Authors: Xianhui Wang, Phillip Tran, Michelle R. Kapolowicz, Thomas Lu, Ginger Stickney, Arnold Starr, Hamid Djalilian, Fan-Gang Zeng
Categories: Article, Cochlear implant, Facial nerve stimulation, Tinnitus, Hyperacusis, Speech perception, Side effects
Source: Cochlear implants international
Authors: Xianhui Wang, Phillip Tran, Michelle R. Kapolowicz, Thomas Lu, Ginger Stickney, Arnold Starr, Hamid Djalilian, Fan-Gang Zeng
Cochlear implants can restore functional hearing but may also cause side effects such as facial nerve stimulation, sound sensitivity or reactive tinnitus. The present study aimed to establish a general framework for optimizing cochlear implantation parameters to manage these side effects while maximizing speech perception performance. A second objective was to elucidate on how side effect origins can impact treatment outcomes.
Eight adult cochlear implant subjects were referred to us with intolerable side effects that rendered device usage difficult or even impossible. New maps were created by implementing strategies such as reducing stimulation levels, increasing pulse duration, reducing stimulation rate, altering channel gains and frequency maps, deactivating problematic electrodes, or a combination of the above. Outcomes were measured in terms of side effect reduction and changes in speech performance.
Facial nerve stimulation was eliminated in three of five subjects, with the remaining two showing partial improvement. Sound hypersensitivity was eliminated in two of two subjects. Tinnitus was alleviated in three of four subjects, while the remaining one with cerebellar malformation experienced no change. Speech performance was either maintained or improved in all subjects. Except for the subject with cerebellar malformation who chose to explant the device that produced reactive tinnitus, all subjects were able to use the implant effectively without bothersome side effects.
Facial nerve stimulation is usually related to electric current spread of the implant on the same side, which can be effectively managed by customized strategies. In contrast, the origins of sound sensitivity and reactive tinnitus can be more variable and therefore, are more difficult to manage.
Customized mapping can alleviate cochlear implant side effects without compromising speech performance.
The cochlear implant (CI) is a safe and effective medical device that restores partial hearing for over one million individuals with various degrees of hearing loss (Zeng, 2022). The implant converts acoustic signals into electric current that can be perceived by the user. Optimal performance can be achieved by adjusting stimulation parameters such as pulse amplitude, pulse duration, pulse shape, stimulation rate, stimulation mode, number of maxima, frequency allocation, dynamic range, and channel gain. These parameters are then saved as a program (referred to as a “map”) onto the CI external processor. This process is often iterative, with adjustments being made over many months to account for user experience, performance, and desired comfort.
Side effects may occur for some users, complicating the mapping process and impeding device usage (Cohen et al., 1988; Venail et al., 2008). Meta-analyses have shown that facial nerve stimulation (FNS) can occur in up to 16% of users (Berrettini et al., 2011; Kelsall et al., 1997; Venail et al., 2008; Weber et al., 1995) and often presents as twitching or pain around the eye, mouth, nasolabial fold, or forehead (Kelsall et al., 1997). Subjective Tinnitus, the perception of a phantom sound, has also been shown to develop post-activation (Quaranta et al., 2004; Venail et al., 2008). Sound sensitivity to cochlear implants involves complaints usually about mechanic or unnatural sound quality that typically resolves over time, but sometimes about hypersensitivity that does not resolve over time (e.g., Caldwell et al., 2017; Zeng et al., 2014). Additionally, vestibular problems may arise in the form of dizziness or even nausea, affecting the user’s ability to perform daily activities (Venail et al., 2008). Longstanding issues with poor speech understanding can lead to frustration with the device and increase the potential for discontinuing its use (Moberly et al., 2016).
The clinical map provided for an individual CI user is often based on thresholds and maximum comfort levels (MCL) of stimulation, with a focus on optimizing speech performance and listening comfort. However, if side effects do occur, the map needs to be adjusted to reduce or eliminate their impact on the user. At present, there is no standard guideline for dealing with cochlear implant side effects. In some cases, clinicians may disable individual or groups of electrodes that cause the side effect. However, this comes at a cost of possible reduced speech understanding due to limited spectral resolution (Muckle & Levine, 1994). Additionally, the causes and management strategies for each side effect vary in complexity. For example, facial nerve stimulation is generally understood to result from current spread to the facial nerve, making it more predictable and manageable with parametric adjustment such as in pulse duration and amplitude. Conversely, hypersensitivity and tinnitus lack a clear, consistent cause, often presenting in patients with normal cochlear anatomy and implant function otherwise. These side effects require a highly individualized approach, as successful treatments for one patient may not be effective for others. Thus, treatment outcomes for hypersensitivity and tinnitus can vary widely and may not be as predictable as those for facial nerve stimulation. Moreover, when the complexity of physiological causes of side effects is combined with the technical challenges of cochlear implant programming, it can take considerable time to find an effective map. For some patients, the delay in addressing side effects may lead to discontinuation of device use.
