Plasticity of human motion processing mechanisms following surgery for infantile esotropia

Pergamon

0042-6989(95)00144-1

Vision Res. Vol. 35, No. 23-24, pp. 3279 3296, 1995 Copyright © 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0042-6989/95 $9.50 + 0.00

Plasticity of Human Motion Processing Mechanisms Following Surgery for Infantile Esotropia ANTHONY M. NORCIA,*t RUSSELL D. HAMER,* A R T H U R JAMPOLSKY,* DEBORAH OREL-BIXLER~ Received 9 November 1994; in revised form 1 May 1995

Monocular oscillatory-motion visual evoked potentials (VEPs) were measured in prospective and retrospective groups of infantile esotropia patients who had been aligned surgically at different ages. A nasalward-temporal response bias that is present prior to surgery was reduced below pre-surgery levels in the prospective group. Patients in the retrospective group who had been aligned before 2 yr of age showed lower levels of response asymmetry than those who were aligned after age 2. The data imply that binocular motion processing mechanisms in infantile esotropia patients are capable of some degree of recovery, and that this plasticity is restricted to a critical period of visual development.

Infantile esotropia

Strabismus surgery Visual evoked potentials Motion processing Binocular vision

INTRODUCTION

Patients with a history of early onset esotropia exhibit a number of abnormalities in their oculomotor responses that imply the presence of a nasalward-temporalward asymmetry at one or more stages of motion processing. First, their monocular optokinetic nystagmus (MOKN) is asymmetric, having a characteristically weak temporalward slow phase (Schor & Levi, 1980; Mein, 1983; Kommerrell, 1988; Demer &von Noorden, 1988; Reed, Steinbach, Anstis, Gallie, Smith & Kraft, 1991; Westall & Shute, 1992). Second, their pursuit eye movements are asymmetric--they fail to track smoothly the temporalward motion of small targets (Tychsen, Hurtig & Scott, 1985; Tychsen & Lisberger, 1986), and pursuit acceleration in the step-ramp task has significantly higher openloop gain for nasalward vs temporalward motion (Tychsen & Lisberger, 1986). Third, infantile esotropia patients report that targets moving nasalward appear to be of a higher velocity than temporalward moving targets with the same objective velocity (Tychsen & Lisberger, 1986). Tychsen and Lisberger (1986) suggested that asymmetries in pursuit eye movements and in perceived velocity imply that analogous asymmetries must be present at the level of visual cortex. Norcia, Garcia, Humphry, Holmes, Hamer and Orel-Bixler (1991) have

found evidence for a cortical motion asymmetry in patients with a history of early onset esotropia treated after 2 yr of age. They found that monocular oscillatorymotion visual evoked potentials (VEPs) from these patients contained both first (F~) and second harmonic (F2) components, while normal, mature responses were dominated by F2. The presence of a F~ component is consistent with a directional asymmetry of cortical responsiveness. Moreover, the F~ components had opposite temporal phase in the two eyes, indicating that largest response from each of the eyes had resulted from opposite directions of motion. The absolute direction of the asymmetry, nasalwards or temporalwards could not be recovered from the FI data, since there is an inherent 180 deg phase ambiguity introduced by the unknown polarity of the response. Kommerrell, Ulrich and Bach (1995) have replicated Norcia et al.'s results for visually mature infantile esotropia patients A similar pattern of asymmetric VEP responsiveness occurs early in normal development (Norcia, et al., 1991; Jampolsky, Norcia & Hamer, 1994; Norcia, Hamer & Orel-Bixler, 1990a). Depending on the spatial and temporal conditions of the test, normal infants achieve a symmetric motion response between 5 months and approx. 2 yr of age (Norcia et al., 1990). Normal infants also show asymmetric MOKN (Atkinson, 1979; Naegele & Held, 1982) and nystagmus is elicited by directionally ambiguous reversing gratings viewed monocularly (Teller, Succop & Mar, 1994). Unlike normal neonates or patients with early onset strabismus, patients with onset of esotropia after age 2 do not show pronounced nasalward-temporalward direction biases in their motion VEP (Hamer, Norcia,

*Smith Kettlewell Eye Research Institute, 2232 Webster Street, San Francisco, C A 94115, U.S.A. t T o w h o m all correspondence should be addressed [Email [email protected]]. :~School of Optometry, University of California--Berkeley, Berkeley, CA 94720, U.S.A. 3279

3280

ANTHONY M. NORCIA et al.

Orel-Bixler & Hoyt, 1993), MOKN (Westall & Shute, 1992) or pursuit eye movements (Tychsen et al., 1985; Sokol, Pelli, Moskowitz & Reese, 1991). The different patterns of VEP and oculomotor responses in early and late onset esotropia patients suggests that there is a developmental sensitive period during which an interruption of binocularity is likely to result in directional asymmetries in cortical motion responsiveness and in oculomotor behavior. The magnitude of the motion VEP asymmetry can be reduced by full-time alternate occlusion prior to surgery for infantile esotropia (Jampolsky et al., 1994) and Westall and Shute (1992) have reported that part-time patching can modify the degree of OKN asymmetry in amblyopic patients. These results also suggest that the mechanisms underlying the various developmental motion asymmetries are experience dependent and furthermore, that they exhibit plasticity typically associated with developmental sensitive/ critical periods (Hubel & Wiesel, 1970; Sherman & Spear, 1982). The present report describes the results of oscillatorymotion VEP recodings made in patients who had undergone surgery for infantile esotropia, a form of constant strabismus that occurs before 6 months of age (Costenbader, 1950; Ciancia 1962; Taylor, 1973; von Noorden, 1988). The status of visual motion processing mechanisms of patients operated on between 9 months and 3 yr of age was compared before and after surgery in a prospective series and oscillatory-motion VEPs from infantile esotropia patients who achieved early vs late alignment were also compared in a retrospective series. A more normal motion VEP response developed after early and accurate re-alignment of the eyes, suggesting the existence of a developmental critical period for binocular, direction-selective mechanisms in human visual cortex.

METHODS

Patient selection

Patients with an onset of constant esotropia prior to 6 months of age were studied. No other clinical variables were used to define the patients as having infantile esotropia, although additional clinical signs known to be associated with infantile esotropia were noted when present (see Table 1). All patients had been born within 3 weeks of term, without complications and none had a history of CNS abnormalities. One group of patients (n = 14) was studied prospectively, before and after strabismus surgery. Patients in this group were between 9 months and 3.6yr at the time of surgery (mean of 1.5 yr with a SD of 38 weeks; median age of 1.3 yr). Two retrospective groups were studied: the first comprised 11 patients with onset of constant strabismus before 6 months of age by parental report, but who did not achieve alignment during the first 2 yr of life. The average age of this group at the time of testing was 13.7 yr (range 9-26 yr) and all of these patients required two or more surgeries. The clinical features of this group are summarized in Table 2. The second retrospective group (n = 9) also had an onset of constant esotropia before 6 months of age by report of the parents but in each of these cases surgery was performed before 20 months of age. In 7 of these 9 patients a single surgery was sufficient to maintain alignment to within 10Aup to the time of testing, but two required an additional surgery 7 yr after their initial surgeries at 9 months. The clinical features of these patients are summarized in Table 3. Adult and infant control observers

Motion responses of the esotropic patients were compared to those of normal adults and normal infants tested on the same apparatus. The 12 normal adults each

TABLE 1. Patient characteristics for prospective sample

Patient GB RC BF NH ALL ANL CPA CPE KR ES NS KS JS CT Averages:

Pre-op VEP age (weeks)

Horizontal deviation at pre-op VEP (Prism D)

Pre/post DVD?

Surgery age (weeks)

68 188 50 111 74 39 47 69 72 39 86 125 61 51

60 55 55 30 20 60 50 50 45 25 60 50 35 30

No/yes No/yes No/yes No/no No/yes No/no No/yes No/no No/yes No/no No/yes No/yes No/no No/no

68 188 50 115 74 39 47 69 72 49 86 128 61 51

77.14

44.64

78.36

Type of surgery (Op No. l//Op No. 2) MRi, MRi, MRi, MRi MRi, MRi, MRi, MRi MRi, MRi MRi, MRi, MRi, MRi

IOs IOs//SRi LRi IOs//SRi, LLR LRi LRi lOs IOs//LIO, LRi* IOs lOs

Post-op VEP age (weeks) 210 251 173 171 139 209 70 106 225 169 215 237 149 98 173.00

Deviation at Post-op VEP (prism D) - 12 25 0 2 -14 0 0 0 - 8 2 12 - 10 0 0 6.07

The age at surgery and type of surgery are indicated, along with the deviations present at the time of the pre-operative and post-operative recordings.

3281

STRABISMUS SURGERY AFFECTS MOTION PROCESSING TABLE 2. Patient characteristics for late-alignment, retrospective sample Surgery age(s)

Patient BIB TG ZM EM RP MR KS MS CS KS CT

9.5, 9.6 yr 2, 6.3, 8.7 yr 11.4 months, 2, 6.7, 7.7, 10.2, 13.5yr 22.1 months, 5.2, 6.6 yr 2, 4, 10.7, l l . 8 y r 17.3, 34.7, 39.5 months 12, 26,6 yr 1.5, 19.l yr 2.0, 11.1, 12.2yr 12.4, 21.4 months 10, 17 months 8.4 yr

Pre/post DVD?

