HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 92/4/2333    most recent 00174.2004v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (6) Citing Articles via Google Scholar Google Scholar Articles by Peterka, R. J. Articles by Merfeld, D. M. Search for Related Content PubMed PubMed Citation Articles by Peterka, R. J. Articles by Merfeld, D. M.

Origin of Orientation-Dependent Asymmetries in Vestibulo-Ocular Reflexes Evoked by Caloric Stimulation

¹Neurological Sciences Institute, Oregon Health & Science University, Portland, Oregon 97006; and ²Jenks Vestibular Physiology Laboratory, Massachusetts Eye and Ear Infirmary, Department of Otology and Laryngology, Harvard Medical School, Boston, Massachusetts 02114

Submitted 24 February 2004; accepted in final form 25 May 2004

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
A caloric stimulus evokes primarily a horizontal vestibulo-ocular reflex (VOR) when subjects are in a supine or prone orientation with the horizontal semicircular canal plane oriented vertically. In both monkeys and humans, the magnitude of VOR eye movements is greater in the supine than in the prone orientation, indicating that some factor or factors, other than the conventionally accepted convective stimulation of the horizontal canals, contributes to the generation of the VOR. We used long-duration caloric irrigations and mathematical models of canal-otolith interactions to investigate factors contributing to the prone/supine asymmetry. Binaural caloric irrigations were applied for 7.5 or 9.5 min with subjects in a null orientation with horizontal canals in the earth-horizontal plane (control trial), or with the subject's pitch orientation periodically changing between null, supine, and prone positions with each orientation held for 30 s (caloric step trial). The control trial responses identified a small response attributable to a direct thermal effect on vestibular afferent activity that accounted for only 15% of the observed prone/supine asymmetry. We show that the gravito-inertial force resolution hypothesis for sensory integration of canal and otolith information predicts that the central processing of canal and otolith information produces an internal estimate of motion that includes both a rotational motion component and a linear acceleration component. These components evoke a horizontal angular VOR and linear VOR, which combine additively in the supine orientation, but subtract in the prone orientation, thus accounting for the majority of the observed prone/supine asymmetry.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
It has long been known that, in humans, horizontal vestibulo-ocular reflex (VOR) responses to caloric irrigations are larger with the subject lying in a supine position than in a prone position (Coats and Smith 1967). A similar supine greater than prone asymmetry has been demonstrated in squirrel monkeys (Minor and Goldberg 1990; Paige 1985). Although an opposite prone greater than supine asymmetry has been reported in rhesus (Böhmer et al. 1992, 1996) and cynomolgus monkeys (Arai et al. 2002), this opposite asymmetry is attributable to very strong cold water caloric irrigations that were sufficient to drive the horizontal canal afferent activity in the irrigated ear to an inhibitory cutoff in the supine orientation. Böhmer et al. (1992) demonstrated a supine greater than prone asymmetry similar to the asymmetry observed in humans and squirrel monkeys when less intense irrigations were used.

The supine greater than prone asymmetry in eye movement responses has often been attributed to a direct thermal effect on afferent nerve activity, which increases neural discharge for warm caloric irrigation and decreases neural discharge for cold caloric irrigation, thus adding to or subtracting from the primary convective effect resulting from changes in endolymph density (Coats and Smith 1967; Hood 1989). However, two studies in squirrel monkeys demonstrated that a direct thermal effect could not account for the entire prone/supine asymmetry (Minor and Goldberg 1990; Paige 1985). Both studies postulated the existence of another factor, dependent on head position and affecting both the convective and thermal components of the afferent neural discharge.

Paige (1985) suggested that this factor may represent a positional (possibly otolith) dependent modification of the horizontal canal VOR, or might reflect a stimulus artifact. Minor and Goldberg (1990) also proposed the existence of a position-dependent factor that they postulated was likely attributable to an otolith-mediated modulation of the horizontal VOR time constant. Specifically, they observed that the monkey horizontal VOR time constant was shorter during yaw rotation about an earth-vertical axis when the whole body was pitched 45° nose down (ND) compared with 45° nose up (NU). Because the VOR acceleration gain is related to the product of VOR velocity gain and the VOR time constant, the short ND VOR time constant should produce a lower ND VOR acceleration gain if the VOR velocity gain remains constant. Therefore because the conventional caloric stimulus is effectively a very low frequency stimulus, where VOR responses are expected to be proportional to rotational acceleration (Wilson and Jones 1979), the NU/ND caloric response asymmetry would be expected to correspond to the predicted VOR time constant asymmetry, and thus to a VOR acceleration gain asymmetry.

However, differences in horizontal VOR time constants, derived from pre- and postrotatory constant velocity rotations, appear small in humans for vertical axis rotations with NU and ND head tilts (Fetter et al. 1986). Therefore as pointed out by Minor and Goldberg (1990), a position-dependent modulation of the horizontal VOR time constant could make only a limited contribution to the prone/supine caloric asymmetry observed in humans. Indeed, Minor and Goldberg (1990) also pointed out that the prone/supine caloric asymmetry reported in humans (Clarke et al. 1988; Coats and Smith 1967) is smaller than that in the squirrel monkey, and suggest that the caloric response in humans may not include a position-dependent component. That is, they suggested that the prone/supine caloric asymmetry in humans might be entirely attributable to a direct thermal effect, as originally hypothesized by Coats and Smith (1967). However, more recent results do show a consistent and significant difference between horizontal VOR time constants derived from postrotatory tilts into supine and prone orientations (Zupan et al. 2000), suggesting that it may be premature to dismiss this potential contribution to the prone/supine caloric asymmetry.