We have been referred many cochlear implant users with side effects over the last thirty years. To alleviate side effects, we have combined and integrated practical strategies from experienced clinicians and researchers with basic knowledge about spatial selectivity of electric stimulation (e.g., Grill & Mortimer, 1996; Tang et al., 2011). To address facial nerve stimulation, the primary strategy is to reduce the current level by increasing pulse duration (Zeng et al., 1998). However, using longer pulse duration can sometimes lead to an automatic reduction in stimulation rate. Therefore, a secondary adjustment to the stimulation rate may be needed. If these adjustments are ineffective, the culprit electrode would need to be deactivated. To address sound hypersensitivity, the primary adjustment is to lower the maximal comfort level, and additionally change the input dynamic range, gain or compression parameters. For tinnitus management, parametric adjustments depend on individuals, with general adjustments being made similar to those used for managing facial nerve stimulation and sound hypersensitivity. Additional adjustments may be needed from channel gain to electrode deactivation. In general, we aim to retain as many electrodes as possible. If electrode deactivation is unavoidable, adjustments such as frequency allocation would be made to optimize speech recognition and quality.
In this study, we report our experience with eight CI users who experienced one or more side effects from cochlear implant stimulation including facial nerve stimulation, sound hypersensitivity, and reactive tinnitus. We highlight two complex cases that involved multiple side effects due to central or unknown causes. A flowchart outlining the general approach is provided in the Methods section. Given the highly personalized adjustment, details are presented for each specific case in the Results section.
Eight CI users, aged between 18 and 89 at the time of testing, participated in the study (Table 1). Each subject was experiencing at least one side effect after CI activation. Facial nerve stimulation was the most common side effect (5 out of 8 S1, S2, S3, S4, S7), followed by tinnitus (4 S5, S6, S7, S8), hypersensitivity (2 S4, S5), and dizziness (1 S7). Five out of eight subjects had one side effect, and three had two or more. All subjects gave written informed consent to participate in the study under an experimental protocol approved by the University of California Irvine’s Institutional Review Board.
Basic audiological and medical examinations were performed. Audiological assessment primarily used speech recognition in quiet. In most cases, HINT sentences were used, consisting of 25 phonemically balanced lists of 10 sentences each, spoken by a male talker (Nilsson et al., 1994). Each sentence includes five to six keywords, allowing scoring at both word and sentence levels, with vocabulary approximately at a first-grade level. For three cases (S1, S2, and S8), AzBio sentences were used, which contain less contextual information than the HINT sentences (Spahr et al., 2012). In S8’s case (bilateral CI user), speech reception threshold was measured with HINT sentences in steady-state noise. All speech scores were based on two test lists. Radiological imaging was conducted for two cases with suspected abnormalities in cochlear implant positioning. Post-operative computed tomography (CT) of the temporal bone was obtained for S7 and S8 to verify electrode position in the cochlea, and magnetic resonance imaging (MRI) was used for S7 to examine anatomical brain structures. Patient history and radiological imaging results were requested conforming to the HIPAA regulation.
Fig. 1 shows the systematic approach to identify electrodes that cause a side effect and to adjust stimulation parameters for side effect management. The clinician begins with the default map, typically the original programming settings that the implant user reports experiencing side effects with. To start, impedance measurements are taken to identify and deactivate any short or open-circuit electrodes. From there, each electrode is systematically evaluated to determine its role in causing side effects. Using default stimulation parameters, each electrode is tested, with stimulation levels gradually increased from the minimum to the maximum comfortable level. If a side effect is detected, adjustments specific to the side effect are made (see Specific parametric adjustment below). If these adjustments do not alleviate the side effect, the electrode is deactivated. Remaining electrodes are integrated to form a new map, with further adjustments to global settings such as gain, volume, and frequency allocation to optimize speech performance. Once adjustments yield acceptable performance, the new map may be provided to the patient for extended acclimatization and follow-up re-evaluation. In our study, patients used their CIs for at least two weeks before a follow-up consultation to report any issues, including remaining side effects.
The following strategies were implemented for managing facial nerve (1) Examine and identify the specific electrodes and electric stimulation parameters that evoke facial nerve stimulation; (2) For those electrodes that stimulate the facial nerve, manipulate stimulation parameters from increasing pulse duration to lowering pulse amplitude and rate as well as changing electrode mode and electric waveform to reduce or eliminate the side effect; (3) If none of the manipulations can reduce the side effect to a negligible level, then deactivate the electrode(s); (4) Stimulation manipulation and electrode deactivation need to be balanced against maintaining speech performance.
