Abstract
purpose. To examine the waveform characteristics of 37 subjects with manifest
latent nystagmus (MLN) and determine the manner in which visual
feedback influences the nature of the waveform.
methods. Binocular recordings of the eye movements of all subjects were
undertaken using an infrared tracking system. Subjects viewed
the target binocularly and monocularly in primary gaze. The effect of
visual feedback on the nature of the MLN waveform was examined by
either removing the fixation target or by progressively stabilizing the
target in relation to the retina. This progressive stabilization was
achieved by feeding back the eye movement signal to move an otherwise
stationary target.
results. Four types of MLN were distinguished on the basis of the fixation
characteristics seen during binocular and monocular viewing. First,
under binocular viewing conditions, subjects could theoretically
exhibit stable fixation (type 1 MLN). In addition, three other MLN
types were recorded during binocular fixation: conjugate horizontal
square-wave jerks (type 2 MLN), conjugate torsional nystagmus (type 3
MLN) and conjugate horizontal jerk MLN waveforms (type 4 MLN).
Monocular viewing always gave rise to a conjugate horizontal jerk MLN
waveform for each of the four types of MLN. More than 80% of the
subjects exhibited either type 3 or type 4 MLN, both of which conform
with previous classic descriptions of MLN. Much less common was type 2
MLN. Type 1 MLN (conventionally referred to as a latent nystagmus)
appeared to be a rare occurrence. In addition to the two classic linear
and decelerating MLN slow phases, four additional slow-phase shapes
with either saccadic or pendular elements were recorded and described.
Removing visual feedback generally reduced the mean slow-phase velocity
and the number of fast phases. For each subject some variability of the
slow-phase class was documented from session to session.
conclusions. Four types of MLN have been described. Their differences are based on
their binocular oculomotor behavior, and it is proposed that type 1 MLN
and type 4 MLN represent the absolute states and types 2 and 3 the
intermediate levels of the MLN spectrum. All types of MLN appear to be
strongly visually driven and are largely dependent on the attentional
state of the subject and the target conditions. Six different classes
of slow phase were found among the four MLN types. The introduction of
visual feedback had an immediate effect on the subsequent slow phase or
fast phase. It is likely that adaptation mechanisms are in play after a
period of visual feedback.
The two most common types of benign nystagmus seen in infancy are
congenital nystagmus (CN) and manifest latent nystagmus
(MLN).
1 2 3 4 In both conditions the oscillations are
typically conjugate, horizontal, and jerk. Differential diagnosis is
made on the basis that the CN slow phases are typically of an
increasing exponential velocity form, whereas in MLN the form is
decelerating or linear.
5 6 In addition to its
distinguishing slow phase, the fast phase of MLN always beats toward
the viewing eye. MLN is also closely associated with the presence of
strabismus and dissociated vertical divergence.
6
A third, but less common, type of infantile nystagmus is latent
nystagmus (LN).
2 5 6 7 In this disorder, during binocular
viewing conditions it is said that the eyes are steady, but during
monocular viewing, bilateral conjugate jerk nystagmus becomes manifest.
As in MLN, the LN slow phase is either a decreasing or linear velocity,
with the fast phase directed toward the viewing eye. Our experience
over the past 20 years in examining more than 300 subjects with
nystagmus has been that MLN is not uncommon (10%–15%), whereas LN is
a rare occurrence. This is in agreement with other
reports.
4 5 6
Over the years, we have carefully examined the eye movements of 37
individuals with MLN and have found that the oscillations do not fall
neatly into the prevailing definitions. The purpose of this study was
therefore to investigate the nature of MLN and describe the waveform
characteristics in detail. To this end, we examined whether there is
just one type of MLN and only two possible classes of slow phases. We
also considered where LN should be classified within the spectrum of
infantile nystagmus. Finally, by varying visual feedback, we examined
how visual signals influence the MLN oscillation. In our results, MLN
was a complex oscillation, and the nature of MLN was greatly affected
by the presence of a target and by the amount of retinal image movement
experienced by the subject.
Thirty-seven subjects (aged 18–67 years) with MLN took part in
this study. All underwent full clinical investigation
(Table 1) . The tenets of the Declaration of Helsinki were followed in
this research program. Informed consent was obtained from all subjects
after the nature and possible consequences of the study had been
explained.
