Iatrogenic inner ear dehiscence associated with lateral skull base surgery: a systematic analysis of drilling injuries and their causal factors

Abstract

Purpose

Drilling injuries of the inner ear are an underreported complication of lateral skull base (LSB) surgery. Inner ear breaches can cause hearing loss, vestibular dysfunction, and third window phenomenon. This study aims to elucidate primary factors causing iatrogenic inner ear dehiscences (IED) in 9 patients who presented to a tertiary care center with postoperative symptoms of IED following LSB surgery for vestibular schwannoma, endolymphatic sac tumor, Meniere’s disease, paraganglioma jugulare, and vagal schwannoma.

Methods

Utilizing 3D Slicer image processing software, geometric and volumetric analysis was applied to both preoperative and postoperative imaging to identify causal factors iatrogenic inner ear breaches. Segmentation analyses, craniotomy analyses, and drilling trajectory analyses were performed. Cases of retrosigmoid approaches for vestibular schwannoma resection were compared to matched controls.

Results

Excessive lateral drilling and breach of a single inner ear structure occurred in 3 cases undergoing transjugular (n=2) and transmastoid (n=1) approaches. Inadequate drilling trajectory breaching ≥1 inner ear structure occurred in 6 cases undergoing retrosigmoid (n=4), transmastoid (n=1), and middle cranial fossa approaches (n=1). In retrosigmoid approaches the 2-cm visualization window and craniotomy limits did not provide drilling angles to the entire tumor without causing IED in comparison to matched controls.

Conclusions

Inappropriate drill depth, errant lateral drilling, inadequate drill trajectory, or a combination of these led to iatrogenic IED. Image-based segmentation, individualized 3D anatomical model generation, and geometric and volumetric analyses can optimize operative plans and possibly reduce inner ear breaches from lateral skull base surgery.

Introduction

Lateral skull base (LSB) approaches are commonly used to surgically access pathologies of the posterior fossa, the temporal bone, cerebellopontine angle (CPA), and jugular foramen. Reaching these areas requires careful and precise drilling through the temporal bone in close proximity to delicate inner ear structures including the auditory and vestibular systems, the facial nerve and vessels such as the sigmoid sinus, jugular bulb, and carotid artery. Surgeons must rely upon their expert understanding of temporal bone anatomy to guide their drilling approach in order to preserve the integrity of these structures. The vast degree of natural variation in patient anatomy coupled with possible distortion from the primary disease or prior skull base surgery can further complicate the anatomical relationship of these structures.

Recently, attention has been drawn to symptomatic effects of drilling injuries to the middle ear, or iatrogenic inner ear dehiscences (IED).[2, 9, 13] Inner ear dehiscences can result in hearing loss, vestibular dysfunction, and 3^rd window phenomena such as Tullio phenomenon (vertigo or nystagmus induced by loud sounds), Hennebert sign (vertigo or nystagmus induced by pressure changes in the external auditory canal), hypersensitivity to auditory stimuli, and autophony, among others.[12] Inner ear breaches from lateral skull base surgeries were initially reported in the 1990s, with incidences of inner ear injury ranging from 10 to 30% of patients undergoing a retrosigmoid approach for vestibular schwannoma resection. [3, 7, 24, 26] Over the past 30 years, the incidence of inner ear injury is expected to decrease with recent advances in imaging technology, surgical technique, and intraoperative navigation. Recent studies reporting surgical complications from lateral skull base approaches rarely mention the integrity of inner ear structures,[1, 20] and so the current incidence of IED remains unknown.[9] Thus, a systematic analysis of the factors that lead to these surgical complications was undertaken.

Accordingly, using patient specific anatomical segmentation, 3D visualization, and geometric analysis, this study aims to elucidate why iatrogenic IED complications may occur. Our investigation centered on differentiating drilling injuries from drilling trajectory from errant lateral or deep strikes of the drill and aims to present currently available strategies to help prevent inner ear breaches and other drilling injuries during surgery.

