Caloric Tests

The otoscopic findings and the pure tone audiometry (PTA) data of all patients were evaluated. The mean hearing level (MHL) was calculated as the average hearing threshold at 0.5, 1, 2, and 3 kHz^{19 (link)}. Normal hearing was defined as an MHL ≤ the 25-decibel hearing level (dB HL).
For all patients presenting with typewriter tinnitus, vestibular function tests (VFTs), including the bithermal caloric test; recording of ocular vestibular evoked myogenic potentials (oVEMPs) and cervical VEMPs (cVEMPs); and the rotational chair test, were recommended at the initial visit. Of the 21 patients, 10 underwent at least one of these VFTs. Vestibular function was considered abnormal when: (1) canal paresis (CP, %; calculated using Jongkee’s formula) was > 25%²⁰ ; or (2) when the VEMPs were reduced or absent on the affected side (oVEMP or cVEMP asymmetry ratio >40% or >33%, respectively^{21 (link)}); or (3) when the slow harmonic acceleration (SHA) phase of the rotational chair test exhibited a reduced gain or an increased phase lead, in at least three consecutive frequencies of those tested (0.01, 0.02, 0.04, 0.08, 0.16, 0.32, and 0.64 Hz), using the manufacturer-provided normative data as references (Neuro Kinetics, Inc., Pittsburgh, PA, USA)^{22 (link)}.

We conducted a prospective cross-sectional study (convenience sample), between February 2015 and May 2020, of all cases presenting with acute dizziness at the ED in a tertiary referral center. 1677 patients were screened for Acute Vestibular Syndrome (AVS) as part of a large cross-sectional study (DETECT—[Dizziness Evaluation Tool for Emergent Clinical Triage]), of which 152 met the inclusion criteria and were enrolled. Inclusion criteria consisted of a state of continuous dizziness, associated with nausea or vomiting, head-motion intolerance, new gait or balance disturbance and nystagmus. Patients were excluded if they were younger than 18 years, if symptoms lasted < 24 h or if the index ED visit was > 72 h after symptom onset. Figure 1S (Appendix) shows a flow diagram with all screened patients, inclusions and exclusions of dizzy patients. All enrolled patients underwent when feasible a thorough physical examination, Caloric Testing and vHIT testing. All patients received an MRI either at the index visit or a second, delayed MRI if there was no acute MRI indicated based on clinical grounds or if the first MRI was non-diagnostic. The delayed MRI served as a reference standard for stroke detection. Enrolled patients were clinically re-evaluated between day 3 and day 10, at day 30 and day 90. All images were reviewed by a certified second blinded neuroradiologist, discrepancies were resolved by consensus and inter-rater concordance reported. Figure 1 shows the two investigated tests, the required equipment, stimulation modalities and recording setup.Fig. 1

Technical setup for the caloric exam compared to the Video-Head Impulse test. Diagram comparing the technical setup for the caloric exam with that of the vHIT; calorics are performed in the dark on a patient in a supine position and head rest positioned at 30° from horizontal. The outer ear canal on each side is irrigated sequentially for 30 s (at 30° C cold and 44° C warm water) and the resulting eye movements recorded for a duration of 3 min using VOG goggles. The whole procedures takes up to 30 min including waiting intervals of 5 min between irrigations. The vHIT is performed in a normal lit room on a upright sitting patient. The head is moved rapidly from side to side (20 times in an impulse-like motion) and eye movements are recorded using adapted vHIT-goggles. When done correctly, the vHIT takes under 5 min

We performed caloric tests irrigating sequentially both ears with warm (44 °C) and cold (30 °C) water for 30 s and a total water volume of 250 ml (Vario Otopront device) in patients lying 30° supine (Fig. 2S, supplementum, panel A). Intervals between irrigations were 5 min long, starting first with warm irrigation on the right ear. Convection flows of inner ear fluid (particularly in the horizontal semicircular canal) produced horizontal eye movement responses (nystagmus), which were recorded in darkness (blocked visual fixation) with a calibrated VOG device (EyeSeeCam, Munich). The Cut-off for pathologic Caloric responses was 20% asymmetry [17 ], which was calculated using Jongkee’s formula [18 (link)] after correcting for spontaneous nystagmus.
In contrast, vHIT was performed solely on the lateral canal by fast passive horizontal head movements (high frequency, 10–20° head excursion in 100–300 ms corresponding to a 1000–6000°/sec² acceleration) in room light during visual target fixation at > 1 m distance. We recorded head and eye movement velocity with a head mounted infrared highspeed camera (EyeSeeCam, Munich) connected to a laptop by USB (Fig. 2S, Panel B). VOR gain values were derived from eye velocity divided by head velocity at 60 ms after HIT onset [19 (link)]. vHIT exams were classified as abnormal based on VOR gains (Gain < 0.79 based on own laboratory normative values) and the presence of corrective saccades. Additionally we collected information on age, gender, duration of symptoms, and other associated relevant otological or neurological symptoms.