Binocular horizontal and vertical eye movements were monitored
using a head-mounted infrared limbal tracker (IRIS 6500; Skalar
Medical, Delft, The Netherlands). The analog output was filtered
through a 100-Hz low-pass filter, digitized to 12-bit resolution, and
then sampled at between 1- and 5-msec intervals. The system was linear
to ±20° and had a resolution of 0.03°. Eye movements were
calibrated by presenting a pursuit stimulus that moved horizontally
over a range of ±10° in a sinusoidal manner at 0.24 Hz for three
cycles or 0.32 Hz for four cycles. Subjects were instructed to follow
this as accurately as possible and by subsequently plotting eye
position against target position, the eye position data were
calibrated. A chin rest with supplementary cheek supports was used to
stabilize the head position. Head movements were well controlled in
relation to earth (<0.1° in amplitude). Fundus video recordings were
also performed on a selected number of subjects to assess the torsional
components of any oscillation.
Experiment 1: The Nystagmus Characteristics under Binocular and
Monocular Viewing Conditions
Both eyes were recorded when 37 subjects fixated a composite
bull’s eye and cross target (5.5°) which was back projected onto a
screen that subtended 105° × 41° when viewed from 114
cm.
8 The stationary target was presented in the primary
position. After fixation was recorded with both eyes open, each eye was
fully occluded in turn. Subjects were instructed to look at the center
of the target. The nature of the slow phase was judged on the basis of
eye velocity and eye acceleration profiles.
The characteristics of MLN were examined in 11 subjects under
three test conditions: target present, target absent, and target under
servocontrol. When the target was absent, the subject was requested to
direct the gaze at its remembered position. The amount of retinal image
motion experienced by each subject was controlled by varying the
amounts of eye position feedback to a mirror galvanometer and thus
target position. Feedback gain (fbg) is defined as target velocity/eye
velocity. When the target position was decoupled from the subject’s
nystagmus the fbg was equal to 0. Feedback gains greater than 0 but
less than +1.0 decreased the retinal image movement. The retinal image
was stabilized when the fbg was equal to +1.0. Subjects were instructed
to keep the target clear and on the screen throughout each viewing
period. All investigations with the target under servocontrol were
performed under monocular viewing conditions only. Further details of
the experimental arrangement can be found in our recent
publication.
8
Our waveform analysis indicated that MLN is not a single entity
but is best divided into four distinct categories
(Fig. 1) . These categories were distinguished on the basis of the fixation
characteristics seen during binocular and monocular viewing. Type 1 MLN
represents the absolute case in which the eyes are stable
during binocular viewing, but when either eye is covered, the eyes
oscillate in a manner consistent with MLN. In the past, this MLN type
has been referred to as LN
2 3 4 5 6 7. Of the 37 subjects in our
study, none exhibited type 1 MLN.
In type 2 MLN, horizontal conjugate square-wave jerks are seen during
binocular viewing
(Fig. 2) , whereas type 3 MLN exhibits torsional nystagmus during binocular
viewing
(Fig. 3) . As in type 1 MLN, subjects with type 2 MLN and type 3 MLN always
displayed conjugate horizontal jerk MLN oscillations during monocular
viewing. A subject with type 4 MLN showed decelerating or linear
slow-phase jerk MLN waveforms during both binocular and monocular
viewing
(Fig. 4) . That is, unlike MLN types 1, 2 and 3, the type 4 waveform shape was
unaffected by monocular occlusion.
Occasionally some subjects displayed compound oscillations. For
example, the eye movements of a subject could exhibit
elements of both type 2 and type 4 MLN during binocular viewing
(Fig. 5) . We called this a type 2–type 4 MLN hybrid.
The distribution of the four different MLN types found in our randomly
chosen subject pool are shown in
Figure 6 . Type 3 MLN (32.4%) and type 4 MLN (48.7%) were found to be the most
common. No type 1 MLN was found in our study population. Two subjects
exhibited type 2 MLN in combination with either type 3 or type 4 MLN.
Apart from subjects belonging to the MLN hybrid group, subjects
exhibited a single constant MLN type throughout all recording sessions.
Six classes of MLN slow phases were distinguished
(Fig. 7) . Apart from the classic linear slow phase (class IIA) and
decreasing-velocity slow phase (class IA), we identified two with
saccadic elements (classes IB and IIB) and two with pendular components
(classes IIIA and B). A class I MLN slow phase has either a
conventional decreasing-velocity slow phase or a decreasing-velocity
slow phase with a preceding saccade. Class II MLN slow phases have
either a linear slow phase or a linear slow phase with a preceding
saccade, and a class III MLN slow phase exhibits strong pendular
components. The intensity of the pendular component determines whether
the class III MLN slow phase is a class IIIA or a class IIIB MLN slow
phase.
