Bilateral vestibulopathy (BV) is a chronic disorder in which the vestibular function is bilaterally severely reduced or absent (1). It has many etiologies, varying from toxic (e.g., gentamicin ototoxicity), infectious (e.g., meningitis) and genetic (e.g., DFNA9), to inner ear disease (e.g., Menière’s disease, auto-immune inner ear disease) and neurodegenerative disease (e.g., CANVAS) (2). BV patients report a spectrum of symptoms, of which chronic unsteadiness and/or oscillopsia are most frequently reported (3, 4). These symptoms often result in a decreased quality of life and a high socio-economic burden on society (4, 5). BV can be diagnosed using the diagnostic criteria of the Bárány Society. These criteria include a chronic vestibular syndrome with symptoms of unsteadiness when walking or standing (possibly combined with oscillopsia), and a bilaterally reduced or absent angular vestibulo-ocular reflex function. This latter should be documented by at least one of the following three tests: video Head Impulse Test, caloric test or rotatory chair test (6). Currently, only lateral semicircular canal function is included in the diagnostic criteria of BV, not otolith function. This implies that in patients diagnosed with BV, otolith function can still be present (7–9).

Vestibular evoked myogenic potentials (VEMPs) measure dynamic otolith function by stimulating the Type I hair cells at the striola (10, 11). VEMPs are electromyographic responses to air-conducted sound or bone-conducted vibration of the skull, which most likely reflect otolith function. Two types of VEMPs can be measured: cervical VEMPs (cVEMP) and ocular VEMPs (oVEMP). Cervical VEMPs comprise inhibitory responses from the ipsilateral sternocleidomastoid muscle, mainly evaluating function of the saccule. Ocular VEMPs comprise excitatory responses from the contralateral inferior oblique extra-ocular muscle, mainly evaluating function of the utricle. The testing paradigm and interpretation of VEMPs are difficult to standardize (12). Since VEMP response characteristics (amplitude, latency, threshold) depend on, e.g., stimulus type, muscle-contraction and age, each laboratory should obtain its own normative data. However, even after correcting for differences in muscle contraction, variability in VEMPs can be large in normal subjects (13).

In BV, VEMPs vary widely. BV can lead to reduced or absent VEMP responses, but in a significant number of patients VEMPs are within the normal range (7–9). The number of reduced, absent or present VEMPs in BV patients differs between studies. This most likely reflects the heterogeneity of testing paradigms, outcome measures and patient populations used in these studies (7). For example, some etiologies like aminoglycoside toxicity might be associated with otolith abnormalities (14). Moreover, it remains difficult to perfectly understand VEMPs in BV due to the large variability in normal subjects. After all, VEMPs might be reduced due to BV, but still be within the “broad” normal range, leading to false negative results (7). Additionally, age significantly affects VEMPs, resulting in a high rate of absent responses in normal subjects above the age of 60 years (12). Since the age of the BV population is relatively high (8), an absent VEMP response in a BV patient might reflect age, BV, testing paradigm, or a combination of these factors. In these patients, the influence of BV in the observed VEMP responses remains unknown. Furthermore, as stated above, otolith function is not included in the diagnostic criteria of BV. Patients with disorders predominantly affecting otolith function, might therefore be missed and not included in BV studies (7, 15, 16).

It was previously found that different acoustic stimulus frequencies, evoke different VEMP responses (17). For example, cVEMP responses in the affected ears of patients with Menière’s disease, demonstrate a significantly higher cVEMP threshold at tone bursts of 500 Hz than at 1,000 Hz, compared to normal subjects (18, 19). This implies that “frequency tuning” exists in the vestibular system. This “frequency tuning” is also affected by age. In young normal adults, the largest VEMP responses are obtained around 500 Hz. In older adults (≥ 60 years), the largest VEMP responses are more often obtained at 750 and 1,000 Hz in the majority of cases. It was therefore recommended to test 750 and 1,000 Hz tone burst frequencies, in case absent responses are found at 500 Hz (20).

In the last decades, a novel treatment was proposed to treat BV: the vestibular implant. This is a (not yet clinically available) strategy to partially restore vestibular function by stimulating the vestibular nerves, using surgically implanted electrodes (21–23). The electrodes can be implanted inside or close to the semicircular canals (24, 25), or inside the otolith organs (26). Surgically positioning electrodes in a semicircular canal or otolith organ, can destroy the (residual) function of that specific organ. Consequently, otolith implantation is currently only considered in case of bilaterally absent cVEMP and oVEMP responses (27). In BV subjects it is therefore imperative to understand, before considering vestibular implantation, whether otolith function is present or not. However, previous VEMP studies in BV patients mainly tested at 500 Hz (8, 9, 14). This might imply that some BV patients that were considered to have absent VEMP responses, might have had preserved VEMPs at other test frequencies.

