Eye movements in patients with superior canal dehiscence syndrome align with the abnormal canal

Subjects.

We studied 11 subjects, eight men and three women, aged between 35 and 61 years (mean 42 years). All patients were diagnosed with SCD syndrome based on the appearance of either or both sound- and pressure-induced vertigo and nystagmus, with evidence of superior canal dehiscence on a high-resolution CT scan of the temporal bones.14,15⇓ All subjects gave informed, written consent, and the experimental procedure was approved by the Joint Committee on Clinical Investigation of The Johns Hopkins University School of Medicine, Baltimore, MD.

Recording the three-dimensional angular head and eye position.

Head and eye movements were recorded in three dimensions with the magnetic search coil technique.18,19⇓ In 10 patients, we recorded the position of both eyes, and in the remaining patient, we recorded from one eye only. The calibration procedures have been described elsewhere.20,21⇓

Experimental procedure.

Subjects were seated level and upright so that their eyes were in the center of the magnetic field recording system. We maintained the subjects in this position by asking them to bite on a custom-made bite bar that was firmly attached to the chair. While in this position, we measured the subject’s orientation side-on, by recording the angle between earth-horizontal and Reid’s line (which connects the inferior orbital margin and the center of the external auditory canal). Across the group of patients, Reid’s line was pitched nose-up by 10° to 19° relative to the earth-horizontal plane. At the start of each test, we recorded eye movements in complete darkness to measure any spontaneous nystagmus.

Analysis of the axis of eye rotation.

We only analyzed nystagmus with a mean slow phase eye speed >2°/s. The calculated axis of eye velocity for nystagmus <2°/s varied considerably from beat to beat and was subject to error from system noise and fluctuations in the underlying spontaneous nystagmus. The horizontal, vertical, and torsional components of the spontaneous nystagmus, if present, were subtracted from the sound- or pressure-induced nystagmus data. The eye velocity data were then mathematically rotated about a horizontal (interaural) axis21 to simulate the subject sitting with Reid’s line earth-horizontal. The orientation of Reid’s line in each subject determined the angle through which the data were rotated. We rotated the data both to bring all subjects back to a common frame of reference and to compare directly the axis of nystagmus with the anatomic axis of each canal that is referenced to Reid’s line.22 We then calculated the median eye velocity axis for each subject by taking the median roll component, median pitch component, and median yaw component of the (normalized) eye velocity axis for the entire population of nystagmus slow phases. We then normalized this to create the median response axis, and we checked its validity in each subject by graphically superimposing it onto the eye velocity axis for the entire population of slow phases for that subject to ensure that it truly represented the nystagmus data. We then calculated the angle between the median axis for the left and right eyes, and between each median axis and the anatomic axis of the superior canal by computing the inverse cosine of the inner product of each pair of axes.

Tone bursts.

Pure tones were delivered through head phones attached to a portable audiometer (Model 120; Beltone Electronics Corp, Chicago, IL). Both the audiometer and head phones had recently been calibrated. For each subject, we tested a range of frequencies from 125 to 6 kHz at 110 dB HL (hearing level) and recorded the induced eye movements in complete darkness. The stimulus duration was typically 10 to 20 seconds and was marked by the experimenter triggering a button, and this trigger signal was sampled by the computer. At the frequency that caused maximal nystagmus, we delivered a range of stimulus intensities from 70 to 110 dB HL in complete darkness and recorded the eye movements. In 10 patients, the tones were also delivered with a light-emitting diode (LED) located at 1 m directly in front of the subject and at eye level. The LED was illuminated to determine the effect of visual suppression on the nystagmus.

Valsalva maneuver.

To test the effect of changes in intracranial pressure,23 we asked each subject to perform a Valsalva maneuver straining against a closed glottis, without pinching the nostrils. The tests were carried out in darkness with the subject restrained by the bite bar.

Testing the rotational response of each semicircular canal.

We tested the response of each of the six semicircular canals to rapid head rotations, using the head impulse test.24,25⇓ The head impulse is a low-amplitude (10° to 15°), high-acceleration (3,000 to 6,000°/s²) head rotation (delivered by the clinician) in the plane of one pair of canals. Ideally during a head impulse, the canals should generate an eye rotation equal in speed and opposite in direction to the head rotation. We defined the gain of the vestibulo-ocular reflex as the slope of the eye velocity versus head velocity curve,26 when both values were geometrically projected into the plane of the canal being tested.25

Spontaneous nystagmus.

