Evaluation of a Totally Implantable Cochlear Implant in Adults With Bilateral Sensorineural Hearing Loss: A Feasibility Study

Abstract

Objectives:

To investigate the feasibility of a newly designed investigational totally implantable cochlear implant, the Cochlear^™ totally implantable cochlear implant (TICI) Research System, which enables hearing without the use of an external sound processor.

Design:

A prospective, single-treatment arm, repeated measures feasibility study to investigate clinical benefit and safety. Ten adult participants with bilateral moderately severe-to-profound sensorineural hearing loss were implanted unilaterally with the TICI Research System. Co-primary endpoints evaluated external hearing (EH) mode (i.e., sound processor on) and invisible hearing (IH) mode (i.e., sound processor off) performance using words in quiet (60 dB SPL) and sentences in babble noise (60 dB SPL, +10 dB SNR) from the preoperative baseline to 6-month post-activation. The same speech perception tests were re-measured at 12-month post-activation. At 9-month post-activation, performance using phonemes in quiet was measured at a wider range of speech input levels (45, 55, 65, and 75 dB SPL), such as would be encountered in real-life settings. Patient-reported outcomes were evaluated using the Speech, Spatial, and Qualities of Hearing scale, Global Health Utilities Index mark 2/3, and Patient Satisfaction Survey. Device data logging captured time of use statistics in each hearing mode. Safety outcomes (adverse events and device deficiencies) were reported for the study duration (up to 2 years post-implantation).

Results:

Speech performance with the TICI Research System for both EH and IH modes at 6 months was significantly improved over baseline (n = 9). For words in quiet, there was a mean improvement of 38.5% (95% confidence interval [CI], 24.6 to 52.5) and 25.2% (95% CI, 13.7 to 36.7) for EH and IH, respectively. For sentences in noise, there was a mean improvement of 74.8% (95% CI, 53.4 to 96.2) and 64.5% (95% CI, 42.9 to 86.0) for EH and IH, respectively. For words in quiet, performance was found to be significantly better for EH compared with IH, with marginal mean differences of 13.3% (p < 0.001) and 12.0% (p = 0.004) at 6 and 12 months, respectively. Likewise, for phonemes in quiet at 9 months, EH was significantly better than IH with differences of 33.9% (p < 0.001), 21.7% (p < 0.001), and 9.8% (p = 0.002) at 45, 55 and 65 dB SPL, respectively, with no significant difference at 75 dB SPL (p = 0.384). For sentences in noise, at 6 months, EH had significantly better performance than IH with a marginal mean difference of 10.3% (p = 0.021), with no difference shown at 12 months (p = 1.000). Speech, Spatial, and Qualities of Hearing scale showed a significant mean improvement of 1.84 (95% CI, 1.00 to 2.65) compared with baseline. Health Utilities Index mark 2/3 showed variability in scores across the participant cohort, and Patient Satisfaction Survey ratings showed that overall, participants were satisfied with the device. Data logging showed participants' use of IH varied from 22% to 88% of total listening time (median 49%). Device failures occurred in 2 participants due to a microphone-related fluid ingress issue. Both participants were reimplanted with conventional Cochlear Nucleus ® devices. No other safety issues were detected.

Conclusions:

This study demonstrated the feasibility of the TICI Research System to provide patient benefit in both IH and EH modes. No fundamental problem with the use of a transcutaneous microphone was detected. These findings support further development of the TICI design and approach.

INTRODUCTION

The cochlear implant (CI) has been demonstrated to be an effective intervention for restoring functional hearing in adults and children with severe-profound sensorineural hearing loss (Sharma et al. 2020; Tang et al. 2024). Multiple studies have reported that most recipients achieve significant improvements in speech performance and hearing-related quality of life outcomes when compared with preoperative levels (McRackan et al. 2018; Boisvert et al. 2020; Cuda et al. 2024). Nevertheless, of those who could be considered as eligible candidates for a CI, only a minority receive an implant (Nassiri et al. 2022; Zhan et al. 2024). While the barriers to cochlear implantation are multifaceted, a contributing factor could be related to the design of the CI itself.

Current commercial CI designs require the user to wear an external sound processor to power and control the internal implant. Limitations inherent in an external prostheses include: (i) discomfort and the inability to use the device during vigorous physical activities, in water or while sleeping; (ii) potential for traumatic damage to the external device; (iii) need for ongoing maintenance; and (iv) cosmetic appearance, which results in an inability of the user to control the visibility of their deafness and disclosure of their hearing loss to others (Lo et al. 2024) and can be considered a barrier to uptake of CIs in adults (Bierbaum et al. 2020; Ebrahimi-Madiseh et al. 2020).

To address these potential limitations, the concept of a CI that allows “Invisible Hearing” without the need for use of an external sound processor was first investigated in adult patients using a prototype design of a totally implantable cochlear implant (TICI) known as TIKI (Briggs et al. 2008). TIKI offered two modes of hearing: an external hearing mode (EH) in which the CI is used with an external sound processor (Esprit 3G, Cochlear); and an invisible hearing mode (IH), avoiding the need for an external sound processor through incorporation of an internal rechargeable Li-ion battery, an internal microphone mounted on the main implant package, and sound processing capabilities contained within the implant body electronics. The TIKI system also included a separate external charger for the Li-ion battery, while the external sound processor had the functionality to recharge the internal battery when in use.

In that first-time-in-human pilot study, 3 adult participants were implanted with the TIKI device (Briggs et al. 2008). Surgical placement of the device was straightforward in all cases, wound healing was uneventful despite the larger implant package, and there were no surgical or postoperative complications. Speech perception outcomes improved in all participants relative to preoperative performance, with benefits comparable to those for conventional CIs as reported in the literature at the time of TIKI development. Speech scores were higher in EH mode compared with IH mode, although all 3 participants reported benefit with the IH mode. Participants reported using IH mode in specific situations, such as when alone at home, during physical activities, or at night. Body noise interference, due to internal microphone sensitivity, was found to affect speech understanding and usability of IH mode. To date, all 3 subjects have continued to use their TIKI and the IH mode. No problems with recharging of the internal battery or other device functionality have been reported. In summary, the Briggs et al. (2008) TIKI study served as a proof of concept, demonstrating overall surgical safety and the feasibility of using an implantable rechargeable battery. On the basis of these learnings from TIKI, the design of the totally implantable device was modified, and this led to the development of the investigational (non-approved) Cochlear^™ TICI Research System (Cochlear Ltd, Sydney, Australia).

