Head Position–Dependent Changes in Ocular Torsion and Vertical Misalignment in Skew Deviation

Objectives  To investigate whether ocular torsion and vertical misalignment differ in the upright vs supine position in skew deviation and to compare these findings with those in trochlear nerve palsy.

Methods  Ten patients with skew deviation, 14 patients with unilateral peripheral trochlear nerve palsy, and 12 healthy subjects were prospectively recruited. With subjects first in the upright position and then in the supine position, ocular torsion was measured by double Maddox rods and vertical misalignment was measured by the prism and alternate cover test.

Results  In patients with skew deviation, the abnormal torsion and vertical misalignment in the upright position decreased substantially with change to the supine position, whereas in patients with trochlear nerve palsy, it changed little between positions. Torsion was decreased by 83% in patients with skew deviation, 2% in patients with trochlear nerve palsy, and 6% in healthy subjects (P < .001). Similarly, vertical misalignment was decreased by 74% in patients with skew deviation and increased by 5% in patients with trochlear nerve palsy and 6% in healthy subjects (P < .001).

Conclusions  Our findings provide the basis for additional clinical tests to support the classic 3-step test: ocular torsion and vertical misalignment that decrease from the upright position to the supine position indicate skew deviation, whereas torsion and vertical misalignment that do not change significantly between positions indicate trochlear nerve palsy.

Skew deviation is a vertical strabismus caused by supranuclear lesions and is often associated with ocular torsion and head tilt, constituting the ocular tilt reaction.^1-4 It is typically caused by damage to the brainstem tegmentum or cerebellum or by acute peripheral vestibulopathy.^4-10 Skew deviation has been attributed to an imbalance of the vestibulo-ocular reflex projections from the utricles in the inner ears to ocular motoneurons (ie, the utriculo-ocular reflex); however, direct evidence for this is lacking. The utricles lie roughly in the horizontal plane, and they normally detect static positions (tilts) of the head. We postulate that if skew deviation is caused by an imbalance of the utriculo-ocular reflex, the abnormal torsion and vertical misalignment might be head position dependent. The purposes of this study were to investigate whether ocular torsion and vertical misalignment differ in the upright vs supine position in patients with skew deviation and to compare these findings with those in patients with unilateral peripheral trochlear nerve palsy.

Ten patients with skew deviation and 14 with unilateral peripheral trochlear nerve palsy were recruited. A complete history was taken and detailed ophthalmic and neurologic examinations were performed. When indicated, appropriate tests were performed to rule out myasthenia gravis, thyroid ophthalmopathy, or other orbital diseases. Magnetic resonance images with gadolinium enhancement and slice thickness of at least 5 mm were obtained in all of the patients.

In this investigation, skew deviation was diagnosed as follows: (1) a vertical misalignment (with or without head tilt or fundus torsion), the pattern of which is inconsistent with that found in palsy of 1 or more cyclovertical muscles; (2) lesions in the posterior fossa detected by magnetic resonance imaging; and (3) neurologic symptoms and signs (although in our standard practice, neurologic findings are not necessary diagnostic criteria). Patients with acute peripheral vestibulopathy were not included because we do not encounter these patients in our neuro-ophthalmology practice.

Unilateral peripheral trochlear nerve palsy was diagnosed as follows^11-13: (1) deficient depression of the hypertropic eye in adduction; (2) incomitant hypertropia that increased with adduction of the hypertropic eye and with head tilt toward the hypertropic eye; (3) presence of objective or subjective excyclodeviation, ie, torsional diplopia; (4) absence of any neurologic symptoms and signs suggestive of lesions in the trochlear nucleus or fascicle; and (5) negative magnetic resonance imaging results. Exclusion criteria were the following: (1) history of diplopia or strabismus dating to infancy or early childhood; (2) prior strabismus surgery; or (3) history of head trauma that could lead to skew deviation, trochlear nerve palsy, or both.