VEP age (yr)

Deviation at post-op VEP (Prism D)

No/no NA/yes No/yes NA/yes NA/no No/yes NA/yes NA/yes NA/yes No/yes Maybe/yes

11.7 8.7 13.8 8.2 17.l 9.7 26.7 19.0 13.7 11.8 8.5

25 4 -12 Ortho 12 Ortho - 14 35 -2 - 15 Ortho

Ages at the relevant surgeries are listed along with the deviation at the time of the VEP recording.

had optotype acuity of at least 6/6 in each eye, had no interocular acuity differences exceeding 0.33 octaves, were orthophoric on cover, alternate cover and 4 ~ base-out prism tests, and each demonstrated 40 sec arc stereoacuity on the Titmus test. Twenty-two presumptively normal infants between 25 and 104 weeks of age also participated. Each infant had been born at term (_+3 weeks) after an uncomplicated pregnancy and delivery. The infants were screened for high hyperopia, anisometropia and normal eye-alignment using an offaxis photorefractor (Day & Norcia, 1986) and each appeared to be normal on this measure. Informed consent was obtained directly from the adult observers and from the parents of the infants and children.

VEP recording and procedure Motion VEPs were measured monocularly in response to vertical sinusoidal gratings displayed on a video monitor. Monocular viewing was achieved in older children and adults by occluding one eye with a translucent diffuser. Viewing was done with the fixing eye in adduction to minimize any latent nystagmus. Infants were occluded with an opaque patch (Opticlude). Their fixation was controlled through the use of small, noisy toys suspended on strings in front of the monitor. Attempts were made to hold the esotropic infants so that the fixating eye was in adduction, but this was not always successful. We did not observe any stimulus evoked eye

movements, although some of the esotropic infants did appear to have latent nystagmus. The experimenter acquired data when the corneal reflection of the video monitor was centered in the patient's pupil. The gratings were presented on a 10 × 20 deg field wi~h a space-average luminance of 80 cd/m 2 and a Michelson contrast of 80%. The viewing distance was 1 m. The gratings were shifted synchronously with the video frame between two positions separated by 90 deg of spatial phase ( 1 5 m i n a r c for l c/deg gratings and 5 m i n a r c for 3 c/deg gratings) at either 6 H z (83.3 msec per position) or l0 Hz (50 msec per position). The 6 and l0 Hz oscillations thus had 12 or 20 changes of direction per sec. The spatial frequency of the grating was scaled to be at least 5 times lower than the observer's expected acuity based on age: infants and toddlers in the prospective sample were tested with l c/deg gratings presented at 6 Hz; children and adults in the retrospective group were also tested with 3 c/deg gratings presented at 10 Hz. The gratings were displayed for 10 sec and several trials were recorded from each eye. The direction of grating motion was always the same at the start of each trial and all data collection was strictly time-locked to the stimulus. The E E G was digitized to 8 bits of accuracy at 180 Hz. Variable-gain amplifiers were used to adjust the total system gain under program control to ensure that the analog E E G filled, but did not exceed, the range of the

TABLE 3. Patient characteristics for early-alignment, retrospective sample

Patient EB IB BEB ME KL NM VM DS JT

Surgery age(s) 12,8 months 1 yr 11.3 months 9 months, 7.9 yr 16.3 months 13 months 13 months 19.2 months 9 months, 7.4 yr

Pre/post DVD?

VEP age (yr)

Deviation at post-op VEP (Prism D)

No/yes NA/yes No/yes NA/no Yes/yes NA/yes NA/yes No/yes NA/yes

4.9 18.1 9.0 9.4 7.4 9.6 9.6 9.5 26.0

0 to - 5 - 5 to 5 - 5 Ortho 10 Ortho Ortho 8 Ortho

Ages at the relevant surgeries are listed along with the deviation at the time of the VEP recording.

ANTHONY M. NORCIA et al.

3282

analog-to-digital converter. Two bipolar derivations were used: O, vs 3 cm to the left of Oz at O~ (CH~); Oz vs 3 cm to the right of Oz at 02 (CH2). This montage has been used previously for studies of infant spatial vision (Norcia & Tyler, 1985; Norcia, Tyler & Hamer, 1990b) and it generally provides higher signal-to-noise ratio than monopolar recordings (Orel-Bixler & Norcia, 1987).

VEP data analysis The EEG was subjected to Fourier analysis to extract the amplitude and phase of the evoked response at F~ (6 or 10 Hz) and F2 (12 or 20 Hz). Equal responsiveness to each direction of motion implies that the response waveform repeats itself exactly at the direction reversal rate which is twice the stimulus frequency. The lowest frequency component of the symmetric response will thus be at ~ . If the response is not equal for the two directions of motion, the waveform will repeat itself exactly at the stimulation rate--6 or 10Hz--and the lowest frequency component in the response will be at F~. In each case, higher harmonic components may also be present. A discrete Fourier transform (DFT) was calculated on the basis of each 10 sec data record by multiplying the EEG f(t) point-by-point by sine and cosine waveforms of the appropriate frequency. The real coefficient of the DFT is 1 u ~¢~ = ~ ,=~lf(t)cOs 2rcnvot and the imaginary coefficient is

so that all signal and noise frequencies were exact integer multiples of the 0.1 Hz frequency resolution of the DFT. Since a rectangular window was used, there was no leakage of energy between the signal and noise frequencies (see Harris, 1978 for details.) Each 10-see trial generated four VEP records comprised each combination of recording channel and harmonic. All analyses were based on vector averages of three or more individual trials recorded under the same conditions. The degree of symmetry of the motion VEP was quantified by comparing the relative proportion of F~ (asymmetric component) and F2 (symmetric component) amplitudes. An asymmetry index was calculated by dividing the F~ amplitude by the sum of the FI and F2 amplitudes. The asymmetry index ranges between 0 and 1, with higher values corresponding to greater degrees of asymmetry. Significance tests and error estimates for each vector average were determined by the T~r<: test of Victor and Mast (1991) if at least four trials were available in a given condition. The error estimate for the asymmetry index F,(F, + F2) is

(g~ + F2)2~[\ F, /t + \ F, J where ae, is the SE. When fewer than four trials were available, the amplitude at the response frequency was compared to the average amplitude recorded at the noise frequencies adjacent to the recording frequency. A response was considered to be significant if its signal-to-noise ratio was >3:1 (Norcia, Tyler & Clarke, 1985).

1 L f ( t ) s i n 2rtnVot RESULTS where j is the trial number, N is the number of data samples (1800), n is the harmonic number (1 for F~ and 2 for F2) and v0 is the fundamental frequency (either 6 or 10 Hz). The response amplitude V~ for the j t h trial, nth harmonic is

v.j _ _

j

2

-

+

~ 2

and the response phase qS~ is ~b~ = tan-' (~J"~.

t,d,,)

The vector average amplitude for M trials is

V:w V \

} +

and the vector average phase is q~v = tan I "~dM--I ~Jn No data window was used and the resolution of the DFT was therefore 0.1 Hz. The mean amplitude at a pair of adjacent nonharmonic frequencies was used as a noise estimate for each response frequency. These frequencies were located _+2 Hz from each response frequency. Care was taken

Frequency domain signature of cortical motion asymmetry A nasalward-temporalward response asymmetry produces a characteristic signature in the frequency domain: the steady-state VEP contains significant odd-harmonic components that are 180 deg out of phase in the two eyes. Figure 1 compares the amplitude spectra of a normal, visually immature infant (top panel), a visually mature infantile esotropia patient (age 12 yr) who was aligned after age 2 (middle panel) and a normal adult (bottom panel), each recorded monocularly at 6Hz, l c/deg. The normal 12 week-old infant's monocular response was dominated by first harmonic (Fj) and second harmonic (F2) components, as was the response of the 12 yr old infantile esotropia patient. F 3 and F4 components were present in the response of the visually mature infantile esotropia patient, but not in the normal infant response. The normal adult response contained both F2 and F4 components, with non-significant activity at F~ and F 3. All data reported below was derived solely from analyses of F~ and F2 components since these were the largest and most reliable components of the response across subject groups. Figure 2 presents polar plots of the oscillatory-motion VEP responses that show both the amplitude and the

STRABISMUS SURGERY AFFECTS MOTION PROCESSING

INFANT (12 wks)

1

10

7.5

5.0 ¸

"2

2.5 ¸

l,h

~IIi.ll,k,=.,,.iu,.,.hh,,,i,.,,Ll.,i.,................

0

6

12

18

24

30

INFANTILE

O3

F2

ET

>O 0.75 O 0

"E'~ 0.50. . 0.25

F3

<

lli il.L, 0

a,, .u=,JLLJit ,J ...... .i=a.

.

0

6

12

18

24

30

1 F2

NORMAL ADULT

0.75.