Materials and methods

Patient selection

Formal approval of the study was provided by our institution’s Institutional Review Board (#2017P002162), and the study was performed in accordance with HIPAA. A search was conducted using our healthcare system’s database for cases of vestibular schwannoma, glomus jugulare tumor, and endolymphatic sac tumor procedures performed by multiple providers at three major tertiary academic medical centers from 2005 to 2018. Inclusion criteria included patients undergoing hearing preservation approaches but who ultimately had a symptomatic complication of inner ear dehiscence as determined by neurotology and neuroradiology physicians reviewing postoperative Computer Tomography (CT). Demographic information obtained included age, gender, signs and symptoms, intervention, pathology, and post-operative evaluation. When available, follow-up information was collected as well. Due to size of the inner ear structures, only CT images with slice thickness <1.0mm were acceptable for 3D segmentation and analysis, and patients with inadequate CT imaging were excluded, as were patients undergoing translabyrinthine approaches due to the inherent breach of the inner ear utilized in this approach.

Image processing

Image registration, segmentation, and individualized 3D anatomic models were generated for each patient from DICOM data using 3D Slicer, a free, open source image processing software (USA, version 5.0.3, https://www.slicer.org/).[11] Preoperative CT, Magnetic Resonance Imaging (MRI), and postoperative CT images were loaded into 3D Slicer. The inner ear structures were segmented using semi-automatic image intensity-based methods and were confirmed by a neurotologist and neuroradiologist. Structures segmented included the cochlea, vestibule, superior semicircular canal (SSCC), lateral semicircular canal (LSCC), posterior semicircular canal (PSCC), endolymphatic sac (ELS), otic capsule, facial nerve, malleus, incus, stapes, sigmoid sinus (SS), and internal carotid artery (Fig. 1). The limits of the otic capsule were determined by extending the margins of the inner ear structures by 1.0mm as a radiological buffer. Segmentation was similarly conducted on tumors when applicable. In the cases of vestibular schwannoma, tumor volumes were divided into cisternal and canalicular components.

Fig. 1.

(A) 3D mesh model of preoperative CT: 1-mm radiological buffer (blue) around vestibular structures (yellow). Tumor (star) superimposed from preoperative MRI. (B) 3D mesh model of post-operative CT showing posterior semicircular canal, superior semicircular canal, common crus, and vestibule (in yellow) breached by drilling (red arrow). Additional structures displayed include carotid (labeled), sigmoid sinus (SS), facial nerve (burnt orange), ossicles (magenta), and retrosigmoid craniotomy (labeled)

Preoperative CT and MRI scans were registered to post-operative CT using mutual information-based rigid registration and implemented using the BrainsFit module in 3D Slicer.[22] Landmarks were matched on the preoperative CT and MRI, and postoperative CT images to compute the target registration error (TRE).

Trajectory analysis

3D mesh models were generated in 3D Slicer software and were presented to the neuroradiologist and neurotologic surgeon to view from multiple angles and manipulate in space. The elements of the surgical trajectory leading to dehiscence were identified. Linear vectors and geometric analysis were completed on 3D models generated from postoperative images to formally assess the drilling trajectories (Fig. 2).

Fig. 2.

(A) Postoperative axial CT demonstrating breach (arrow) of the inner ear labyrinth (yellow). The vestibular schwannoma (VS) was segmented from preoperative MRI and superimposed. (B) Drilling trajectory analysis as shown on 2D axial slice. Sigmoid sinus (SS) and 1-mm border of otic capsule (blue). Cisternal and intracanalicular tumor separated by Line A. Line B measures 2cm from the posterior border of the sigmoid sinus. Volume of tumor obscured behind otic capsule from limits of craniotomy (line D) was 38.5mm³ (star), whereas a drilling trajectory from the 2-cm viewing window (line C) would have obscured an additional 17.5mm³ of tumor. Utilizing trajectories from this craniotomy would not have allowed for line-of-sight exposure of the tumor portion extending deep into the IAC

Due to the popularity of retrosigmoid approaches to the internal auditory canal (IAC) and since these cases represented most cases with IED in our cohort (4/9 patients), a more detailed analysis was conducted. These cases were compared to controls matched by tumor size, tumor depth in the IAC, and patient demographics. Electronic fiducial markers were placed on the postoperative 3D model 2cm from the posterior border of the sigmoid sinus and at the posterior edge of the craniotomy to determine optimal line of sight vectors to the tumor from these landmarks. From these vectors, a line was drawn to the tumor tangent to the edge of the otic capsule to establish the maximum tumor volumes accessible within the IAC under “safe” conditions, leaving the otic capsule intact. Tumor volumes remaining obscured behind the otic capsule under these parameters were compared to controls.

