Cochlear implant surgery and perioperative dizziness is associated with utricular hyperfunction

2.1 Study design

The SVV was estimated with a new VR application designed by MacDougall et al. [6]. This system presents a series of paintings to the patient in a head-mounted VR system, and their task is to align these to the visual vertical. The system was well-suited to the study, as it is mobile, and could be done readily by patients on the ward, as early as one day after implantation. The test-retest reliability of this system has been validated with results comparable, or even more superior to that of traditional SVV devices [7, 52].

For the main part of our study, we used the SVV system on a group of cochlear implant recipients immediately before surgery, and again at 1-day, 1-week and 6-weeks later. At each postoperative follow up, participants’ perceptions of dizziness were classified using a custom algorithm (Fig. 1). Associations between the SVV measurements and dizziness were determined with a mixed linear model. As a secondary analysis, we also tested the SVV on a group of normal hearing adult participants, across two time points in order to better contrast these with the implant-related effects on the SVV.

OPEN IN VIEWER

Fig. 1

Postoperative diagnostic algorithm. Postoperatively, the diagnostic algorithm was used to classify participant symptoms into possible aetiology groups based on the type of dizziness as well as any other triggers or specific accompanying symptoms. The boxed numbers represent the number of participants who were classified into each aetiology group, at each follow up time point. The shaded curved boxes represent the differential diagnoses for participants who reported dizziness symptoms that were new or different to their baseline after surgery, with the total values tallied at the bottom of the figure. ^†BPPV; Benign Paroxysmal Positional Vertigo. ^‡The vestibular migraine history was taken according to the classification criteria formulated by the Committee for Classification of Vestibular Disorders of the Bárány Society and the Migraine Classification Subcommittee of the International Headache Society (IHS).

Ethics for both parts of the study was approved by the Human Research and Ethics Committee at the Royal Victorian Eye and Ear Hospital (RVEEH) (Part 1: HREC 10/969H, Part 2: HREC 16/1299H). All procedures were performed in accordance with the Declaration of Helsinki and participants could discontinue at any point during the study.

The “Curator SVV” system utilises three pieces of equipment: a Samsung S6 phone containing the mobile phone app, a Samsung Gear VR headset (Innovator Edition for S6), and a Bluetooth remote controller. Subjects were required to have sufficient visual acuity to see the borders of the test images unaided or corrected with contact lenses.

During the test, a series of 10 different paintings were presented against a black background, with each image rotated away from the vertical axis at a specific starting angle between –12 to 12 degrees. Participants were seated upright and asked to align the paintings to their perceived gravitational vertical using the remote controller. The output of the device records the difference (in degrees) between the participants’ perceived gravitational vertical and the true vertical (0 degrees), termed the response angle. In total, 10 response angles are measured for each test, one for each image. The average of these 10 response angles was used to calculate the overall deviation away from the visual vertical (referred in this paper as the ‘SVV angle’).

2.3 SVV testing before and after CI surgery

Forty-one adult CI recipients with a mean age of 65 years (range 25–85 years, 20M:21F) were recruited into the study between April 2019 and October 2019 (Fig. 2). Participants on the cochlear implant waiting list at either the RVEEH or St Vincent’s Private Hospital, East Melbourne were invited to participate. Formal vestibular testing prior to surgery was not a criterion for inclusion. There were no exclusion criteria, except for those unable to see the images clearly, as noted above.

OPEN IN VIEWER

Fig. 2

Flowchart of participants recruited into the cochlear implant arm of the study showing the number of cases recorded at each follow-up time point.

Participants were tested at four intervals: preoperatively, 1-day (D1), 1-week (W1) and 6-weeks (W6) after surgery. At each time interval, dizziness symptoms were classified using a diagnostic algorithm (Fig. 1) and the SVV test was performed.

An algorithm was developed that sought to assist with classification of the various presentations of dizziness that have been associated with CI (Fig. 1). The presentation of interest was postoperative “CI related dizziness”, however well described alternate aetiologies (e.g. benign paroxysmal positional vertigo (BPPV) and vestibular migraine) have been associated with CI, and for these the internationally recognised diagnostic criteria were applied [31, 50]. For the remaining reports of dizziness, the diagnostic category was determined by case-conference between the investigators. The algorithm was structured to differentiate between the possible diagnoses in a systematic way, to increase consistency.

SVV angle directions were standardised such that a positive (+) value indicated a deviation away from the operated ear.

