eyelid position. Treatment of amblyopia is only effective in main- Albert W Biglan, MD

RESULTS FOLLOWING TREATMENT OF THIRD CRANIAL NERVE PALSY IN CHILDREN* BY Linda A. Schumacher-Feero, MD (BY INVITATION), K W Yoo, MD, (BY INVITATION), Fernando Mendiola Solari, MD (BY INVITATION) AND Albert W Biglan, MD ABSTRACT Purpose: To investigate the etiology, sensory, motor, and cosmetic results of treatment for oculomotor (CNIII) palsy in children. Methods: We conducted a retrospective review of the clinical records of children with a diagnosis of CNIII palsy who were followed up in our practice between 1981 and 1996. Results: During the 15-year period, 49 children with 53 affected eyes were followed for a mean of5.5 years. CNIII palsywas congenital in one third of cases and secondary to postnatal trauma in another third. Thirty-three of the eyes were affected before visual maturation (age 8 years) and 27 eyes developed amblyopia. None of the 6 eyes with amblyopia in which visual acuity could be quantitated had measurable improvement of Snellen acuity after treatment. Overall, visual acuity was between 6/5 and 6/12 at the last follow-up visit in 56% of affected eyes. Ocular alignment was greatly improved after recessresect procedures on the horizontal rectus muscles, but binocular function was difficult to preserve or restore. Blepharoptosis improved after levator palpebrae muscle resection or eyelid suspension procedures. Conclusions: CNIII palsy may undergo partial resolution in children, but surgical treatment is frequently necessary. Although surgery can result in cosmetically acceptable alignment of the eyes, it rarely results in restoration or achievement of binocular function. Multiple procedures are often necessary to maintain good ocular alignment. Several surgical procedures may be needed to correct related blepharoptosis and maintain an acceptable eyelid position. Treatment of amblyopia is only effective in main- 'From the Department of Ophthalmology, University of Pittsburgh School of Medicine (Dr Schumacher-Feero and Dr Biglan), the Dong A University College of Medicine, Pusan, Korea (Dr Yoo), and the Department of Ophthalmology, Hospital Del Nino, Lima, Peru (Dr Mendiola). Supported in part by a grant from the Children's Hospital of Pittsburgh, One Children's Place, Pittsburgh, Pennsylvania (Dr Biglan). TR. AM. OPHTH. SOC. VOL. XCVI, 1998

456 Schunacher-Feero et al taining the level of visual acuity present at the onset of the CNIII palsy, and improvement in acuity is difficult to achieve. INTRODUCTION Third cranial nerve (CNIII) paresis or palsy occurs uncommonly in children. Miller,' in a review of more than 3 million records kept at the Wilmer Eye Institute, was able to identify only 28 patients with this condition who were first seen when younger than 20 years of age. Whereas CNIII palsy is usually caused in adults by such life-limiting conditions as neoplasm, microvascular disease, and aneurysms of the central nervous system, these conditions rarely cause CNIII palsy in children.2 Instead, CNIII palsy in children is most often congenital (due to adverse intrauterine events or a traumatic delivery) or the result of postnatal trauma, infection, or migraine.',5 The goals of managing CNIII palsy in children are to promote optimal development of visual acuity and binocular function. Long-term planning is required to achieve these goals and includes managing diplopia and preventing suppression of vision.6 Children with congenital CNIII palsy have a high incidence of amblyopia.57 Amblyopia is challenging to manage and is caused by paresis of the extraocular muscles, dysfunction of the levator palpebrae muscle leading to obstruction of the visual axis, and impairment of accommodation. In one study, 9 of 12 children with CNIII palsy had amblyopia.7 In another report of 11 eyes with congenital CNIII palsy, 5 eyes had a visual acuity of 6/12 or better after treatment only.5 Results following surgical procedures for strabismus are reported infrequently and usually focus on analysis of the effect achieved by a particular surgical procedure. Gottlob and associates8 treated 7 children with a superior oblique transposition procedure, and orthotropia in primary gaze was achieved in only 4. Later, Biglan and Walden9 reported that ocular alignment was achieved in 8 of 9 patients with CNIII palsy who underwent a modified Knapp procedure. One child, who was 11 years old at the time of surgery, achieved excellent fusion and stereo acuity. To the best of our knowledge, this is the only report documenting effectiveness of a surgical procedure in restoring binocular function in a child with congenital CNIII palsy. Blepharoptosis is a common consequence of CNIII palsy and can be problematic to treat.10-'3 The eyelid may cover the pupil, causing or exacerbating amblyopia management, and repair of this condition may be complicated by exposure keratitis and a poor Bell phenomenon due to dysfunction of the superior rectus or inferior oblique. The prognosis for binocular vision in visually immature children with severe, uncorrected ptosis is guarded.