The aim of the present study was to investigate the efficacy, safety, and patient-reported outcomes of the TICI Research System in both the EH and IH modes in adult recipients with bilateral moderately-severe to profound sensorineural hearing loss.

MATERIALS AND METHODS

Investigational Device

The internal (implanted) package of the TICI Research System is shown in Figure 1 and consists of an electronics assembly enclosed in a hermetically sealed titanium case and a receiver coil, over-moulded by silicone elastomer. The electronics assembly is coupled to the Cochlear Contour Advance ® electrode array and a subcutaneous pendant microphone.

Fig. 1.

Cochlear TICI Research System consisting of a subcutaneous microphone and a Contour Advance electrode array.

The system utilizes a Cochlear Nucleus ® 6 (N6) behind-the-ear sound processor (with modified firmware). In EH mode, the system operates as a conventional CI. Acoustic signals are obtained from the external sound processor microphone, signal processing is performed within the sound processor, and stimulation data and power are supplied via the transmitting coil to the implanted electronics package. In IH mode, no external sound processor is worn, acoustic signals are obtained from the implanted subcutaneous microphone, and signal processing and stimulation are carried out by the implanted electronics package operating on an internal battery power supply. To facilitate ease of switching between hearing modes, EH mode is automatically activated whenever the N6 sound processor is worn and the RF coil is connected. The system automatically reverts to IH mode when the N6 sound processor RF coil is removed.

The subcutaneous microphone is a modified version of that used in the Cochlear Carina ® active acoustic implant (with additional shielding and partly overcoated with silicone elastomer). It is implanted directly posterior to the external auditory canal and is active only while the IH mode is used. The pendant microphone assembly incorporates two conventional electret microphone transducers. The transducer on the upper (skin) surface is mounted within a titanium diaphragm-covered cavity. It is designed to be mostly sensitive to air-conducted sound, whereas the second transducer is mounted within a closed cavity on the bone side of the package, designed to receive mostly bone-conducted vibration. The two-transducer design allows for signal processing to actively reduce the level of vibration-based bone-conducted signals, thereby addressing a major limitation identified in the earlier prototype TIKI microphone.

Power is provided by an internal Li-ion rechargeable battery integrated into the titanium electronics package. The internal battery is automatically charged by the RF signal delivered by the N6 sound processor when the system is used in EH mode, or by using a dedicated external charger. When fully charged, the internal battery is specified to provide power sufficient for up to 15 h of use in IH mode when new. For more than 90% of patients implanted, the internal battery is estimated to provide sufficient power for 12 h of continuous IH use. After 10 years of operation, this is estimated to reduce to 8 h of IH use. It is therefore anticipated that at some future point in time, battery capacity will be further reduced and IH mode may no longer be supported. In this situation, EH mode will continue to provide equivalent benefits to those provided by the standard commercial CI for the normal lifetime of the device. Magnetic resonance imaging is currently contraindicated for the TICI Research System.

Participants

Adult participants (≥18 years of age) were recruited from those with bilateral moderately-severe to profound post-lingual sensorineural hearing impairment attending the CI clinic, with ≤15 dB difference between masked air conduction and bone conduction hearing thresholds (i.e., air-bone gap). Participants were required to have a prior history of hearing aid (HA) use in the ear to be implanted, or were referred for a 30-day HA trial before their eligibility for enrollment was assessed. Individuals with evidence of severe-to-profound hearing loss before 5 years of age, a preexisting cochlear or bone conduction implant, or a preexisting medical condition that requires serial magnetic resonance imaging scans were excluded from study participation.

Ten participants (male:female = 5:5) were implanted with the TICI Research System between September 2018 and March 2020. Participant demographics and baseline characteristics are summarized in Table 1 and 2 respectively. The mean age at implantation was 60.4 years (range 41 to 73 years), and the mean duration of severe-profound hearing loss in the implanted ear was 15.1 years (range 2 to 58 years). Before implantation, HAs were still in use in the to be implanted and contralateral ear in 8 and 9 participants, respectively.

TABLE 1.

Participant demographics

Item	Ear	Mean (SD)	Median (Min, Max)
Age at study enrolment (yr)	NA	60.4 (12.96)	67 (41.0, 73.0)
Age at diagnosis of severe-to-profound HL (yr)	Implanted	33.5 (15.44)	38.5 (11.0, 55.0)
	Contralateral	33.1 (16.73)	35.5 (11.0, 57.0)
Duration of severe-to-profound HL (yr)	Implanted	15.1 (17.52)	8.5 (2.0, 58)
	Contralateral	9 (6.44)	7.0 (2.0, 22)

HL, hearing loss.

TABLE 2.

Participant baseline characteristics

Hearing History	Implanted Ear	Contralateral Ear
Hearing loss onset
Progressive	8	8
Sudden	2	2
Primary etiology
Genetic	3	3
Meniere’s disease	1	1
Sudden sensorineural	1	1
Otosclerosis	2	2
Trauma	0	1
Unknown	3	2

Figure 2 shows the group mean baseline (pre-implant) air conduction pure-tone audiometric thresholds in the implanted and contralateral ears (n = 10), indicating that the poorer hearing ear was implanted in most cases. The number of participants with non-measurable thresholds at audiometer limits (i.e., either no response or vibrotactile only at maximum audiometer levels) for the ear to be implanted were 250 Hz (n = 5), 500 Hz (n = 2), 1000 Hz (n = 2), 2000 Hz (n = 4) and 4000 Hz (n = 5) and for the contralateral ear; 250 Hz (n = 3), 500 Hz (n = 1), 2000 Hz (n = 1), and 4000 Hz (n = 2). For the mean calculation, non-measurable thresholds were given the value of 125 dB HL.

Fig. 2.

Pre-implant baseline group mean unaided pure-tone audiometric thresholds and SDs for the ear to be implanted (circles) and the contralateral ear (triangles).

