C) The response traces were divided into corresponding 5 eccentricity rings and averaged. Note that the scaling of patches is different from that used for mfVEP and the central patch is substantially larger. Also how the peak-to-peak measurements were made is illustrated in the top trace.

Fixation stability

The subjects' fixational eye movements were measured using a dual Purkinje image eye (dPi) tracker as they fixated monocularly on the target identical to the mfVEP stimuli with a small red fixation point at the center of the stimuli. Each eye was measured individually. Eye movements were not measured concurrently with the mfVEP recording since it was impractical to use Veris 5.1 system along with the dPi eye tracker. Eye movements were recorded in 8-second epochs over a period of 20 minutes, and subjects were allowed to blink normally during the measurements. The output of the eye tracker was sampled digitally at 120 Hz and stored for offline analysis. During the offline analysis, blinks were manually located and deleted from further analysis. The standard deviation of the remaining fixational traces was calculated and compared with VA and mfVEP/mfERG amplitudes ( Figure 3). 

Figure 3

 

Examples of normal and abnormal fixation during monocular viewing. Top panel, representative traces showing the retina image positions over 8 seconds of recording while fixating with the normal control eye (A), control eye with 2 Hz artificial nystagmus (B), and amblyopic eye (C). Notice the jerk nystagmus and/or saccadic oscillations. Traces in the second panel show the corresponding fixation pattern while viewing with the fellow eye. Bottom panel. The distribution of retinal image positions through many trials for the normal control eye (D), control eye plus 2 Hz nystagmus (E), and amblyopic eye (F). The normal eye exhibited a narrow distribution of retinal image positions where at least 80% were located within 0.5 degree from the fixation point during the 8 seconds of recording. The amblyopic eye and normal eye with artificial nystagmus showed broader distributions of retinal image positions. Note that eye motion in controls with the nystagmus simulator was subtracted from simulated nystagmus motion in order to determine residual retinal image jitters.

Simulation of nystagmus

Two control subjects with normal fixation patterns were used in the experiment. To simulate the latent nystagmus that is typically observed in amblyopic subjects, we optically induced a jerk waveform with a motorized Risley prism (4 Hz/1.3 degs and 2 Hz/0.7 degs) while measuring mfVEP/mfERG responses, and the response amplitudes were then compared to the retinal image motion during the simulated nystagmus of the same amplitude and frequency. Also we used the stimulator tubes of the dPi tracker to generate the jerk nystagmus waveform and measured eye movements while subjects fixated a target identical to the mfVEP. Retinal image motion was then calculated by subtracting the eye motion from the target motion. 

Results

Fixation stability

To examine the possibility that much of the mfVEP response deficits in amblyopes may be explained by unsteady fixation during mfVEP recordings, we measured the residual eye movements during monocular fixation for each eye using our dual Purkinje image eye tracker. Figure 3 illustrates the representative 8-second traces of eye positions for each eye for a normal control without the simulator, a control with the 2-Hz/0.7-deg simulator, and a strabismic amblyope. In both eyes of normal controls without the nystagmus simulator, the distribution of eye positions, thus signifying retinal image positions, was narrow and at least 80% were located within 0.5 degrees from the mean fixation point during the period of 8 seconds of measurement. The standard deviation was 0.10 degrees for both eyes ( Figures 3A and 3D). 

The control subject with 2 Hz/0.7 deg simulator exhibited jerk nystagmus of 2 Hz and 0.7 deg in amplitude ( Figure 3B). In these subjects with the simulator, their eye position motion was subtracted from simulated nystagmus motion in order to determine their residual retinal image jitters. The distribution of their retinal image positions was much broader than that for normal controls without the simulator, and the standard deviation of retinal image positions for the manipulated eye was 0.34 degrees compared to 0.11 degrees for the fellow eye ( Figure 3E). Interestingly, the distribution of retinal image positions in a strabismic amblyope was even broader, and the standard deviation was 0.50 degrees for the affected eye compared to 0.13 degrees for his fellow eye ( Figures 3C and 3F). 

In 4 out of 6 amblyopic subjects, the standard deviations of retinal image positions were between 0.3 and 0.5 degrees for the affected eye and 0.08 and 0.18 degrees for the fellow eye ( Figure 4). These values were similar to the standard deviations in the distribution of retinal image positions for the manipulated eye in control subjects with nystagmus simulators, i.e., between standard deviations for 2 Hz/0.7 deg and 4 Hz/1.3 deg ( Figure 4). Interocular differences in retinal image positions for these 4 amblyopic subjects were statistically significant (paired t-test, p < 0.01). 