Twelve healthy subjects served as control subjects (mean [SD] age, 31 [7] years; median age, 29 years; age range, 20-45 years; 8 women). The research protocol was approved by the Research Ethics Boards of The Hospital for Sick Children and University Health Network. Informed consent was obtained from all of the subjects.

Ocular torsion was quantitatively measured using double Maddox rods. While wearing a trial frame, a red Maddox rod lens was placed in front of each eye. The direction of the glass rods was aligned with the 90° marks of the trial frame. The subject was instructed to fixate on a small white penlight located 1 m away in the midsagittal plane at eye level. This distance was chosen so that a standardized target could be used in both the upright and supine positions. Through the Maddox rods, the images of the penlight appeared as 2 red horizontal streaks. If one or both horizontal streaks were perceived as slanted, the subject was instructed to rotate the lens(es) until the streak(s) became perfectly horizontal. Three trials were performed while the subject sat in an upright position (ie, subjects were not allowed to adopt their usual abnormal head posture, if present), and the trials were repeated while the subject lay in a supine position with the head maintained in the neutral position. The mean value of the final position of the lenses from the 3 trials was used for analysis. The mean within-subject variability was 0.20° in patients with skew deviation and 0.21° in patients with trochlear nerve palsy.

The magnitude of vertical strabismus was measured by the prism and alternate cover test. While sitting upright with the head erect (ie, subjects were not allowed to adopt their usual abnormal head posture, if present), the subject fixated on a single letter e of optotype size 3/32 inch (12-point font size) located 1 m away in the midsagittal plane at eye level. Prisms of increasing power were placed over the deviated eye while the cover alternated between the eyes. The highest prism strength when no refixation movement occurred was recorded. The test was repeated with the subject lying in a supine position.

To compare torsion between the upright and supine positions, a torsion change index (TCI) score, defined as (T_up − T_supine)/(T_up + T_supine), was calculated, where T_up is torsion in the upright position and T_supine is torsion in the supine position. A TCI score of 1.0 indicates complete disappearance of torsion when changing from the upright position to the supine position, whereas a TCI score of 0.0 indicates no change between positions. Similarly, a vertical change index (VCI) score, defined as (V_up − V_supine)/(V_up + V_supine), was calculated, where V_up is vertical misalignment in the upright position and V_supine is vertical misalignment in the supine position. The TCI scores were analyzed using a 3 (between) × 2 (within) nested repeated-measures analysis of variance. The between-subjects factor was the diagnosis (skew deviation, trochlear nerve palsy, and healthy) and the within-subjects factors were the hypertropic and hypotropic eyes. The VCI scores were analyzed using analysis of variance. If significant (P < .05), the specific effects of each factor were analyzed further with post hoc Tukey tests.

For patients with skew deviation (n = 10), the mean (SD) age was 25 (18) years (age range, 9-63 years) (Table 1). Seven were women. The mean (SD) duration of symptoms was 48 (42) months (range, 15-145 months). Four patients had lesions affecting both the cerebellum and medulla, 3 had primarily cerebellar lesions, and 2 had primarily brainstem lesions (Table 2). In a 10th patient with a history of acute lymphoblastic lymphoma, leptomeningeal enhancement was evident on magnetic resonance imaging. All of the 10 patients had abnormal ocular torsion in 1 or both eyes (see later). Of the 10 patients, 6 (60%) had abnormal head tilt toward the side of the hypotropic eye; the other 4 had no head tilt. Thus, the complete ocular tilt reaction was present in 60% of patients. Five patients had bilateral lesions in the posterior fossa, 3 had unilateral lesions in the medulla and/or cerebellum with contralesional hypertropia, and 2 had unilateral pontomesencephalic lesions with ipsilesional hypertropia. The vertical strabismus was comitant in 8 patients, incomitant in 1, and alternating in 1. Five patients had no spontaneous nystagmus, 1 had an upbeat nystagmus in the primary position, and 4 had gaze-evoked nystagmus. There were no difficulties measuring torsion and vertical strabismus in the upright and supine positions in any patients, with or without nystagmus.

