Identifying Slim Modiolar Electrode Tip Fold-over with Intracochlear Electrocochleography

Abstract

Objective:

To evaluate the predictive value of intracochlear electrocochleography (ECochG) for identifying tip fold-over during cochlear implantation (CI) using the slim modiolar electrode (SME) array.

Study Design:

Prospective cohort study

Setting:

Tertiary referral center

Methods:

From July 2022 to June 2023, 142 patients, including adults and children, underwent intracochlear ECochG monitoring during and after SME placement. Tone-bursts were presented from 250 Hz to 2 kHz at 108 to 114 dB HL. A fast Fourier transform (FFT) allowed for frequency-specific evaluation of ECochG response. ECochG patterns during insertion and post-insertion were evaluated using sensitivity and specificity analysis to predict tip fold-over. Intraoperative plain radiographs served as a reference standard.

Results:

Fifteen tip fold-over cases occurred (10.6%) with significant ECochG response (>2μV). Sixty-one cases without tip fold-over occurred (43.0%) with significant ECochG response. All tip fold-overs had both a non-tonotopic post-insertion sweep and non-robust active insertion pattern. No patients with robust insertion or tonotopic sweep patterns had tip fold-over. Sensitivity of detecting tip fold-over when having both non-robust insertion and non-tonotopic sweep patterns was 100% (95% CI: 78.2% to 100%), specificity was 68.9% (95% CI: 55.7% to 80.1%), and the overall accuracy was 72.0% (95% CI: 60.5% to 81.7%).

Conclusion:

Intracochlear ECochG monitoring during cochlear implantation with the SME can be a valuable tool for identifying properly positioned electrode arrays. In cases where ECochG patterns are non-robust on insertion and non-tonotopic for electrode sweeps, there may be a concern for tip fold-over, and intraoperative imaging is necessary to confirm proper insertion.

Introduction

Minimizing insertion trauma and improving tonotopic selectivity are critical factors that have informed electrode array design in modern cochlear implants (CI) ^1,2. The slim modiolar electrode (SME) is a pre-curved array designed for a perimodiolar position³. While this design may improve rates of hearing preservation and post-operative speech perception^4,5, this flexible internal array has resulted in increased rates of tip fold-over.

Tip fold-over occurs when the apical end of the electrode array fails to properly deploy and instead folds on itself within the cochlea. Figure 1 shows plain films of the SME after a tip fold-over and subsequent correction with proper insertion. This complication necessitates reinsertion and potential need for backup devices. Tip fold-over may also increase cochlear trauma which negatively affects CI performance⁶. Tip fold-over occurs in 4–10% of perimodiolar electrodes^7–9, and is identified primarily through intraoperative plain radiographs^10–12. While plain radiographs are the gold standard for identifying tip fold-over, they are not real-time feedback for the surgeon, increase operative time, and contribute to radiation exposure for both the patient and the medical personnel.

Figure 1. Plain films of implanted slim modiolar electrode (SME).

Dots are superimposed on each electrode. Left, Tip fold-over with acute turn at the apex and incomplete modiolar wrap. Right, Reinserted SME in appropriate perimodiolar position.

Electrocochleography (ECochG) is a tool that uses acoustically evoked electrical potentials to quantify residual function of the cochlea and auditory nerve^13,14. While ECochG has historically been used to diagnose inner ear pathology like Meniere’s disease^15,16, recent work has used frequency-specific tone-burst stimuli to characterize cochlear electrophysiology in CI recipients^17,18. ECochG monitoring can provide direct and rapid feedback to the surgeon on the status of the CI within the cochlea^19,20. Intracochlear ECochG using the SME electrode array has shown utility as a tool for hearing preservation, prognosticating CI speech perception outcomes, and understanding the in vivo, tonotopic place map of the cochlea^21–23.

Only recently has the use of ECochG been explored for the detection of tip fold-over ²⁴. Using extracochlear ECochG, heterogenous responses were demonstrated in 5 patients with tip fold-over and the recommendation of these investigators was for future work utilizing intracochlear ECochG.

The primary hypothesis of this study was that intraoperative ECochG monitoring can demonstrate distinct patterns of response that can effectively identify tip fold-over of the SME. By providing more immediate feedback to the surgical team, ECochG may offer a more efficient alternative to intraoperative imaging, thereby potentially reducing delays, costs, and intraoperative radiographs.

Methods

Study Population & Inclusion Criteria

This prospective study involved the use of ECochG monitoring in both adult and pediatric patients who underwent cochlear implantation with the SME (CI632; Cochlear Corporation, Sydney, Australia). Patients with malformed cochlear anatomy or obstruction, those without a patent external auditory canal, and those undergoing revision CI were excluded from the study. Informed consent was obtained preoperatively from all participants. The research protocol was approved by the Institutional Review Board at Washington University in St. Louis (#202007087).

Surgical Technique & ECochG Setup

A standard surgical approach was used for all implantations with either round window or extended round window cochlear openings. After induction of anesthesia and patient positioning, an ER3–14A foam earplug connected to an ER3–21 sound delivery tube was placed in the external auditory canal (Etymotic, Elk Grove Village, IL). The Nucleus CP910 sound processor (Cochlear Corporation, Sydney, Australia) was used to generate acoustic stimuli. After routine surgical draping to exclude the sound delivery system, a transmastoid facial recess approach was used to access the cochlea. Following adequate exposure of the round window (RW), the internal receiver stimulator component of the CI was immobilized under the pericranium/temporalis muscle region. Using a sterile ultrasound drape, the telemetry coil was aligned with the receiver stimulator. Measurements along the SME array were obtained with the Cochlear Research Platform version 1.2 (Cochlear Corporation, Sydney, Australia). All ECochG recordings were made with tone-burst stimuli alternating in rarefaction and condensation phases (30 averages; rise/fall time, 3 ms; recording epoch, 19 ms; sampling rate, 20 kHz).

Multifrequency insertion ECochG protocol

The SME was inserted according to the manufacturer’s recommendation, and multifrequency insertion monitoring was actively recorded from the most apical electrode (ICE22) as previously described²³. Sound delivery alternated between 250 Hz and 500 Hz delivered at 112 dB HL and 108 dB HL respectively per the maximum system output. The initial 250 Hz ECochG amplitude, as the SME was positioned just within scala tympani at the start of insertion, was the starting response. The final response was the amplitude when the SME was fully inserted.

Significant response was defined as an initial 250 Hz ECochG FFT amplitude >2 μV. Below this threshold, the noise floor from the operating room was not distinguishable from cochlear response. Additionally, there were cases when the initial response was large (>100 μV). This led to saturation of the cochlear response (at this stimulus intensity) with resulting obscuring of response patterns. An initial 250 Hz ECochG FFT amplitude between 2–100 μV was identified for pattern recognition which accounted for 70% of all cases with measurable ECochG response.

The topographic arrangement of frequency within the cochlea, also known as tonotopy, has been foundational for understanding the pattern of cochlear activation to broad ranges of acoustic stimuli^25,26. The cochlear location that is optimally responsive to a specific frequency at high stimulus is the “best frequency” (BF) ²⁷. ECochG patterns are strongest at the electrode closest to the BF; the 250 Hz BF location is apical to a fully-inserted SME in essentially all cases²³. Therefore, we would anticipate that during insertion as the distal electrode in the SME is advanced apically within the cochlea towards the 250 Hz BF, the 250 Hz ECochG response should have a robust rise in amplitude that is maximal at full insertion²⁸. When this does not occur, it raises concern that the distal end of the SME has folded on itself.

With each cochlea varying in ECochG responses, a ratio of the final 250 Hz ECochG response over the starting response was evaluated during insertion monitoring. The optimal ratio for identifying tip fold-over was determined based on a receiver operating characteristic analysis^29,30. Three times greater amplitude in final response relative to the initial response was defined as the cutoff above which the ECochG response was considered a robust insertion pattern. For example, an initial 4 μV response at the most basal aspect of scala tympani should increase to at least 12 μV response by full insertion. When this response is limited, this indicates a non-robust insertion pattern. Figure 2 shows examples of these multifrequency insertion patterns.

Figure 2. Multifrequency insertion electrocochleography patterns.

Recording time on x-axis, and fast Fourier transformation (FFT) amplitude on y-axis. A, Robust insertion during a proper insertion. B, Non-robust insertion during a tip fold-over.

Although 500 Hz ECochG was simultaneously monitored, based on cochlear size and insertion depth, the most apical electrode may or may not pass the BF for 500 Hz even with proper insertion²³. This leads to variability, and often an expected drop, in the signal for 500 Hz at full insertion. For these reasons, 500 Hz ECochG signal amplitude during insertion was not considered a consistent predictor of tip fold-over.

Post-insertion ECochG protocol

ECochG recorded along the electrode array has shown responses that match tonotopic expectations²⁷. Following insertion, the electrodes along the array were conditioned and recordings were made across every other internal electrode contact (i.e., ICE22, ICE20, ICE18, etc.). Sequential tone-burst stimuli were delivered at specific frequencies (250, 500, 1000, and 2000 Hz) between 108 to 114 dB HL per maximum system output. This sweep across the electrodes of the SME allowed for assessment of how ECochG response varied throughout the array at each frequency. These “electrode sweeps” acted as an electrophysiologic surrogate for evaluating how the CI was positioned within the cochlea. The BF electrode for each stimulus was determined across the array and used as a surrogate for tonotopy. An expected tonotopic sweep pattern would have the BF electrode of 250 Hz apical to the BF electrode for 500 Hz, the BF electrode for 500 Hz apical to the BF electrode for 1000 Hz, and so on.

To determine a non-tonotopic electrode sweep pattern, we established the following criteria: 1) the BF electrode for 250 Hz stimulus was not located at the apical end of the of the array (ICE22), 2) as frequency increases, the BF electrode for each stimulus did not progress basally as would be tonotopically expected, as described above. Figure 3 shows examples of these post-insertion electrode sweep patterns.

Figure 3. Post-insertion electrode sweep electrocochleography patterns.

For each frequency, internal electrode on x-axis, and fast Fourier transformation (FFT) amplitude on y-axis. A, Tonotopic sweep during a proper insertion. B, Non-tonotopic sweep during a tip fold-over.

After ECochG monitoring, intraoperative plain radiographs were then completed to confirm post-insertion electrode position for all cases. In cases of tip fold-over, this necessitated immediate removal of the device with attempts at reimplantation or use of a backup device with an alternate electrode array.

ECochG signal analysis

During implantation, real-time feedback from ECochG recordings was provided from spectral analysis within the Cochlear Research Platform version 1.2 (Cochlear Corporation, Sydney, Australia). For comparative analysis between subjects, the ECochG responses were stored as averaged condensation and rarefaction phases and processed offline. To analyze the ongoing portion of each frequency response, we used a fast Fourier transform (FFT) in MATLAB R2023a (MathWorks Corp., Natlick, MA) with custom scripts. A significant response was defined as exceeding three times the standard deviation of the noise floor, based on established standards from prior studies^28,31. The noise floor was confirmed to be ~0.1 μV.

Statistical Analysis

Analysis was performed with SPSS version 29 for Windows (IBM Corp., Armonk, NY). We performed normality tests via histogram visualization and Shapiro-Wilk tests. Sensitivity and specificity analyses were performed with test positive defined as an abnormal ECochG pattern (non-robust insertion and/or non-tonotopic sweep) and disease positive defined as tip fold-over. Alpha level for all statistical tests was set at 0.05 and all tests were two-tailed.

Sample Size Calculation

A 10% prevalence of tip fold-over in the population was taken from prior literature^7–9. Given no prior work evaluating intracochlear ECochG to predict tip fold-over, effect size was based primarily on clinical judgment. Based on our experience with hundreds of ECochG monitored cases, the sensitivity for abnormal ECochG patterns in predicting tip fold-over was estimated to be 98%. With a type I error of 5% and a marginal error of 10%, an estimated sample size was calculated to be 75 patients³².

Results

Patient Demographics

Fifteen (10.6%) tip fold-over cases, confirmed by intraoperative plain film, were monitored with significant ECochG response (i.e., 2–100 μV amplitude on initial 250 Hz ECochG FFT during insertion monitoring). Sixty-one (43.6%) cases without a radiographically confirmed tip fold-over and with significant ECochG response were also consecutively collected over the same time. Sixty-six (46.5%) cases either had no measurable response or had ECochG responses that did not satisfy the parameters as previously described. Demographic and audiologic data are available in Table 1.

Table 1.

Baseline characteristics of subjects

	Tip fold-over (n=15)	No tip fold-over (n=61)
Laterality, n (%)   Left	7 (46.7)	34 (55.7)
Right	8 (53.3)	27 (44.3)
Age at implant, mean (SD)   Years	64.9 (18.4)	66.1 (21.2)
Sex, n (%)   Female	7 (46.7)	26 (42.6)
Male	8 (53.3)	35 (57.4)
Etiology of hearing loss, n (%)   Congenital	1 (6.7)	5 (8.2)
Idiopathic	4 (26.7)	6 (9.8)
Infectious	1 (6.7)	3 (4.9)
Meniere’s disease	2 (13.3)	4 (6.6)
Meningitis	0 (0)	1 (1.6)
Medication/radiation therapy	0 (0)	4 (6.6)
Noise-induced	1 (6.7)	9 (14.8)
Presbycusis	5 (33.3)	16 (26.2)
Sudden sensorineural	1 (6.7)	13 (21.3)
Duration of hearing loss, mean (SD)   Years	26.0 (14.1)	22.5 (18.6)
Duration of severe profound hearing loss, mean (SD)   Years	5.7 (7.3)	6.3 (7.0)

