Methods: Binocular decorrelation was imposed using prism-goggles for durations of 3 to 24 weeks (in 6 experimental, 2 control monkeys). Behavioral recordings were obtained, followed by neuroanatomic analysis of ocular dominance columns and binocular, horizontal connections in the striate visual cortex (area V1).

Results: Concomitant, constant esotropia developed in each monkey exposed to decorrelation for a duration of 12 to 24 weeks. The severity of ocular motor signs (esotropia-angle; dissociated vertical deviation; latent nystagmus; pursuit/optokinetic tracking asymmetry; fusional vergence deficits), and the loss of V1 binocular connections, increased as a function of decorrelation duration. Stereopsis was deficient and motion visual evoked potentials were asymmetric. Monkeys exposed to decorrelation for 3 weeks showed transient esotropia but regained normal visuomotor behaviors and binocular V1 connections.

Conclusions: Binocular decorrelation is a sufficient cause of infantile esotropia when imposed during a critical period of visuomotor development. The systematic relationship between severity of visuomotor sign, and severity of V1 connectivity deficit, provides a neuroanatomic mechanism for several of these signs. Restoration of binocular fusion and V1 connections, after short durations of decorrelation, helps explain the benefits of early repair in human strabismus.

Strabismus (pathologic misalignment of the visual axes) is an important public health problem, depriving 4% to 5% of children of the lifelong advantages bestowed by normal binocular vision.^1–3 The leading form of developmental strabismus is concomitant (nonparalytic) esotropia, which has a bimodal, age-of-onset distribution.^3,4 The largest peak—comprising about 40% of all strabismus—occurs before age 12 months in children who are predominantly emmetropic (the 3- to 4-year-old, “later-onset” age group is predominantly hyperopic). Nonaccommodative, early onset esotropia may be considered the paradigmatic form of strabismus in all primates, as it is also the most frequent type of natural strabismus observed in monkeys.⁵

The cause of infantile-onset esotropia is unknown. Detailed functional and structural studies of the extraocular muscles, orbital connective tissues, and brainstem circuitry in primates with concomitant esotropia have not revealed causal anomalies.^6,7 The striate cortex (area V1) is the first locus in the central nervous system for binocularity,^8,9 and binocular connections in V1 are important for generating the error signals that guide eye alignment.^10–12 Neonates who suffer insults that, directly or indirectly, perturb binocular inputs into V1 have a risk of developing strabismus that is 20 to 100 times higher than that in normal infants.^13–16 These observations led the author to propose a neural mechanism: infantile esotropia could be caused by simply perturbing the development of binocularity in V1, leading downstream, immature motor pathways to decompensate in the direction of innate (nasalward or convergent) biases.

The first goal was to test this hypothesis by rearing neonatal monkeys under conditions that exposed them to periods of sensorial, binocular decorrelation, but left their motor pathways and eye muscles completely intact. The second goal was to determine whether the duration of binocular decorrelation predicted the severity of any motoric maldevelopment (or the probability of recovery to normal function). These intervals of decorrelation were designed to emulate “early” vs “delayed” strabismus repair in human infants, with the aim of contributing behavioral and neuroanatomic information that could help clarify an issue of clinical controversy.^17–22

Infantile esotropia in humans is accompanied by a constellation of sensorimotor abnormalities, including deficits of stereopsis and fusional vergence, latent nystagmus, dissociated vertical deviation (DVD), as well as nasotemporal asymmetries of optokinetic tracking, smooth pursuit, and motion visual evoked potentials.^23–26 A good animal model should possess many of the disease features observed in human patients. For this reason, comprehensive behavioral testing was performed to detect (and quantify) these signs. After behavioral testing, binocular connections within area V1 were analyzed using neuroanatomic labeling methods. The goal of the anatomic experiments was to determine if abnormal binocular behaviors were related systematically to abnormal V1 connections.

FIGURE 1 Infant monkey wearing prism-goggles to induce binocular image decorrelation.

A total of 8 monkeys were studied: 6 experimental and 2 controls. The 6 experimental monkeys (Table 1) were divided into 3 prism-rearing groups: 3-week (2 animals), 12-week (1 animal), and 24-week (3 animals) groups. In each group, experimental animals wore prism goggles (for periods of 3 to 24 weeks) to impose horizontal and vertical binocular noncorrespondence (decorrelation) of 11.4º (20 prism diopters) in each eye; 11.4º base-in in one eye, and 11.4º base-down in the other eye. The 2 controls wore goggles with plano lenses. At 4 to 6 months of age, the monkeys were shipped to Washington University in St Louis, Missouri, where they were trained to perform visual fixation and tracking tasks without goggles, using a positive-feedback reward (a small bolus of fruit juice).²⁹ Cycloplegic refractions revealed a refractive error ≤ +3.00 spherical equivalent in each of the experimental and control animals.

TABLE 1 VISUAL AND OCULAR MOTOR CHARACTERISTICS OF THE 8 MACAQUE MONKEYS IN THE STUDY

At age 1 year, eye coils were implanted29 and ocular motor as well as sensory testing initiated, which proceeded typically for 3 to 6 months. Monocular visual acuity was measured using spatial sweep visual evoked potentials (SSVEP)30,31 (without correction for refractive error).