We recorded spontaneous nystagmus in the dark, with a mean slow phase eye speed (±1 SD) of 0.8 ± 0.5°/s, which we considered to be within normal limits. Removing the spontaneous nystagmus had little effect on the trajectory of the sound- and pressure-induced nystagmus, yet we removed it from the data because we were interested in the response to sound and pressure stimuli, not the underlying baseline nystagmus. For all sound- and pressure-induced nystagmus, the mean angle (±1 SD) between the axis of eye velocity with spontaneous nystagmus included and the axis of eye velocity with spontaneous nystagmus removed was 8.6 ± 7.0° for the left eye and 7.9 ± 7.7° for the right. However there was no systematic shift of the axis in any particular direction, and the angle between the eye velocity axis and the closest anatomic canal axis was not different between the data with spontaneous nystagmus included and the data with spontaneous nystagmus removed (two-tailed t-test, p > 0.3).

Responses to sound.

During the presentation of pure tones to a maximum intensity of 110 dB HL, eight patients developed nystagmus and three did not. These three patients did, however, develop nystagmus in response to a Valsalva maneuver. Two patients had bilateral symptomatic SCD and developed sound-induced nystagmus from stimulating either ear. Figure 1 shows the eye movements from one such patient in response to a 3-kHz tone presented to the right ear and left ear. On stimulation of the right ear, the patient developed nystagmus with upward, counterclockwise slow phases, and on stimulation of the left ear, he developed nystagmus with upward, clockwise slow phases. We describe all eye movements in this study with respect to the patient. For example, a counterclockwise torsional eye movement implies rotation of the superior pole of the eye to the patient’s left. The median slow phase speed was 2.8°/s during stimulation of the right ear and 2.6°/s during stimulation of the left. The nystagmus was conjugate, and specifically the vertical position of each eye was tightly coupled (see figure 1), indicating that there was no vertical disconjugacy (or skew deviation).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Horizontal (H), vertical (V), and torsional (T) eye positions were recorded with scleral search coils and are plotted against time for the left eye (thick traces) and right eye (thin traces), in a patient with bilateral superior canal dehiscence syndrome. The thick black line (bottom) indicates the duration of a 3-kHz tone that was presented at 110 dB HL (hearing level) in darkness to the right ear (top traces) and to the left ear (bottom traces). The scale markers indicate a 5° eye rotation and a 5-second time interval. Raw data are shown, without removal of spontaneous nystagmus. Positive direction for the horizontal, vertical, and torsional axis is defined as left, down, and clockwise (rotation of the superior pole of the patient’s eye toward his right side). In response to a tone presented to the right ear, the patient developed nystagmus with upward, counterclockwise slow phases, consistent with excitation of the right superior canal. In response to a tone presented to the left ear, the patient developed nystagmus with upward, clockwise slow phases, consistent with excitation of the left superior canal. The nystagmus was sustained for the duration of the tone and was conjugate, without evidence of vertical (skew) disconjugacy. The median slow phase nystagmus speed was 3°/s.