Results

A total of nine patients met the inclusion criteria. Five patients were female with a mean age of 50 (median 55, range 22–67). All patients underwent 3D segmentation analysis for IED from LSB surgery postoperatively. Indications for surgery were vestibular schwannoma resection (n=5), vagal schwannoma resection (n=1), paraganglioma jugulare resection (n=1), endolymphatic sac tumor resection (n=1), and endolymphatic sac decompression for refractory Meniere’s disease (n=1). Surgical approaches were retrosigmoid (n=4), transjugular (n=2), transmastoid (n=1), and other transtemporal (n=2). Inner ear dehiscences consisted of PSCC (n=6), SSCC (n=4), common crus (n=4), vestibule (n=4), and cochlea (n=2). Mean tumor volume was 3.59cc (median: 1.77cc, range 0.14cc–15.34cc). A summary of cases is displayed in Table 1.

Table 1.

Cases analyzed for iatrogenic dehiscence. superior semicircular canal (SSCC), posterior semicircular canal (PSCC)

Case	Age	Ear	Pathology	Approach	Structure breached	Total tumor volume	Cause for IED
1	40, M	R	Vestibular schwannoma	Retrosigmoid	SSCC, PSCC, common crus	3088 mm3	Drilling trajectory
2	61, M	L	Vestibular schwannoma	Retrosigmoid	SSCC, PSCC, common crus, vestibule	136 mm3	Drilling trajectory
3	67, M	L	Vestibular schwannoma	Retrosigmoid	SSCC, PSCC, common crus, vestibule	1772 mm3	Drilling trajectory
4	60, F	R	Vestibular schwannoma	Retrosigmoid	SSCC, PSCC, common crus, vestibule	194 mm3	Drilling trajectory
5	57, M	L	Vestibular schwannoma	Transtemporal (middle cranial fossa)	SSCC, vestibule	**	Drilling trajectory
6	49, F	L	Meniere’s disease	Transmastoid	PSCC	---	Drilling trajectory
7	36, F	L	Endolymphatic sac tumor	Transtemporal (combined transmastoid, middle cranial fossa)	PSCC	3335 mm3	Lateral drilling
8	22, F	L	Paraganglioma jugulare	Transjugular	Cochlea	1257 mm3	Lateral drilling
9	55, F	L	Vagal schwannoma	Transjugular	Cochlea	15,345 mm3	Lateral drilling + Deep drilling

^**No preoperative imaging available to assess tumor volume

PreOp CT–PostOp CT registration average TRE for the 7 fiducial landmarks was 0.8 mm (range 0.3–1.7), and PreOp MRI–PostOp CT registration average TRE using 5 fiducial landmarks was 0.9mm (range 0.4–1.5). These TRE values are comparable to the resolution of the CT scans, which correspond to a single 3D pixel (voxel).

Causes for IED in transjugular, transmastoid, and transtemporal approaches were identified as non-ideal extent of lateral drilling (n=3), extent of drilling depth (n=1), and drilling trajectory (n=2). In the cases of dehiscence due to lateral and deep drilling, the drilling angles were appropriate for reaching the pathology safely, but ultimately the inner ear structures were traversed and breached due to excessive drilling. In the two remaining cases, the angle of drilling taken was insufficient to reach the pathology without traversing the inner ear structures. This occurred in a transmastoid approach to the posterior fossa for endolymphatic sac decompression and in a middle cranial fossa approach for vestibular schwannoma resection.

Causes for IED in all retrosigmoid approaches (n=4) were identified as drilling trajectory non-ideal to reach the lateral extent of the tumor within the IAC. In these cases, the average volume of tumor within the IAC was 170mm³. The tumor invaded into the lateral third (fundus) of the IAC in all cases. The lateral extent of the tumor was able to be safely exposed in one of the cases from the limits of the 2cm viewpoint window, and in two of the cases when drawing the vector from the posterior limit of the craniotomy.

