conclusion drawn from these observations was that MST neurons promote esotropia (i.e., a bias for nasalward vergence) when binocularity fails to develop in V1. The mechanism is attractive, because it ties together the

5nasalward biases of vergence, pursuit/OKN and gaze holding (latent nystagmus) in cortical regions vulnerable to perinatal damage.

Outputs from the cortical areas noted earlier (V1, MT/ MST) and related cortical areas descend to brainstem visual relay and premotor neuron pools immediately adjacent to the motor nuclei (Fig. 5.5) [106]. Even in the absence of cortical maldevelopments, the vergence system is unbalanced, favoring convergence. Midbrain premotor neurons driving convergence outnumber those driving divergence, by a ratio of 3:2.

5.1.16Repair of Strabismic Human Infants: The Historical Controversy

Is repair of binocular V1 connections possible, restoring normal fusion and stereopsis, while preventing or reversing the constellation of ocular motor maldevelopments? The answer to this question is rooted in a debate between two competing twentieth century schools of treatment philosophy, derived from the eminent British strabismologists, Claude Worth and Bernard Chavasse. Worth postulated in 1903 that esotropic infants su ered “an irreparable defect of the fusion faculty” [107]. Their brain was congenitally incapable of achieving substantial binocular vision. Early surgical treatment was therefore unfounded because it was futile. Chavasse on the other hand – attracted by the Pavlovian physiology of the 1920 and 1930s – believed that the brain machinery for fusion was present in esotropic infants, but the development of “conditioned reflexes” for binocular fusion were impeded by factors such as weakness of the motor limb [108]. He postulated (in his text published in 1939) that if the eyes could be realigned during what he believed to be a period of reflex learning, binocular fusion could be restored.

5.1.17Repair of High-grade Fusion is Possible

New knowledge of stereopsis development in the 1980s bolstered the rationale in favor of early surgery, as articulated by disciples of Chavasse in the U.S., most notably August Costenbader, Marshall Parks, and a series of Parks’ trainees [109, 110]. The new knowledge prompted a gradual reexamination of old data and inspired important case studies – in the 1980 and 1990s – on the e cacy of early strabismus surgery [111–114]. These reports

showed that if stable, binocular alignment was not achieved until age 24 months, the chances of repairing stereopsis were nil. If stable alignment was achieved by age 6 months, the chances of repairing stereopsis were good, and a substantial percentage of the infants regained robust stereopsis, i.e., random dot stereopsis with thresholds on the order of 60–400 arcsec.

Scrutiny of early alignment data in infantile esotropia has produced more refined and forceful conclusions. Figure 5.6a is replotted data on stereopsis outcomes in over 100 consecutive infantile esotropes [112]. The Y-axis is prevalence of stereopsis after surgical alignment, and the X-axis is age of onset or duration of misalignment before surgery. The dashed line at 40% represents the average prevalence of stereopsis when all infants operated upon by 2 years of age are grouped together, without regard to age at correction or duration before correction. The noise in the data – relating age at alignment to stereopsis outcome – is related to the fact that onset of strabismus is idiosyncratic, varying considerably from infant to infant, and distributed randomly in the interval 2–6 months of age.There is no systematic relationship between age of onset of esotropia and subsequent attainment of stereopsis. However, when the data is reanalyzed with strict attention to duration of misalignment, a strong correlation is evident between shorter durations of misalignment and restoration of stereopsis (Fig. 5.6b). Excellent outcomes are achievable in infants operated upon within 60 days of onset of strabismus (“early surgery”) [112]. The clinical dictum that follows is that age at surgery should be tailored to age of onset and not chronological age.

Esotropic infants who regain high grade stereopsis also regain robust fusional vergence [112–114]. Clinical observation also suggests that they have a lower prevalence of recurrent esotropia (or exotropia), pursuit/OKN asymmetry, motion VEP asymmetry, latent nystagmus, and dissociated vertical deviation (DVD). However, ocular motor recording is di cult to perform in children and detailed, quantitative information is lacking.