Eye movements were recorded using standard magnetic search coil techniques.^33,34 The monkey sat in a primate chair in the middle of field coils. The head restraint was locked to preclude head movement and the room was lit with dim background illumination. Eye position was calibrated at the start of each recording session by using a calibration coil and by having the animal maintain eye position within a 2º window of target position. The target was a laser spot subtending ~ 0.05º projected onto the back of a translucent screen located 50 cm in front of the animal. The calibration sequence was repeated separately for each eye.

Recordings were performed under conditions of binocular and monocular viewing. Monocular viewing was achieved by use of liquid-crystal shutter goggles which cycled from transparent to opaque (or the reverse) in 80 microseconds (0.08 msec).35 Voltages proportional to horizontal and vertical eye position were digitized at 500 Hz. Eye velocity signals were obtained by passing the eye position signals through a Finite Impulse Response filter (DC to 90 Hz) and differentiated. Angular resolution of the system was about 0.05º. Experiments were controlled and the data were acquired and analyzed with the aid of a computer and interactive signal processing software (Spike2 for Macintosh, Cambridge Electronic Design, United Kingdom, and Igor Graphics, Wave Metrics, Lake Oswego, Oregon).

FIGURE 2 Visual stimuli used to test eye alignment and stable fixation, horizontal smooth pursuit, and horizontal optokinetic nystagmus. The moveable stimuli (spot or stripes) were back-projected on a translucent tangent screen.

FIGURE 3 Visual stimuli used to test short-latency disparity-vergence, accommodative-disparity vergence, and stereopsis. Monkeys wore liquid-crystal (LC) shutter goggles and viewed correlated-random-dot patterns, displayed on a video monitor, to provide dichoptic (more ...)

FIGURE 4 Labeling of ocular dominance columns (ODCs) and horizontal axonal connections in striate cortex (area V1) of a normal primate. [3H] proline was injected into the vitreous cavity of one eye and transported anterogradely from eye-to-V1 layer 4C, labeling (more ...)

The counting method is laborious (~ 100 microscope man-hours/injection site). It was necessary for establishing a strict relationship between labeling patterns viewed at low power, and soma counts at higher power, in representative injections. The volume of injections that needed to be assessed in both hemispheres in each monkey precluded analysis using this technique in all of the monkeys. For this reason, an alternative method was employed. Soma counting revealed a systematic relationship between the overall pattern of labeling, viewed at low power in individual sections, and the soma counts. When the pattern appeared as a “sunburst” distribution of label across ODCs (ie, an oval, densest at the injection center and diminishing uniformly in intensity of labeling with increasing distance from the center), soma counts showed comparable numbers of labeled somata in right vs left eye ODCs. The diminished intensity of labeling, with increasing distance from the injection center, represents diminishing numbers of connections. Because the “sunburst” pattern showed uniform diminishment across both right and left eye ODC territory, the “sunburst” pattern of labeling was designated the “binocular connection” pattern. Alternatively, when the pattern appeared as a “skipping” distribution of label across ODCs (also densest at the center but fluctuating in intensity of labeling across every other row of ODCs), soma counts showed significantly higher numbers of labeled somata in ODCs, which had the same ocularity as the injected ODC, and significantly lower numbers of labeled somata in the ODCs of opposite ocularity. The “skipping” pattern of labeling was therefore designated the “monocular connection” pattern.

The BDA injections in each V1 hemisphere of each monkey were scored as “sunburst/ binocular” or “skipping/monocular” by 2 observers masked to the animal’s behavioral status. Approximately 15% of the injections had ambiguous patterns that could not be scored as either “sunburst” or “skipping” and were excluded. The results of the labeling-pattern scoring are tabulated as bar graphs for each monkey group.

FIGURE 27 Severity of behavioural and neuroanatomic deficit with increasing duration of binocular decorrelation. For the behavioural measures, the regression lines plot the magnitude of the deficit in multiples of standard deviation, normalized to the average value (more ...)

The results of behavioral testing are described first, and the results of neuroanatomic analysis second. The behavioral results proceed from eye alignment to testing for the constellation of signs that typify the infantile esotropia syndrome in humans. The results in each section compare behaviors by the duration of binocular-decorrelation imposed: control, 3 weeks duration, 12 weeks duration, and 24 weeks duration (hereinafter designated 3-wk, 12-wk, and 24-wk).

FIGURE 5 Eye alignment at age 1 year measured under conditions of binocular (neither eye covered) viewing. Group means ± SD for straight-ahead (primary position) gaze.

FIGURE 6 Eye alignment at age 1 year measured during automated, single-cover testing at 5 cardinal gaze positions. Control monkey AY and 3-wk monkey SY had small heterophorias. 12-wk monkey GO was esotropic, shown here when fixating with the right eye. This animal (more ...)