Third Nerve Palsy in Children Over the past 20 years, one of the authors (A.WB.) has observed an improvement in the prognosis of children with partial, acquired CNIII palsies. To test the validity of this impression, we studied the course of CNIII palsy among patients with partial or complete oculomotor nerve palsy. We analyzed the prevalence of amblyopia, good visual acuity, satisfactory ocular alignment, and binocular function. We assessed the effect of procedures used to correct strabismus and blepharoptosis in 2 groups of patients: those with partial CNIII palsy and those with complete CNIII paralysis. From our results, we developed guidelines for management of CNIII palsy in children. METHODS.457 Medical records of patients treated between 1981 and 1996 by members of our university-based pediatric ophthalmology practice were reviewed to identify patients younger than 18 years of age with a diagnosis of CNIII palsy. The patient's sex and age at onset of the palsy and age at presentation, the length of patient follow-up, and the etiology of the palsy were recorded, as well as whether the palsy was complete or partial and whether clinical signs of aberrant regeneration were present. These data were tabulated. A palsy was complete if no adduction or vertical action of the superior or inferior rectus or the inferior oblique was recorded, the pupil was dilated and unresponsive to light, and ptosis was present. A partial palsy was characterized by limited adduction and vertical movements of the globe on ductions, a normal or dilated but responsive pupil, and a normal or ptotic eyelid. In each case, best-corrected visual acuity at presentation and subsequent visits, the methods used to assess visual acuity (selected according to the patient's age and level of cooperation or responsiveness), presence or absence of amblyopia, and, if present, the type and response to treatment were recorded.14 For this study, visual acuity was classified as "good" if corrected Snellen visual acuity was between 6/5 and 6/12, "fair" if between 6/15 and 6/30, and "poor" if the best-corrected visual acuity was equal to or worse than 6/60. Because of age or mental status, some children could be assessed using the fixation response. For purposes of analysis, central and maintained fixation was considered "good" vision, central but not maintained vision was considered "fair," and inability to fixate on the target was considered "poor" vision. The degree of ocular misalignment was assessed in very young or nonresponsive patients by using the Krimsky light reflex test. In older, cooperative patients, horizontal and vertical ocular alignments were measured at 6 m and at X m using the alternate prism and cover test with best optical correction in place.

458 Schumacher-Feero et al Horizontal ocular alignment was considered "good" if it was +/ 10 prism diopters (PD) from orthotropia in primary gaze, "fair" if horizontal ocular deviation was between 11 and 18 PD, and "poor" if it was greater than +/ 18 PD. Vertical alignment was considered "good" if it was within 2 PD of orthotropia in primary gaze, "fair" if between 3 and 5 PD of orthotropia, and "poor" if there was a vertical deviation of greater than 5 PD of hypertropia. When strabismus surgery was performed, the procedure and alignment in primary gaze 2 months following surgery and at the most recent visit were recorded. Binocular function was assessed with spectacle correction in place. The Worth four-dot test was administered at 6 m and at X m. Stereo acuity was measured using the Titmus stereo acuity test at A m. On initial evaluation, the degree of blepharoptosis was classified as "mild" if the level of the eyelid differed by 2 mm or less from normal (taken as the level of the fellow eyelid for those with unilateral blepharoptosis), "moderate" for differences between 3 and 4 mm, and "severe" for more than 4 mm ofptosis or if the eyelid occluded the visual axis. Because of a patient's young age or inability to cooperate with measurement, levator excursion frequently could not be quantitated. After surgical correction, the eyelid position was considered to be "good" if less than 2 mm of ptosis was present and "poor" if greater than 2 mm of ptosis was present a minimum of 6 months following surgery. RESULTS Over 65,000 computer-coded clinical records were reviewed; 45 children who had a unilateral palsy and 4 who had bilateral CNIII palsies were identified, for a total of 49 patients with 53 affected eyes. Seventeen patients (35%) were female. In 31 patients (33 eyes), onset of CNIII palsy was before 8 years of age; in 18 eyes, palsy was congenital. The mean age at onset was 59 months (range, birth to 198 months). The mean age at presentation was 69.5 months (range, 2 weeks to 200 months). CNIII palsy was partial in 32 (60%) of the 53 eyes and complete in the remaining 21 eyes (40%). ETIOLOGY OF CRANIAL NERVE III PALSY The etiology of CNIII palsy was congenital in most cases (Table I). Trauma was the second most common cause. Eight eyes had concurrent palsy of at least one other cranial nerve. Trauma was the etiology in 5 of these 8 eyes. Palsy was the cause in the remaining 3 eyes by an arterial venous malformation (1) or a brain-stem tumor (2). In addition to these 2 eyes with multiple cranial nerve palsies due to tumor, isolated CNIII palsy

Third Nerve Palsy in Children 459 was due to tumor in 4 other eyes. Of these 6 eyes in which CNIII palsy was due to tumors, 4 had gliomas, 1 had a "lipofibroma" of the orbit, and 1 had a "nonspecified brain stem tumor." TABLE I: ETIOLOGY OF CNIII PALSY IN 53 EYES ETIOLOGY PARTIAL PALSY COMPLETE PALSY TOTAL EYES Congenital 11 7 18 Trauma 7 10 17 Tumor 5 2 7 Vascular 4 2 6 Meningitis/encephalitis 4 0 4 Idiopathic 1 0 1 Total eyes 32 21 53 FOLLOW-UP TIME AND NERVE REGENERATION The mean follow-up period was 5.5 years (range, 0 to 19 years). Forty-five of 53 eyes were followed up for a minimum of 6 months, and 22 were followed up for more than 5 years. Clinical signs of aberrant regeneration of oculomotor nerve fibers were present in 17 eyes. Four of the eyes with partial CNIII palsies had complete spontaneous resolution within 1 to 28 months of onset. None of the eyes with complete CNIII palsy showed spontaneous resolution. VISUAL ACUITY Visual acuity at the time of presentation was considered good in 29 (55%) of the 53 eyes (Table II). The fixation response was central, steady, and maintained in 13 of these eyes. At the most recent evaluation, 30 eyes (57%) had visual acuity that was considered good by tests of letter or symbol recognition (Table III). Among the 30 eyes with good visual acuity at the most recent evaluation, the proportion with good vision was nearly identical among the 20 eyes (55%) with onset of CNIII palsy after 8 years of age when compared TABLE II: VISUAL AcumTY AT PRESENTATION IN 53 EYES wrm CNIII PALSY VISUAL Acurry PARTIAL PALSY COMPLETE PALSY TOTAL EYES Good (6/5-6/12) 18 11 29 Fair (6/15-6/30) 9 6 15 Poor (6/60 or worse) 2 2 4 Unable to assess 3 2 5

