Detection of VOR dysfunction during the gaze stabilization test: Does target size matter?

Abstract

BACKGROUND:

The Gaze Stabilization Test (GST) identifies vestibulo-ocular reflex (VOR) dysfunction using a decline in target recognition with increasing head velocity, but there is no consensus on target (optotype) size above static visual acuity.

OBJECTIVE:

To determine the optimal optotype size above static visual acuity to be used during the GST in subjects with unilateral vestibular dysfunction and healthy individuals.

METHODS:

Eight subjects with unilateral vestibular dysfunction (UVD) and 19 age-matched, healthy control subjects were studied with the standard GST protocol using two optotype sizes, 0.2 and 0.3 logMAR above static visual acuity (ΔlogMAR). Maximal head velocity achieved while maintaining fixation on both optotypes was measured. Sensitivity, specificity and receiver-operator characteristic area under the curve (ROC AUC) analyses were performed to determine the optimal head velocity cut off point for each optotype, based on ability to identify the lesioned side of the UVD group from the control group.

RESULTS:

There was a significant difference in maximal head velocity between the UVD group and control group using 0.2 ΔlogMAR (p = 0.032) but not 0.3 ΔlogMAR (p = 0.061). While both targets produced similar specificities (90%) for distinguishing normal from subjects with UVD, 0.2 ΔlogMAR targets yielded higher sensitivity (75%) than 0.3 logMAR (63%) and accuracy (86% vs 80%, respectively) in detecting the lesioned side in subjects with UVD versus controls with maximal head velocities≤105 deg/s (p = 0.017). Furthermore, positive likelihood ratios were nearly twice as high when using 0.2 ΔlogMAR targets (+ LR 10) compared to 0.3 ΔlogMAR (+ LR 6.3).

CONCLUSION:

The 0.2 ΔlogMAR optotype demonstrated significantly superior identification of subjects with UVD, better sensitivity and positive likelihood ratios than 0.3 ΔlogMAR for detection of VOR dysfunction. Using a target size 0.2logMAR above static visual acuity (ΔlogMAR) during GST may yield better detection of VOR dysfunction to serve as a baseline for gaze stabilization rehabilitation therapy.

Keywords

Gaze stabilization test vestibular testing vestibulo-ocular reflex (VOR)visual acuity vestibular rehabilitation

1 Introduction

The vestibulo-ocular reflex (VOR) allows for maintenance of clear vision with natural head movements (e.g. locomotion, rapid head turns). To avoid oscillopsia (blurred or jumping vision), the VOR produces compensatory eye movements equal and opposite to head rotation to maintain visual targets on the fovea. Everyday activities, such as walking and running, elicit high velocity head rotations for which images on the retina must be stabilized [6, 15]. Subjects with loss of vestibular function struggle to maintain gaze stability with high velocity head movements and may compensate with non-vestibular gaze mechanisms, such as smooth pursuit at velocities < 90 deg/s or central pre-programmed saccadic eye movements [3, 32]. However, such compensation mechanisms become inadequate when head velocity exceeds 90 degrees per second and, as a consequence, visual acuity worsens for those with vestibular disease [7 , 13].

The Gaze Stabilization Test (GST) was developed as a reliable, computerized functional assessment of VOR contribution to visual acuity by measuring peak head velocity while maintaining visual acuity [7]. Previously, the Dynamic Illegible “E” test assessed VOR dysfunction clinically by determining the change in visual acuity during high velocity head movements, classically defining an abnormal test as a drop of at least three lines on a standard Snellen chart [17]. Subsequently, to allow for control of predictive behavior and minimize pre-programmed eye movements, the computerized Dynamic Visual Acuity Test (DVAT) and GST were developed to precisely monitor head velocity by presenting a target (optotype) at random, brief periods to minimize pre-programmed eye movements while employing high frequency (1–4 Hz) head movements, like those required during daily life activities (ADLs) [36]. The DVAT uses a single head velocity threshold, typically above 85 deg/s [24, 36]. while optotype size varies according to performance. In contrast, the GST uses increments of peak head velocities based on performance with a fixed optotype size 2-3 lines above static acuity. Thus, the GST provides clinicians with velocity-specific data associated with the VOR contribution to gaze stabilization.

It has been suggested that protocols similar to the GST lead to enhanced VOR adaptation that may result in improved patient gaze stability during activities of daily living (ADL) following vestibular rehabilitation therapy (VRT) [27]. However, to date, there is no consensus on the best optotype size to use in GST. The standard protocol consists of presenting the patient with an optotype 2-3 lines (0.2 to 0.3 ΔlogMAR) above static visual acuity (SVA). Most studies have utilized an optotype two lines above SVA while others have opted for three lines above SVA (Table 1). In this study, the 0.2 ΔlogMAR optotype was compared to 0.3 ΔlogMAR in order to determine which optotype size was better at distinguishing subjects with impaired VOR function from healthy control subjects.

Table 1
Optotype sizes used in prior GST research studies and variance in achieved head velocities

Average head velocities (deg/sec)

Healthy controls (n) Patients (n) % UVD p<0.05?