5.1.18Timely Restoraion of Correlated Binocular Input: The Key to Repair

Eye movement studies of strabismic infant monkeys have helped fill gaps in clinical knowledge. The studies have shown that normal motor and sensory pathway development can be restored when the timeliness of therapy conforms to that of early surgery in humans [47, 115]. If binocular image correlation is restored in strabismic monkeys within 3 weeks of onset of strabismus (the equivalent of 3 months in humans), fusional vergence,

a

Stereopsis

with

Children

%

b

Stereopsis

with

Children

%

100

80

60

40

20

0

100

80

60

40

20

0 0-2 3-5 6-8 9-11 12-18 19-24

Duration of Misalignment (months)

pursuit/OKN and gaze holding return to normal (Fig. 5.6c). The repair of ocular motor behavior occurs with repair of stereopsis and restoration of normal motion responses (motion VEPs). If decorrelation persists in strabismic monkeys until the equivalent of 12 months’ duration in humans, esotropia and stereoblindness persist. Prolonged-decorrelation animals exhibit latent nystagmus, pursuit/OKN asymmetry, motion VEP asymmetry, and DVD. The quality of behavioral repair correlates with the quality of neuroanatomic repair in V1 (Fig. 5.6c). “Early repair” monkeys (i.e., those who have shorter durations of decorrelation) have a normal complement of binocular horizontal excitatory connections between ODCs of opposite ocularity, and “delayed repair” (longer durations of decorrelation) monkeys a paucity. The restoration of binocular connections in V1 of “early repair” monkeys appears to have equally benefical e ects on downstream areas of extrastriate cortex (MT/MST) driving the ocular motor neurons of the brainstem. The benefit is evident as symmetric nasotemporal eye tracking, stable gaze holding, and more normal fusional vergence.

Fig. 5.6 Repair of random-dot stereopsis after surgical realignment of the eyes in children with infantile esotropia, and analogous findings in strabismic monkeys. (a) Prevalence of stereopsis as a function of age-of-onset of strabismus. No systematic relationship is evident. (b) High prevalence (~80%) of stereopsis in infants who were aligned within 2 months of onset of strabismus. Probability of stereopsis was negligible in infants who had durations of strabismus exceeding ~12 months. Redrawn from data of Birch et al. [112]. (c) Magnitude of behavioral deficits increases systematically as a function of decorrelation-duration in monkeys. One week of monkey visual development is equivalent of 1 month in humans. Pur Asymm horizontal pursuit asymmetry; Nyst velocity of latent nystagmus; Stereo random dot stereopsis deficit; Eso angle of esotropia; DVD magnitude of dissociated vertical deviation; V1 binoc reduction in binocular connections between RE and LE ODCs in V1 (striate cortex)

5.2Visual Cortex Mechanisms in MicroEsotropia (Monofixation Syndrome)

As outlined earlier, recent data on early correction of infantile strabismus suggests that it is a curable disorder. But early surgery is the exception rather than the rule of current clinical practice in the U.S. and Europe. The majority of infants who have esotropia are corrected 6 or more months after onset of misalignment. The chances of rescuing bifoveal fusion after this interval are slim. Most infants are aligned to within 8 PD of orthotropia (microesotropia) and regain a degree of subnormal stereopsis and motor fusion, i.e., monofixation syndrome.

Monofixation syndrome occurs as a primary disorder (prevalence 1%) or, more commonly, as a secondary phenomenon, after delayed treatment of large magnitude strabismus [116, 117]. The syndrome also occurs in monkeys [118]. The major sensory and motor features of monofixation syndrome are listed in Table 5.4. Neural mechanisms for the first two features listed in Table 5.4 are not di cult to explain. Receptive fields in V1 – representing the fovea – are tiny and have narrow tolerances. Any defocusing or other decorrelation of one eye’s inputs would produce a conflict in neighboring V1 columns and promote suppression of ODCs corresponding to the weaker eye. The fovea subtends ~5° of the retinotopic map of V1, thus a suppression scotoma of ≤5° makes sense. Feature two, subnormal stereopsis,

could be explained along similar lines. Stereoscopic thresholds increase exponentially from the fovea to more eccentric positions along the retinotopic map of the visual field. If foveal ODCs are suppressed and parafoveal ODCs are left to mediate stereopsis; stereopsis is degraded but not obliterated. But it is features three and four of the monofixation syndrome, the visuomotor signs, that are most intriguing. If binocular development is perturbed so that right and left eye foveal ODCs (receptive fields) do not enjoy perfectly correlated activity, why should the fall back position of visual cortex be set so predictably ~2–4° (~4–8 prism diopters or PD) of micro-esotropia (Fig. 5.7)? And if the heterotropia exceeds that range, why is fusional vergence typically absent?