The following strategies were implemented for managing sound (1) Examine each electrode to identify the specific electrode(s) and electric stimulation parameters that cause sound hypersensitivity; (2) Lower the MCL on these electrodes or even deactivate them if necessary; (3) Change sensitivity and volume control to further decrease sound sensitivity, while adjusting frequency maps to match electric hearing to residual acoustic hearing.
Although cochlear implantation has been known to generally suppress tinnitus (e.g., Chang and Zeng, 2012; Miyamoto and Bichey, 2003), it could worsen or even cause tinnitus in relatively rare cases. These cases, if present, are usually complicated and difficult to manage. To manage tinnitus, the parametric strategies for managing facial nerve stimulation and sound hypersensitivity were first employed. For those who did not benefit from adjustment, individualized adjustments were made additionally. Details of those adjustments are presented in the results section.
The method of adjustment was used to obtain thresholds and most comfortable levels for each electrode during cochlear implant stimulation. The threshold was measured by increasing the stimulation from a subthreshold level until just audible, then decreasing it from high to low until just inaudible. The threshold was determined as the average of the levels of being audible and inaudible. The MCL was determined by only the ascending method, in which the stimulation was increased always from a level below the MCL.
A scaling procedure was also used to determine the loudness of the stimulation or the magnitude of the side effect like tinnitus. The scale varied from 0 to 10, with 0 being no perception or side effect and 10 being intolerable loudness or side effect.
We organized this section according to the various side effects of facial nerve stimulation, sound hypersensitivity, and tinnitus. We first described the general strategies for managing the specific side effect, including each subject’s complaints. We then detailed mapping manipulations and their outcomes. We highlighted both successes and limitations of these strategies and outcomes of the two complex cases involving likely central origins of the side effects.
S1 was an 81-year-old male whose hearing loss was related to otosclerosis and presbycusis. He received a CI (Med-El Concert implant and Opus 2 processor) in his left ear. He had been wearing a hearing aid (HA) in his right ear. Since activation of his CI, the subject had been experiencing FNS in visible twitching progressively from lower left jaw to upper left eye with increasing stimulation level. The FNS side effect prevented him from using the CI in daily life. He was referred to us six months after CI activation in December 2012. Fig. 2A shows the standard map that caused FNS. We noted that all 12 electrodes were activated, with relatively short pulse duration being 10.42–20.42 μs. Other mapping parameters were set at (not shown in Fig. 2A): default volume = 100%, log frequency band = 70–8500 Hz, maplaw = 1000.00, AGC compression = 1 and default sensitivity = 75%.
We systematically increased the amplitude of the short pulses from threshold to MCL on each electrode and found that electrodes #1–9 all evoked FNS, with electrodes #4–6 being the worst in terms of evoking FNS at soft-to-moderate loudness levels. We then increased the pulse duration from 10.42–20.42 μs to 33.75–87.08 μs and decreased the pulse rate from 1967 Hz to 1130 Hz. Initially, we selected a relatively large increase in pulse duration (approximately ten times the original pulse duration) to rigorously assess its potential for reducing FNS. Once the effectiveness was confirmed, we minimized the pulse duration as much as possible while still avoiding FNS. The stimulation rates were adjusted accordingly. These manipulations eliminated FNS on all electrodes except for electrode #4 and 5, which were deactivated in a revised map (Fig. 2B). Although the CI alone did not produce any intelligible speech performance, the CI improved the bimodal performance from 28%-correct in AzBio sentences with hearing aid alone to 49% correct with both the implant and hearing aid. The patient was able to use the revised map without any FNS for one month until he noticed a bothering background low-pitched humming in his CI.
We suspected that the humming background might be caused by the low-frequency acoustic stimulation in the 70–8500 Hz frequency range or modulating components due to electrode interactions (Shannon, 1983; Tang et al., 2011). A new map was created by increasing the low-cut frequency from 70 to 300 Hz to reduce the low-frequency acoustic component, while deactivating five of the electrodes to reduce interactions (electrode #2, 5, 6, 8, 10; Fig. 2C). As a result of this electrode deactivation, we were able to increase the pulse duration for the remaining electrodes (#1,3,4,7,9,11) to 100–200 μs and reduce the stimulation rate to 473 Hz. The additional adjustment of pulse duration was to improve speech performance. This new map eliminated both the background humming and FNS. A speech test showed 15%-correct AzBio recognition with the hearing aid and 41% with both devices, which were similar to the previously revised map that eliminated FNS but produced the humming background.