Study Design

The study was a prospective, single-treatment arm feasibility study with a repeated measures design. It was conducted at the Royal Victorian Eye & Ear Hospital, Melbourne, Australia in collaboration with the HEARnet Clinical Studies (The University of Melbourne, Parkville, Australia) and was sponsored by Cochlear Ltd (Sydney, Australia). The clinical investigation protocol was reviewed and approved by the Royal Victorian Eye and Ear Hospital Human Research Ethics Committee (Reference number 16/1304H) and was conducted in accordance with the National Statement on Ethical Conduct in Human Research (NH&MRC 2007, updated 2018 and 2023), the ethical principles of the Declaration of Helsinki, and the International Standard ISO 14155 Clinical investigation of Medical Devices for Human Subjects—Good Clinical Practice. Written informed consent was obtained from all participants before enrollment. An Independent Data Monitoring Committee also reviewed safety data at regular intervals during the study.

Surgery

Given the inclusion of the pendant microphone and larger size of the investigational implant as compared with the existing Nucleus Profile^™ cochlear implant, before commencement of this study, cadaver studies were undertaken to assess positioning and the extent of the bony recess necessary both for the receiver stimulator package and the microphone. Results demonstrated that the investigational implant could be safely sited in the standard position for an implant package at 45° postero-superior to the mastoid. The microphone could then be positioned inferior to the receiver stimulator package on the posterior aspect of the mastoid. This allowed the microphone cable to be recessed within the mastoid cavity.

As with the previous prototype TIKI study (Briggs et al. 2008), the usual minimally invasive surgical approach was modified to accommodate the larger device package. Initial device placement and incision planning were carried out using a non-sterile silicone template of the implant and microphone. The usual incision was extended approximately 3 cm superiorly. The scalp flap was elevated anteriorly, and a posteriorly based offset fibro-periosteal flap was then elevated. After routine cortical mastoidectomy and facial recess approach, a subperiosteal pocket was created for the antenna, and a recess was drilled for the implant package, with a channel drilled to the mastoid cavity for the electrode and microphone cables. A further bony recess was then drilled for the microphone, with a metal dummy microphone used to confirm size and fit. The implant package was then positioned, and the Contour electrode inserted using an advanced off-stylet technique via an extended round window cochleostomy. The electrode cable was then coiled in the mastoid. The microphone was then positioned in its recess, the malleable screw fixture arms bent to contact bone and secured with two self-tapping screws. The microphone lead was coiled in the mastoid lateral to the electrode cable. The fibro-periosteal flap covering both the microphone and the implant package, and the skin incision were repaired with interrupted Vicryl sutures.

Programming

The mean time between surgery and the initial activation of the TICI Research System was 15.7 days (range 13 to 22 days). Electrode stimulation parameters were configured to reduce power consumption and remain within the voltage compliance limits dictated by the internal battery voltage. Threshold (T) and Comfort (C) levels were measured using standard clinical procedures, followed by loudness balancing at the C-level. C-levels were then adjusted using live voice to ensure a comfortable volume and acceptable sound quality for the EH program. These T- and C-levels were then transferred to the IH program. Outside the fitting session, participants were able to adjust volume settings in EH mode either using the sound processor or a remote control. There was no user access to volume settings for IH mode.

To address the potential issue that the acoustic sensitivity of the implanted microphone might be impacted by skin loading over its diaphragm, the fitting software included a calibration protocol for measuring and comparing its response with that of the external microphone on the sound processor. This enabled the signal processing to be configured to equalize the responses of the two microphones. Body noise was addressed through the use of a body noise canceler preprocessing algorithm, implemented using the two transducers in the implanted microphone, and an active noise suppression filter, to reduce internal (body) noises. Fitting parameters were adjusted throughout the study with the fine-tuning of T- and C-levels, re-calibration of the implanted microphone, and adjustments to the settings of the body noise canceler.

In EH mode, the modified N6 sound processing included the standard features available in the commercial device: Automatic Sensitivity Control (ASC), Adaptive Dynamic Range Optimization (ADRO), Signal to Noise Ratio Noise Reduction (SNR-NR), Automatic Scene Classifier System (SCAN), and SCAN’s associated directionality settings. IH mode sound processing supported the same features as EH mode except for SCAN/directionality, which was not supported for the subcutaneous microphone input.

Whilst previous studies had shown the implantable microphone to be safe and reliable when used in the Carina middle ear implant (Bruschini et al. 2016), for use in the TICI Research System, a patient safety process was incorporated. A diagnostic test designed to detect any fluid ingress failures to the microphone wire (or other electrical leakage paths), was run automatically each time EH mode was activated. In the event that a measured leakage resistance was lower than the predefined threshold, the implant was designed to lock, thereby eliminating the potential for any unwanted stimulation and/or auditory percepts to be delivered to the user.

Outcome Measures

The primary objective of the study was to investigate the efficacy of the TICI Research System as a feasible treatment in terms of restoring speech perception performance in adult patients with sensorineural hearing loss. The co-primary endpoints for the clinical investigation were mean speech perception performance for words in quiet and sentences in noise for both EH and IH modes at 6-month post-activation in the unilateral listening condition, as compared with the baseline preoperative condition. Baseline speech performance was measured using the participant’s own HA in the ear to be implanted. Speech was presented from the front, and the contralateral ear was plugged. The same speech perception tests were re-measured at 12-month post-activation.

Speech test materials included: two lists (2 × 50 words) of CNC monosyllabic words (Petersen & Lehiste 1962) presented in quiet at 60 dB SPL (female speaker, scored for whole words correct); and one list of 20 sentences from the AuSTIN sentences test (Dawson et al. 2013) presented at 60 dB SPL (female speaker, scored for morphemes correct) with competing four-talker babble noise presented from the front at a fixed signal to noise ratio of +10 dB.

In addition to the 6- and 12-month post-activation assessments, speech performance was also assessed at 9-month post-activation using CNC words presented at 45, 55, 65, and 75 dB SPL (two lists, 2 × 50 words, female speaker, scored for phonemes correct). This additional testing was aimed at comparing performance in both EH and IH mode more systematically across a wider range of speech input levels, such as would be encountered in real-life settings.

For all speech tests, calibration of presentation levels was performed with a sound level meter (NTi Audio XL2) measured at the center of the listening position (without the participant in place). The stimulus used for calibration was a ⅓ octave narrow-band noise centered at 1 kHz that had the same root-mean-square level as the speech corpus.

Aided sound field thresholds were measured using warble tones at frequencies 250, 500, 1000, 2000, and 4000 Hz with the contralateral ear plugged. EH and IH mode aided thresholds were measured on occasions when T- and C-levels were changed by the clinician.

