1. INTRODUCTIONTOP
Tinnitus is the conscious perception of sound heard in the absence of physical sound sources external or internal to the body (Eggermont & Tass, 2015). Epidemiological studies report that tinnitus roughly affects 10 % of the adult population (Hall et al., 2015) and severely disturbs the quality of life of about 1–2 % of adults by producing anxiety, annoyance, irritability, disturbed sleep patterns, and depression (Cobo, 2015; Diges, Simón, & Cobo, 2017; Van de Heyning et al., 2007; Vio & Holme, 2005).
Similar to in phantom limb pain, tinnitus perception seems to be the correlate of maladaptive attempts of the brain at reorganization due to deprived sensory input. Therefore, hearing loss (HL) is the most important risk factor for developing tinnitus (Kleinjung, Steffens, Struz, & Langguth, 2009). The central auditory system compensates for diminished input by upregulating its responsiveness in central circuitries. Central compensation that follows reduced auditory nerve activity may occur first at the level of the auditory brainstem, from where altered activity patterns then spread to ascending auditory nuclei. Electrophysiological and functional imaging measurements in humans and animals suggest the following neural correlates of tinnitus in the auditory system (Eggermont, 2012):
| • | Increased neural synchrony (hypersynchrony) |
| • | Increased spontaneous firing rates (hyperactivity) |
| • | Reorganization of tonotopic map |
Tinnitus can occur at both sub-cortical and/or cortical levels, suggesting two different tinnitus subtypes: cochlear and central (Noreña, 2011; Milloy, Fournier, Benoit, Noreña, & Koravand, 2017). Cochlear tinnitus results from a hyperactivity at the acoustic nerve and is the subtype taking place in salicylate induced tinnitus in animal models. Central tinnitus, on the other hand, outcomes due to cortical changes (mainly hypersynchrony and tonotopic map reorganization) due to HL, and is the subtype happening in noise induced tinnitus in animal models (Noreña, 2011).
Tinnitus and HL are highly correlated. According to Eggermont (2014), the prevalence of tinnitus is a cubic-root function of the prevalence of significant HL (HL > 25 dB from 500 Hz to 4 kHz). However, although chronic tinnitus is often accompanied by some kind of hearing deficit, it is still unknown how HL can actually produce tinnitus, as many HL impaired people do not develop tinnitus. The intriguing relationship between tinnitus and HL is even more disconcerting as 25 % of tinnitus participants in a research study had normal hearing up to 8 kHz (Roberts, Moffat, & Bosnyak, 2006).
Many researchers have investigated how the tinnitus occurrence and the HL curve (the audiogram) shape are related (König, Schaette, Kempter, & Gross, 2006; Langguth et al., 2017; Schecklmann et al., 2012; Sereda et al., 2011; Shekhawat, Searchfield, & Stinea, 2014). For example, the perceived frequency of tinnitus, also called the tinnitus pitch (TP in the following), is usually associated with frequencies showing HL, i.e., high pitch tinnitus is associated to high frequencies HL, and low pitch tinnitus with low frequencies HL (Shekhawat et al., 2014).
Different theories have been proposed to relate different types of HL with TP (Schecklmann et al., 2012). Roberts, Bosnyak, Bruce, Gander, and Paul (2015), based in similarity judgments, reported that tinnitus subjects matched their TP near the edge frequency of the audiogram, that is, the frequency at which the HL commences. Thus, one theory proposes the edge frequency as the mechanism triggering the tinnitus by a lateral inhibition imbalance, which results in an over representation of this edge frequency at cortical level (reorganization of the tonotopic map). According to this theory, the TP should correspond to the edge frequency of the HL.
Shekhawat et al. (2014) proposed the frequency associated to dead region as the most likely audiometric characteristic related to TP. Previously, Weisz, Hartmann, Dohrmann, Schlee, and Noreña (2006) demonstrated that 72.7 % of tinnitus sufferers had dead regions. The dead region is the cochlea zone where the inner ear cells (IHC) are not functioning. In fact, IHC damage has been identified as a prerequisite for auditory pathway deafferentation and tonotopic reorganization. Post mortem studies have demonstrated that IHC damage starts roughly at HL = 50 dB (Shekhawat et al., 2014). Therefore, the frequency at which HL = 50 dB (F50 in the following) was proposed by Shekhawat et al. (2014) as the most probable audiometric correlate of TP. Other possible audiometric correlates of TP were analyzed by Shekhawat et al. (2014), namely, the frequency at which HL approximately begins, that is the frequency for HL = 20 dB (F20 in the following), and the frequency at which HL is maximum (Fmax in the following). Notice that F20 could be assimilated to the edge frequency proposed by Roberts et al. (2015).