In 2017, Med-El introduced MAESTRO 7 fitting software that allowed triphasic stimulation, which had been shown to reduce FNS (Bahmer et al. 2017). Indeed, we fitted S1 with a map using triphasic pulses that successfully produced comfortable loudness without any FNS for all electrodes except #5 and 6 (Fig. 2D). The triphasic stimulation also allowed the number of active electrodes to be increased from 6 to 10. Importantly, S1’s speech score using the triphasic stimulation map with the CI alone improved to 44% correct in AzBio words. Indeed, successful management and outcomes of S1 were a result of not only technological advances that allowed stimulation manipulations, but also close collaboration and cooperation between researchers, clinicians and patients. In this case, 31 different maps were created by the researchers and clinicians and tested by the patient for at least one week over a one-year period between 2012 and 2013.
S2 was an 89-year-old male whose cause of hearing loss was unknown. He received a CI (Cochlear CI532 and CP1000 processor) in his left ear after 15 years of HA use. He continued to use a HA on his right ear. He was referred to us because of twitching around the left eye during everyday CI use, even at conversational levels. We examined the original mapping and found that electrodes 1–4 were deactivated due to high impedance (>20 kΩ) across MP1, MP2, MP1+2, and common ground modes. Despite the deactivated electrodes and bothersome twitching, S2 was a good CI user, achieving 96% correct on AzBio sentences in quiet using the CI only.
We also noted that the original map used 25 μs default pulse duration for all remaining electrodes (#5–22). We systematically increased the pulse amplitude at this pulse duration and found four culprit electrodes (#15–18), which produced FNS at high stimulation levels. Because we did not want to jeopardize the good baseline performance, we slightly increased the pulse duration for electrodes #16–17 from 25 μs to 37 μs, further lowered the C-level by 7 current units for electrodes #16–17 and 3 units for electrodes #15 and 18 while leaving the pulse duration unchanged. All other electrodes and the processor setting were unchanged in the revised map. As a result of these changes, S3 showed an immediate elimination of FNS during conversation using the new map. At a two-week follow-up, S2 reported that the FNS had not returned. Re-test of his speech performance showed no reduction in speech intelligibility (95% correct on AzBio sentences in quiet with CI only). This case exemplifies how small parametric adjustments can effectively eliminate FNS while preserving speech performance.
S3 was an 18-year-old female who was born deaf due to Mondini dysplasia, an abnormality of the inner ear with incomplete partitioning of the cochlea and a reduced number of turns. She used hearing aids in both ears at age one. She received a CI (Cochlear CI24R(CS) and CP920 processor with ACE strategy) on the right ear at age two and a second CI (Cochlear CI24RE(CA) and CP920 with ACE) on the left ear at age five. The left implant failed after 4 years, and a revision surgery was performed with the same implant. S3 was referred to us because she had been experiencing persistent, mild, but bothersome, twitching around the left eye since activation.
Examination of the original map in the left CI showed that all 22 electrodes were working and activated, with higher impedances for the middle electrodes (10–12 kΩ for #11–14) than the other electrodes (<6 kΩ). Systematically increasing the pulse amplitude revealed that the FNS was present for all electrodes at varying loudness levels, with twitching starting at perceived ‘loud’ levels for basal electrodes, ‘medium’ for middle electrodes, and ‘soft’ for apical electrodes.
We created two new maps for S3. The first map increased the pulse duration for the higher impedance electrodes #11–14 to 100 μs, then remapped the threshold and MCL for all electrodes. This new map resulted in an immediate reduction in FNS, and the subject reported that the sound was clearer in quality but had reduced overall volume.
The second new map further increased pulse duration to 150 μs for all electrodes, decreased stimulation rate from 500 to 250 Hz, and reduced the number of peaks picked in ACE from 8 to 4. This map almost eliminated the FNS, with only infrequent, mild ‘tickling’ sensations around the left eye. However, S3 reported a subtle low-frequency echo with the second map. We further lowered the MCL for all electrodes, which eliminated both the twitch and the echo, but resulting in softer sound and poorer sound quality than the first map.
S3 was sent home with both new maps. The subject returned for a one-month follow-up favoring the first map for the better sound quality while still having some FNS. However, the FNS was no longer bothersome to the subject. Importantly, speech intelligibility with the first map was improved to 85% sentences and 93% words correct with HINT in quiet, compared with 60% sentences and 86% words with her original map.
S4 was a 57-year-old female with a congenital hearing loss due to her mother contracting Rubella during pregnancy. She had been wearing bilateral hearing aids since childhood until receiving a CI (Med-El Synchrony with SONNET processor) on the right ear in 2015. She had been wearing the CI for 3 years when she was referred to us because she was experiencing vibrations at the base of her skull and sound hypersensitivity, making her very tired and uncomfortable from CI usage in daily life. She also reported having non-bothersome tinnitus. The subject was diagnosed with hypothyroidism in 2007 and had been taking levothyroxine pills. At the time when S4 was referred to us, she was in the process of reducing her intake of this drug, which, according to her report, worsened sound quality and sensitivity. A hearing assessment revealed poor performance with speech understanding for her CI ear (25% sentences, 63% words using HINT in quiet). When using both CI and HA, her scores improved (90% sentences, 94% words).