For patients with unilateral peripheral trochlear nerve palsy (n = 14), the mean (SD) age was 36 (14) years (age range, 14-63 years) (Table 1). Six were women. The mean (SD) duration of symptoms was 34 (36) months (range, 3-145 months). Of the 14 patients, 9 (64%) exhibited an abnormal head tilt toward the hypotropic eye; the other 5 had no head tilt. No patients had any neurologic findings.

In the upright position, all of the 10 patients with skew deviation had abnormal ocular torsion (Figure 1A), defined as torsion higher than the 95% confidence interval in healthy subjects (ie, > 2.4°; see later). Abnormal torsion was found in both eyes in 3 patients (30%; patients S3, S6, and S10), in the hypotropic eye only in 5 (50%; patients S1, S2, S4, S5, and S9), and in the hypertropic eye only in 2 (20%; patients S7 and S8). When torsion was observed, an excyclodeviation was always measured in the hypotropic eye (n = 8) and an incyclodeviation in the hypertropic eye (n = 5). In the 6 patients with ocular tilt reaction, 5 had conjugate torsion of both fundi (patients S2, S3, S5, S6, and S10). The abnormal torsion either decreased substantially or disappeared completely in the supine position with the head in its neutral position (Figure 1A) in all of the patients. For example, in patient S9, the 10° of excyclodeviation of the hypotropic right eye in the upright position decreased to 2° of excyclodeviation in the supine position using the double Maddox rods test (no torsion was found in the hypertropic left eye). Consistent with this subjective finding, fundus photography (RetCam; Massie Laboratories, Inc, Pleasanton, California) of the hypotropic right eye revealed a reduction from 12° (upright) to 2° (supine) of excyclodeviation (Figure 2).

Excyclodeviation was found in the hypertropic eye in 10 of 14 patients with unilateral peripheral trochlear nerve palsy in the upright position (71%; patients T1-T9 and T11) (Figure 1B). It was abnormal (ie, > 2.4°) in 8 of these 10 patients (patients T2-T9). In contrast to patients with skew deviation, the abnormal torsion in patients with trochlear nerve palsy did not change significantly between positions. For example, in patient T4, the hypertropic right eye exhibited 10° of excyclodeviation in the upright position, which increased slightly to 11° of excyclodeviation in the supine position (no torsion was found in the hypotropic left eye).

Figure 1C shows that in healthy subjects, up to 6° of cyclodeviation was present; however, this cyclodeviation did not exhibit any significant difference between head positions. The mean (SD) torsion in healthy subjects was 1.2° (2.1°) (95% confidence interval, 0.0°-2.4°).

In patients with skew deviation, the mean (SD) TCI score was 0.85 (0.27) in the hypertropic eye and 0.81 (0.20) in the hypotropic eye (Figure 3); these scores were significantly higher than those in patients with unilateral peripheral trochlear nerve palsy (mean [SD] TCI score: hypertropic eye, 0.01 [0.09]; hypotropic eye, 0.07 [0.15]) and in healthy subjects (mean [SD] TCI score: right eye, 0.08 [0.11]; left eye, 0.01 [0.01]) (P < .001).

Similar to the patterns of torsional change, the vertical misalignment decreased substantially or disappeared completely in patients with skew deviation (Figure 4A), but it did not change significantly from the upright position to the supine position in patients with unilateral peripheral trochlear nerve palsy (Figure 4B) or in healthy subjects (Figure 4C) who had minimal vertical phoria (< 2 prism diopters) only, ie, no tropia. The mean (SD) VCI score (Figure 5) was 0.74 (0.34) in patients with skew deviation, −0.05 (0.09) in patients with unilateral peripheral trochlear nerve palsy, and −0.06 (0.14) in healthy subjects (P < .001).

The principal finding in this investigation is that the abnormal torsion and vertical misalignment in patients with skew deviation are head position dependent—they decrease substantially or disappear completely when the head changes from an upright erect position to a supine position. This head position–dependent reduction in torsion and vertical misalignment was present whether the patients had isolated cerebellar lesions, isolated brainstem lesions, or combined lesions. In contrast, in unilateral peripheral trochlear nerve palsy, there are little or no changes in torsion or vertical deviation in different head positions.

In healthy humans, static lateral head tilt activates the utricles, which results in sustained conjugate low-amplitude counterroll of the eyes and a small vertical deviation.^14-16 Head rotation toward the right shoulder is accompanied by counterrolling of both eyes combined with slight upward movement of the right eye and downward movement of the left eye.¹⁶ This reflexive vertical divergence is minimally compensatory to downward displacement of the right eye and upward displacement of the left eye relative to the earth-horizontal plane due to the head tilt.¹⁶

The ocular tilt reaction is a pathological synkinetic triad of skew deviation, ocular torsion, and head tilt. It has been attributed to lesions in the vestibular organ and its nerve as well as central connections within the brainstem or cerebellum that asymmetrically disrupt the utriculo-ocular pathway.^1-3,10 Brandt and Dieterich⁹ reviewed 56 patients with skew deviation. They found abnormal ocular torsion in all of the patients, with both eyes affected in 50%, only the hypotropic eye affected in 31%, and only the hypertropic eye affected in 19%.⁹ When abnormal torsion was present, the hypotropic eye was invariably excyclodeviated and the hypertropic eye incyclodeviated.⁹ Only 61% of patients had abnormal head tilt toward the hypotropic eye, ie, the complete ocular tilt reaction. Rostral pontomesencephalic lesions were associated with ipsilesional hypertropia and caudal pontomedullary lesions with contralesional hypertropia.⁹ The clinical characteristics of our patients with skew deviation are in close agreement with those described by Brandt and Dieterich.⁹

The hair cells on the maculae of the utricles have a wide range of polarization vectors that are sensitive to head motion in different directions, and they are oriented roughly in the horizontal plane. Thus, the utricles are responsible for detecting horizontal head displacement (ie, right-left and fore-and-aft head movements) as well as static head tilt (ie, head position with respect to gravity). Normally, balanced signals from the utriculo-ocular pathway are used to align the head's vertical axis and the eye's vertical meridian with the absolute earth-vertical plane (gravity) when stationary. Damages to the utriculo-ocular pathway lead to an erroneous internal estimate of the absolute earth-vertical plane (gravity), ie, the brain erroneously computes that the head is tilted despite the facts that the head is in an upright erect position and that the utricles lie in the horizontal plane. The triad of head tilt, skew deviation, and abnormal torsion seen in ocular tilt reaction represents a righting response, the goal of which is to realign the vertical axes of both the head and the eyes to the internal estimate, albeit erroneous, of the absolute earth-vertical plane.⁴

It is difficult to precisely predict the effects of head orientation on torsional and vertical eye positions in a diseased state and whether these effects are dependent on the location of lesions for several reasons. First, the morphological arrangement of maculae in the utricles is complex—the axes of polarity of the hair cells are multidirectional and they reverse direction across a curvilinear midline landmark in the maculae, called the striola.^17,18 Thus, signals of opposing directions of stimulation can be elicited from either side of the striola (lateral or medial) in each utricle and from either labyrinth (right or left). Second, the exact anatomy of the utriculo-ocular pathway, by which signals from the utricles reach the ocular motoneurons, is not well understood. Although the utriculo-ocular pathway has traditionally been thought to be disynaptic,¹⁹ there is growing evidence that more complex polysynaptic pathways involving an extensive network within the cerebellum may play a more important role.²⁰ We postulate that when changing from an upright position to a supine position, the orientation of the utricles changes from the earth-horizontal plane to the earth-vertical plane. This new orientation of the utricle with respect to the absolute earth-vertical plane (gravity) leads to a saturation or reduction in the overall activities of the utriculo-ocular reflex, such that any asymmetry of the reflex (as in skew deviation) is minimized. This overall saturation or reduction in activities and asymmetry of the reflex in turn lead to a reduction in torsion and vertical misalignment in skew deviation. Although the exact mechanism remains to be elucidated, our findings that torsion and vertical misalignment decrease when changing from an upright position to a supine position in patients with skew deviation are consistent with the currently accepted role of the utricles in detecting changes in head orientation and provide support that skew deviation is caused by disruption of the utriculo-ocular reflex.

In isolated unilateral peripheral trochlear nerve palsy, the utriculo-ocular pathway remains intact. Thus, the magnitude of vertical deviation and excyclodeviation in the hypertropic eye remains the same in different head positions. The contralateral head tilt commonly seen in unilateral trochlear nerve palsy is a compensatory mechanism that exploits the normal intact utriculo-ocular reflex to minimize the magnitude of vertical deviation and diplopia. Sydnor et al²¹ found that the head tilt in unilateral trochlear nerve palsy disappears in the supine position. Because there is no change in utricular input with roll tilts when supine, the utriculo-ocular reflex is no longer effective in reducing the vertical deviation in unilateral trochlear nerve palsy in the supine position; thus, the head tilt disappears.²¹

Trochlear nerve palsy is typically diagnosed with the clinical 3-step test.¹³ In contrast, the vertical misalignment in skew deviation does not follow any set patterns; it may be comitant or incomitant, or it may even be alternating with right hypertropia on right gaze and left hypertropia on left gaze.^7,22,23 Some skew deviations are known to mimic trochlear nerve palsy during the 3-step test.^24,25 Conversely, trochlear nerve palsy with spread of comitance may simulate a comitant skew deviation because in both conditions the head is usually tilted toward the side of the hypotropic eye. Because both skew deviation and trochlear nerve palsy may result from brain trauma or from lesions in the posterior fossa,²⁶ differentiating these 2 conditions can be challenging. In this situation, indirect ophthalmoscopy is useful—the fundus is excyclodeviated in the hypertropic eye in patients with trochlear nerve palsy, but it is incyclodeviated in the hypertropic eye in patients with skew deviation (if torsion is present). In addition, the fundus of the hypotropic eye in patients with skew deviation may be excyclodeviated.^9,27,28

The double Maddox rods test as well as the prism and alternate cover test are simple and quick. They also have the advantage of not requiring pupillary dilation or indirect ophthalmoscopy, which may not be readily available or feasible for nonophthalmologists, including neurologists and orthoptists. Although the magnitude (but not direction) of torsion estimated by double Maddox rods (subjective torsion) may differ from the measurement by fundus photography (objective torsion),²⁹ our results provide the basis of additional tests that could be used clinically to differentiate skew deviation from trochlear nerve palsy: ocular torsion and vertical misalignment that decrease from the upright position to the supine position indicate skew deviation, whereas torsion and vertical misalignment that do not change significantly between the upright and supine positions indicate trochlear nerve palsy. Head position–dependent changes in torsion and vertical misalignment suggest skew deviation and warrant investigation for a lesion in the posterior fossa as the cause of vertical diplopia.

Correspondence: Agnes M. F. Wong, MD, PhD, FRCSC, Department of Ophthalmology and Vision Sciences, The Hospital for Sick Children, 555 University Ave, Toronto, ON M5G 1X8, Canada (agnes.wong@utoronto.ca).

Submitted for Publication: November 8, 2007; final revision received January 10, 2008; accepted January 10, 2008.

Financial Disclosure: None reported.

Funding/Support: This work was supported by the Department of Ophthalmology and Vision Science, The Hospital for Sick Children (Drs Buncic and Wong), a New Investigator Award (MSH 55058) (Dr Wong) and grants MOP 67104 and MOP 57853 (Dr Wong) from the Canadian Institutes of Health Research, and the University Health Network Ophthalmology Practice Plan (Dr Wong).