Figure 4

 

Fixation stability for all subjects. Figure shows the standard deviations of the distribution of retinal image positions. Filled circles represent the amblyopic eyes and filled squares represent the normal control eyes with artificial nystagmus. Open symbols represent the fellow eyes of amblyopic patients or the normal eyes of control subjects.

Two amblyopic subjects (Aniso-2 and Strab-4) exhibited unsteady fixation in both eyes. Interestingly, in Aniso-2 the standard deviations for the distribution of retinal image positions were nearly identical for the two eyes but were similar to those for the affected eye of the first 4 amblyopic subjects (i.e., SD = 0.3 deg). In ST-4, the fixation instability was exceptionally large in both eyes ( SD = 1.2 deg for the amblyopic eye and 0.75 deg for the non-amblyopic eye). Although the standard deviation for the non-amblyopic eye was higher than that in other amblyopic subjects, it was still smaller than that for the amblyopic eye. 

mfVEP

Representative mfVEP responses and interocular probability plots are illustrated in Figure 5 for an amblyopic subject (Strab-3), a normal subject with 2.0 Hz/0.7 deg artificial nystagmus, and a normal control. The mfVEP responses to stimulation of the affected eye of this amblyopic subject were severely depressed for the great majority of sectors within the first two rings representing the central 5° ( Figure 5C, red) compared to those for the fellow eye (blue). Consequently, the interocular probability plot for this subject demonstrates that the differences in amplitude between the affected and the fellow eye were significant in the great majority of sectors within the central 5 degrees ( Figure 5F). However, we did not find systematic differences in the waveform of mfVEP responses between the amblyopic versus non-amblyopic eye, perhaps because we always analyzed the best-channel responses. These observations in the amblyopic subjects were in sharp contrast to the comparable data from the normal controls where there were no significant interocular differences in response amplitudes ( Figures 5A and 5D). 

Figure 5

 

Examples of mfVEP responses (top) and interocular probability plot (bottom). Top panels, the mfVEP response traces from both eyes are shown for the normal control subjects (A), normal subjects with 2 Hz artificial nystagmus (B), and the amblyopic subjects (C). Red traces represent left eye (A), normal control eye with artificial nystagmus (B), and amblyopic eye (C), respectively. Blue traces show the corresponding traces for the fellow eyes. Bottom panel, interocular probability plots for the normal subjects (D), normal subjects with 2 Hz nystagmus (E), and amblyopic subjects (F). Majority of the sectors showing significant interocular differences are located within the central 5 degrees.

In a normal subject wearing a nystagmus simulator, the mfVEP responses to stimulation of the manipulated eye were also severely depressed compared to the fellow eye ( Figures 5B and 5E). In this measurement, the degree and the frequency of artificial nystagmus were maintained at 0.7 degrees and 2.0 Hz, within the range of nystagmus that was exhibited by the amblyopic subjects of this study. The significant interocular differences in mfVEP amplitudes were confined to the most central 12 sectors (i.e., the first ring) representing the central 5 degrees. 

The summary plot in Figure 6 shows that five amblyopic subjects (Aniso-1 and Strab-1, Strab-2, Strab-3, and Strab-4) exhibited a relatively large proportion of sectors having significant interocular differences in response amplitude for the central 5 degrees (repeated measures ANOVA, ring1 vs. ring2 F = 11.51, p = 0.0145; ring1 vs. ring3 F = 20.57, p < 0.01). Moreover, in Strab-2 and Strab-3, over 20% of sectors outside the first ring, i.e., 5–20 central degrees showed significant interocular differences in mfVEP amplitudes. Similarly in both normal subjects with artificial nystagmus of small magnitude (2 Hz, 0.7 degrees), significant interocular differences were found in about 60% of the sectors within the central 5 degrees, whereas only about 20% of all the sectors between 5 and 20 degrees showed significant interocular differences. With larger simulated nystagmus (4 Hz/1.5 deg), 100% of all sectors within the central 5 degrees and 40–70% of sectors between 5 and 20 degrees in both subjects, and in one of the control subjects, 35% of the sectors outside of the central 20 degrees showed significant interocular differences in mfVEP amplitudes. Interestingly, for the only amblyopic subject (Aniso-2) that did not have interocular differences in the distribution of retinal image positions, although both eyes exhibited unsteady fixation ( Figure 4), mfVEP responses were similar between the amblyopic and the fellow eye. Taken together, these mfVEP data for amblyopic subjects agree well with results on their fixation stability (compare Figure 4 with Figure 6). 