The relative incidences of the six slow phases strongly depended
on the viewing conditions and the MLN type of the subject. Histograms
illustrating the percentage of incidences of oscillations seen when
three subjects with different MLN types viewed a target in the primary
position are shown in
Figure 8 . The distributions seen during binocular viewing clearly defined the
MLN type. As expected, the distributions changed dramatically during
monocular viewing with none of the MLN types displaying a unique
distribution. The distributions of the classes of slow phases (classes
I–III) seen during monocular viewing occasionally varied from session
to session.
Figure 9 illustrates the eye movement traces when three subjects with types 2,
3, and 4 MLN monocularly viewed a target, in the primary position. The
effects of either removing the target or stabilizing the retinal image
are shown respectively in columns two and three. For all three MLN
types, removal of the target reduced the intensity of the MLN and
decreased the mean slow-phase velocity. The application of a fbg of+
1.0 also modified the nystagmus.
In an effort to explore the effect of visual feedback more
systematically, three feedback gain (fbg) conditions were investigated.
These were 0.0, 0.5, and 1.0 fbg and were equivalent to normal retinal
image motion (fbg = 0), a 50% reduction in retinal image motion
(fbg = 0.5), and a stabilized retinal image state (fbg =
1.0).
Figure 10 illustrates how these three test conditions affected the class of MLN
slow phases for three subjects with types 2, 3, and 4 MLN. For subjects
with type 2 MLN, the +0.5 and +1.0 fbgs greatly modified the subjects’
oscillations which reverted to those seen normally during binocular
viewing (i.e., square-wave jerks). In contrast, the +0.50 and +1.0 fbg
conditions did not appear to change the incidence of the dominant MLN
slow-phase class for subjects with type 3 and type 4 MLN.
The response to the introduction of a change in visual feedback
appeared to be specific to each subject, and
Figure 11 illustrates each of the six variations that we found. In
Figure 11A the
subject who had type 3 MLN never exhibited any apparent change in the
right-beating MLN. In comparison, the monocular left-beating nystagmus
seen before the onset of feedback in a subject with type 2 MLN, changed
to square-wave jerk oscillations when feedback was introduced
(Fig. 11B) . This response, whereby the monocular oscillations either
partially or completely reverted to the binocular state was typical of
all subjects with type 2 MLN. The same response was also elicited when
the target was removed during binocular viewing. Of note, when the
feedback was introduced, there was often a delay of up to 3 seconds
before the onset of the square-wave jerks. Thereafter, for the subject
in
Figure 11B the left eye slowly drifted nasally 4° or so. In the
third feedback-related affect
(Fig. 11C) , the introduction of feedback
to a subject with type 3 MLN brought about a large increase in the
amplitude of the right-beating MLN. Once again, there was a measurable
latency—2 seconds in this case—before the onset of the
large-amplitude oscillations.
Figure 11D illustrates how the
introduction of a +1.0 fbg to a subject with type 4 MLN brought about a
shift in the eye position of approximately 8°. This shift was
mediated by leftward saccades. By comparison, a subject with type 3 MLN
(Fig. 11E) who also showed a gaze shift, achieved the new eye position
by a series of slow eye movements. Finally, a subject with type 4 MLN
(Fig. 11F) exhibited a change in the nystagmus intensity which is
superimposed on a large-amplitude, low-frequency slow eye movement.
The saccade-mediated gaze shift seen in
Figure 11D for the
subject with type 4 MLN was investigated further by examining how the
level of the fbg influenced the change in the eye position.
Figures 12A 12B 12C 12D illustrate that the shift was directly related to the
level of the fbg, so that the +1.0 fbg condition brought about a 6°
shift in the eye position.
Removal of the visual feedback brought about a gaze shift in the
direction opposite to that seen when the feedback was introduced. This
eye position change was achieved by either an extended slow phase
(
Figs. 13A 13B ) or a saccade
(Fig. 13C) .
The purpose of this study was to perform both a qualitative and
quantitative examination of the waveform characteristics of MLN. We
believe that the four MLN types described in this study represent a
continuum from the stable binocular state of a subject with type 1 MLN
to the sustained MLN oscillations seen during binocular viewing for a
subject with type 4 MLN. These four categories have been based on the
binocular and monocular fixation characteristics of 37 subjects with
MLN when they viewed a stationary target presented in the primary
position.
In common with past reports, we concur that type 1 MLN (or LN) is
an uncommon occurrence.