Safety outcomes were assessed throughout the study by means of the recording and reporting to the local HREC of all adverse events (AEs) and device deficiencies. Reporting covered the 2-year study period for all participants and extended beyond this as per standard obligations of the manufacturer. Data were collated at a date by which time participants had been implanted for between 28 and 44 months. As noted, an Independent Data Safety Monitoring Board also reviewed safety data at regular intervals throughout the study.

The Speech, Spatial, and Qualities of Hearing scale (SSQ-12) (Noble et al. 2013) and Health Utilities Index Mark 2/3 (HUI 2/3) (Horsman et al. 2003) were administered preoperatively and at 6-month post-activation to assess patient-reported outcomes with the TICI Research System. The SSQ-12 contains 12 questions covering aspects of speech perception, spatial hearing, and more general qualities of hearing such as listening effort. A change of more than 1.0 on the scale represents a clinically meaningful improvement. The HUI 2/3 is a generic health profile and preference-based system used for the purposes of measuring health status, reporting health-related quality of life, and producing utility scores. It is a 15-item questionnaire for a self-administered, self-assessed, 1-week health status assessment. A change of 0.1 is indicative of the minimum clinically important change. For SSQ-12 and HUI 2/3, outcome measures used the participant’s own HAs as the preoperative baseline (described as the best aided condition, i.e., bilateral or unilateral HA depending on the participant). At 6-month post-activation, the responses were based on use of the TICI Research System generally (i.e., not specifically to use of either EH or IH mode).

The Tinnitus Handicap Inventory (THI) is a 25-question measure that identifies difficulties the patient may be experiencing because of their tinnitus, and allows classification into severity of handicap caused by tinnitus based on their total THI score (Newman et al. 1996). A change score of at least seven points has been considered to denote reliable clinically significant improvement on the THI (Zeman et al. 2011). This was administered preoperatively and at 6-month post-activation to assess changes in tinnitus.

The Patient Satisfaction Survey (PSS) was created for the purpose of the study and designed to capture patient feedback and assess overall satisfaction when using the EH and IH modes. It was administered at 6-month post-activation and comprised 22 questions that related to independence (e.g., personal security and confidence), ease of use, hearing ability, and the ability to always hear (i.e., when EH mode is impractical). The participants' free text responses were collated and summarized qualitatively. Overall satisfaction of IH and EH modes was assessed with a question using a five-point Likert response scale (1: Extremely Dissatisfied, 2: Dissatisfied, 3: Neutral, 4: Satisfied, 5: Extremely Satisfied).

Statistical Analyses

Co-primary endpoints compared EH mode and IH mode hearing performance at 6-month post-activation to the preoperative baseline for both CNC words in quiet and AuSTIN sentences in noise. In EH mode, a superiority investigational design was used to demonstrate that hearing performance was superior to preoperative performance in the unilateral aided listening condition. These were evaluated using the Intent-to-Treat (ITT) population. For IH mode, a non-inferiority investigational design was used to demonstrate that hearing performance was non-inferior to performance preoperatively in the unilateral aided listening condition. Assessment of non-inferiority was based on a two-sided 95% confidence interval (CI) for the mean change in the outcome (6-month post-activation less preoperative baseline value), and non-inferiority was concluded if the lower limit of the confidence interval exceeded the non-inferiority margin (−10%). This non-inferiority margin has been used in previous clinical studies (Gilden et al. 2015; Slager et al. 2018) and is based on clinical consensus. This was conducted in the Per Protocol (PP) population. If non-inferiority was demonstrated, the analyses proceeded to a test of superiority in the ITT population.

For the co-primary endpoints, sample size estimation was conducted using PASS 14 statistical software, assuming a one-sample t test and a one-sided 0.025 α level. Retrospective data were used to estimate the SD of change (Briggs et al. 2008; Runge et al. 2016). The analysis indicated a minimum sample size of 6 participants was required to provide adequate power for the co-primary endpoints at a 90% power level. An increased sample size of up to 10 participants was planned, to allow for the possibility of subject attrition as well as enabling generalization of outcomes, both performance and safety, with the TICI Research System to a wider adult clinical population.

Speech perception data, collected for EH and IH modes at 6- and 12-month post-activation (CNC words in quiet and AuSTIN sentences in noise), were analyzed using a linear mixed effects model to allow for unequal sampling across time points (9 participants at 6 months, 7 participants at 12 months). Words correct were modeled as the dependent variable with implant mode (EH, IH) and time point (6, 12 months) as within-subject factors.

Speech perception data, collected for EH and IH modes at the 9-month time point (CNC phonemes in quiet at 45, 55, 65, and 75 dB SPL), were analyzed using a two-way repeated measures analysis of variance (RM-ANOVA), modeling words correct as the dependent variable with implant mode (EH, IH) and phoneme presentation level (45, 55, 65, and 75 dB SPL) as within-subject factors.

Aided thresholds, collected for EH and IH modes, were analyzed using a two-way RM-ANOVA, with thresholds as the dependent variable and implant mode (EH, IH) and stimulus frequency as within-subject factors.

Following RM-ANOVA and linear mixed effects modeling, assumptions of normality and equal variance of residuals were assessed with the Shapiro–Wilk test and studentized Breusch–Pagan test, respectively. Post hoc comparisons were conducted when interactions or main effects were significant according to a Chi-squared test, and p values were corrected for multiple comparisons using the Tukey method.

The safety objective of the study was to evaluate the safety of the TICI Research System for use in an adult population. All implanted participants (ITT) were considered in the safety dataset for the purposes of medical/surgical and device-related AEs at 2 years post-activation.

A further objective of the study was to investigate whether the TICI Research System was a feasible treatment for restoring hearing in terms of patient-reported outcomes. All participants (ITT) were considered. PSS, THI and HUI2/3 were assessed with descriptive statistics and a qualitative assessment. The SSQ-12 was analyzed using a two-tailed paired t test, comparing baseline to 6-month post-activation scores.