Figure 6

 

Plots showing the percentage of sectors having significant interocular differences in three retinal eccentricity rings. Filled circles represent the amblyopic eyes and filled squares represent the normal control eyes with artificial nystagmus.

mfERG

It is widely agreed that neural alterations responsible for amblyopia occur beyond the lateral geniculate nucleus (LGN), i.e., in visual cortex (Delint et al., 1998; Hendrickson et al., 1987; Hess et al., 1985; Kiorpes, 2006; Kiorpes & McKee, 1999). Thus, the mfERG responses of our amblyopic subjects are anticipated to be “normal.” However, if the fixation of the amblyopic eye of our subjects is not stable during measurements as demonstrated above, unsteady fixation is likely to disrupt mfERG responses to stimulation of the central few degrees. We examined this possibility by recording mfERG in our amblyopic subjects, and the results were compared to those obtained in normal control subjects with and without artificial nystagmus. 

There were no interocular differences for the normal controls with steady fixation ( Figures 7A and 7D). However, the representative traces of mfERG responses and the peak-to-peak amplitudes for the amblyopic and non-amblyopic eyes for different rings show that there were substantial interocular differences in amplitudes for stimulation within the first eccentric ring (covering the central 5°) in the amblyopic subject (Strab-3) ( Figures 7C and 7F) and the normal control with artificial 2 Hz/0.7 deg nystagmus ( Figures 7B and 7E). 

Figure 7

 

Examples of mfERG responses (top) and interocular difference plot (bottom). Top panels show the response traces from both eyes of the normal control subject (A), normal subjects with 2 Hz/0.7 deg artificial nystagmus (B), and amblyopic subjects (C). Bottom panels show how the response amplitude of each eye and the interocular differences in response amplitudes varied with retinal eccentricity for the normal control subjects (D), normal subjects with one eye having 2 Hz/0.7 deg nystagmus (E), and amblyopic subjects (F).

Figure 8A illustrates the average amplitudes of mfERG responses in the amblyopic eye and the fellow eye for the 5 rings for our amblyopic and normal control subjects with 2 Hz/0.7 deg and 4 Hz/1.3 deg nystagmus simulators and without simulators. The response reductions in the amblyopic eye relative to the fellow eye were largely confined to Ring 1 (repeated measures ANOVA, amblyopic eye vs. fellow eye, ANOVA, F = 66.37, p < 0.01; ring effect, ring1 vs. ring2, F = 6.2, p = 0.05, ring1 vs. ring3, 4, or 5 F = 14.8, p = 0.015; F = 24.25, p < 0.01, F = 21.28, p < 0.01) except in Strab-3 where there was a considerable interocular difference for Ring 2. Curiously, in one strabismic amblyope (Strab-4), the mfERG responses for both the amblyopic and fellow eyes were virtually identical and were not different from controls without simulators. Fixation in this subject was exceptionally poor in both eyes due to slow drifts of fixation in addition to jerky nystagmus and/or saccadic oscillations during the 8-second measurement period, but the relative interocular differences in the retinal image jitters were smallest among all strabismic subjects. 

Figure 8

 

mfERG response and normalized interocular differences in the response amplitudes. (A) mfERG response amplitudes for both eyes are shown, triangles represent amblyopic eyes or control eyes with artificial nystagmus. Circles represent the fellow eyes in amblyopic subjects and squares represent the normal eyes in control subjects. The shades of the symbols represent the retinal eccentricity with darker ones close to central retina. (B) Normalized interocular difference in mfERG response amplitude retinal eccentricity. Circles represent amblyopic subjects and squares represent control subjects with artificial nystagmus.

In both control subjects with the nystagmus simulator, there were consistent, sizable interocular differences in all rings favoring the control eye (i.e., without the simulator). However, the absolute size of amplitude reductions in the manipulated eye was largest for the first two rings in both subjects. Also the interocular differences were greater with the 4-Hz/1.3-deg simulator than the 2-Hz/0.7-Hz simulator ( Figure 8B). More importantly, the magnitude of interocular differences in mfERG amplitudes for the first two rings in control subjects with both simulators was similar to that in our amblyopic subjects. 

