Cross-axis adaptation improves 3D vestibulo-ocular reflex alignment during chronic stimulation via a head-mounted multichannel vestibular prosthesis

Abstract

By sensing three-dimensional (3D) head rotation and electrically stimulating the three ampullary branches of a vestibular nerve to encode head angular velocity, a multichannel vestibular prosthesis (MVP) can restore vestibular sensation to individuals disabled by loss of vestibular hair cell function. However, current spread to afferent fibers innervating non-targeted canals and otolith end organs can distort the vestibular nerve activation pattern, causing misalignment between the perceived and actual axis of head rotation. We hypothesized that over time, central neural mechanisms can adapt to correct this misalignment. To test this, we rendered five chinchillas vestibular deficient via bilateral gentamicin treatment and unilaterally implanted them with a head-mounted MVP. Comparison of 3D angular vestibulo-ocular reflex (aVOR) responses during 2 Hz, 50°/s peak horizontal sinusoidal head rotations in darkness on the first, third, and seventh days of continual MVP use revealed that eye responses about the intended axis remained stable (at about 70% of the normal gain) while misalignment improved significantly by the end of 1 week of prosthetic stimulation. A comparable time course of improvement was also observed for head rotations about the other two semicircular canal axes and at every stimulus frequency examined (0.2-5 Hz). In addition, the extent of disconjugacy between the two eyes progressively improved during the same time window. These results indicate that the central nervous system rapidly adapts to multichannel prosthetic vestibular stimulation to markedly improve 3D aVOR alignment within the first week after activation. Similar adaptive improvements are likely to occur in other species, including humans.

Fig 1

Responses to head rotation before, during and after 1 week of prosthetic stimulation using a multichannel vestibular prosthesis. Column 1: Head rotational velocity (black) and mean eye rotational velocity components (horizontal, left-anterior/right-posterior [LARP], and right-anterior/left-posterior [RALP]) of a normal chinchilla during 2 Hz, 50°/s peak whole-body rotations in darkness. Top, middle and bottom rows show data during whole-body rotations about the horizontal, LARP, and RALP axes, respectively, adapted from Della Santina et al., 2007. Column 2: Head and eye velocity measured four weeks after electrode implantation and bilateral gentamicin treatment in a chinchilla, with the prosthesis set to fire at constant rates (mean 60 pulse/s on each channel) that are not modulated by head rotation. Columns 3–5: Head and eye velocity measured on the 1^st, 3^rd, and 7^th day of continuous motion-modulated prosthetic stimulation. Note the progressive (though still incomplete) reduction in wrong-axis eye movement components. Each trace is the mean of ≥5 cycles; SD <5°/s at each time point for each trace. The first half-cycle in each plot is the response to excitation of the left labyrinth.

Fig 2

Changes in axis of eye response for one chinchilla over 7 days of stimulation. Each data vector depicts the 3D axis and magnitude of aVOR responses during whole-body 2 Hz, 50°/s peak head rotations in darkness about the mean horizontal, LARP, and RALP semicircular canal axes. Vector length indicates the peak response velocity; for comparison, thick axes depict inverse of 50°/s peak head rotations about each canal axis. Data shown are for peak excitatory responses; curved arrows in inset show direction of aVOR response to excitation of the corresponding semicircular canal in the left labyrinth. Progression over time toward the ideal response is apparent for each axis of head rotation.

Fig 3

Gain of each component of 3D aVOR responses for each of N=5 chinchillas during 2 Hz, 50°/s peak whole-body rotations about the horizontal, LARP and RALP canal axes on the 1^st, 3^rd, and 7^th days of continuous stimulation using a head-mounted multichannel vestibular prosthesis. Before: values measured to head movement with the prosthesis stimulating at constant pulse rates, independent of head velocity, 3–5 hours after prosthesis activation. After: values measured 4–48 hours after the prosthesis was powered off on Day 7. Normal: Responses of 5 normal chinchillas.

Fig 4

Mean± SD gain (N=5 chinchillas) for each component of the peak 3D aVOR eye movement response during 2 Hz, 50°/s peak whole-body rotations about the horizontal, LARP and RALP canal axes on the 1^st, 3^rd, and 7^th days of continuous stimulation using a head-mounted multichannel vestibular prosthesis. Before: values measured to head movement with the prosthesis stimulating at constant pulse rates, independent of head velocity, 3–5 hours after prosthesis activation. After: values measured 4–48 hours after the prosthesis was powered off on Day 7. Normal: Responses of 5 normal chinchillas.

Fig 5

Mean± SD gain (top) and eye/head misalignment angle (bottom) for N=5 chinchillas, displayed for each component of the peak 3D aVOR eye movement response during 50°/s peak whole-body rotations in darkness at 0.2–5 Hz about the horizontal, LARP and RALP canal axes on the 1^st, 3^rd, and 7^th days of continuous stimulation using a head-mounted multichannel vestibular prosthesis. The 3D aVOR response maintained gain of the desired component while improving misalignment over 7 days of continuous prosthesis for every stimulus frequency, axis and animal examined.

Fig 6

Changes in disconjugacy during adaptation to prosthetic stimulation. (a) Head rotational velocity and mean eye rotational velocity components (horizontal, LARP, and RALP) for each eye of a normal chinchilla during 2 Hz, 50°/s peak whole-body rotations in darkness about the mean horizontal semicircular canal axis on Days 1 and 7 of continual prosthesis use. For each component of eye rotation, differences between the two eyes are reduced by Day 7. (b) 3D aVOR response axes corresponding to data in Panel A. L=left eye R=right eye. By Day 7, responses for both eyes align better with each other and to the horizontal head rotation axis (+Z). (c) Mean ±SD disconjugacy angle (between the 3D axes of rotation for the two eyes) during 2 Hz whole-body 50°/s peak sinusoidal head rotations in darkness about the horizontal (■), LARP (◆) and RALP (●) semicircular canal axes. Disconjugacy of the aVOR response improved over 7 days of continuous prosthesis. Each data point for prosthesis animals is the mean of 3–5 cases; N=5 for normal animals.