0.2 ΔlogMAR optotype

Kaufman et al. [17] 197.0 (50) – – –

Honaker et al. [15]^# 123.25 (15) 74.64 (15) 0 yes

Lee and Honaker [18] 149.5 (20) – – –

Ward et al. [42]^$ 123– 157 (40) – – –

Ward et al. [43] 95.5 (86) – – –

Honaker and Shepard [16]^$ 114.7– 154.7 (87) – – –

Whitney et al. [44]^dotplus 124.4 (20) 79.4 (12) 92 yes

Pritcher et al. [31]^& 107– 127 (40) 90– 93 (15) 80 yes

0.25 ΔlogMAR optotype

Voelker et al. [41] 141.3 (17) 86.3– 105.7^ł (17) 100 yes

Mohammad et al. [27] – 124– 152 (16) 75

0.30 ΔlogMAR optotype

Goebel et al. [7] 147 (14) 83.85– 112.45^ł (14) 100 yes

	Average head velocities (deg/sec)
0.2 ΔlogMAR optotype
Kaufman et al. [17]	197.0 (50)	–	–	–
Honaker et al. [15]^#	123.25 (15)	74.64 (15)	0	yes
Lee and Honaker [18]	149.5 (20)	–	–	–
Ward et al. [42]^$	123– 157 (40)	–	–	–
Ward et al. [43]	95.5 (86)	–	–	–
Honaker and Shepard [16]^$	114.7– 154.7 (87)	–	–	–
Whitney et al. [44]^dotplus	124.4 (20)	79.4 (12)	92	yes
Pritcher et al. [31]^&	107– 127 (40)	90– 93 (15)	80	yes
0.25 ΔlogMAR optotype
Voelker et al. [41]	141.3 (17)	86.3– 105.7^ł (17)	100	yes
Mohammad et al. [27]	–	124– 152 (16)	75
0.30 ΔlogMAR optotype
Goebel et al. [7]	147 (14)	83.85– 112.45^ł (14)	100	yes

^&Left and right-sided data separated in study. ^$Young and old controls separated in study. ^#Based on Dynamic Gait Index fall risk. ^dotplusResults from “New Gaze Stabilization Test”. ^łLesioned vs non-lesioned side separated in study.

2 Materials and methods

2.1 Study subjects

After obtaining approval from the Human Research Protection Office of Washington University in Saint Louis, subjects were recruited from a database of subjects seen by providers diagnosed with unilateral vestibular loss at the principal investigator’s (J.A.G.) clinical institution. All subjects had documented unilateral vestibular dysfunction (UVD) based on greater than or equal to 49% bithermal binaural caloric test asymmetry on closed loop irrigation at the same institution, exceeding the clinical standard of greater than 30% asymmetry. Control subjects were recruited from the Volunteer for Health Database at Washington University School of Medicine and did not have a history if imbalance or vestibular dysfunction. All control subjects were age-matched to subjects with UVD. All study subjects were 18 years of age or older, had no history of cervical dysfunction restricting them from rapidly moving their head, had no history of a central nervous system disorder, and had normal corrected vision. The exact etiology of subjects’ vestibular loss, the duration of symptoms, or the extent of any vestibular rehabilitation therapy were unavailable for review at the time of this study nor for retrospective review following the study. As such, subjects were not explicitly excluded from this study if they had a history of vestibular rehabilitation therapy.

2.2 Gaze stabilization (GST) protocol

The two study groups, controls and subjects with UVD, participated in identical GST procedures.

The VisionTracker System (Bertec Corporation, Columbus, OH, USA) was used to present the optotype to the subject via a laptop computer throughout the test. The subject was seated in a chair located 10 feet from a computer screen. This optotype presentation distance allowed for minimal change in viewing angle due to variations in subject height. The study was conducted with room lights turned off to minimize any distraction to the subject. All subject responses were recorded by the test administrator by using a wireless remote.

Each subject underwent six randomized trials, each consisting of three parts: Static Visual Acuity (SVA) testing, Visual Processing Time (VPT) test, and the Gaze Stabilization Test (GST).

SVA was measured by having study subjects maintain gaze on a circle in the middle of the computer screen using best corrected vision. The optotype (letter “E”) was presented for a short period of time (1.5 s), then study subjects were asked to report the direction optotype (up, down, right, left). The size of the optotype was changed until the subject’s static visual acuity was determined using a computer-generated modified Hughson-Westlake procedure (i.e. 0.05 logMAR increase for incorrect/non-responses and 0.10 logMAR decrease for correct responses until threshold reached) [5]. Visual acuity was recorded in log of the minimal angle of resolution (logMAR) wherein 0.0 logMAR corresponded to 20/20 on the Snellen chart. Based on the subject’s determined SVA, the optotype used for the rest of the GST protocol was adjusted to either 0.2 (two lines) or 0.3 (three lines) logMar above their static acuity (ΔlogMAR).

Next, study subjects underwent the Visual Processing Time (VPT) test. This test determined the minimum target duration required to perceive the optotype on the retina. The optotype was presented for variable duration (30 ms–70 ms), and study subjects were asked to report the target orientation with their head still. A cutoff of < 70 ms was used to document time for target recognition to avoid any potential non-vestibular pre-programmed eye movements that could occur with target presentation greater than 70 ms during the GST. Once the VPT was determined, the Visual Display Time (VDT) was calculated as VDT = VPT + 35 ms, up to a maximum of 70 ms. Then, the optotype was displayed for the fixed duration of VDT during the GST protocol. No subject had a VDT greater than 70 ms.

