The aim of this study was to explore in detail the vestibular function in a large cohort of children and young adults with genetically confirmed FRDA, most of them being early onset patient (40/43), and to combine these observations to clinical, genetic, auditory and oculomotor evaluations. Our main findings are:
FRDA patients have vestibular dysfunction. Our patients’ canal function is abnormal at high and middle head velocities (assessed by HIT and EVAR) but normal for low head velocities (assessed by bithermal caloric test). Otolith function (assessed by cVEMP with bone and air conducted stimuli) is severely impaired.
FRDA patients have abnormal neural conduction in the central auditory pathways: 46% of our FRDA patients have ABRs dys-synchronisation.
Our FRDA patients present oculomotor abnormalities similar to those observed in previously published series [10, 12, 38]: fixation instability, saccades dysmetria, and saccadic pursuit.
These audio-vestibular and oculomotor anomalies are more frequently observed in subjects with longer disease duration and more severe forms of the disease (FARS, ICARS and SARA scores).
Previous studies have shown that vestibular anomalies are common to most FRDA patients [10, 11] but little is known concerning the peripheral or central origin of this deficit. We hereby discuss our vestibular results and see how these observations can help localize the neurophysiological impairment.
Our study shows that majority of our FRDA patients present partial vestibular impairment of canal function for high (HIT) and middle head velocities (EVAR) with mostly normal low head velocities canal function (bithermal caloric test). They also have severe impairment of otolith vestibulospinal function.
The time constant of EVAR responses is greatly reduced and this is linked to longer disease duration in our population. This observation corroborates previous findings of bilateral VOR deficit in FRDA patients . Time constant of EVAR response is closely linked to the concept of “velocity storage”. This concept was introduced in 1979 by Raphan et al. to account for the fact that the VOR response to a rotatory impulsion followed by a constant velocity rotation lasts longer than the actual initial peripheral vestibular stimulation . In normal subjects, the time constant of the observed nystagmus is two or three times longer than that of the corresponding activity in the vestibular nerve. The velocity storage concept suggests the presence of a neural integrator in the central vestibular system that extends the VOR . This function is implemented in the cross commissural pathways between the right and left sided vestibular nuclei and under control of the cerebellum (e.g. the nodulus and uvula). Shorter time constant can therefore be linked to alteration of those central structures. But peripheral vestibular lesions can also lead to shorter time constant  and this hypothesis can be supported in our population by the observation of a shorter time constant in subjects with abnormal response to bithermal caloric test compared to those with normal caloric responses.
The prevalence of abnormal VOR at high head velocities might be under evaluated in our study since clinical HIT only detects canal impairment with at least 75% of canal functional loss. Even though the possible under evaluation, abnormal clinical HIT with clearly visible catch-up saccades is observed in more than half of our population. This test evaluates the reflex pathway from the vestibular canal receptors to the oculomotor nuclei. It is thought to be a purely peripheral vestibular test because it is too fast for other oculomotor control systems to participate (such as smooth pursuit, optokinetic, cervico-ocular reflex) . However, abnormal clinical HIT has also been reported in pontine cerebellar stroke involving vestibulo ocular pathways or cerebellar lesion of the nodulus . One of the limitations of clinical HIT is that, in FRDA, it can be difficult to differentiate true positive clinical HIT (where catch up saccade comes from an impaired canal) and false positive HIT (coming from the overlap of a hypermetric saccade or a saccadic pursuit with the VOR). Especially since the majority of our patients had abnormal saccades (either hyper or hypometric), abnormal eyes movements and saccadic pursuit. Video recorded HIT (vHIT) could differentiate false positive HIT (of central origin) from true positive HIT (of peripheral origin) and detect partial canal functional loss (less than 75%). This test was not available in our department at the time of the study.
To incriminate the cerebellar dysfunction for the anomalies observed in the VOR is tempting. But all of our subjects (except one) could perfectly inhibits VOR and OKN with fixation, and this observation does not support the hypothesis of a cerebellar impairment. Neurons important in VOR-suppression are located in the cerebellar flocculus and paraflocculus . The ability to perform a VOR-suppression by fixation clearly differentiates FRDA from other cerebellar ataxia .
Two previous publications mention a higher prevalence of bilateral hyporeflexia in FRDA population with reduced caloric responses respectively observed in 66% (8/12) and 25% (4/16) of adults with FRDA [12, 38]. Unlike these observations, only two of our subjects (2/40, 5%) showed bilateral hyporeflexia and one a complete bilateral areflexia (1/40, 2,5%). The short duration of the disease in our young population might explain the low prevalence of caloric dysfunction, and this prevalence could increase as the disease evolves. A normal caloric function does not exclude peripherical vestibular impairment neither does it mean that the vestibular signs observed are strictly central. Indeed, preserved low frequency canal function can be observed in some peripherical inner ear diseases with selective canal impairment (reduced middle and high frequency canal function with normal low frequency canal function - e.i. Usher type I syndrome) .