Alternatively, Schecklmann et al. (2012) suggested that tinnitus is caused by homeostatic plasticity, which compensates for deprived sensory input by increasing spontaneous firing rate and neural synchrony in the corresponding auditory pathway. According to this theory, the TP should correspond to the frequency for maximum HL, that is, Fmax.
Therefore, studies of relationship between audiometric characteristics, obtained from the HL curve shape, and tinnitus features, mainly the TP, are still relevant. Hence, the aim of this article is to provide the results of such a study in a cohort of 34 tinnitus volunteers, which undertook joint audiometric and tinnitus measurements in our laboratory. When analyzing the relationship between tinnitus features and audiometric characteristics, the following issues should be taken into consideration:
| 1. | The procedure to assess the tinnitus features, for instance, the tinnitus pitch. |
| 2. | The way the audiometric characteristics are obtained, namely, which attributes from the HL are used and how they are measured. |
| 3. | The statistical methods used for testing tinnitus features and audiometric characteristics. |
2. MATERIAL AND METHODSTOP
2.1. ParticipantsTOP
The study was approved by the Research Bioethics Subcommittee of the Spanish National Research Council (CSIC) and was conducted in accordance with the Spanish Law of Data Protection (RD1720/2007). 34 volunteers with tinnitus (21 men, age 51 ±14 years, 13 women, age 45 ± 11 years) were recruited through Spanish Tinnitus Associations and Tinnitus Clinics. Subjects with audiological surgical history (otosclerosis, tumors, head trauma…) were excluded. All participants in this study gave their written informed consent.
Columns 1–3 of Table 1 show the assigned number, sex and age, respectively, of each participant.
Table 1: Audiometric characteristics of participants.
| Subject | Sex | Age | HL subtype | F20left ear | F20right ear | F50left ear | F50right ear | Fmaxleft ear | Fmaxright ear |
|---|---|---|---|---|---|---|---|---|---|
| (*) | (Hz) | (Hz) | (Hz) | (Hz) | (Hz) | (Hz) | |||
| 1 | F | 45 | F | n/a | n/a | n/a | n/a | 125 | 125 |
| 2 | M | 63 | HS | 1498 | 1928 | 2599 | 1499 | 8000 | 8000 |
| 3 | M | 39 | F | 499 | 748 | n/a | n/a | 1500 | 750 |
| 4 | M | 44 | HS | 4665 | 4996 | 5998 | 9000 | 6000 | 8000 |
| 5 | M | 53 | HS | 2498 | 2496 | 3998 | 7980 | 8000 | 8000 |
| 6 | M | 50 | HS | 3496 | 998 | 7712 | 9000 | 8000 | 8000 |
| 7 | M | 43 | HS | 3664 | 6796 | 9000 | 9000 | 8000 | 8000 |
| 8 | F | 50 | HS | 6661 | 249 | 9000 | 9000 | 8000 | 8000 |
| 9 | F | 41 | CS | 167 | 208 | 500 | 749 | 6000 | 8000 |
| 10 | F | 67 | HS | 249 | 4496 | 9000 | 9000 | 6000 | 8000 |
| 11 | M | 28 | F | n/a | n/a | n/a | n/a | 6000 | 6000 |
| 13 | F | 45 | CS | 125 | 125 | 9000 | 2995 | 750 | 8000 |
| 14 | M | 62 | HS | 417 | 437 | 3363 | 2666 | 6000 | 6000 |
| 15 | M | 31 | ST | 1083 | 750 | 1624 | 1278 | 2000 | 2000 |
| 16 | M | 68 | CS | 499 | 624 | 5997 | 7996 | 8000 | 8000 |
| 17 | F | 47 | HS | 1498 | 4748 | 5330 | 6000 | 6000 | 6000 |
| 18 | M | 65 | HS | 125 | 1498 | 999 | 9000 | 8000 | 8000 |
| 19 | F | 61 | F | 749 | 5991 | n/a | n/a | 1000 | 8000 |
| 20 | M | 41 | F | 6991 | n/a | n/a | n/a | 8000 | 125 |
| 21 | M | 47 | ST | 2398 | 2332 | 3500 | 3249 | 3000 | 4000 |
| 22 | F | 45 | CS | 156 | 125 | 2798 | 3495 | 4000 | 6000 |