Finally, the Gaze Stabilization Test (GST) required study subjects to use active head rotations in the horizontal (yaw) plane to determine the maximum head velocity at which the subject maintained visual acuity. This test was performed using both 0.2 and 0.3 ΔlogMar in a randomized order, 3 times each (6 trials overall). In one study subject, the test administrator utilized passive head movements to maintain appropriate head velocity and amplitude since head accelerations during active rotations remained too high despite prompting by a metronome to appropriately trigger the optotype presentation. Study subjects wore a headband with a lightweight, wireless rate sensor that communicated with software (Balance Advantage [Bertec Corporation, Columbus, OH, USA] and 3-Space Wireless 2.4 GHz rate-sensor (TSS-WL v2.0, YEI Technology, Inc [Portsmouth, Ohio, USA]) which monitored the velocity of head rotation and presented the optotype when desired head velocity was reached. In all instances, optotype orientation and timing of presentation were unpredictable. The duration of optotype presentation was based on the study subject’s VDT. As before, study subjects were asked to report the orientation of the optotype after each presentation. For each trial, study subjects were prompted by the software, using an audible metronome, to move their head at varying speeds to determine the highest velocity at which they can report 3/5 correct responses, using a modified Hughson-Westlake algorithm [5]. Possible target head velocities ranged from 30–150 deg/sec. Of note, the test administrator was blinded to the lesioned side when testing subjects with UVD. While the test administrator was not blinded to the study subject’s group (control vs UVD), the severity of vestibular dysfunction which has been shown to have a direct relationship on dynamic visual acuity was not known to the tester [16].

2.3 Statistical analysis

Power and sample size estimates for the study were based the original GST study by Goebel et al. [7]. To estimate the expected between group differences for the GST, published data from lesioned side of their subjects with UVD (mean = 83.45, SD = 28.66 deg/s) was compared to their healthy controls (mean = 147.40, SD = 38.45). To detect this 64-point mean difference, at a power of 80% using a two-sided significance level of 5%, six subjects per group were needed. Standard descriptive statistics were used to describe the study population and the distribution of scores for each of the tests. An independent t-test was used to compare SVA and VPT measurements between controls and subjects with UVD. A dependent t-test was used to compare SVA, VPT, and GST measurements within each study group trials when using the 0.2 ΔlogMAR or 0.3 ΔlogMAR optotype as well as comparing measurements obtained from the lesioned vs non-lesioned side of subjects with UVDs. An independent t-test was used to compare the mean highest achieved GST scores between control and subjects using the 0.2 ΔlogMAR or 0.3 ΔlogMAR optotype. Effect size was estimated by Cohen’s d, where 0.2 = small effect, 0.5 = medium effect, and 1.0 = large effect [30]. As there were no significant difference between left and right-sided scores for the control subjects, GST scores in this group were calculated as the mean of combined left and right-sided measurements. For subjects with UVD, the GST data was separated into rightward and leftward head movement for analysis of the lesioned and non-lesioned side tested. Receiver operating characteristic (ROC) curves were used to assess the accuracy of using either optotype for the GST and to identify the ideal cutoff values with the best discrimination between controls and subjects. The ROC is a plot of sensitivity versus 1-specificity of the values of a test. The area under the ROC curve (AUC) ranges from 0.5 (accuracy not better than obtained by chance alone) to 1 (perfect accuracy). Each point on the ROC curve can serve as a cutoff point with its own sensitivity and specificity value. Theoretically, the optimal cutoff is the point with the best combination of sensitivity and specificity. This cutoff value was chosen for each test to dichotomize the test values into normal and abnormal. In this study, accuracy, based on the cut off, is derived from the formula (TP + TN)/(TP + TN + FP + FN), where TP = True positive; FP = False positive; TN = True negative; FN = False negative. The alpha level for all tests was set at 0.05. All statistical analysis was performed using the IBM SPSS Statistics, 21.0 (IBM Corp, Armonk, NY).

3 Results

Nineteen healthy control subjects (median age, 63; range 52 –77 yrs) and 9 subjects with UVD (median age, 65; range 45 –77yrs) were enrolled in this study. There was one patient with UVD who did not complete trials using the 0.2 ΔlogMAR optotype due to testing fatigue and was excluded from the analysis, leaving 8 subjects with UVD. Female participation consisted of 74% of the control group and 67% of the UVD group. There was no significant difference in age between the study groups. All caloric asymmetry values for the UVD group were greater than 71% (median, 100%; range, 71–100%). For 7 UVD subjects, there was an average of 1.33 years from the date of caloric testing to GST testing for the UVD group (range 2 months-2 years and 4 months). For one subject, the date of caloric test could not be located due to changes this study’s institution electronic medical record. Further demographic details on the subjects with UVD can be found in Table 2.

Table 2
Demographics of subjects with unilateral vestibular dysfunction (N = 8)

Subject # Age Sex % Caloric weakness Side of UVD Duration from caloric test (days)

1 56 M 87 Right 649

2 72 M 100 Left 193

3 77 F 71 Right 754

4 45 F 100 Right 694

5 70 F 100 Left 69

6 55 F 100 Right 188

7 69 M 100 Left Not available

8 69 M 80 Right 866

Subject #	Age	Sex	% Caloric weakness	Side of UVD	Duration from caloric test (days)
1	56	M	87	Right	649
2	72	M	100	Left	193
3	77	F	71	Right	754
4	45	F	100	Right	694
5	70	F	100	Left	69
6	55	F	100	Right	188
7	69	M	100	Left	Not available
8	69	M	80	Right	866

3.1 Static visual acuity (SVA) and visual processing time (VPT)

Table 3 summarizes distribution of the median scores for SVA and VPT for both optotypes. Overall, when using the 0.2 ΔlogMAR optotype, there was no significant difference in SVA between the control (mean = 0.117, SD = 0.130) and UVD group (mean = 0.185, SD = 0.170, t(25) = –1.13, p = 0.268, d = 0.48), nor a significant difference in VPT between the control (mean = 34.0, SD = 5.85) and UVD group (M = 33.4, SD = 5.57, t(25) = 0.245, p =0.808, d = 0.10, respectively). When using the 0.3 ΔlogMAR optotype, there was no significant difference in SVA between the control (mean =0.095, SD = 0.125) and UVD group (mean = 0.178, SD = 0.163, (t(25) = –1.37, p = 0.66, d = 0.58), nor a significant difference in VPT between the control (mean = 32.9, SD = 5.67) and UVD group (mean =30.9, SD = 2.65, t(25) = 0.906, p = 0.373, d = 0.38).

