Effects of visual fixation on gaze stability. A, A normal subject fixates on a small red target light in a dark room that is turned off at the arrow; subsequently, she attempts to fixate on the remembered location of the target. Increased drift occurs after the light is turned off. Gaze positions are relative, having been offset for clarity of display. B, Effect of electronically delaying the visual consequences of eye movements by 480 milliseconds in a normal subject, starting at time zero. The subject developed spontaneous ocular oscillations at about 1.0 Hz; details of the method are described elsewhere.8 C, Acquired pendular nystagmus during attempted fixation of a stationary target in a patient with multiple sclerosis. The frequency of these oscillations is more than 6 Hz. D, Effect of electronically delaying visual feedback by 480 milliseconds in the patient whose nystagmus is shown in C. The oscillations of her nystagmus (essentially unchanged) were superimposed on growing oscillations at about 0.67 Hz, which were induced by the electronic manipulation. Positive deflections indicate rightward and upward rotations. Scales differ among the parts. RH indicates right horizontal; LH, left horizontal; RV, right vertical; and LV, left vertical.
Hypothetical scheme that accounts for how a positive feedback loop (of gain, K) could govern the time constant of an inherently leaky brainstem neural integrator (LNI). The effects of varying the value of K are as follows. If K is appropriate, neural integration would be perfect and the eyes would be held steady in their new position in the orbit after an eye movement. If K is too small, the integration becomes imperfect (leaky) and the eyes drift back, with a negative exponential time course, toward the central position; gaze-evoked nystagmus results. If K is too large, the neural integrator becomes unstable and the eyes drift away from the central position with a positive exponential time course (increasing velocity) also causing nystagmus. One possibility is that the time constant of the network of cells in the nucleus prepositus hypoglossi–medial vestibular nucleus is controlled by a feedback loop that passes through the cerebellar flocculus.
Stahl JS, Averbuch-Heller L, Leigh RJ. Acquired Nystagmus. Arch Ophthalmol. 2000;118(4):544-549. doi:10.1001/archopht.118.4.544
LEONARD A.LEVINMD, PhDFrom the Departments of Neurology (Drs Stahl and Leigh), Neuroscience (Dr Leigh), and Biomedical Engineering (Dr Leigh), Department of Veterans Affairs Medical Center and University Hospitals, Case Western Reserve University, Cleveland, Ohio; and the Departments of Neurology and Ophthalmology, Rabin Medical Center, Tel Aviv, Israel (Dr Averbuch-Heller).
Traditionally, acquired forms of nystagmus have been classified in descriptive terms based on their clinical features and recorded waveforms. In the past 20 years, the mechanisms of several major forms of nystagmus have been elucidated; animal and mathematical models for these ocular oscillations have been developed. These advances, which owe much to modern anatomical, physiological, and pharmacological techniques, have enhanced the diagnostic value of nystagmus and provide the basis for developing rational therapy.
For more than a century, clinicians have found acquired nystagmus a perplexing subject—the clinicopathologic connections were often unknown, and treatment options were few and unreliable. Given these diagnostic and therapeutic limitations, it seemed reasonable for clinicians to limit their evaluation of nystagmus to the observable characteristics—"phenomenology." Research for the past 3 decades has led to progress in understanding the pathophysiological features of several forms of nystagmus and developing treatments.1 The first advance was the development of methods to reliably measure eye rotations in 3 planes.2,3 Such measurements allowed precise characterization of nystagmus waveforms, which enabled clinicians to distinguish between acquired and congenital forms of nystagmus4 and to differentiate nystagmus from conditions that mimic it, such as saccadic oscillations.1 Since 1987, the development of experimental models of nystagmus, by inducing discrete lesions or pharmacological inactivation,5 has expanded our understanding of the pathophysiological features of acquired nystagmus. Before summarizing these developments, we first define certain important terms and relate nystagmus to visual needs. We will then concentrate on 3 major mechanisms that have been identified to cause nystagmus. Finally, we will summarize how these concepts can be applied to the development of new therapies.