0.50.

0.25 t

0

6

12

Hz

18

24

30

FIGURE I. Representative amplitude spectra for monocular oscillatory-motion VEPs recorded from a normal, visually immature infant (top), a visually mature infantile esotropia patient (age 12 yr middle) and a normal adult (bottom), The monocular motion VEP is dominated by F~ and F2components in the normal infant and in the infantile esotrope. The normal adult response is dominated by the F2 component. Higher harmonic components were present in the older observers, but not in the 12 week-old infant.

phase of the evoked response at F l (upper panels) and F2 (lower panels). The data illustrated in the figure are typical both of normal infants at different stages of *A patient was considered to have a nasalward-temporalward asymmetry if the F~ components were statisticallysignificant in both eyes within a recording channel and if the phase differencebetween them was 180 deg, within the measurement error.

3283

development (a, b) and of infantile esotropia patients prior to surgery (c). The length of each vector indicates the response amplitude for a 10 sec trial, and the angle made with the axes indicates the response phase. Right eye (RE) and left eye (LE) responses are indicated by the solid and dashed vectors respectively. Each data set was recorded at 6 Hz, l c/deg. Figure 2(a) illustrates the motion VEP from a 10 week-old normal infant. The F1 components are highly significant and are 180 deg out of phase between the two eyes, forming a characteristic "bowtie" configuration in the upper plot of Fig. 2(a).* By contrast, the mature motion VEP has a large F 2 component, and a small or absent Fj component, as seen in the responses from a 31 week old, normal infant [Fig. 2(b)]. Figure 2(c) shows the motion VEP data recorded from a 38 week-old esotropic infant (NS) prior to surgery. At the time of recording, the patient had a 35 ~ alternating esotropia. The response of the esotropic infant shows the characteristic bowtie signature in the monocular F~ responses associated with a nasalward-temporalward response asymmetry [upper plot, Fig. l(c)] The motion asymmetry in this patient was present beyond the normal age of maturation for these spatio-temporal conditions (Norcia et al., 1990a; Jampolsky et al., 1994).

Prospective studies Cross-sectional results. For each recording session, asymmetry indices were calculated separately for each combination of eye and recording channel. Asymmetry indices for which both F1 and F2 components did not reach statistical significance on the T~rc test or those with less than four trials per eye were excluded and the comparison was done between the recording immediately preceding surgery and the last recording available (requiring a minimum follow-up of 23 weeks with a mean follow-up of 95 weeks). An initial within-subjects, repeated measures analysis of variance (ANOVA) was performed to determine if the responses from the left and right eyes, and from the leftand right-hemisphere recording channels were equivalent. There was no a priori reason to expect that left or right eyes would be differentially affected by strabismus or by treatment, or that left- and right-hemisphere derivations would differ. This A N O V A utilized all patients from Table 1 who had complete data sets (n = 8), regardless of their age at surgery or their clinical outcome. The factors analyzed were eye (left and right; Eye), recording channel (CHI, CH2; Chan) and recording session (pre-operative, and last post-operative; Session). The patients who were not included in the analysis either did not have complete recordings for both eyes at each session, or did not have a statistically reliable asymmetry index for each eye/channel combination. This analysis showed a non-significant trend for the left eyes to have higher asymmetry indices than right eyes (Eye: F = 4.86; P = 0.06) and no significant effect of recording derivation (Chan: F = 2.00; P = 0.20). There was a highly significant main effect of Session (F = 20.86; P = 0.0026), with the post-operative asymmetry indices

3284

ANTHONY M. NORCIA a.

b,

Normal

Normal (31 wks)

(10 wks)

et al.

C°

Infantile Esotropia (38 wks)

F1

FIGURE 2. Polar plots of oscillatory-motion VEP responses from three infants. For each infant, the response is comprised both first (Fl, top plot) and second harmonic (F2, bottom plot) components. Each line represents the amplitude (length) and phase (angle relative to horizontal axis) of a single 10-sec motion VEP trial (RE, solid lines; LE, dashed lines). • and O plot the vector average of the individual trial responses for the RE and LE, respectively. (a) The pattern of response typical of immature normal infants (age 10 weeks). The response has a prominent F~ component, consistent with an asymmetric cortical response to the oscillatory motion stimulus. Moreover, the F t responses are 180 deg out of phase between the two eyes which is indicative of a nasalward-temporalward response bias. (b) The pattern of response typical of a normally maturing infant (age 31 weeks). The mature response shows no evidence of a nasalward-temporalward asymmetry. The response is dominated by responses at F 2, whereas the F~ component is not statistically significant. (c) The developmental motion asymmetry in a 38 week old infantile esotropia patient prior to surgery. As in the case of the normal immature response, this patient shows prominent (F~) response components that are 180 deg out of phase between the two eyes. The developmental motion asymmetry in this patient was present beyond the normal age of maturity (approx. 20 weeks).

b e i n g lower t h a n the p r e - o p e r a t i v e indices. T h e m e a n s for the m a i n effect o f Session are s h o w n in Fig. 3(a). T h e r e was also a s i g n i f i c a n t Session x Eye × C h a n i n t e r a c t i o n ( F = 11.5; P = 0.01). N o n e o f the o t h e r intera c t i o n s were significant. A s e c o n d A N O V A r e c o r d e d the Eye v a r i a b l e a c c o r d i n g to e y e - d o m i n a n c e as i n d i c a t e d b y the p a t i e n t ' s

a.

b.

t

.8 X

"o --= .6

i

clinical record. Eye d o m i n a n c e ( d o m i n a n t o r n o n - d o m i n a n t ) was b a s e d o n the existence o f a n o n - a l t e r n a t i n g d e v i a t i o n o r fixation preference. If n o p r e f e r e n c e was seen o n a given visit, d o m i n a n c e was a s s i g n e d o n the basis o f the p r e v i o u s visit(s). T h e r e c o r d i n g c h a n n e l s were r e c o r d e d so as to reflect p o s s i b l e differences between crossed a n d u n c r o s s e d c o r t i c a l p r o j e c t i o n s (cf.

Contra •

Ipsi

.4

.2 0

Preop.

I

P

I

Postop.

Preop.

Postop.

FIGURE 3. Pre- and post-operative group-mean asymmetry indices for the eight infants with complete data sets. The VEP motion asymmetry is significantly lower after surgery (a). (b) The Session x Pathway interaction effect that results from a greater decrease in asymmetry after surgery as measured by the ipsilateral electrode derivations.

STRABISMUS SURGERY AFFECTS MOTION PROCESSING Ciancia, Borrone, Schuarzberg & Garcia, 1988). The Oz-O 2 derivation records the gradient between the midline and the right hemisphere. When the left eye is viewing, the right hemisphere receives the crossedpathway projection from the left eye. Similarly, the O..-O t derivation records the gradient between the midline and the left hemisphere which receives the uncrossed projection. It should be noted that there is no independent evidence that the derivations used in this study are differentially selective for the hemisphere they overlie. However, recoding variables in this fashion allows one to test for topographic differences that may have the symmetry relationships implied by the partially decussated projection of the visual hemifields onto striate cortex. The LE-CH1 and RE-CH2 records were grouped together as were the LE-CH 2 and RE-CH, records. These recoded derivations will be referred to as ipsilateral and contralateral with respect to the viewing eye. The same main effect of Session was found, as expected, after the Eye and Chan variables were recorded according to dominance and projection (ipsilateral vs contralateral; Pathway). The three-way interaction between Eye, Chan and Session seen in the first analysis was resolved into a Session x Pathway interaction ( F = 11.5; P =0.01). This interaction is plotted in Fig. 2(b). The reduction of the motion asymmetry after surgery was larger for the ipsilateral derivations than for the contralateral derivations. There were trends for the ipsilateral derivations to show lower asymmetry indices (F = 4.0; P = 0.086) overall and for the dominant eye to show less asymmetry (F = 4.9; P = 0.09). The general pattern of results found in the patients who had complete data sets was also seen in the group as a whole: group mean asymmetry indices for all patients combined were lower after surgery, the ipsilateral derivations showed lower indices post-operatively than did the contralateral derivations and the dominant eyes tended to have lower asymmetry indices than the non-dominant eyes. Longitudinal results: individual infants. The overall decline in motion asymmetry demonstrated in the group data can also be seen within individual infant's data. All patients in the prospective study showed a statistically significant decrease in the asymmetry index for at least one eye/channel measure. In some cases, asymmetry indices fell to within normal limits. Nevertheless, the magnitude of the decrease in the motion asymmetry index differed substantially both between individuals and within individual patient's data (between conditions). Figure 4 shows data from four patients who exhibited the most consistent decreases in motion asymmetry. In this figure and the next, horizontal and vertical eye alignment data ( 0 ) and motion asymmetry indices (©) are displayed on a time-line relative to the time of surgery (A). Half-filled circles on the alignment curves indicate intermittent deviation. A star denotes a significant change in spectacle prescription and a " D " denotes the presence of dissociated vertical deviation (DVD). The surgeries for the four patients in Fig. 4 all occurred

