Abstract
Purpose.:
In uncorrected anisometropia, protracted dichoptic stimulation may result in interocular inhibition, which may be a contributing factor in amblyopia development. This study investigates the relationship between interocular interactions and anisometropic amblyopia.
Methods.:
Three visual functions (low-contrast acuity, contrast sensitivity, and alignment sensitivity) were measured in the nondominant eye of 44 children aged 5 to 11 years: 10 with normal vision, 17 with anisometropia without amblyopia, and 17 with anisometropic amblyopia. The dominant eye was either fully or partially occluded. The difference in nondominant eye visual function between the full-and partial-occlusion conditions was termed the interaction index. The index of each visual function was compared between subject groups. A higher index indicates stronger inhibition of nondominant eye function with partial occlusion of the dominant eye. Amblyopic children had 6 months of therapy (refractive correction and occlusion), and the reduction in interocular difference in high-contrast acuity was regarded as the treatment outcome. The relationships of the interaction index with the degree of anisometropia, the severity of amblyopia, and the treatment outcomes were examined.
Results.:
The acuity interaction index was significantly higher in anisometropic children with amblyopia than in those without (P = 0.003). It was positively correlated with the degree of anisometropia (r s = 0.35, P = 0.042) and the amblyopic treatment outcomes (r s = 0.54, P = 0.038). No such difference or association was found between the contrast sensitivity or alignment sensitivity interaction index and anisometropic amblyopia.
Conclusions.:
Interocular interactions are associated with amblyopia, the degree of anisometropia, and amblyopia treatment outcomes, but these associations are visual function dependent.
In the visual system, monocular signals interact in several ways during the processing that underpins sensory fusion. Interocular interaction of this kind can be classified broadly into two categories: excitatory and inhibitory. Studies have demonstrated that visual performance with two eyes exceeds that of the better eye when identical images are presented to each eye.
1,2 This phenomenon is termed binocular summation,
3 which is an example of an excitatory interaction. However, such apparent summation is not always obtained, particularly if the stimulation of each eye occurs at different times or differs in spatial detail.
4 When stimuli presented to each eye differ in luminance, chromatic, and/or spatial properties, perception may alternate between the two eyes (exclusive dominance), or may be a “patchwork” continuously changing over time (mosaic dominance), a phenomenon known as binocular rivalry,
5 and an example of inhibitory interaction. Independent stimulation of each eye is known as dichoptic stimulation (or dichoptic masking), and the resulting effects are underpinned by interocular interactions. Studies using dichoptic stimulation have shown that visual functions of one eye can be improved (e.g., summation) or reduced (e.g., rivalry) with a masking stimulus presented to the fellow eye.
6 –10
Uncorrected anisometropia causes blurred visual input to one eye in particular. During visual development, this may result in amblyopia, which occurs in approximately 3% of the adult population,
11 with deficits in a range of visual functions including optotype acuity, contrast sensitivity, vernier acuity, and binocularity.
12 –15 In some types of anisometropia, such as spherical myopic anisometropia, monovision can be achieved, with the less ametropic eye used for distant vision, whereas the more ametropic eye is used for near vision. In these observers, neither eye is relatively disadvantaged (unless the more ametropic eye has extremely high refractive error), and amblyopia is unlikely to develop. It is possible, however, that binocularity is abnormal, and there may be a lack of interocular interaction, because the monocular images differ. In uncorrected hyperopic anisometropia, the less ametropic eye may be able to achieve a clear retinal image in distant vision, but the more ametropic eye receives relatively blurred input and may never have a sharp image focused on its retina. This is a form of dichoptic stimulation and may lead to inhibition of the more ametropic eye by the fellow eye and the development of amblyopia.
The severity of amblyopia correlates positively with the degree of anisometropia,
16 but exceptions have been reported.
17 Thus, there may be other factors that influence the development of anisometropic amblyopia, and these factors may be key to understanding this development. In normal visual systems, interocular interactions may differ with different amounts of monocular defocus. Binocular summation may occur with a fogging lens of equal to or less than +1.50 DS in front of one eye, but inhibition may occur at higher powers.