Systematic increase of pulse amplitude on each electrode showed that different electrodes were associated with the skull vibration and hypersensitivity. On the one hand, the skull vibration was isolated to the basal electrodes (#8–12), with the strongest vibration being experienced with electrodes #11–12. Because these two electrodes produced the vibration at all levels, even with the extended pulse duration of 150 μs, they were deactivated. We also lowered the MCL by five charge units for electrodes #8–10 to ensure only sound but no vibration sensation. On the other hand, the sound hypersensitivity was isolated to the apical electrodes (#1–5), with the most uncomfortable sensation occurring on electrode #1. This uncomfortable sensation could be minimized by lowering the MCL on all electrodes except for electrode #1, which still produced the same hypersensitivity at the lowest loudness level. As a result, electrode #1 was also deactivated; electrodes #2–5 were remapped to reduce the MCL by 5 charge units in a new map.
After one month of use, the subject responded positively to the new The vibrations were eliminated, and sounds were tolerable. She reported needing to adjust the volume and sensitivity on her device to cope with certain environments, such as the restaurant being too loud and a lecture hall being too soft. Despite these changes and adjustments, speech intelligibility remained unchanged between the old and new 30% sentences and 60% words correct with the HINT in quiet using only the CI, and 90% sentences and 96% words correct using both CI and HA.
A side note is worth mentioning. During the SARS-CoV-2 pandemic, the subject reported that her speech processor needed to be repaired. Upon receiving the repaired processor, which had only the old map, she immediately experienced all the side effects and wished for us to provide her with the new map. Following Covid-safe practices, we were able to load our map onto her device, which immediately eliminated the side effects while maintaining the same good speech performance with bimodal hearing. S4 has been active in the CI patient advocacy group and used her learning experience with us to educate other CI users to improve their device effectiveness and experience.
S5 was a 44-year-old female who suffered unilateral hearing loss and tinnitus in the left ear from an unknown cause. Despite having normal hearing thresholds (≤10 dB HL) in the right ear, she received a CI (Med-El PULSARci100 and Opus 2 processor) in her left ear to treat tinnitus in December 2010. Post-surgically, her left ear still had usable residual hearing below 750 Hz (40 dB HL at 125 Hz, 35 dB at 250 Hz, 75 dB at 500 Hz, 90 dB at 750 Hz and 95 dB at 1000 Hz). The subject was referred to us in April 2012 because the implant produced sound distortions, which she described as “echoes” that limited her CI usage in daily life. The HINT sentence intelligibility with CI alone was only 15% correct in quiet. In addition, her right, normal-hearing ear was sensitive to loud sounds.
Systematic examination of her original map showed that the two most basal electrodes (#11 and 12) were deactivated due to an incomplete insertion. Except for these two extra-cochlear electrodes, all other electrodes showed normal loudness growth with default settings. We suspected that the echoing is a result of two mismatches between acoustic and electric hearing. A mismatch between electric and acoustic hearing may result in echo-like or distorted perceptions which are oftentimes described as ‘unnatural’ or ‘fatiguing’ by some patients. These effects may worsen sound tolerance and auditory comfort. Aligning electric and acoustic hearing creates a more integrated, natural perception that may reduce these negative effects, improving sound tolerance. In the case of S5, the overlap between her residual low-frequency hearing and the CI frequency map may have caused her to hear both versions of low-frequency sounds in her left ear. In addition, we noticed that S5’s pitch perception in the right, normal-hearing ear was inconsistent with her abnormal CI pitch perception in the left ear (Zeng et al., 2014). Therefore, we increased both the low and high-cutoff of the input frequency range from 300–6000 Hz to 700–8500 Hz so that the CI would not respond to sounds lower than 700 Hz while extending the high-frequency range from 6000 to 8500 Hz. A new map was created and reported by the subject to be more natural, and importantly without any echoing.
We suggested that she try a HA in addition to the CI in the left ear and she did. A follow-up visit after a year and a half since the customized new map showed that her HINT sentence recognition increased to 45% in quiet using the CI alone and was as high as 90% using both the CI and HA (via direct connect and headphones to avoid hearing speech in the good ear). We also hypothesized that her routine usage of the CI and HA in the left ear would alleviate hyperacusis in her right ear because of restored balance in loudness between the ears. This hypothesis could not be tested because the subject did not schedule any follow-up visits.