A mixed linear model was performed with the “fitlme” function in Matlab 2019b (Mathworks, MA, USA) [38] to determine factors affecting the variance in SVV angles. The time of testing, presence or absence of dizziness at each time point, age of the patient, and outcomes of the 10 trials contributing to each SVV test were entered as categorical variables (fixed factors) into the mixed linear model, with the patient as a grouping variable, assuming a random intercept per patient. SVV angle was the dependent factor, and the model method was restricted minimum likelihood (REML).

In order to contrast the CI-related SVV responses, we also tested thirty-four healthy volunteers with no known self-reported hearing loss or imbalance. Participants had a mean age of 43 years (range 24–76 years). Participants were tested twice using the “Curator SVV” system; the median time interval between the two test days was 7.5 days. The SVV angle was calculated for each test and used for analyses (SPSS Statistics, Version 26.0 IBM; 2019). Given that all data points were normally distributed (as determined by the Shapiro Wilk test, p > 0.05), parametric analyses were performed to compare the participant results over time, using a paired t-test.

3.1 SVV testing after CI surgery

Forty-one participants were recruited into the cochlear implant arm of the study. Demographic data for these participants is shown in Table 1. Participants ranged in age from 25–85 years, averaging 64.8±14.1 years of age. Twenty-one participants received Cochlear Ltd.’s Thin Perimodiolar™ electrode, five received another perimodiolar electrode, Cochlear’s Contour Advance™, and the remainder a thin and flexible lateral wall (Cochlear’s Thin Straight™) electrode. Prior to surgery, seven participants had had a history of imbalance or vertigo of uncertain cause, two had Meniere’s Disease, one a history of BPPV and one vestibular neuronitis.

Subjective reports of dizziness were determined preoperatively and again at each time point after surgery using the algorithm in Fig. 1. In total, 25/41 (54%) participants experienced new symptoms of dizziness at some point during the follow-up period that they perceived to be different or of greater magnitude to any preoperative symptoms. 13/38 (34%) participants reported subjective symptoms at 1-day post-op, compared to 18/39 (46%) at 1-week, and 11/37 (30%) at 6-weeks (Fig. 1). For more detail regarding the pattern of dizziness symptoms across the three follow-up time points see Supplementary Figure 1. Participants with “CI-related dizziness” described experiencing unsteadiness, imbalance or feeling “wobbly” while on their feet. The majority of cases were mild lasting seconds to minutes and began almost immediately after surgery (if reported at 1-day post-op). Symptoms were most severe when walking (with many needing assistance to get to the bathroom), getting up from a seated position or when moving in the dark. Most cases of imbalance improved with time, with only four reporting ongoing imbalance at 6-weeks post-op. Two cases of “acute labyrinthine dysfunction” were reported in the study using the diagnostic algorithm, both occurring immediately after surgery (1-day post-op). The vertigo was constant and debilitating, with nausea and vomiting. By the 1-week follow up, the severity of the symptoms had reduced, however both participants still complained of dizziness triggered mainly by head movements (e.g. getting out of bed), with episodes lasting only seconds. These reports may allude to the possibility of BPPV developing after CI, a phenomenon that has been documented in previous CI studies as well. Reasons as to why this occurs after surgery is still unclear, proposed theories suggest it could be due to bone dust particles travelling from the scala tympani into the scala media, otoconia dislodging after electric stimulation of the implant, or from the vibratory trauma created after drilling into the cochlea [16, 32].

Figure 3 presents SVV angles for 32 participants in which data was acquired at all four time points. It is apparent that prior to implantation, there was greater variability in the estimate of the SVV than expected from normal subjects where response angles are usually within 2° of vertical. One day after implantation the SVV deviated significantly away from the implanted ear in most cases when compared to their preoperative baseline. The mean SVV angle returned to preimplantation levels over the next 6 weeks, although for some individuals the SVV deviated most one week after implantation. A few participants exhibited marked variability in the SVV angle over this 6-week period.

OPEN IN VIEWER

Fig. 3

The subjective visual vertical for 32 participants following cochlear implantation. The heavy black line shows the mean values and the vertical bars are +/–1 standard error. Positive deviations are away from the implanted ear. The coloured lines represent individual participants’ results, at the following time points: Pre: Preoperative; D1: One day after surgery, W1: One week after surgery; W6: Six weeks after implantation.