In figure 2, we plotted the axis of eye rotation for each slow phase of nystagmus depicted in figure 1. To calculate the axis, we averaged the horizontal, vertical, and torsional components of eye velocity during each slow phase of nystagmus. The axis of eye rotation is determined by the relative magnitude of these three components. We plotted the axis of eye rotation with the anatomic axis of each canal22 (see figure 2). The boxes represent the range of superior canal orientation (±2 SD).22 In response to tones presented to the right ear, the axis of eye rotation during the slow phases of nystagmus aligned with the right superior canal and the angle between the median slow phase axis of the right eye and the presumed anatomic axis of the right superior canal (determined from the data of Blanks et al.22) was 11.8°. During the presentation of tones to the left ear, the axis of eye rotation aligned with the left superior canal, and the angle between the median slow phase axis of the left eye and the presumed anatomic axis of the left superior canal was 3.4°. The white symbols represent the axis of left eye rotation, the black symbols represent the axis of right eye rotation, and their proximity to each other confirms the conjugacy of the response (see figure 2). The mean angle (±1 SD) between the left and right eye velocity axes for all slow phases of nystagmus (see figures 1 and 2⇓) was 13.9 ± 7.3° during right ear stimulation and 7.2 ± 5.5° during left ear stimulation. Figure 2 also shows the beat-to-beat variability in the eye velocity axis, which was most prominent when the slow phase speed was <5°/s.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. The axis of slow phase eye velocity corresponding to the data shown in figure 1. The sphere represents the patient’s head, as viewed from the right side. The positive direction of the horizontal axis (H) travels upward from the top of the head, the torsional axis (T) straight ahead from the patient’s nose, and the vertical axis (which is obscured by the sphere) from the patient’s left ear. Positive rotation is defined as left (H), down (V), and clockwise (T), or rotation of the superior pole of the patient’s eye toward his right side. The anatomic axis of each of the right superior (RS), left superior (LS), right lateral (RL), and left lateral (LL) semicircular canals is shown.22 The posterior canal axes are obscured by the sphere. The box around the axis of each superior canal indicates the region (±2 SD) from the mean orientation of that axis.22 Each data point represents the mean eye velocity axis for one slow phase of nystagmus after removal of spontaneous nystagmus and rotation of the data to simulate Reid’s line being earth-horizontal. Presentation of a tone to the right ear caused nystagmus with an axis close to that of the right superior canal: left eye (white asterisks), right eye (black squares). Presentation of a tone to the left ear caused nystagmus with an axis close to that of the left superior canal: left eye (white crosses), right eye (black circles). The rotation axis of the left eye and right eye were close to each other. There was considerable beat-to-beat variability in the axis of eye rotation in this patient, largely because the nystagmus was of low velocity (slow phase eye speed <5°/s). The calculation of axis for each slow phase of nystagmus is therefore subject to the effects of system noise and to variation in the velocity of the underlying spontaneous nystagmus.

For the group of 10 affected ears in eight patients with sound-induced nystagmus, the mean slow phase eye speed was 4.6 ± 3.4°/s for the left eye and 4.9 ± 4.2°/s for the right. During stimulation of eight ears, the eye velocity axis aligned with the anatomic axis of the ipsilateral superior canal. For those responses, the mean angle (±1 SD) between the median eye velocity axis for each patient and the anatomic axis of the ipsilateral superior canal was 13.5 ± 8.1° for the left eye and 8.1 ± 4.6° for the right. During stimulation of the remaining two ears, the eye velocity axis was better aligned with the axis of the lateral canal on the affected side than the superior canal on that side. In both these patients, the dehiscence of the affected superior canal was large (5.0 and 5.5 mm). The rotational response of the affected superior canal of these two patients was markedly reduced, with a gain of 0.51 in one and 0.33 in the other (normal range 0.7 to 1.1).25 The eye movements were conjugate for the entire group of patients. The median axis of rotation calculated from eye velocity recorded from the right eye was aligned with that calculated from the left eye. The mean angle (±1 SD) between the median slow phase axis of each eye was 9.9 ± 7.2°. The eye position traces showed no consistent skew deviation across the patient group. In all but one patient, the magnitude of skew deviation, or the difference in vertical eye position, was <0.5° (<0.9 prism diopters). In the remaining patient, the magnitude of skew was 0.5° to 2.0° (0.9 to 3.5 prism diopters), and both eyes rotated in the same direction but by different amounts. In this patient with right-sided SCD syndrome, the vertical component of eye velocity was greater in the right eye, and the torsional component was greater in the left.

The minimum intensity of sound required to generate nystagmus was 100 to 110 dB HL. Of the 10 affected ears, nine responded to a relatively narrow range of frequencies, from 1 to 3 adjacent frequencies available on a standard audiometer. One patient responded to a large range of frequencies, from 125 Hz to 4 kHz. The frequency that elicited the maximum velocity of nystagmus varied from 125 Hz to 4 kHz across patients, and there was no correlation between that frequency and the size of the dehiscence, as measured on high resolution CT scans of the temporal bones (r² = 0.24, p > 0.10).

Valsalva maneuver.