Datalogging reports, downloaded from the sound processor at each postoperative session, allowed insights into the duration of use in EH mode, IH mode, and Off Air (when the implant was turned off and not in use). The duration in each mode was summed over the 2-year period post-implantation. For each mode, data were converted to a proportion of total time, and are presented in units of hours per day, therefore representing the daily average use over the study. To investigate whether IH mode performance predicted the IH utilization, a Pearson correlation analysis was performed using IH benefit (defined as the speech score difference between the modes, i.e., IH-EH) and the proportion of time in IH mode (i.e., IH/ [IH + EH]). This was completed for both words in quiet and sentences in noise at 6 and 12 months.

All 10 participants recruited into the study were included in the ITT population. Only participants who completed the study per the protocol were included in the PP population. The 3 participants excluded did not meet the inclusion criterion of ≤15 dB difference between masked air conduction and bone conduction hearing thresholds. It was considered appropriate to still include these participants in the analysis of study outcomes, as bone conduction thresholds were at or near maximum audiometer limits in the non-implanted ear and only occurred at isolated frequencies. While these 7 participants had baseline results, only 6 of these participants had a 6-month post-activation result available due to Coronavirus Disease restrictions preventing the visit, thus yielding 6 participants with paired differences for the PP population.

RESULTS

Speech Perception

Words in quiet scores at baseline and 6 months for the ITT population are presented as box plots in Figures 3A, B. For EH mode, the baseline mean word score was 1.9% as compared with 40.4% at 6 months, an improvement of 38.5% which was statistically significant (95% CI, 24.6 to 52.5) (refer Fig. 3A). For IH mode, the baseline mean word score was 1.9% as compared with 27.1% at 6 months, an improvement of 25.2%, which was statistically significant (95% CI, 13.7 to 36.7) (refer Fig. 3B).

Fig. 3.

CNC word performance (% correct word scores presented at 60 dB SPL). A, Baseline (HA) and 6-mo post-activation using EH mode (n = 9). B, Baseline (HA) and 6-mo post-activation using IH mode (n = 9). C, 6- and 12-mo post-activation using EH and IH modes (n = 7). Boxplots show % correct word scores, with boxes and whiskers showing first quartile, median, and third quartile, and minimum and maximum scores. The group mean score is shown as a filled circle. Statistical difference between each of the listening modes is shown ***p < 0.001, **p < 0.01, *p < 0.05. EH indicates external hearing; HA, hearing aid; IH, invisible hearing; ns, nonsignificant.

Words in quiet scores at 6 and 12 months are presented as box plots in Figure 3C. Mean word scores for EH and IH modes at 12-month post-activation were 51.4% and 39.4% respectively. It should be noted that 9 participants were tested at 6 months, and 7 participants were tested at 12 months. A linear mixed effects model was fitted, and assumptions of normality (p = 0.051) and equal variance (p = 0.44) were upheld. A significant effect of hearing mode (p < 0.001) and a significant effect of time point (p = 0.024) were found. The interaction of hearing mode and time point was not significant (p = 0.791). Post hoc comparisons for the main effect of time point revealed an estimated marginal mean difference of 6.8% between 6 and 12 months (averaged across hearing modes). The estimated marginal mean difference between IH and EH was 12.7% (averaged across time points), and within each time point was 13.3% at 6 months (p < 0.001) and 12.0% at 12 months (p = 0.004). This analysis suggests that while intelligibility for both IH and EH improved between the 6- and 12-month time points, the difference between IH and EH remained consistent over the same period.

To further examine the change over time, a paired t test was used to compare the difference between IH and EH for the 6 participants who were tested at both 6 and 12 months. The t test was not significant (p = 0.815), further indicating that the difference between IH and EH was consistent over time.

Sentences in noise scores at baseline and 6 months in the ITT population are presented as box plots in Figures 4A, B. For EH mode, the baseline mean score was 2.8% as compared with 77.6% at 6 months, a statistically significant improvement of 74.8% (95% CI, 53.4 to 96.2). In this case, due to the non-normality of the data, a Wilcoxon signed-rank test was used (refer Fig. 4A). For the IH mode, the baseline mean score was 2.8% as compared with a mean of 67.3% at 6 months, a statistically significant improvement of 64.5 % (95% CI, 42.9 to 86.0) (Fig. 4B).

Fig. 4.

Sentence in noise performance (% correct morpheme scores presented at 60 dB SPL, +10 SNR). A, Baseline (HA) and 6-mo post-activation using EH mode (n = 9). B, Baseline (HA) and 6-mo post-activation using IH mode (n = 9). C, 6- and 12-mo post-activation using EH and IH modes (n = 7). Boxplots show % correct morpheme scores, with boxes and whiskers showing first quartile, median, and third quartile, and minimum and maximum scores. The mean score is shown as a filled circle. Statistical differences between each of the listening modes are shown ***p < 0.001, **p < 0.01, *p < 0.05. EH indicates external hearing; HA, hearing aid; IH, invisible hearing; ns, nonsignificant.

Sentences in noise scores at 6 and 12 months are presented as box plots in Figure 4C. The group mean scores for EH and IH mode at 12 months were 80.7% and 76.3% respectively. Note that 9 participants were tested at 6 months and 7 participants were tested at 12 months. A linear mixed effects model was fitted, and the effect of hearing mode, time point, and their interaction was all nonsignificant (p = 0.069, p = 0.822, and p = 0.160, respectively). Assumptions of normality (p = 0.961) and equal variance (p = 0.413) were upheld. While the model's main effects and interactions were nonsignificant, the effect of hearing mode was close to the significance threshold of 0.05, suggesting that hearing mode may have had an influence on the scores. To explore the data further, post hoc comparisons were conducted using a conservative Bonferroni correction to protect against type I errors (false positives). At 6 months, hearing mode had a significant effect on sentence scores (p = 0.021); the estimated marginal mean difference between IH and EH was 10.3%. At 12 months, the effect of hearing mode was nonsignificant (p = 1.000, estimated marginal mean difference 1.4%). This tends to suggest there may have been a difference between IH and EH at the 6-month time point that was not present at the 12-month point. To further examine the change over time, a Wilcoxon signed-rank test was used (due to non-normal distribution) to compare the difference between IH and EH for the 6 participants who were tested at both 6 and 12 months. The Wilcoxon signed-rank test was significant (p = 0.031), further indicating that the difference between IH and EH changed over time.