S6 was a 51-year-old female who was first diagnosed with mild hearing loss in both ears at eight years of age. Her hearing loss progressed to moderate-to-profound at age 30, and she was fitted with bilateral hearing aids. She received a CI (Nucleus Contour Advance and Freedom speech processor) on her right ear at age 49 while continuing to wear the HA on the left ear. She had no prior tinnitus history but reported that “I developed tinnitus two days after my cochlear implant activation. It started out as just ringing in my right ear and then after about three months I developed head noise as well.” Her tinnitus handicap inventory score was 50 out of 100. Furthermore, the implant did not provide any speech intelligibility, making her reluctant to use the device.
Upon referral from S6’s surgeon, we examined her original map and found unusually narrow dynamic ranges for all electrodes (mean=10±5 SD, range=3–17 Clinical Units or CU for pulse duration of 25 μs using MP1+2 mode). Both the sensitivity and volume settings were also relatively low at 5. We suspected these settings effectively converted dynamic sounds, including speech, into relatively flat-amplitude electric stimulation, which could be less effective for tinnitus suppression (Reavis et al., 2012). Systematic examination also revealed that the most basal electrodes (#1–4) produced uncomfortable sounds that aggravated her tinnitus.
To address the narrow dynamic range issue, we increased the pulse duration from 25 μs to 150 μs and remapped the T and C levels using the ACE strategy (changed from SPEAK). This increase in pulse duration increased the dynamic range from 10 to 40 CUs for all electrodes. We also increased her sensitivity to 10 and volume to 9 to boost audibility for the soft speech sounds. To address the tinnitus issue, we reduced the basal electrode gain progressively starting with −1.5 for electrode #7 to −10 for electrode #1. We also reduced the dynamic range to 32, 24, 16 and 8 CUs for electrode #4, 3, 2 and 1, respectively. The subject liked this new map, describing it as “clearer and more natural” than the original map. Speech intelligibility with the new map was still poor immediately after 0% for HINT sentences in quiet. Importantly, she reported decreased tinnitus sensitivity to the implant. The new map was sent to her audiologist in the referral surgeon’s office. S6 was satisfied with the new map and has not asked for additional help with tinnitus issues.
S7 was a 65-year-old female who has had progressive hearing loss since childhood. She had a pulmonary issue at birth, which might result in hearing loss and a type of nerve spasm on the left side of her body. She has worn a hearing aid on her right ear most of her life but relied on her unaided left ear for telephone communication. The hearing loss in her left ear deteriorated to a profound level with 0%-word recognition; as a result, she received a CI (Cochlear CI512 and CP1000 Processor) in the left ear in 2011. However, this device failed after three months and she was reimplanted (the same implant and processor) in 2012. Both the first and second implant stimulation produced twitching on the left side of her face. She also reported unusual sound sensations after the second implantation. The sound sensations ranged from a constant tone and a soft melody to “loud, piercing, rattling” tinnitus and sometimes even “vibrations” in the head. Dizziness was often present such that she needed to lie down. The severe episodes typically lasted 10–15 minutes. A CT scan showed normal full insertion of all electrodes in the cochlea. The re-implanted device and processor also passed the integrity test. Still, the implant provided no speech intelligibility.
She was referred to us in 2014. Systematic testing and evaluation revealed multiple issues and findings. First, an impedance check (Fig. 3A) showed a shorted electrode (#3) and an open electrode (#12). Then, an electrode-by-electrode evaluation showed that stimulation from the apical half of the electrodes was responsible for left-sided facial twitches. In contrast, the basal half of the electrodes produced similar unpleasant sound sensations that the subject called “tinnitus” because they could also occur when the implant was off. During her time with us, one of the authors, Dr. Starr, a neurologist, observed that the subject had an imbalanced gait and posture. A drawing test revealed problems with hand coordination and fine motor skills for her left side (Fig. 3B). Radiological imaging further revealed a Chiari malformation, with part of her cerebellum protruding into the foramen magnum (Fig. 3C). Dr. Starr suggested that her tinnitus and dizziness are symptoms of the Chiari malformation, and probably exacerbated by the cochlear implant and facial nerve stimulation.
A preferred map was created, based on brief trial and subjective feedback from 74 maps during her extensive visits from 2014 to 2018 (Fig. 3D). The new map increased the pulse duration from 25 μs to 200 μs for all electrodes and decreased the per electrode stimulation rate from 500 Hz to 250 Hz. These two manipulations resulted in immediate elimination of the left-side facial twitches, even at perceived ‘loud’ electric stimulation sensations. To alleviate the unpleasant sound sensations caused by the basal electrodes, we reduced the number of active electrodes in this region from eleven to four (#2,4,7,10). Although the new map still produced low speech intelligibility (5% HINT words correct), she liked it for “sounding good” without any FNS. The new map did not affect her tinnitus or dizziness.