Eight patients developed nystagmus immediately after a Valsalva maneuver that involved straining against a closed glottis. On release of the pressure, they all developed nystagmus consistent with stimulation of the affected superior canal ( figure 3). The mean angle (±1 SD) between the anatomic axis of the affected superior canal and the median eye velocity axis was 15.6 ± 7.7° for the left eye and 16.8 ± 8.9° for the right. The mean slow phase eye speed was 7.3 ± 4.6°/s for the left eye, and 6.0 ± 4.5°/s for the right. The eye movements were conjugate, and the mean angle (±1 SD) between the median slow phase velocity axis of each eye was 7.2 ± 4.2°. Specifically, the eye position traces showed no consistent pattern of skew deviation, and in all eight patients, the magnitude of skew deviation was <0.5°. While straining during the Valsalva maneuver itself, the patients had minimal nystagmus in the opposite direction.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. The median axis of slow phase eye velocity immediately after release of pressure from a Valsalva maneuver, when the patient was straining against a closed glottis. The sphere represents the patient’s head, as viewed from the right side, and corresponds to that shown in figure 2. Data from each patient are plotted with a unique symbol to compare the rotation axis of the left eye (white) with that of the right eye (black). (The white + symbol represents data from the patient in whom we recorded the position of the left eye only). The eye velocity axes are calculated after removal of spontaneous nystagmus and after rotation of the data to simulate Reid’s line being earth-horizontal. The axis of eye rotation corresponded to the anatomic axis of the affected superior canal in all patients, and there was no consistent disparity between the rotation axis of each eye.

The Valsalva- and tone-induced nystagmus in one patient with bilateral SCD provide a further understanding of the mechanisms underlying these responses. The high-resolution temporal bone CT in this 61-year-old woman demonstrated a large (5.5 mm) dehiscence on the left and a smaller (3.2 mm) dehiscence on the right. The corresponding rotational gains for the superior canals were 0.33 on the left and 0.68 on the right (normal range 0.7 to 1.1). Tones of 3000 Hz presented to each ear individually evoked a nystagmus with mean slow phase eye speed of 3°/s for the left ear and 11°/s for the right ear. The nystagmus was aligned with the plane of the right superior canal for tones on the right and with the plane of the left horizontal canal for tones on the left. The Valsalva-induced nystagmus had a mean slow phase eye speed of 3°/s and aligned with the plane of the left superior canal. Responses to rotations and to tones were diminished for the left superior canal, probably because of the large size of the dehiscence. The Valsalva-induced response indicates that the left superior canal was more responsive to direct changes in intracranial pressure perhaps owing to the larger area of dehiscence.

The analysis of the evoked eye movements in these patients establishes a close correspondence between the axis of eye rotation and the axis of the affected superior semicircular canal. In all the patients with Valsalva-induced nystagmus and in seven of the nine patients with tone-induced nystagmus, the eye movement axis correlated with the anatomic axis of the affected canal. The large size of the dehiscence and associated lower rotational responses from the dehiscent superior canal provide an explanation for the failure of the tone-evoked responses to align with the affected canal in two patients.

Visual suppression of nystagmus.

We tested 10 patients both in complete darkness and in the presence of a visual LED target. In five patients, the presence of an LED almost completely suppressed the nystagmus, and these patients instead had a displacement of torsional eye position away from the stimulated ear, which was sustained for the duration of the tone ( figure 4). At the onset of the tone (see figure 4 arrow, top traces), there was an upward, counterclockwise eye movement. Within 250 milliseconds, the vertical component was corrected by a voluntary saccadic eye movement back to the LED target, but the torsional deviation persisted for the duration of the tone. In darkness, the same patient developed nystagmus with horizontal, vertical, and torsional components (see figure 4, bottom traces). In another two patients, there was partial suppression of the vertical component of nystagmus in the presence of an LED, and the resulting nystagmus had a more pronounced torsional axis than did the nystagmus in darkness. In the remaining three patients, the presence of a visual target did not alter the axis of the nystagmus, which corresponded to the anatomic axis of the affected superior canal. Nystagmus in the same direction and with a similar magnitude as that recorded in darkness was also observed by the examiner when each of the patients wore Frenzel goggles.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Horizontal (H), vertical (V), and torsional (T) positions of the left eye are plotted against time, in a patient with right superior canal dehiscence syndrome. The thick black line (bottom) indicates the duration of a 500-Hz tone that was presented at 110 dB HL (hearing level) to the right ear and indicates a different time base for the two conditions (light-emitting diode [LED] on and LED off). In the top traces, the tone was presented for 13.0 seconds in the presence of a visual LED target. In the bottom traces, the tone was presented for 9.5 seconds in darkness. The scale marker indicates a 5° eye rotation. Raw data are shown, without removal of spontaneous nystagmus. Positive direction for the horizontal, vertical, and torsional axis is defined as left, down, and clockwise. In the presence of an LED, the patient developed a sustained counterclockwise torsional eye movement with very little nystagmus. There was also an initial upward eye movement at the onset of the tone (top arrow), which was corrected by a voluntary saccadic eye movement soon after. In darkness, the patient developed nystagmus with horizontal, vertical, and torsional components.