3285

between 39 and 51 weeks of age and each patient achieved a post-operative deviation of < 10A that was stable over a period of at least 6 months prior to the final post-operative VEP measurement. This degree of alignment will be referred to as "criterion" alignment. For two of the four patients (CPA and CT) criterion alignment was achieved immediately after surgeries at 47 and 51 weeks respectively. For the other two patients (ES and ANL) criterion alignment was achieved by the 40th post-operative week at ages 89 and 79 weeks, and was then maintained throughout the period of VEP testing. The sub-panels in Figs 4 and 5 immediately below the eye alignment data show the VEP data for each recording session. Pre- and post-operative motion asymmetry indices for the RE and LE (open symbols, solid curves) for each of the two recording channels (CH~, lefthemisphere; CH2, right-hemisphere), along with the SEMs for each of the asymmetry index measurements are shown. The horizontal shaded bands indicate the 95% confidence limits for the median asymmetry indices of 22 normal infants between 25 and 104 weeks of age. Patient CPA [Fig. 4(a)] had an onset of esotropia by 1 month of age according to parental report. The deviation was first documented by an ophthalmologist as a 45 a left esotropia (without spectacle correction). The patient underwent full-time alternate occlusion up to the time of surgery which was performed at 47 weeks of age. CPA's deviation 1 day prior to surgery was approx. 50A of esotropia (with spectacle correction) with free alternation. One week after surgery, CPA's eye alignment was orthotropic with approx. 10Aof intermittent right hypertropia. Criterion alignment was maintained over the next 60 weeks. During this time the patient wore full spectacle correction and was managed with a diminishing regimen of part-time alternate occlusion. Three VEP measurements were made prior to surgery, all of which manifested abnormally high asymmetry indices. The final pre-operative recording was performed the day before surgery. The motion asymmetry indices at this time were all abnormal (~>0.54 for each of the four measures). In addition to the abnormal asymmetry indices, characteristic bowtie configurations were present in the F~ data from both the second and third pre-operative recordings. Post-operatively, CPA's motion VEPs became significantly less abnormal. Subsequent to surgery, six of the eight measures recorded from the LE had significantly lower asymmetry indices than the immediate pre-operative index. One week after surgery, the RE was recorded and the asymmetry indices at this time were both greater than or equal to the immediate pre-operative values. However, all eight of the subsequent measures for the RE were significantly lower. A bowtie configuration was present in the F1 data only on the fourth post-operative session. At 23 weeks after surgery, both RE measures were within the normal range, but both LE measures fell outside the normal limits. At that time, CPA had ~ 10a of right hypertropia with no detectable horizontal deviation.

.......

.

~

47 wks

V

-30

,

6~0

,

I

<,:

0 -30

+[

£D

10

20 '

i 40

Age relative to surgery (weeks)

0

,

>"

>r1"

x

>"

o

->

-20

Ol

-30

30

6o

-30

"~

m

~

0

60 3O

B

FIGURE 4(A, B). Caption on facing page.

70

l

I

, 50

~

~]

+o

,

rr >I-,

, 30

,

I

.......~............ ] .....

,

I[

k

V

~

I VERTICAL

..................................... .~ ...............................D I

•

0

-10

~

:

I HORIZONTALI I

>rr

-20

~

=

CPA

0 .............................................. ~ . . . . . : .

30

6o

x t

t",

I.U

_0

z-~o

0

60 3O

A

0

Y

V

49 wks

40

I

60

I

I

80 Age relative to surgery (weeks)

20

I

o

ES

100

T

I

olom

120

I VERTICAL I

I HORIZONTALI

Z ©

Z .<

> Z :=

oo

......

~.~

={

>-

"r

)-

Z m

ii

.

.

.

.

, -20

?' 0

Age

relative

2'0

.

to surgery

4'0

~ ~

~

(weeks)

.....................

I

, 60

I

8'0

I

VERTICAL

]

100

.............. ~ , ) ..................

cl

Iii

-30

z

n-

>-

a

ILl

o -20

x 11

0-

-30

30

60

0

A

...

V 0

3 9 wks

3o °I

D

, 20

I

Age

40

I

:

, 80

I

relative to surgery

60

I

=

ANL

I

, 120 (weeks)

, 100

I

:

:

, 140

I

, 160

I

VERTICAL

180

,4 olo~

I

- ' .....

...j

Z

0

:z

© ,.}

C3 ,...]

,.< >

,4 7~ >

He'zonAL I

FIGURE 4. Longitudinal eye alignment and VEP asymmetry indices in four patients showing a decline in motion VEP asymmetry on multiple measures. Eye alignment is plotted during the period of VEP recordings, with • indicating constant deviations and O indicating intermittent deviations. The horizontal and vertical components of the deviation are plotted separately. Age is relative to the nitial surgery. Each surgery is marked by an inverted triangle near the abscissa. DVD is noted by the " D " symbols below the axis. Any significant changes in spectacle correction are indicated by the ~sterisks. VEP asymmetry indices are plotted for each combination of eye and recording channel, along with the error estimate (___ l SE). The horizontal gray bands mark the limits for the normal asymmetry index. Details of each case are described in the text.

-40

O

0

~ ~

I

=

>-

~

:

rr I-

0

I

:::

51 wks

-

I HORIZONTAL I

•r -

.3O

0

30

........

CT

~-

> ii .'a

60

-3o

0

60

C

3288

ANTHONY M. NORCIA el al.

Patient ES [Fig. 4(b)] had strabismus at birth according to parental report. Esotropia was documented by an ophthalmologist at 8 months of age (30a, alternating, without spectacle correction). At the time of the final pre-operative VEP (45 weeks), as well as on the day prior to surgery, the patient had a basic deviation of 30a right esotropia, with a significant A-pattern (20A difference between up and down gaze). Full-time alternate occlusion had been instituted at 9 months (35 weeks) of age and was maintained until surgery was performed at 11 months (49 weeks) of age. One week after surgery, ES had a residual deviation of 12a right esotropia. By 3 months after surgery, criterion alignment had been achieved. Alignment was maintained using full spectacle correction and asymmetric, part-time occlusion until 1 yr post-surgery, at which time occlusion therapy was discontinued. At the time of the final VEP measurement (2.3 yr after surgery), ES had 1-2 ~ of right esotropia with intermittent periods of motor fusion. Three motion VEP recordings were made prior to surgery, although a LE measurement was not obtained on the third visit. All of these manifested abnormally high asymmetry indices and the bowtie configuration was present for the two sessions in which data from both eyes was obtained. Post-operatively, ES's motion VEPs became significantly less abnormal and the bowtie configuration was no longer present. For the LE, four of the eight eye/channel measures were significantly lower than the pre-operative values, and each was within normal limits. The same trend occurred in the RE--six of the eight measures were significantly lower than the measures from the final pre-operative session. At the time of the final recording 2.3 yr (120 weeks) after surgery, the (dominant) LE-CHI measure was within the normal range, but the other three measures were not, in contrast to the previous visit (at 100 weeks) where three of four measures had been normal. Patient CT [Fig. 4(c)] had strabismus at birth according to parental report. Strabismus was documented by an ophthalmologist at 4 months of age (~30 a left esotropia without spectacle correction). Part:time alternate occlusion was instituted for 2 months, at which time a full-time occlusion regime was initiated and maintained until the time of surgery (12 months). One week after surgery, CT had no measurable strabismus, and remained orthotropic over the subsequent year of measurements. No occlusion therapy was used after surgery. CT had only a small amount of spherical hyperopia and was never prescribed spectacles. Two motion VEP recordings were made prior to surgery. The LE had an abnormally high asymmetry index at the time of the first recording. However, by the time of the second recording (on the day prior to surgery), three of the four asymmetry indices were within normal limits, and a bowtie configuration that had been present in the FI data from the first session was not present (cf. Jampolsky et al., 1994). The RE exhibited a significant reduction in motion asymmetry relative to the immediate pre-operative measurements, and all values were within the normal range at the last post-operative