Nystagmus is defined as repetitive, to-and-fro involuntary eye movements that are initiated by slow drifts. Some forms of nystagmus are normal. Thus, nystagmus that occurs in response to rotation of an optokinetic drum or rotation of the body in space acts to preserve clear vision. Eye movements that point the retinal fovea at an object of interest are called foveating; those that move the fovea away from the object are called defoveating. In pathological nystagmus, each cycle of movement is usually initiated by an involuntary, defoveating drift of the eye away from the object of interest, followed by a return movement. Both the defoveating and refoveating movements may be slow (nonsaccadic), as in pendular nystagmus, or the defoveating movement may be slow and the refoveating movement fast (saccadic), as in jerk nystagmus. The direction of jerk nystagmus is often defined by the fast phase (eg, downbeat nystagmus), which is more prominent on bedside examination. In general, however, it is the slow phase that takes the eye off the visual target and thus reflects the underlying disorder. Nystagmus should be distinguished from saccadic intrusions. When repetitive (as in ocular flutter, opsoclonus, repetitive square wave jerks, or macrosaccadic oscillations), these disorders superficially resemble nystagmus but differ in that the initial, pathological, defoveating movement is saccadic. In describing any form of nystagmus, it is important to note its trajectory (how much the eye rotates horizontally, vertically, or torsionally around the line of sight).
A useful way to approach nystagmus is to start by considering the requirements that eye movements must normally satisfy to permit a clear and stable view of the environment. Simply stated, the angle of gaze (corresponding to the line of sight, originating at the fovea) must be pointed at the object of regard, and held fairly steadily. Just how steadily gaze must be held depends on what is being viewed. For visual targets consisting of sharp edges (eg, Snellen optotypes), gaze drift exceeding about 5° per second degrades vision,6 and may lead to the illusion of motion of the seen world (oscillopsia).
In health, at least 3 separate mechanisms can be identified that normally collaborate to prevent deviation of the line of sight from the object of regard. The first mechanism is fixation, which has 2 distinct components: the visual system's ability to detect retinal image drift and program corrective eye movements and the ability to suppress unwanted saccades that would take the eye away from the target. The second mechanism is the vestibulo-ocular reflex, by which eye movements compensate for head perturbations at short latency, and thus guarantee clear vision during natural activities, especially locomotion. The third mechanism is the gaze-holding system, which programs a constant level of muscle activity to counteract the elastic pull of the extraocular suspensory structures and muscles, which would tend to return the eye toward a central position in the orbit. Proper operation of this gaze-holding system is critical for maintaining stable, eccentric eye positions. Disorders of any of these 3 systems may disrupt steady gaze and lead to nystagmus. Furthermore, since the pathogenesis of each of these 3 types of nystagmus differs, certain features characterize each disorder.
Nystagmus invariably occurs in complete blindness, and may also accompany incomplete visual impairment due to lesions anywhere along the visual pathways (Table 1). Visual loss may generate nystagmus in 2 ways—through loss of visual inputs to the fixation system that are used to detect and immediately correct ocular drift and through loss of visual signals that are used, over the long-term, to tune the ocular motor systems. Each of these effects will be considered separately.
The first of these components may be regarded as "visual fixation," and it is easily demonstrated when a normal subject attempts to fixate on the remembered location of a target after the room is switched to darkness: the eye drifts off target several times faster than when the subject was actually viewing it (Figure 1, A). This visual fixation mechanism, by which smooth eye movements correct for drifts of gaze, depends on the motion-vision system (especially portions of the cerebral cortex, such as the middle temporal area or V5). Although such visually mediated eye movements are important for maintaining steady fixation, they have one important limitation—a response time of longer than 70 milliseconds. If this response time is delayed further by disease of the visual system, then the brain's attempts at correcting eye drifts might actually add to the retinal error rather than reducing it, leading to ocular oscillations. Experimentally, delaying the visual consequences of eye movements (using electronic feedback) induces oscillations in normal subjects (Figure 1, B).7,8 It is tempting to ascribe to this "visual-delay mechanism" the pendular oscillations (Figure 1, C) that occur in patients with optic nerve demyelination (due to multiple sclerosis). If this mechanism were indeed responsible, then simulating additional delays (as previously mentioned) should alter the frequency of the pendular oscillation. In fact, experimentally adding visual delays in patients with pendular nystagmus induced a superimposed low-frequency oscillation (similar to that generated in normal subjects) but did not alter the original nystagmus (Figure 1, D).8 Thus, conduction delays in the visual motion system are unlikely to be the main pathogenetic mechanism leading to acquired pendular nystagmus.