Sound-induced head movements.

We recorded the response to loud tones with the head unrestrained in four patients to assess the direction of any sound-induced head movement that occurred. Two patients developed a head movement, in addition to the nystagmus, at the onset of the tone. The magnitude was small (approximately 1°), and the direction of the head rotation was similar to the direction of the slow phase of ocular nystagmus, with an upward and a torsional component ( figure 5). The other two patients had no consistent head movement in response to sound.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5. Horizontal (H), vertical (V), and torsional (T) positions of the head and the left eye are plotted against time in a patient with left superior canal dehiscence syndrome. The thick black line (bottom) indicates the duration (3.8 seconds) of a 500-Hz tone that was presented at 110 dB HL (hearing level) to the left ear in darkness. Head position during three successive presentations of the tone is shown (top traces), and eye position is shown for only one presentation of the tone (bottom traces). The scale marker indicates a 1° head rotation and a 5° eye rotation. Positive direction for the horizontal, vertical, and torsional axis is defined as left, down, and clockwise. At the onset of the tone (arrow), the head consistently rotated upward and clockwise, in the same direction as the ocular slow phases of nystagmus. The magnitude of the rotation was approximately 1° for the head, and 5° for the eye. There were no head movements in response to tones presented to the right ear.

Rotational response of the superior canal.

We measured the response of each superior canal to rapid head rotations or head “impulses” in nine patients.25 There was a correlation between the size of the dehiscence and the functional status of the superior canal. Seven canals had a dehiscence <5.0 mm in length, and the mean gain for rotation in the plane of the affected canal was 0.84 ± 0.13, which was not different from the gain of the six intact canals (0.97 ± 0.09, p > 0.05). Five canals had a dehiscence of ≥5.0 mm, and in this group the mean rotational gain was 0.47 ± 0.13, which was lower than that of intact canals (p < 0.001).

Superior semicircular canal specificity–axis of the induced nystagmus.

Eight of the 11 patients in this study had Valsalva-induced eye movements. In each case, the axis of eye velocity aligned (within 16.8°) with the anatomic axis of the affected superior canal (see figure 3). Eye movements were induced by tones in eight patients (10 affected ears). In eight of these affected ears, the axis of eye velocity aligned with the anatomic axis of the ipsilateral superior canal. The presence of a defect in the bone overlying the superior canal creates a mobile ‘third window’ in the labyrinth and a low-impedance pathway for sound and pressure energy, akin to the round window of the cochlea.14,27⇓ The 8.1° to 16.8° disparity between the eye velocity axes and the anatomic axis of the superior canal might be owing in part to interindividual differences in anatomy. The boxes (see figures 2 and 3⇑) represent the superior canal axes based on the measured canal orientation angles (±2 SD) from 10 subjects,22 giving an angle of 13.7° between the mean axis and each corner of the box. It would be helpful to measure the canal orientation in each patient using high-resolution CT scans; however, techniques for doing this have yet to be validated.

