Hearing impairment in MELAS: new prospective in clinical use of microRNA, a systematic review

We conducted a literature search on PubMed, Scopus and Google Scholar using the following keywords: “MELAS, Hearing loss, Hearing Impairment, Temporal Bone, Otoacustic Emission (OTOAE), Auditory Brain Answer (ABR), and microRNA (miR)”. A total of 250 articles were found.

After screening their abstracts, 38 papers were read entirely, 15 were excluded because not relevant to this study, and 32 were selected to be included in our review.

In patients with MELAS, SV displays severe atrophy that affects all turns of the cochlea [10]; additionally, the SGs are reduced in number when compared with SGs in gender- and age-matched healthy subjects [10, 11]. The residual SV cells show vacuole formation and small dark cells which are normally not present in the structure; residual SGs are affected by several degenerative processes such as loss of cell membrane outline and loss of nuclear definition [12].

Takahashi et al. [10] reported that Organ of Corti showed no alterations, and that inner and outer hair cells were normal in number and function; however, these findings could be due to the fact that in this study patients were under 30 years old; unfortunately, other MELAS temporal bone studies descriptions are scarce.

The greater involvement of SV and SGs compared to hair cells might be due to the fact that in these structures the concentration of mitochondria is higher than in hair cells [13].

Mitochondrial mutations in the temporal bone have been studied, but the reported rates are inconsistent across studies. Takahashi et al. reported that SV and Organ of Corti were the most affected structures, with a load of mutations between 78% and 85%, respectively [10].

Koda et al. [11], instead, reported a higher mutation load in SGs than in hair cells and SV and this is partially consistent with data observed in human temporal bone, where SV is more affected by damage than SGs. This is consistent, in part, with the findings reported by Takahashi et al. [10], who in the Organ of Corti observed a mutation load (indicative of mitochondria disorders) higher than in the SV.

We suggest that these inconsistencies can be explained by mitotic segregation. The random distribution of mitochondria at the time of cell division modifies the distribution of mitochondria [14]; thus, the temporal bone changes can be very different among patients affected by MELAS mutation, which might explain the different phenotypes.

Both MELAS mutation or mtDNA deletion in mitochondria modify the production of cytochrome oxidase complex IV. It has been shown that the resulting biochemical deficit of cytochrome although not directly responsible for SG and SV loss, is directly correlated with an increase of Reactive Oxigens Species (ROS) production [15] that induces damage in different parts of the cochlea. The ROS can act on different structures of the cochlea thereby damaging SV, SGs and hair cells, which could further explain the lack of consistency observed in the temporal bone studies [16].

The auditory tests of patients with MELAS that have been analyzed in temporal bone studies present flat and down sloping curves always associated with altered word discrimination. The auditory tests show a progression in SNHL correlated with time since mitochondrial disease onset carries a direct relationship with the aggressiveness of the pathology [17‐20]. The down-ward sloping curve is observed even when the number of cells of the Organ of Corti is preserved [12]. This can be explained by a reduced function of the hair cells, probably related to the same degeneration observed in the residual SV and SGs.

SNHL in patients with MELAS is commonly bilateral [5, 6, 19, 20]; the unilateral form is present only in 2% of cases [4, 17]. In both forms, SNHL affects high frequency in the onset (75%) and, then involves mild and low frequency [4‐6, 17, 19, 20]; in the remaining 25% of case SNHL affects all frequencies in its onset [4, 17].

In clinical studies, patients with MELAS are evaluated, in addition to PTA, with a number of other tests, including Transient Evoked Otoacustic Emission (TEOAE) [4], Otoacustic Emission (OTOAE) [4, 17], Auditory Brain Response (ABR) [4, 5, 17], Psycoacusting Tuning Curves (PTC) [5], Distortion Product Otoacustic Emission (DPOAE) [6, 17], electrocochleography [6, 17], and electrically-evoked compound action potentials [5].