Table 3
Distribution of test scores for 0.2 and 0.3 ΔlogMAR optotypes

Test UVD group Control group^*

0.2 ΔlogMAR 0.3 ΔlogMAR P value 0.2 ΔlogMAR 0.3 ΔlogMAR P value

SVA (logMAR)

Mean (SD) 0.18 (±0.17) 0.18 (±0.16) 0.574 0.12 (±0.13) 0.10 (±0.13) 0.207

VPT (ms)

Mean (SD) 33.44 (±5.57) 30.94 (±2.65) 1.92 34.03 (±5.85) 32.85 (±5.67) 0.510

GST (deg/sec)

UVD Lesioned side^∧

Mean (SD) 90.50 (±39.98) 109.63 (±39.87) 0.076 128.71 (±20.33) 141.75 (±22.5) 0.001

UVD Non-lesioned side^∧

Mean (SD) 119.70 (±44.16) 128.50 (±37.21) 0.338

Test	UVD group	Control group^*
SVA (logMAR)
Mean (SD)	0.18 (±0.17)	0.18 (±0.16)	0.574	0.12 (±0.13)	0.10 (±0.13)	0.207
VPT (ms)
Mean (SD)	33.44 (±5.57)	30.94 (±2.65)	1.92	34.03 (±5.85)	32.85 (±5.67)	0.510
GST (deg/sec)
UVD Lesioned side^∧
Mean (SD)	90.50 (±39.98)	109.63 (±39.87)	0.076	128.71 (±20.33)	141.75 (±22.5)	0.001
UVD Non-lesioned side^∧
Mean (SD)	119.70 (±44.16)	128.50 (±37.21)	0.338

^*Scores were calculated as the mean of the right and left-side measurements. ^∧Within-group UVD lesion vsnon-lesioned side GST comparison p-values = 0.002 and 0.092 for 0.2ΔlogMAR and 0.3ΔlogMAR optotypes, respectively.

3.2 Gaze stabilization test (GST)

Figure 1 summarizes the data for both groups using the two optotypes. When utilizing the larger optotype size (0.3 ΔlogMAR), both study groups were able to achieve higher head velocities while maintaining visual acuity. Control subjects using the 0.3 ΔlogMAR optotype achieved a significantly higher mean maximal head velocity of 141.75 deg/s (SD 22.5) than using the 0.2 ΔlogMAR optotype (mean = 128.71 deg/s; SD = 20.3, t(18) = 3.80, p = 0.032, d = 0.61). In contrast, for subjects with UVD, there was no significant difference comparing 0.2 ΔlogMAR (mean = 90.50 deg/s, SD = 40.0) or 0.3 ΔlogMAR (mean = 109.63 deg/s, SD = 39.9) peak head velocities on the lesioned side (t(7) = –2.07, p = 0.076, d = 0.48). On the non-lesioned side, there was also no significant difference between 0.2 ΔlogMAR (mean = 119.71 deg/s, SD = 44.2) or 0.3 ΔlogMAR (mean = 128.50 deg/s, SD = 37.2) peak head velocities (t(7) = –1.03, p = 0.338, d = 0.22). In addition, Fig. 1 also demonstrates wider variability among subjects with UVD when using the 0.2 ΔlogMAR optotype, regardless of side tested. When compared to testing the non-lesioned side, peak head velocity was significantly lower on the lesioned side when utilizing the 0.2 ΔlogMAR optotype (t(7) = –5.01, p = 0.002, d = 0.69), but not for 0.3 ΔlogMAR optotype (t(7) = –1.95, p = 0.092, d = 0.49).

Fig. 1

Mean GST scores and variability (in SD) for healthy controls and UVD subjects using 0.2 and 0.3 ΔlogMAR optotypes; (a.) 0.2 ΔlogMAR optotype, control vs UVD Lesioned side; (b.) 0.2 ΔlogMAR optotype, control vs UVD Non-lesioned side; (c.): 0.3 ΔlogMAR optotype, control vs UVD Lesioned side; (d.) 0.3 ΔlogMAR optotype, control vs UVD Non-lesioned side.

3.3 Performance analyses of optotypes used in the gaze stabilization test (GST)

Receiver operating characteristic (ROC) curve analyses were performed to measure the performance of both optotype sizes in detecting subjects with UVD based on various head velocity thresholds (Fig. 2). The optimal cut off point for each optotype was determined based on ability to identify the lesioned side of the UVD group from the control group. This consisted of choosing a maximal head velocity with the best combination of sensitivity and specificity, properly classifying test values into normal or abnormal. The 0.2 ΔlogMAR optotype was slightly more accurate than the 0.3ΔlogMAR optotype in distinguishing the lesioned side of the UVD group from the controls, attributed mostly to the former’s higher sensitivity.