The second component of the visual influence on gaze control concerns the need for continuous calibrating and optimizing all types of eye movements. This optimization depends heavily on visual projections to the cerebellum—the
"ocular motor repair shop."9 The cerebellum receives visual signals from motion-vision areas of the cerebral cortex via the pontine nuclei.10,11 In addition, visual signals for calibration probably also pass to the cerebellum via the inferior olivary nucleus on climbing fibers. However, proper calibration of the ocular motor system requires that visual signals be compared with eye movement commands, and the latter probably reach the cerebellum from the cell groups of the paramedian tracts, which lie diffusely throughout the midline of the brainstem and receive input from all premotor structures that project to ocular motoneurons.12 Thus, lesions at any part of this visual-motor "calibration" pathway might deprive the brain of signals that are essential to hold each of the eyes precisely on the object of regard; the result would be drifts of the eyes away from the target, leading to nystagmus. Although these mechanisms normally act to stabilize gaze, occasionally their response to disease may cause ocular oscillations, such as the "dissociated nystagmus" that occurs as a feature of internuclear ophthalmoplegia. Thus, patching the eye affected by adduction paresis abolishes the nystagmus, whereas patching the normal eye increases it.13
One special case of nystagmus associated with a specific visual system disorder is seesaw nystagmus; this is a pendular oscillation in which half a cycle consists of elevation and intorsion of one eye and synchronous depression and extorsion of the other eye, with the vertical and torsional movements reversing during the next half cycle.1 Seesaw nystagmus is usually associated with parasellar lesions, such as pituitary tumors, implicating crossing fibers of the optic nerves in its pathogenesis. Congenital seesaw nystagmus has been reported in a mutant strain of dogs that lack an optic chiasm14 and in patients with a similar developmental defect.15 Acquired seesaw nystagmus has also been documented in a patient with progressive visual loss due to retinitis pigmentosa.16 In each of these contexts, the loss of crossed visual inputs may underlie the development of the unusual seesaw pattern. Under natural conditions, seesaw eye movements occur when subjects view a target located off the midsagittal plane during an ear-to-shoulder head roll.17 Crossed visual inputs are presumably necessary to keep this response calibrated and, if removed, might lead to instability and the oscillations of seesaw nystagmus.
The vestibulo-ocular reflex plays an essential role in maintaining gaze stability during locomotion. The relatively simple circuitry connecting the vestibular organ of the inner ear (which transduces head acceleration) to the extraocular muscles allows compensatory eye movements to be generated at a latency of less than 15 milliseconds.18 This circuitry achieves high sensitivity and wide dynamic range by maintaining a high resting discharge and basing its calculation of head movements on the differences in head activity emanating from left and right vestibular organs. However, these characteristics also render the system vulnerable to disorders that create left-right imbalances. Any reduction of activity from one side (for instance, due to infection or ischemia) is interpreted by the brain as a head rotation away from the affected side. Such imbalance produces a nystagmus that is mixed horizontal-torsional, reflecting the sum of sensitivity vectors from the 3 semicircular canals.19 When the imbalance arises in the periphery, it can be attenuated by fixation, because visually mediated eye movements are functioning normally.
Imbalances in the vestibular system may also arise due to central lesions. The resulting nystagmus differs from that produced by peripheral lesions in several respects: it is less likely to be attenuated by visual fixation (because the mechanisms mediating the visual inputs are usually involved in the damage), and the plane of the nystagmus may be purely vertical (a pattern that would be difficult to achieve by any naturally occurring lesion in the periphery). Downbeat nystagmus is usually associated with lesions of the vestibulocerebellum and underlying medulla.1 Upbeat nystagmus is most commonly reported with medullary lesions (including the perihypoglossal nuclei, the adjacent medial vestibular nucleus, and nucleus intercalatus20—all structures important for gaze holding).