Zwirner et al. [4] observed that MELAS patients suffered from a mild form of SNHL affecting the high frequencies [18] and from moderate to severe SNHL forms involving all frequencies. The word recognition test score was normal in subjects with mild SNHL and abnormal in subjects with moderate to severe SNHL, with a score depending on the SNHL severity. Patients with SNHL with a loss of 40 dB showed normal OTOAE; those with moderate to severe SNHL presented no OTOAE response. In this study, ABR was recorded using a stimulus that consisted of alternating clicks presented at a rate of 16.7/s and were generated by square-wave electrical pulses of 0.1 milliseconds duration. Stimuli were presented monaurally at 80, 90, and 100 dB normal Hearing Level (nHL). Average values of 2000 trials were obtained on stimulation of each ear. In all patients, ABR was normal in latency and amplitude.

Kullar et al. [5] reported that 8/11 MELAS patients with mutation m3243A > G suffered from SNHL, which ranged from mild/moderate hearing loss in high frequencies (5/11 patients) to severe/profound hearing loss (3/11 subjects) spanning all frequencies. These results can be described in terms of the auditory threshold shapes described above, and summarized by both flat and down-ward sloping curves (Fig. 1). Those patients showed complete absence of TEOAEs in all forms of SNHL, reflecting a completely loss of function in outer hair cells. In this study, ABR was recorded using a click stimulus with alternating polarity delivered at a suitable sensation level to give a clear response. The sensation level was predetermined by the mean hearing level from each ear at 2/4 kHz: 440 dB Hearing Level (HL) used click stimulus at 70 dB nHL, 40–60 dB HL used click stimulus at 80 dB nHL, and 460 dB HL used click stimulus at 90 dB nHL. Contralateral masking was applied when required. In two of the 3 patients with profound SNHL, the ABR was not recordable; in the remaining patients, ABR waves displayed normal latency and amplitude even in patients with SNHL. PTC, which allows functional evaluation of inner and outer hair cells at the same time, showed no tip shifts in patients with normal hearing and mild SNHL; shifts at 1 kHz were observed in patients with moderate to severe SNHL in 66% of cases. The shift at 1 kHz indicates complete loss of inner and outer hair cells in the middle turn of the cochlea (Fig. 2).

Santarelli et al. [6] analyzed data from 10 patients with MELAS. They reported a flat threshold curve in all patients. Only 20% of subjects suffered from severe to profound SNHL, while the remainder of patients showed mild to moderate SNHL. DPOAE were detected in 1 ear in 6 out of 10 (60%) of patients. DPOAE responses were identified only at low frequencies in 3 out of the 6 subjects. The DPOAE testing results indicate that outer hair cells function is preserved in some portion of the cochlea, and in particular in HCs in basal turn. One of the two patients with severe to profound SNHL showed preserved DPOAE in both ears, but the ABR waves were not detectable.

Electrocochleography showed normal results in both ears in terms of potential peak amplitude but, the potential displayed lower amplitude when compared with potentials recorded from patients with normal hearing. In the other MELAS patients (i.e., patients with moderate SNHL) electrocochleography testing showed potentials similar to those recorded from normal hearing subjects in terms of peak amplitude, while the evoked potential was altered and resembled the shape typically recorded from patients with hearing impairment.

Sue et al. [17] analyzed 20 patients with MELAS and found moderate to profound SNHL in 78% of patients. The auditory threshold presented a down-ward curve at the onset of SNHL, which then became flat with the disease progression and aging. Only 50% of patients displayed normal speech recognition, suggesting a good retro-cochlear function. ABR was performed using a rarefaction click stimuli; stimulus intensity was at 65 to 70 dB above hearing thresholds or at maximal stimulator output (110 dB) if the hearing threshold was above 40 dB. ABR showed absent or delayed wave I in one ear at least in 61% of patients, but waves III and V were always present. Electrocochleography was performed in 11 patients and the test result was found to be normal in 64% of patients; ABR was not recordable from two patients and in the last two the click evoked electrocochleogram was broad. DPOAE were not detectable when observed in the range of frequency interested by severe to profound SNHL in 7/11 patients, but responses were present and electrically recordable when the SNHL was within 40 dB.