A comparison of tumor characteristics of cases and controls can be found in Table 2. All control cases allowed for 100% exposure of the IAC tumor with a safe drilling trajectory within the 2-cm viewing window, and all craniotomies in control cases provided adequate exposure to attain the 2-cm viewing window in its entirety.

Table 2.

Controls were matched for retrosigmoid approaches for vestibular schwannoma resection according to best fit of tumor volume as well as tumor volume within the internal auditory canal (IAC) as measured in 3D Slicer. Though control cases showed similar tumor invasion into the lateral 1/3 of the IAC, the anatomical positioning of the inner ear labyrinth in relation to the tumor in these cases allowed for safe access to the lateral extent of the tumor without breaching the otic capsule

	Age	Ear	Total tumor vol	Vol IAC tumor (mm3)	% IAC invaded by tumor	Vol tumor obscured behind 2.0-cm window	Vol tumor obscured behind limit of craniotomy	Craniotomy length at level of vestibule
Case 1	40, M	R	3088 mm3	158 mm3	89%	51.55 mm3	38.5 mm3	3.8 cm
Control 1	57, F	L	2409 mm3	139 mm3	80%	0 mm3	0 mm3	3.8 cm
Case 2	61, M	L	136 mm3	136 mm3	100%	23.43 mm3	5.64 mm3	4.5 cm
Control 2	51, M	L	199 mm3	199 mm3	90%	0 mm3	0 mm3	3.9 cm
Case 3	67, M	L	1772 mm3	269 mm3	91%	28.8 mm3	28.8 mm3	3.4 cm
Control 3	47, F	L	2504 mm3	275 mm3	78%	0 mm3	0 mm3	3.6 cm
Case 4	60, F	R	194 mm3	118 mm3	78%	<5.0 mm3	0 mm3	3.0 cm
Control 4	56, M	L	399 mm3	216 mm3	83%	0 mm3	0 mm3	3.7 cm

Discussion

Since Lloyd B. Minor’s description of superior semi-circular canal dehiscence in 1998,[19] greater attention has been paid to symptomatology from third window defects in the otic capsule. Though strategies to avoid damaging the inner ear have been “drilled” into surgeons during training, these drilling injuries do occur,[2, 7, 9, 13, 24, 26] and this complication has been chronically underreported.

By viewing the operative scene in 3D, investigators can manipulate the scene to view the anatomy from multiple angles, including the operative viewpoint. Using models generated within 3D Slicer software, one can easily visualize the drilling trajectories performed at an angled approach that cannot be easily appreciated from an axial, sagittal, or coronal CT slice.

Drilling trajectory

Analysis of the drilling trajectory revealed that in 6 cases, the drilling angle was the determinative factor in causing inner ear dehiscence. In these cases, the preoperative plan and the chosen trajectory of the drill as analyzed by both 2D images and 3D reconstructions of the postoperative scans placed the inner ear structures directly in the path to the site of pathology.

The risk of recurrence for vestibular schwannoma is strongly associated with the completeness of resection,[4, 10] making it imperative that surgical planning determine a suitable path for achieving the most complete tumor resection. Retrosigmoid approaches are popular for accessing vestibular schwannomas as well as other pathologies of the CPA due to their versatility and excellent visualization of the area. [23] When resecting tumors that deeply invade the fundus of the IAC, however, the retrosigmoid approach often places the semicircular canals and vestibule directly in the line of sight.[14] The cases presented in this analysis demonstrate patients with challenging anatomy, and the surgical plan and intraoperative guidance were insufficient to protect the inner ear labyrinth from drilling injury.

The angle of approach to the tumor created by the retrosigmoid craniotomy was determined to be the major causal factor for IED. We chose to center our analysis on the intracanalicular portion of the tumor because this part of the tumor resection requires drilling of the temporal bone which places the inner ear structures at greatest risk. Though the tumor volumes within the IAC may be small in comparison to the cisternal portion, it has been shown that tumor remnants from the fundus of the IAC have a high likelihood of becoming a nidus for recurrence.[4, 5]