To characterize the performance of the implanted microphone in terms of speech intelligibility across a wider range of speech presentation levels that would be encountered in real-life situations, the performance-intensity function for CNC phonemes in quiet was measured in 6 participants at 9-month post-activation using both EH and IH modes. Phoneme scores are presented as box plots for each of the four presentation levels (Fig. 5). The two-way RM-ANOVA revealed a significant interaction between hearing mode and presentation level [F(3,15) = 38.722, p < 0.001]. Post hoc paired comparisons between IH and EH at each of the presentation levels were conducted. The performance of EH was significantly better than that of IH at three of the four presentation levels. The difference was 33.9% (p < 0.001), 21.7% (p < 0.001), and 9.8% (p = 0.002) at 45, 55, and 65 dB SPL, respectively. There was no significant difference between IH and EH at the highest presentation level of 75 dB SPL (p = 0.384).

Fig. 5.

CNC phoneme performance (% correct phoneme scores presented at 45, 55, 65, and 75 dB SPL) at 9-mo post-activation using EH and IH modes (n = 6). Boxplots show % correct phoneme scores, with boxes and whiskers showing first quartile, median, and third quartile, and minimum and maximum scores. The group mean score is shown as a filled circle. Statistical difference between each of the listening modes is shown ***p < 0.001; **p < 0.01, *p < 0.05. ns indicates nonsignificant.

Aided Thresholds

Aided thresholds are presented in Figure 6. These were measured at 6-month post-activation and were used to compare the threshold differences between EH and IH modes across the frequency range from 250 to 4000 Hz. Aided thresholds were measured on occasions when T- and C-levels were modified by the clinician. The 6-month data set (n = 10) was created from the 4 participants who had thresholds measured at 6 months and from the 6 participants who had thresholds measured at 3 months. Given that the T- and C-levels were not modified at the 6-month time point for these 6 participants, the use of the 3-month thresholds was deemed acceptable. A two-way RM-ANOVA was performed, which revealed a significant effect of hearing mode (p < 0.001), a nonsignificant effect of frequency (p = 0.113), and a significant interaction between hearing mode and frequency (p < 0.001), indicating there was a difference in aided thresholds obtained with IH and EH mode, and the difference varied with frequency. Post hoc analysis revealed a significant difference between IH and EH at each frequency. The mean difference at 250 Hz was 3.5 dB (p < 0.007), at 500 Hz was 4.0 dB (p < 0.002), at 1000 Hz was 8.5 dB (p < 0.001), at 2000 Hz was 14.0 dB (p < 0.001), and at 4000 Hz was 11.0 dB (p < 0.001).

Fig. 6.

Group mean aided sound field thresholds for EH (gray circles) and IH (white circles) combined at 6-mo post-activation (n = 10). The black lines indicate the SD. Statistical differences between each of the listening modes are shown, ***p < 0.001; **p < 0.01. EH indicates external hearing; IH, invisible hearing.

Safety Outcomes

A total of 62 AEs were reported over the course of the study. Of these, 21 were reported as being related to the ear or device. Seven of the AEs were classified as Serious Adverse Events. Five of these were unrelated to the device or procedure and included: two cases of atrial fibrillation, one case of hernia surgery, one case of abdominal pain, and one case of pregnancy/childbirth. The remaining two Serious Adverse Events were device failures in 2 participants, which occurred at 15- and 22-months post-surgery, respectively.

In both of these cases, device lockdown occurred automatically as a result of potential fluid ingress detected through a Microphone Fluid Ingress Test. The device lockdown was not designed to be reversible and, as such, both participants had their experimental devices explanted and were successfully reimplanted with a Cochlear Nucleus CI612 device, as per the experimental protocol. Following the explanation, thorough testing for root cause showed that in both cases, there was a failure of the insulation between the microphone chassis and the microphone wire.

Of the remaining 14 AEs reported as being related to the ear or device, there were nine cases associated with temporary pain, soreness and/or discomfort over the implant site, one case of discomfort with the external speech processor, one case of a ringing sound in the implanted ear, one case of dizziness post CI surgery, one case of a stretched feeling at the implant site, and two cases of headache, all of which were resolved.

Of the 8 study participants who were using the TICI Research System at study completion, to date, 7 continue to use their implant in both EH and IH modes, and 1 participant is deceased due to an unrelated comorbidity.

Questionnaires

All participants completed the patient-reported outcomes protocol (n = 10). For the SSQ-12 and Quality of Life (HUI 2/3), the preoperative baseline to 6-month post-activation comparison relates to everyday experiences with device(s), so it does not indicate benefits specific to EH or IH mode. The baseline and 6-month group mean SSQ-12 scores were 3.46 (95% CI, 2.74 to 4.19) and 5.30 (95% CI, 4.26 to 6.35), respectively. The mean score increase of 1.84 (95% CI, 1.00 to 2.65) was statistically significant and was considered a clinically meaningful improvement. The preoperative baseline group mean (SD) Quality of Life (HUI 2/3) score was 0.69 (0.14) and the 6-month score was 0.78 (0.12), indicating a mean (SD) change in HUI 2/3 score of 0.10 (0.14). Two participants showed no change, 2 participants showed negative change of approximately 0.05 (below the range of what is deemed qualitatively meaningful), 2 subjects showing changes above the 0.03 threshold that has been deemed indicative of “noticeable improvement in quality of life” (Furlong et al. 2001; Müller et al. 2021) and 4 participants showed change of 0.1 or above, which is deemed a minimum clinically important change (Müller et al. 2021). While indications of improved quality of life were apparent for some participants, the small sample size is not sufficient to make authoritative statements about the overall mean Quality of Life benefit.

Most participants (7 out of 10) self-reported no tinnitus symptoms. Three participants reported tinnitus at baseline (1 moderate, 1 mild, and 1 very mild in severity), with a clinically meaningful reduction in symptoms at 6 months as reflected by decreasing THI scores for all 3 participants. One participant reported a reduction in tinnitus severity from moderate to very mild at 6-month post-activation.