460 Schumacher-Feero et al to those with an onset of CNIII palsy before 8 years of age (57%) (Table IV). TABLE III: VISuAL Acurry AT MOST RECENT EVALUATION IN 53 EYEs with CNIII PALSy VISUAL AcurrY PARITIAL PALSY COMPLETE PALSY TOTAL EYEs Good (6/5-6/12) 19 11 30 Fair (6/15-6/30) 7 4 11 Poor (6/60 or worse) 2 5 7 Unable to assess 4 1 5 TABLE IV: VISUAL Acurry AT MOST RECENT EVALUATION IN 53 EYEs wrrm CNIII PALSY BY AGE AT ONSET VISUAL ACUITY UNDER 8 YEARS (N=33) OVER 8 YEARS (N=20) Good (6/5-6/12) 19 11 Fair (6/15-6/30) 9 2 Poor (6/60 or worse) 3 4 Unable to assess 2 3 AMBLYOPIA Among the 33 eyes in which the diagnosis of CNIII palsy was made before the child was 8 years old, 27 eyes were considered to have some degree of amblyopia. Treatment consisted of occlusion therapy for 25 eyes and glasses alone in 2 patients with anisometropia. For comparison, the results of amblyopia treatment were evaluated separately for children whose visual acuity was initially determined using the flxation response (17) or measured by Snellen acuity (6). Visual acuity at the initial evaluation was not obtainable in 4 eyes. Among the 17 eyes in which visual acuity was determined at the initial evaluation using the fixation response, 8 had central fixation at the time of the first visit but later developed unsteady fixation, and 6 had central fixation but at least a moderate preference for the uninvolved eye. The remaining 3 eyes in which the visual acuity was determined by fixation response were not able to fixate a target. Of these 3 eyes, at the most recent evaluation 1 had fair visual acuity, 1 had achieved central, steady fixation, and 1 was still unable to fixate a target. At the most recent evaluation among the 14 eyes that initially had central fixation, 5 eyes had good vision, 2 had fair vision, and 1 had poor visual acuity. Six maintained or achieved central, steady fixation. In the remaining 6 eyes treated for amblyopia, Snellen acuity was mea-

Third Nerve Palsy in Children sured at the initial evaluation. Treatment resulted in maintaining the visual acuity in 3 eyes, but 2 eyes lost 1 or 2 lines of Snellen visual acuity. No follow-up was available for 1 eye. Of the 27 eyes with amblyopia, 9 eyes had blepharoptosis severe enough to be considered a risk for inducing deprivation amblyopia. Conditions other than amblyopia accounting for decreased visual acuity included optic atrophy (2 eyes), an optic nerve coloboma (1 eye), a macular scar (1 eye), and uncorrected incyclotorsion (4 eyes). Despite the multiple hindrances to achievement of good visual acuity, 16 of the 27 eyes (59%) had either good Snellen acuity or central and steady fixation at the most recent evaluation. OCULAR ALIGNMENT AND STRABISMUS SURGERY 461 At the initial evaluation, horizontal alignment was good in only 6 (12%) of the 49 patients (Table V). This increased to 30 (61%) of the patients at the most recent evaluation (Table VI). The vertical alignment of the eyes was good in 24 (50%) of the 49 patients at the initial evaluation and in 35 (71%) at the most recent evaluation. TABLE V: OcuLAR ALIGNMENT AT INITIAL EVALUATON IN 49 CHILDREN WITH CNIII PALSY HRIZONTAL PARTIAL COMPLETE TOTAL NO. OF) AUGNMENT PALSY (N=31) PALSY (N=18) CHILDREN (N=49) Good (0-10 PD) 5 1 6 Fair (11-18 PD) 1 0 1 Poor (>18 PD) 19 14 33 Unable to assess 6 3 9 VERTICAL ALIGNMENT Good (0-2 PD) 14 10 24 Fair (3-5 PD) 3 1 4 Poor (>5 PD) 11 4 15 Unable to assess 3 3 6 PD, prism diopters. Partial CNIII Palsy Among the 31 children (32 eyes) with partial CNIII palsy, 15 eyes did not undergo strabismus surgery. Four of these eyes had complete spontaneous resolution, and the remainder had partial recovery, which made surgery unnecessary. The remaining 17 eyes underwent a mean of 1.5 sur-

462 Schumacher-Feero et al TABLE VI: OcuLAR ALIGNMENT AT MOST RECENT EVALUATION IN 49 CHILDREN WITH CNIII PALSY HORIZONTAL PARTIAL COMPLETE TOTAL NO. OF) ALIGNMENT PALSY (N=31) PALSY (N=18) CHILDREN (N=49) Good (0-10 PD) 20 10 30 Fair (11-18 PD) 2 2 4 Poor(>18PD) 4 4 8 Unable to assess 5 2 7 VERTICAL ALIGNMENT Good (0-2 PD) 20 15 35 Fair (3-5 PD) 0 0 0 Poor(>5 PD) 5 2 7 Unable to assess 6 1 7 PD, prism diopters. gical procedures to correct horizontal, vertical, and cyclorotatory alignment. Of the 14 eyes for which an operation for horizontal alignment was performed, the initial operation was a recession/resection procedure in 9 eyes and a modified Knapp procedure with vertical transposition of the medial and lateral rectus muscles in 5 (Table VII).9 Four eyes required a second muscle recession/resection operation to correct horizontal alignment. A surgical procedure to correct vertical alignment was used alone or combined with a horizontal recession/resection in 7 eyes. TABLE VII: CORRECTIVE STRABISMUS PROCEDURES IN 32 EYES OF CHILDREN WTIH CNIII PALSY HORIZONTAL CNIII PROCEDURE No. OF No. OF ALIGNMENT PALSY EYES REOPERATIONS Partial Recess/resect 9 3 Modified Knapp 5 2 Complete Recess/resect 15 17 Modified Knapp 3 2 VERTICAL ALIGNMENT MUSCLE TREATED Partial Superior obhque 5 0 Vertical rectus 2 0 Complete Superior oblique 1 0 Vertical rectus 3 0