recording. There was a transient increase in the asymmetry index for the LE-CH2 measure 8 weeks after surgery. Patient ANL [Fig. 4(d)] had strabismus by 3 months of age according to parental report. Strabismus was documented by an ophthalmologist at 6 months of age (~50 a alternating esotropia, without spectacle correction). Full-time alternate occlusion was instituted and maintained until surgery was performed at 9 months (39 weeks). ANL's deviation (without spectacle correction) 1 day prior to surgery was approx. 60Aof alternating esotropia with significant abduction restrictions evident in both eyes. One week after surgery, ANL had a residual deviation of 106 of left esotropia. Criterion alignment was achieved by 9 months post-operatively (full spectacle correction and part-time alternate occlusion were used). However, during this period the deviation never exceeded 10a (recall that criterion alignment was < 10a). Occlusion therapy was discontinued at the time orthotropia was achieved. A motion VEP recording was made on the day prior to surgery. At this time four trials were obtained from the RE, but only two trials were obtained from the LE. The RE motion asymmetry indices were significantly abnormal in this recording. By the first post-operative measurement both asymmetry indices had decreased significantly. For the contralateral derivation (CH~), the asymmetry remained significantly below the pre-operative measure for all subsequent measures over the next 2.5yr. For the ipsilateral derivation, the final two measures, though both within the normal range, were not statistically different from the preoperative value. Although the LE preoperative data were insufficient to apply the T~rc test [hence, no preoperative LE data are shown in Fig. 4(d)], the signal-to-noise ratios indicated significant responses that were 180deg out of phase with the RE data. In the post-operative recordings, a bowtie was only present in the FI data of the third post-operative VEP recording. At the time of the last post-operative recording, the RE values were both in the normal range, but the LE values were just above it. Eye-dominance, when present, had always been in favor of the RE. Figure 5 shows data from the only patient (BF) who maintained highly abnormal motion asymmetry indices in spite of accurate, stable surgical alignment at an early age (12 months). Patient BF presented at age 3 months with a right esotropia (with ability to alternate) that varied between 20~ and occasional orthotropia in primary gaze. The esotropia became constant (40a LET) sometime between BF's 6- and 8-month visits, at which time asymmetric full-time alternate occlusion was initiated. One day prior to surgery, BF's deviation was approx. 55A of right esotropia. Bifoveal fusion was achieved within 5 weeks after surgery. This excellent alignment was stable for the 2.4 yr during which postoperative motion VEP recordings were made. By that time, BF had developed a small amount of DVD (,~ 5A). The preoperative VEP recording was made on the day before surgery. Three of the four motion asymmetry

3289

STRABISMUS SURGERY AFFECTS MOTION PROCESSING

BF 60

I HORIZONTAL I

30 ~.

0

A w

v

~.

=

=

¢

_-

=

-_

T

z -30 O

50 wks

> 60 LU D 3O

I

VERTICAL I

..........................................................................................................................................................

D -30

l

~'

i

,

I

l

I

,

D I

I LE CH 1 I X LU a Z

V pW

IRE CH1]

x a z

T

rr

0

I

-40

-20

'~

0

I

I

I

I

I

I

I

20

40

60

80

100

120

140

Age relative to surgery (weeks)

160 01042

FIGURE 5. Longitudinal eye alignment and VEP asymmetry indices from the only patient showing who showed a very small decline in VEP asymmetry in spite of nearly optimal alignment. All symbols as in Fig. 4.

indices for this session were highly abnormal, including the presence o f a marginally significant bowtie relationship between the R E and L E responses. F o r the RE, the postoperative a s y m m e t r y indices showed improvements on six o f the eight measures. However, all o f the a s y m m e t r y indices f r o m the last three post-operative VR 35/23-24--E

sessions remained abnormally high in spite o f a nearly optimal surgical outcome.

Retrospective studies Long-term stability. Figure 6 shows m o t i o n V E P responses in polar f o r m recorded f r o m four infantile

3290

ANTHONY M. NORCIA et al.

esotropes who had surgery between 11 and 14 months of age and who had achieved criterion alignment. Their VEPs were recorded in response to 6 H z , l c/deg gratings, 3.8-8.5 yr after surgery. None of these patients showed evidence of a nasalward-temporalward response asymmetry and all of their asymmetry indices were within normal limits. F~ components (upper panels) were small relative to the F2 components in each patient and the F2 responses occurred at similar phase lags with respect to the stimulus (lower panels). Patients N M and VM [Fig. 6(a, b)] were identical twins who were esotropic at birth by parental report. Family photographs confirmed that large angle esotropia was present in each child before 6 months of age. Each child had surgery at 13 months of age with good outcome. At the time of these recordings, VM exhibited m o t o r fusion at both distance and near, but had 106 of DVD. Patient N M had a small left esotropia of 1-26 with a left hypertropia of 56, and 106 of bilateral D V D at distance. At near, N M had 56 of left esotropia. Their motion asymmetry indices were in the normal range for each eye/channel measure (NM, 0.13-0.35; VM, 0.09-0.4). N M had demonstrated 80-100 sec arc stereo acuity on the Titmus test. However, neither child could see depth in random-dot stereograms. Patient EM [Fig. 6(c)] was esotropic at birth by parental report. Constant esotropia was documented by an ophthalmologist at 6 months of age. At 9 months, EM presented with 80-1006 of esotropia, a substantial V-pattern, and significant restrictions of abduction in both eyes. Full-time alternate occlusion was initiated and maintained until the time of surgery. Surgery was performed for 806 of esotropia at 11 months of age. After surgery, EM held criterion alignment for 3.7 yr, maintaining bifoveal fusion for the last 3 yr. Except for two

a

b

visits (4.2 and 7.2 yr after surgery) E M ' s alignment never exceeded 86 of esotropia throughout the entire 8.3-yr post-operative period. Five and one-half years after surgery, after 3 full years of bifoveal fusion, EM had a transient re-occurrence of a small left esotropia (1-2~), but nevertheless demonstrated a significant degree of stereopsis (Titmus stereoacuity of 100 sec arc). At 6.5 yr after surgery, she had regained fusion and had normal visual acuity in each eye. A VEP measurement at that time produced normal motion asymmetry indices [Fig. 6(c), RE, 0.143; LE, 0.135]. Patient EB [Fig. 6(d)] had a large angle (506) constant right esotropia, with significant abduction restrictions, documented by 5 months of age. Full-time alternate occlusion was initiated at the time of the first visit. After 2 months, the regime was decreased to part-time occlusion until surgery at 14 months of age. At the time of surgery, the deviation was 25 ~ of alternating esotropia. One week after surgery, EB had a residual right esotropia of 126. Two months after surgery criterion alignment was achieved, and was maintained over the next 16 months. Ten months later (2.2 yr after surgery) a 106 right exotropia had developed. Management with spectacles and part-time occlusion brought the alignment back to criterion over the next 2 visits (by 3.8 yr after surgery), although a small amount of D V D (56) was then noted. At the time of the motion VEP measurement, EB had 56 of right exotropia, as well as a small amount (5 A) of right hypertropia and D V D tested at distance. At near, a small (56) left esotropia was noted. The motion VEP measurements made at this time had asymmetry indices within normal limits (0.08 0.18). Earl), es late alignment. Finally, the motion VEPs of normal adults (n = 11) were compared to two categories of patients: those who were not aligned during the first

c

d

FIGURE 6. Retrospective motion VEP recordings from four patients who were accurately aligned before age 2. The upper diagram in each panel plots the amplitude and phase of the Ft component and the lower diagram plots the F2 component. Individual vectors plot the data from each 10-sec trial (RE, solid lines; LE, dashed lines). • and © plot the vector average of the individual trial responses for RE and LE respectively. Each patient showed a normal motion VEP in response to 1c/deg gratings undergoing oscillatory motion at 6 Hz. Details of each case are described in the text.

STRABISMUS SURGERY AFFECTS MOTION PROCESSING

a.

3291

b.

.8 X r-

.6 J E

E .4 (,9

<

.2 I

I

I

I

I

Normal

Early

Late

Contra

Ipsl

0

FIGURE 7. Group results from the retrospective series. (a) The group mean asymmetry indices for normal observers, and for early- and late-alignment patient groups. Both patient groups have higher than normal asymmetry indices, but the magnitude of the abnormality is significantly less in the early alignment group. (b) Illustrates that early surgery patients had lower asymmetry indices for the contralateral derivations than for the ipsilateral derivations.

2 yr of life (late alignment group, n = 11), vs those who had had a successful surgery before age 18 months of age (early alignment group, n = 9). The comparison of asymmetry indices was made for responses to 10Hz, 3 c/deg motion since pilot testing indicated that the motion VEP asymmetry is larger and more persistent at higher spatial and temporal frequencies. Recording at 10Hz, 3 c/deg was thus deemed more likely to show residual motion asymmetries that might remain after treatment. An initial ANOVA examined the effects of patient category (Cat), eye (Eye) and recording channel (Chan). Twenty-eight of the 31 observers had complete data sets for all eye/channel measures. The mean asymmetry indices for each group are plotted in Fig. 7(a). Both early and late alignment groups had significantly abnormal asymmetry indices; but the magnitude of the abnormality was significantly less in the early alignment group (main effect of patient category: F = 75.7; P = 0.0001). There was also a significant Eye x Chan x Cat interaction (F = 6.2; P = 0.007). The interaction effect was clearly dominated by the early treatment group. ANOVAs were performed for each group after recoding the Chan variable into ipsi- and contralateral derivations and the Eye variable into dominant and non-dominant eye, as was done for the prospective group. These ANOVAs indicated that the normal and late treatment groups showed no main effects or interaction effects involving eye or pathway. In the early alignment group, the two contralateral derivations had significantly lower asymmetry indices than the ipsilateral derivations, consistent with surgery having differential effects on the crossed vs uncrossed pathways [P = 0.022; effect shown as Fig. 7(b)]. No other main effects or interactions were significant. Finally, the level of asymmetry did not differ between those patients in the late treatment group with < 10A of residual deviation compared to those who had more than 10~ (n = 6 and 5 respectively). This suggests that, although accurate alignment may be necessary for