A similar pattern of head nystagmus developed in pigeons after bilateral surgical fenestration of the superior canals.27 Presentation of a loud tone close to either ear elicited head nystagmus, with an axis of rotation parallel to that of the fenestrated canal closer to the sound source.27 The axis of the nystagmus suggests that the nystagmus was generated by the superior canal on the stimulated side. Ewald’s first law,28 a fundamental principle of vestibular physiology, predicts that when stimulated, each canal produces an eye rotation about an axis parallel to that of the canal. For example, the lateral canal produces nystagmus that is predominantly horizontal (with some torsional component), and the diagonally oriented superior and posterior canals produce a mixed vertical/torsional nystagmus. There is qualitative evidence in the monkey, cat, and rabbit that this relationship holds.17 However, there is sparse evidence that this important principle is valid in humans.29-31⇓⇓

The axis of eye rotation corresponded to the anatomic axis of the affected posterior canal in a study of five patients with benign paroxysmal positional vertigo (BPPV) from the posterior canal.29 The axis of nystagmus in one patient with a postoperative defect in the bone overlying one posterior canal was aligned, within 19°, with the axis of that canal.30 Another study of five patients each with a solitary defect in the bone overlying the lateral canal and one patient with a solitary defect in the posterior canal found that the axis of nystagmus was aligned with that of the affected canal.31 In the same study, however, the authors reported two patients each with a defect in the bone overlying both the anterior and lateral canals. In those patients, the axis of nystagmus was not aligned with the axis of either canal, nor was it intermediate between the axis of the affected canals. The authors did not measure the rotational response of each canal, and the final axis of nystagmus might have been influenced by unresponsiveness in one or more canals. That influence has been demonstrated in pigeons, where the axis of head nystagmus was altered by inactivating the stimulated canal, either by applying cocaine to the surgical opening of the canal or by destroying the sensory epithelium of that canal, without destroying the entire labyrinth.27 In two of our patients with a large dehiscence (≥5.0 mm), sound-evoked eye movements corresponded more closely to the lateral than the superior canal, and in both patients, the eye movements evoked by head rotation in the plane of the dehiscent superior canal were markedly deficient. This deficiency might have shifted the axis of nystagmus away from the superior canal.

Responses to Valsalva maneuver.

Eight patients developed nystagmus consistent with stimulation of the affected superior canal, immediately after release of pressure from a Valsalva maneuver when straining against a closed glottis. During the maneuver itself, there was low-velocity nystagmus in the opposite direction. The asymmetry in eye velocity between the per- and post-Valsalva conditions could be owing both to the more rapid pressure change on release and to the physiologic asymmetry in the response of individual semicircular canals (Ewald’s second law—excitatory responses are larger than inhibitory responses).28 Presumably this Valsalva maneuver increases intracranial pressure and exerts force directly at the site of the dehiscence, causing deflection of the cupula of the superior canal toward the ampulla, in its “off” direction. On release of the pressure, the cupula is deflected in its preferred “on” direction, causing stimulatory nystagmus.14 Both on- and off-direction responses have been found in a patient with the Tullio phenomenon7 and experimentally by Ewald when he surgically fenestrated canals in different animals and applied positive and negative pressure at the site of the opening.28

Association of the Tullio phenomenon with PLF.

Many recent clinical reports suggested an association between sound- and pressure-induced vertigo and PLF, or leakage of perilymphatic fluid from the oval or round window of the inner ear.5,32⇓ This association is problematic for the following reasons: there is controversy surrounding the diagnosis of PLF,33 the predicted axis of eye rotation in such patients is unknown, and repair of a suspected PLF does not necessarily improve the symptoms.5 The clinical situations in which the diagnosis of PLF has been most conclusively established have been as a complication after stapedectomy,34 in association with congenital abnormalities of the inner ear such as the Mondini malformation,35 and after temporal bone fractures.36 In these cases, fluctuating sensorineural hearing loss has most often been noted in association with vertigo.

In one series of 17 patients with SCD syndrome, five patients were initially misdiagnosed as having PLF and underwent middle ear surgery for that condition.15 The clinical observation of nystagmus with upward torsional slow phases in patients with symptoms of sound- or pressure-induced vertigo or oscillopsia should suggest the diagnosis of SCD.

A role for the utricle?