ECochG pattern specificity and sensitivity analysis

See Tables 2, 3, and 4 for full breakdown of ECochG patterns and tip fold-over. A non-robust insertion pattern was observed in all tip fold-over cases and in 36 non-tip fold-over cases (59.0%). The sensitivity of a non-robust insertion pattern to predict tip fold-over was 100% (95% confidence interval (CI): 78.2% to 100%), the specificity was 41.0% (95% CI: 28.6% to 54.3%), the positive predictive value (PPV) was 15.8% (95% CI: 13.3% to 18.9%), the negative predictive value (NPV) was 100% (95% CI: 86.3% to 100%), and the overall accuracy was 46.9% (95% CI: 35.3% to 58.7%).

Table 2.

Contingency table of non-robust insertion pattern with tip fold-over

		Tip fold-over				Total
		Yes		No
		N	%*	N	%*	N	%*
Non-robust insertion	Yes	15	100%	36	59.0%	51	67.1%
	No	0	0.%	25	41.0%	25	32.9%
Total		15	100%	61	100%	76	100%

^*Percentages based on total in each column.

Table 3.

Contingency table of non-tonotopic sweep pattern with tip fold-over

		Tip fold-over				Total
		Yes		No
		N	%*	N	%*	N	%*
Non-tonotopic sweep	Yes	15	100%	25	41.0%	40	52.6%
	No	0	0%	36	59.0%	36	47.4%
Total		15	100%	61	100%	76	100%

^*Percentages based on total in each column.

Table 4.

Contingency table of both non-robust and non-tonotopic patterns with tip fold-over

		Tip fold-over				Total
		Yes		No
		N	%*	N	%*	N	%*
Both non-robust and non-tonotopic pattern	Yes	15	100%	19	31.1%	34	44.7%
	No	0	0%	42	68.9%	42	55.3%
Total		15	100%	61	100%	76	100%

^*Percentages based on total in each column.

A non-tonotopic sweep pattern was observed in all tip fold-over cases and in 25 non-tip fold-over cases (41.0%). The sensitivity of a non-tonotopic sweep pattern to predict tip fold-over was 100% (95% CI: 78.2% to 100%), the specificity was 59.0% (95% CI: 45.7% to 71.5%), the PPV was 21.3% (95% CI: 16.7 to 26.8%), the NPV was 100% (95% CI: 90.3% to 100%), and the overall accuracy was 63.1% (95% CI: 51.3% to 73.9%).

The presence of both non-robust insertion pattern and non-tonotopic sweep pattern was observed in all tip fold-over cases and in 19 non-tip fold-over cases (31.1%). The sensitivity of having both a non-robust insertion pattern and non-tonotopic sweep pattern to predict tip fold-over was 100% (95% CI: 78.2% to 100%), the specificity was 68.9% (95% CI: 55.7% to 80.1%), the PPV was 26.3% (95% CI: 19.7% to 34.1%), the NPV was 100% (95% CI: 91.6% to 100%), and the overall accuracy was 72.0% (95% CI: 60.5% to 81.7%). There were no tip fold-over events identified when the ECochG patterns were either robust during insertion and/or tonotopic for post-insertion electrode sweeps.

Adverse Event Reporting

There were no adverse events related to ECochG monitoring identified during this study.

Discussion

Despite advances in CI design, tip fold-over remains a real concern that must be identified intraoperatively. In this study, we assessed the predictive value of ECochG monitoring for identifying tip fold-over during implantation with the SME array for both pediatric and adult patients. To our knowledge, this is the first study to evaluate the relationship with intracochlear ECochG responses and tip fold-over.

The results of our study demonstrate that when ECochG insertion patterns are robust and the electrode sweep reveals a tonotopic configuration, there were no cases of tip fold-over. By contrast, a non-robust insertion pattern and non-tonotopic electrode sweep pattern should raise the suspicion for a tip fold-over. These findings have important clinical implications, as they offer a potential means to identify and prevent tip fold-over intraoperatively, ultimately improving outcomes for CI recipients⁶. Furthermore, these findings contribute to the growing body of literature on the use of ECochG monitoring during CI implantation, highlighting its potential value as a clinical tool for optimizing CI placement.

