Functional biomarkers that distinguish between tinnitus with and without hyperacusis

To the Editor: We recently observed that tinnitus is associated with reduced auditory input that fails to increase neural gain due to diminished stimulus-evoked responses.1–3 This was in contrast to views that suggested a homeostatic increase in neural gain to generate central hyper-excitability leading to tinnitus.4 A curative therapy for tinnitus currently does not exist. Its progress is mostly impeded by the existing controversial views about the neural correlate of tinnitus that, depending on predictions, either would require the suppression or the enhancement of brain activity. We hypothesized that different neural correlates of tinnitus, whether with or without the co-occurrence of hyperacusis, contributed to this dilemma. To test this hypothesis, we recruited 43 controls and 50 audiologically examined tinnitus patients with andwithout a co-occurrence of hyperacusis (Tables S1 and S2) and performed brainstemaudiometry (ABR) and functional imaging of brain activity (fMRI). Among the group of 50 tinnitus patients, 20 could be identified with the co-occurrence of hyperacusis (T+H group) from the HKI hyperacusis questionnaire (Figure 1A).5 The overall score of the Goebel and Hiller Score (G-H-S) tinnitus questionnaire6 was significantly higher for the T+H group than the tinnitus-only patients (T group) (Figure 1B, p< .001***) for nearly all subscores (Figure S1, p < .002***). In the T+H group (Figure 1D) but not the T group (Figure 1C), auditory perceptional difficulty became worse for patients with self-rated tinnitus loudness ≤15 dB HL (Figure 1E, Figure S2). The T and T+H group differences in annoyance and distress were not linked to differences in hearing sensitivity (Figure S3A– C, p > .5), since pure tone audiometry (PTA) thresholds (Supporting Material1) were not different between groups. In contrast, supra-threshold ABR by brainstemevoked response audiometry (BERA)1,2 revealed group differences: In T group, significantly reduced ABR wave V amplitude together with significantly prolonged interpeak latency (IPL) I–V (Figure 1F,H,J; Table S3) and reduced

ABR wave V/I ratios ( Figure 1I), were found. In contrast, T+H group showed a significantly higher ABR wave III and wave V amplitude at 75 dB compared to controls (Figure 1G,H; Table S3) with no difference in ABR wave V/I ratio ( Figure 1I). Questioning if these group differences were reflected in BOLD fMRI responses, stimulus-induced BOLD fMRI signals were recorded from anatomically predefined ROIs in ascending auditory regions (Table S4) in response to binaurally exposed (i) rock music, (ii) LF-chirp, (iii) HF-chirp, and (iv) BB-chirp stimuli ( Figure 2). A significant reduction in BOLD fMRI signals in lower auditory brainstem regions (SOC, partly CN) (Figure 2A-C; music, LF stimuli) revealed as a characteristic feature of tinnitus in both groups.
From the MGB upwards, BOLD fMRI signals between groups differed, remaining reduced in the T group in the MGB ( 7 such as the mammillary body (Mam. Body), the dorsal posterior insula (DpIns), and the postcentral gyrus regions with the parietal operculum (PO 1 , PO 2 ) responded with reduced BOLD fMRI signals to music stimuli in the T group, but not in the T+H group (Figure 2J,L; music), suggesting more response activity in pain regions to sound in the T+H group. Interested if evoked BOLD fMRI responses were related to BOLD signals at rest, as hypothesized, 8 we strikingly observed that the number of correlations of BOLD signals at rest (r-fcMRI) (  regions such as the CN, SOC, IC ( Figure 3D), between MGB and the anterior AC-I regions BA41 and BA42 (Figure 3E), between the AC-I and regions controlling emotional distress, particularly the amygdala ( Figure 3G), and between the AC-I and attention-controlling regions such as BA45 and BA46 ( Figure 3H) was significantly lower in the T group compared to the T+H group. Few regions with negative correlations that were lower in the T group and T+H group may need further future specification.
To summarize, as a most characteristic functional biomarker of the T group, the present study identified (i) delayed and reduced ABR wave V; (ii) reduced evoked BOLD fMRI responses in the MGB, AC-I, and regions of sound identification as the BA13P and hippocampus, particularly specific in response to HF-chirp stimuli; and (iii) reduced number of positive connectivities between subcortical and cortical auditory regions (Figure 4, red). As a characteristic functional biomarker for tinnitus with a co-occurrence of hyperacusis (i) enhanced ABR wave III and ABR wave V for high sound intensities; (ii) elevated evoked BOLD fMRI responses in the MGB, AC-I, BA13P, and hippocampus particularly for LF-chirp stimuli; and (iii) greater number of positive connectivities between subcortical and cortical auditory regions compared to the T group ( Figure 4, blue) were found. Group differences were independent of G-H-S group differences (Supporting Material). We conclude that the overall reduced and delayed auditory-specific responsiveness in the T group is best corroborated by previous assumption of a loss of fast (high-SR) auditory fiber processing in tinnitus frequency channels leading to re-emergence of hyperexcitability through loss of tonic parvalbumin interneuron in deprived regions. 9 This would lead to diminution of memory-linked contrast amplification and elevated noise, and as a result would promote further alertness and attention to the phantom noise, as reviewed. 9 With the co-occurrence of hyperacusis, a more widespread signal amplification process appears to proceed through overactive thalamo-cortical activity that may trigger an excitation spread to limbic and pain regions, and results in overattention to increased loudness at all sound frequencies, as was also previously hypothesized. 10 The findings may eventually lead to new differential clinical diagnosis of tinnitus, a prerequisite for achieving a successful, personalized curative therapy for tinnitus with and without hyperacusis, when regarding suggestions for altered strategies to find treatment predictors. 9

A C K N O W L E D G M E N T S
This work was supported by the Deutsche Forschungsgemeinschaft: KN 316/13-1, RU 713/6-1, KL 1093/12-1. We thank Hyun-Soon Geisler, Karin Rohbock, and Iris Köpschall for excellent technical assistance. English language services were provided by stels-ol.de.

C O N F L I C T O F I N T E R E S T
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. F I G U R E 1 Hyperacusis Questionnaire total score (HKI) and Tinnitus Questionnaire total score (G-H-S). The figures represent mean ± SD between control (n = 29, gray), T group (n = 30, red), and T+H group (n = 20, blue) for (A) HKI score, and (B) Tinnitus Questionnaire total score (G-H-S) (see Supporting Material Methods). Mann-Whitney U-test was used to calculate the group differences. Two-tailed Spearman correlation of tinnitus loudness with the Tinnitus Questionnaire subscore "auditory perceptual difficulties" for individual participants: (C) T group (red), and (D) T+H group (blue). The box plot (E) shows median, range (whiskers), and quartiles (box) of the Tinnitus Questionnaire Auditory Perceptual Difficulty (Aud. Perc. Diff. score) for patients with self-rated tinnitus loudness intensity ≤15 dB HL. In the T+H group (D) but not the T group (C), the auditory perceptional difficulty correlated quite well with the self-rated tinnitus intensity (for the correlations of all subscores see Figure S2), due to an auditory perceptional difficulty that in the T+H group became particularly annoying for self-rated tinnitus when loudness was ≤15 dB HL (E). Mann-Whitney U-test was used to calculate the group differences between T group (n = 16, red) and T+H group (n = 10, blue). (F) ABR wave amplitude changes and latency shifts. Averaged ABR wave I, III, V, and VI amplitudes at 75 dB nHL (upper panels), and latency as function of stimulus level (lower panels). Lines connect data points of the same experimental group. Repeated measures two-way ANOVA was used to test for group differences within single ABR waves and Holm-Sidak's multiple comparison test for pairwise differences between control (n = 43, gray) and T group (n = 30, red), (G) between control and T+H group (n = 20, blue), and (H) between T group and T+H group. For wave I, III, and VI, distinct responses were limited to 65 and 75 dB nHL stimuli. The box plots (I) show median, quartiles, and range of ABR wave ratio V/I at 75 dB nHL for control (n = 43, gray), T group (n = 30, red), and T+H group (n = 20, blue). Interpeak latency (IPL) between wave I and wave V peak is shown in the bar graphs (J) as mean ± SD. Details of statistical results are given in Table S3. The overall differences that emerged from the Tinnitus Questionnaire (G-H-S) for the T and the T+H groups were not related to gender (Table S3), left-or right-handedness of the participants, tinnitus laterality, tinnitus intensity, tinnitus frequency, nor with the age of the participants (Table S1). The reduced ABR wave V responses in the T group were reflected in reduced ABR wave V/I ratio ( Figure 1I), confirming assumptions of reduced central neural gain with elevated response variability in tinnitus. 9 *p < .05, **p < .01, ***p < .001, ****p < .0001 from Holm-Sidak's multiple comparison test; (*) statistical test close to statistical significance (p < .066 in Holm-Sidak's multiple comparison test). ABR, auditory brainstem response; Aud. Perc. Diff., auditory perceptual difficulties; dB, decibel; G-H-S, Goebel and Hiller Score; HL, hearing level; nHL, normalized hearing level; RM, repeated measures two-way ANOVA; SD, standard deviation BOLD fMRI differences for sub-cortical regions A

E T H I C S A P P R O VA L A N D C O N S E N T T O PA R T I C I PAT E
The ethics committee of Tübingen University (Faculty of Medicine) and University Hospital Tübingen (ethical approval number 264-2016BO1) approved this study. Trial registration: German Clinical Trials Register DRKS0006332. Written informed consent was obtained from all participants at their first visit.

C O N S E N T F O R P U B L I C AT I O N
All the authors have read the manuscript and approved its submission to Clinical and Translational Medicine.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from Marlies Knipper upon reasonable request.