Fig. 2

Receiver operator characteristic (ROC) curve for GST predicting unilateral vestibular loss (UVL) when 0.2(a.) and 0.3 ΔlogMAR (b.) optotypes are utilized. The chosen cutoff is designated by the arrow. A GST value of ≤105 deg/s represents an abnormal test result for 0.2 ΔlogMAR optoype (a.) and ≤126 deg/s for the 0.3 ΔlogMAR optotype (b.).

Head velocities less than or equal to 105.92 deg/s using a 0.2 ΔlogMAR optotype were categorized as abnormal VOR function with 75% sensitivity and 90% specificity. This amounted to an accuracy of 86%, a positive likelihood ratio of 10 and negative likelihood ratio of 0.27. Head velocities less than or equal to 126.33 deg/s using a 0.3 ΔlogMAR optotype demonstrated lower sensitivity (63%), similar specificity (90%), and was able to maintain a consistent specificity and accuracy of 80% with increased head velocity. Positive likelihood ratio of 6.3 was lower than the 0.2 ΔlogMAR optotype and the negative likelihood ratio was higher at 0.41.

Overall, both optotype sizes performed well and had relatively similar area under the curve (AUC) measurements, as the absolute difference in AUC values was only 0.046. The AUC corresponds to the percent chance that maximal head velocity achieved using the specified optotype during the GST can distinguish between a patient with UVD and a healthy control.

ROC curves when using 0.2 ΔlogMAR and 0.3 ΔlogMAR optotypes to determine the non-lesioned side of subjects with UVD from controls did not reach statistical significance (p = 1.00 and 0.457, respectively) and were not used for further performance analysis provided results would yield no better accuracy than chance.

4 Discussion

This study demonstrated that while 0.2 ΔlogMAR and 0.3 ΔlogMAR optotypes performed in similar fashion for UVD and control groups, the optotype 0.2 logMAR above static visual acuity produced a higher sensitivity and higher likelihood ratio than 0.3 log MAR above static visual acuit for both study subject groups. Of note, the 0.2 ΔlogMAR optotype is smaller than the published target size used in the Dynamic Illegible “E” test or the first GST study [7, 17]. The results of the present study suggest that using the 0.2 ΔlogMAR optotype may be the better target size to use during the GST to evaluate VOR impairment prior to gaze stabilization therapy and to monitor progress.

4.1 Comparing optotype performance metrics

While both optotype sizes performed reasonably at detecting UVD, yielding the same specificity, the optimal GST cut off values differed greatly and may be of diagnostic interest. Compared to utilizing the 0.3 ΔlogMAR optotype, a lower maximal achieved head velocity (105 deg/s) provided the most optimal cutoff point to discriminate between controls and subjects with UVD when utilizing the 0.2 ΔlogMAR optotype. By using a cut-off point of 126.3 deg/s, the 0.3Δ logMAR optotype used in this study performed similar to Goebel et al. [7], who found similar sensitivity and specificity, while using the 0.3 ΔlogMAR optotype with a 90.5 deg/s cut off point. This suggests that the 0.3 ΔlogMAR optotype may create a test too easy for subjects with unilateral vestibular dysfunction to achieve adequate gaze stabilization at higher head velocities due to the increased target size.

4.2 Subject-based GST performance

Both healthy subjects and subjects with UVD maintained gaze stabilization at higher head velocities when the larger optotype, 0.3 ΔlogMAR above SVA, was used during the GST. As expected, maximal head velocities achieved during the GST were significantly lower for subjects with UVD regardless of the optotype chosen. Also, there was more variation in head velocities elicited by subjects with UVD, which may be due to variations in vestibular compensation among that group, with a 1.33-year average duration from caloric diagnostic testing. It is unclear how this may have affected the study results, as this study was not designed to investigate the effect of vestibular compensation nor rehabilitation strategies on GST results. Nonetheless, maximal head velocities among subjects with UVD in this study were not statistically different when comparing results using the 0.3 and 0.2 ΔlogMAR optotypes, which may be due to this study’s small subject sample size, the undefined, yet presumably heterogenous level of compensation among the subject sample, or both.

To our knowledge, this is the first case-control study investigating the effect of optotype size used in the GST. Mohammad et al. [23] looked at the effect of optotype on GST performance using various combinations of size and perception time among only healthy controls. They found healthy controls performed better on the GST with the 0.3 ΔlogMAR optotype in the horizontal plane. The authors suggested that subjects were more likely to perform worse when GST used the smaller 0.2 ΔlogMAR optotype because the target was too small for the subjects to recognize, even at slower head velocities. This notion may infer differences in foveation based on target size [10, 34].

4.3 GST expectations for unilateral vestibular disease vs healthy controls

The present study produced similar results as previous studies looking at GST maximal achieved head velocities in subjects with vestibular disease and healthy controls (Table 1). These results showed a similar spread in head velocities among subjects with UVD using the 0.2 ΔlogMAR optotype, but average GST head velocities were closest to those of Whitney et al. [39]. Nonetheless, as in the present study, lower GST head velocities (< 100 deg/sec) in subjects with vestibular dysfunction were reported in nearly every study which used the 0.2 ΔlogMAR optotype (Table 1).