A mixed linear model was performed to examine relations between SVV tilts, time of testing, and new-onset dizziness. For this analysis, the raw traces (10 trials per test) from all 41 participants were used. In the mixed linear model, data were grouped by patient, and for individuals a random intercept was assumed. SVV tilts were found to be dependent upon the time of testing (F(3,1485) =41.4, p < 0.001), with the mean difference, relative preoperative levels, of 2.17° on postoperative day 1 (CI:1.69–2.66, p < 0.001), 0.889° at 1 week (CI:0.390–1.39, p < 0.001), and –0.250° at 6 weeks (CI:–0.720–0.220, p = 0.298). SVV angles were significantly affected by the new-onset dizziness (F(1,1485) = 3.89, p = 0.049) with tilts away from the implanted ear of 0.519° (CI:0.020–1.018°) in dizzy participants. The trial number (of the 10 repetitions per test) had no significant impact upon the SVV angles (F(9,1485) = 1.034, p = 0.41). Patient age was not significant (F(1,1485) = 1.98, p = 0.21).

The CI-related outcomes may be contrasted to those of thirty normal-hearing subjects tested with the same SVV system. The SVV was tested on two separate occasions. The mean SVV for the first test was 0.125±1.44°, and that of the second, 0.237±0.981°. The paired t-test showed no statistical significance (p = 0.591). These data demonstrate that in normal-hearing subjects the SVV is similar from one test to another, which contrasts to the SVVs elicited in CI recipients. These data are presented in Supplementary Figure 2.

4.1 SVV testing after CI surgery

The main finding was that the SVV tilted away from the operated ear in most CI recipients; an observation consistent with the outcomes of smaller studies where patients had undergone either CI or stapedectomy surgery [18, 47]. The SVV deviations typically returned to preoperative levels over the next 6 weeks. New-onset dizziness was associated with significantly greater SVV tilts away from the implanted ear.

The normal range of SVV or subjective visual horizontal (SVH) settings for healthy subjects is within a band of +/–2 degrees of gravitational vertical or horizontal [11, 14]. It is noteworthy to mention that some of the participants demonstrated very large SVV values preoperatively; this may have been due to pre-existing vestibular dysfunction as well as extra-vestibular issues. However, although their baseline level of utricular function was not within the normal thresholds, their 1-day SVV results were still significantly changed and often tilted even further away from normal range when compared to their preoperative levels, suggesting that the surgery had still caused a change to their otolith function. Two participants showed consistent postoperative tilts towards the implanted ear implying reduced utricular function which may result from surgical trauma at implantation.

The SVV measurements suggest that cochlear implant surgery typically causes utricular hyperfunction in the operated ear. The SVV results reflects the function of a subset of afferents within the utricular macular that exhibit a regular resting discharge, and respond to low frequency linear acceleration and tilt, thereby assessing static otolith function [4, 10]. An SVV tilt away from the implanted ear is the response anticipated when there is an increase in the activity of these afferents. The most likely reason for increased afferent activity would be biasing of the sensory receptors. One way in which the utricular macular could be biased is from an intraoperative loss of perilymph [33]. The macula floats on perilymph and is suspended across the labyrinth on the membrana limitans [48]. Thus, any loss of fluid is likely to lead to a bowing of the macula and a bias of the receptors. However, while this could lead to intraoperative changes in utricular function, it is an unsatisfactory explanation for the SVV tilt 24 hours after surgery. Perilymph rapidly refills the labyrinth and there is no direct clinical evidence to support the idea of a continuous peri-lymphatic leak after inner ear surgery. Alternatively, utricular function might increase in association with endolymphatic hydrops. Endolymphatic hydrops has been reported after many types of inner ear stress including CI [22, 45] and stapedectomy surgery [24]. It has been observed in the weeks following experimental CI, where it was thought to relate to the inflammation associated with inner-ear surgery [43].

The return of the SVV to preoperative levels likely result from either of two mechanisms anticipated in the first 6 weeks after implantation. First, as the inner ear recovers from implantation, there is a reduction in inflammation and the associated endolymphatic hydrops. A resolution of hydrops over this period of time has been demonstrated in experimental cochlear implantation [43]. Second, central compensation is expected to normalise the perception of the visual vertical over this period. This has been demonstrated after surgical removal of vestibular schwannomas or surgical labyrinthectomy [12]. Immediately after these surgical procedures there is torsion of the eye towards the operated eye due to the loss of spontaneous activity in the deafferented labyrinth, but over 6 weeks the eye returns to a normal orientation offering an alternative theory that the SVV may normalise irrespective of whether the utricle recovers or not. Thus, it is still unclear whether recovery of the SVV to normal values at 6-weeks post-op is due to recovery from a temporary loss of utricular function or through the mechanisms of central compensation and would require further investigation.