Alternatively, recent experimental and modeling results (Merfeld et al. 1999; Zupan and Merfeld 2003; Zupan et al. 2000, 2002) offer a different explanation for the existence of an otolith-mediated prone/supine caloric asymmetry based on the gravito-inertial force (GIF) resolution hypothesis. Specifically, the GIF resolution hypothesis predicts that caloric stimulation of the horizontal canals will evoke both an angular VOR and a linear VOR response. The linear VOR response arises from the central processing of canal and otolith information that attempts to resolve the apparent canal-otolith sensory conflict caused by an angular rotation cue from the horizontal canal that is not accompanied by a corresponding change in otolith afferent activity signaling a head tilt with respect to gravity. (More detailed explanation is provided in the DISCUSSION.) The prone/supine caloric asymmetry is consistent with the GIF resolution hypothesis because this hypothesis predicts that the linear VOR component always adds to the angular VOR magnitude in the supine position, and always subtracts from the angular VOR magnitude in the prone position.

A final possible factor, which has not been previously considered as a possible explanation for the prone/supine asymmetry, is a difference in vertical eye position during prone/supine caloric irrigation. Using passive rotation testing, Fetter et al. (1986) showed that horizontal eye velocity is attenuated by the cosine of the angle made between the optical axis and the plane of head rotation. A prone/supine caloric response asymmetry could occur if there were systematic differences in the vertical orientation of the eyes in their orbits in prone versus supine head orientations.

To identify the origin and the magnitude of the factors contributing to the prone/supine asymmetry in humans, we conducted long-duration caloric irrigation experiments with rapid changes in whole body position into supine and prone orientations. This type of caloric test protocol is referred to as a caloric step stimulus (Bock et al. 1979; Formby and Robinson 2000). A constant thermal gradient is assumed to be established across the vestibular labyrinth after a 2- to 3-min, constant-temperature caloric irrigation (Formby and Robinson 2000). Once a constant gradient is established, a step change of position from a "null" orientation (horizontal canal plane perpendicular to earth-vertical) into a supine or prone orientation induces a step change in convective force acting on the cupula of the horizontal canal. The simple torsion-pendulum model of the canal indicates that this step change in force causes a deflection of the cupula that follows second-order dynamics (Steinhausen 1933; Wilson and Jones 1979), but with the low-frequency dynamics determined by a single dominant time constant. The cupula deflection excites or inhibits canal afferents and generates VOR eye movements with an exponential rise in slow-phase velocity (Formby and Robinson 2000). Additional peripheral [e.g., adaptation (Goldberg and Fernandez 1971)], central [e.g., velocity storage (Merfeld et al. 1993; Raphan et al. 1979; Robinson 1977)], and multisensory interaction (Zupan et al. 2000) factors are expected to contribute to the dynamic VOR response. If head orientation is maintained in the null orientation throughout a long-duration caloric irrigation, then VOR eye movements recorded after 2–3 min should be indicative of the caloric response attributable to the direct thermal effect.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Subjects

Twelve healthy subjects with no history of peripheral or central vestibular disorders participated in this study (8 males and 4 females, age 23–49 yr). Subjects gave their informed consent before being tested using a protocol approved by the Institutional Review Board at Oregon Health & Science University and in accordance with the 1964 Helsinki Declaration. Before taking part in the experiments, subjects were informed of the high likelihood that these tests would induce motion sickness. Subjects were asked questions related to their sensitivity to motion sickness during various activities so they could judge their willingness to participate.

Caloric stimulus

Two Brookler-Grams (Grams Medical Products, Costa Mesa, CA) closed-loop caloric irrigators were modified to provide constant-temperature irrigations for 10 min. The modifications included the addition of external reservoirs with circulating heaters and cooling fans for the pump and pump motor. Calibration runs with monitoring of irrigator tip temperature showed that the temperature reached a steady state in about 1 min and was maintained ±0.2°C for 10 min.

Irrigations were applied binaurally for the entire duration of each trial with one ear maintained at a temperature 4°C above normal body temperature and the other ear 4°C below body temperature. Left ear warm (41°C) and right ear cool (33°C) irrigations (LW-RC) were used in 11 subjects. Left ear cool and right ear warm irrigations (LC-RW) were also performed in 3 of these 11 subjects and in one additional subject. The 3 subjects tested with both irrigation protocols underwent the second irrigation protocol (LW-RC or LC-RW) 6 to 31 days after the first test session.

Binaural irrigations were chosen for 3 reasons. First, binaural irrigations provided a more natural push-pull activation of afferents from the horizontal semicircular canals. This avoided a potential sensory conflict between rotational information from the 2 horizontal canals that may have interfered with the interactions between canal and otolith inputs. Second, preliminary studies showed that monaural irrigations resulted in asymmetric dynamic responses that were related to excitation versus inhibition of the irrigated ear and not related to prone versus supine orientations. Specifically, the VOR time constant was always shorter when the irrigation evoked inhibitory canal activation (by convective forces) versus excitatory activation. This effect was independent of supine versus prone orientation and of warm versus cold irrigation. Such a time constant asymmetry is consistent with previously observed time constant asymmetry in canal afferent responses in the squirrel monkey (Goldberg and Fernandez 1971). Third, studies of peripheral afferent physiology in the squirrel monkey (Goldberg and Fernandez 1971) also indicate that the sensitivity of canal afferents to inhibitory accelerations is less than that to excitatory accelerations. Use of a binaural caloric stimulus minimized the influence of afferent sensitivity asymmetry as a factor contributing to the observed prone/supine asymmetry in the caloric-evoked VOR.

Pitch-tilt stimulus

The subject was seated and restrained in a kneeling position on a race car-type seat mounted in a hydraulically powered rotation device (2,300 Nm position servo-controlled actuator) with the earth-horizontal pitch-tilt rotation axis aligned with the subject's interaural axis. The subject's head was secured using a bite-bar assembly attached to the motion device and an adjustable foam-lined head restraint. A mold of each subject's teeth was formed on the bite bar using a dental-impression compound (3M Express, 3M Dental Products, St Paul, MN).