The high degree of variability of the angles and position of the inner ear labyrinth in relation to the IAC and the posterior fossa place some patients at greater risk for inner ear injury than others.[8, 14, 15, 21] Laine and Palva recommend that by extending the craniotomy to 2cm from the midline, one can attain adequate exposure to the fundus of the IAC regardless of the anatomical variation of the inner ear in relation to the IAC.[15] Though this maximal exposure would provide the ideal angle for reaching the fundus of the IAC, this would require a high degree of cerebellar retraction which is undesirable as this may cause traction injury to cranial nerves and ischemic injury to the cerebellum. Blevins and Jackler argue that the cerebellum, and not the craniotomy, is the limiting factor for IAC exposure.[3] From intraoperative measurements of retrosigmoid craniotomies, Blevins and Jackler concluded the average realistic view-point of the surgeon between the sigmoid sinus and average cerebellar relaxation yielded a distance of 1.5–2.0cm. With the assumption of a viewpoint of 1.5cm behind the sigmoid sinus, they argue that an average of 0.3cm (32% of the total IAC length) must be left unexposed. Our analysis, as demonstrated in Fig. 2, took two points from which to assess the feasibility of the retrosigmoid approach in resecting these tumors: one vector from the 2-cm viewpoint measurement demonstrating the maximal line of sight the surgeon can expect from non-retracted cerebellar relaxation (line C), and one vector illustrating the optimal line of sight from the posterior border of the craniotomy to demonstrate the maximal amount of tumor accessible with the use of cerebellar retraction (line D).

In each of the four cases of retrosigmoid craniotomies, three or more inner ear structures were breached, suggesting that these inner ear breaches did not occur by an errant strike of the drill, but rather due to a more pernicious drilling trajectory that did not adequately realize the inner ear architecture. Visual assessment of the models in 3D Slicer further demonstrates that these patients’ anatomy situated the vestibular structures directly in the path to the fundus of the IAC from the retrosigmoid approach. (Supplemental Video 2) In half of the cases, more than 10% of the IAC tumor component would have remained obscured even in the case of utilizing the maximal angle provided by the craniotomy, despite necessitating severe cerebellar retraction, as demonstrated by the star in Fig. 2. In one case, the maximal operative viewpoint at 2.0cm (without cerebellar retraction) would have placed 33% of the IAC tumor component obscured behind the inner ear structures. These findings suggest that even before drilling the posterior fossa, the angles of approach provided by the retrosigmoid craniotomy were not favorable for achieving total tumor exposure and attempts to reach the innermost tumor within the fundus were destined to cause the surgeon to breach the inner ear.

These findings contrast with the control cases, all of which demonstrated adequate operative viewing window to provide clear drilling access to 100% of tumor under both non-retracted cerebellar (2 cm from sigmoid sinus) conditions as well as retracted cerebellar (total dimension of the craniotomy) conditions. These data suggest that the position of the inner ear labyrinth in relation to the IAC as well as the craniotomy position can contribute to placing the vestibular labyrinth in danger of drilling injury. With careful preoperative analysis, the surgeon can recognize the limits of IAC access from a retrosigmoid approach and may be advised to opt for a different approach. If the surgeon does pursue a retrosigmoid approach for these patients with challenging anatomy, we recommend they defer line of sight exposure of the fundus of the IAC and instead utilize an endoscope and angled instruments that can allow for access to the deeper recesses of the IAC while allowing for more conservative drilling.[6, 21, 18]

When assessing the patient who sustained iatrogenic IED from a transmastoid approach, the drilling trajectory chosen was insufficient to access the posterior fossa safely and resulted in a breach of the posterior semicircular canal. (Fig. 3) With an improved surgical plan, the appropriate angle could have been directed more medially (line C), or alternatively the operative plan could have determined an approach to the anterior border of the sigmoid sinus, (line D) and then proceed with drilling in an anteromedial fashion to expose the endolymphatic sac before contacting the posterior canal.

Fig. 3.

