Different bimodal neuromodulation settings reduce tinnitus symptoms in a large randomized trial

Abstract

More than 10% of the population suffers from tinnitus, which is a phantom auditory condition that is coded within the brain. A new neuromodulation approach to treat tinnitus has emerged that combines sound with electrical stimulation of somatosensory pathways, supported by multiple animal studies demonstrating that bimodal stimulation can elicit extensive neural plasticity within the auditory brain. More recently, in a large-scale clinical trial, bimodal neuromodulation combining sound and tongue stimulation drove significant reductions in tinnitus symptom severity during the first 6 weeks of treatment, followed by diminishing improvements during the second 6 weeks of treatment. The primary objective of the large-scale randomized and double-blinded study presented in this paper was to determine if background wideband noise as used in the previous clinical trial was necessary for bimodal treatment efficacy. An additional objective was to determine if adjusting the parameter settings after 6 weeks of treatment could overcome treatment habituation effects observed in the previous study. The primary endpoint at 6-weeks involved within-arm and between-arm comparisons for two treatment arms with different bimodal neuromodulation settings based on two widely used and validated outcome instruments, Tinnitus Handicap Inventory and Tinnitus Functional Index. Both treatment arms exhibited a statistically significant reduction in tinnitus symptoms during the first 6-weeks, which was further reduced significantly during the second 6-weeks by changing the parameter settings (Cohen's d effect size for full treatment period per arm and outcome measure ranged from - 0.7 to - 1.4). There were no significant differences between arms, in which tongue stimulation combined with only pure tones and without background wideband noise was sufficient to reduce tinnitus symptoms. These therapeutic effects were sustained up to 12 months after the treatment ended. The study included two additional exploratory arms, including one arm that presented only sound stimuli during the first 6 weeks of treatment and bimodal stimulation in the second 6 weeks of treatment. This arm revealed the criticality of combining tongue stimulation with sound for treatment efficacy. Overall, there were no treatment-related serious adverse events and a high compliance rate (83.8%) with 70.3% of participants indicating benefit. The discovery that adjusting stimulation parameters overcomes previously observed treatment habituation can be used to drive greater therapeutic effects and opens up new opportunities for optimizing stimuli and enhancing clinical outcomes for tinnitus patients with bimodal neuromodulation.

Figure 1

Timeline of different parameter settings for tinnitus treatment in the TENT-A2 study. (a) The Lenire bimodal neuromodulation device developed by Neuromod Devices is shown (CE-marked Class IIa; previously referred to as MBT in the study). Sound stimulation is delivered through wireless headphones and electrical stimulation is presented to the anterior-dorsal surface of the tongue via a 32-site electrode array. Both stimulation approaches are controlled using a battery-powered controller. Bimodal stimulation activates auditory and somatosensory pathways, as well as converging centers in auditory, limbic and attentional brain regions implicated in tinnitus. (b) Timeline of four treatment arms with different stimulation parameters over time (see Supplementary Table S1 for detailed description). Arm 1 and arm 2 are included for the primary endpoint analyses. Arm 3 and arm 4 are included for additional analyses. Evaluations of device safety and efficacy were conducted at interim and final visits with post-treatment follow-ups planned at weeks 18, 38 and 64 from enrollment.

Figure 2

Participant flow diagram. For the primary endpoint analyses, within-arm comparisons from start of treatment to the 6-week assessment for THI and TFI were performed with per-protocol analysis (PPA) for those who were compliant to treatment (at least 18 h of device usage over 6 weeks of treatment period). Only arm 1 was included for the within-arm primary endpoint analyses whereas the other arms were included as additional analyses. Between-arm comparison was performed only between arm 1 and arm 2 for the primary endpoint analyses for changes in THI scores from start of treatment to the 6-week assessment using intention-to-treat (ITT) with imputation for missing values described in the Methods. Additional exploratory analyses for THI and TFI were performed at other time points during treatment and up to week 64 (12 months post-treatment) that are presented in this paper, with compliance and participant number information provided where appropriate for those analyses. Follow-up (FU) visits are depicted in Fig. 1.

Figure 3

Within-arm and between-arm analyses for primary endpoints at 6-weeks. The mean differences in THI and TFI scores across participants for arm 1 and arm 2 from baseline to 6-week assessment for the within-arm cases, or the mean differences between arm 1 and arm 2 for the change in scores from enrollment to 6-week assessment for the between-arm cases are plotted with the ± 95% CI. Within-arm analyses were based on per-protocol analysis that included treatment-compliant participants (≥ 18 h treatment at 6-weeks) with baseline and 6-week scores. The baseline score corresponds to the average of the scores at screening and enrollment visits. Within-arm analyses for arm 1 and arm 2 showed a highly statistically significant reduction in THI and TFI scores (i.e., improvement in tinnitus symptom severity) based on paired two-tailed t-tests (P < 0.0001; 95% CI does not overlap the vertical line at zero). Between-arm analyses were conducted using an intention-to-treat analysis with changes in THI or TFI scores from the enrollment to 6-weeks assessment and tested with multiple regression using enrollment score as a covariate. Missing data were imputed using the Markov chain Monte Carlo multiple imputation method (further details are provided in the Methods), which leads to n values that match the enrolled numbers for each arm. There was no significant difference between arm 1 and arm 2 (P > 0.05; 95% CI crosses vertical line at zero). Note that primary endpoint analyses included within-arm changes for THI and TFI for arm 1 and between-arm differences for arm 1 and arm 2 only for THI, but that the other within-arm and between-arm comparisons are presented in this figure for completeness. Parameter settings (see Supplementary Table S1 for details): arm 1—PS1 (synchronized bimodal stimulation, includes pure tones and background noise stimuli); arm 2—PS6 (includes only pure tones with no background noise stimuli and with varied interstimulus delays in the range of 700–800 ms).

Figure 4

Long-term clinical efficacy of bimodal neuromodulation. Change in THI score (a) or TFI score (b) from baseline to the different time points are plotted up to the 12-month post-treatment visit (Week 64) for participants who were treatment-compliant (≥ 18 h treatment at interim and ≥ 36 h treatment at final) and attended all visits. Mean change values and 95% CIs are plotted for each arm. All data points with CIs are substantially below the zero line, supporting that bimodal neuromodulation with sound and tongue stimulation achieves significant reduction in tinnitus symptoms that is sustained for 12 months after treatment ended. For clearer visualization, data points and error bars were jittered in time.

Figure 5

Additional reduction in tinnitus symptoms with adjustment of parameter settings at the midway point during treatment. Absolute THI or TFI scores (distribution of values as violin plots) are presented at different time points during the treatment period at Week 0 (baseline), Week 6 (interim), Week 12 (final) and Week 64 (12-month post-treatment) for pooled data across arm 1 and arm 2 for THI (a) and TFI (b). Data is included for participants who were treatment-compliant (≥ 18 h treatment at interim and ≥ 36 h treatment at final) and correspond to the same data shown in Fig. 4. Different stimulus combinations were implemented in the first versus second 6-week period. Asterisks correspond to significant reductions in THI or TFI scores based on a Wilcoxon signed-rank test (P < 0.05). Significant P-values accounting for multiple comparisons based on the Bonferroni correction are labeled with asterisks in order from left to right in each plot: (a) P < 0.00001, P < 0.00001; (b) P < 0.00001, P = 0.0047.

Figure 6

Within-arm analyses for the four different arms at 6-weeks and at 12-weeks. (a) The mean difference in THI or TFI score across participants for each arm from baseline to 6-week assessment are plotted with the ± 95% CI. Within-arm analyses were based on a per-protocol analysis that included treatment-compliant participants (≥ 18 h treatment at 6-weeks) with baseline and 6-week scores (note that the baseline score corresponds to the average of the scores at screening and enrollment visits). Within-arm analyses for all arms, except for arm 3 for THI and arm 4 for TFI, showed a statistically significant reduction in THI or TFI score (i.e., improvement in tinnitus symptom severity) based on paired two-tailed t-tests (P < 0.05; 95% CI does not overlap the vertical line at zero). Cohen’s d effect size and 95% CI are also listed for each arm for THI and TFI. Parameter settings (see Supplementary Table S1 for details): arm 1—PS1 (synchronized bimodal stimulation, includes pure tones and background noise stimuli); arm 2—PS6 (includes only pure tones with no background noise stimuli and with varied interstimulus delays in the range of 700–800 ms); arm 3—PS7 (same as PS6 except four tones presented simultaneously over a larger frequency range); arm 4—PS9 (same sound stimuli as in PS6 but no tongue stimulation). (b) Similar to (a) except data is presented from baseline to 12-week assessment for treatment-compliant participants (≥ 36 h treatment at 12-weeks). Parameter settings (see Supplementary Table S1 for details): arm 1—PS1 changed to PS4 (includes pure tones and background noise stimuli and with varied interstimulus delays in the range of 0–30 ms); arm 2—PS6 changed to PS10 (includes only wideband noise stimuli without pure tones and with varied interstimulus delays in the range of 0–30 ms); arm 3—PS7 changed to PS4 (includes pure tones and background noise stimuli and with varied interstimulus delays in the range of 0–30 ms); arm 4—PS9 changed to PS6 (same sound stimuli as PS9 with addition of tongue stimulation).

Figure 7

Changes in tinnitus symptom severity for arm 4 between interim and final visits after adding tongue stimulation to sound-only stimulation. Mean change in THI or TFI and 95% CIs are plotted between 6-week (interim) and 12-week (final) visits for all participants who were compliant to treatment (defined as ≥ 18 h at interim and ≥ 36 h at final). There is a significant improvement in tinnitus symptom severity after adding tongue stimulation (PS6) to sound-only stimulation (PS9), supporting the criticality of combining tongue stimulation with sound to drive additional therapeutic benefit for tinnitus. Post-hoc statistical analysis was based on a paired two-tailed t-test (P < 0.05; 95% CI does not overlap the vertical line at zero; P = 0.0039 for THI and P = 0.0180 for TFI). Cohen’s d effect size and 95% CI are also listed for THI and TFI on the right side of the figure. Note that n is equal to 13 instead of 14 as in Fig. 6 for arm 4 because compliance criteria of both 18 h and 36 h (rather than just one of them) must be satisfied by each participant to be included in the analysis.

Figure 8

Long-term clinical efficacy of bimodal neuromodulation. (a,b) Scatter plots of THI or TFI scores are shown that include each treatment-compliant individual at baseline versus end of treatment (12-week final visit) for all arms pooled together to represent bimodal neuromodulation in general. (c,d) Scatter plots of THI or TFI scores are shown that include each treatment-compliant individual at baseline versus 12-month post-treatment visit for all arms pooled together. All individuals who completed at least the assessments displayed in each figure were included to maximize the total number of subjects plotted in each figure, leading to a large sample size for each plot. Compliance to treatment is defined as ≥ 36 h treatment at final. Data points are jittered for visibility.

Figure 9

Satisfaction rates in using treatment device. Two questions relating to the participant’s satisfaction (a) or acceptability (b) of the treatment device were asked at the final visit when the treatment ended, in which the percentage of YES or NO responses are shown. The “other” category refers to cases where the participant could not answer YES or NO or did not feel comfortable in committing to an answer.