The Effect of Round Window vs Cochleostomy Surgical Approaches on Cochlear Implant Electrode Position: A Flat-Panel Computed Tomography Study

Importance  The round window insertion (RWI) and cochleostomy approaches are the 2 most common surgical techniques used in cochlear implantation (CI). However, there is no consensus on which approach is ideal for electrode array insertion, in part because visualization of intracochlear electrode position is challenging, so postoperative assessment of intracochlear electrode contact is lacking.

Objective  To measure and compare electrode array position between RWI and cochleostomy approaches for CI insertion.

Design, Setting, and Participants  Retrospective case-comparison study of 17 CI users with Med-El standard-length electrode arrays who underwent flat-panel computed tomography scans after CI surgery at a tertiary referral center. The data was analyzed in October 2015.

Exposures  Flat-panel computed tomography scans were collected between January 1 and August 31, 2013, for 22 electrode arrays. The surgical technique was identified by a combination of operative notes and imaging. Eight cochleae underwent RWI and 14 cochleae underwent cochleostomy approaches anterior and inferior to the round window.

Main Outcomes and Measures  Interscalar electrode position and electrode centroid distance to the osseous spiral lamina, lateral bony wall, and central axis of the modiolus.

Results  Nine participants were men, and 8, women; the mean age was 54.4 (range, 21-64) years. Electrode position was significantly closer to cochlear neural elements with RWI than cochleostomy approaches. Between the 2 surgical approaches, the RWI technique produced shorter distances between the electrode and the modiolus (mean difference, −0.33 [95% CI, −0.29 to −0.39] mm in the apical electrode; −1.42 [95% CI, −1.24 to −1.57] mm in the basal electrode). This difference, which was most prominent in the first third and latter third of the basal turn, decreased after the basal turn.

Conclusions and Relevance  The RWI approach was associated with an increased likelihood of perimodiolar placement. Opting to use RWI over cochleostomy approaches in CI candidates may position electrodes closer to cochlear neural substrates and minimize current spread. These findings need to be interpreted in light of the increased potential for osseous spiral lamina trauma with reduced distances between the electrode array and modiolus.

Cochlear implantation (CI) is a common surgical procedure used to restore sound perception in adults and children with severe deafness. These devices replace the function of hair cells by generating electrical impulses in response to sound stimuli. There are currently more than 324 000 people with CIs worldwide, with approximately 50 000 CI procedures occurring each year.^1,2 Successful CI outcomes are influenced by a number of intrinsic and extrinsic factors. Among the intrinsic factors, limited duration of deafness,³ high preimplant speech recognition,⁴ and postlingual status⁵ are predictive of postimplantation performance. Beyond patient characteristics, operative aspects such as electrode insertion depth,^6,7 electrode array kinking, interscalar excursions,⁸ intracochlear trauma, and distance from modiolus⁷ are known to affect audiological outcomes.

Although CI procedures were originally intended only for patients with total deafness because the procedure could damage and destroy any residual hearing, there has been a greater emphasis on surgical approaches that minimize intracochlear trauma and optimize placement of electrode contacts with respect to spiral ganglion neurons of the cochlea. The “soft” surgical approach focuses on preserving residual hearing in the cochlear apex with modification to various components of surgical technique such as blood and bony dust entry, steroid use, surgical site of insertion, perilymph leakage and suctioning, and depth of insertion.⁹

Atraumatic surgery has since been a major goal of CI surgery, with a growing trend toward round window insertion (RWI)^10-12 instead of cochleostomy approaches. However, there is no consensus on the optimal approach for CI insertion, much of which is due to a lack of literature comparing these 2 techniques. Cochleostomy approaches anterior and inferior to the round window (RW) were previously championed as the superior approach because of RW proximity to the osseous spiral lamina and membranes, intracochlear fluid dynamics, and less disruption to the cochlear aqueduct.⁹ More recently, cadaveric dissections and clinical outcome studies have suggested that RWIs might be advantageous because this surgical technique results in less traumatic insertions than the traditional cochleostomy approach.^11,12

One of the challenges with surgical insertion of CIs is the lack of visibility once the electrode array enters the cochlea, which precludes visualization of both insertion dynamics and final electrode array position within the cochlea. Newer imaging methods, such as flat-panel computed tomography (FPCT), have been proposed as a means of obtaining high-resolution images of CI electrode arrays after surgery and thus permitting assessment of final electrode array position. These methods represent an improvement over several previous imaging studies of electrode position.¹³ In this study, we used high-resolution FPCT imaging to quantify differences in CI electrode position for cochleostomy and RWI surgical approaches.