18,19 It is possible that these types of interaction also occur in uncorrected anisometropia, but whether there is a relationship between them and anisometropic amblyopia is unknown. Previous studies have found that interocular interactions differ between amblyopic and normal observers. However, high interindividual variation has been reported in these interactions in animals and humans with amblyopia.
8–9,20
Thus, previous work has enhanced understanding of the way in which stimulation (or occlusion) of one eye may affect vision in the fellow eye in the normal visual system, and it has been established that interocular interactions are abnormal in amblyopia. The present study was conducted to improve understanding of the role (if any) of interactions in the development of anisometropic amblyopia, by examining (1) whether these interactions differ in children with normal vision and anisometropic children with or without amblyopia; (2) the relationship between these interactions, the degree of anisometropia, and the severity of amblyopia; and (3) the relationship between these interactions and the response to amblyopia therapy.
Visual stimuli were generated with a graphics card (VSG 2/5; Cambridge Research Systems, Cambridge, UK) externally connected to a laptop (HP 8530P; Hewlett Packard, Palo Alto, CA) and were displayed on a gamma-corrected 21-in. cathode ray tube monitor (Trinitron GDM-F520; Sony, Tokyo, Japan). The refresh rate was 120 Hz. The stimuli were viewed through Ferro-electric shutter goggles (FE-1; Cambridge Research Systems). The goggles were worn using an elasticized strap and were held in place by an assistant to reduce their weight and to minimize discomfort. For anisometropic children with and without amblyopia, refractive error was always corrected using trial lenses. Three visual functions were measured: (1) low-contrast acuity (20% Weber contrast when viewing through the goggles); (2) contrast sensitivity; and (3) alignment sensitivity. The mean room illuminance was 4.78 ± 2.76 lux (Konica T-10 illuminance meter; Minolta, Tokyo, Japan).
Each visual function was measured in the full- and partial-occlusion conditions. In the full-occlusion (monocular) condition, an opaque eye patch was used to cover the nontested eye. In the partial-occlusion condition, a square central patch (78% Weber contrast) was used as a partial occluder (see below) presented at the center of the monitor and visible to the nontested eye only. The goggles were worn in both viewing conditions and were synchronized with the monitor so that alternate frames were presented to each eye (e.g., odd-numbered frames to right eye, even-numbered frames to left eye). Thus, each eye viewed the stimuli (or occlusion) at a refresh rate of 60 Hz. The background luminance of the monitor was fixed at 170 cd/m2, and this level was reduced to approximately 21 cd/m2 at each eye when viewing through the goggles.
Acuity was measured using a single letter E constructed in a 5 × 5 grid, in which each stroke and gap was one fifth of the dimension of the square grid. The letter was presented to the tested eye only at one of four possible orientations on each trial (right, left, up, or down). The square partial occlusion was presented to the nontested eye in the partial occlusion condition only. A fusion lock (a ring target at 78% Weber contrast) with a width of 0.1° was constantly presented to both eyes to ensure that both eyes were in alignment in the partial-occlusion condition. In the full-occlusion condition, the fusion lock was visible by the tested eye only, because the fellow eye was occluded. Suppression markers were four lines, with two lines presented to each eye at a peripheral location (
Table 2). They were used in the partial-occlusion condition only, to check for suppression of either eye.
Table 2. Stimulus Parameters and Staircase Procedures Used in the Main Experiment
Table 2. Stimulus Parameters and Staircase Procedures Used in the Main Experiment
| Test | Stimulus Parameters (deg) | Staircase Procedures | Step Size |
| Partial Occlusion Size | Fusion Lock Size | Suppression Markers | Determination of Individual Start Levels | Determination of Thresholds |
| Location | Size | Staircase Rules | Termination Rules | Staircase Rules | Termination Rules |
| Acuity test | 2.3 | 4 | 6 | 0.3 | 1/1 Single staircase | 4 Reversals | 2/1 Double staircase | 10 Reversals | 0.08 logMAR |
| Contrast sensitivity test | 3.5 | 8 | 10 | 2.5 | 1/1 Single staircase | 4 Reversals | 2/1 Double staircase | 8 Reversals | 3.5 dB |
| Alignment sensitivity test | 9.5 | 15 | 19.5 | 1 | 1/1 Double staircase | 4 Reversals | 1/1 Double staircase | 8 Reversals | 1.5 arcmin |
Contrast sensitivity was measured with a modified temporal two-alternative, forced-choice method with a Gabor stimulus (vertical, at 6 cyc/deg). The circular Gabor patch subtended 3.5°, with the SD of the Gaussian envelope 0.65°. Alignment sensitivity was measured using three Gabor patches at 65% Michelson contrast (vertical, at 6 cyc/deg). The upper and lower Gabor patches were in vertical alignment. The central Gabor patch was displaced either to the left or right relative to this alignment on each trial. The square partial occlusion, the fusion lock, and the suppression markers were applied only in the partial-occlusion condition for the acuity, contrast, and alignment sensitivity tests (see
Table 2 for parameters in each visual function test).