With the device's pitch orientation at 0°, the subject's head was oriented to place Reid's plane earth-horizontal [Reid's plane is defined by the upper margins of the external auditory meati and the infraorbital margin (Haslwanter et al. 1994)]. Standard conventions were followed to define the positive directions of eye and head rotations (Paige and Tomko 1991a).

Two types of trials were used: control trials and step trials. For the control trials, the subject's whole body was quickly tilted +20° (head oriented 20° ND), thus placing the horizontal semicircular canals in the earth-horizontal plane (Curthoys et al. 1977) and in the null orientation that provides no convective stimulation of the horizontal canals. Caloric irrigations were initiated at the onset of the tilt and maintained for 7.5 min with the subject remaining in a 20° ND orientation.

For the caloric step trials, the subject's whole body was tilted +20° (ND), as the caloric irrigations were initiated, and remained in this null orientation for 120 s to allow stabilization of the thermal gradient in the ears (Formby and Robinson 2000). The subject was then sequentially pitched and held for 30 s in a series of orientations that were pitched ±90° from the original +20° orientation (i.e., +110° for the prone position and –70° for the supine position). The sequence also included 30-s null-orientation intervals between prone and supine orientations. Pitch position changes were accomplished in 1.5 s. Caloric irrigations were started at the onset of the initial 20° ND tilt and maintained for 9.5 min while the subject changed pitch position. Two orientation sequences were used: 1) prone, null, supine, null (repeated 3 times total); and 2) supine, null, prone, null (repeated 3 times total). The final null orientation was maintained for 90 s. (See Fig. 1 for schematic representations of each of the stimuli.) The prone and supine orientations were chosen to provide maximal excitatory or inhibitory convective stimulation to the horizontal semicircular canals. The null orientation was selected to provide zero convective stimulation.

View larger version (33K): [in this window] [in a new window]   FIG. 1. Horizontal slow-phase eye velocity (SPV) recorded in one subject who underwent a full test session for 2 irrigation protocols. Positive eye velocities correspond to slow-phase eye movements to the left. This subject first underwent LC-RW irrigations (A–C), and then underwent LW-RC irrigations 6 days later (D–F). Control trials (C and F) were always performed between the 2 step trials. Step trials started either with a pitch into the supine orientation (A and D) or with a pitch into the prone orientation (B and E).

Eye movement recording and analysis

Binocular eye movements were recorded using infrared illumination for small video cameras (Machine Vision Hyper CCD Cameras, model CV-36SH) mounted on the bite-bar assembly. Reid's plane was the horizontal reference plane for the eye movement coordinate system. Details of the eye movement recording system, image analysis, and calibration procedures are described in Zupan et al. (2000).

Data analysis was performed for the left eye. Horizontal and vertical eye position data were digitally filtered and differentiated to yield horizontal and vertical eye velocity. Fast phases were automatically removed from the eye velocity data using a computer algorithm based on peak acceleration detection, with manual editing by experienced personnel. Fast phase removal yielded slow-phase eye velocity (SPV). Positive horizontal and vertical SPV indicated eye movements with slow phases directed to the left and downward, respectively.

To quantify responses to each caloric step trial, individual horizontal VOR time constants and response amplitudes were determined using least-squares fitting techniques. For tilt from null to supine (or prone) orientation, the time interval for the fit was defined as the onset of tilt movement to the beginning of a decrease in horizontal slow-phase velocity. For most cases, this corresponded to the onset of the tilt back to upright; for a few cases, however, the SPV started to decrease before the onset of the tilt to upright, and therefore the fitting intervals in these cases were shorter. After subtraction of the SPV present at the onset of the tilt, the slow-phase velocity was fitted over this time interval with a function of the form: A(1 – e^–t/TC), where t is the time variable, TC is the VOR time constant, and A is the amplitude of the exponential fit.

For tilt from supine (or prone) to null orientation, the time interval for the fit was from the maximum horizontal SPV after tilt onset (usually within 2 s of the tilt onset) to the beginning of the next tilt (or for a time interval of 31 s if the last tilt of the trial). The slow-phase velocity was fitted with a function of the form: B + Ae^–t/TC, where t is the time variable, TC is the VOR time constant, and A and B define the amplitude and offset of the exponential fit, respectively.

An alternative model-based curve fit was also performed on the mean SPV obtained across subjects (Fig. 8A). The model was a simplified representation of the GIF resolution hypothesis that included an angular VOR component represented by a response amplitude factor, a VOR time constant, and a VOR adaptation time constant. The position-dependent linear component included a separate amplitude factor and a time constant. The fit was performed by simulating the SPV response to the caloric step stimulus (using Matlab Simulink, The MathWorks, Natick, MA) and using the Matlab's optimization fit routine "fmins" to iteratively adjust the model parameters to minimize the mean-squared error between the model response and the mean SPV data.

View larger version (32K): [in this window] [in a new window]   FIG. 8. Simplified GIF resolution model and model prediction for a LW-RC caloric step irrigation. A: model represents angular VOR (AVOR) dynamics with a first-order low-pass filter and a VOR adaptation operator and linear VOR (LVOR) dynamics with a low-pass filter. Pitch orientation dependency of the LVOR is represented by the "Tilt Switch." LVOR and AVOR components sum to produce the final horizontal VOR slow-phase velocity. B: a least-squared-error fit of the model is compared with the mean (n = 8) experimental SPV response to the LW-RC caloric step stimulus for the last 2 steady-state cycles of pitch orientation changes. Curve fit procedure allowed the angular VOR time constant (AV) to assume different values for the null-to-prone, null-to-supine, prone-to-null, and supine-to-null portions of the stimulus. All other model parameters were not permitted to change value as a function of pitch orientation.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

Examples of control and caloric step responses

Horizontal slow-phase eye velocities recorded in one subject who performed all trials for both LW-RC and LC-RW caloric irrigation protocols are shown in Fig. 1. For both LW-RC and LC-RW irrigations, eye movement responses in the supine orientations (–70°) were consistently stronger than responses in the prone orientations (+110°). Control trials produced small eye movements that were in opposite directions for LW-RC and LC-RW irrigations. These responses are representative of responses obtained from all subjects for both control and step trials.