Seventeen participants with sensorineural hearing loss had previously undergone CI with 22 Med-El standard 12-electrode contact arrays (31.5 mm linear insertion length, 2.4 mm between contacts). Nine participants were men, and 8, women; the mean age was 54.4 (range, 21-64) years. A standard posterior tympanotomy approach was used for all cases. The surgical technique was identified using a combination of operative notes and computed tomography (CT) visualization (Figure 1) in the coronal oblique, sagittal oblique, and axial oblique sections. The implantation approach varied between pure RWIs and cochleostomies anterior and inferior to the RW (eTable in the Supplement and Figure 1). In this study, the term cochleostomy is defined as a separate opening into the cochlea and not an extension of the RW. The Johns Hopkins University institutional review board approved the study protocol and we obtained written informed consent from all study participants.

Flat-panel computed tomography data sets were collected between January 1 and August 31, 2013.¹⁴ The FPCT (DynaCT; Siemens) scans were performed on a flat-panel angiography system (Axiom Artis Zee; Siemens) with commercially available software (Syngo DynaCT; Siemens). Collimated 20-second head FPCT scans were taken using the following parameters: 109 kV, small focus, 200° rotation angle, and 0.4° per frame angulation step. A commercially available workstation (Leonardo DynaCT InSpace 3D Software; Siemens) was used for postprocessing. High-resolution secondary reconstructions were created using the following parameters: manually generated voxels of interest to include only the electrode array; voxel size of 0.07 to 0.08 mm with the creation of secondary reconstructions, 512 × 512 section matrix; HU and EE kernel types; and very smooth, normal, auto, and sharp image characteristics. DICOM data processing was performed using open source imaging software for Mac OSX (OsiriX; Pixmeo). The curved multiplanar reconstruction interface in OsiriX allowed for visualization and reformatting of a 3-dimensional (3D) data set. The platform provided 3 orthogonal views of the cochlea in the sagittal oblique (Figure 2A), coronal oblique (Figure 2B), and axial oblique sections. Digital composite images were produced in MATLAB R2015B from preoperation conventional multislice CT images and postoperation FPCT 3D reconstructions (Figure 3).

We conducted all image analyses in a blinded fashion with respect to insertion approaches. Any extracochlear electrodes were excluded from this study. We developed standardized steps to measure electrode centroid distance from the center axis of the modiolus, osseous spiral lamina, and lateral bony wall (the bony curvature of the cochlear canal orthogonal to the electrode of interest). First, all 3 axes were centered on an individual electrode. Next, the 3D curved multiplanar reconstruction windows were rotated and positioned to provide a coronal oblique, sagittal oblique, and axial oblique cross-sectional view of the cochlea. Using the coronal oblique and sagittal oblique windows, we identified the osseous spiral lamina. Within the sagittal oblique cross-sectional window, we recorded the bony canal’s roof and floor. The sagittal oblique and coronal oblique windows were manipulated to provide visualization of the lateral and medial edges of the bony canal. Measurements were taken between the electrode centroid and (1) the center of the modiolus, (2) the osseous spiral lamina, and (3) the lateral bony wall.

All statistical analyses were performed using SPSS Statistics, version 23 (IBM). Extreme outliers were assessed by inspection of a box plot for values greater than 3 box-lengths from the edge of the box. Shapiro-Wilk tests (P > .05) of normality and normal Q-Q plots were used to demonstrate normal distribution. Homogeneity of variances was determined using the Levene test (P > .05). A 1-way analysis of variance (ANOVA) was used to assess whether differences in measurements between RWI and cochleostomy approaches were statistically significant. When the assumption for homogeneity of variances was violated (P < .05), we used a modified version of the ANOVA for statistical analysis. A probability of P ≤ .05 was used for determining statistical significance.

We evaluated 22 FPCT data sets among 17 patients; there were 8 RWIs and 14 cochleostomy insertions. The electrode centroid distances to intracochlear landmarks (central axis of the modiolus, osseous spiral lamina, and lateral bony wall) were statistically significantly different between the 2 surgical approaches (F[1744] = 7.010, P = .008). Our findings demonstrate electrode placement to be significantly closer to the central axis of the modiolus (Figure 4A) and to the osseous spiral lamina (Figure 4B) among the RWI group than the cochleostomy group. This difference was proportionally largest in the basal turn; RWI’s most basal electrode was a mean of 1.42 (95% CI, 0.24 to 1.57) mm more medially situated than the cochleostomy’s most basal electrode (Table). Electrode 2 was a mean of 0.83 (95% CI, 0.70 to 0.96) mm closer to the central axis of the modiolus in the RWI group than the cochleostomy group. Statistically significant differences in electrode distance from the central axis of the modiolus for electrodes 7, 8, and 9 were −0.27 (95% CI, −0.17 to −0.37) mm, −0.33 (95% CI, −0.28 to −0.37) mm, and −0.33 (95% CI, −0.29 to −0.39) mm, respectively (Table). These differences between the surgical approaches were proportionally largest in the first basal turn.