The 12-wk and each of the 24-wk monkeys also showed DVDs,^65,66 ranging in magnitude from 2.0 to 7.1º. The 12-wk monkey GO manifested, in addition to the DVD, a dissociated horizontal deviation when fixating with the left eye (right-eye dissociated horizontal deviation (DHD), Figure 6).^67,68 As in human esotropes, the DVDs differed in magnitude in the 2 eyes. Figure 6 shows a larger DVD in monkey GO (12-wk) when fixating with the right eye, and a larger DVD in monkey HA (24-wk) when fixating with the left eye. Only the DHD of the right eye in monkey GO conformed to a V pattern; the esotropia measured with both eyes uncovered showed no noteworthy pattern.

Concomitancy of the horizontal and vertical deviations across 5 cardinal gaze positions is shown for each of the duration groups in the bar graphs of Figure 7. The magnitude of the heterophorias is plotted in the control and 3-wk monkeys, and the magnitude of the heterotropias in the 12-wk and in the 24-wk monkeys. Magnitudes were substantially greater in the 12- and 24-wk animals (ANOVA, P = .008).

FIGURE 7 Horizontal and vertical eye misalignment measured during single-cover testing at 5 gaze positions in the 4 groups of monkeys. Group means ± SD.

To appreciate the significance of these findings, it is important to emphasize a methodological point: the direction of the horizontal strabismus cannot be explained as adaptation of the fusional vergence system to optical prisms (prism adaptation).^69,70 As illustrated in Figure 8, the horizontal prism was positioned base-in, displacing the image off the fovea and onto the eccentric nasal hemiretina in each experimental animal. If the animals had attempted to overcome the horizontal component of the decorrelation by prism adaptation, they would have diverged their eyes into an exotropic position. But each of the monkeys (Table 1 and Figure 5) developed convergent strabismus. The finding of DVDs in the strabismic animals can, likewise, not be explained as vertical prism adaptation. Cover-testing did not show a hypertropia in one eye and a hypotropia in the other; both eyes had hyperdeviations (Figure 6), which is the hallmark of DVD.

FIGURE 8 Absence of prism-adaptation, and deterioration to esotropia caused by noncorresponding binocular images.

As noted in the “Methods” section, the infant monkeys manifested esotropic deviations after wearing the prisms for a period of ~ 1 to 2 weeks (observed by infrared viewing during daily goggle cleaning and estimated by Hirschberg measurement to be, on average ~ 8 º to 10º). A physical esotropia of ~ 10º added to a base-in prism displacement of 11º sums to a total magnitude of horizontal-image noncorrespondence of ~ 21º. Removing the goggles after 3 to 24 weeks of prism-rearing achieved, therefore, a reduction in binocular noncorrespondence of ~ 50%. Extrapolating to an esotropic human infant, this was the sensorial equivalent of correcting the esotropia by 50% (ie, surgical undercorrection). The results of Figures 5 through 7 show that a 50% correction was effective for recovery of normal alignment if carried out by age 3 weeks, but was ineffective if delayed to age 12 to 24 weeks. Unfortunately, the time-course for the recovery of eye alignment after goggle removal in the 3-wk monkeys (hours, days, or weeks) is not known. By age ~ 4 months, when they were shipped from the nursery in Atlanta to the author’s lab in St Louis, eye alignment by Hirschberg measurement was orthotropic.

FIGURE 9 Stable fixation vs latent/manifest latent nystagmus in horizontal eye position tracings from 4 monkeys; fixating monocularly with the right eye (RE) or left eye (LE) covered, and viewing binocularly. A and B, Stable fixation in control monkey WE and 3-wk (more ...)

Average slow-phase velocities of the latent and manifest-latent nystagmus are shown in Figure 10. The largest mean velocity, 2.0º/sec was recorded during monocular viewing in the 24-wk monkeys (ANOVA, P = .02). Nystagmus velocity during binocular viewing was 26% to 51% slower. Nystagmus intensity (the product of frequency × amplitude) increased systematically with increasing duration of decorrelation (Figure 11; ANOVA, P = .001).

FIGURE 10 Velocity of latent and manifest-latent nystagmus in the 4 groups of monkeys. Group means ± SD; pooled responses of right and left eye.

FIGURE 11 Intensity of latent and manifest-latent nystagmus in the 4 groups of monkeys. Group means ± SD; pooled responses of right and left eye.

FIGURE 12 Horizontal smooth pursuit for nasalward vs temporalward target motion, viewing with the right eye (left eye occluded). In both the control (A) and 3-wk monkey (B), horizontal pursuit was equally robust for nasalward vs temporalward motion. In contrast, (more ...)