Discussion

Unsteady fixation in amblyopic subjects

Abnormal fixation patterns and ocular motility of amblyopes have been routinely detected in clinical settings, and unsteady fixation in amblyopic subjects has been well investigated with a variety of quantitative methods (Bedell & Flom, 1985; Bedell et al., 1990; Ciuffreda et al., 1979; Schor & Flom, 1975; Schor & Hallmark, 1978; Schor & Westall, 1984; Westall & Aslin, 1984). For example, Westhall and Aslin (1984) found that the deviations in eye positions from the fixation point for the amblyopic eye was far larger than that for the non-amblyopic eye, and that introducing a patterned background did not improve the fixation of the affected eye. Our results are consistent with this and other earlier observations. 

However, to our knowledge, the present study represents the first quantitative investigation of variability in eye positions of the amblyopic eyes using a dual Purkinje image eye tracker while the amblyopic subjects were viewing the stimuli used for the mfVEP recordings. Moreover, this is also the first attempt to measure the fixation patterns of “normal” subjects wearing nystagmus simulators, and this manipulation gave us a precise estimate of the “pure” effects of unsteady fixation on acuity, mfVEP, and mfERG. It is worth noting, however, that the retinal image movements that occur under simulated nystagmus are somewhat different in waveform from those that occur in amblyopes. Firstly, amblyopes can exhibit a variety of waveforms of nystagmus. Our simulated nystagmus was designed to match typical behavior but did not exactly mimic an individual's waveform. Secondly, the image motion imposed by the nystagmus simulator drives an OKN response that partially negates the image motion. The retinal image motion is therefore the difference between the simulated nystagmus waveform and the eye movement response to it and can be a fairly complicated waveform. The important consequence of both real and simulated nystagmus is the loss of retinal locus specificity in stimulus presentation, and the exact waveform is probably less important than the overall variability of eye position as described here by RMS error. 

There is considerable evidence to suggest that neural signals recorded with mfVEP methods largely represent the electrical activity originating from the population of neurons in the primary visual cortex (V1) (Fortune & Hood, 2003; Hood et al., 2006; Slotnick et al., 1999; Zhang & Hood, 2004). This means that the responses of V1 neurons viewing the central 5 degrees with small receptive-field centers (RFC) (0.2°–1.5° in monkeys) and large suppressive RF surrounds (about 4° in monkeys) (Cavanaugh, Bair, & Movshon, 2002; Sceniak, Hawken, & Shapley, 2001; Sceniak, Ringach, Hawken, & Shapley, 1999; Zhang et al., 2005) could be severely affected by a relatively small fixation instability observed in this study. Also with substantial shifts in eye positions, small central checkerboard stimuli may land on perifoveal regions where cells with larger RFCs are abundant and these cortical neurons do not respond well to small stimuli. Our VEP/ERG data unambiguously confirm this prediction, and our results can be readily explained without invoking major developmental alterations of receptive-filed properties in V1 neurons that are driven by the affected eye. 

In a related matter, early investigations on the neural basis of strabismic amblyopia in anesthetized and paralyzed animals were able to avoid major problems with unsteady fixation. However, these studies have given us, at best, inconsistent results, partly because these studies used non-primate species for their animal models and/or often lacked behavioral documentation of amblyopia (Crewther & Crewther, 1990; Hubel & Wiesel, 1965; Van Sluyters & Levitt, 1980). Moreover, developmental alterations in the spatial response properties of V1 neurons in monkeys due to early strabismus that have been demonstrated thus far are of very small magnitude (Kiorpes et al., 1998; Kiorpes & Movshon, 2003) or all together absent (Watanabe et al., 2005), and dramatic ocular dominance shifts away from the affected eye in V1, common to amblyopia resulting from monocular form deprivation or severe anisometropia, is largely absent in strabismic monkeys that developed amblyopia (Hubel & Wiesel, 1965, 1977; Smith et al., 1997; Watanabe et al., 2005). 