That central lesions are more likely to generate vertical, rather than horizontal, nystagmus probably reflects fundamental differences between the anatomical and pharmacological features of the lateral ("horizontal") and vertical semicircular canal systems. In the horizontal system, activities from the left and right sides balance each other, while in the vertical system, the geometry of the canals is such that there is a small residual imbalance that tends to drive the eyes upward.21 Central mechanisms—chiefly inhibitory connections from the flocculus and paraflocculus—cancel this normal peripheral imbalance. Experimental lesions to the vestibulocerebellum unmask the latent imbalance and produce downbeat nystagmus.22 Another fundamental difference between vertical and horizontal systems lies in the fact that, in the horizontal system, the left-right balance is determined by the activities of the left and right horizontal canals, whereas in the vertical system, the activity of the anterior canal on one side must be balanced by the activity of the posterior canal on the other side. Because the central pathways mediating anterior and posterior canal signals differ in anatomical and probably in pharmacological features, it is possible to impair anterior or posterior canal pathways differentially, leading to a vertical imbalance. Thus, lesions aimed at pathways of the posterior canal system as they cross the midline produce downbeat nystagmus.23 Likewise, the cerebellar flocculus inhibits anterior canal pathways but not posterior canal pathways,24 and consequently experimental flocculectomy leads to a vertical imbalance and downbeat nystagmus.22 Differences in the pharmacological properties of anterior and posterior canal pathways probably account for the vertical jerk nystagmus produced by numerous substances (alcohol25 and nicotine26).
Yet another form of acquired nystagmus ultimately attributable to vestibular mechanisms is periodic alternating nystagmus (PAN), a spontaneous horizontal nystagmus, present in primary gaze, that reverses direction approximately every 2 minutes.27 Although rare, acquired PAN is perhaps the best understood of all forms of nystagmus and was the first for which an effective treatment was identified.28 This nystagmus arises from disinhibition of a central "velocity storage" mechanism that normally enhances the raw vestibular signal during sustained body rotations. Ablation of the cerebellar nodulus and uvula in monkeys causes excessive velocity storage, prolongs the duration of rotationally induced nystagmus, and produces PAN when the animal is in darkness.29 The periodic reversal in direction probably reflects unsuccessful attempts of the vestibular "repair mechanisms" to halt the spontaneous nystagmus. Visual stabilization mechanisms should suppress the nystagmus, but disease of the cerebellum that causes PAN usually inflicts concomitant damage on these mechanisms; in rare cases, eye disease prevents fixation and allows PAN to develop.30 Pharmacological evidence suggests that the normal inhibitory actions of the nodulus and uvula are mediated by γ-aminobutyric acid (GABA).31 Baclofen, a GABABagonist, usually abolishes PAN, probably by restoring the inhibitory tone that would normally be supplied by inputs from the nodulus and uvula.27,28
To maintain clear vision of an object located off to one side, the eyes must be held steadily at eccentric positions in the orbit. This requires a tonic contraction of the extraocular muscles to oppose the elastic forces imposed by the orbital tissues, which would tend to return the eye to a central position. To achieve this tonic muscle contraction, the ocular motor system must synthesize a signal proportional to the desired eye position. Although extraocular motor neurons do encode desired eye position,32 premotor neurons that send vestibular,33 saccadic,34 and pursuit35 signals to the ocular motor neurons encode eye velocity, not position. This difference implies that a mathematical integration, converting eye velocity commands to eye position commands, must be performed by the nervous system.36,37 It is established that, for horizontal eye movements, this neural integration depends heavily on the nucleus prepositus hypoglossi–medial vestibular nucleus (NPH-MVN) region in the caudal pons and medulla.5,38 For vertical gaze holding, the interstitial nucleus of Cajal plays an analogous role.39 In addition, the cerebellum, especially the flocculus, makes an important contribution to vertical and horizontal neural integrators.22 Lesions of any of these structures impair the ability to hold the eyes in eccentric gaze, resulting in the centripetal slow phases and centrifugal corrective fast phases that define gaze-evoked nystagmus.1