| 23 | M | 42 | F | 2498 | 2996 | n/a | n/a | 3000 | 3000 |
| 24 | M | 37 | F | 3998 | 7981 | n/a | n/a | 6000 | 8000 |
| 25 | M | 69 | HS | 498 | 249 | 1999 | 7197 | 6000 | 8000 |
| 26 | M | 44 | F | 5991 | n/a | n/a | n/a | 6000 | 6000 |
| 27 | F | 26 | F | n/a | n/a | n/a | n/a | 6000 | 8000 |
| 28 | F | 29 | ST | 3997 | 3331 | 5000 | 5000 | 6000 | 4000 |
| 29 | M | 58 | HS | 1996 | 2331 | 5327 | 6796 | 8000 | 8000 |
| 30 | F | 42 | F | 333 | n/a | n/a | n/a | 6000 | 125 |
| 31 | F | 41 | F | 5994 | n/a | n/a | n/a | 6000 | 8000 |
| 32 | M | 36 | HS | 2499 | 2428 | 4990 | 6000 | 6000 | 6000 |
| 33 | M | 75 | HS | 125 | 125 | 1374 | 2444 | 6000 | 8000 |
| 34 | F | 50 | HS | 2331 | 1998 | 7994 | 5994 | 8000 | 6000 |
| (*) F=Flat, HS=High-steep high-frequency, CS=Continuously steep, ST= Scotoma | |||||||||
2.2. Audiometric measurementsTOP
Each subject underwent HL measurements by pure-tone audiometry of both ears. HL curves were measured with the Clinic Audiometer GSI 60, using pure tones at 11 pre-specified frequencies (125, 250, 500 750, 1000, 1500, 2000, 3000, 4000, 6000, and 8000 Hz). 32 of 35 subjects have HL similar for left and right ears. On the other hand, 3 of 35 (18, 21, and 25) have HLleft ear and HLright ear significantly different. Looking at the HL curves of participants, four subtypes can be defined:
| 1. | Roughly flat HL (Figure 1; participants 1, 3, 11, 19, 20, 23, 24, 26, 27, 30, and 31). |
| 2. | High-steep high-frequency HL (Figure 2) (participants 2, 4, 5, 6, 7, 8, 10, 14 17, 18 (left ear), 25 (more in left ear), (29, 32, 33, and 34). |
| 3. | Continuously steep HL (Figure 3; participants 9, 12, 13, 16, and 22). |
| 4. | HL with a scotoma (Figure 4; participants 15, 21, and 28). |
Figures 1–4 depict mean HL curves (superimposed to individual HL) for these four subtypes. Column 4 of Table 1 shows the HL subtype of each participant.
|
Figure 1: Mean HL (superimposed to individual HL) for flat HL subtype and (a) right ear, (b) left ear. |
|
Figure 2: Mean HL (superimposed to individual HL) for high-steep high-frequency HL subtype and (a) right ear, (b) left ear. |
|
Figure 3: Mean HL (superimposed to individual HL) for continuously steep HL subtype and (a) right ear, (b) left ear. |
|
Figure 4: Mean HL (superimposed to individual HL) for HL with scotoma subtype and (a) right ear, (b) left ear. |
2.3. Audiometric characteristicsTOP
The frequencies at which HL attains 20 dB, F20, 50 dB, F50, and its maximum, Fmax, for left and right ears, are summarized in Table 2. To improve the estimation of F20 and F50, finer HLs are linearly interpolated first from measured HLs. This interpolation improves the estimation of the cutting of HL curves with HL = 20 and HL = 50. However, despite this interpolation, it was not possible to find F20 and F50 for some cases (mainly for flat HL subtypes), as HL curves do not reach these values. In these cases, n/a is used for the corresponding F20 and F50 values. The HL curves of some participants deserve special consideration. Firstly, HLs are greater than 20 dB for subject 13, in both ears, and for subject 18, in left ear. In these cases, we assign arbitrarily F20 = 125 Hz. Secondly, HL < 50 in subjects 6, 7, 8, 10, 13, 18, 21, 28, and 32. In these subjects, F50 are assigned to the closer frequency at which HL = 40 or 45 dB.
Table 2: Comparison of F20, F50, Fmax, and TP for the different HL subtypes.
| HL subtype | Subject | F20 (Hz) | F50 (Hz) | Fmax (Hz) | TP (Hz) |
|---|---|---|---|---|---|
| HS | 2 | 1748 | 2049 | 8000 | 5000 |
| 4 | 4830 | 7500 | 7000 | 4500 | |
| 5 | 2498 | 3998 | 8000 | 4400 | |
| 6 | 3496 | 712 | 8000 | 12500 | |
| 7 | 3664 | 9000 | 8000 | 7000 | |
| 8 | 3455 | 9000 | 8000 | 8000 | |
| 10 | 249 | 9000 | 6000 | 5000 | |
| 14 | 437 | 2666 | 6000 | 3100 | |
| 17 | 1498 | 5330 | 6000 | 5200 | |
| 18 | 811 | 5000 | 8000 | 3900 | |
| 25 | 498 | 1999 | 6000 | 3500 | |
| 29 | 1996 | 5327 | 8000 | 5500 | |
| 32 | 2499 | 4990 | 6000 | 3100 | |
| 33 | 125 | 1909 | 7000 | 3500 | |
| 34 | 2165 | 6994 | 7000 | 7500 | |
| CS | 9 | 167 | 500 | 6000 | 3000 |
| 12 | 416 | 2664 | 8000 | 4000 | |
| 13 | 125 | 5997 | 750 | 3800 | |
| 16 | 561 | 6998 | 8000 | 8300 | |
| 22 | 145 | 3145 | 5000 | 2000 | |
| ST | 15 | 917 | 1451 | 2000 | 2400 |
| 21 | 2332 | 3249 | 3000 | 12000 | |
| 28 | 3664 | 5000 | 5000 | 3000 |
2.4. Tinnitus featuresTOP
The tinnitus characteristics were assessed on the basis of the responses of the participants to the clinical evaluation sheet. The interview to participants included temporal (variability), spectral (pitch), and spatial (location) aspects of their tinnitus. Furthermore, additional information of participants was obtained, including anamnesis (clinic history), tinnitus severity through Visual Analog Scale (VAS) and a Spanish version of the Tinnitus Handicap Inventory (THI; Herráiz, Hernández Calvín, Plaza, Tapia, & de los Santos, 2001). The anamnesis included information about the history and descriptive characteristics of the tinnitus, their possible etiology, previous tinnitus treatments, and relevant comorbidities.
The type and pitch of the tinnitus were evaluated using the self designed Graphical User Interface (GUI) of Figure 5. Using this GUI, tones, ringing and hissing sounds can be easily generated. This is accomplished by creating a band-pass filtered noise. Two parameters (central frequency and bandwidth) determine the type of sound. The bandwidth is defined as a percentage of the central frequency. For instance, for tones the band-pass is very narrow (0.1 %). Ringing are narrowband noises (bandwidth lesser than 10 %) while hissing are wideband noises (bandwidth greater than 10 %).
|
Figure 5: MATLAB GUI for tinnitus pitch matching. |
For tinnitus pitch matching, participants were sat in front of the computer where the GUI runs. Firstly, they are trained in how tones, ringing, and hissing sound. Then, they are asked to identify roughly which sound is more similar to their tinnitus. After that, subjects are trained on the effect of central frequency and bandwidth on the sounds. Finally, a bracketing procedure is used to match as close as possible the sound generated by the GUI to its own tinnitus. It must be mentioned that some participants referred several types of sounds. In this case, the GUI is run consecutively to match all the sounds perceived by the participant.
Table 3 summarizes the resulting tinnitus characteristics of participants. Tinnitus is located either to left ear, right ear, bilateral (both ears), or the centre head. Some participants hear different types of tinnitus in each ear. A variety of etiologies were identified by the subjects, including sensorineural HL, conductive HL (otitis, Eustachian tube dysfunction), stress, noise, head trauma, and sinusitis. Some of the participants referred several possible origins of their tinnitus. When a tinnitus trigger is not clearly identified, the etiology is referred as idiopathic.
Table 3: Tinnitus features of participants.
| Subject | Sex | Age | Tinnitus duration (years) | Tinnitus location | Tinnitus sound | Tinnitus pitch (Hz) | Tinnitus bandwidth (%) | THI | VAS | Tinnitus Etiology |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | F | 45 | 0.7 | Bilateral | Ringing | 2200 | 1 | 60 | 75 | Idiopathic |
| 2 | M | 63 | 10 | Bilateral | Hissing | 5000 | 30 | 4 | 30 | HL induced |
| 3 | M | 39 | 2 | Left ear | Tonal | 6000 | 16 | 55 | Idiopathic | |
| 4 | M | 44 | 2.8 | Central | Ringing | 4500 | 1 | 74 | 75 | HL induced |
| 5 | M | 53 | 4 | Left ear | Ringing | 4400 | 4 | 10 | 15 | Stress. HL induced |
| 6 | M | 50 | 7 | Left ear | Tonal | 12500 | 20 | 25 | HL induced | |
| 7 | M | 43 | 3 | Left ear | Ringing | 7000 | 3 | 34 | 65 | Idiopathic |
| 8 | F | 50 | 10 | Bilateral | Tonal | 8000 | 22 | 30 | Head trauma induced | |
| 9 | F | 41 | 14 | Left ear | Tonal | 3000 | 38 | 55 | HL induced | |
| 10 | F | 67 | 1 | Left ear | Tonal | 4000 | 36 | 75 | Idiopathic etiology | |
| Ringing | 6000 | 5 | ||||||||
| 11 | M | 28 | 14 | Bilateral | Tonal | 9000 | 42 | 60 | Head trauma induced | |
| 12 | M | 72 | 11 | Left ear | Tonal | 4000 | 54 | 50 | Noise induced | |
| 13 | F | 45 | 8 | Bilateral | Tonal | 3800 | 1 | 64 | 70 | Idiopathic |
| Ringing | 250 | |||||||||
| 14 | M | 62 | 10 | Right ear | Ringing | 3100 | 1 | 18 | 65 | HL induced |
| 15 | M | 31 | 0.3 | Bilateral | Hissing | 2400 | 17 | 88 | 85 | Noise induced |
| 16 | M | 68 | 16 | Bilateral | Tonal | 8300 | 30 | 75 | Stress. HL induced | |
| 17 | F | 47 | 12 | Left ear | Tonal | 5200 | 50 | 75 | Eustachian tube dysfunction | |
| 18 | M | 65 | 20 | Bilateral | Ringing | 300 (RE) | 5 | 52 | 70 | Otitis. HL induced |
| 7000(LE) | 7 | |||||||||
| 19 | F | 61 | 0.4 | Left ear | Hissing | 1000 | 25 | 14 | 25 | Noise induced |
| 20 | M | 41 | 1 | Left ear | Ringing | 8000 | 1 | 50 | 60 | Idiopathic |
| 21 | M | 47 | 8 | Right ear | Ringing | 12000 | 5 | 10 | 55 | Noise induced |
| 22 | F | 45 | 0.3 | Central | Hissing | 100 | 60 | 56 | 55 | HL induced |
| 2000 | 50 | |||||||||
| 23 | M | 42 | 0.1 | Left ear | Ringing | 7000 | 10 | 36 | 65 | Otitis |
| Tonal | 7000 | |||||||||
| 24 | M | 37 | 1.1 | Left ear | Ringing | 7000 | 2 | 26 | 30 | Eustachian tube dysfunction |
| 25 | M | 69 | 0.8 | Left ear | Hissing | 3500 | 20 | 48 | 60 | Stress. Noise induced |
| 26 | M | 44 | 1.3 | Left ear | Tonal | 8800 | 62 | 70 | Stress. Idiopathic | |
| 27 | F | 26 | 0.17 | Central | Hissing | 1500 | 20 | 62 | 70 | Stress. Eustachian tube dysfunction |
| Tonal | 8000 | |||||||||
| 28 | F | 29 | 0.25 | Bilateral | Hissing | 3000 | 20 | 96 | 75 | Stress. Acoustic trauma. |
| 29 | M | 58 | 0.25 | Left ear | Tonal | 5500 | 22 | 35 | HL induced | |
| 30 | F | 42 | 0.42 | Bilateral | Hissing | 125 (LE) | 20 | 32 | 55 | Stress. |
| Tonal | 4000 (RE) | |||||||||
| 9000 (RE) | ||||||||||
| 31 | F | 41 | 1.3 | Left ear | Hissing | 1000 | 40 | 70 | 70 | Stress. Sinusitis. Eustachian tube dysfunction |
| 32 | M | 36 | 0.33 | Left ear | Tonal | 3100 | 50 (day) | 82 | Head trauma. Noise induced | |
| 80(night) | ||||||||||
| 33 | M | 75 | 0.25 | Central | Hissing | 4000 | 30 | 80 | 84 | Noise induced |
| Tonal | 3500 | |||||||||
| 34 | F | 50 | 1.5 | Bilateral | Tonal | 7500 | 65 | 76 | HL induced |
2.5. Data analysisTOP
Numerical proportions of different types of HL and tinnitus are appraised first by pie charts. Then, scatter plots and Spearman rank correlation analysis are applied to paired tinnitus and audiological variables. Spearman rank correlation is used to identify and test the strength of relationships between these variables (Diges, Simón, & Cobo, 2017). Positive Spearman correlation coefficients (ρ) between x and y variables denote that both variables increase monotonically, and vice versa, a negative correlation coefficient indicates that when x increases y decreases monotonically. The correlation between the variables is considered to be very weak for | ρ | ≤ 0.2, weak for 0.2 < | ρ | ≤ 0.4, moderate for 0.4 < | ρ | ≤ 0.6, strong for 0.6 < | ρ | ≤ 0.8, and very strong for | ρ | > 0.8.
3. RESULTSTOP
Numerical proportion analysis of each HL subtype (Figure 6) show that 44 % (15 of 34) of the participants have HL curves roughly flat at low frequencies and high steep at high frequencies; 32 % (11 of 34) of the participants have more or less flat HLs. HLs are continuously steep for 15 % (5 of 34) of participants. And, for the other 9 % (3 of 34), HLs have a scotoma at 4–6 kHz.
|
Figure 6: HL curves subtypes. |
Numerical proportional analysis applied to tinnitus laterality, tinnitus sound, and tinnitus etiology, of 23 subjects of HS, CS, and ST HL subtypes, affords the results depicted in Figure 7. Concerning the tinnitus laterality, Figure 7a, 43 % of subjects (10 of 23) allocate their tinnitus to left ear, in 35 % of subjects (8 of 23) the tinnitus is bilateral, 13 % (3 of 23) perceive the tinnitus in the head (central), and only 2 of 23 (8 %) assign their tinnitus to the right ear. Notice that there are people with several types of tinnitus, allocated to distinct parts of the head. In these cases, tinnitus is assigned to the dominant (more intense) tinnitus. Regarding the tinnitus sound, Figure 7b, the more frequent is tonal (39 %, 9 of 23), followed by ringing (35 %, 8 of 23), and hissing (26 %, 6 of 23). As before, when subjects refer to several tinnitus sounds, the more prominent is assigned. Finally, the predominant tinnitus etiology (Figure 7c) was sensorineural HL (HL induced in Table 3), with a percentage of 39 % (9 of 23), followed by noise (30 %, 7 of 23), idiopathic (13 %, 3 of 23), conductive HL (9 %, 2 of 23), and head trauma (9 %, 2 of 23). Notice that stress (see Table 3) was considered a comorbid effect and not a triggering cause of tinnitus.
|
Figure 7: (a) Tinnitus laterality, (b) Tinnitus sound, and (c) Tinnitus etiology. |
Table 2 summarizes the mean F20, F50, and Fmax for the participants with HS, CS, and ST subtypes, together with the corresponding tinnitus pitches. For subjects with unilateral tinnitus, the F20, F50, and Fmax values of the corresponding ear are chosen. For subjects with either bilateral or central tinnitus, the mean of left and right ears is selected.
Figure 8 shows the mean differences between TP and F20, TP and F50, and TP and Fmax. For HS HF and CS HL subtypes, it can be seen that F20 underestimates, Fmax overestimates, and F50 is the closest estimator of the TP. For ST HL subtype, the three variables underestimate the tinnitus pitch. Figure 9 shows the average HL curves for the three HL subtypes with the values of F20, F50, Fmax, and TP superimposed.
|
Figure 8: Mean differences between TP and F20, F50 and Fmax for (a) high-steep high-frequency HL subtype, (b) continuously steep HL subtype, and (c) HL with scotoma subtype. |
|
Figure 9: Mean HL curves for the three subtypes, with the values of F20, F50, Fmax, and TP overimposed. |
Figures 10–12 show scatter plots for F20 versus TP, F50 versus TP, and Fmax versus TP, for the three HL subtypes, respectively. Again, for the HS and CS subtypes, it can be seen that F20 underestimates TP, Fmax overestimates TP, and F50 is the best estimator of TP. For HL curves with scotoma, neither F20, F50 nor Fmax approach sufficiently to TP. However, taking into account that ST subgroup has only three participants, this assertion does not have enough statistical power. Since scatter results are similar for HS and CS HL subtypes, we could integrate both in just a subgroup. Figure 13 shows a joint scatter plot for both subtypes. Table 4 summarizes the ρ and p values obtained when applying Spearman rank correlation to the paired variables F20-TP, F50-TP, and Fmax-TP for the join HS and CS HL subgroup. As it can be seen, there exists a positive correlation between audiometric (moderate for F20 and F50 and strong for Fmax) and tinnitus (TP) features.
Table 4: Spearman rank correlation analysis between paired audiometric-tinnitus variables for the subjects with HS and CS HL subtypes.
| Paired variables | r | p |
|---|---|---|
| F20 versus TP | 0.59 | 0.006 |
| F50 versus TP | 0.49 | 0.027 |
| Fmax versus TP | 0.65 | 0.002 |
|
Figure 10: Scatter plot of (a) F20 versus TP, (b) F50 versus TP, and (c) Fmax versus TP, for high-steep high-frequency HL subtype. |
|
Figure 11: Scatter plot of (a) F20 versus TP, (b) F50 versus TP, and (c) Fmax versus TP, for continuously steep HL subtype. |
|
Figure 12: Scatter plot of (a) F20 versus TP, (b) F50 versus TP, and (c) Fmax versus TP, for HL with scotoma subtype. |
|
Figure 13: Scatter plot of (a) F20 versus TP, (b) F50 versus TP, and (c) Fmax versus TP, for high steep high-frequency (blue) and continuously steep (red) HL subtypes. |
4. DISCUSSIONTOP
Our categorization of HL subtypes (Figure 6) is slightly different to that of Nicolas-Puel et al. (2002). In a similar way to them, we consider high-steep high-frequency and flat HL subtypes. However, the other two subtypes, namely, continuously steep and scotoma HL, could differ of the low-frequency HL and dead ear considered by Nicolas-Puel et al. König et al. (2006) and Shekhawat et al. (2014) only considered tinnitus subjects with high-steep high-frequency HL and continuously steep HL, respectively. Serena et al. (2011), on the other hand, distinguished between 0-break “broken-stick” HL (similar to our flat HL), 1-break “broken-stick” (similar to our HS HL), and 2-break “broken-stick”, without matching to our subgroups. Langguth et al. (2017) defined four HL subgroups; namely (1) normal hearing (0–20 dB HL); (2) mild/moderate HL (25–50 dB HL), representing mostly outer hair cell loss; (3) severe/profound HL (> 50 dB HL), representing outer and inner hair cell damage; and (4) no data available.
The most prevalent HL subtype in our cohort is high-steep high-frequency HL (44 %), followed by flat HL (32 %), continuously steep HL (15 %) and scotoma HL (9 %). If HS and CS would be included in a joint subtype, then the prevalence of both should be 59 %. Regarding the flat HL curves, it might be emphasized that we have measured just up to 8 kHz. Therefore, HL flat up to 8 kHz does not exclude the occurrence of losses above 8 kHz which can potentially trigger tinnitus, mainly at high frequencies (Weisz et al., 2006). High frequency losses (8–16 kHz) have been recently interpreted as an early indication of cochlear synaptopathy in humans (Milloy et al., 2017). Cochlear synaptopathy, also named hidden HL, is the selective loss of synaptic connections between high-threshold and low-spontaneous rate IHCs with the auditory nerve, showing or not threshold elevations, due to the loss of hair cells-auditory nerve synaptic connection (Liberman & Liberman, 2015). The loss of these connections can reach 40–50 % without elevating hearing thresholds. As HL is considered the most likely trigger of tinnitus, this subgroup is further excluded of the correlational analysis between audiometric and tinnitus features.
Concerning the tinnitus lateralization (Figure 7a), surprisingly 43 % of subjects with HS, CS, or ST HL perceived their tinnitus in the left ear, in contrast to subjects perceiving their tinnitus in the right ear, only 8 %. In the cohort of Scheklmann et al. (2012) there was also a bias toward the left ear (72 versus 45 subjects). In the cohort of Shekhawat et al. (2014), on the contrary, 103 participants had the predominant tinnitus towards the right ear, versus 83 subjects with their tinnitus towards the left ear. Therefore, the high bias of the unilateral tinnitus towards the left ear in our cohort might be considered merely casual.
The tinnitus sound of our sample differs also from other databases. In our cohort (Figure 7b), pure tone and ringing have a similar prevalence (39 % and 35 %, respectively), while hissing is less prevalent (26 %). In the Tinnitus Archive of the Oregon Health State University (OHSU), tonal, ringing, and hissing are also the more frequent tinnitus sounds, but with a different prevalence. When more than one predominant sound is reported, ringing is prominently the more prevalent sound.
Notice also that, when flat HL subtype is excluded, sensorineural HL is the most possible origin of tinnitus referred by participants (39 %), followed by noise (30 %) (Figure 7c).
Our results concur with those of Shekhawat et al. (2014) regarding the audiometric feature which correlates better with the tinnitus pitch (Figures 8 and 9), at least for the HS and CS HL subtypes. In effect, our results confirm that F20 (the frequency at which audiometry crosses HL = 20 dB) and Fmax (the frequency at which HL is maximum) under- and overestimate, respectively, the tinnitus pitch. Spearman rank correlation analysis confirms that tinnitus pitch increases monotonically with F20, F50, and Fmax (Figures 10–12 and Table 4), but that it is better correlated with F50. This corroborates the hypothesis that tinnitus pitch is matched to the frequency at which hearing loss reaches 50 dB HL (Shekhawat et al., 2014).
5. CONCLUSIONSTOP
This article contains a preliminary correlational study between audiometric characteristics and tinnitus features in a sample of 34 human subjects with tinnitus. Following the current trend of tinnitus heterogeneity, subjects are categorized first in four subgroups, taking into account the shape of HL curves, namely flat HL, high-steep high-frequency HL, continuously steep HL, and HL with scotoma. The more prevalent subgroup was the high-steep high-frequency HL. Excluding the flat HL subgroup, three audiometric features are calculated from the HL curves: F20, the frequency at which HL = 20 dB, F50, the frequency at which HL = 50 dB, and Fmax, the frequency at which HL is maximum.
Tinnitus characteristics include tinnitus laterality (left ear, right ear, bilateral, or central), tinnitus sound (tonal, ringing, or hissing), tinnitus etiology, THI, VAS, and tinnitus pitch (TP). Tinnitus laterality, sound, and etiology were evaluated by the responses of participants to an interview. TP was assessed by matching the tinnitus of participants to a band-filtered noise generated by a GUI. Depending on the bandwidth of the filter, tones (very narrow), ringing (narrow), or hissing (wide) sounds were generated.
Correlational studies included paired audiometric-tinnitus variables analysis. Spearman rank correlation analysis confirmed that TP increases monotonically with the three audiometric variables. Mean variables, TP-F20, TP-F50, and TP-Fmax were also analyzed. Results demonstrated that F20 understimates TP, Fmax overstimates TP, and F50 is the best estimator of TP. Therefore, our results confirm that F50, the frequency at which IHC damage begins, is likely the best predictor of the associated tinnitus pitch.
ACKNOWLEDGMENTSTOP
The enthusiastic participation of volunteers in this study is kindly recognized. Financial support was provided by CSIC through grant 201750E037.