Relationships between acuity, mfVEP, mfERG, fixation stability deficits in amblyopes

Close examinations of all of our results together give several interesting insights into the functional anomalies associated with strabismic and anisometropic amblyopia. In the summary diagram of Figure 9, interocular differences in mfVEP/mfERG responses to central 5 degree stimulation and fixation stability of each subject are plotted as a function of interocular differences in their visual acuity after normalizing each data point against the respective value obtained in control subjects with 4 Hz artificial nystagmus (i.e., those subjects that exhibited largest deficits in nearly all measures). The relative magnitude of mfVEP deficits in amblyopes was well correlated with interocular differences in fixation stability and in visual acuity of these subjects ( r > 0.8, p < 0.01). Thus, the observed mfVEP deficits in amblyopes may be largely attributed to their unsteady fixation during the recording, and if developmental anomalies in V1 do exist, they may have been masked by the effects of such unsteady fixation during the recording. 

Figure 9

 

Relationships between fixation instability, mfVEP, and mfERG. In all subjects, interocular differences in the mfVEP responses of the central 5 degrees, the mfERG responses of the central 5 degrees, and the standard deviations of the eye positions were normalized against the respective values for normal controls with 4 Hz nystagmus. These normalized values from different measurements for each subject were plotted against the interocular difference in visual acuity (logMar) of that subject. Data for Aniso-2 and Strab-4 are illustrated but not included in calculation of correlations because these subjects exhibited qualitatively different fixation patterns from other experimental subjects.

Interestingly, interocular differences in mfERG were also proportional to fixation instability and acuity. This result in mfERG appears to suggest that the retinal functions of our amblyopes may be abnormal as several earlier ERG studies in human amblyopes reported (Arden et al., 1980; Arden & Wooding, 1985; Sokol & Nadler, 1979). However, more recent studies in amblyopic monkeys and humans found that the retinal structures and the functions of the amblyopic eyes are indistinguishable from those in normal primates (Delint et al., 1998; Hendrickson et al., 1987; Hess et al., 1985; Kiorpes, 2006; Kiorpes & McKee, 1999). Taken together, the deficits in mfERG observed in this study are also likely to have resulted from unsteady fixation in our amblyopes during the mfERG measurements. A similar conclusion was reached in a previous study of pattern ERG in amblyopes (Hess et al., 1985), although these investigators did not directly record the variability in eye positions of their subjects. 

The relative magnitude of the mfVEP deficits in amblyopes or in controls with 2 Hz/0.7 deg artificial nystagmus was 10 to 31 % larger than their mfERG deficits although the size of both mfVEP and mfERG deficits generally paralleled the degree of their fixation instability and acuity ( Figure 9). The observed differences between retina and cortex may have resulted from a simple “cortical magnification” of mfERG anomaly caused by fixation instability, small differences in visual stimuli between mfERG and mfVEP, and/or the mfVEP responses representing the central few degrees may be more sensitive to unstable fixation during measurements. Alternatively, the differences between mfERG and mfVEP deficits in amblyopes may reflect a true developmental change that is unique to V1. This issue could be satisfactorily resolved if the fixation of amblyopes is reasonably stabilized during mfVEP measurements. 

Conclusions

Although amblyopic subjects are known to exhibit anomalous responses to central visual stimulation in psychophysical tasks, mfVEP, and fMRI, the interpretation of anticipated results must take the effects of unstable fixation during measurements into consideration in order to reveal the true nature and extent of sensory neural deficits in amblyopes resulting from early abnormal visual experience. This critical point may be generalized to similar studies where central vision is compromised and fixation stability is a potential issue. Finally, although our result may not rule out V1 as a major locus of processing deficits in amblyopia, the relatively tight correlation between the severity of mfVEP deficits and the degree of fixation instability may favor an emerging view that the major neural limits imposed on strabismic and/or anisometropic amblyopes appear to reside beyond V1 (Chino et al., 2003; Kiorpes, 2006; Kiorpes et al., 1998; Kiorpes & McKee, 1999; Kiorpes & Movshon, 2003; Levi, 2006). 

Acknowledgments

Supported by grants from the National Eye Institute (EY-08128, RR-07146), the NEI Summer Research Program, and the Benedict-McFadden Professorship (YMC). We thank Dr. Laura Frishman for the use of her mfVEP/ERG facility, Dr. Don Hood for customized mfVEP software, and Dr. Harold Bedell for providing us with a nystagmus simulator. We also thank Nickki Polnik, Lance Wright for data acquisition, and Hope Queener for programming. 

Commercial relationships: none. 

Corresponding author: Yuzo M. Chino. 

Email: [e-mail]. 

Address: College of Optometry, University of Houston, Houston TX 77204-2020, USA.