Control trials

Data for control trials were available in 7 subjects for LW-RC irrigations and in 3 subjects for LC-RW irrigations. Horizontal SPV showed a transient time course in all subjects for the first 150–200 s, followed by a low-amplitude, steady-state SPV throughout the remaining 200–250 s of the trial (Fig. 2A). Slow-phase velocities were of similar amplitude but opposite in direction for LW-RC and LC-RW irrigations (Fig. 2B).

View larger version (34K): [in this window] [in a new window]   FIG. 2. Horizontal SPV (mean ± 1SD) evoked during control trials. A: mean eye velocity obtained in the 7 subjects tested with LW-RC irrigations. B: mean eye velocity obtained in the 3 subjects tested with both LW-RC and LC-RW irrigations.

The steady-state component of the responses measured at the end of the control trials was in the direction expected from a direct thermal effect on all but one trial. The steady-state SPV was to the right for 6 of the 7 subjects tested with LW-RC irrigations. The amplitude of the steady-state component ranged from –2.5 to 0.5°/s (mean of –1.3°/s, 0.9 SD, n = 7), and the mean was significantly different from zero (t-test, P < 0.01). As expected for a direct thermal effect for LC-RW irrigations, steady-state slow-phase velocity was to the left for the 3 subjects tested (range: 1.0 to 1.8°/s, mean of 1.5°/s, n = 3; mean significantly different from zero, P < 0.05). These same 3 subjects had steady-state SPV of comparable magnitude, but opposite direction during LW-RC irrigations (range: –1.4 to –1.8°/s, mean –1.6°/s; mean significantly different from zero, P < 0.05). The overall results from these control trials indicate that the direct thermal effect on afferent nerve activity and subsequent eye movements was small under our caloric irrigation conditions.

Caloric step trials

HORIZONTAL SPV. Figures 3 and 4 show normalized mean horizontal SPV responses to LW-RC and LC-RW caloric step trials, respectively. SPV data from each subject were normalized by dividing by the peak SPV magnitude before averaging so that the dynamic characteristics of the mean response were not dominated by the subjects with the largest responses. Small transient responses in horizontal SPV were observed during the first 120 s of caloric step trials (Figs. 3 and 4), consistent with those seen in control trials (Fig. 2). The first pitch into the supine or prone orientation produced an exponential rise in SPV in a direction consistent with convective canal stimulation (i.e., SPV to the left for prone head orientation and to the right for supine head orientation with LW-RC irrigation; Fig. 3). After the return to null orientation, the SPV exponentially declined toward a value close to zero. The peak amplitudes of SPV responses for the first and second pitches to prone/supine orientations were smaller than those in the subsequent 4 prone/supine orientations (Figs. 3 and 4). The smaller responses for the first and second pitches may be attributed to the fact that a steady-state temperature gradient was not established until about 200 s, as indicated by the control trial responses (Fig. 2). Therefore detailed data analysis was performed only for the last 4 pitch orientations of a step trial.

View larger version (23K): [in this window] [in a new window]   FIG. 3. Normalized horizontal SPV (mean ± 1SD) evoked during caloric step trials with LW-RC irrigations. Data normalization before averaging consisted of dividing the SPV signal obtained for each trial by its maximum value during that trial. A: data obtained in the 8 subjects tested with a step trial starting with a pitch-to-prone orientation. B: data obtained in the 7 subjects tested with a step trial starting with a pitch-to-supine orientation.

View larger version (24K): [in this window] [in a new window]   FIG. 4. Normalized horizontal SPV (mean ± 1SD) evoked during caloric step trials with LC-RW irrigations (n = 3). A: data obtained with a step trial starting with a pitch-to-prone orientation. B: data obtained with a step trial starting with a pitch-to-supine orientation.

Individual and average responses to step trials showed a large prone/supine asymmetry in SPV (Figs. 3 and 4, Table 1). The magnitude of SPV amplitude (average SPV over the last 15 s of each pitch orientation) was on average about 40% smaller during the prone orientation than during the supine orientation for both LW-RC and LC-RW irrigations (Table 1). For the 9 subjects with data from LW-RC irrigations, the ratio of SPV during prone orientation to SPV during supine orientation varied from 0.4 to 0.78, and there was a significant difference in response magnitude between supine and prone orientations (paired t-test, P < 0.001). For the 3 of the 9 subjects who were also tested with LC-RW irrigations, the ratio varied from 0.32 to 0.95.

Vertical eye position, measured for each subject over the last 15 s of each tilt, varied from –14.9 to 10.6° during prone orientation (mean: –0.13°, SD: 7.0° for LW-RC irrigations and mean: –0.74°, SD: 7.0° for LC-RW irrigations) and from –18.0 to 16.9° during supine orientation (mean: –1.6°, SD: 9.4° for LW-RC irrigations and mean: 1.18°, SD: 8.1° for LC-RW irrigations). The corrected horizontal SPVs are shown in Table 1 (numbers in parentheses). Because the corrected horizontal SPVs differ only slightly from the uncorrected horizontal SPVs, we conclude that vertical eye position did not account for the prone/supine asymmetry in horizontal SPV.

VERTICAL SPV. Nine of the 10 subjects who underwent caloric step trials showed some vertical nystagmus in the null orientation. The direction of the nystagmus was consistent within subjects across trials (including trials for LW-RC and LC-RW irrigations, which were performed on different days). Seven subjects had upbeat nystagmus (mean SPV: 1.8°/s, range: 0.6–3.9°/s), and 2 had downbeat nystagmus (–0.3 and–0.6°/s). The magnitude of this nystagmus is consistent with the magnitude of vertical positional nystagmus previously reported without caloric stimulation (Bisdorff et al. 2000).