We created a separate map to manage her tinnitus (Chang & Zeng, 2012). The subject first performed a tinnitus match task, in which a fixed 1000-Hz pulse train was delivered to each of the active electrodes. Two electrodes were matched to her tinnitus. We then created a “tinnitus” map for her, in which only these two electrodes were activated over the entire frequency range. The subject could control the overall stimulation loudness via the processor’s volume control. The subject was instructed to use this specific map when her tinnitus was particularly bad. She liked the map and found that this map could reduce or even eliminate her tinnitus temporarily under certain conditions such as noisy restaurants and driving.
Over the next three years, we had attempted multiple mapping sessions, hoping to improve her speech performance with the CI or with the CI+HA while alleviating her tinnitus. Despite these adjustments, including attempts to increase active electrodes and reduce channel interactions, 67 maps were created over the three-year span, but speech performance remained poor, with tinnitus impacting her daily life. In 2018, the subject returned and informed us that she had her second CI explanted due to the poor performance and continued tinnitus. This case illustrated the limitation of cochlear implantation in treating centrally-originated side effects.
S8 was a 52-year-old female who had progressive hearing loss since she was four years old. She lost hearing completely in the right ear in her late 20s and the left ear in her late 30s. She recalled the onset of tinnitus in the right ear during middle school, which was described as a “stable, loud, high frequency” sound. She did not recall any tinnitus in her left ear. At age 28, she received a CI in the right ear (Cochlear CI24M and N7 CP1000 processor), and at age 38, a second CI in the left ear (Cochlear CI512 and N7 CP1000 processor). She developed a reactive tinnitus side effect after receiving the second CI: “I have two implants, yet I find myself using only the right one. The left implant, when turned on, makes my tinnitus in the right ear become intolerably loud – even my husband’s voice becomes nerve racking…So I have to take it off.” Her speech intelligibility with the right implant was excellent (90% correct HINT sentence-in-quiet and 87% AzBio sentence-in-quiet) but was 0% with the left implant. A CT scan showed no abnormality of cochlear structures or electrode placement in either ear.
S8 was referred to us for her left implant triggering tinnitus in the right ear in June 2022. An impedance and integrity check found no abnormality in either implant, suggesting that this cross-over tinnitus effect was not due to hardware malfunctioning. We systematically investigated this reactive tinnitus phenomenon. Unlike other cases with obvious impedance or dynamic range issues, S8’s maps appeared normal, yet her tinnitus was reactive even to low-level sounds. This prompted our investigation into the stimulus-tinnitus relationship across electrodes. First, a loudness growth function was measured on each electrode of the left implant such that four levels of stimulation were determined to produce equal loudness of 1 (barely audible), 3 (soft), 5 (medium) and 7 (loud) on each electrode (Fig. 4A). Second, for each loudness level of stimulation delivered to the left implant, the subject reported the tinnitus loudness in her right ear (Fig. 4B). The “barely audible” stimulation delivered to the left implant did not affect the tinnitus in the right ear (see the unchanged baseline tinnitus = thin blue horizontal line). However, even the “soft” stimulation started to cause an increase in her right tinnitus on more basal electrodes (>#17), with the greatest increase in loudness from 2 to 5 on electrode #10 (thick blue line). The “medium” and “loud” stimulation in the left implant monotonically increased her right tinnitus, reaching 4–6 and 5–7 on the loudness growth function for electrodes #14–1, respectively (thick dark blue and purple lines). This systematic investigation confirmed the subject’s report that stimulation of the basal electrodes of the left implant increased her right tinnitus.
Despite this initial adjustment of pulse duration, stimulation rate, and pulse amplitude, residual tinnitus persisted. Further deactivation of electrodes from #13 to #3 eliminated tinnitus. The new map contained the nine apical electrodes and the two most-basal electrodes (Fig. 4C). The idea was to preserve as many electrodes on the left implant as possible while maintaining the reactive tinnitus on the right side at a tolerable level. The gain on the two most basal electrodes (#2 and #1) was also reduced so their stimulation could provide some useful high-frequency speech cues but not cause significant reactive tinnitus. S8 reported that this particular map produced a barely noticeable increase in tinnitus from the baseline, yet it improved her speech recognition with the left implant alone from 0% to 40% correct for both the HINT and AzBio sentences in quiet. Importantly, the previously unusable left implant, when combined with the right implant, improved the speech reception threshold in noise by 3 dB when compared with the right implant alone.
Table 2 summarizes 11 strategies applied to eight subjects and their impact on side effects and speech outcomes. All strategies resulted in a decrease, if not complete elimination, of the main side effects. Six of the 11 strategies improved speech performance (S1_1, S1_2, S3_1, S5, S7_1, S8), while the rest did not decrease the speech outcome.