normalization of motion asymmetries, it may not be sufficient unless it is achieved within some finite critical period of plasticity. DISCUSSION Motion processing mechanisms accessible in the human VEP are significantly abnormal prior to treatment for infantile esotropia. Directional biases that are physiologic in young infants are present in infantile esotropia patients beyond the age at which they would have disappeared in infants with normal ocular alignment (Norcia et al., 1990a, 1991; Jampolsky et al., 1994). In the prospective study, the magnitude of the directional bias decreased significantly after surgery in the group data and to varying degrees in individual patients. Moreover, the decrease in motion asymmetry was not a transient response to surgical alignment, but was maintained for an average of 95 weeks (1.8 yr) in the prospective group and for several years in the patients shown in Fig. 6. The cortical mechanisms generating the directionally biased VEP responses in infantile esotropia thus retain a significant degree of plasticity. Significantly lower levels of motion asymmetry were seen in patients in the retrospective group who had been accurately aligned before 2 yr of age compared with those who had been aligned later. As a group, patients who achieved early alignment had lower asymmetry indices at 10 Hz, 3 c/deg than those who did not. The difference between the early- and late-treatment groups is consistent with the existence of a critical period in the development of binocular, direction selective mechanisms, during which accurate alignment may be effective in promoting recovery of a more normal response. This notion is supported by the observation that within the late-treatment retrospective group, the subgroup of patients who had experienced accurate alignment ( < 10A of residual deviation) after age 2 fared no better than the subgroup with poorer alignment. Taken together, the

3292

ANTHONY M. NORCIA et al.

prospective and retrospective results demonstrate that motion-processing mechanisms are plastic within a developmental critical period and that they are dependent on the development of normal binocular interactions.

Individual differences The group results notwithstanding, it is important to note that individual differences exist, both between patients and within patients (between eye/channel conditions). For example, considerable differences in the absolute magnitude of the VEP motion asymmetry exist prior to surgery, both across patients and between eyes of individual patients. This heterogeneity in the VEP responses prior to treatment suggests that infantile esotropia may not be a unitary entity, but rather a multi-factorial disorder with variable sensory expressions in addition to the well known variability in the oculomotility of these patients (cf. Greenwald, 1992). The sensory response to surgery is also not uniform. In the prospective study, the reductions in motion asymmetry after surgery differed in magnitude across individual patients and within eyes of individual patients, but the effect was significant in at least one measure from each patient at the time of the final recording (Figs 4 and 5). The variability in the response to treatment may be due to many factors associated with the outcome of strabismus surgery--individual differences in the patients' underlying pathophysiology, variability in the timing and accuracy of the alignment procedure and complications due to refractive error and/or amblyopia. Individual differences in the response to pre-operative alternate occlusion therapy were also noted (Jampolsky et al., 1994) and it is possible that the varying response to preoperative management could contribute variability to the postoperative results. Our sample is much too small to sort out the effects of these different factors.

Eye dominance and hemiretinal effects There was a trend toward lower motion VEP asymmetries in the dominant eyes of infants in the prospective study. However, an eye dominance effect was not found in either of the older retrospective groups. Westall and Shute (1992) noted that when MOKN asymmetries were present, they tended to occur most frequently in the non-dominant eye. One might suspect that the greater asymmetry in the non-dominant eye might be associated with amblyopia, although Westall and Shute (1992) found no such relationship in their MOKN data. Significant eye/channel interactions were found in both the prospective surgery group and in the earlyalignment retrospective group, but not in the normal adults or the late-alignment group. The early-alignment retrospective group is similar to the prospective group in that their ages of surgery and realignment were comparable. While these two groups both show topographic differences in the magnitude of motion asymmetry, the pattern differs in the two groups. In the prospective group it was the ipsilateral recordings that showed the lowest asymmetry indices after surgery while in the

retrospective group it was the contralateral recordings that showed the lowest asymmetry indices. As noted previously, there is no independent evidence that the recording derivations used in the present study isolate responses from the hemispheres they overlie. Furthermore, Barrett, Blumhardt, Halliday, Halliday and Kriss (1976) have reported that the pattern-reversal response of adults may show paradoxical localization. It is not known whether young children show the same pattern of lateralization of the response. It is also possible that the apparent localization of the generator for the 6 Hz, 1 c/deg stimulus used in the prospective group differs from that for the 10 Hz, 3 c/deg stimulus used in the retrospective group. Finally, Ciancia et al. (1988) found significant hemispheric asymmetries in the monocular pattern reversal response of infantile esotropes. The localization was consistent with a crossed pathway dominance in some patients and an uncrossed dominance in other patients. Further study of infantile esotropia patients with hemifield stimuli would be needed to determine whether the differences can legitimately be attributed to hemiretinal differences in the development of motion processing mechanisms.

Implications for theories of the etiology of infantile esotropia One could argue that the abnormal motion VEPs observed in esotropic infants are due to a fundamental incapability of these patients to develop a normal motion response, perhaps due to a genetic predisposition. This view is reminiscent of Claude Worth's theory of strabismus etiology which suggested that strabismus was due to the absence of a "central fusion faculty" (Worth, 1905). A similar argument could be made for binocular direction-selective mechanisms. Our observation that infantile esotropia patients show a decline in the degree of asymmetry of their motion VEP after surgery argues against a fundamental incapability of the motion pathway in the majority of patients. This recovery appears to be long-lasting in most patients with stable, accurate alignment and can extend into visual maturity. It should also be noted that disruption of binocular input in otherwise normal monkey neonates is sufficient to cause persistent motion VEP asymmetries (Brown, Wilson, Tigges & Boothe, 1992), and thus a genetic predisposition is not necessary for their production. One could also argue that the substrate for a normal motion VEP is present in infantile esotropia patients at birth, hut that maturation of the motion response is delayed and is independent of the state of eye alignment or binocular interaction. If this were true, one would not expect the age of surgery to make any difference in the degree of motion asymmetry. Yet the retrospective data showed that the magnitude of the motion asymmetry was significantly greater in the late-treatment group, independent of the ultimate quality of postoperative eye alignment. These results appear to rule out models that posit independence between eye alignment and the degree of cortical motion asymmetry. They also indicate

STRABISMUS SURGERY AFFECTS MOTION PROCESSING that infantile esotropia patients do not spontaneously lose their motion abnormalities over time. A third position would have that motion processing asymmetries are the result of secondary changes in visual cortex following the disruption of correlated binocular input by strabismus. There are at least two factors that could operate here--one is the interruption of binocularity caused by misregistration of corresponding receptive fields and another is the development of abnormal binocular interactions. In natural and experimental strabismus these two factors are confounded. In monkeys, a prolonged interruption of correlated binocular input by alternating monocular exposure prevents the development of symmetrical motion VEPs (Brown et al., 1992) and MOKN (Brown et al., 1992; Rastelli, Matsumoto, Tychsen & Boothe, 1993) and thus interrupted binocular vision, by itself, is sufficient to block the development of a normal motion response. Jampolsky et al. (1994) have argued that both interruption and abnormal binocular interaction secondary to infantile esotropia play a role in delaying the development of cortical motion mechanisms. They found that the magnitude of the VEP motion asymmetry prior to surgery was lower in patients who had been alternately occluded compared to infants of similar age who had not been occluded. Both groups had interrupted binocular vision--one group because of a large angle, constant strabismus and the other because of alternate occlusion--but the competitive interactions that were constantly present in the non-occluded group were reduced or absent in the alternate occlusion group. Finally, one could also argue that an epigenetic delay of the motion pathway could help to precipitate strabismus which, once in place, might block further maturation. Our data cannot discriminate a purely secondary mechanism--strabismus causing the failure of motion mechanisms to develop--from a model in which developmental delays in motion processing mechanisms play a precipitating role in the development of the strabismus, which then prevents further maturation (cf. Tychsen, 1993). Our data would suggest that if the latter model is applicable, the delayed mechanisms are plastic and subject to later manipulations of eye-alignment and binocular interaction. One would have to study the development of motion processing mechanisms prior to the onset of strabismus in order to determine if they differ from those of normal neonates.

Possible mechanisms o f the motion V E P asymmetry Latent nystagmus. Latent nystagmus (LN) is frequently associated with a diagnosis of infantile esotropia. The presence of LN is thus likely to correlate with the presence of a motion VEP asymmetry, at least in infantile esotropia. The direction of causality between LN and VEP motion asymmetries or whether such a causal relationship exists, is not known. In LN, the eye drifts nasalward and saccades temporalward. Such eye motion would tend to impose an asymmetry in both the effective displacement amplitude and the effective velocity of the grating. Such stimulus-based asymmetries

3293

would switch to the opposite direction between monocular viewing conditions, possibly mimicking a nasalward-temporalward response bias. Norcia et al. (1991) were unable to induce significant VEP asymmetries in normal observers who were executing tracking movements designed to simulate typical LN waveforms (e.g. 1 deg/sec sawtooth motion) and thus a simple imagemotion asymmetry is not likely to be the sole source of the VEP asymmetry. It is also possible that LN, if manifest under binocular viewing conditions, produces some form of long-term adaptation of the cortical response. Simulated nystagmus in normals would not mimic such an effect. For this to happen, the adaptation would have to occur only in the viewing eye, since with alternate fixation, each eye would become adapted to each direction of motion because of conjugacy. Motion adaptation is not blocked by rivalry suppression (Lehmkuhle & Fox, 1975) and thus adaptation under alternate fixation and suppression would result in equal adaptation to both directions if strabismic suppression acts after the site of motion adaptation, as does rivalry suppression. If the LN model of the VEP asymmetry is correct, one would have to argue that normal infants either have manifest LN, or that they have stimulus-evoked nystagmus during testing. While the former is highly unlikely, the latter is not inconceivable, given the reports that infants show nystagmus when monocularly viewing a reversing grating (Teller et al., 1994). However, stimulusevoked nystagmus was never seen during our infant recordings and our targets are not well suited to eliciting nystagmus. Fixation targets were used at all times during the recording and the jittering grating is not directionally ambiguous as is a reversing grating. It is therefore unlikely that motion VEP asymmetries are an epiphenomenon associated with nystagmus. Visual cortex. VEP directional biases, such as those that are present in normal infants and in patients with disrupted binocularity, may be based on a population of direction selective cells that have failed to attain normal binocularity. One can infer that the mechanisms tapped in the present study are normally binocular simply because the abnormalities in the motion VEP are correlated with abnormalities in the development of binocular vision. Chandna, Norcia and Peterzell (1993) provided direct evidence that the oscillatory-motion VEP is tapping binocular, direction selective mechanisms by showing that a direction specific motion aftereffect could be seen in the oscillatory-motion VEP after adaptation to unidirectional drift, and that this aftereffect transferred interocularly. Neurons that are both binocular and direction selective are common as early as area V1 in monkey (Poggio & Talbot, 1981). Cortical areas V2 (Burkhalter & Van Essen, 1986; Poggio, Gonzalez & Krause, 1988), V3 (Fellemann & Van Essen, 1987; Poggio et al., 1988), MT (Maunsell & Van Essen, 1983a, b) and MST (Roy, Komatsu & Wurtz, 1992) also contain populations of cells that are both directionally selective and sensitive to binocular disparity. It is not yet possible to localize the VEP with confidence to a

3294

ANTHONY M. NORCIA et al.

particular cortical area, but our derivations would favor striate or first tier extra-striate areas. Relationship of motion following asymmetries

VEP, perceptual and ocular

A correlation between perceptual, VEP and oculomotor biases is certainly present in broad terms, but the precise details of their relationship are at present unclear. At the simplest level, it is not yet clear whether the motion VEP asymmetry is due to stronger responses for nasalward motion or for temporalward motion. VEP methods other than the one used in the present study will be needed to determine the absolute direction of the cortical asymmetry. If one puts aside for a moment the directional ambiguity of the steady-state VEP, it is clear that perceptual, VEP and O K N biases in strabismic patients are each found mainly in patients with early onset strabismus. Cortical motion biases at any of a number of levels could have a direct effect on either O K N or pursuit eye movements and perception. Extrastriate areas in the superior temporal sulcus such as MT and MST are known to be involved in the generation of pursuit eye movements (Lisberger, Morris & Tychsen, 1987) and they also project to the nucleus of the optic tract which is involved in the production of O K N (Hoffmann, Distler & Erickson, 1991). Area V1 projects directly both to extrastriate areas, such as V2 and MT and also to the N O T (Hoffmann et al., 1991). V1 is likely to dominate our recordings because it is so much larger than MT or MST (Van Essen & Anderson, 1990) and because it directly underlies our recording electrodes. A recent examination of the relationship between perceptual and M O K N biases (Hartmann, Succop, Buck, Weiss & Teller, 1993) did not find a good correlation between the two in a reversing grating paradigm-O K N asymmetries were always larger than perceptual asymmetries and could exist in the absence of a perceptual asymmetry. Roberts and Westall (1990) found that patients with M O K N asymmetry did not have perceived velocity biases. The relative lack of a perceptual bias in Hartmann et al.'s task and in Roberts and Westall's study and its presence in Tychsen and Lisberger's (1986) perceived velocity estimates is puzzling, but may be due to differences between the stimuli used in the different studies [extended, low spatial frequency gratings in Hartmann et al. (1993) and Roberts and Westall (1990) and small line-targets in Tychsen and Lisberger (1986)]. Comparisons between oculo-motor and perceptual measures of bias and the motion VEP bias have yet to be done and could be quite informative. Relationship to other studies of binocularity in infantile esotropia The change in the motion asymmetry index following surgery is consistent with plastic changes in binocular, direction selective mechanisms that are contingent on the restoration of more normal binocular interaction during a critical period of development. Previous work has indicated that other forms of binocularity are enhanced after early restoration of binocular alignment in infantile

esotropia. Interocular transfer of the tilt-aftereffect is greater with early as opposed to late surgery in infantile esotropia (Banks, Aslin & Letson, 1975) and stereopsis and binocular fusion are more common after early rather than late correction (Uemura, 1973; Taylor, 1973; Ing, 1981, 1983; Bateman, Parks & Wheeler, 1983), although it has been reported that completely normal stereopsis is rarely attained in infantile esotropia (Ing, 1983; Parks, 1984). It is not clear at present what relationship exists between the binocular mechanisms being tapped by the motion VEP and other binocular mechanisms such as stereopsis. As noted above, many cells in visual cortex are jointly tuned for direction and binocular disparity. However, many of these cells prefer vertical disparity and are thus not involved in stereopsis. There are also other cells that are disparity selective, but not direction selective (Poggio et al., 1988). It would not be surprising therefore, if there were multiple sub-populations of binocular cells, each with its own critical period, stimulus requirements and efferent targets in the oculomotor system. We have not yet made extensive measurements over the whole spatio-temporal response surface of direction selective mechanisms. Recovery of motion symmetry was only partial at 10 Hz, 3 c/deg but appeared complete at 6 Hz, 1 c/deg in patients with stable alignment before 2 yr of age who were studied long after their surgery. It would be of interest to compare stereopsis and motion VEP asymmetry indices over the entire spatio-temporal surface to determine whether or not they share the same tolerances for age of alignment, residual deviations and for quality of binocular interaction. A definitive answer to the question of what constitutes the necessary and sufficient conditions for normal maturation of motion pathways is not yet possible. Answering this question awaits the determination of the full range of spatio-temporal sensitivity of the relevant motion mechanisms, as well as an evaluation of the effects of treatment variables such as pre-operative management technique, timing and method of alignment, and postoperative management. Since the status of motion processing mechanisms has not generally been considered in the management of esotropia it is premature to state the ultimate effects of re-alignment procedures on motion processing mechanisms. Nonetheless, it can be said that their status appears to be modifiable by treatment. REFERENCES

Atkinson J. (1979). Development of optokinetic nystagmus in the human infant and monkey infant. In Freeman, R. D. (Ed.), Developmental neurobiology of vision (pp. 277-287). New York: Plenum. Banks, M. S., Aslin, R. N. & Letson, R. D. (1975). Sensitive period for the development of human binocular vision. Science, 190, 675-677. Barrett, G., Blumhardt, L., Halliday, A. M., Halliday, E. & Kriss, A. (1976). A paradox in the lateralization of the ~isual evoked response. Nature, 261, 253 255. Bateman, J. B., Parks, M. M. & Wheeler, N. (1983). Discriminant analysis of congenital esotropia surgery. Ophthalmology, 90, 1146-1153.

STRABISMUS SURGERY AFFECTS MOTION PROCESSING Brown, R. J., Wilson, J. R., Tigges, M. & Boothe, R, G. (1992). Motion perception in monkeys: correlation between physiological and behavioral asymmetry following alternating monocular occlusion. Presented at the International Primatological Society Meeting, Strasbourg, France, 16 21 August. Burkhalter, A. & Van Essen, D. C. (1986). Processing of color, form and disparity information in areas VP and V2 of ventral extrastriate cortex in the macaque monkey. Journal of Neuroscience, 6, 23275351. Chandna, A., Norcia, A. M. & Peterzell, D. H. (1993). VEP correlates of direction-specific binocular mechanisms. Investigative Ophthalmology and Visual Svience (Suppl.), 34, 1054. Ciancia, A. O. (1962). La Esotropia con limitacion bilateral de la abduccion en el lacante. Archivos de Oftalmologia de Buenos Aires, 36, 207 211. Ciancia, A. O., Borrone, R., Schuarzberg, D. E. & Garcia, H. A. (1988). Asymmetrical Visual Evoked Potentials (VEP) in congenital esotropia with bilateral limitation of abduction. Binocular Vision, 3, 15--22.

Costenbader, F. D. (1950). The management of convergent strabismus. In Allen, J. H. (Ed.), Strabismus ophthalmic symposium. St Louis, Mo.: Mosby. Day, S. H. & Norcia, A. M. (1986). Photographic detection of amblyogenic factors. Ophthalmology, 93, 25-28. Demer, J. L. & v o n Noorden, G. K. (1988). Optokinetic asymmetry in esotropia. Journal of Pediatric and Ophthalmological Strabismus, 25, 286 292. Fellemen, D. F. & Van Essen, D. C. (1987). Receptive field properties of neurons in area V3 of macaque extrastriate cortex. Journal of Neurophysiology, 57, 889-920. Greenwald, M. J. (1992). Surgical management of essential esotropia. Ophthalmological Clinics North America, 5, 9 22. Hamer, R. D., Norcia, A. M., Orel-Bixler, D. & Hoyt, C. S. (19933. Motion VEPs in late-onset esotropia. Clinical Vision Sciences, 8, 55-62. Harris, F. J. (1978). On the use of windows for harmonic analysis with the discrete Fourier transform. Proceedings IEEE, 66, 51 83. Hartmann, E. E., Succop, A., Buck, S. L., Weiss, A. H. & Teller, D. Y. (1993). Quantification of monocular optokinetic nystagrnus asymmetries and motion perception with motion-nulling techniques. Journal of the Optical Society of America A, I0, 1835 1840. Hoffmann, K.-P., Distler, C. & Erickson, R. (1991). Functional projections from striate cortex and superior temporal sulcus to the nucleus of the optic tract (NOT) and dorsal terminal nucleus of the accessory optic tract (DTN) of macaque monkey. Journal of Comparative Neurology, 313, 707 724. Hubel, D. H. & Wiesel, T. N. (1970). The period of susceptibility to the physiological effects of unilateral eye closure in kittens. Journal of Physiology, London, 206, 419-436. Ing, M. R. (1981). Early surgical alignment for congenital esotropia. Transactions of the American Ophthalmology Society, 79, 625 663. Ing, M. R. (1983). Early surgical alignment for congenital esotropia. Ophthalmology, 90, 132 135. Jampolsky, A., Norcia, A. M. & Hamer, R. D. (1994). Preoperative alternate occlusion decreases motion processing abnormalities in infantile esotropia. Journal of Pediatric and Ophthalmological Strabismus, 31, 6 17. Kommerrell, G. (1988). Ocular phenomenon in infantile strabismus. In Lennerstrand, G., von Noorden G. K. & Campos, E. C. (Eds), Strabismus and amblyopia (pp. 99-109). Werner-Gren International Symposium Series. London: Macmillan. Kommerrell, G., Ulrich, D. & Bach, M. (1995). Asymmetry of motion VEP in infantile strabismus and in central vestibular nystagmus. Documenta Ophthalmologica. In press. Lehmkuhle, S. W. & Fox, R. (19753. Effect of binocular rivalry suppression on the motion aftereffect. Vision Research, 15, 855 859. Lisberger, S. G., Morris, E.J. & Tychsen, L. (1987). Visual motion processing and sensory-motor integration for smooth pursuit eye movements. Annual Review of Neuroscience, 10, 97 129. Maunsell, J. H. R. & Van Essen, D. C. (1983a). Functional properties of neurons in middle temporal visual area of the macaque monkey.

3295

I. Selectivity for stimulus direction, speed and orientation. Journal of Neurophysiology, 49, 1127 1147. Maunsell, J. H. R. & Van Essen, D. C. (1983b). Functional properties of neurons in middle temporal visual area of the macaque monkey. II. Binocular interactions and sensitivity to disparity. Journal of Neurophysiology, 49, 1148 1166. Mein, J. (1983). The asymmetric optokinetic response. British Orthoptic Journal, 40, 1-4. Naegele, J. R. & Held, R. (1982). The postnatal development of monocular optokinetic nystagmus in infants. Vision Research, 22, 341 346. von Noorden, G. K. (1988). A reassessment of infantile esotropia. XLIV Edward Jackson Memorial Lecture. American Journal of Ophthalmology, 105, 1-10. Norcia, A. M. & Tyler, C. W. (19853. Spatial frequency sweep VEP: Visual acuity in the first year of life. Vision Research, 25, 1399 1408. Norcia, A. M., Hamer, R. D. & Orel-Bixler, D. (1990a). Temporal tuning of the motion VEP in infants. Investigative Ophthalmology and Visual Science (Suppl.), 31, 10. Norcia, A. M., Tyler, C. W. & Clarke, M. (1985). Digital filtering and robust regression techniques for estimating sensory thresholds from the evoked potential. IEEE Engineering in Medicine and Biology, 4, 26 32. Norcia, A. M., Tyler, C. W. & Hamer, R. D. (1990b). Development of contrast sensitivity in the human infant. Vision Research, 30, 1475-1486. Norcia, A. M., Garcia, H., Humphry, R., Holmes, A., Hamer, R. D. & Orel-Bixler, D. (1991). Anomalous motion VEPs in infants and in infantile esotropia. Investigative Ophthalmology and Visual Science, 32, 436 439. Orel-Bixler, D. & Norcia, A. M. (1987) Differential growth of acuity for steady-state pattern reversal and transient pattern onset-offset VEPs. Clinical Vision Sciences, 2, 1-9. Parks, M. M. (1984). Congenital esotropia with a bifixation result: Report of a case. Documenta Ophthalmologica, 58, 109--114. Poggio, G. F. & Talbot, W. H. (19813. Mechanisms of static and dynamic stereopsis in foveal cortex of the rhesus monkey. Journal of Physiology, London, 315, 469--492. Poggio, G. F., Gonzalez, F. & Krause, F. (1988). Stereoscopic mechanisms in monkey visual cortex: Binocular correlation and disparity selectivity. Journal of Neuroscience, 8, 4531-4550. Rastelli, A., Matsumoto, B., Tychsen, L. & Boothe R. G. (1993). Long-term instability of vergence eye movements in monkeys who lacked normal binocular experience in infancy. Investigative Ophthalmology and Visual Science (Suppl.), 34, 1188. Reed, M. J., Steinbach, M. J., Anstis, S. M., Gallie, B., Smith, D. & Kraft, S. (1991). The development of optokinetic nystagmus in strabismic and monocularly enucleated subjects. Behavioral Brain Research, 46, 31-42. Roberts, N. & Westall, C. (1990). OKN asymmetries in amblyopia-their effect on velocity perception. Clinical Vision Sciences, 5, 383 389. Roy, J.-P., Komatsu, H. & Wurtz, R. H. (1992). Disparity sensitive neurons in monkey extrastriate area MST. Journal of Neuroscience, 17, 2478-2492. Schor, C. M. & Levi, D. M. (1980). Disturbances of small-field horizontal and vertical optokinetic nystagmus in amblyopia. Investigative Ophthalmology and Visual Science, 19, 668-683. Sherman, S. M. & Spear, P. D. (1982). Organization of visual pathways in normal and visually deprived cats. Physiological Review, 62, 738 855. Sokol, S., Pelli, E., Moskowitz, A. & Reese, D. (1991). Pursuit eye movements in late onset esotropia. Journal of Pediatric and Ophthalmological Strabismus, 28, 82-86. Taylor, D. M. (1973). Congenital esotropia: Diagnosis and management (pp. 19, 22, 83, 84, 87). New York: Intercontinental Medical Book Corp. Teller, D. Y., Succop, A. & Mar, C. (1994). Infant eyemovement asymetries: Stationary gratings elicit temporal-to-nasal optokinetic nystagmus in 2-month-old infants. Vision Research, 34, 1859 1864.

3296

ANTHONY M. NORCIA et al.

Tychsen, L. (1993). Motion sensitivity and the origins of infantile strabismus. In Simons, K. (Ed.), Early visual development, normal and abnormal (pp. 364-390). New York: Oxford University Press. Tychsen, L. & Lisberger, S. G. (1986). Maldevelopment of visual motion processing in humans who had strabismus with onset in infancy. Journal of Neuroscience, 6, 2495-2508. Tychsen, L. R., Hurtig, R. & Scott, W. E. (1985). Pursuit is impaired but the vestibulo-ocular reflex is normal in infantile esotropia. Archives of Ophthalmology, 103, 536-539. Uemura, Y. (1973). Surgical correction of infantile esotropia. Japenese Journal of Ophthalmology, 17, 50 59. Van Essen, D & Anderson, C. H. (1990). Information processing strategies and pathways in the primate retina and visual cortex. In Zornetzer, S. F., Davis, J. L. & Lau, C. (Eds), An introduction to

neural and electronic networks (pp. 43-72). San Diego, Calif.: Academic Press. Victor, J. D. & Mast, J. (1991). A new statistic for steady state evoked potentials. Electroencephalogy and Clinical Neurophysiology, 7g, 378-388. Westall, C. A. & Shute, R. H. (1992). OKN asymmetries in orthoptic patients: Contributing factors and effects of treatment. Behavioral Brain Research, 49, 7744. Worth, C. (1905). Squint: Its causes, pathology and treatment (4th edn). London: John Bale, Sons and Danielsson.

Acknowledgements--Research conducted at the Smith-Kettlewell Eye Research Institute, supported by NIH grants EY06579 and EY06883.