In the supine and prone orientations, vertical SPV was considerably smaller than horizontal SPV. For LW-RC irrigations, the mean vertical SPV was 3.0°/s (SD 4.9°/s, n = 9) for the prone orientation and 0.6°/s (SD 7.2°/s, n = 9) for the supine orientation. For LC-RW irrigations, the mean vertical SPV was 1.1°/s (SD 2.4°/s, n = 3) for prone orientation and 11.1°/s (SD 11.2°/s, n = 3) for supine orientation.¹ Vertical SPVs were variable in amplitude and direction between subjects (Fig. 5) but were consistent within each trial for each subject. Some subjects had nystagmus beating in the same direction for both supine and prone positions, whereas others had nystagmus beating in the opposite direction for supine and prone positions (Fig. 5).

View larger version (19K): [in this window] [in a new window]   FIG. 5. Vertical SPVs measured in prone, null, and supine orientations for LW-RC irrigations (9 subjects). Vertical SPV was averaged for the whole duration of a tilt (i.e., over a 30-s interval) because some subjects had increasing or decreasing vertical SPV during the whole duration of a tilt, whereas others had mostly constant vertical SPV.

The thermal component to the eye movement response was estimated using 2 techniques. First, a direct measure of the thermal component was obtained from the control-trial steady-state responses. The second, indirect estimate of the direct thermal effect assumed that 1) only thermal and convective components contribute to the VOR and 2) thermal and convective components combine additively. If these assumptions are true then: VOR_supine = VOR_convective + VOR_thermal and VOR_prone = VOR_convective – VOR_thermal, and the thermalcomponent of the VOR response can be calculated as:VOR_thermal = (VOR_supine – VOR_prone)/2. This calculation gave a mean eye velocity of –9.9°/s for LW-RC irrigations (SD: 5.2°/s, n = 9), and a mean eye velocity of 7.4°/s for LC-RW irrigations (SD: 5.8°/s, n = 3). These values are about 6 times larger than the estimate of the direct thermal contribution obtained from the steady-state component of SPV measured during control trials (about 1.5°/s). Measures of the direct thermal effect from both the control trials and step trials were available from 7 subjects. There was a significant difference between these 2 measures of the direct thermal effect (paired t-test, P < 0.01). The discrepancy between measures of the direct thermal effect from control and caloric step trials indicates that another factor, in addition to a direct thermal effect, must be contributing to the large prone/supine asymmetry observed during caloric step trials.

Horizontal eye movements and motion sickness

Five of the 11 subjects tested with LW-RC irrigation and one of the 4 subjects tested with LC-RW irrigation were not able to complete a full test session because of motion sickness. Although not part of the experimental design, the results from these 6 subjects were compared with the results from the 6 subjects who were able to complete a full test session to determine whether test results revealed systematic differences between responses from more-susceptible and less-susceptible motion-sick subjects.

The mean horizontal SPV measured over the last 15 s of supine and prone tilt were significantly larger in the 6 motion-sick subjects (mean: 71°/s, SD: 11°/s for supine orientation; mean: 43°/s, SD: 10°/s for prone orientation) than in the other 6 subjects (mean: 42°/s, SD: 16°/s for supine orientation; mean 24°/s, SD: 12°/s for prone orientation) (t-test, P < 0.01 for supine orientation; P < 0.02 for prone orientation). There were no significant differences between the 2 subject groups for the ratio (prone SPV/supine SPV) or for the difference (supine SPV – prone SPV). It is not surprising that subjects with a larger response amplitude would be more likely to become motion sick because one would expect a larger sensory conflict with a larger rotational response.

Time constants for tilts from null-to-prone/supine orientation and from prone/supine-to-null orientation were not significantly different between the motion-sick subjects and the subjects able to complete full testing. However, as an anecdotal observation, one of the motion-sick subjects had a response that differed noticeably from the average results in other subjects. This subject was distinguished from other subjects in having a long VOR time constant, particularly for the prone-to-null tilts (24.9 s) and supine-to-null tilts (18.7 s). However, this subject also had a large response amplitude in addition to a long time constant. Therefore we cannot be certain that the long time constant contributed to the motion-sickness susceptibility of this subject.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
For both LW-RC and LC-RW irrigations, the magnitude of the horizontal SPV demonstrated a clear dependency on head orientation with a larger SPV magnitude for the supine orientation. Horizontal SPV magnitude in the prone orientation was 41% smaller than that in the supine orientation during LW-RC irrigations and 39% smaller during LC-RW irrigations (Table 1). Our results are in agreement with previous human studies on the effect of body position on caloric responses (Clarke et al. 1988; Coats and Smith 1967). Using a monaural caloric stimulus (40-s irrigation), Coats and Smith (1967) found that horizontal SPV magnitude was 36 and 48% smaller during prone orientation than during supine orientation for cold and warm irrigations, respectively (Table 1 of Coats and Smith 1967). Using binaural continuous irrigations, Clarke et al. (1988) found SPV in the prone orientation that was about 33% smaller than the responses in the supine orientation (based on peak slow-phase velocity estimated from Fig. 2 in Clarke et al. 1988). These results in humans are similar to the result seen in a single rhesus monkey, which showed a 33% smaller response in the prone orientation when mild warm irrigations were performed (Böhmer et al. 1992).

Larger prone/supine asymmetries have been reported in squirrel monkeys (Minor and Goldberg 1990; Paige 1985). Using a monaural ice-water caloric stimulus, Paige et al. (1985) reported responses in the prone orientation 78% (for a 5-s irrigation) and 63% (for a 10-s irrigation) smaller than in the supine orientation. Similarly, Minor and Goldberg (1990) reported monaural caloric responses 71% smaller for prone than for supine orientation during cold irrigations and 73% smaller during warm irrigations.²²

In the following, we discuss various factors that could have contributed to the observed asymmetry of eye velocity magnitude with head orientation. We will show that the GIF resolution hypothesis that characterizes interactions between otolith and semicircular canal cues is consistent with experimentally determined features of the VOR evoked by caloric step irrigations.