In the study by Vandana et al. [19], 6 children and 2 adults with MELAS were investigated; 3 out of 8 patients suffered from moderate to severe SNHL; in two cases the SNHL was subclinical; 1 presented a mild SNHL. All patients presented a down-ward sloping curve and OTOAE were absent in 50% of subjects. Auditory evoked potentials were recorded using a standard protocol. Only in 1 patient ABR showed absence of signal, which is indicative of retrocochlear disease.

In a large cohort study, Iwanicka-Pronicka et al. [20] showed that the shapes of PTA were correlated with specific mitochondrial mutations. They observed a down-ward sloping curve in patients with m.1555A > G and a pantonal shape with slight down-ward sloping at the high frequencies in the patients with mutation 3243A > G. Their results were statistically significant (p < 0.05). A prevalence of moderate SNHL was observed in m.3243A > G, and the 97% of patients with this mutation had a family history of hearing loss.

Overall, the studies described above show the limitations of the pure tone auditory test and of the electrophysiological tests. A comparative analysis of results highlighted that there are major inconsistencies between the outcome of OTOAE/TEOAE/DPOAE and ABR testing.

The presence of OTOAE/TEOAE waves were reported even for SNHL with threshold higher of 40 dB [4, 5], where absence of response would be expected. Other studies showed absence of OTOAE/TEOAE response only when SNHL is moderate to severe (> 40 dB) [4, 19].

ABR waves follow a similar trend. Some studies described either normal latency and amplitude in presence of moderate to severe SNHL [3, 18], or instead reported absent in mild forms of SNHL [4, 5, 16].

MicroRNAs (miRs) are endogenous, small sequences of non-coding RNA [21], that have been shown to modulate a wide range of biological function. MiRs regulate post transcriptional mRNA expression binding the 3′-untraslated region of the complementary mRNA sequence and acting as gene modulator [22]. Change in their concentration has been observed in several diseases, including inflammation and aging [23]. miR levels increasing are specifically related to the damaged structure [22]. Their levels associated to hearing disorders have been investigated [24‐27] using miRs, which, due to their high stability in blood, can be easily identified [25].

We speculate that miR levels can be the expression of damage but at the same time they might influence the mitochondrial metabolism by acting on it; they could downregulate the Sirtuin (SIRT1) action by increasing ROS [24], suppress the function of Blc-2 by increasing the apoptosis in cells [25], or increase the function of Bak by causing cell death through increased apoptosis [26] (Fig. 3), and then they might modulate the expression of specific genes by increasing apoptosis [28]. The increase of miRs 34a, 29b, 76, 96, 183 and 431 have been identified as potential markers of hearing damage in animal studies [21‐26]; among them, only miR34a has been validated in humans [27].

In humans, miR34a increase has been found to be correlated with hearing loss in aging. In particular, its concentration in blood is anti-correlated with the scores of Pure Tone Averages testing [24]. miRs s16- 5p, 24-3p, and 185-5p were identified in subjects with SNHL exposed to noise and the increase in their levels was correlated with the severity of SNHL [23].

miR increase was directly correlated with reduced responses or completely absence of OTOEA; when the hearing damage was electrically identified, the miR 34a and miR-29b levels were also increased in blood [24, 25], showing a specific correlation between the miRs level and the altered response in OTOAE.

The miRs that express cochlear damage are very specific for each structure as shown in Fig. 4, but so far only miRs sensitive to general damage (miR 76) or to hair cells and/or SG damage 34a 96 have been tested in humans.

Jong et al. identified the role of miR-299-3p in the aging vessel process [29] but, until today, nobody use it to evaluate the function of stria vascularis both in animal or human studies; we think that it could be useful due to similar cells present in both structures (vessel and SV).

Meseguer et al. [30] showed the power of miR − 9/9*as a detector of brain damage in humans. Using cybrids from two patients affected from mutation 3243A > G and m8344 A > G, they found that the overexpression of this small molecule was able to increase the mitochondrial dysfunction in MELAS and at the same time provide a measure of brain degeneration. We think that this miR could be helpful to investigate the SNHL in the retrocochlear portion due to its possible increased level when a brain degeneration is ongoing.

In conclusion, the studies described above suggest that miRs can help identify the cells involved in SNHL. Change in miR levels is the expression of cells damage but at the same time miRs can directly modulate the mitochondria metabolism by increasing apoptosis.