The 3-wk monkey shown in Figure 12B also demonstrated symmetrical pursuit for both temporalward and nasalward stimulus motion. In contrast, in the 12-wk (Figure 12C) and 24-wk (Figure 12D) animals, a nasotemporal asymmetry of pursuit was evident. Their responses also tended to be more variable and less machinelike. Pursuit for temporalward stimulus motion was less robust, evident as eye position lagging target position (upper panels) and eye velocity (lower panels) that was, on average, 16% to 49% lower than that for nasalward motion. Pursuit latencies did not differ between monkey groups (ANOVA, P = .24)

Nasotemporal pursuit performance was compared for target velocities of 15º/sec and 30º/sec, across monkey groups, by calculating a nasal bias index (NBI, defined as the difference, divided by the sum, of nasalward and temporalward mean smooth eye velocity; see “Methods” section). An NBI of zero would indicate symmetrical pursuit, a positive NBI stronger nasalward, and a negative NBI stronger temporalward pursuit. The NBIs for the 4 groups of monkeys are shown in the bar graphs of Figure 13. Pursuit was symmetrical in both the control and 3-wk groups, who had NBI ≤ 1. Pursuit was nasally biased in the 12-wk and 24-wk monkeys, with the largest NBI in the 24-wk animals (ANOVA for mean temporalward velocities, P = .04).

FIGURE 13 Nasalward biases of smooth pursuit as a function of decorrelation duration. Group means ± SD; pooled responses for right and left eye viewing.

FIGURE 14 Horizontal optokinetic nystagmus for nasalward vs temporalward target motion, viewing with the left eye (right eye occluded). In both the control (A) and 3-wk monkey (B), responses were equally robust for nasalward vs temporalward stripe motion. In contrast, (more ...)

Figure 15 plots a NBI for OKN in each of the 4 monkey groups. Symmetrical nasotemporal OKN responses in the control and 3-wk monkeys produced NBI ≤ 2. The NBI in the 12-wk monkey was 10 times higher (= 20.2) and in the 24-wk 18.5 times higher (= 37.1; ANOVA for mean temporalward SPV, P = .000). It is important to note that the directional biases of pursuit and OKN cannot be explained, as latent nystagmus drifts, merely adding to a nasalward response and subtracting from a temporalward response. Latent nystagmus slow-phases, averaging ~ 0.75º/sec in the 12-wk monkey, and ~ 2.0º/sec in the 24-wk monkeys, would produce NBI of 2.5 and 6.7, respectively, when factored into a perfectly symmetric response to 30º/sec target motion. The actual NBI for 30º/sec pursuit in these animals was 5.1 (12-wk) and 22.5 (24-wk). For OKN, the actual NBI was 20.2 (12-wk) and 37.1 (24-wk). Thus, the actual NBIs were 104% to 554% higher than could be accounted for by any influence of nasalward-drift nystagmus.

FIGURE 15 Nasalward biases of optokinetic nystagmus (OKN) as a function of decorrelation duration. Group means ± SD; pooled responses for right and left eye viewing.

The stimulus used to evoke short-latency disparity vergence was a large, correlated-dot pattern displayed on a video monitor, viewed through liquid-crystal shutter goggles (Figure 3). By varying the horizontal offset of the dot patterns viewed by the right vs left eye, pure binocular disparity could be produced, devoid of any accommodative, monocular cues. Figure 16 shows short-latency convergence and divergence eye velocity tracings in response to disparities of − 0.5 to + 1.0º, presented at time-0 on the y-axis. Each tracing represents an average of 10 trials at that disparity. The control (WE) and 3-wk monkey (TE) had appropriate convergent (+) and divergent (−) velocity deflections at ~ 60 msec, with the 3-wk animal generating more robust velocities than the control. In contrast, the 12- (GO) and 24-wk (QN) animals displayed erratic vergence-velocity baselines and no convincing disparity responses (ANOVA, velocities at +0.5º, P = .01). The 12- and 24-wk monkeys were also tested in sessions using base-out Fresnel prisms to offset the esotropia and optically re-align the eyes. Performance did not improve.

FIGURE 16 Velocity of short-latency fusional vergence in response to binocular disparities of −0.5 to +1.0º. The control and 3-wk monkey had comparable, robust responses. The 12- and 24-wk monkeys had negligible or inverted responses, with noisier (more ...)

Figure 17 shows change in short-latency vergence, for monkeys in each group, over a wider range of disparities. The control (WE) and 3-wk monkey (TE) had the most systematic responses, with the largest amplitude divergence and convergence responses in the 3-wk monkey. The responses of the 12- and 24-wk strabismic animals were unsystematic and at some disparities inverted (“wrong-way”), not coupled to either the magnitude or sign of the stimulus (ANOVA, amplitudes at +1º and +2º, P = .04).

FIGURE 17 Amplitude of short-latency fusional vergence in response to binocular disparities of −1 to +2º. The control and 3-wk monkey had comparable response profiles for disparities of −0.5 to +2º. The 12- and the 24-wk monkeys (more ...)

As shown in Figure 18, control monkey AY executed accurate, stereotyped convergence eye movements when the target stepped from far to near. The animal converged an average of 3.35 ± 0.5º, or 96% of the near-target vergence angle. The right eye adducted an average of 1.72º and the left eye an average 1.63º. The average latency of the initial response was 136 ± 11 msec. The 3-wk monkey SY (Figure 18) also executed robust convergence, albeit not as systematically as normal monkey AY. In the 10 trials shown, the near-target was situated minimally to the right of SY’s midsagittal plane, which required the left eye to converge (adduct) 0.6º more than the right. (The near-target in these sessions was aligned as accurately as possible on the midsagittal plane [“cyclopean center”] of the head with the goal of requiring perfectly symmetric vergence in the 2 eyes. However, the head restraint device in each animal held the skull firmly but not rigidly, and owing to the powerful neck musculature of macaques, head translational displacements of ~ 1 mm to the right or left of the midsagittal plane were unavoidable.) The convergence response in SY was 3.01 ± 1.1º, or 86% of target vergence, at an average latency of 165 ± 17 msec.