5.2.1Neuroanatomic Findings in Area V1 of Micro-Esotropic Primates

Studies of ODCs and neuronal axons in area V1 have revealed a possible mechanism. The overall pattern and width of ODCs in V1 (~400 mm [0.40 mm]) is the same in normal and strabismic monkeys [70, 78]. Horizontal axon length was measured for neurons within the V1 region corresponding to visual field eccentricities of 0–10° (i.e., the representation of the fovea, parafovea and macula). The length is similar in both normal and strabismic monkeys, on average ~7 mm [70, 119]. In a primate with normal eye alignment, the ODC representing the foveola (or 0° eccentricity) of the left eye is immediately adjacent to the column representing the foveola of the right eye. The side-by-side arrangement of the “foveolar” columns in normal V1 is well within the range of horizontal axonal connections needed to allow those ODCs to communicate for high-grade binocular fusion.

In a primate with microesotropia and a right eye fixation preference (Fig. 5.7), a neuron within a foveolar (0°) column of the fixating, right eye must link up with a nonadjacent column representing the pseudo-foveola of the deviated, left eye. Based on retinotopic maps of V1 in macaque monkey, a horizontal axon ~7 mm in length could join ODCs (and receptive fields) that were up to but not further than 2.5° apart, or converting deg to PD, not more than 4.4 PD. Shown here is a 2-dimensional map representing V1 from the right cerebral hemisphere (left visual hemi-field) of a microesotropic macaque. The sulci and gyri have been unfolded and the visual field representation superimposed using standard retinotopic landmarks. One horizontal axon, originating within the foveal representation at 0–1° eccentricity, could link to a receptive field shifted 2.5° or 4.4 PD distant (Fig. 5.7). Two neurons strung together could join receptive fields 5° or 8.7 PD apart. The conclusion that emerges is that the 4–8 PD “rule” of the monofixation syndrome is explicable as a combination of innate V1 neuron size and V1 topography. The visuomotor system of the strabismic primate appears to achieve subnormal, but stable binocular fusion so long as the angle of deviation is confined to a distance corresponding to not more than one to two V1 neurons [119].

5.2.2Extrastriate Cortex in Micro-Esotropa

Neuronal response properties of the vergence-related region of extrastriate visual cortex, MST, may also explain the 2.5°-microesotropia rule in monofixation syndrome. MST receives downstream projections from disparity-sensitive cells, both in V1 and in MT. The majority of binocular neurons in V1, MT and MST encode absolute disparity [82, 120]. Absolute disparity

Fig. 5.7 (a) Monofixator/ microesotrope exhibits a deviation of the visiual axes on cover testing of approximately 4 PD (~2.5°), which in this case is shown as a left eye microesotropia (dark arrowhead pseudofovea position in deviated eye). When fusional vergence or prism adaptation is tested in such a patient, the angle of deviation tends to persistently return to that 2.3° angle. (b) Two-dimensional map representing V1 from the right cerebral hemisphere (left visual hemi-field) of a microesotropic primate. The sulci and gyri have been unfolded and the visual field representation superimposed using standard retinotopic landmarks. One horizontal axon (average length ~7mm), originating within the foveal representation at 0–1° eccentricity, could link to a receptive field shifted 2.5° or 4.4 PD distant. Two neurons strung together could join receptive fields 5° or 8.7 PD apart. The conclusion that emerges is that the 4–8 PD “rule” of monofixation/ microesotropia syndrome is explicable as a combination of innate V1 neuron size (one to two axon lengths) and V1 topography

a

2.5° (4.4 PD)

Left Esotropia

sensitivity (the location of an image on each retina with respect to the foveola, or 0° eccentricity) guides vergence, as opposed to relative disparity sensitivity (the location of an image in depth with respect to other images), which is necessary for stereopsis. The largest population of vergence-related neurons in MST of normal monkeys drives the eyes to ~2.5° of convergent (crossed) disparity [82]. (The next largest population encodes ~2.5° of divergence.) Normal primates have the strongest short-latency vergence responses to convergent disparities of ~2.5° [121].

Insults that impair the development of binocular connections in immature V1 would be expected to impair the (downstream) development of the entire population of

binocular MST neurons. The probability of surviving an insult would be the greatest for the most populous neurons: those encoding ~2.5° (~4.4 PD) of convergence. In the presence of a generally weakened pool of disparitysensitive neurons, the vergence system may default to the vergence commanded by the surviving population. A 2.5° convergence angle could be kept stable (preventing deterioration to large angle strabismus) by the next most populous remaining neurons, those encoding 2.5° of divergence. These mechanism are attractive because they can account for the direction, approximate magnitude, and stability of microesotropia, with retention of a capacity for fusional (e.g., prism) vergence responses evoked by disparities >2.5°.