Possible factors contributing to the prone/supine asymmetry in horizontal SPV

DIRECT THERMAL EFFECT. The prone/supine asymmetry in eye movement responses has often been attributed to a direct thermal effect on afferent nerve responses. In our study, control trials consisting of long-duration caloric stimulation in the null orientation were performed to investigate any contribution of such a thermal effect to the eye movement responses evoked during caloric step trials. Our experimental design assumed that the steady-state SPV evoked by the control-trial irrigations provided a direct measure of the direct thermal effect. If this assumption is correct then the SPV attributed to the direct thermal effect was much smaller than the SPV attributed to convective effects and was too small to account for the prone/supine asymmetry. This small direct thermal effect accounted for only about 15% of the difference between the SPV magnitude in the supine (mean 49°/s with LW-RC irrigations) and prone (mean 29°/s) positions [15% = 2 x (control trial steady-state SPV)/(supine SPV – prone SPV)].

However, there is reason to question whether the steady-state control-trial SPV provided an accurate measure of the direct thermal effect. Specifically, we expected to see an exponential increase in SPV magnitude over time as the continuous irrigations established steady-state thermal gradients across the labyrinths. However, we actually recorded a more complex transient response in the first 100 s where SPV initially developed in a direction opposite to that expected for a direct thermal effect, but then reversed and ultimately reached a steady-state SPV after about 200 s with the expected direction (Fig. 2). The direction of the initial transient we observed is consistent with the results reported by Aw et al. (1998) who showed low-amplitude rightward SPV in the first 60 s after the initiation of warm caloric irrigation of the right ear with the subject positioned in the null orientation.

The origin of this initial transient response is unclear and it is possible that it could influence the estimated SPV attributed to the direct thermal effect. For example, if this initial transient were the result of some unknown process that had a nonzero steady-state response in the null orientation, then our estimate of the direct thermal effect would obviously be in error. However, if this unknown process produced a steady-state SPV that was independent of pitch orientation, then our conclusion remains valid that the steady-state SPV amplitude identified in our control trials is not large enough to account for the prone/supine asymmetry seen on caloric step trials.

Alternatively, if this unknown process contributed a response that changed with pitch orientation, then the above conclusion could be wrong. A worst-case example might be that this unknown process did not contribute to the SPV in the supine or prone orientations, but did contribute in the null orientation in a manner that reduced the steady-state SPV on control trials. In this case, our measure of the direct thermal effect would be too small. Therefore in this worst-case example, the direct thermal effect might truly be responsible for the prone/supine asymmetry observed in caloric step trials, but because this unknown process interfered with our estimate of the direct thermal effect, our conclusion would be wrong about the limited contribution of the direct thermal effect. Future research, both experimental and theoretical, is required to determine the cause of the initial transient response seen in control trials.

Another consideration in evaluating the accuracy of our estimate of the direct thermal effect is that our estimate depends on accurately positioning the horizontal canals in the null orientation. Anatomic measurements in 10 subjects (20 canals) showed a SD of about 6° in the naso-occipital inclination of the canals (Blanks et al. 1975). Given this variability, the tight distribution of our steady-state responses is surprising. This unexpectedly tight distribution may be because the SD of 6° includes both anatomic variability as well as the influence of measurement "noise". The tight distribution of our steady-state thermal responses suggests that anatomic variability may be less than the measurement noise.

DIFFERENCE IN VERTICAL EYE POSITION WITH HEAD ORIENTATION. It has previously been shown that vertical eye position can affect the measurement of horizontal eye velocity during whole body rotation (Fetter et al. 1986). Only small differences in vertical eye position (<7°) were observed in most of our subjects during prone and supine orientation. After correction of the measured horizontal eye velocity to take into account vertical eye position, the large asymmetry in horizontal eye velocity for prone and supine orientation still remained (Table 1). Therefore we conclude that small differences in vertical eye position during supine and prone orientation did not contribute to the asymmetry observed in slow-phase horizontal velocity.

CHANGES IN VOR TIME CONSTANT WITH HEAD ORIENTATION. For conventional, transient monaural caloric irrigations, Minor and Goldberg (1990) suggested that the prone/supine asymmetry in peak slow-phase velocity observed in squirrel monkeys results from the large difference in VOR time constants for symmetrical head tilt (i.e., time constant of 28 s for 45° NU rotation vs. 17 s for 45° ND rotation). That is, if the VOR velocity gain remains constant, then the VOR acceleration gain, which is related to the caloric irrigation response amplitude, changes in proportion to the VOR time constant. Previous experimental results suggest that a time constant change could not explain the prone/supine caloric asymmetry in humans because humans, unlike squirrel monkeys, have similar VOR time constants for symmetrical head tilts (Fetter et al. 1986, 1992).

However, more recent experimental results in humans do show a significant difference between the supine and prone VOR time constants (Table 3 in Zupan et al. 2000). Specifically, this study showed a mean ND postrotatory time constant of 5.22 s and a mean NU postrotatory time constant of 7.00 s. These time constants are very similar to the null-to-prone and null-to-supine time constants that we identified using exponential fits to caloric step responses (5.44 and 7.33 s, respectively, Table 2). Following the logic of Minor and Goldberg (1990), the prone/supine ratio of these time constants (5.22/7.00 = 0.75 from postrotatory results or 5.44/7.33 = 0.74 from caloric step results) should predict the ratio of prone/supine SPV from the caloric step responses if this change in time constant is indeed responsible for the observed prone/supine caloric asymmetry. However, the experimental prone/supine SPV ratio of 0.60 (Table 1) is substantially smaller than the time constant ratio. Therefore even if the time constant change does contribute to the observed prone/supine caloric asymmetry, it cannot account for all of the asymmetry and some other orientation-dependent factor must be present. Below we consider such a factor (i.e., the GIF resolution hypothesis) and then go on to show that this hypothesis can explain both the prone/supine SPV asymmetry and the apparent time constant asymmetry.