2 It is likely that in the past,
small-intensity type 2 or type 3 MLN oscillations were classified as a
type 1 MLN because of the absence of high-resolution eye movement
recordings. The conventional waveform descriptions of MLN are typical
of type 3 and type 4 MLN.
4 5 6 In both cases a nystagmus is
present during both binocular and monocular viewing. It is the change
of the nystagmus from a torsional to a horizontal decelerating or
linear slow-phase jerk oscillation seen in type 3 MLN that
differentiates it from type 4 MLN. In the latter case, the nystagmus
remains principally in the horizontal plane, and the slow phase remains
decelerating or linear during both binocular and monocular viewing.
Using fundus video oculography, we found that in type 3 MLN, the
torsional nystagmus seen when both eyes were open was no longer
detectable during monocular fixation when the nystagmus motion was in
the horizontal plane.
The presence of square-wave jerks during binocular viewing
defines type 2 MLN and differentiates it from the other three types. It
is tempting to propose that type 2 MLN represents a stage between types
1, 3, and 4 MLN and may reflect the differences in the underlying
mechanisms responsible for each of the four categories of MLN. In a
previous study we stated that the incidence of physiological
square-wave jerks in the normal population is approximately
30%.
9 These saccadic oscillations were more commonly seen
when otherwise oculomotor normal subjects viewed the target in mesopic
conditions or when they were tired. To date, there have been two
reports of square-wave jerks preceding the postnatal appearance of
congenital nystagmus.
10 11
Occasionally, our subjects intentionally or unintentionally changed
from their binocular state to looking out of either eye. For subjects
with type 2 or type 3 MLN, this immediately changed the binocular
oscillations (i.e., square-wave jerks or torsional nystagmus) into an
oscillation with an MLN waveform and, depending on the fixing eye, a
change in the direction of the fast phase was noted. As can be seen in
Table 1 , none of the 37 subjects who took part in this study was truly
binocular, because all exhibited squints of one sort or another and
more than 30% had symmetrical or asymmetrical dissociated vertical
divergence.
None of the six classes of MLN slow phases described in this
study had an increasing velocity. The variety of slow phases that we
have found in MLN support the findings of earlier
reports.
3 6 12 The microsaccades seen in the class IB and
class IIB slow phases have been previously shown to be saccadic in
nature.
9 In agreement with other studies we encountered,
on the rare occasion, odd nystagmus beats with a runaway slow
phase.
3 4 5 6 The histograms seen in
Figure 8 clearly
illustrate that each subject did not exhibit a single unique MLN slow
phase class, but rather displayed a variety of classes, with one being
dominant. Great variations in the distribution of the classes of MLN
slow phase were apparent within our subject group. The great
variability of MLN slow phases is not surprising and is akin to that
seen in subjects with congenital nystagmus, when they can
exhibit up to four different congenital nystagmus
waveforms.
2 13 14 In both, the case of congenital and
manifest latent nystagmus, the slow phase appears to be strongly
influenced by the presence of the target
(Figs. 9 10) and to a lesser
degree whenever feedback is applied to the target
(Fig. 11) . In a
recent study we have shown that during periods of visual disengagement,
such as +1.0 fbg, there can be a decline in the slow-phase eye
velocity.
12 In addition, and in common with
previous studies,
6 12 13 14 15 16 we propose that
attentional and/or voluntary mechanisms can drive MLN or CN.
The eye position shifts triggered by either the introduction of or the
removal of visual feedback is idiosyncratic. Apart from subjects with
type 2 MLN, there appears to be no specific response for any one of the
other MLN types. However, the responses may provide insights into
saccadic programming.
17 Consider, for example,
Figure 11D where the switching off of visual feedback occurs before the quick
phase of the MLN. At this time the saccadic amplitude and velocity have
already been programmed, yet the saccadic amplitude is modified
midflight by the change in visual feedback. This is compatible with the
nonballistic behavior of some saccades, and a sampled data model of
saccadic generation predicts that it is possible to increase the
magnitude of a saccade during the first 70 msec of saccadic
programming.
4 18 19 20
It is noteworthy that the eye position shifts seen after the
removal of feedback were generally greater than those seen in response
to the original introduction of visual feedback
(Fig. 13) . This may
well reflect adaptation mechanisms, so that during feedback there was a
modification of slow eye movement control which became inappropriate
once the feedback was switched off. That is, the slow phases that make
up an MLN are not simply the result of a passive drift between fast
phases but are part of a continuous active control system.
The authors thank Jon Whittle for his comments and Anne Bjerre for
her assistance.