The PSS showed mean (SD) ratings of 4.3/5 (0.48) for EH mode and 3.3/5 (1.49) for IH mode, indicating that overall, participants were satisfied with the device at 6-month post-activation. A qualitative analysis of participant responses in the PSS indicated that water-related improvements in capability provided by IH mode were highly valued. Six participants stated that the ability to hear while swimming improved their sense of safety, particularly if they were supervising children, and 4 participants mentioned the usefulness of the IH mode for showering. Two subjects rated the discretion and cosmesis provided by using the IH mode highly, enabling them to be “treated like everyone else” and “more confident.” The ability to wear items such as bicycle helmets, hats, and glasses more easily when in IH mode was also valued. Conversely, there were participants who were less concerned with aesthetic issues and more interested in the practical hearing outcomes for the device, so they were more likely to opt for EH mode at times when they wanted to maximize their hearing ability, including accessing the directionality features available in this mode. The PSS also highlighted that there were differences among the participants regarding their need to hear at night. Three participants indicated a strong preference for not hearing at night to sleep better, four had no preference, and the remaining three used IH mode on the occasions when they wanted an increased sense of personal security. There were reports of the IH mode being too sensitive or noisy for sleeping, and a low sensitivity map was provided for this purpose.

Datalogs

Datalogs for each participant are displayed in Figure 7. The number of hours spent in EH and IH modes was summed over the 2-year period and presented as daily average hours (h/day) of use (these values are indicated for each participant). Usage patterns varied between participants, with time in IH ranging from 3.3 to 13.2 h/day (median 6.7), time in EH ranging from 1.3 to 11.9 h/day (median 6.8), and total time on air ranging from 7.0 to 15.3 h/day (median 13.9). The percentages shown in Figure 7 indicate the proportion of time on-air spent in IH mode for each participant, ranging from 22 to 88% (median 49 %). When averaged across the entire group, time using IH mode (6.6 h/day) was greater than the time spent in EH mode (5.8 h/day). To investigate whether IH mode performance predicted IH utilization, a Pearson correlation analysis was performed, with no significant relationships found at 6 or 12 months for either words in quiet (r² = 0.198 at 6 months, r² =0.360 at 12 months) or sentences in noise (r² = 0.248 at 6 months, r² =0.278 at 12 months), suggesting that IH speech performance, in this study, was not predictive of IH utilization.

Fig. 7.

Daily use categorized into EH, IH, and Off-Air modes for each participant, as extracted from sound processor datalogging, averaged for the 2-yr post-implantation. Total time on air shown in hours per day, percentages show the proportion of time on-air spent in IH mode. EH indicates external hearing; IH, invisible hearing.

DISCUSSION

The present study aimed to investigate the efficacy, safety, and patient-reported outcomes for the TICI Research System in both the EH and IH modes in a group of 10 adult recipients with bilateral moderately-severe to profound sensorineural hearing loss. Co-primary endpoints compared EH and IH mode performance for words in quiet and AuSTIN sentences in noise measured at 6-month post-activation to the preoperative baseline scores. For EH mode, all statistical hypotheses were upheld (n = 9 ITT) with superiority achieved, demonstrating a significant improvement in speech perception as compared with preoperative baseline scores. Similarly, for IH mode, all statistical hypotheses were upheld with both non-inferiority (n = 6 PP) and superiority (n = 9 ITT) achieved, demonstrating a significant improvement in speech perception scores as compared with baseline preoperative scores. These findings of improvements in speech understanding and hearing-related quality of life outcomes after cochlear implantation are consistent with those reported in the literature (McRackan et al. 2018; Boisvert et al. 2020; Cuda et al. 2024), and are not surprising given that, in the present study, the mean baseline word score for the 10 participants was 2.2% (range 0 to 6%).

To place the EH mode mean word scores of 40.4% and 51.4% at 6 and 12 months, respectively into a broader perspective, a review of outcomes for adults with post-lingual hearing loss, implanted at the Royal Victorian Eye and Ear Hospital (n = 382) (Leigh et al. 2016), reported a mean postoperative CNC word score of 45% (range 0 to 94%, median 46%), for words spoken by a male speaker and presented at 65 dB SPL. A scoping review of CI outcomes conducted by Boisvert et al. (2020) found that the average word score performance, typically evaluated at 12-month post-implantation and at test presentation levels ranging from 60 to 70 dB SPL, was improved from 8.2% pre-implant to 53.9% post-implantation. Results from the present study are comparable, particularly considering that participants were tested at a softer presentation level of 60 dB SPL and using a female speaker, both of which would increase the test difficulty.

For sentences in noise performance, it is more difficult to directly compare scores with those reported in other studies, due to the wide range of test paradigms typically used for speech testing (e.g., speech material, noise types, signal to noise ratio), all of which have been shown to have an impact on the outcomes.

Whilst the results for both EH and IH modes showed benefit over baseline scores, the comparison between IH and EH modes is instructive in understanding the benefits and limitations of the performance of the implanted microphone. EH mode scores were significantly higher than IH mode scores for words in quiet, by 13.3% and 12.0%, at 6 and 12 months, respectively. However, the period between 6- and 12-month post-implantation assessments, words in quiet score performance, averaged across both IH and EH modes, improved by 6.8%, but the difference between IH and EH modes did not, indicating that further experience with CI stimulation was beneficial, regardless of hearing mode. The analysis of the sentence in noise performance provided a more nuanced statistical insight. While the EH mode scores were superior to IH mode at 6 months (10.3%), by 12 months the difference was reduced (1.4%) and was not significant. While the gap between hearing modes diminished at 12 months, the power to observe differences between hearing modes and changes over time may have been limited due to test ceiling effects, given that 4 participants scored above 80% for IH and EH modes at both 6- and 12-month time-points. Testing at a lower signal to noise ratio may have allowed better insight and will be included in future studies.

A loudness difference between IH and EH modes was reported by participants based on their use of the device outside of the clinic. Loudness and level were controlled during the trial by regularly calibrating the microphone to address potential changes in the microphone response within the body. The difference in loudness between IH and EH modes was unexpected and is likely to be due to an underlying issue with the equalization process and/or measurement. These differences in loudness remained unresolved during the study and are likely to have contributed to the observed differences in speech performance between IH and EH modes. A further contributing factor to the speech performance differences between IH and EH modes is associated with the inherent noise floor of the implanted microphone. While the implanted microphone response was equalized in sensitivity across all frequencies, its response exhibited a steep roll-off in the high frequencies, and the equalization used for compensation also boosted the inherent noise floor. It is likely that the higher noise floor of the IH microphone also contributed to the observed speech performance and aided threshold differences, particularly impacting high-frequency components of speech, compounded by the use of a female speaker.

The characterization study of the implanted microphone performance at 9 months aimed to provide a deeper insight into the differences between EH and IH modes, by assessing speech test outcomes across a range of presentation levels, such as would normally be encountered during everyday use of the device. At the highest presentation level of 75 dB SPL, there was no significant difference between EH and IH mode scores. At lower presentation levels, a significant difference was found, and the extent of the difference widened as the presentation level was lowered. The gap between IH and EH modes was smallest at 65 dB SPL (9.8%), and largest at the lowest presentation level of 45 dB SPL (33.9%). Likely, both the overall loudness of IH mode and the inherent noise floor of the implanted microphone contributed to these observations, perhaps via different mechanisms. A loudness difference alone is expected to cause a simple horizontal shift in the performance-intensity function. Meanwhile, the noise floor is expected to impact low presentation levels more than higher levels due to the masking of high-frequency speech sounds, causing a tilt in the performance-intensity function. Both aspects (shift and tilt) were observed in the comparison of the performance-intensity function between IH and EH modes, supporting the hypothesis that both overall loudness and high noise floor were likely contributors to IH mode speech performance limits observed in this study. The differences in characterization score at different levels are important findings, as they demonstrate that given more equivalent loudness levels of the IH and EH microphones and a reduction of the noise floor of the IH microphone can be achieved, then the performance of the IH mode can be equivalent to the EH mode.

Patient-reported outcomes were consistent with positive hearing benefits and patient satisfaction ratings, with improved quality of life reported for some participants. Overall, participants reported that when using IH mode, some body noise was noticeable but not intrusive, and its presence was generally well accepted. This would suggest that the inclusion of the body noise reduction technology was at least partly effective in addressing this issue, which was reported in the earlier TIKI study (Briggs et al. 2008). Datalogs showed varied IH usage across participants, indicating that different user needs in terms of access to IH and EH modes and internal battery recharging requirements should be considered in future TICI designs.

In terms of safety outcomes, despite the larger internal package of the investigational TICI implant, there were no serious perioperative complications related to the device or surgical placement. A range of post-implantation AEs, including sensitivity and soreness over the site of the implant, some postoperative dizziness, and two cases of headaches, were recorded and resolved successfully. These events were adjudged by the Independent Data Safety Monitoring Board to be consistent with those reported for adult patients in the CI Clinic after cochlear implantation.

There were two device failures reported at 15- and 22-months post-surgery. Consistent with good clinical practice for any study investigating the safety and efficacy of a prototype investigational device, the possibility of a higher device failure rate as compared with the standard clinical device, and the potential consequence of re-implantation, were risks that patients were informed of preoperatively. In the present study, based on reliability data for both the Carina microphone and Nucleus CI, the incidence rate of failure was expected not to exceed 0.0025%. As reported, the incidence of device failure in the study was 2/10 investigational devices implanted (20%), which exceeded the safety endpoint.

Due to activation of the automated lockdown safety process, neither of the two participants experienced any inadvertent stimulation or auditory percepts. However, the lockdown process, as designed in the investigational device, resulted in a permanent loss of primary function, meaning that it was not technically possible for the TICI Research System to operate in EH mode. As a consequence, both participants were reimplanted with Cochlear Nucleus CI612, as per the experimental protocol for a failed device, which had been discussed with the participants during their informed consent process. This is a consideration for future TICI design, where the intended purpose of the lockdown mechanism would be to allow continued use in EH mode, even in the situation of microphone failure.

Further root cause investigation of the two explanted devices to identify fault mechanisms and identify any necessary remediation in the design of the device revealed that, in both cases, a failure of insulation between the microphone chassis and microphone wire was the cause. The root cause investigation did not identify any fundamental problem with the use of a transcutaneous microphone, but highlighted the need for improved electrical insulation protection of the microphone. This issue is currently being addressed for any future devices.

Limitations

Given the variability in patient-reported outcomes and usage patterns reported with the TICI Research System, it is relevant that future TICI devices should be assessed in studies with larger cohorts and should include assessments across a range of acoustic environments to better assess overall benefits. Regarding patient-reported outcomes, the SSQ-12 and HUI 2/3 relate to everyday experiences using the device and were not designed to indicate benefits specific to EH or IH mode, which limited the conclusions that could be made about each mode. Future studies should consider the inclusion of patient questionnaires that make a direct comparison between IH and EH modes, and which include a more detailed analysis of scenarios and usage times for both IH and EH modes. The inclusion of a method of capturing body noise annoyance and/or intrusiveness, such as a Likert-type scale, would be useful for future studies to allow statistical analysis rather than relying on free-text participant feedback as was done in the present study. Given the reliance of IH mode on a rechargeable battery, longer-term follow-up of battery recharging performance is also important to be included in future studies.

CONCLUSIONS

The results demonstrate the feasibility of the investigational TICI Research System in providing functional hearing for moderately-severe to profound post-linguistic sensorineural hearing loss in adults. Compared with pre-implant aided speech perception scores, significant group mean improvements were observed for speech perception in quiet and in noise, with both EH and IH modes, post-implantation. Direct comparison between IH and EH modes revealed that while some test conditions showed equivalent performance (e.g., high presentation levels), overall EH mode had significantly better speech perception performance compared with IH mode. This is likely due to differences in loudness and/or microphone noise floor, which could be addressed in future TICI design iterations to improve performance. Patient-reported benefits were observed, with all participants using IH mode on a regular basis over the duration of the study, and feedback indicated that the level of body noise in the IH mode was acceptable. Despite speech outcomes in the EH mode being overall superior, the IH mode was reported by subjects as offering benefits over the use of a conventional CI in specific situations. Device failures occurred in two participants. Neither participant experienced any inadvertent stimulation due to the automated device shutdown process, and both participants were able to be reimplanted with conventional Cochlear Nucleus devices. Root cause analysis identified a problem with microphone-related fluid ingress issue to be addressed in future devices. No other safety issues were detected.

In summary, this study demonstrated the feasibility of the TICI Research System to provide patient benefit in both EH and IH modes. No fundamental problem with the use of a transcutaneous microphone was detected. These findings support further development of the TICI design and approach.

ACKNOWLEDGMENTS

The authors thank all patients participating in the study. A special thank you also to all clinic staff supporting the study at RVEEH. The authors also acknowledge the editorial assistance provided by Kerrie Plant, Paul Boyd, Josie Wyss, Beth Elks, and Beejal Vyas-Price during the development of this article.