Third Nerve Palsy in Children Among the 17 of 32 eyes with partial CNIII palsy treated by alignment surgery, the mean time from onset of the palsy to surgery was 29 months (range, 0 to 326 months). Excluding a single eye that underwent strabismus surgery more than 20 years after onset of CNIII palsy, the mean time to surgery was 12 months. The mean time to reoperation was 60 months (range, 3 to 188 months). Complete CNIII Palsy Among the 18 patients (21 eyes) with complete CNIII palsy, 3 eyes did not undergo strabismus surgery. One of these eyes had no light perception and was enucleated, 1 had a complex and variable ocular motility pattern secondary to a brain-stem malformation, and 1 eye was in a patient with bilateral CNIII palsies and a CNVI palsy. The remaining 18 eyes underwent a mean of 2.3 alignment procedures. The initial surgical procedure to correct strabismus in 15 eyes was a recession/resection of the horizontal extraocular muscles; a second such operation was required in 10 and a third operation was required in 5 to obtain good alignment. One eye underwent 4 recession/resection procedures without obtaining satisfactory alignment. In the remaining 3 eyes, the initial surgery was a modified Knapp procedure, and 2 required an additional recession/resection procedure. Of the 21 eyes with a complete CNIII palsy, 4 required separate procedures for vertical alignment. One eye underwent a recession/resection procedure combined with a Humelscheim procedure. Botulinum A toxin was injected into the lateral rectus muscle in 3 eyes, with limited success. The mean time from onset of palsy to the initial alignment surgery was 6 months (range, 0 to 11 months). The mean time to a subsequent alignment procedure to treat a complete CNIII palsy was 23 months (range, 2 to 122 months). BINOCULAR FUNCTION 463 A fusion response at distance and at near occurred in 4 of the 31 patients with partial CNIII palsy (Table VIII). Two of these children had spontaneous resolution of their CNIII palsy. None of the children with a complete palsy were able to demonstrate a fusion response at 6 m. Ten patients were unable to provide a response or were not tested. The remaining patients demonstrated suppression or diplopic responses on testing. Two (6%) of the 31 patients with partial CNIII palsies had stereo acuity better or equal to 100 seconds of arc. No patient with a complete CNIII palsy was able to demonstrate stereo acuity better than 100 seconds of arc. Half of those tested had no measurable stereo acuity, including 7 (22%) of the 31 children with partial CNIII palsy and 11 (61%) of the 18 children with complete CNIII palsies.

464 Schumacher-Feero et al TABLE VIII: BINOCULAR FUNCTION OF 49 CHILDREN WITH CNIII PALSY AT THE MOST RECENT EVALUATION FUSION PARTIAL COMPLETE TOTAL NO. OF) PALSY (N=31) PALSY (N=18) CHILDREN (N=49) Distance (6 meters) 4 0 4 Near (% meter) 5 3 8 Unavailable 4 6 10 STEREO ACUITY 40-100 seconds of arc 2 0 2 >100 but <800 seconds of arc 7 2 9 >800 seconds of arc 6 1 7 Unavailable 7 6 13 PD, prism diopters. BLEPHAROPTOSIS Partial CNIII Palsy A total of 8 of 32 eyes with partial CNIII palsy underwent a mean of 1.4 procedures to correct blepharoptosis. Fourteen eyes (44%) with partial CNIII did not have ptosis. Among the 18 remaining eyes, blepharoptosis was mild in 9 eyes and moderate in the other 9 eyes. Of the eyes with mild ptosis, 2 underwent levator palpebrae muscle resection. In 1 eye the degree of ptosis was not improved; a second levator resection procedure was performed and had a good result. Four of the 9 eyes with moderate ptosis underwent maximal levator resections, and 2 underwent external TABLE IX: BLEPHAROPTOSIS IN 23 CHILDREN WITH CNIII PALSY EXTENT INrrIAL No. OF INITIAL TOTAL EYEs WITH OF PALSY DEGREE EYES PROCEDURE No. OF GOOD FINAL OF PTOSIS OPERATIONS LID PosrmoN LEVATOR RESECTON EXTERNAL SUSPENSION PARTIAL Mild 9 2 0 3 9/9 Moderate 9 4 2 6 8/9 Severe 0 0 0 0 0/0 COMPLETE Mild 1 1 0 1 1/1 Moderate 7 6 0 8 6/7 Severe 8 5 3 12 7/8

Third Nerve Palsy in Children sling procedures (Table IX). The mean time from onset of the partial CNIII palsy to surgery for correction of ptosis was 21 months (range, 5.5 to 45 months). In 3 of the 8 eyes, ptosis repair was undertaken at the time of alignment surgery. The mean time between surgery to correct ocular alignment and surgery to correct ptosis in the remaining eyes was 9 months (range, 5 to 20 months). Reoperation with a second levator resection operation was needed for 1 eye with mild and 2 eyes with moderate blepharoptosis. The mean time to reoperation was 3.8 years. Complete CNIII Palsy Five of the 21 patients with complete CNIII palsies did not have a minimum follow-up period of 6 months following ptosis surgery and were not included in the analysis. One eye with moderate ptosis did not have surgery on the eyelid. The remaining 15 eyes with complete CNIII palsy underwent a mean of 1.4 procedures to correct ptosis. Ptosis was classified as mild in 1 eye with aberrant regeneration. Levator resection was performed to correct ptosis, and postoperatively the eyelid position was considered good. Ptosis was moderate in 7 eyes affected by complete CNIII palsy. Six eyes were treated by levator resection. In 1 of these eyes the eyelid position was poor postoperatively, and two additional resections did not achieve a good eyelid position. Ptosis was severe in 8 eyes with complete CNIII palsy. Five eyes underwent levator resection, and 3 were treated with an external eyelid suspension procedure. Two of the eyes undergoing levator resection and 1 eye treated with a sling procedure had poor results. Two of these 3 underwent a Fassanella- Servat procedure, and 1 of these 2 required an additional levator resection to achieve good eyelid position. The last of the 3 patients with poor eyelid position following levator resection underwent a sling procedure. Complications of ptosis repair included 1 case each of eyelid infection, symblepharon formation, and redundant conjunctiva as well as 3 cases of exposure keratitis. The mean time from onset of complete CNIII palsy to correction of ptosis was 14 months (range, 0 to 58 months). In 7 of the 15 eyes, ptosis repair was performed at the time of the initial alignment surgery to reduce the risk of repeated anesthesia. For the remaining eyes, ptosis repair was undertaken a mean of 15 months (range, 2 to 45 months) after alignment. The mean time to reoperation on the eyelid was 2 years. DISCUSSION 465 Palsy of the oculomotor nerve is rarely seen in children, and comprehen-