Recent studies, using the technique of pharmacological inactivation, have clarified the neurotransmitters involved in the process of neural integration of ocular motor signals. Microinjection of the GABAAagonist, muscimol, into the NPH-MVN severely impairs neural integration for horizontal movements, resulting in gaze-evoked nystagmus.40,41 Recently, Arnold and colleagues42 reported the effects on gaze stability of microinjections of 8 different drugs into the NPH-MVN of monkeys. Agents with either agonist or antagonist actions at GABA, glutamate, and kainate receptors all caused gaze-evoked nystagmus, while agents acting at the glycine receptor (glycine and strychnine) had no effect. In contrast, when muscimol was injected near the center of the MVN, the eyes sometimes drifted away from the central position with increasing velocity waveforms ("runaway nystagmus"), indicating that the actions of the neural integrator had become excessive (unstable). Clinically, patients who show nystagmus with increasing velocity waveforms have cerebellar, not brainstem, lesions.43,44 Taken together, this experimental and clinical evidence is consistent with the idea that the network of neurons within the NPH-MVN is under cerebellar feedback control; this idea is schematized in Figure 2. Disruption of this system may lead to gaze-evoked nystagmus (a leaky integrator) or runaway nystagmus (an unstable integrator).
Finally, instability in the neural integrator might also be the basis for pendular nystagmus, such as occurs in multiple sclerosis (Figure 1, C). If this were the case, then saccadic eye movements, which transiently inhibit some constituent neurons in the neural integrator,45 might be expected to "reset" (phase shift) the oscillations. Das and colleagues46 have noted that, in patients with multiple sclerosis, saccades did cause phase shifts of the pendular oscillations in proportion to their size. Thus, in patients with multiple sclerosis, the oscillations of acquired pendular nystagmus might arise from within the neural integrator. However, acquired pendular nystagmus occurring in other disorders, such as oculopalatal myoclonus, may have a separate pathogenesis.1
The studies previously outlined have suggested approaches to the treatment of nystagmus and its visual consequences that are either symptomatic or pathophysiological. One symptomatic approach is to weaken the extraocular muscles by surgical means or botulinus toxin injections.47- 50 The drawback of this approach is that it also impairs the normal eye movements that are required for clear vision during natural activities (eg, the vestibulo-ocular reflex). Optical methods that seek to stabilize images on the retina will have similar limitations,51 unless special technology is used to distinguish abnormal from normal eye movements and negate just the former.52 A pathophysiological approach is to administer drugs that selectively suppress nystagmus. The initial reports that GABAergic projections contributed to the neural integrator40,41 prompted a double-blind study to evaluate 2 GABAergic agents, baclofen and gabapentin.53 The latter agent was shown to be an effective treatment in some patients with acquired pendular nystagmus. Another open-label study showed that memantine, a drug with glutamate-antagonist effects, was also effective in suppressing acquired pendular nystagmus.54 Further studies to compare drugs that influence either GABA or glutamate mechanisms as a treatment for pathological nystagmus seem warranted.
Accepted for publication November 24, 1999.
This study was supported in part by grants K08 EY00356 (Dr Stahl) and EY06717 (Dr Leigh) from the National Institutes of Health, Bethesda, Md; the German-Israeli Foundation for Scientific Research and Development, Jerusalem, Israel (Dr Averbuch-Heller); the Office of Research and Development, Medical Research Service, Department of Veterans Affairs, Washington, DC (Dr Leigh); and the Evenor Armington Fund, Cleveland, Ohio (Dr Leigh).
Reprints: R. John Leigh, MD, Department of Neurology, University Hospitals, 11100 Euclid Ave, Cleveland, OH 44106-5040 (e-mail: firstname.lastname@example.org).