When subjects obtain lower GST peak velocities (< 100 deg/sec), there is likely significant contribution from compensatory non-VOR gaze stabilization strategies such as catch -up saccades and smooth pursuit [33]. Catch-up saccades and smooth pursuit work synergistically to maintain foveation by changing the distance of eye placement relative to the target center [26, 40]. Studies show that while the smooth pursuit mechanism of large target size is primarily driven by integrated motion signals, [18] catch-up saccade frequency must increase to foveate small objects [10]. This may be because smaller objects are known to produce weak signals in the primary motion processing region of the brain, the middle temporal visual area [4]. As it relates to this study or other studies using dynamic visual acuity testing, one should be cautious when using optotypes smaller than 0.2 ΔlogMAR, as results may be confounded the increased contribution of catch-up saccades to the central angular VOR (aVOR), provided the extent of vestibular compensation from subjects with UVD. Alternatively, optotypes larger than 0.3 ΔlogMAR may allow for increased contribution of smooth pursuit to the aVOR, and, therefore, may allow for inaccurate assessment of the peripheral aVOR pathway.

4.4 Effect of lesioned vs non-lesioned side on GST

Interestingly, this study did not find a statistical difference between maximal achieved head velocities between non-lesioned and lesioned side of the UVD group using the 0.3 ΔlogMAR optotype. Prior studies comparing non-lesioned side during GST found significant asymmetry [7, 36]. While it is known that the VOR on the non-lesioned side can be mildly deficient following unilateral vestibular loss, [9] significant compensation can be expected for low and mid-velocity rotations [29]. However, the non-lesioned side will show marginal VOR gain decline during high-velocity head movements. This decreased VOR gain is thought to be due to a combination of reduced tonic firing of afferents in the contralesional canal, inability for the CNS to compensate for the loss of ipsilesional inhibitory signals, secondary to Eward’s Second Law, as well as reduced sensitivity of Type I neurons in the contralesional vestibular nuclei [25 , 35]. However, studies in cats with unilateral vestibular loss show that the central vestibular neurons retain their rotational sensitivity, [14] evidenced by improvements seen in gaze stabilization exercises [14, 33]. Badaracco et al. [1] showed that behavioral compensation via vestibular physical therapy, consisting of adaption, substitution, and habitualization, can significantly increase the achieved maximal head velocities during GST. Thus, the insignificant asymmetry found using the 0.3 ΔlogMAR optotype in this study is likely due to compensation.

4.5 Implications for vestibular rehabilitation

In addition to gait and balance testing, symptomatic and functional improvement in subjects with unilateral vestibular loss undergoing VRT largely involves outcomes from validated, self-assessments (Vertigo Symptom Scale, Dizziness Handicap Inventory) [19]. Excluding vestibular neuritis, one would not expect to see any substantial improvements in low-frequency vestibular function tests (i.e. calorics or rotatory chair testing). To better trend objective progression in gaze stabilization, vestibular physical therapists have found improvements in functional patient outcomes by utilizing objective measurements of dynamic visual acuity, such as the DVAT, video head impulse testing (VHIT), and GST [1 , 31] Significant improvements in VOR gain observed during VHIT, which involves passive head rotations, are not commonly observed after in those with chronic unilateral vestibular hypofunction or those with acute dysfunction undergoing late gaze stabilization exercises [21, 33]. While both the DVAT and GST quantify known vestibular dysfunction well, Voelker et al. [36] found the GST was more specific at identifying abnormal gaze stability than the DVAT. The present study suggests that utilizing the smaller 0.2 ΔlogMAR optotype during the GST may yield more accurate results. Furthermore, the GST may provide enhanced benefit for monitoring the effect of gaze stabilization exercises since its protocol is similar to that of the “VOR x 1”’ exercises where the patient is expected to progress with increased head velocity while maintaining fixation [20]. Finally, when using the GST in a VRT capacity, this study supports the notion that by using the smaller 0.2 ΔlogMAR optotype with subjects, clinicians may better eliminate the contribution of smooth pursuit to compensatory gaze stability strategies, given mean peak velocities reached above 100 deg/s [33].

4.6 Limitations

This study had certain limitations which should be considering when interpreting the data. There was no control for variations in vestibular compensation that may have been seen in the data, as subjects with a history of previous or ongoing therapy for their UVD were not excluded. In addition, inclusion of subjects with UVD was not based on the etiology or duration of their UVD. There may also have been a component of non-vestibular related mechanisms, such as vestibular migraine or persistent postural perceptual dizziness (PPPD), that was not explicitly screened during the time of testing which may have affected test performance in both groups [2, 11]. The study had many subjects over 60, wherein prior studies have shown GST measurements to be considerably lower in healthy subjects in this age group [37, 38]. While this does not affect the validity of the present study, as subjects with UVD were age-matched to controls, average maximal head velocities reported here may differ from those found in younger adults. One study subject required passive head rotation, but this should not have affected their performance, as the timing and direction of the optotype presentation were randomized as to limit predictability. Finally, this study did not examine changes in GST scores with either optotype during gaze stabilization therapy to directly assess the performance of one optotype over the other.

5 Conclusion

Target size 0.2 logMAR above static visual acuit (ΔlogMAR) yielded a significant difference in maximum head velocities between healthy control subjects and subjects with UVD during the GST. Additionally, the 0.2 ΔlogMAR optotype was more sensitive and demonstrated a greater likelihood ratio for identifying the side of vestibular loss than 0.3 ΔlogMAR. This finding suggests using 0.2 ΔlogMAR during the GST in subjects with known unilateral VOR dysfunction to establish a baseline prior to and while undergoing vestibular rehabilitation therapy. Future research is warranted to determine the effect of optotype size on changes in peak head velocities for subjects with UVD undergoing a structured vestibular rehabilitation program.