Responses to otolithic stimulation are abnormal in almost all subjects with absent or high threshold cVEMPs responses. However, when a response was observed, latencies were normal in all but one subject. Although cervical VEMPs have mostly been applied to peripheral vestibular disorders, many studies report abnormal c-VEMPS response in various central pathologies, usually associated with long latencies [41, 45, 46]. The existence of mostly normal latencies in our population is in favor of a peripherical vestibular impairment. However, we did not test the robustness of the cVEMP responses with longer stimulation duration. Observation of signs of fatigability of the cVEMP responses could be an argument for neuropathy at the level of the vestibulospinal pathway.
The peripheric or central origin of our observed vestibular dysfunctions cannot be clearly determined based on our results. The vestibular impairment limited to middle and high head velocities but respecting low head velocities responses as well as the histopathological findings of no hair cells lesions but gliosis in the vestibular nucleus  and abnormalities of the spiral ganglion  and vestibular nerve  in FRDA patients could support the possibility of a vestibular neuropathy with central extension to the vestibular nuclei. But an association to a possible alteration of the peripheral vestibular system is not excluded especially since normal latencies were found in most of our cVEMPs responses. In order to determine the origin of the vestibular lesion further studies should be done in this population focusing mainly on:
Clinically apparent hearing loss is relatively uncommon in patients with FRDA [4, 14, 48, 49] and, if present, is generally mild. In previous studies, only 8 to 13% have demonstrated elevated hearing thresholds [4, 14, 15]. Thirty-seven of the 40 tested subjects (92.5%) had normal pure tone audiometric thresholds (mean thresholds < 20 dB). Three subjects only had a mild bilateral SNHL.
However, even if FRDA patients have a normal pure tone audiometry they frequently report speech perception problems [15, 17]. These difficulties can be linked to disorders in the central auditory pathways. Some studies on FRDA patients report speech difficulties associated with absent or distorted auditory brainstem responses while pre-neural response from the cochlea hairs cells are normal [19, 20]. In the literature, the incidence of abnormal ABR varies from 30 to 100% [15, 17]. In our series, 51% demonstrated abnormal ABR (either increased latencies or dys-synchronization). Most papers link this observation to the presence of an auditory neuropathy/dysynchrony . Neuro histological studies in patients with FRDA support this hypothesis. One study shows damage to the cochlear nerve without damage to inner ear structures  and another shows spiral ganglion degeneration . A recent paper suggests that the Friedreich’s auditory neuropathy is linked to reversible energetic failure explaining why FRDA patients can have normal ABRs in response to short duration testing but abnormal ABRs for prolonged series of stimulation . In our population, we observed dys-synchronization of the ABR waves appearing for long series of stimulation in 19% of our tested ears. Information regarding the hearing evaluation were extracted from a descriptive database and detailed information about the ABR measurements such as waves latencies could not be retrieved. Further studies with precise stimulation protocol, such as the one used in , should be done in order to document more precisely the ABR fatiguability.
Ell et al. observed a significant decrease in hearing acuity in the low frequency range in the Friedreich population compared to normal controls  in absence of conductive problems. The same pattern is found in our population, particularly the ones with abnormal ABRs (Fig. 2). Audiograms with increased low-frequencies thresholds are typically found at the beginning of inner ear diseases such as Meniere’s disease, or in certain forms of hereditary hearing loss. But such reverse slope audiograms are also described in both adults and children with auditory neuropathy spectrum disorder (ANSD) [50, 51]. Several explanations have been given for these reverse slope audiogram in ANSD: impairment of auditory processing of temporal information (since the timing of auditory nerve discharge play a role, particularly, in the encoding of low spectral acoustic signals) or neuropathological processes affecting particularly the apical fibers encoding for low-frequency sounds because of their longer course outside of the cochlear nucleus and greater axonal diameter . This reverse slope audiogram was more pronounced in our subjects with abnormal ABR and this additional observation support the hypothesis of ANSD in FRDA patients. Signs of auditory neuropathy were more frequently observed in subjects with longer disease duration and seem to be associated with increased ICARS, FARS and SARA scores. Therefore, hearing difficulties can be considered as a sign of disease severity and disease progression.
Our findings confirm that children and young adults with FRDA likely face communication and educational challenges due to auditory neuropathy. For these patients, auditory tasks that are more complex than pure tone detection (most notably speech perception especially in noisy environment) may be dysfunctional far beyond what is indicated by the audiogram. All patient with FRDA should benefit from an auditory assessment including at least a pure tone audiometry, DPOAEs, ABR and a speech in noise test. In case of hearing difficulties, the introduction of personal FM system and communication training (i.e. listening tactics, lip-reading cues) proved to be helpful [15, 17].
The oculomotor abnormalities we found in our FRDA patients include fixation instability, dysmetric saccades and saccadic pursuit. The type and prevalence of the ocular motor anomalies observed are very similar to previously published series [10, 12, 38]. As opposed to other published series , we did not observe an increase of the overall mean saccade latencies for our population. This discrepancy might be explained by the relatively young age of our population associated to shorter progression of the disease. We did, however, observed a great variability in saccade latencies (range 56.9 ms to 391 ms). Analysis of saccadic accuracy demonstrated that 76% of our subjects were hypermetric and 12% hypometric. As previously observed , hypermetric and hypometric saccades were frequently combined in the same individuals.