Opinion: The neural basis of vestibular evoked myogenic potentials. The cVEMP is a specific indicator of saccular function

Jonas Bruun Kjærsgaard^1,2*Herman Kingma^1,2

• ¹Department of Otolaryngology, Head & Neck Surgery and Audiology, Aalborg University Hospital, Aalborg, Denmark
• ²Department of Clinical Medicine, Aalborg Universitet, Aalborg, Denmark

In a recent review Curthoys et al. 1 reviewed animal studies examining responsivity of the vestibular system to sound and state that, beyond doubt, the cVEMP is a specific indicator of saccular function. However, in our opinion, the debate about the origin of the cVEMP is not closed. In our review 2 , published 9 months earlier we paid special attention to the issue that a normal cVEMP does NOT exclude loss of saccular function, especially of that of the clinically most relevant sustained saccular system (detecting head tilt and linear accelerations). Also, we draw a more conservative conclusion, that substantial contributions of non-saccular vestibular end-organs to the cVEMP response cannot be ruled out 2 . As this may leave researchers and clinicians in confusion, we want to share the key differences in interpretation of the cited animal studies and the clinical relevance of cVEMP. Clinical relevance of cVEMP: is cVEMP really a test of saccular function? One highly clinically relevant issue is not addressed in the review by Curthoys et al. 1 In contrast to what is suggested in the title of their review, it is a simplification to state that the cVEMP is a specific indicator of saccular function. Curthoys et al. 1 state that cVEMPs result predominantly from direct stimulation of striolar hair cells and irregular afferents, i.e. the transient otolith system, which is only a small part of the otolith system with a thin otolith membrane and small otoconia. Based on the transfer function of the otolith systems, high frequency cVEMP stimuli hardly move the thick otolith membrane with larger otoconia (high membrane mass and stiffness) on top of the extra-striolar hair cells; in line with that regular extrastriolar afferents have a low sensitivity to the high frequency transient/vibratory stimuli used in VEMP testing [3][4][5] . Therefore, cVEMP testing hardly -if at allprobes the function of the clinical very important sustained otolithic system, which is responsible for detecting head tilt and linear acceleration. Clinically, this distinction is crucial: cVEMP's results cannot be equated with preserved saccular function as suggested by Curthoys et al. 1 , just as normal bone conduction does not ensure functional hearing in daily life. Hence, the presence of cVEMP is NOT a specific indicator of saccular function in toto, it might only reflect normal function of striolar hair cells and irregular afferents. This directly limits the clinical relevance of the cVEMP. Are other hair cells in the labyrinth be involved to? Regarding the other discussion addressed in the review by Curthoys et al. about the actual origin of the cVEMP: is this response really only related to direct stimulation of saccular hair cells, or may other hair cells in the labyrinth contribute? We agree with Curthoys et al. that the later animal studies by the Bordeaux group [Didier, Cazal and colleagues] convincingly demonstrate the sound sensitivity of fibres coming from the saccule in the guinea pig model 1 . The question is however, to which extend the saccule may selectively be activated by sound. Curthoys et al. 1 specifically highlights a finding of Cazal et al. 6 where they, after semi-selective damaging the ampullae and the utricular maculae, were still able to elicit a non-cochlear electrophysiological response. The number of hair cells in the utricular macula and the ampullae were reduced by approximate half of that of the controls, saccular hair cell count was reduced by one fourth 6 . So, these experiments supported the relevance of saccular hair cell contributions, but do not exclude ampullar of utricular hair cell contributions. Also, the responses were measured from the round window niche 6 page 612, 614 and not from the saccular nerve in contrast to what Curthoys et al. state 1 page 03 . Therefore, the response seen could still be composed of contributions from all still intact hair cells in sacculus, utriculus and ampullae, and do not selectively show saccular responses. In conclusion, the experiments by Cazal et al. 6 are unable to rule out a substantial contribution by ampullar and utricular hair cells to the cVEMP response, which Curthoys et al. suggests 1 . To our knowledge, there are no studies at hand that show that ablation of the sacculus invariably abolishes the cVEMP, which would be the real proof that the cVEMP is a crucial indicator of saccular hair cell activation. We are still awaiting such evidence similar as stated by Welgampola & Carey 7 .We will address now three key aspects regarding the studies dealing with the origin of the cVEMP. Curthoys et al. state that "if semicircular canal neurons are activated at clinical test frequencies and intensities, it detracts from the specificity of the cVEMP as a test of saccular function". 