Electrode distance from the osseous spiral lamina followed a similar pattern (Figure 4B) to electrode distance to the central axis of the modiolus in that electrode locations in RWIs were significantly closer to the spiral lamina than the electrodes after a cochleostomy approach. This was most prominent in the 3 most basal electrodes, 1 through 3, with a mean difference of −1.17 (95% CI, −1.51 to −1.20) mm, −0.63 (95% CI, −0.32 to −0.94) mm, and −0.44 (95% CI, −0.36 to −0.52) mm, respectively, between RWI and cochleostomy (Table). Within the basal turn, electrodes 6 and 8 also demonstrated statistically significant differences between the 2 surgical approaches; in the RWI group they were a mean of 0.54 (95% CI, 0.37 to 0.70) mm and 0.45 (95% CI, 0.36 to 0.55) mm, respectively, closer to the osseous spiral lamina than in the cochleostomy group (Table).

With the exception of the most basal electrode (mean difference of 0.39 [95% CI, 0.32 to 0.46] mm between RWI and cochleostomy), we did not observe a statistically significant difference in electrode distance from the lateral bony wall (Table and Figure 4C). The lack of significant findings may be attributed to surgical differences in cochleostomy placement and interindividual variability in bony canal diameter.

In this retrospective case-comparison study, RWIs were associated with shorter distances between electrodes and target neural elements in the basal portion of the electrode array. Between the 2 surgical approaches, mean differences were as large as 1.42 mm, and this observation persisted among electrode distances from medially located landmarks.

Ideally, CI relies on the response of surviving spiral ganglion neurons to electric stimulation by the CI. Unfortunately, a substantial fraction of electrical current does not get delivered to the modiolus because of tissue impedance.^15,16 In such cases, shorter distances between the current stimulus and the spiral ganglion may improve stimulus transmission while also reducing broad current spread,¹⁷ thereby improving spectral resolution. There is evidence that channel interaction is a limiting factor in CI performance and that reducing channel interactions improves performance scores.^18,19 Modiolar hugging electrode arrays attempt to address CI current spread and localization^20,21; these modifications were designed to reduce the distance between the stimulus and the target neural substrate, thereby increasing the likelihood of efficiently transmitting the electrical current. Unfortunately, the trade-off with modiolar hugging electrode arrays is an increased probability of inducing intracochlear damage and destroying residual hearing.^12,22

Lateral wall CIs and mid-scala CIs (ie, Advance Bionics HiFocus Mid-Scala Electrode Array) have been designed with the intention of increasing electrode distance from the modiolus to protect intracochlear structures. Although these electrodes seem to demonstrate preliminary success in acquiring proximal perimodiolar position and lateral wall location, the greater electrode distances come with a compromise of increasing the level of current required for suprathreshold stimulation.^19,23 Furthermore, the limited studies evaluating mid-scalar electrodes report episodes of through-fracturing of the interscalar partition with a cochleostomy approach but not with an RWI approach.^23-25

The results from this study suggest that Med-El standard electrode arrays (nonmodiolar hugging) achieve closer proximity to the central axis of the modiolus and the osseous spiral lamina via RWI. Nowadays, intraoperative imaging equipment is commonly equipped with flat-panel detectors (ie, Siemens Orbic 3D C-arms, Medtronic O-arms). These technological advances allow for real-time imaging in the operating room, making it more feasible to monitor electrode array passage at the time of the insertion. A prospective study evaluating the use and advantages of intraoperative FPCT units for CI may be a worthwhile undertaking.

There are important limitations to consider regarding this study. Although all of the cochleostomies were placed anterior and inferior to the RW, the study results do not account for millimeter differences in the location of the cochleostomy. We acknowledge that any variability in positioning relative to the RW could influence final intracochlear electrode position and alter the robustness of the statistical significance found in this study.

Metallic artifacts were unavoidable with FPCT scans. We used high-resolution FPCT settings and tailored reconstruction parameters to reduce electrode artifacts. The remaining metallic artifact mainly obscured the lateral bony wall in direct proximity to the electrode and did not interfere with visualization of the bony capsule or the osseous spiral lamina. We attempted to control for bloom distortion by using digital composite imaging to compare undistorted preoperative CT scans with postoperative FPCT scans (Figure 3). Ultimately, our study lacks histologic correlation because our population involves CI users and not cadaveric specimens. Previous studies directly evaluated the results of radiographic imaging (multislice CT and FPCT) in CI users with temporal bone gross histologic features²⁶ and histological microgrinding imaging.²⁷ More recent studies used FPCT to image the CI in vivo^14,28,29 with no histological confirmation. Nonetheless, the lack of histological confirmation in our study is an important limitation of this study method and should be taken into consideration when interpreting the study findings. In addition to using preoperative CT scans to control for postimplantation scans, we rotated FPCT secondary reconstructions along the modiolus axis and the basal turn axis to deduce the trajectory of the bony canal. By integrating 3D structures, we were able to reasonably predict the location of the lateral bony wall. Although the basilar membrane and cochlear ducts could not be reliably visualized from the FPCT images with the electrode array present, the osseous spiral lamina was easily visualized on the sagittal oblique axis.

In this study, we observed that differences between electrode distances to the osseous spiral lamina for RWI and cochleostomy approaches were proportionally greater than the electrode distance to the modiolus. There are plausible reasons for this observation. First, the center axis of the modiolus is not necessarily in the center of the sagittal view of the electrode array. Anatomical studies and histological sections demonstrate interindividual variability in angulation of the modiolus, and this was observed in our study.³⁰ In addition, the measurement of the distance between the electrode centroid and the osseous spiral lamina will vary depending on the location of the electrode within the bony canal. Electrodes that are located near the floor or the roof of the bony canal will result in shorter distances between the electrode and the orthogonal landmarks. Therefore, surgical differences in cochleostomy placement may confound electrode centroid measurements to the lateral bony wall, osseous spiral lamina, and the central axis of the modiolus. Finally, the 2 modes of measurement operate on different scales and a 0.5- mm difference will be influenced by the range on the y-axis. Ultimately, differences in electrode contact positions between the 2 surgical approaches are greatest in the first third and latter third of the basal turn regardless of which measurement is used (electrode centroid to central axis of the modiolus and electrode centroid to osseous spiral lamina).

Despite the aforementioned limitations, this study represents the first effort to use modern imaging techniques to perform a quantitative comparison of electrode distance to bony landmarks between RWI and cochleostomy approaches. This study provides new evidence that RWIs may reduce the distance between electrode contact and the modiolus in the basal turn of the cochlea. Further studies should examine the relationship between electrode distance and preservation of neural structures of the cochlea, as well as the clinical significance of these findings with respect to CI performance.

Various surgical approaches have been used in CI—with the 2 most common techniques being cochleostomy and RWI. In this FPCT study, RWIs were associated with shorter distances between electrodes and target neural elements, particularly in the basal portion of the electrode array. Prospective studies are needed to establish the clinical significance for both surgical insertion approaches. This study provides imaging-based evidence supporting differences in electrode proximity to neural substrates between the RWI and cochleostomy approaches in CI insertions.

Accepted for Publication: April 29, 2016.

Corresponding Author: Charles J. Limb, MD, Department of Otolaryngology–Head and Neck Surgery, University of California San Francisco School of Medicine, 2380 Sutter St, First Floor, San Francisco, CA 94115 (charles.limb@ucsf.edu).

Published Online: June 23, 2016. doi:10.1001/jamaoto.2016.1512.

Author Contributions: Ms Jiam and Dr Limb had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Jiam, Limb.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Jiam, Jiradejvong, Pearl.

Critical revision of the manuscript for important intellectual content: Jiam, Limb.

Statistical analysis: Jiam.

Obtained funding: Limb.

Administrative, technical, or material support: Jiradejvong, Pearl, Limb.

Study supervision: Limb.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Ms Jiam reports research funding from Med-El Corporation (PI, Dr Limb) and Siemens Research Corporation (PI, Dr Pearl) during the conduct of the study. Dr Pearl reports grants from Siemens Corporate Research outside the submitted work. Dr Limb reports grants from Med-El Corporation during the conduct of the study and consultant/research support from Advanced Bionics Corporation outside the submitted work. No other disclosures are reported.

Funding/Support: Funding for this study was provided by research grants from Med-El Corporation (PI, Dr Limb) and Siemens Corporate Research (PI, Dr Pearl).

Role of the Funder/Sponsor: Med-El Corporation and Siemens Corporate Research were indirectly involved in the collection of the data by funding the costs of FPCT acquisition. The 2 corporate funders were not involved in the design and conduct of the study; management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.