466 Schumacher-Feero et al sive guidelines for management and outcome are difficult to find.'-"" In adults, the conditions causing CNIII palsy frequently are progressive or recurrent and often lead to profound disability or death. Therefore, treatment for CNIII palsy is often deferred or not recommended. In children, CNIII palsies are frequently congenital or a result of trauma or infection. Partial or full recovery from the CNIII palsy occasionally occurs, and in most cases life expectancy is not diminished. Dysfunction of CNIII with its resultant strabismus, blepharoptosis, and lack of accommodation causes symptoms that are functionally and cosmetically disabling. These defects also have a profound impact on the development of vision and binocular function. The goals of this study were to examine the success of treatment of CNIII palsy in children. While we used surgical principles recommended for management of CNIII palsy in adults, our goals for treating children in this study were to promote or maintain good visual acuity and to achieve optimal development of binocular function. The causes of CNIII palsy in this study are similar to those reported elsewhere. Palsy in our study was most frequently congenital or due to postnatal trauma. In a series of 28 children reported by Keith,4 CNIII palsy was most often due to trauma, infection, and idiopathic causes, and in a series of 30 children studied by Miller,' CNIII palsy was most often congenital or due to traumatic or inflammatory causes. Congenital, traumatic, and infectious causes were also the most frequent causes of CNIII palsy in 38 children studied by Ing' at the Hospital for Sick Children in Toronto, and Harley3 found that among children at St. Christopher's Hospital in Philadelphia, the most common causes of CNIII palsy were congenital, traumatic, inflammatory, and neoplastic. Hamed20 found that children with congenital CNIII palsy had a high incidence of associated neurologic diseases. However, Kodsi and Younge21 found that of the 35 children presenting to the Mayo Clinic with CNIII palsy, tumors were the single most common cause of the condition, followed by cryptogenic causes. This discrepancy may reflect referral bias. In children who have onset of a CNIII palsy before the age of 8 or visual maturity, the attainment of good visual acuity may be adversely influenced by the development of strabismic and deprivation amblyopia as well as the inability to accommodate. Ancillary problems such as exposure keratitis following ptosis repair and the prevalence of anisometropic and astigmatic refractive errors all contribute to the disappointing visual prognosis.2' Others have documented the deleterious effects of early-onset CNIII palsy on visual development. For example, Victor7 found a high prevalance of amblyopia among children with congenital CNIII palsy, and the best visual acuity among his patients was 6/21 in the affected eye. Ing

Third Nerve Palsy in Children 467 and colleagues5 also found amblyopia to be common among children with CNIII palsy but more responsive to treatment; visual acuity was 6/12 or better in 5 of his 11 cases. In this series, amblyopia was prevalent, occurring in 27/33 eyes (82%) in children under 8 years of age. This study confirms that visual acuity is likely to be impaired in children with CNIII palsy. The assessment of visual acuity at the initial examination may be difficult, especially in young children or those recovering from an acute episode of infection or trauma. In these situations, the fixation response may be the only measure of visual acuity that can be obtained. Because we had to use this method to measure visual acuity initially in most patients younger than 8 years, it was difficult to obtain a precise quantitative evaluation of change in vision from the initial to the final evaluation. In those eyes in which quantitative measures of visual acuity could be obtained, none had an improvement in visual acuity. Overall, the proportion of visually immature children with CNIII palsy and amblyopia who had quantifiable visual acuity better than or equal to 6/12 was the same at the final as at the initial evaluation (50% in each case), despite efforts to improve visual acuity by treating amblyopia. The finding of no alteration in visual acuity from the initial to the final evaluation seemed to be independent of whether CNIII palsy was partial or complete (Table II) and the age of the patient at onset of the palsy (Table III). Overall, 57% of children younger than 8 years at onset compared to 55% of those older than 8 years had a final visual acuity of 6/12 or better. We recommend meticulous attention to prevention and treatment of amblyopia, recognizing that this appears to be effective in maintaining the visual acuity but not improving it. In this study, all patients had amblyopia in the eye affected by CNIII palsy. However, this is not always the case.'72 Caution should be exercised in treating amblyopia aggressively when the eye affected by CNIII palsy is amblyopic, because converting the eye affected by CNIII palsy to the eye of preference will cause large, undesirable secondary deviations. We had greater success treating strabismus than improving visual acuity. Defects of the oculomotor nerve cause an eye to assume a hypotropic and exotropic position. Aberrant regeneration and lack of function of the inferior oblique muscle produce a complex combination of horizontal, vertical, and cyclotorsional strabismus. In this study, only 6 of 49 patients had good horizontal ocular alignment at the initial evaluation, and 24 had good vertical alignment. At the most recent evaluation, 30 patients had good horizontal and 35 had good vertical alignment. This success was achieved by multiple surgical procedures. Over the 5.5-year follow-up period, a mean of 2.3 procedures were necessary to align eyes with complete palsy compared with a mean of 1.5 procedures for eyes with a partial CNIII palsy which underwent surgery. These results were usually achieved by