Another factor which may affect SVV results at 6-weeks post-op is the cochlear implant device itself. Electrical stimulation of the cochlear implant device has been demonstrated in studies to affect neural function postoperatively, bringing about changes in neural adaptation, plasticity, impedance and conductivity [2, 49]. In our study, CI activation occurred at the 2-week postoperative period. Thus, when the SVV and questionnaire was conducted at the 6-week mark, the test was conducted with the CI turned on in order to replicate the daily environment that patients were exposed to. Given this, it is also possible that the auditory stimulation of the CI could have affected outcomes reported by patients at this time point. However, whether this would have contributed a significant amount is unclear, especially given that most patients’ symptoms had already resolved by this point.

As the paper focuses solely on utricular function in the perioperative CI period, these data do not inform whether utricular function remains stable over time. However, previous literature suggests that beyond the first few weeks of surgery, utricular function is similar to preoperative values; and when it has changed, there is no clear relationship to the clinical symptoms. This suggests that the resolution of utricular function observed here at 6-weeks is likely to persist, or alternatively, that compensation has occurred [1, 3, 8].

4.2 Weaknesses

Some participants were noted to have pre-existing dizziness symptoms, and this made it difficult to determine whether their postoperative dizziness symptoms were an extension of their ongoing problems or new from the surgery. This also posed a challenge when assessing SVV results; because some participants had abnormal preoperative SVV results, we could not use the traditional SVV thresholds of +/–2 degrees to determine whether the results were abnormal or not. Instead, we used each participants’ preoperative result as their baseline, and compared postoperative results to their first test, therefore looking for relative changes rather than absolutes. Additionally, it is conceivable that the anaesthetic may still have affected patient’s perception of balance 1 day after surgery. Similarly, other vestibular organs may also contribute to dizziness symptoms either saccular or semi-circular canal in origin. While previous studies suggest that semi-circular canal is not affected shortly after CI surgery [42], saccular function in the perioperative period is not as clear. While saccular function has been assessed after CI using cVEMPs [1, 28], such studies are usually conducted in the long term, often months after CI, and also do not correlate results with patient symptoms. There are very few studies which have tested cVEMPs within the first week after CI surgery. Moreover, cVEMP testing can be quite uncomfortable to perform in the perioperative period as it can cause patients significant neck discomfort. This gap in literature may be worth pursuing in the future once more convenient measures of saccular function become available.

4.3 Clinical significance

Dizziness was associated with greater SVV tilts after implantation. These findings suggest utricular hyperfunction may contribute to the perioperative dizziness experienced by some cochlear implant recipients. Clinically, this knowledge will enable clinicians to better track patients’ recovery over time. Depending on the direction of SVV tilt, we are able to delineate between the patients whose dizziness symptoms are due to a temporary irritative change in utricular function, compared to those with those with more permanent vestibular loss. As seen in this study, those with tilts away from the implanted ear tend to recover over time, with the majority resolving by the 6-week postoperative mark. In these cases, offering symptomatic treatment for patient’s symptoms would be reasonable. However, SVV tilts towards the implanted ear suggest that a more damaging injury has occurred, which may take longer to recover, or necessitate more intensive intervention such as vestibular rehabilitation.

4.4 Future directions

The SVV test provides accurate information on static utricular function by reflecting the activity of regularly firing neurons; it may be of interest to know how the irregular firing utricular neurons are affected after CI, allowing us to assess whether dynamic function is also impacted. This can be achieved using other vestibular tests such as oVEMPs. Although more challenging technically, oVEMPs testing in conjunction with the SVV would give us a more holistic impression of utricular function after implantation. For the future, we hope to look for modified methods of bone conduction that would be acceptable for testing in patients in the perioperative period in order to accurately and safely perform oVEMPs. This would add to our understanding of vestibular dysfunction after cochlear implantation and may help predict who will become dizzy, and who may benefit from targeted interventions to mitigate the effects. Additionally, it may be worth exploring the effect that CI electrical stimulation has on SVV measurements by repeating the test at each follow-up time point, once when the CI is switched on, and then again with the device off to see if electrical stimulation plays a role in affecting utricular function.