Figure 1 shows target and partial-occlusion stimuli for each visual function test.
The central target and the partial occlusion used in this study are a form of dichoptic stimulation, since different stimuli are presented to each eye. In the partial-occlusion condition, the square patch occluded the central visual field, but was not a conventional occluder. It is referred to as partial occlusion because only the central field is occluded and to distinguish from full occlusion with an opaque eye patch.
Since all subjects were naïve, a training session was conducted on each subject before the main experiments. The stimulus duration was 400 ms during training and was reduced to 140 ms for the main experiments to minimize the effects of eye movements.
22 However, the subjects were allowed to take their time to give responses after each trial. The experimental tasks were identical in the training and the main experiments. These tasks were (1) acuity: indicate by pointing the orientation of the E target; (2) contrast sensitivity: verbally report whether the Gabor patch was presented in interval one or two; and (3) alignment sensitivity: point to the offset direction (left or right) of the central Gabor patch. All responses were input to the program by the examiner (XJL). No feedback was given.
The viewing distance was 4, 2, and 1 m for acuity, contrast sensitivity, and alignment sensitivity tests, respectively. To maintain viewing distance, subjects were carefully monitored by observation, and a backrest was used. Subjects were instructed to hold their viewing distance during experiments and were reminded every time their position was observed to have changed.
Acuity was found to be significantly poorer in the partial- than in the full-occlusion condition in all three subject groups, although this difference was slight in the children with normal vision and in the anisometropic children without amblyopia (
Fig. 2; children with normal vision:
F 1,9 = 8.159,
P = 0.019; anisometropic children without amblyopia:
F 1,16 = 6.052,
P = 0.026; amblyopic children:
F 1,16 = 8.373,
P = 0.011). Contrast sensitivity was significantly lower (worse) in the partial- than the full-occlusion condition in the children with normal vision (
F 1,9 = 13.203,
P = 0.005) and the amblyopic children (
z = −2.817,
P = 0.005), but this difference was not found in the anisometropic children without amblyopia (
F 1,16 = 0.700,
P > 0.05). Alignment sensitivity was not significantly different between the two viewing conditions in any subject group (children with normal vision:
z = −1.478; anisometropic children without amblyopia:
F 1,16 = 0.392; amblyopic children:
F 1,16 = 0.179;
P > 0.05).
Contrast sensitivity was significantly higher (better) in the children with normal vision (full: χ2 2 = 19.946, U = 11.0, z = −3.716; partial: χ2 2 = 22.073, U = 14.0, z = −3.565; P < 0.001) and the anisometropic children without amblyopia (full: U = 33.0, z = −3.840; partial: U = 21.0, z = −4.254; P < 0.001) than in the amblyopic children, but it was not significantly different between the children with normal vision and anisometropic children without amblyopia (full: U = 77.5, z = −0.377; partial: U = 77.0, z = −0.402; P > 0.017; note that P < 0.017 was the significance level corrected for multiple comparison). Surprisingly, neither acuity (full: χ2 2 = 7.241; partial: χ2 2 = 7.942; P > 0.017) nor alignment sensitivity (full: χ2 2 = 2.468; partial: χ2 2 = 0.387; P > 0.017) was significantly different between the three subject groups.
Acuity and contrast sensitivity were both found to be reduced in the partial- compared with the full-occlusion condition in children with normal vision and in those with anisometropic amblyopia, indicating greater inhibitory interactions (perhaps also weaker excitatory interactions) in the partial- compared to the full-occlusion condition. Although acuity was poorer in the partial- than the full-occlusion condition in anisometropic children without amblyopia, the difference was small (< 0.03 logMAR), and no such difference was found for contrast sensitivity in this group.
According to previous work,
18 acuity was improved when allowing light input to the nontested eye. This situation occurred in the partial-occlusion condition of the present study, but visual function was worse in this condition compared with full occlusion, indicating that in the partial-occlusion condition, signals generated by the nontested eye had a negative influence on those of the tested eye. This negative influence may be due to one or a combination of at least four factors:
First, perhaps the bright periphery produces an interocular inhibitory effect. Denny et al.
33 suggested that binocular summation (an excitatory effect) could reflect in part a suppressive influence from the occluded eye in monocular viewing. Optimal visual function could be achieved when tonic interocular suppression is removed from a dark-adapted eye by light adaptation. In our partial-occlusion condition, the bright periphery may have allowed light adaptation of the nontested eye, but visual function was reduced in this condition, indicating that any interocular interactions occurring here differ from those demonstrated by Denny et al.
Second, the bright periphery may have enhanced an inhibitory effect of the central (partial) occlusion, thus raising the threshold for the tested eye. The size of the partial occlusion was determined based on a pilot experiment, in which stronger inhibition was achieved using 2° versus 4° occlusion. However, in that pilot experiment, the size of the light surround co-varied with the size of the occlusion, and the relative effects of the central dark and surrounding light regions were unclear.
Third, contour interaction
34 may have arisen from the presence of the partial-occlusion edge presented to the nontested eye around the target of the tested eye.
10 However, foveal contour interaction is limited in its spatial extent and in this study exceeded that limitation.
35,36 Moreover, Simmers et al.
37 found contour interaction with optotypes of high contrast, but not with those of low contrast, and thus it seems unlikely to have been a significant influence on acuity in our study. However, contour interaction or a related form of suppression due to interocular stimulus difference may have played a role in the reduction in contrast sensitivity in the partial-occlusion condition.
Finally, a higher level function such as attention may be involved. The partial occlusion may raise visual attention in the nontested eye and increase the threshold of the tested eye, whereas this would not be expected to occur when the nontested eye is fully occluded.
The interaction index indicated the level of interocular interactions in terms of the relative impact of full and partial occlusion of the dominant eye on function of the nondominant eye. Different values of the index indicated different levels of dichoptic masking effect engendered using the setup of the present study, which may suggest a difference in inhibitory (e.g., rivalry) or excitatory (e.g., summation) interaction. However, our experiment cannot determine whether a difference in interaction index between groups reflects enhanced inhibition or reduced excitation.
The results presented in
Figure 3 indicated that the inhibition of the nondominant eye acuity due to partial occlusion of the dominant eye was stronger (or excitation was weaker) in anisometropic children with amblyopia than in those without. These differences between the subject groups in the full- and partial-occlusion conditions may reflect the differences in organization and function of the visual systems in these groups. For example, previous research has indicated that in the amblyopic visual system, the percept of the amblyopic eye is inhibited along with a reduction in binocular neurons and a shift of representation toward the dominant (nonamblyopic) eye.
20,38 Thus, signals from the nonamblyopic eye would dominate the resulting perception. If this were the case for the amblyopic visual system in the present study, the partial occluder presented to the nonamblyopic eye would dominate perception, at the expense of the target presented to the amblyopic eye. In the full-occlusion condition, input to the nonamblyopic eye was blocked with the opaque cover, and thus the target would be more easily detectable by the amblyopic eye, resulting in the difference in acuity between the two viewing conditions and the high acuity interaction index in this group of subjects. In the anisometropic children without amblyopia, however, instead of inhibition of the more ametropic eye, perhaps input from this eye was not inhibited or was weakly inhibited. This possibility is consistent with our finding of a low acuity interaction index in this group of subjects.
As shown in
Figure 4, the acuity interaction index correlated positively with the degree of anisometropia. It is possible that subjects with lower degrees of anisometropia did not have a significant interocular difference in image quality, and hence binocular fusion was likely to be achieved. In contrast, subjects with higher degrees of anisometropia had greater interocular image differences which were less easily fused to yield a unified percept. Another consideration is that the image size difference in uncorrected anisometropia depends on whether the anisometropia is due to refractive or axial length differences between the two eyes.