FIGURE 18 Accommodative-disparity vergence in response to a step-change in position of target from far to near (vergence demand ~ 3.5º, or ~ 1.75º/eye). Target jump to near position occurred at “0 msec” on the x-axis. The responses (more ...)

The accommodative-disparity vergence responses of the esotropic monkeys GO (12-wk; esotropia ~ 7º) and HA (24-wk; esotropia ~ 13º) differed strikingly from that of the control and 3-wk monkey. Rather than smoothly converging, both of the strabismic monkeys (Figure 18) employed a variable, disjunctive-saccade strategy. The saccades in monkey GO adducted the fixating-right eye to the near target. The saccades in monkey HA adducted the fixating-left eye toward the near target, but consistently overshot target position. The saccades were disjunctive: in 12-wk monkey GO they achieved convergence of an average 1.1º, and in 24-wk monkey HA convergence of an average 0.52º. The mean latencies of the saccadic responses (168 ± 18 msec in GO; 175 ± 14 msec in QN) were comparable to the vergence responses of the control and 3-wk monkeys (ANOVA; P = .07). Convergence performance, measured as vergence amplitude as a percentage of near target position, is summarized for the 4 monkeys in the bar graph of Figure 19. The convergence of the strabismic 12- and 24-wk monkeys was deficient compared to the control and 3-wk monkey (n = 50 trials, ANOVA P = .01). The behavior of the 12- and 24-wk esotropic monkeys was similar to that reported in monkeys with natural infantile esotropia.⁴² Natural esotropes also execute disjunctive saccades to refixate near targets and have subnormal vergence.^42,76 Saccadic peak velocities were compared in control and experimental monkeys, for saccade amplitudes from 5º to 25º. The esotropic monkeys had peak velocities (“main sequence plots”⁴⁰) that fell within the range recorded in controls, confirming absence of lateral rectus paresis or abducens palsy.

FIGURE 19 Accommodative-disparity convergence amplitude as a percentage of target step to the near position. The control and 3-wk monkey had comparable responses. The 12-wk monkey showed a subnormal response; the 24-wk monkey showed an even weaker response. Mean (more ...)

To test accommodative vergence, binocular disparity cues were eliminated by recording vergence to the near target under conditions of monocular viewing (ie, the paradigm remained the same, but one eye was occluded). In normal adult monkeys, monocular viewing reduces the amplitude of the vergence response by ~ 20%.^42,77 In the control and 3-wk monkeys of the current study, monocular viewing reduced the amplitude of the convergence responses an average of 25% and 19%, respectively (data not shown). The disjunctive-saccade behavior of the 12- and 24-wk monkeys was unaltered, reinforcing the point that their saccades—even with both eyes open—were driven by monocular position and blur cues, rather than sensitivity to binocular disparity.

The synkinetic relationship between accommodative-convergence and accommodation is expressed clinically as a ratio (AC/A, in prism diopters/sphere diopters with 1 prism diopter = 0.57º).⁴⁰ A “stimulus AC/A ratio” is measured in the typical clinical setting with the assumption that lens accommodation matches the accommodative demand of the near target to eliminate all blur. Measurement of an actual “response AC/A” requires recording with an optometer.⁷⁸ A stimulus AC/A ratio was measured in each of the monkeys with the assumption that they carried out the 2.0 D change of accommodation demanded to focus on the near target (the distant target [1 meter] demanded 1.0 D of accommodation and the near target [0.33 meter] demanded 3.0 D, where D = 1/target distance in meters). The stimulus AC/A ratios in control monkeys averaged 2.31 and in the 3-wk monkeys 2.24. These values agree both with stimulus AC/A ratios and response AC/A ratios (2.0 to 3.0) reported for normal, adult macaques.^77,42 The very weak accommodative vergence response of the 12-wk esotropic monkey GO yielded a low stimulus AC/A ratio: 0.55. The stimulus AC/A in the two 24-wk monkeys was lower, an average of 0.27.

Figure 20A shows the responses of a control monkey (AY). He displayed stereo-sensitivity (ie, a minimum of 70% correct) over the entire range tested, with responses falling to chance level only for interleaved, zero-disparity, “catch” trials. The 3-wk monkey (SY) also showed normal stereo-sensitivity (Figure 20B). In contrast, 24-wk monkey HA proved to be stereo-blind (Figure 20C), as was 24-wk monkey QN (not shown). Optically realigning the eyes of the 24-wk monkeys with Fresnel prism did not improve their performance. (Broken eye movement coils near the end of behavioral testing prevented measurement of stereopsis in 12-wk monkey GO). Figure 21 plots the responses to larger (Figure 21A) and smaller (Figure 21B) disparities. Performance declined for smaller disparities, even in the control animal.