The vestibular system responds to sensory inputs across a range of frequencies going from low frequencies of approximately 0.05 Hz (e.g. postural sway) to high frequencies around 5 Hz during rapid head movements [28 , 29 (link)]. Canal function was evaluated at different velocities with 1- bithermal caloric test (33 °C and 44 °C) for low velocities, 2- earth vertical axis rotation (EVAR) with 40°/s² acceleration and deceleration along a vertical axis for medium velocities [30 (link)] and 3- head impulse test (HIT) to evaluate the 6 semicircular canals at high velocities [30 (link)]. Otolithic function was assessed by using cervical vestibular evoked myogenic potential (c-VEMP) with short tone bursts (750 Hz, 4.1/s and 6-ms duration) delivered by air and bone conduction with a control of the electromyogram level for each stimulation (Difra-Neurosoft® system Bruxelles, Belgium) [31 (link)]. This later characteristic allows the selection of the responses obtained for the same level of EMG. The EVAR test was performed using a computer-driven rotatory chair (SAMO®, Caen, France), and the vestibulo-ocular responses (VORs) were recorded by electronystagmography more adapted to young children than video recordings. Description of the EVAR experiment apparatus and protocol has been described in detail in previous publications [31 (link)].
For the bithermal caloric test, the Jongkees formula was applied [32 (link)]. Values for relative valence and directional preponderance for children were considered normal when < 15%. The responses to bithermal caloric test were categorized as either normal, absent bilaterally, partially and symmetrically impaired (bilateral symmetric hyporeflexia), or partially asymmetrically impaired (one side being either areflexic or hyporeflexic compared to the other side).
The HIT test was either normal (no catch-up saccade observed) or abnormal (presence of a catch-up saccade in at least one direction). Visible catch-up saccades indicate that the canal has lost at least 75% of its function.
For the c-VEMP, we studied the P and N latencies (ms), amplitude of P-N (mV) and the response thresholds (dB). The VEMP results could either be normal (present on both sides with a symmetric P-N amplitude at 100 dBHL), absent bilaterally (with no detectable P and N at 120 dBHL), partially symmetric (positive responses with thresholds > 100 dBHL), or partially asymmetric (difference of thresholds between sides > 10 dBHL and/or P-N amplitude difference exceeding 100 μV between the 2 ears).
For the EVAR test, we measured the time constant and maximal initial slow phase velocity of the VOR. Canal VOR time constant was calculated from the curve of the decay of slow phase velocities over time (measured from the start of the fastest slow phase until the extinction of the nystagmus); it corresponded to the time necessary to cover about one-third of the total area subtended by the slow phase velocity/time curve [31 (link)].

During the caloric test, EEG and nystagmus data were concurrently captured.

Saline electrode caps (EGI, Geodesic Sensor Net, 64 channels, 1,000 Hz) were used to record EEG data throughout the caloric test, amplified by a Net Amps 400 amplifier. The distribution of electrodes was based on the international 10–20 system, and the impedance was guaranteed to be less than 50 kΩ during the experiment. The volunteers were reminded before the experiment to refrain from unnecessary head and body movements while collecting data to prevent significant data contamination. EEG data for PBA was defined as the data from 50 s before stimulation, for POVA as the data from 20 s centered on the time point of the strongest nystagmus, for POFS as the data from 10 s following the turn-on of the fixation lamp, and for POR as the data from 10 s following the turn-off of the fixation lamp.

As illustrated in Figure 1A, volunteers were asked to remain awake with their eyes open during the experiment while lying supine in a dark field with their heads raised by 30° on a firm cushion to place their horizontal semicircular canal in a vertical posture. The caloric test was performed by perfusing each ear with cold (24°C) and warm (50°C) gases. Gas perfusion was performed from the external auditory canal in four conditions: cold gas into the right ear (RC), cold gas into the left ear (LC), warm gas into the right ear (RW), and warm gas into the left ear (LW). The duration of each perfusion was 60 s. It is important to ensure that the nystagmus of the subject due to the previous perfusion has completely subsided before each gas perfusion (no less than 5 min). After reaching the maximum nystagmus, the fixation lamp was turned on to suppress the vestibular activation. For each condition, we divided the entire experiment into four phases: the phase before stimulation (PBA), the phase of vestibular activation (POVA), the phase of fixation suppression (POFS), and the phase of recovery (POR), as shown in Figure 1B.