(A) 3D mesh model of the operative perspective of a patient who underwent transmastoid approach for endolymphatic sac decompression. The red arrow highlights dehiscence of the posterior semicircular canal (yellow). (B) The chosen drilling path (Lines A and B) to reach the endolymphatic sac (star) vs alternative drilling angles (Lines C and D) that could have been taken from the limits of the chosen craniotomy to reach the posterior fossa without breaching the posterior semicircular canal

Proximity awareness

The remaining 3 cases of iatrogenic IED were determined to have been caused by inappropriate drilling depth or inappropriate lateral drilling. In these cases, it was determined that the surgeon was perhaps unaware of the proximity of the drill to the inner ear structures intraoperatively. Assessment of the anatomy in 3D Slicer suggested that the surgical plan and drilling approach were sufficient to provide adequate access to the pathology, but that the surgeon’s drill veered too deeply or too laterally, thus breaching the inner ear.

Emerging intraoperative image guidance technologies can provide a protective perimeter to segmented structures. Though this technology remains in mostly experimental use, it has been shown as a useful strategy for informing the surgeon of their proximity to sensitive organs. Studies such as Voormolen et al. have shown risk reduction in skull base surgery by providing an auditory and visual alert on a monitor to notify the surgeon when the drill tip advances within a protective perimeter of preoperatively segmented inner ear structures.[25] Other groups are incorporating augmented reality to project patient-specific anatomic models onto the patients in real time[16, 17] and can provide an intuitive, visual understanding of the drill’s relationship to the inner ear structures obscured behind opaque bony structures. The use of careful preoperative surgical planning combined with 3D image-based evaluation of the planned trajectory and/or applying emerging intraoperative guidance technology shows promise to reduce the likelihood of iatrogenic inner ear dehiscence and provide more precise, safer surgeries in the future.

In closing, we would like to acknowledge some strengths and weaknesses of this study. This study utilized widely available, free 3D Slicer software. The geometric analysis conducted in this study is limited by the accuracy of image registration and the manual alignment of segmentations between MRI and CT images, which is subject to human error. However, our average TRE reached sub-millimeter accuracy, which is less than one single voxel error, and is below the customarily reported error for intraoperative navigation technologies in otologic surgery. In addition, segmentation processes are limited to the resolution of the scans and are time-consuming, therefore possibly limited in their practicality for routine use in preoperative planning. Lastly, we acknowledge the complexity of these surgeries, coupled with the retrospective nature of this analysis, making data extrapolation from 3D segmentations to the actual surgery inherently difficult. There are many surgical variables and details to consider; however, we did our best to present the data in a most useful and objective way to better understand the nature of iatrogenic IED in the hopes of preventing their occurrence in the future.

Conclusion

Iatrogenic inner ear dehiscence is an underreported surgical complication of lateral skull base surgery potentially causing hearing loss, vestibular dysfunction, and other debilitating third window symptoms. Drilling injuries to the inner ear or vestibular labyrinth may be attributable to drilling trajectory and drill tip proximity awareness. In retrosigmoid approaches for resection of vestibular schwannomas, tumor deeply invading the lateral third of the IAC may remain obscured behind the otic capsule, increasing the risk of drilling injury in the pursuit of complete tumor exposure. Careful preoperative analysis can illuminate the limits of a chosen surgical approach, and the incorporation of an endoscope with angled instruments may improve access to areas that would be unsafe for line-of-sight drilling. Though many promising technologies are in development, many of these are in research stages and not yet applicable to general practice, we recommend a preoperative 3D representation of the patient anatomy for clear visualization of the anatomical relationships of nearby structures to provide a patient-specific surgical plan.

Abbreviations

3D

Three dimensional

CT

Computed tomography

CPA

Cerebellopontine angle

ELS

Endolymphatic sac

HIPAA

Health Insurance Portability and Accountability Act

IAC

Internal auditory canal

IED

Iatrogenic inner ear dehiscence

LSB

Lateral skull base

LSCC

Lateral semicircular canal

MRI

Magnetic resonance imaging

PSCC

Posterior semicircular canal

SSCC

Superior semicircular canal

SS

Sigmoid sinus

TRE

Target registration error

VS

Vestibular schwannoma

Biography

John Day Arkansas, USA

1. Day JD, Kellogg JX, Fukushima T, Giannotta SL (1994) Microsurgical anatomy of the inner surface of the petrous bone: neuroradiological and morphometric analysis as an adjunct to the retrosigmoid transmeatal approach. Neurosurgery 34:1003–1008

Appendix

Standards for Reporting Qualitative Research (SRQR)

checklist: