Vagus Nerve Stimulation Paired with Tones for the Treatment of Tinnitus: A Prospective Randomized Double-blind Controlled Pilot Study in Humans

The aim of the pilot study was to evaluate the effect of Vagus Nerve Stimulation (VNS) paired with sounds in chronic tinnitus patients. All participants were implanted and randomized to a paired VNS (n = 16) or control (n = 14) group. After 6 weeks of home therapy, all participants received paired VNS. The device was used on 96% of days with good compliance. After 6 weeks, the paired VNS group improved on the Tinnitus Handicap Inventory (THI) (p = 0.0012) compared to controls (p = 0.1561). The between-group difference was 10.3% (p = 0.3393). Fifty percent of the participants in the paired VNS group showed clinically meaningful improvements compared to 28% in controls. At one year, 50% of participants had a clinically meaningful response. The therapy had greater benefits for participants with tonal and non-blast induced tinnitus at the end of 6 (24.3% vs. 2%, p = 0.05) and 12 weeks (34% vs. 2%, p = 0.004) compared to controls with 80% and 70% responding at 6 months and 1 year, respectively. Adverse effects were mild and well-tolerated and the therapy had a similar safety profile to VNS for epilepsy. VNS paired with tones may be effective for a subgroup of tinnitus patients and provides impetus for a larger pivotal study.

For the tinnitus pitch match, a paired-comparison procedure was used. First, 0.5 kHz and then a 4-kHz pulsed tone were presented and the participant was asked "Of the two tones presented to you, which has a pitch that is most like the most prominent pitch of your tinnitus?" If the participant said the higher tone, the next two tones compared would be 2 and 8 kHz. If the participant said the lower tone, the next two tones compared would be 0.25 and 1 kHz. Once the frequency region was established, half octave steps were used. When the participant reported the same tone as being similar twice, this value was recorded as the most prominent pitch. This sequence was repeated 3 times and the average recorded.
Loudness Severity: For the loudness severity, participants were asked to rate the loudness of their tinnitus on a scale from 0-100; 0 meaning no tinnitus and 100 indicating loudest tinnitus that they can imagine. This estimation was performed for both ears (or documented as only occurring in one ear).
Loudness Matching: For loudness matching, the estimation was performed for both ears (or documented as only occurring in one ear). Starting at 10 dB above the participants hearing threshold, they were asked to rate the loudness of their tinnitus compared to a 1000-Hz pulsed tone using an ascending method of limits. The tone was presented. Then, the participant was told "Tell me if your tinnitus is louder or softer than the loudness of my tone." If the participant said louder, the level was increased by 2 dB, if they said softer, the level was reduced by 2 dB. The test was stopped when the tinnitus loudness had been bracketed and the loudness recorded as the last test level (dB HL). This was repeated 5 times during each measurement session and averaged to get a final value. The same 2-dB step size procedure was used to determine a 1000-Hz hearing threshold. This test was performed in the ear with tinnitus, or in the one with the loudest presentation if in both ears; if equal in both ears, the right ear was used.
Minimum Masking Level (MML): Measurements were made of the amount of noise required to mask a participant's tinnitus (Minimum Masking Level (MML)). With a pulsed (duration less than 5 seconds) broadband noise, the threshold and minimum masking level (MML) required to mask the tinnitus was measured with an ascending method of limits. First, the threshold for the noise was measured as described above with a 2 dB step size. For minimum masking level, the noise is presented at threshold and the participant asked "Tell me if you can hear your tinnitus above my noise." If the participant said yes, the noise was increased by 2 dB. The test was stopped when the tinnitus was masked. The last test level was the MML. This procedure was repeated 7 times and each participant's mean MML was used for comparisons. This was also repeated for the participant's other ear, and then documented in both ears for white noise.

Outcome measures
The THI [1] is a 25-item self-administered questionnaire that aims to quantify the impact of tinnitus on quality of life by measuring its effects on everyday function. Respondents are asked to answer the questions with 'Yes' (4 points) 'Sometimes' (2 points) or 'No' (0 points). A higher THI score (maximum 100) is indicative of a greater tinnitus handicap. The THI is a widely used, well validated and widely used outcome measure [2]. This study used THI as the primary outcome measure.
The THQ is a scale comprised of 27 items and is a well-established measure for the assessment of a broad spectrum of tinnitus-related psychological complaints [3]. Participants were asked to indicate agreement on a scale from 0 (if you strongly disagree) up to 100 (you strongly agree) [4][5].
The Tinnitus Functional Index (TFI) [6] is comprised of 25 items according to how they felt over the past week. Each item is scored on an 11-point scale with descriptors at either end of the scale. The sum of all scores is divided by 2.5 to give a global score out of 100. Higher scores reflect greater impact (worse) on daily functioning.

Adverse Event Reporting
All clinical signs and symptoms present at the time of enrollment and/or during the baseline period (prior to implantation) were documented. At the time of implantation and at each subsequent study visit, adverse events were recorded based on information volunteered by the participant and observations made by the health care staff. All events were recorded even if they were not considered to be related to study treatment, implantation, or the implantation system.

Sample size determination
This Pilot study was not powered to demonstrate statistically significant between-group results; it was based on typical sizes for IDE device pilot studies and was agreed upon with FDA. Typical study sizes for between-group differences of neurostimulation devices would be between 100 and 200 subjects. IDE pilot studies are typically between 10 and 50 subjects, and often do not have a separate control group (IDE pilot studies are often not randomized, parallel studies).
With respect to safety, a sample of 30 allowed adequate power to detect the incidence of rare safety and device events. A sample of 30 patients yields 95% probability that the study will reveal at least one occurrence of all events or complications that occur in patients at a rate of 9.5% or greater. In addition, implantation and follow-up of 30 patients for 6 weeks will yield 1,260 days of exposure. In this case, the threshold for detection decreases to a very unlikely event, one that occurs in only 0.2375% of days. In other words, if an event has a probability of 0.002375 or 1 event in 422 days of exposure we have a 95% probability of seeing this event in a sample of 30 patients with 1,260 days of exposure. As patients continue into the Long-Term portion of the study, the ability to detect rare events increases as the exposure increases.

Randomization Allocation Sequence
The CRO statistician created the randomization. Each site had one individual who was the "Programmer" that set the device to its correct settings. This was the only person at the site who was unblinded. The Programmer called a number at the Contract Research Organization (CRO) to receive the assignment as each subject to be randomized. Once proper identification information was given for both the Programmer and the Patient (Patient code and implant date), the Programmer was given the appropriate randomization assignment. After the Programmer set the device to the appropriate group settings, they emailed a confirmation back to the CRO for final verification. The randomization code was not given to any other individual until the blind was broken at the end of the study.

Description of the similarity of interventions
The interventions were similar. All subjects had the same surgical procedure and device implanted. Both groups received vagus nerve stimulation and tones via headphones. The difference was in the timing and length of the stimulation and tones.   (4). Recovery was within one week for intubation, within 12 weeks for two due to surgery, and is ongoing in two cases (one year assessment has not yet occurred but is planned). 2 -Neck lump was reported both prior to surgery and after surgery in one subject, but was noted as possibly related to surgery.

Recruitment started in February and
Note: One lead fracture did occur and resulted in a 3-month delay in implants (reported to both the FDA and the study DSMB).