468 Schumacher-Feero et al recession/resection procedures, with graded supraplacement of the insertions of the rectus muscles to elevate the globe in cases with hypotropia and exotropia.9 Of the patients with partial CNIII palsy, 47% did not require any alignment surgery because of spontaneous resolution or partial recovery. After the recession/resection procedure, incyclotorsion caused by unopposed action of the superior oblique muscle was treated with procedures designed to weaken its action. Others have treated CNIII palsy with transposition of the superior oblique tendon to the superior border of the medial rectus.15'2324 In this series, 1 eye was treated with this procedure, and although the eye was centered, the diplopia and strabismus outside of primary gaze were unacceptable. The 1 eye in this series treated with a Hummelscheim procedure (vertical rectus muscle transposition) had disappointing results. This procedure can be expected to have limited success because it transfers the forces of the paretic inferior and superior rectus muscles nasally to correct exotropia. Transferring the force of paretic muscles is unlikely to result in lasting eye alignment. This procedure also carries the additional risk of causing anterior segment ischemia if additional horizontal rectus muscle surgery is performed. We continue to advocate use of a large recession/resection procedure with graded supraplacement of the insertions when CNIII palsy is partial and unilateral, particularly when the medial rectus muscle continues to have some function. If necessary, supplemental injections of botulinum A toxin into the lateral rectus muscle help to further reduce abducting forces and temporarily improve alignment in primary gaze. However, the best results with botulinum toxin occur when the opposing muscle is functional. When CNIII palsy is complete or bilateral, achieving satisfactory alignment is more problematic and requires more surgical procedures. Our preference for the initial surgery is a large recession/resection designed to cripple the lateral rectus muscle; this procedure was also used in all of the eyes that required reoperation for horizontal alignment. Five of 11 eyes undergoing a second recession/resection procedure required a third operation. In some patients, satisfactory alignment could not be achieved in primary gaze despite multiple surgeries. Two patients were treated with prisms after surgery. Response to this treatment was unsatisfactory, and these patients eventually elected to discontinue this method of treatment. Seventeen children with acquired palsy of CNIII showed signs of aberrant oculomotor nerve regeneration, a situation that complicates surgery to correct ocular alignment and blepharoptosis. In theory, synkinesis of the medial rectus muscle and the levator palpebrae muscle could help to elevate a droopy eyelid with attempts to adduct the eye.25 This

Third Nerve Palsy in Children 469 may occur when the preferred eye for fixation is the paretic eye. If this eye is slightly exotropic at the resting position, attempts by the patient to center the eye or adduct it will cause the eyelid to elevate. In these uncommon situations, it may be desirable to leave the affected eye in a slightly exotropic position. The experience of one of the authors (A.W.B.) who has used this procedure to treat adults with CNIII palsy and aberrant regeneration suggest that the procedure works better in theory than in practice. It is ideal to preserve or reestablish binocular function in children with CNIII palsy. Our success in achieving this goal was poor. The probability of achieving good binocular function was notably higher for those with partial CNIII palsy; children with complete CNIII palsy are quite unlikely to achieve either fusion at distance or high levels of stereo acuity. In very young children with congenital CNIII palsy, satisfactory alignment that promotes binocular function may not be accomplished before the developmental period for achievement of binocular function has passed. Nevertheless, one child in our study with a congenital CNIII palsy did achieve fusion at distance and near following a single recession/resection procedure performed at age 8 months. When the etiology of CNIII palsy in a child is unknown or there is a reasonable expectation of recovery, we recommend waiting a minimum of 6 months (mean time in this study was just under 9 months) after onset of CNIII palsy before surgical correction of strabismus. This delay allows time for confirming the etiology of the palsy, for recovery to occur, and for stabilization of the angle of strabismus. Aberrant regeneration may occur during this time and lead to an alteration in the surgical plan. During this waiting period amblyopia may occur. Earlier surgical intervention may be warranted, however, when the CNIII palsy is known to be due to such causes as tumor resection with unequivocal and irreparable damage to the third nerve nucleus or its efferent pathways. It has been suggested that patients with CNIII palsy due to trauma undergo a 1-year observation period before surgical treatment of ptosis." An excessively long waiting period poses a high risk for visual deprivation amblyopia in children who are younger than 8 years. In these children, earlier surgical intervention is warranted. Correction of ptosis is usually undertaken after alignment of the globe because of the possibility that repositioning of the extraocular muscles might lead to alteration in the position of the eyelid, especially when aberrant regeneration is present. In several of the patients with congenital, traumatic, or tumor-induced CNIII palsies, ptosis was corrected at the time of the primary alignment procedure. Three of these 9 eyes eventually required a second eyelid correction procedure, results of which were similar to those for children who underwent eyelid repair subsequent to ocular alignment. Overall, 22% of eyes

470 Schumacher-Feero et al required a second procedure for repair of ptosis irrespective of whether the palsy was partial or complete. We recommend a maximal resection of the levator palpebrae muscle as the initial procedure. In severe cases, we attempted to include levator tissue above Whitnall's ligament when placing sutures from the tarsus back through the levator.10,13 This procedure was initially effective in elevating the eyelid, but it occasionally had to be repeated several years later. In this series, experience with sling procedures or suspension procedures external to the orbital septum was disappointing: The cosmetic and functional results were only fair, and the effect produced by the sling tended to dissipate with time. To counteract the loss of effect over time, overcorrection of the eyelid position was attempted at the time of surgery. This, in turn, placed the eye, often with poor elevating capability, at high risk for developing exposure keratitis. In cases of complete congenital CNIII palsy, early use of a sling procedure may be necessary to provide the child with a satisfactory but temporarily unobstructed visual axis. The findings in this study are limited by the problems inherent in any retrospective review, including surgeon bias. To decrease the effect of this factor, patients were examined by multiple surgeons throughout the follow-up period. A preprinted data recording form to standardize observations of visual acuity, alignment, refraction, and binocular function was used throughout the study period. In addition, photographs were taken of many of the children to document the degree of ptosis and strabismus and were available for review. Owing to the relatively short length of followup and the nature of our practice, we are unable to make statements regarding the future needs for additional surgery for ocular alignment and ptosis repair during adulthood. In addition, 12 of the 27 (44%) of the children with amblyopia were still under the age of 8 years at their most recent visit. With continued treatment, these children may show an improvement in their visual acuity beyond the acuity documented at the time of the initial evaluation. This outcome would be very encouraging and will require further study as these children pass the age of visual maturity. In summary, the results of this study indicate that the visual acuity present shortly after the onset of CNIII palsy in children generally predicts visual acuity 5 or more years later and that Snellen visual acuity will be 6/12 or better in almost two thirds of children. Persistence in administering occlusion therapy and encouraging the use of spectacles, including bifocals, can help to prevent amblyopia and preserve vision. Generally, however, these measures were not effective in improving vision. Ocular alignment can be reestablished surgically, with the goal that both eyes can be used to fixate objects in primary gaze and in the reading position. The failure of children in this study to achieve high levels of binocular function is disappointing and suggests that the benefits of such

Third Nerve Palsy in Children 471 surgery may be more cosmetic than functional. Nevertheless, good function was achieved in some children, including 1 child with a congenital palsy Ȯnce a satisfactory ocular alignment has been achieved, severe blepharoptosis should be corrected with a levator resection to provide an unobstructed visual axis. Treatment may be delayed in patients with partial CNIII palsy and minimal obstruction of the visual axis. Because of the lack of, or reduced innervation to, the levator muscle, maximum external levator resection procedures and internal suspension procedures will be necessary. Frequent follow-up is recommended, especially for children whose vision has not yet matured. REFERENCES 1. Miller N. Solitary oculomotor nerve palsy in childhood. Am J Ophthalmol 1977;83:106-111. 2. Rush JA, Younge BR. Paralysis of the cranial nerves III, IV, and VI: Cause and prognosis in 1,000 cases. Arch Ophthalmol 1981;99:76-79. 3. Harley RD. Paralytic strabismus in children. Etiologic incidence and management of the third, fourth and sixth nerve palsies. Ophthalmology 1980;87:24-43. 4. Keith CG. Oculomotor nerve palsy in childhood. Aust N ZJ Ophthalmol 1987;15:181-184. 5. Ing EB, Sullivan TJ, Clarke MP, et al. Oculomotor nerve palsies in children. J Pediatr Ophthalmol Strabismus 1992;29:331-336. 6. Mizen TR, Burde RM, Klingele TG. Cryoptogenic oculomotor nerve palsies in children. Am JOphthalmol 1985;100:65-67. 7. Victor DI. The diagnosis of congenital unilateral third nerve palsy. Brain 1976;99:711-718. 8. Gottlob I, Catalano RA, Reinecke RD. Surgical management of oculomotor nerve palsy. AmJ Ophthalmol 1991;111:71-76. 9. Biglan AW, Walden PG. Results following surgical management of oculomotor nerve palsy with a modified knapp procedure. Ophthalmic Surg 1985;16:759-764. 10. Berke RN. Results of resection of the levator muscle through a skin incision in congenital ptosis. Arch Ophthalmol 1959;61:177-201. 11. Krohel GB. Blepharoptosis after traumatic third-nerve palsies. Am J Ophthalmol 1979;88:598-601. 12. Beyer CK, McCarthy RW. Surgical treatment of ptosis in acquired third nerve paralysis. Ann Ophthalmol 1981;13:373-376. 13. Malone TJ, Nerad JA. The surgical treatment of blepharoptosis in oculomotor nerve palsy. AmJ Ophthalmol 1988;105:57-64. 14. von Noorden GK. Examination of patient. III. Sensory signs, symptoms and adaptations in strabismus. In: von Noorden GK, ed. Binocular Vision and Ocular Motility: Theory and Management of Strabismus. 5th ed. St Louis: Mosby; 1996:216-223. 15. Scott AB. Transposition of the superior oblique. AmJ Orthop 1977;5:11-14. 16. Helveston EM. Surgical Management of Strabismus: An Atlas of Strabismus Surgery. 4th ed. St Louis: Mosby; 1993:502-508. 17. Pratt-Johnson JA, Tillson G. Paralytic and paretic strabismus. In: Pratt-Johnson JA, Tillson G. Management of strabismus and amblyopia: A practical guide. New York: Thieme Medical Publishers; 1994:174-177.

472 Schumacher-Feero et al 18. Eggers HM. Oculmotor (third nerve) paralysis. In: Frauenfelder FT, Roy FH, Grove J, eds. Current Ocular Therapy 4. Philadelphia: Saunders; 1995:549-550. 19. Buncic JR. Third nerve palsy. In: Good VW, Hoyt CS, eds. Strabismus Management. Boston: Butterworth-Heinemann; 1996:287-295. 20. Hamed LM. Associated neurologic and ophthalmic findings in congenital oculomotor nerve palsy. Ophthalmology 1991;98:708-714. 21. Kodsi SR, Younge BR. Acquired oculomotor, trochlear and abducent cranial nerve palsies in pediatric patients. AmJ Ophthalmol 1992;114:568-574. 22. Kazarian EL, Flynn JT. Congenital third nerve palsy with amblyopia of the contralateral eye. J Pediatr Ophthalmol Strabismus 1978;15:366-367. 23. Metz HS, Yee D. Third nerve palsy: superior oblique transposition surgery. Ann Ophthalmol 1973;5:215-218. 24. Saunders RA, Rogers GL. Superior oblique transposition for third nerve palsy. Ophthalmology 1982;89:310-315. 25. O'Donnell FE, Del Monte M, Guyton DL. Simultaneous correction of blepharoptosis and exotropia in aberrant regeneration of the oculomotor nerve by strabismus surgery: a new, simplified ptosis correction for selected cases. Ophthalmic Surg 1980;11:695-697. DISCUSSION DR DAVID R. STAGER. I want to thank Al Biglan and his coauthors for getting this very thought-provoking paper to me and others in this audience in a timely manner. This is one of the most difficult clinical entities with which pediatric ophthalmologists deal, involving most areas of our subspecialty. As we all know, amblyopia severity depends on the age at onset, the duration without treatment, and the age at adequate treatment. Of the three types of amblyopia (deprivation, strabismic, and refractive), patients with congenital III nerve palsy could have all three. In the present study, there was a high incidence of amblyopia in even the older patients, indicative of some additional factor being responsible. Also, standard amblyopia treatment does not seem to be effective, despite Dr Biglan's conscientious care. The easiest area to miss is adequate refractive correction for the accommodative weakness. Limited or absent vertical ductions and vertically contracted binocular visual fields may make bifocals useless without manual manipulation. Still, there must be other factors that lead us to such dismal success in our treatment. One factor may be reduced fine-motor control of fixation and following. This could also explain the apparent amblyopia that occurs in patients beyond the amblyopia age. Other explanations might be nystagmus, disruption to the normal tear layer and corneal epithelium, or concurrent retinal, optic nerve, or CNS pathology. Poor compliance with adequate treatment as well as a desire to avoid overtreatment are additional factors limiting amblyopia therapy. Ptosis is a special problem because of poor Bell's phenomenon and exposure keratitis. Early ptosis correction may reduce the incidence of

Third Nerve Palsy in Children 473 exposure keratitis by stimulating adaptation mechanisms to basic tear secretion and the corneal epithelium. An early temporary sling is advocated by Dr Biglan's group. In older patients, a similar temporary removable sling or tape may determine the ability of the eye to withstand the exposure. An internal sling off Whitnall's ligament works well. However, a full Beard procedure or a "chicken Beard" may create better symmetry of the lids. Strabismus problems are complicated because of the paretic horizontal, vertical, and torsional components. Horizontally, an R&R is the procedure of choice. The volume of horizontal surgery needs to be increased from the standard numbers for nonparalytic strabismus procedures. With recessions of up to 13 mm or more, a marginal myotomy or, ultimately, a disinsertion may be necessary. I avoid a hang-back or hang-loose suture, since those muscles can reattach to the globe in an unpredictable manner. Regarding the vertical hypotropia with intorsion, the SO tenotomy, as suggested by Del Monte and Guyton, usually corrects this without the need for the vertical transposition of the horizontal recti. Transposing the superior oblique muscle to the globe nasally can lead to an unsightly and permanent hypertropia. Fusion of any significance is often impossible to attain. Patients with an abnormal head position preoperatively will probably fuse postoperatively. If they do not have an abnormal head position, then fusion is not an issue and surgery is done at any age the parents would like. Cosmetically, these are very devastating physical deformities, and it is important to do all we can to restore an acceptable appearance. I would just have a couple of questions for Dr Biglan. First, what is the incidence of abnormal head posture in the congenital and acquired groups, and is it related to amblyopia? Second, in a patient with aberrant regeneration that elevates the lid on adduction, what success have you had in operating on the noninvolved eye to stimulate adduction and lid elevation in the involved eye? Although this paper provides a gold mine of good information, we must all realize that each case is individual, necessitating a customized approach. ZANE POLLARD, MD. What do you think about operating on the uninvolved eye? Of course, it may be difficult to explain this to patients and talk them into such surgery. I have done this a few times with some success. JOHN FLYNN, MD. Dr Biglan, my question is, how many patients in your group had aberrant regeneration? Secondly, in congenital third nerve palsies, there are two things that set them apart from acquired palsy. First, they have lid nystagmus when they attempt adduction. I cannot explain

474 Schumacher-Feero et al the phenomenon, but it is real. Secondly, the pupil is never a blown pupil. It is a mid-dilated pupil. One thing that is helpful here cosmetically is to use weak doses of Pilocarpine which the patients can, by titrating the doses, make the pupils as equal as possible. It is a small point, but it is important from a cosmetic standpoint to do anything you can for these patients. Congratulations. ALBERT BIGLAN, MD. I would like to thank Dr David Stager for his comments. Many of these patients have other problems. One of these patients had a traumatic macular scar. Another had severe optic nerve atrophy. Others have associated central nervous system injuries. Two patients had radiation for brain tumors and have ocular problems associated with that. One of our problems is to have patients look cosmetically acceptable with their glasses on. This sometimes requires surgical elevation of the lids. I cannot give you a lot of data on the head posture these patients had. The most significant problem with head posture was a young man I did a KNAPP procedure on and did not weaken the superior oblique. This patient had a right third nerve palsy, but had his head tilted to the right side. He had incyclotorsion on fundus examination. I did a right superior oblique tenotomy, weakening the superior oblique. This corrected the head posture. The second question was about surgery for the other eye. I have tried this in adults, but not in children. The success I have had in adults has left much to be desired. Regarding Dr Flynn's comments on aberrant regeneration, I cannot give you the exact incidence of this. I agree with him that using Pilocarpine helps to manage the pupil disparity. I would like to thank the members of the program committee for the opportunity to share this information with you. Thank you.