The objective of this study was therefore to investigate whether multi-frequency VEMP testing (500, 750, 1,000, and 2,000 Hz) would improve the detection of present otolith responses in BV patients.

This study investigated whether multi-frequency VEMP testing (500, 750, 1,000, and 2,000 Hz) would improve the detection of present otolith responses in BV patients, compared to only testing at 500 Hz. It was demonstrated that more present otolith responses were obtained with multi-frequency testing. Present cVEMPs were more often present than present oVEMPs. Most present VEMP responses were found when testing at 500, 750, and 1,000 Hz, while 2,000 Hz resulted in fewer present VEMP responses and (for cVEMPs) significantly higher thresholds. These results show that multi-frequency VEMP testing should be considered in BV patients, in case evaluation of (residual) otolith function is indicated.

Multi-frequency VEMP testing improves the detection of present VEMP responses in BV patients, emphasizing the need to test “beyond” the 500 Hz stimulus. These findings are congruent with previous literature in healthy subjects, in which it was illustrated that 500 Hz is not always the best frequency for VEMP testing (20). This results from “frequency-tuning” of the vestibular system, in which different acoustic stimulus frequencies evoke different VEMP responses. Frequency-tuning is patient specific, since it might depend on multiple factors. These factors include, among others: etiology, stage of disease, age, and stimulation paradigm (18, 20). Since multi-frequency VEMP testing improves the detection of present VEMP responses in BV patients, it might be somehow analogous to ice water caloric testing (1). After all, multi-frequency VEMPs and ice water calorics can both be used to detect residual function in case no responses are obtained in the “routine tests” (respectively 500 Hz VEMPs and bithermal caloric testing). Furthermore, this study showed that fewer present VEMP responses and (for cVEMPs) significantly higher thresholds were found when testing at 2,000 Hz. This was expected, since in healthy subjects 2,000 Hz has also less robust responses and significantly higher thresholds than the other tested frequencies (35). This might not directly imply that VEMP testing at 2,000 Hz should be abandoned completely: one ear only demonstrated a 2,000 Hz oVEMP response, without any responses at the other tested frequencies (oVEMPs and cVEMPs). Future research should be conducted to investigate whether testing at other frequencies is beneficial.

Present VEMP responses were frequent in this BV population. Seventy-eight percent demonstrated at least one present VEMP response in at least one ear, and in 46% a present VEMP response was found bilaterally. A high percentage of present VEMP responses in BV patients is consistent with previous studies. However, direct comparison is difficult because of different BV populations, stimulation paradigms and outcome measures. Nevertheless, these results illustrate two important aspects. First, BV is currently a diagnosis based on lateral semicircular canal function: it does not include vertical semicircular canal function, or otolith function as measured by VEMPs (6, 9, 14). This implies that patients with predominantly affected vertical semicircular canal and/or otolith function, might be missed (7, 15, 16, 36). However, diagnosing (isolated) otolith dysfunction is still challenging. The clinical presentation of otolith dysfunction is not yet well understood and no consensus has been reached on this possible clinical entity (37–39). After all, absent VEMP responses are not necessarily causally related to vestibular symptoms, and do not rule out involvement of other structures (37). It would therefore be advised to further investigate the possible clinical entity of otolith dysfunction (38). Secondly, this study shows that VEMP responses are often still present in BV patients. On-the-one-hand this may imply that otolith function is relatively spared in BV patients compared to lateral semicircular canal function. It could be hypothesized that, e.g., otoliths are less affected by certain vestibular disorders. On-the-other-hand it might have nothing to do with otoliths being less affected than semicircular canals, but with VEMP testing itself. It could be hypothesized that VEMPs are relatively stronger stimuli to the otoliths than, e.g., bithermal caloric testing or video head impulse testing to the semicircular canals. In other words: although otolith function might be affected, a response remains present due to the strong nature of the stimulus. As a result, a present VEMP response is obtained, while hypofunction of the lateral semicircular canals is detected by the caloric test and/or video head impulse test. Additionally, VEMPs are considered to test Type 1 hair cells, while the caloric test might mainly test Type 2 hair cells (10). A dissociation between these tests could therefore also result from a difference in affected hair cell type. Furthermore, the interpretation of the tests may play a role. VEMPs are currently not able to detect subtle changes in otolith function, even with good normative data (7). This results from the large range of normal responses in healthy subjects (7). Therefore, except for superior semicircular canal dehiscence syndrome (40), VEMPs are clinically often interpreted as an “on–off” response: response are considered present or absent (27). It is however not justified to compare tests with different outcomes: categorical “on–off” (VEMPs) vs. numerical (e.g., caloric test slow phase eye velocities, video-head impulse test vestibulo-ocular reflex gain). For example, in cases with reduced (but still minimally present) otolith and lateral semicircular canal function, a reduced but present VEMP response might be obtained. This would then be classified as “normal.” The numerical outcomes of the caloric test and/or video head impulse would indicate a reduced function. These tests would then be classified as “abnormal.” However, if the caloric test and video head impulse test would be interpreted based on just the presence of a response, these tests would have also been classified as “normal” (e.g., a video head impulse test gain of 0.4 is still a response). Therefore, the dissociation between otolith and lateral semicircular canal findings in BV patients, should be interpreted with care. It might even be incorrect to state that, e.g., otolith function would be less affected than lateral semicircular canal function in BV patients.