Level of evidence

III

Conflicts of interest

Joel A Goebel, MD has served as a consultant for Barron Associates and received lecture honoraria from Micromedical Technologies. None of the authors declare any sponsorship or financial relationships related to this study.

References

Badaracco

, Labini

F.S.

, Meli

, et al., Vestibular rehabilitation outcomes in chronic vertiginous patients through computerized dynamic visual acuity and Gaze stabilization test, Otol Neurotol 28 (2007), 809–813.

Baker

B.J.

, Curtis

, Trueblood

, et al., Vestibular functioning and migraine: comparing those with and without vertigo to a normal population, J Laryngol Otol 127 (2013), 1169–1176.

Bhansali

S.A.

, Stockwell

C.W.

, Bojrab

D.I.

, Oscillopsia in patients with loss of vestibular function, Otolaryngol Head Neck Surg 109 (1993), 120–125.

Britten

K.H.

, Heuer

H.W.

, Spatial summation in the receptive fields of MT neurons, J Neurosci 19 (1999), 5074–5084.

Carhart

, Jerger

J.F.

, Preferred Method For Clinical Determination Of Pure-Tone Thresholds, Journal of Speech and Hearing Disorders 24 (1959), 330–345.

Crane

B.T.

, Demer

J.L.

, Human gaze stabilization during natural activities: translation, rotation, magnification, and target distance effects, J Neurophysiol 78 (1997), 2129–2144.

Goebel

J.A.

, Tungsiripat

, Sinks

, et al., Gaze stabilization test: a new clinical test of unilateral vestibular dysfunction, Otol Neurotol 28 (2007), 68–73.

Gottshall

K.R.

, Hoffer

M.E.

, Tracking recovery of vestibular function in individuals with blast-induced head trauma using vestibular-visual-cognitive interaction tests, J Neurol Phys Ther 34 (2010), 94–97.

Halmagyi

G.M.

, Curthoys

I.S.

, Cremer

P.D.

, et al., The human horizontal vestibulo-ocular reflex in response to high-acceleration stimulation before and after unilateral vestibular neurectomy, Exp Brain Res 81 (1990), 479–490.

10.

Heinen

S.J.

, Potapchuk

, Watamaniuk

S.N.J.

, A foveal target increases catch-up saccade frequency during smooth pursuit, Journal of Neurophysiology 115 (2016), 1220–1227.

11.

Helm

M.R.

, Vestibulo-ocular reflex abnormalities in patients with migraine, Headache 45 (2005), 332–336.

12.

Herdman

S.J.

, Schubert

M.C.

, Das

V.E.

, et al., Recovery of dynamic visual acuity in unilateral vestibular hypofunction, Arch Otolaryngol Head Neck Surg 129 (2003), 819–824.

13.

Herdman

S.J.

, Tusa

R.J.

, Blatt

, et al., Computerized dynamic visual acuity test in the assessment of vestibular deficits, Am J Otol 19 (1998), 790–796.

14.

Heskin-Sweezie

, Farrow

, Broussard

D.M.

, Adaptive rescaling of central sensorimotor signals is preserved after unilateral vestibular damage, Brain Res 1143 (2007), 132–142.

15.

Honaker

J.A.

, Lee

, Shepard

N.T.

, Clinical use of the gaze stabilization test for screening falling risk in community-dwelling older adults, Otol Neurotol 34 (2013), 729–735.

16.

Honaker

J.A.

, Shepard

N.T.

, Age effect on the Gaze Stabilization test, J Vestib Res 20 (2010), 357–362.

17.

Kaufman

D.R.

, Puckett

M.J.

, Smith

M.J.

, et al., Test-retest reliability and responsiveness of gaze stability and dynamic visual acuity in high school and college football players, Phys Ther Sport 15 (2014), 181–188.

18.

Lee

, Honaker

J.A.

, Development of a new gaze stabilization test, J Vestib Res 23 (2013), 77–84.

19.

Leigh JR

Z.D.

, The neurology of eye movements, Oxford University, New York, 2006.

20.

Longridge

N.S.

, Mallinson

A.I.

, The dynamic illegible E (DIE) test: a simple technique for assessing the ability of the vestibulo-ocular reflex to overcome vestibular pathology, J Otolaryngol 16 (1987), 97–103.

21.

Longridge

N.S.

, Mallinson

A.I.

, The dynamic illegible E-test. A technique for assessing the vestibulo-ocular reflex, Acta Otolaryngol 103 (1987), 273–279.

22.

, Watamaniuk

S.N.J.

, Heinen

S.J.

, Illusory motion reveals velocity matching, not foveation, drives smooth pursuit of large objects, J Vis 17 (2017), 20.

23.

McDonnell

M.N.

, Hillier

S.L.

, Vestibular rehabilitation for unilateral peripheral vestibular dysfunction, Cochrane Database Syst Rev 1 (2015), Cd005397.

24.

Meldrum

, Jahn

, Gaze stabilisation exercises in vestibular rehabilitation: review of the evidence and recent clinical advances, J Neurol 266 (2019), 11–18.

25.

Michel

, Laurent

, Alain

, Rehabilitation of dynamic visual acuity in patients with unilateral vestibular hypofunction: earlier is better, Eur Arch Otorhinolaryngol 277 (2020), 103–113.

26.

Mitsutake

, Sakamoto

, Ueta

, et al., Effects of vestibular rehabilitation on gait performance in poststroke patients: a pilot randomized controlled trial, Int J Rehabil Res 40 (2017), 240–245.

27.

Mohammad

M.T.

, Whitney

S.L.

, Marchetti

G.F.

, et al., The reliability and response stability of dynamic testing of the vestibulo-ocular reflex in patients with vestibular disease, J Vestib Res 21 (2011), 277–288.

28.

Mohammad

M.T.

, Furman

J.M.

, Sparto

P.J.

, et al., The effect of optotype size and velocity parameters on the performance of healthy young adult subjects on the gaze stabilization test, Otol Neurotol 34 (2013), 1090–1095.

29.

Newlands

S.D.

, Perachio

A.A.

, Compensation of horizontal canal related activity in the medial vestibular nucleus following unilateral labyrinth ablation in the decerebrate gerbil. I. Type I neurons, Exp Brain Res 82 (1990), 359–372.

30.

Orban de Xivry

J.-J.

and Lefèvre

, Saccades and pursuit: two outcomes of a single sensorimotor process, The Journal of Physiology 584 (2007), 11–23.

31.

Pritcher

M.R.

, Whitney

S.L.

, Marchetti

G.F.

, et al., The influence of age and vestibular disorders on gaze stabilization: a pilot study, Otol Neurotol 29 (2008), 982–988.

32.

Rinaudo

C.N.

, Schubert

M.C.

, Figtree

W.V.C.

, et al., Human vestibulo-ocular reflex adaptation is frequency selective, J Neurophysiol 122 (2019), 984–993.

33.

Ris

, Godaux

, Neuronal activity in the vestibular nuclei after contralateral or bilateral labyrinthectomy in the alert guinea pig, J Neurophysiol 80 (1998), 2352–2367.

34.

Sadeghi

S.G.

, Minor

L.B.

, Cullen

K.E.

, Dynamics of the horizontal vestibuloocular reflex after unilateral labyrinthectomy: response to high frequency, high acceleration, and high velocity rotations, Exp Brain Res 175 (2006), 471–484.

35.

Sawilowsky

, New Effect Size Rules of Thumb, Journal of Modern Applied Statistical Methods 8 (2009), 597–599.

36.

Scherer

, Migliaccio

A.A.

, Schubert

M.C.

, Effect of vestibular rehabilitation on passive dynamic visual acuity, J Vestib Res 18 (2008), 147–157.

37.

Schubert

M.C.

, Migliaccio

A.A.

, Della

C.C.

, Santina, Modification of compensatory saccades after aVOR gain recovery, J Vestib Res 16 (2006), 285–291.

38.

Schubert

M.C.

, Migliaccio

A.A.

, Clendaniel

R.A.

, et al., Mechanism of dynamic visual acuity recovery with vestibular rehabilitation, Arch Phys Med Rehabil 89 (2008), 500–507.

39.

Shanidze

, Ghahghaei

, Verghese

, Accuracy of eye position for saccades and smooth pursuit, J Vis 16 (2016), 23.

40.

Smith

P.F.

, Curthoys

I.S.

, Neuronal activity in the contralateral medial vestibular nucleus of the guinea pig following unilateral labyrinthectomy, Brain Res 444 (1988), 295–307.

41.

Voelker

C.C.

, Lucisano

, Kallogjeri

, et al., Comparison of the gaze stabilization test and the dynamic visual acuity test in unilateral vestibular loss patients and controls, Otol Neurotol 36 (2015), 746–753.

42.

Ward

B.K.

, Mohammad

M.T.

, Whitney

S.L.

, et al., The reliability, stability, and concurrent validity of a test of gaze stabilization, J Vestib Res 20 (2010), 363–372.

43.

Ward

B.K.

, Mohammed

M.T.

, Brach

J.S.

, et al., Physical performance and a test of gaze stabilization in older adults, Otol Neurotol 31 (2010), 168–172.

44.

Whitney

S.L.

, Marchetti

G.F.

, Pritcher

, et al., Gaze stabilization and gait performance in vestibular dysfunction, Gait Posture 29 (2009), 194–198.

45.

Xivry

J.-J.O.d.

, Bennett

S.J.

, Lefèvre

, et al., Evidence for Synergy Between Saccades and Smooth Pursuit During Transient Target Disappearance, Journal of Neurophysiology 95 (2006), 418–427.

	Average head velocities (deg/sec)
	Healthy controls (n)	Patients (n)	% UVD	p<0.05?
0.2 ΔlogMAR optotype
Kaufman et al. [17]	197.0 (50)	–	–	–
Honaker et al. [15]^#	123.25 (15)	74.64 (15)	0	yes
Lee and Honaker [18]	149.5 (20)	–	–	–
Ward et al. [42]^$	123– 157 (40)	–	–	–
Ward et al. [43]	95.5 (86)	–	–	–
Honaker and Shepard [16]^$	114.7– 154.7 (87)	–	–	–
Whitney et al. [44]^dotplus	124.4 (20)	79.4 (12)	92	yes
Pritcher et al. [31]^&	107– 127 (40)	90– 93 (15)	80	yes
0.25 ΔlogMAR optotype
Voelker et al. [41]	141.3 (17)	86.3– 105.7^ł (17)	100	yes
Mohammad et al. [27]	–	124– 152 (16)	75
0.30 ΔlogMAR optotype
Goebel et al. [7]	147 (14)	83.85– 112.45^ł (14)	100	yes