FIGURE 20 Random-dot stereopsis for disparities ± 5º. The control and 3-wk monkey had equivalent levels of stereoscopic sensitivity over the entire range. The 24-wk monkey was stereoblind; esotropic monkey tested with prisms to realign the visual (more ...)

FIGURE 21 Random-dot stereopsis for crossed (+) disparities 0.1º to 5º. The control (n = 2) and 3-wk (n = 2) monkeys showed comparable levels of stereopsis at disparities 0.4º or greater. The 24-wk (n = 3) monkeys were stereoblind. Esotropic (more ...)

Mean asymmetry indices for each of the 4 groups of monkeys are summarized in the bar graphs of Figure 22. When tested using 1-cpd gratings that oscillated at 6 Hz, mean asymmetry indices were < 0.25 (indicated by the horizontal dashed line) for the control and 3-wk monkeys, but 0.55 ± 0.32 in the 12-wk, and 0.51 ± 0.29 in the 24-wk esotropic animals (ANOVA, P = .01). Similar results were obtained when testing was performed using 3-cpd gratings that oscillated at 11 Hz. The mean asymmetry index was < 0.40 (dashed line) in controls (0.38 ± 0.05) and 3-wk (0.31 ± 0.03) monkeys, but measured 0.72 ± 0.18 in the 12-wk and 0.76 ± 0.23 in the 24-wk esotropic monkeys (ANOVA, P = .008).

FIGURE 22 Motion visual evoked potential asymmetry indices in the 4 groups of monkeys. For 1 cpd at 6 Hz stimulus, dashed line = 0.25 upper boundary of normal. For 3 cpd at 11 Hz stimulus, dashed line = 0.40 upper boundary of normal. Pooled data for viewing with (more ...)

Figure 23 shows a region of foveomacular area (ie, opercular cortex) V1 in a 3-wk monkey, viewed at low magnification, in which the neuronal tracer BDA was injected into an ODC. The section was cut tangential to the external surface of the cortex, in this case through layer 4B, providing a view equivalent to peeling away the most superficial layers of V1 to look down on the cut ends of ODCs (Figure 4). Viewed in this manner, V1 is arranged as alternating rows of right-eye and left-eye ODCs. The boundaries of these rows (identified in [³H] proline labeled sections, see “Methods”) are indicated by the lines. BDA produced dark, patchy labeling, densest at the injection center (marked by a “+”) and reducing in intensity with increasing distance from the injection site. The dark patches, viewed at higher power (Figure 23), were produced by clusters of labeled pyramidal-neuron bodies and their horizontal, axonal projections. Distal axonal projections, connecting to neurons within the injected ODC, had taken up the BDA label. These projections transported the BDA back (retrograde) to label their neuron bodies in neighboring ODCs. Neurons in immediately neighboring ODCs have the richest plexus of interconnections, accounting for the fall-off of labeled neurons with increasing distance from the injected ODC.

FIGURE 23 Biotinylated-dextran-amine injection site in V1 layer 4B of a 3-wk monkey, viewed at low power. Left, The center of the injection (+) stained darkly due to the high density of labeled neurons. Label intensity diminishes with increasing distance from the (more ...)

To quantify precisely binocular vs monocular connections, counts of labeled pyramidal neurons were obtained from representative BDA injections in V1 layer 4B of the 3-wk monkey (SY) and the 24-wk monkey (EY). Labeled neurons were counted within the injected ODC, and within neighboring ODCs, up to 5 to 7 ODC rows (2.25 to 3.15 mm) away from the injected ODC. The results of this analysis are tabulated in the bar graphs of Figure 24, with distance from the injected ODC (number 0) annotated as increasing ODC number (1 to 7) along the x axis. The 24-wk monkey had more labeled neurons within ODCs of the same ocularity as the injected ODC, and fewer labeled neurons within ODCs of opposite ocularity, ie, neurons making monocular connections substantially outnumbered neurons making binocular connections (t test for paired samples, n = 1576, comparing same to opposite ODCs, P = 0.004). The imbalance of monocular neurons over binocular neurons, so evident in the 24-wk monkey, was not evident in V1 of the 3-wk monkey (Figure 24). Counts of labeled neurons in the 3-wk revealed equivalent numbers within ODCs of the same vs opposite ocularity (t test for paired samples, n = 1365, P = 0.35).

FIGURE 24 Number of labeled pyramidal neuron bodies in same-eye and opposite-eye ocular dominance columns (ODCs) of V1 after injection into a single ODC, 3-wk vs 24-wk monkey. The injected ODC is at zero on the x-axis, and neighboring ODCs are numbered in successive (more ...)

Counting of labeled neurons revealed a relationship between the overall pattern of BDA-labeling, viewed at low power in individual sections, and the neuron counts (see “Methods”). When the pattern appeared as a “sunburst” distribution of label across ODCs (ie, an oval, densest at the injection center and diminishing uniformly in intensity of labeling with increasing distance from the center), counts tended to show comparable numbers of labeled neurons in right vs left eye ODCs. The “sunburst” pattern of labeling was therefore designated the “binocular connection” pattern. Alternatively, when the pattern appeared as a “skipping” distribution of label across ODCs (also densest at the center but fluctuating in intensity of labeling across every other row of ODCs), counts tended to show higher numbers of labeled neurons in ODCs which had the same ocularity as the injected ODC, and lower numbers of labeled neurons in the ODCs of opposite ocularity. The “skipping” pattern of labeling was therefore designated the “monocular connection” pattern.

Figure 25 shows these dichotomous labeling patterns for 8 representative V1 injections: 4 of the “binocular/sunburst” variety in the 3-wk monkeys (left column) and 4 of the “monocular/skipping” variety in the 12- and 24-wk animals (right column). Each injection in V1 of both cerebral hemispheres was scored as sunburst vs skipping for each of the 3-, 12-, and 24-wk monkeys (Figure 26). The most common pattern of neuronal labeling observed in the 3-wk monkeys was “sunburst/binocular,” accounting for 56% of the injections (n = 19). These results are comparable to those reported previously for BDA injections in V1 layers 2, 3, and 4B of normal, adult monkeys. The lowest proportion of “sunburst/binocular” (13%) was found in the monkeys with the most severe behavioral signs of infantile strabismus—the 24-wk animals (n = 22). The 12-wk animals, who had behavioral signs intermediate between the 3- and 24-wk groups, had a proportion of “sunburst/binocular” injections (n = 16) intermediate (38%) between the 3- and 24-wk groups. These proportions differed significantly between groups (proportions test, P = .03).

FIGURE 25 Biotinylated-dextran-amine (BDA) injection sites in V1 layer 4B viewed at lower power. Four “sunburst” (binocular) patterns of labeling, in control and 3-wk monkeys; and 4“skipping” (monocular) patterns in the 12- and 24-wk (more ...)

FIGURE 26 Proportion of V1 injections that corresponded to “sunburst” (binocular) vs “skipping” (monocular) patterns of transported label in the 3-wk, 12-wk, and 24-wk monkeys. Pooled data within groups for multiple injections into (more ...)

The purpose of this study was to answer 2 major questions. First, can esotropia be produced reliably in normal infant primates by imposing binocular decorrelation during an early critical period for visuomotor development? The answer to this question is yes. Each animal exposed to binocular decorrelation by prism-dissociation in the first weeks of life developed concomitant esotropia. In the animals exposed to longer durations of decorrelation, the esotropia was accompanied by a constellation of sensory and motor signs mimicking closely the features observed in strabismic children. Second, does the severity of the signs increase systematically as a function of duration of decorrelation? The answer to this question is also yes; the magnitude of each behavioral and anatomic abnormality tended to increase—often monotonically—with longer decorrelation duration. These data provide information that would be difficult or impossible to obtain from human infants in randomized clinical trials. Using a primate model allowed us to impose precise periods of deprivation, to then record—in hundreds of quantitative, machine-like trials—the resultant behavioral deficits, and to then analyze, at a level of detail unobtainable by any current neuroimaging technique in human, the underlying neural circuitry.

Fusional vergence eye movements in neonates are also immature, maturing during an equivalent period. In the first 2 months of life, alignment is unstable in human infants and responses to step or ramp changes in disparity are inaccurate, favoring overconvergence.^90–93 Transient convergence errors exceed divergence errors by a ratio of 4:1. The fusional vergence response to crossed (convergent) disparity is also intact earlier and substantially more robust than that to divergent disparity. The innate bias favoring convergence persists after full maturation of binocular disparity sensitivity. Fusional convergence ranges in normal, adult primates exceed divergence ranges by a mean ratio of 2:1,^94,95 and brainstem premotor convergence neurons outnumber divergence neurons by a ratio of 3:2.^96,97

Maturation of fusional vergence was impaired in the current study by imposing binocular decorrelation throughout the first weeks of life. Vergence decompensated to esotropia in each animal, without manipulation of the extraocular muscles or motor pathways. The magnitude of the esotropia increased as a function of decorrelation duration. These results imply that esotropia is an expression of an innate, convergence bias. The convergence bias is held in check by maturation of disparity vergence but may be expressed during an early period of vergence instability by impeding cortical connections that mediate disparity sensitivity. The longer these connections are impeded, the more the convergence bias is expressed.

The results reported here reinforce and extend these findings by demonstrating a relationship between the duration of binocular decorrelation and the severity of nasotemporal asymmetries. The animals that retained (control monkeys) or regained (3-wk monkeys) binocular fusion showed symmetric nasotemporal responses. The animals that lost binocular fusion (the 12-wk and the 24-wk monkeys) showed pronounced asymmetries. In these animals, NBIs for both smooth pursuit and OKN increased as the duration of decorrelation increased, as did the asymmetry indices for motion VEPs.

FIGURE 28 Horizontal axonal connections between same- and opposite-eye ocular dominance columns caused by correlated and decorrelated binocular input during an early critical period. Correlated activity (“cells that fire together, wire together”) (more ...)

These experiments in monkeys show that extrinsically imposed binocular decorrelation causes loss of V1 binocular connections, and esotropia. Could intrinsically generated decorrelation cause an analogous loss in some human infants? Several lines of evidence suggest so. The correlated signals that stabilize horizontal connections must flow through immature, monocular geniculostriate fibers. These fibers in infants are vulnerable to cytotoxic insults from a variety of causes,^117–119 and perinatal insults to the geniculostriate pathways are linked strongly to subsequent onset of strabismus.^13–16 Periventricular hemorrhage increases the prevalence of infantile strabismus 50- to 100-fold. Nonspecific cerebral insults, eg, from very low birth weight, increase the risk 20- to 30-fold.^14,120 Apart from cytotoxic injury, normal inequalities of geniculostriate development promote intrinsic binocular decorrelation. In V1 of each hemisphere, projections from the ipsilateral eye (temporal hemiretina) are slower to develop and smaller in number than projections from the contralateral eye (nasal hemiretina).¹²¹ Any perturbation that slowed geniculostriate development would prolong the duration of binocular decorrelation caused by this input inequality. The decorrelation mechanism is appealing because it is parsimonious: a simple, “final common pathway” origin for essential infantile esotropia. The “essential” factors that perturb oligodendrocytes or delay ipsilateral projections could be quite subtle, varying idiosyncratically with individual thresholds from one strabismic infant to another, eg, genetically expressed neurotrophins, gestational nicotine exposure, and mild toxemia.

The 12- and 24-wk esotropic monkeys of the current study had a paucity of V1 layer 4B horizontal connections joining ODCs of opposite ocularity. The output from layer 4B provides a major projection to extrastriate areas MT/MST, which mediate perception of stereopsis and help modulate vergence.^125–129 The missing V1 binocular connections in these animals provide a structural explanation for their deficient disparity-vergence and stereopsis behaviors. Although binocular functions were lost, the esotropic monkeys retained normal monocular (grating) visual acuity in each eye. The lack of horizontal connections between opposite-eye ODCs would not be expected to impair monocular spatial vision, since the monkeys would retain connections between orientation-tuned neurons ^130,131 belonging to the same eye.

Our finding that 56% of labeled injections were binocular in the 3-wk repair monkeys is within the lower range found in normal adult animals. Anatomic labeling of binocular connections corresponds roughly but not strictly to binocular responsiveness measured by electrophysiolgical methods. In normal adult primates, ~ 80% of V1 neurons show some binocular responsiveness, but only 55% to 75% of V1 neurons labeled anatomically show distinct binocular connectivity.⁵⁷ This functional-structural discrepancy suggests that some binocular connections, too faint to allow unambiguous anatomic detection using the methods we employed, are sufficient nonetheless to elicit a weak binocular response using extracellular recording.

Critics of early surgery argue that the esotropia may have resolved spontaneously in infancy if left untreated. The Congenital Esotropia Observational Study found that the strabismus in fact persisted in 98% of infants who had large-magnitude (≥ 40 prism diopters) constant esotropia.^143,144 A follow-up study of infants with smaller, variable-angle esotropia found that surprisingly few convert to normal fusion behaviors.¹⁴⁵ Even small, early-onset, constant esotropias tend predominantly to progress to larger-magnitude angles of misalignment.

Early surgery in humans is believed to enhance sensory outcomes by re-establishing correlated binocular activity during an early critical period for the development of stereopsis.^20,86,146 Although both age at alignment and duration of misalignment are correlated with long-term stereoacuity outcome, the duration of misalignment appears to be the more important factor.^20,146 Durations less than a total of 3 months are associated with good to excellent restoration of stereopsis and motor fusion. The results of the current study provide indirect support for early surgery by showing that a wide range of fusional visuomotor behaviors, and the underlying V1 binocular circuitry, can be repaired in a primate model. The repair was effective when the duration of binocular decorrelation was limited to the human-equivalent of ~3 months.

Funding/Support: Supported by the National Institutes of Health grant 5R01 EY010214 08-11 from the National Eye Institute and a Research to Prevent Blindness Walt and Lilly Disney Scholars Award for Amblyopia Research. Financial Disclosure: None. Other Acknowledgments: We are indebted to Mae Gordon, PhD, Director of the Biostatistics Core, Department of Ophthalmology and Visual Sciences, Washington University School of Medicine, for assistance with statistical analysis. Paul Foeller, MS (Washington University), provided invaluable help with all aspects of animal training, eye movement recording, and neuroanatomic preparation. Dolores Bradley, PhD, (Yerkes Primate Center, Atlanta, Georgia) directed the initial animal rearing. Andreas Burkhalter, PhD, assisted with all phases of the neuroanatomic work. Agnes Wong, MD, PhD, (Washington University and University of Toronto) assisted with eye movement analysis, as did Michael Richards, BS, Leo Sin, BS, and Aasim Hasany, BS (all of the University of Toronto).