Despite the early experimental work demonstrating that the Tullio phenomenon could be caused by a fenestrated semicircular canal, most recent reports suggested that the utricle generates the nystagmus and vertigo in these patients.5-8⇓⇓⇓ Reports suggesting that the signs and symptoms arise from the utricle cited abnormally close proximity between the stapes footplate and utricle5,9-11⇓⇓⇓ with or without abnormal stapes hypermobility8,11⇓ as the explanation. The primary reason for implicating the utricle is the presence of a sustained torsional eye movement, or rotation about the line of sight, for the duration of the tone. The prevailing view is that the utricle produces sustained ocular torsion as part of the OTR,12,13,37,38⇓⇓⇓ whereas the semicircular canals produce nystagmus. Although vestibular nerve recordings in experimental animals demonstrated that the utricle can be stimulated by intense sounds,39-42⇓⇓⇓ we argue that in our patients, and perhaps in others previously reported as having utricle-induced Tullio phenomenon,16 the utricles need not necessarily be involved. We present three arguments in support of this hypothesis. First, the axis of the induced eye rotation in darkness corresponds to the anatomic axis of the dehiscent superior canal. Yet in the presence of a visual LED target, five patients were able to completely suppress their nystagmus and instead exhibited a torsional eye deviation that was sustained for the duration of the tone (see figure 4). This pattern of physiologic visual suppression of nystagmus was reported in a patient with the Tullio phenomenon.43 Previous studies of the Tullio phenomenon that only reported eye movements in light are therefore subject to visual suppression, leaving the impression either of a transient vertical eye movement44 or of sustained ocular torsion,5,8⇓ both of which were attributed to utricular stimulation. Second, we found no consistent vertical disconjugacy, or skew deviation, because the vertical position of each eye was aligned within 0.5° in all but one patient. Skew deviation would be expected if the eye movements originated from the utricle, based on utricular stimulation in the cat12 and the guinea pig,13 or damage to the utricle38 or its central pathways45 in humans. Third, the direction of the sound-evoked head rotation in our study, when present, corresponded to the direction of the slow phases of nystagmus, not to a lateral head tilt that would be expected from utricular stimulation.

Clinical implications.

There are four fundamental clinical messages that emerge from this work. First, the findings support the validity of Ewald’s first law in humans, because the axis of sound- or pressure-induced nystagmus matches that of the affected superior canal. It follows from this observation, made in patients with an identified dehiscence of the superior canal, that nystagmus with a trajectory matching the anatomic axis of a canal is generated by that canal. This relationship enables the origin of BPPV from the posterior or lateral canal and the SCD syndrome to be determined from an analysis of the eye movements. Conversely, if the nystagmus is purely horizontal, purely vertical, or purely torsional or is disconjugate, a lesion of the brainstem or cerebellum, rather than the labyrinth, should be suspected.46,47⇓

Second, the observation of sound- or pressure-induced eye movements in the plane of the superior canal is important in making the diagnosis of the SCD syndrome. In practice, the finding of nystagmus with upward torsional slow phases (or downward torsional quick phases), is indicative of superior canal excitation.17 CT imaging of the temporal bones, although particularly useful when surgery is contemplated, may have a relatively low specificity when images are obtained with 1.0- or 1.5-mm collimation. An artifactual area of dehiscence may be noted with these scans because of partial volume averaging. The specificity of CT scans is improved when images are obtained with 0.5-mm collimation and displayed in the plane of the superior canal.15 The finding of a low threshold for click-evoked vestibular myogenic potentials in patients with SCD syndrome also appears to be helpful in establishing the diagnosis,48,49⇓ but its specificity is unknown.

Third, when performing an examination to detect abnormalities of the labyrinth, eye movements should be observed or recorded in the absence of visual fixation to prevent visual suppression of the nystagmus. Suppression of the nystagmus through visual fixation may lead to the appearance of a sustained torsional deviation of the eyes. Such a finding could be erroneously interpreted as a sign of PLF. Visual fixation is best prevented by monitoring eye movements with infrared video equipment but is also possible using Frenzel glasses or performing ophthalmoscopy while covering the fixating eye.50 The vestibulo-ocular examination should consist of tests for eye movements evoked by tones, Valsalva maneuvers, and tragal compression. In tests for tone-evoked responses, a range of stimulus frequencies should be administered from 125 Hz to 6 kHz at an intensity of 110 dB HL.

Fourth, patients with large regions of dehiscent bone (≥5.0 mm) are likely to have deficient rotational responses from the affected superior canal, whereas those responses in patients with small regions of dehiscence are likely to be normal. We postulate that a large region of bony dehiscence allows the overlying dura of the temporal lobe to compress the membranous superior canal, impeding the flow of endolymph during head rotation.