For example, when the subject is supine (Fig. 6, left panel), LW-RC irrigations produce a clockwise (CW) convective endolymph flow that generates a horizontal canal response signaling a counterclockwise (CCW) head rotation indicated by _caloric. If the subject was truly rotating CCW, the CNS expectation is that the gravity vector g should be rotating to the left with respect to the subject. This expectation is represented by the vector , which indicates the internal neural estimate of the orientation of gravity with respect to the head. However the otoliths, which sense the net combination of linear acceleration and gravity (Young 1984), continue to sense the true direction of gravity g. This apparent "conflict" between the direction of gravity derived from canal and otolith measures can be resolved by the CNS if the nervous system interprets the disparity between g and to mean that the subject is concurrently accelerating linearly in the direction  = – g. The VOR is assumed to include 2 components: a horizontal angular VOR compensatory for _caloric and a horizontal linear VOR compensatory for Â_y (the interaural component of Â). We refer to the linear VOR component as an "induced" linear VOR because it arises as a consequence of canal-otolith interactions rather than in direct response to an actual linear acceleration. For LW-RC irrigations in the supine position, Â_y is directed toward the subject's left and therefore evokes a rightward linear VOR that adds to the rightward angular VOR component. For LW-RC irrigations in the prone position the induced linear acceleration  is directed toward the subject's left, and the linear VOR evokes rightward-directed eye movements that decrements from the magnitude of the angular VOR (Fig. 6, right panel). For both supine and prone LC-RW irrigations, the angular VOR and linear VOR components change direction from those shown in Fig. 6. Therefore the magnitudes of the angular and linear VOR components still add in the supine position, and subtract in the prone position.

View larger version (20K): [in this window] [in a new window]   fig.6. Schematic representation of the predictions of the gravito-inertial force (GIF) resolution hypothesis for LW-RC irrigations in supine (left panel) and prone (right panel) orientations. GIF resolution predicts that canal-otolith interactions will produce an internal estimation of body motion that includes rotational motion (caloric) and linear acceleration (Â) components. Estimate of linear acceleration (Â = – g) results from a difference between the actual otolith measurement of gravity (g) and the misestimation of a tilt with respect to gravity () caused by the caloric rotational cue that should normally lead to a tilt with respect to gravity. The compensatory VOR includes an angular VOR related to caloric and an induced linear VOR related to the horizontal component of Â. The magnitudes of the angular and linear VOR contributions add in the supine orientation and subtract in the prone orientation.

View larger version (25K): [in this window] [in a new window]   FIG. 7. Model predictions of VOR responses to a LW-RC caloric step stimulus identical to the experimental stimulus. Predictions are derived from a mathematical model of sensory integration for spatial orientation (Zupan et al. 2002). Model responses include predictions consistent with the GIF resolution hypothesis, showing development of a position-dependent internal estimate of linear acceleration (C) and a corresponding asymmetric induced linear VOR (E). This linear VOR adds to the symmetric angular VOR (D) to give a horizontal VOR (F, black line) with a prone/supine asymmetry consistent with experimental results (SPV data from Fig. 3A, gray line).

A simplified GIF resolution model is shown in Fig. 8A. This model captures other features of the caloric step response (response adaptation and possible dynamics of the induced linear VOR mechanism) and permits parameterization of experimental responses. The model input is a normalized angular acceleration that depends on pitch position. The model includes angular VOR dynamics with a dominant response time constant (_AV), gain constant (K_AV), and a VOR adaptation time constant (_a). The output of the induced linear VOR component is a low-pass filtered version of the angular VOR (time constant _LV and gain constant K_LV). The "tilt switch" controls the pitch position-dependent contribution to the final combined angular and linear VOR.

A least-square-error technique was used to determine the model parameters that generate the best fits of the average data shown in Figs. 3 and 4. Before fitting, the normalized SPV data in Figs. 3 and 4 were multiplied by a factor of 50 to give SPV values that closely matched mean SPV values (Table 1). Only SPV data from the last 2 cycles of pitch position changes were included in the fit procedure. The fit procedure assumed that the values of _a, _LV, K_AV, and K_LV were constant over time. However, the angular VOR time constant _AV was permitted to assume a different value for each movement into a different pitch orientation. That is, the curve fit provided estimates of 4 angular VOR time constants for movements from the prone-to-null (_AV,PN), supine-to-null (_AV,SN), null-to-prone (_AV,NP), and null-to-supine (_AV,NS) positions.

In all cases, the curve fit matched the experimental SPV data very closely (e.g., Fig. 8B shows fit to LW-RC data from Fig. 3A results). There was relatively little variation in parameter estimates across conditions (Table 3). The mean VOR adaptation time constant across all conditions was _a = 116 s and the mean linear VOR time constant was _LV = 2.0 s. The mean value of K_LV was 0.26, indicating that the contribution of the linear VOR component was 26% of the angular VOR in the supine and prone orientations.

The angular VOR time constants were similar for tilts from prone-to-null orientation and for tilts from supine-to-null orientation (_AV,PN and _AV,SN, with mean ± SD of 9.4 ± 0.8 s across all conditions). Angular VOR time constants were also similar for tilts from null-to-prone and null-to-supine orientation (_AV,NP and _AV,NS, with mean ± SD of 5.6 ± 1.0 s across all conditions), but were notably shorter than for tilts into the null orientation. This pattern of angular VOR time constant changes is consistent with velocity storage being increased in the upright position, as indicated by the lengthening of the human VOR time constant, but not increased in the tilted positions (Fetter et al. 1986, 1992; Zupan et al. 2000). Additionally, the observation that _AV,NP and _AV,NS had similar values is consistent with the idea that, in the tilted position, angular VOR dynamics are not dependent on NU versus ND orientation. A similar result was obtained from postrotatory tilt experiments (Fig. 18.8 in Merfeld and Zupan 2003).

Our model-based interpretation of caloric step responses helps explain the VOR time constant asymmetry used by Minor and Goldberg (1990) to explain the observed prone/supine caloric asymmetry. First, the entire prone/supine SPV asymmetry, both amplitude and time constant asymmetries, can be explained by additive or subtractive contributions of the angular and linear VOR components predicted by the GIF resolution hypothesis. Second, the GIF resolution hypothesis accounts for the variation in time constants obtained by direct measurements of SPV responses (Table 2) for different portions of the stimulus. Because the model simulations shown in Fig. 8 account for the SPV time course during all segments of the caloric step stimulus, we infer that the differences in time constants for tilts from null-to-supine and from null-to-prone orientations, and the differences in time constants for tilts from supine-to-null and from prone-to-null orientations (Table 2), are caused by an interaction between the dynamic responses of the induced linear VOR and the angular VOR. That is, rather than the VOR time constant variation explaining the prone/supine SPV asymmetry, both the prone/supine time constant asymmetry and the SPV asymmetry are seen as a consequence of the canal-otolith sensory interaction, which elicits an angular VOR and an induced linear VOR.

One might consider testing the GIF prediction that a linear VOR component contributes to the caloric response by evaluating whether the response was correlated with vergence, given that the gain of the linear VOR correlates with vergence (Paige and Tomko 1991b). There are 2 considerations that need to be taken into account before performing such an evaluation. First, the gain of the angular VOR also increases with vergence (Snyder and King 1992; Viirre et al. 1986). Therefore even a positive finding would remain ambiguous. Second, the correlation between vergence and the linear VOR gain, observed at 1 Hz and higher (Telford et al. 1997), was not observed at 0.5 Hz. Therefore because the frequency content of the predicted linear VOR responses induced by our caloric stimulation paradigm and the earlier "dumping" studies (Merfeld et al. 1999; Zupan et al. 2000) falls below 0.5 Hz, we would not expect the linear VOR induced by these paradigms to show a strong correlation with vergence.

In summary, the caloric step stimulus was used to investigate the orientation-dependent dynamic properties of the VOR and factors influencing the prone/supine asymmetry previously observed in caloric responses. Our results suggest that a change in horizontal canal afferent activity caused by a direct thermal effect is insufficient to explain the prone/supine asymmetry. Previous studies in squirrel monkeys concluded that some position-dependent, possibly otolith-mediated effect must contribute to the prone/supine asymmetry (Minor and Goldberg 1990; Paige 1985). We demonstrate that canal-otolith interactions could also be responsible for the prone/supine asymmetry observed in humans, and suggest a specific mechanism based on the GIF resolution hypothesis. Therefore the response to a caloric test, which has long been considered to provide clinically relevant information about horizontal semicircular canal function only, can now be seen to involve the processing of both canal and otolith information.

	GRANTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
This work was supported by a National Institute of Deafness and Other Communication Disorders Grant R01 DC-04158.

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
We thank J. Roth and S. Clark-Donovan for assistance.

	FOOTNOTES

 
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

¹ The large mean vertical SPV in the supine orientation for LC-RW irrigations probably demonstrates a sampling bias from testing only 3 subjects, given that mean vertical SPVs were small for the LW-RC irrigations where data from 9 subjects were available.

² In future studies that use caloric irrigations to investigate the mechanisms that contribute to orientation-dependent VOR asymmetries, consideration must be given to the strength of the cold-water irrigations used in experiments. Cold-water irrigations that are strong enough to drive afferent activity into inhibitory cutoff will cause a prone > supine asymmetry that is in a direction opposite to the prone < supine asymmetry predicted by the GIF resolution hypothesis and by direct thermal stimulation. It is likely that strong cold-water irrigations were responsible for the prone > supine asymmetry observed in some previous studies (Arai et al. 2002; Böhmer et al. 1992, 1996).

Address for reprint requests and other correspondence: R. J. Peterka, Neurological Sciences Institute, OHSU West Campus, Building 1, 505 NW 185th Ave., Beaverton, OR 97006 (E-mail: peterkar{at}ohsu.edu).

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Angelaki DE, McHenry MQ, Dickman JD, Newlands SD, and Hess BJ. Computation of inertial motion: neural strategies to resolve ambiguous otolith information. J Neurosci 19: 316–327, 1999.[Abstract/Free Full Text]

Angelaki DE, Wei M, and Merfeld DM. Vestibular discrimination of gravity and translational acceleration. Ann NY Acad Sci 942: 114–127, 2001.[Abstract/Free Full Text]

Arai Y, Yakushin SB, Cohen B, Suzuki J, and Raphan T. Spatial orientation of caloric nystagmus in semicircular canal-plugged monkeys. J Neurophysiol 88: 914–928, 2002.[Abstract/Free Full Text]

Aw ST, Haslwanter T, Fetter M, Heimberger J, and Todd MJ. Contribution of the vertical semicircular canals to the caloric nystagmus. Acta Otolaryngol (Stockh) 118: 618–627, 1998.[CrossRef][Medline]

Bisdorff AR, Sancovic S, Debatisse D, Bentley C, Gresty MA, and Bronstein AM. Positional nystagmus in the dark in normal subjects. Neuro-ophthalmology 24: 283–290, 2000.

Blanks RHI, Curthoys IS, and Markham CH. Planar relationships of the semicircular canals in man. Acta Otolaryngol (Stockh) 80: 185–196, 1975.