Facial nerve stimulation is the most common CI side effect, occurring in up to 16% of users (Berrettini et al., 2011; Venail et al., 2008). The facial nerve stimulation typically involves twitching or pain around the eye, mouth, nasolabial fold, or forehead on the same side as the implant (Kelsall et al., 1997). The fact that the same side is involved indicates a peripheral origin of this side effect, likely caused by current leakage from the problem electrodes to the facial nerve via low-impedance pathways. Two such pathways are the basal electrodes and the tympanic portion of the facial nerve (Niparko et al., 1991) or the middle electrodes and the labyrinthine portion of the facial nerve canal (Bigelow et al., 1998; Frijns et al., 2009; Vanpoucke et al., 2004). In rare cases where facial nerve stimulation occurred over the entire array, the current conduction pathways in and around the cochlea are altered, possibly due to abnormal cochlear geometries like Mondini dysplasia or more conductive vascular bone of otosclerosis (Kelsall et al., 1997; Ramsden et al., 2007). Previous and present studies showed that a combination of increased pulse duration, lowered pulse rate, and reduced number of electrodes were effective in minimizing or even eliminating the facial nerve stimulation. Experimental evidence seems to support these manipulations. For example, a long pulse duration not only decreases the amplitude required for equal sensation but also increases spatial selectivity (Grill & Mortimer, 1996). The increased spatial selectivity may also underlie triphasic pulse stimulation, which produces higher thresholds and MCLs than biphasic pulses, but minimizes facial nerve stimulation (Abdelhamed, 2019; Bahmer et al., 2017; Bahmer & Baumann, 2016). Second, lowering pulse rates may help, as differences in anatomical location and physiological properties between auditory and facial nerves result in varied sensitivities to stimulation parameters such as polarity, duration and rate (e.g., Carlyon et al., 2015; Gärtner et al., 2023). Finally, using fewer active electrodes can reduce electrical channel interaction (Tang et al., 2011) and prevent aberrant current amplitudes from stimulating the facial nerve.
Cochlear implantation is designed to restore functional hearing in individuals with severe-to-profound hearing loss. Although the implant is known to reduce tinnitus and sound sensitivity (Kloostra et al., 2019; Punte et al., 2011; Távora-Vieira et al., 2015), there is no guideline for practical management nor any theories regarding the underlying mechanism at present. Rarely would cochlear implant users develop tinnitus or experience worsening tinnitus after implantation. However, if tinnitus occurred, then the solutions are usually complex and often require customized interventions, with some patients requiring high rate stimulation and others requiring low-rate stimulation (e.g., Chang & Zeng, 2012; Rubinstein et al., 2003; Zeng et al., 2011). In the present study, we were able to reduce the tinnitus side effect by widening the electric stimulation range (S6), providing a map with cochlear stimulation mimicking the sounds of the tinnitus (S7), and eliminating the electrodes that triggered tinnitus on the contralateral ear (S8). Except for S7 with suspected cerebellum malformation, the underlying mechanisms of cochlear implantation causing reactive tinnitus are unknown, making management strategies difficult.
Speech performance is an important factor in managing CI stimulation side effects. On the one hand, poor speech intelligibility and CI side effects are often coupled. The present results showed that in many cases (Table 2), reducing or eliminating the side effects resulted in improved speech performance, which, in turn, led to increased cochlear implant usage. The increased device usage would likely train the brain to improve linguistic and cognitive skills (Lenarz et al., 2012; Moberly et al., 2016). On the other hand, some users had excellent speech performance but were bothered by CI side effects (e.g., S2). In these cases, minimizing the CI effect was needed to ensure that speech performance would not be compromised. For example, S8’s basal electrodes all had side effects, but we intentionally preserved the two most-basal electrodes with reduced gain to allow transmission of high-frequency speech information. Because each subject is unique in terms of side effects, speech performance, and expected outcomes, it is important to consider customized solutions over the standard one-size-fits-all clinical map.
A limitation of the present study is that impedance matrix measurements, which are especially useful for assessing abnormal current flow, were not available at the time of testing. Future studies may incorporate impedance matrices to improve diagnosis and management of FNS and other side effects (e.g., Saoji et al., 2024; Vanpoucke et al., 2004).
Although cochlear implants provide their users the means to perceive sounds, they can also produce undesirable side effects that can lead to discomfort, frustration, and poor outcomes. Treating these side effects is often a cumbersome, time-consuming process not routinely available in a clinical setting. Careful and detailed characterization of the side effects is the first important step for managing side effects. Depending on the side effect, effective strategies include deactivating problematic electrodes, reducing stimulation levels, increasing pulse duration, decreasing pulse rate, altering channel gain and input frequency range, and manipulating the stimulation waveform and electrode configuration. Although no one standard strategy exists for all side effects and for all subjects, clinicians could prioritize these strategies based on the characterization of the side effects to actively reduce their negative impact while maintaining or even improving speech performance.