To investigate the superior oblique (SO) extraocular muscle cross section in normal controls and in SO palsy using high-resolution magnetic resonance imaging (MRI).
Prospective observational study.
At a single academic medical center, high-resolution MRI was obtained at 312 μm in-plane resolution using surface coils in multiple, contiguous, quasi-coronal planes perpendicular to the orbital axis in 12 controls and 62 subjects with SO palsy. Previous strabismus surgery was excluded. Imaging was repeated in central gaze and infraduction. In each image plane along the SO, its cross section was outlined to compute cross-sectional area and the major and minor axes of the best-fitting ellipse. Main outcome measures were SO morphology and ocular motility.
The major and minor axes, cross-sectional area distributions, and volume of the SO belly were subnormal in orbits with SO palsy at most anteroposterior locations ( P = .001), but discriminant analysis showed that palsied SO cross sections segregated distinctly into round and elongate shapes representing isotropic vs anisotropic atrophy, respectively. The major axis was relatively preserved in anisotropic atrophy ( P = .0146). Cases with isotropic atrophy exhibited greater hypertropia in infraversion than central gaze, as well as greater excyclotorsion, than cases with anisotropic atrophy ( P < .05 for all).
Characteristic differences in shape of the palsied SO belly correlate with different clinical features, and may reflect both the degree of differential pathology in the medial vs lateral neuromuscular SO compartments and the basis for diversity in patterns of resulting hypertropia.
Compartmentalization has been recognized recently as a feature of most extraocular muscles. There is a selective pattern of terminal intramuscular motor innervation in horizontal rectus muscles and the inferior rectus muscle. The lateral rectus muscles of humans and monkeys consist of distinct superior and inferior muscle fiber compartments separately innervated by either a superior or inferior abducens nerve division. This anatomic division of lateral rectus muscle into separate compartments has been correlated with multipositional magnetic resonance imaging (MRI) demonstrating differential contractility during ocular counter-rolling and vertical fusional vergence. The potential for transmission of differential forces in the 2 compartments to different insertional points on the globe has been demonstrated by biomechanical measurements showing mechanical independence of passive tensile forces in the compartments of bovine extraocular muscles and tendons, as well as actively generated forces in both the orbital and global layers, and transverse compartments, of bovine extraocular muscles. The frequent occurrence of selective atrophy of the superior compartment of the lateral rectus may be related to the observation of incomitant hypertropia in clinical unilateral lateral rectus weakness, which suggests selective compartmental paralysis or paresis.
Recently, Le and associates extended the theme of compartmentalization to the superior oblique (SO) muscle when they described in this muscle evidence for selective compartmental innervation (Le A, et al. IOVS 2014; 55:ARVO E-Abstract 2559). The trochlear nerve bifurcates external to the SO belly into medial and lateral branches that innervate nonoverlapping compartments of the muscle. The medial compartment inserts anteriorly near the globe equator and thus would have mechanical advantage favoring predominantly torsional action in central gaze, while the lateral compartment inserts posteriorly and would have mechanical advantage favoring a predominantly vertical action. The functional implications of this anatomic finding have been extended using MRI of differential compartmental function of the SO muscle during vertical fusional vergence.
In both acquired and congenital SO palsy, MRI studies have shown the affected SO belly to be atrophic or hypoplastic. Demer and associates reported marked reduction in midorbital cross section in denervated SO muscles in MRI and variable atrophy of muscle fiber following experimental intracranial trochlear neurectomy in monkeys. Atrophy of the SO belly therefore can be considered a finding sufficient to establish the diagnosis of SO palsy. We wondered if the SO impaired by a trochlear lesion might exhibit selective compartmental atrophy similar to the lateral rectus and, if so, supposed that the clinical effects of selective compartmental SO palsy might differ from complete palsy. However, no study yet has attempted to examine the shape of the atrophic SO belly using MRI.
Therefore, this study sought to investigate, using high-resolution MRI in a large number of cases, possible variations in the morphologic pattern of SO atrophy and their relationship to clinical characteristics of SO palsy.
This prospective observational study was conducted at Stein Eye Institute, a single academic medical center at the University of California, Los Angeles (UCLA). Volunteers gave written informed consent according to a protocol approved by the UCLA Institutional Review Board that conformed to the tenets of the Declaration of Helsinki. We studied a control group of 12 normal, orthotropic volunteers recruited by advertisement of mean (±standard deviation, SD) age 46.2 ± 18.3 (range, 14–61) years; there were 5 men and 7 women. Each control underwent comprehensive eye examination to verify normal acuity, ocular motility, stereoacuity, and ocular anatomy; none had undergone any prior ocular surgery except for cataract surgery. We recruited 62 subjects of mean age 38.6 ± 14.1 (range, 11–62) years from an ongoing study of strabismus who were determined to have unilateral SO palsy based on significant ipsilesional reduction of maximum SO cross section on quasi-coronal MRI imaging. Subjects were excluded if there was a history of prior strabismus surgery, or if they could not cooperate for MRI. There were 32 men and 30 women. These subjects had initially presented with complaints of strabismus or diplopia. Eighteen cases of SO palsy were considered congenital, 20 traumatic, and 24 idiopathic. None had undergone strabismus surgery. The mean symptom duration of SO palsy was 57.7 ± 43.1 (range, 2–120) months. Each subject with SO palsy underwent complete ophthalmologic examinations including measurement of binocular misalignment using prism and cover testing and the Hess screen test. Subjective cyclodeviation was measured using double Maddox rods. Ocular versions with the tested eye in adduction were graded on a clinical scale of 0 to ±4, with 0 representing normal duction, −4 representing maximal underdepression or underelevation, and +4 representing maximal overdepression or overelevation.
Magnetic Resonance Imaging
A 1.5T General Electric Signa (Milwaukee, Wisconsin, USA) scanner was used for imaging using T1 or T2 fast spin-echo pulse sequences. Both sequences provide equivalent measurements. Technical details published elsewhere include use of a surface coil array (Medical Advances, Milwaukee, Wisconsin, USA) and fiber optic fixation target. High-resolution (312 μm), quasi-coronal images of 2-mm thickness and matrix of 256 × 256 parallel to the long axis of the orbit were obtained in target-controlled central gaze, supraduction, and infraduction for each eye. Because the scanned eye was centered on an afocal, monocularly viewed target that does not induce convergence, this procedure avoided confounding by strabismus.
Image analysis was similar to published methods. Digital MRIs were quantified using ImageJ (Rasband WS. ImageJ, U.S. National Institutes of Health, Bethesda, MD; rsb.info.nih.gov/ij/ , 1997–2009, accessed February 2009). An ellipse was automatically fit to the SO cross section manually outlined in each image plane where it could be identified. The cross-sectional area and the major and minor axes of the ellipse were automatically determined.
Main outcome measures were quantitative MRI morphometry, ocular versions, and binocular alignment as measured using prism-cover testing. Statistical analyses were performed using SPSS (ver. 16.0 for Windows; SPSS Inc, Chicago, Illinois, USA). Significant effects of groups were evaluated using analysis of variance (ANOVA), with subsequent pair-wise contrasts by unpaired t tests and χ 2 tests.
All palsied SO muscles exhibited significant atrophy by MRI, but as hypothesized, there were 2 different general shapes of the atrophic SO cross sections. Figure 1 illustrates MRI of whole orbital cross sections contrasting palsied and unaffected fellow SO bellies of 2 representative subjects, with a normal control for comparison, obtained at the anteroposterior location where the normal SO exhibits maximum cross section. The atrophic right SO in Figure 1 (Top left) exhibits an elongated, narrowly oval cross section distinct from the rounder SO cross section of the unaffected left eye of the same subject ( Figure 1 , Top right) that in turn is similar to both normal SO muscles of the control subject ( Figure 1 , Bottom left and right). An ellipse fit to the SO cross section in Figure 1 , Top left would be long and narrow, which was considered to represent anisotropic atrophy. In contrast, the atrophic SO in Figure 1 , Middle left is uniformly so, exhibiting a roughly circular cross section. An ellipse fit to the SO cross section in Figure 1 , Middle left would be nearly circular, which was considered to represent isotropic atrophy.
Initial qualitative classification of palsied SO morphology was based on the foregoing 2 shapes. Twenty-six cases of SO palsy demonstrated isotropic atrophy of the SO belly, as illustrated for 12 representative cases in Figure 2 . In 36 remaining cases, the atrophic SO belly exhibited an elongate shape typical of anisotropic atrophy, as illustrated for 12 representative cases in Figure 3 in image plane 0, corresponding to the globe–optic nerve junction in central gaze. Despite individual variations in morphology and size for both normal and involved SO, in each case, the elongate-shaped atrophy of the SO belly is apparent.
The qualitative categorization of palsied SO morphology was then validated by quantitative morphometry of ellipses automatically fit to the SO cross sections. Figure 4 plots the mean length of major (Top left) and minor (Middle left) axes and the mean area (Bottom left) of the involved SO cross section in image planes throughout its anteroposterior extent in the orbit. Figure 4 , Top left demonstrates significant differences in major axes of the SO cross section among the anisotropic atrophy, isotropic atrophy, and normal control groups ( P = .001, ANOVA). This analysis confirms that in the anisotropic atrophy group, the major axis of the SO cross section was modestly but significantly subnormal from image planes 1–5 posterior to the globe–optic nerve junction ( P < .05 for all), but the major axis in the isotropic atrophy group was profoundly reduced throughout the entire length of the SO ( P < .0001 for all) compared to normal. Figure 4 , Middle left shows significantly but similarly subnormal minor axes in the isotropic and anisotropic atrophy groups ( P = .001, ANOVA) throughout the entire SO length except the image plane 12 mm anterior to the globe–optic nerve junction. Figure 4 , Bottom left demonstrates more significant reduction in SO cross-sectional area in the isotropic than anisotropic SO atrophy groups ( P = .001, ANOVA). This reduction was significant from 6 image planes posterior to 1 image plane anterior to the globe–optic nerve junction in the anisotropic atrophy group ( P < .05 for all) but significantly subnormal throughout the entire length of the SO in the isotropic atrophy group ( P < .0001 for all). Figure 4 illustrates that the major (Top right) and minor axes (Middle right) and the mean area (Bottom right) of the SO cross sections of fellow eyes in both the anisotropic and isotropic SO atrophy groups were similar to normal ( P > .2 for all).
Validity of the foregoing exploratory classification of SO cross sections was evaluated by post hoc discriminant analysis based exclusively on morphometry. We plotted the lengths of the major and minor axes of each palsied SO cross section in horizontal and vertical Cartesian coordinates, respectively. As seen in Figure 5 , these points fell into 2 readily separable groups based on major axis length: those with major axis exceeding 3.9 mm were classified as having anisotropic atrophy ( Figure 5 , circles), and those with major axis less than 3.9 mm were classified as having isotropic atrophy ( Figure 5 , squares). There was no overlap between groups, despite variation in minor axis within each group.
The maximum cross-sectional area for the palsied SO of the isotropic atrophy group was 5.81 ± 1.38 mm 2 (mean ± standard error, SE), which was less than those of the palsied SO of the anisotropic atrophy group at 8.21 ± 0.94 mm 2 ( P = .005). Both were considerably less than those of the normal SO in the normal control group (14.13 ± 0.89 mm 2 , P = .0001) and the unaffected fellow SO of the isotropic (17.01 ± 1.22 mm 2 , P = .001) and anisotropic atrophy groups (14.53 ± 0.94 mm 2 , P = .001). The isotropic atrophy group showed 66.2% reduction in maximum SO cross-sectional area, while the anisotropic atrophy group showed 43.5% reduction.
The maximum major axis for the palsied SO of the isotropic atrophy group was 3.71 ± 0.39 mm 2 , which was less than those of the palsied SO of the anisotropic atrophy group at 5.54 ± 0.15 mm 2 ( P = .0146). Both were less than those of the normal control group (5.79 ± 0.18 mm 2 , P = .001, P = .031, respectively) and the unaffected fellow SO of the isotropic (6.17 ± 0.20 mm 2 , P = .001, P = .0121, respectively) and anisotropic atrophy groups (6.36 ± 0.21 mm 2 , P = .001, P = .003, respectively).
The maximum minor axis for the palsied SO of the isotropic atrophy group was 2.11 ± 0.21 mm 2 , which was similar to those of the palsied SO of the anisotropic atrophy group at 2.05 ± 0.11 mm 2 ( P = .693). Both were about a third less than those of the normal SO in the normal control group (2.99 ± 0.12 mm 2 , P = .002, P = .001, respectively) and the unaffected fellow SO of the isotropic (3.18 ± 0.17 mm 2 , P = .001) and anisotropic atrophy groups (3.14 ± 0.16 mm 2 , P = .001).
The SO belly volume was computed in planes ranging from 12 mm anterior to 12 mm posterior to the globe–optic junction ( Figure 6 ). The mean SO volume of the anisotropic atrophy group was 144.3 ± 9.8 mm 3 , which was greater than that of the isotropic atrophy group at 97.5 ± 16.8 mm 3 ( P = .001). The SO belly volume of the anisotropic atrophy group was less than those of the normal control group (269.8 ± 12.5 mm 3 , P = .0001) and the unaffected fellow SO (261.1 ± 6.6 mm 3 , P = .001). The SO belly volume of the isotropic atrophy group was less than those of the normal control group (269.8 ± 12.5 mm 3 , P = .0001) and the unaffected fellow SO (270.3 ± 12.6 mm 3 , P = .0001). The isotropic atrophy group showed 66.3% reduction in SO volume, while the anisotropic atrophy group showed 44.5% reduction.