39 Some subjects with similar degrees of anisometropia had widely varying acuity interaction indices (
Fig. 4a). Perhaps some of those subjects had refractive and some axial anisometropia. When spherical myopes were analyzed separately, the correlation between the degree of anisometropia and the acuity interaction index was stronger (
Fig. 4b), indicating that refractive error type also influenced interocular interaction.
Although we found a relationship between interactions and anisometropic amblyopia, the nature of this relationship is unknown, since amblyopia may result from or cause anisometropia.
40,41 In addition, the refractive history such as the period since age of onset was unclear in the subjects of the present study. This may be an important factor in the development of amblyopia and may affect interocular interactions. Further experiments on a larger sample with clear refractive and amblyopic history may improve understanding of the relationship between interocular interactions and anisometropic amblyopia.
The interaction indices of contrast sensitivity and alignment sensitivity did not differ between any of the subject groups. There are at least three possible explanations for this lack of difference. First, these tests were not familiar to the subjects, and the experimental tasks were relatively difficult to comprehend, contributing to the high interindividual variation of these functions (
Fig. 2) and their interaction indices (
Fig. 3). Second, these visual functions may not have reached maturity in the subjects recruited (
Table 1), and the thresholds measured in both viewing conditions may be too low for any difference in the interaction indices between subject groups to be significant.
42 –44 Moreover, the interactions in these visual functions may be immature in these children, and the interindividual variation in the maturation process may add to the high interindividual variation in the interaction index.
8 Third, the partial-occlusion configuration may have different impacts between acuity and these two aspects of visual functions.
The amblyopic children with a higher acuity interaction index before therapy tended to show a better treatment outcome (
Fig. 5). Treatment outcome could be related to age, since the visual system has more plasticity during early than later childhood.
45 However, the four children who showed improvement in the present study were aged from 6 to 10 years, suggesting that age was not a factor. During occlusion therapy, vision of the nonamblyopic eye was blocked for 6 hours per day; thus, any inhibitory impact on the amblyopic eye due to dissimilar (sharp) images presented to the fellow eye may be reduced. For the amblyopic children with stronger inhibitory interaction before therapy, a reduction in this interaction may play an important role in the recovery from amblyopia. However, in some cases, the amblyopic children with moderate acuity interaction indices before therapy showed great improvement, and some of the subjects with high acuity interaction indices before therapy showed little or no improvement, suggesting that factors not controlled for, such as the age of onset and the severity of amblyopia, also play a role in the response to therapy. It is important to note that interaction may develop differently depending on intervention (refractive correction with and without occlusion, or no treatment). The relationship between interaction and types of intervention cannot be addressed here, because identical treatments were prescribed to all of the amblyopic children.
High-contrast acuity measures before and after treatments were based on a chart with unequal steps between lines and an unequal number of letters per line (see the Methods section). This discrepancy may be a confounding factor in our assessment of acuity improvement in the amblyopic children. In addition, only 4 of the 15 amblyopic children showed improvement in terms of a reduction in interocular difference in high-contrast acuity after therapy. This may be due to the age range of these subjects (mean age, 8.7 ± 1.2 years; eight subjects were older than 9 years), and perhaps reflects poorer compliance than that indicated on the checklist and questionnaire.
Inhibition of nondominant eye acuity due to partial occlusion of the dominant eye was significantly stronger (or excitation was weaker) in anisometropic children with amblyopia than in those without. This effect correlated positively with the degree of anisometropia in anisometropic children (with and without amblyopia). Amblyopic subjects with stronger inhibition before therapy showed a better amblyopic treatment outcome. However, these differences and associations were found in acuity only, but not in contrast sensitivity or alignment sensitivity. A better understanding of the relationship between interocular interaction and anisometropic amblyopia may be achieved with a larger sample, as well as a comparison in this interaction before and after amblyopia therapy.
Disclosure:
X.J. Lai, None;
J. Alexander, None;
M. He, None;
Z. Yang, None;
C. Suttle, None
The authors thank Jin Ping Zheng for assisting with the data collection.