1 We agree, and that also holds for utricular coactivation, which was previously been documented by both Curthoys et al. and Zhu et al. among others [8][9][10][11][12] . There is evidence that all end-organs are activated by sound at intensities within limits of safe exposure 8,[10][11][12] . However, this evidence has been put into question, in the review at hand, based on the use of phase-locking or probability firing as a measure of sensitivity 1 . There are experimentally at least two different ways used in the relevant literature to determine responsiveness with quite different outcomes [8][9][10][11][12]18 . Curthoys et al. have previously defended the choice of using rate-change as the definition of activation over phase-locking or probability of firing 9 . But rate-change sensitivities may only be studied by using sustained tones 9,10 , or long tone-bursts 18 . These types of stimuli are not used for human cVEMP testing, where clicks or short tone-bursts are the standard. So, extrapolation of rate-change sensitivities to the cVEMPs in humans can be questioned. Furthermore, using ratechange will not take into account activation of afferents firing highly synchronized to the stimulus 19 , if their total rate of firing is not exceeded 18 page 6064, figure 7 . Indeed, Young et al., using sustained tones, found that phase-locking thresholds were 10-30 dB lower than that of rate-change thresholds 10 . Equally, McCue and Guinan showed that phase-locking to click stimuli results in considerably lower thresholds than using rate-change to long tone-burst stimuli 18 . All these aspects result in a complex and different interpretation of the contribution of non-saccular afferents to cVEMP 8,[10][11][12] . So, for us it is not at all clear what the relative contribution of non-saccular afferents is in cVEMP in humans, and thereby how specific the saccular contribution can be said to be. Zhu at al. 8,11 used high stimulus intensities in the rat studies, which according to Curthoys et al. 1 do not allow extrapolation to the standard stimulation intensities used in humans. However, Zhu et al. when observing considerable activation of afferents from all five vestibular end-organs 8,11 used an intensity up to 130 dB peak SPL, and not 130 dB SPL as incorrectly stated by Curthoys et al. 8 . This is an important distinction 20 . The ABR threshold to click in the rats used by Zhu et al. are 50 dB peak SPL 21 , which after adding the 80 dB sensational level equals 130 dB peak SPL, just like Zhu et al. state 8,11 . Rosengren et al., examining cVEMP responses in humans using 0.1 msec clicks of 138.7 dB peak SPL, report a crest factor of up to 33.4 dB between peak SPL and the A-weighted, 1 sec integrated SPL 22 . In fact, 130 dB peak SPL, depending on the integration time used, may be equivalent to or less than 100 dB SPL. The first cVEMP studies in healthy people report usage of 0.1 msec clicks delivered at 145 dB SPL 23,24 , which might have been 145 dB peak SPL, which in any case is much higher than the intensities used by the Zhu-group in rats. In standard clinical cVEMP practice the maximum intensity used is 135-140 peak SPL 25 . Thereby, the intensities used by the Zhu-group are similar to or even below the values used clinically and extrapolation to humans is therefore not hindered as Curthoys et al. suggests 1 . The focus on this core issue is shifted away in the review of Curthoys et al. 1 by the description of behavioural threshold differences between the Rat, Guinea Pig and Humans, and related flawed speculations of the physical sound intensity used by the Zhu et al. 8,11 . In its current form, the review advocates for the continued use of cVEMP as a specific measure of saccular function. Yet this endorsement rests on a misrepresentation of the studies and findings by Cazal et al. 6 , Uchino et al. [13][14][15] , and Zhu et al. 8,11 , as demonstrated herein. Moreover, the appropriateness of using rate-change as a criterion for irregular afferent activation has been questioned, given that it necessitates a stimulus paradigm that is not applicable to cVEMP testing.Finally, and most clinically relevant, is the recognition that an intact VEMP does not imply normal macular function. Clinicians should be particularly aware of this limitation when managing patients with vestibular disorders.