Nevertheless, this study demonstrates that the diagnostic criteria of bilateral vestibulopathy should be revised in the future. A more detailed classification could be considered, taking into account all 10 vestibular sensors, not only the lateral semicircular canals. This might facilitate a detailed classification. For example, categories could vary from abnormal vestibular responses in all vestibular sensors, to isolated abnormal responses such as selective vertical canal or otolith impairment (10, 36).

In this study, the presence of present cVEMP responses was higher than present oVEMP responses. Air-conducted cVEMPs are therefore superior to air-conducted oVEMPs to detect present VEMP responses in BV patients. This does not necessarily indicate that, e.g., saccular function is less affected than utricular function. After all, these findings are not consistent with previous literature, in which saccular and utricular function were almost equally affected (14). Most likely, this can mainly be attributed to the stimulation paradigm: air-conducted sound (this study) vs. bone conducted vibration. Bone conducted vibration produces more reliable oVEMP responses than air-conducted sound (12). This implies that the oVEMP findings in this study probably underestimate the presence of present oVEMP responses in BV patients. It was aimed to use bone-conducted vibration in this study, but long lasting equipment failure unfortunately prevented this study from using bone-conducted vibration. Nevertheless, the objective of this study was to investigate the application of multi-frequency stimulation. This could still be accomplished, but these findings are only applicable to air-conducted stimulation.

BV patients demonstrated a significantly higher percentage of absent VEMP responses and significantly higher VEMP thresholds than healthy subjects, when corrected for age. This shows that VEMP responses can be significantly affected in BV patients. Unfortunately, as stated above, the large range of normal responses in healthy subjects does not allow to detect subtle changes in otolith function (7, 12). Therefore, VEMPs are mainly interpreted as an “on–off” response in BV patients (27). Taking these limitations and the results of this study into account, a pragmatic VEMP testing paradigm could be proposed for BV patients, if the clinician would only be interested in the presence or absence of a VEMP response. This paradigm is illustrated in Figure 2.

Its concept focuses on obtaining a relatively quick insight in VEMP responses: start testing at a high sound pressure level (e.g., 130 dB SPL) to maximize the detection rate; use a 13 Hz stimulus rate to minimize testing time (cVEMPs); use multi-frequency testing to improve detection rate; stop when one frequency shows a present response since presence of a VEMP response has been demonstrated. It should be noted that testing at 2,000 Hz may be optional. After all, only one ear demonstrated a present VEMP response at 2,000 Hz, without any responses at the other frequencies (Table 2: oVEMP, left ear). The proposed frequencies in this paradigm are based on this study, and other frequencies (e.g., 1,500 Hz) were not tested. Therefore, other frequencies could also be included. Additionally, based on previous literature it would be proposed to use a bone-conducted stimulus for oVEMP testing (12). However, the reliability of a 13 Hz stimulus for a bone-conducted stimulus should still be determined and therefore a 5 Hz stimulus would still be preferred for oVEMPs.

Multi-frequency VEMP testing improves the detection rate of present otolith responses in BV patients. Therefore, multi-frequency VEMPs should be considered when evaluation of (residual) otolith function is indicated.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Ethical Committee of Maastricht University Medical Center (NL52768.068.15/METC). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

FL: Conceptualization, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing, Data curation, Formal analysis, Project administration, Resources. ML: Investigation, Writing – review & editing, Formal analysis. LS: Writing – review & editing. MJ: Formal analysis, Methodology, Writing – review & editing. VR: Writing – review & editing. ED: Writing – review & editing. AP-F: Writing – review & editing. NG: Writing – review & editing. RB: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.