Non-robust and non-tonotopic ECochG with a properly inserted SME

The cause for a non-robust insertion pattern and non-tonotopic sweep pattern in cases of a well-positioned array on post-insertion x-ray remain speculative. Possible explanations include cochlear trauma during insertion (e.g., inadequate wrap along the modiolus, basilar membrane contact, or scalar translocation not well appreciated on plain film), or underlying pathology that results in the cochlea not behaving in a tonotopic pattern^33–35. While these abnormal responses were seen in well-positioned cases, the suspicion for a tip fold-over should still be raised in these cases and intraoperative imaging is needed to confirm placement of the SME.

While intracochlear ECochG provides valuable information on the tonotopic organization of the cochlea, patients undergoing CI can have inner ear pathology resulting in ECochG non-responsiveness above the noise floor^24,36. This variability may explain why non-robust and non-tonotopic ECochG patterns can be observed even when the internal array is well-positioned on post-insertion imaging. Further studies are needed to better understand the pathophysiology of tonotopic changes in sensorineural hearing loss and to characterize the different patterns of intracochlear ECochG responses.

Robust and tonotopic ECochG: Well-positioned array

Comparing ECochG patterns between the cohorts revealed that every case with both robust insertion response and tonotopic electrode sweeps was well-positioned. While abnormal ECochG had a poor PPV for tip fold-over, robust and tonotopic ECochG had perfect sensitivity and NPV to confirm a properly positioned SME. This finding challenges the initial study hypothesis and suggests that appropriate ECochG can accurately identify a properly positioned electrode within a responsive cochlea, even though abnormal ECochG cannot always predict tip fold-over. As criteria for cochlear implants broaden to include more patients with residual hearing, there will likely be increasing number of patients who have meaningful ECochG responses that fit this “robust and tonotopic” pattern. ³⁷. Although one could argue that cases with robust insertion and tonotopic sweep ECochG response do not need confirmatory imaging (since the chance of tip fold-over is minimal) more data should be collected before reaching such a conclusion. Therefore, ECochG should remain a valuable adjunct to the gold-standard of intraoperative imaging.

Limitations

Despite the relatively low barrier for implementation of ECochG monitoring, it remains a tool that is not available at many CI centers. The lack of availability may limit the generalizability of our findings. Future studies that support the utility of ECochG during CI implantation may help promote wider adoption of this modality.

We acknowledge that there is a high likelihood of selection bias in this study. Patients with favorable anatomy are typically chosen for SME electrodes, while patients with atypical pathology or abnormalities may receive lateral wall or more rigid, styleted arrays. This limitation could reduce the generalizability of our findings to patients with less favorable inner ear pathology or those receiving other types of CI implants.

Even with our large cohort, the low incidence of tip fold-over and variability in ECochG can make pattern recognition challenging. Therefore, future studies involving multi-institutional data could help solidify the conclusions drawn about ECochG patterns and their association with tip fold-over.

Alternative methods of electrophysiologic CI monitoring

In addition to ECochG, there are other novel technologies that show promise in identifying tip fold-over by measuring electrophysiologic data off the SME during implantation. Transimpedance matrix (TIM) and spread of excitation (SOE) measurements have been more recently evaluated for identifying tip fold-over^38,39. The Hans et al., study included 100 patients with 4 tip fold-over events and had encouraging results showing sensitivity of 100% and specificity between 97–98% in identifying tip fold-over for both technologies. One limitation is that pattern recognition criteria were more subjectively defined without explicit cutoffs to what constituted an atypical result which raises concern for underrepresented false positive cases.

Future Directions

Intraoperative electrophysiologic monitoring during cochlear implantation continues to be identified as a useful tool. While TIM and SOE are also exciting technologies, the inability to record these measures simultaneously leaves the clinician to decide between auditory-evoked ECochG, which has been shown to have prognostic value on post-activation CI performance, or non-evoked electrical measures, which have yet to show this same benefit. Future work would ideally integrate various technologies to create a multimodal CI monitoring platform. A multi-technology approach has the potential to provide real-time feedback to the surgeon and improve clinical outcomes for all CI recipients.

Conclusion

Intracochlear ECochG monitoring during cochlear implantation with the SME can be a valuable tool in identifying properly positioned electrode arrays. In cases where ECochG patterns are non-robust on insertion and non-tonotopic for electrode sweeps, there may be a concern for tip fold-over, and intraoperative imaging is necessary to confirm proper insertion. The plain radiograph remains the gold standard of intraoperative confirmation of proper CI positioning; however, ECochG does have promise as a screening adjunct for identifying properly inserted arrays. Further work would involve multicenter studies to reinforce the utility of intraoperative ECochG during cochlear implantation.