Use of DTT to Define Facial Nerve Position in Vestibular Schwannomas

BACKGROUND AND RATIONALE Vestibular schwannomas (VS) are peripheral nerve sheath tumours arising from the vestibulocochlear nerve, which emerges from the brainstem at the base of the brain and enters the structures of the inner ear via a small bony canal within the skull (internal auditory meatus). VS account for 8% (Johnson J et al.) of intracranial tumours and, although benign, these tumours can grow resulting in several potentially debilitating symptoms, including hearing loss, vertigo, tinnitus, facial paralysis and meningitis. Furthermore, large tumours may cause compression on the brainstem, which can, in some cases, have life-threatening consequences (Johnson J et al.). Therefore, patients diagnosed with larger tumours (>1.5cm extrameatal component) are frequently offered surgical resection (as opposed to conservative management or stereotactic radiosurgery). As surgical techniques have improved, the goal of surgery has broadened to include not only tumour removal, but also preservation of the function of the nerves involved by VS. In particular, surgical resection aims to preserve the function of the facial nerve, responsible for innervating the muscles of facial expression, and which runs alongside the vestibulocochlear nerve and is frequently displaced and flattened by VS. Given its juxtaposition, the facial nerve is at high risk of injury during surgery. Injury to the facial nerve (and consequent debilitating facial weakness or paralysis) is a serious potential risk of vestibular schwannoma (VS) surgery. However, at present, the only means of avoiding nerve injury involves a combination of meticulous surgical technique and the use of a facial nerve monitor (a device that can detect activity within the facial muscles when stimulated by a surgeon-held probe); however, despite this, facial nerve injury may still occur. Indeed, until the surgeon begins operating, the course of the facial nerve is unknown. In a large series (>1000 surgical cases) by Sampath et al., the position of the facial nerve position was found most commonly on the anterior middle portion of the tumour; however, in a smaller proportion of cases, the nerve lay in 1 of 7 additional locations (Sampath et al.). Therefore, it follows that preoperative detection and identification of the course of the facial nerve would be extremely useful to guide surgery and reduce the potential risk of neuronal injury.

There now exists a potential means of non-invasively identifying the course of the facial nerve using advanced applications of magnetic resonance imaging (MRI). This involves utilising a technique known as diffusion tensor tractography (DTT), which has the capacity to detect the direction (vector) of diffusivity in a particular volume of tissue (voxel) (Mukherjee et al.). This data can be subsequently used to reveal structures, such as nerve fibre tracts, that are arranged along particular vectors, enabling them to be differentiated from surrounding tissue (Cauley et al.). This cannot be achieved using standard MRI sequences, in which the flattened nerve often becomes indistinguishable from the adjacent VS. Fortunately, as MRI represents the current means of diagnosing and characterising VS, adding an additional DTT sequence to a patient's scan does not represent a significant burden (only approximately 10 additional minutes of scanning time to acquire the DTT data).

Several groups (Choi et al., Taoka et al., Gerganov et al., Zhang et al. and Song et al.) have used DTT to delineate the course of the facial nerve in patients with VS and the majority have reported high success rates (>90% agreement with operative findings). However, many of these studies employed 'deterministic' DTT, which has drawbacks that limit its reliability in the setting of small fibre tracts with complex geometry. In particular, it is unable to accurately track fibres that cross, split or merge because the technique assumes that adjacent voxels have similar vectors. In order to overcome these limitations, we propose using the newer 'probabilistic' DTT technique that calculates a vector for each voxel analysed. It therefore has the capacity to account for complex geometries, such as those encountered in the facial nerve, particularly where its course is heavily distorted by tumour tissue. Zolal et al. used this technique in 21 patients with VS, resulting in an agreement rate of 81% for the position of the facial nerve and 33% for the cochlear nerve (Zolal et al.). However, there is an ongoing need for further studies in this area.

This project seeks to clarify the accuracy, reproducibility and practicality of probabilistic DTT in locating the facial nerve pre-operatively. If the technique proves accurate, it will potentially help with pre-operative planning and avoidance of facial nerve injury during VS surgery. Additionally the project seeks to clarify associated practical issues; in particular, the inter-observer/intra-observer variability (i.e. whether the reading radiologists agree with each other and themselves when analysing DTT images), whether the facial nerve position can be partially inferred on conventional sequences (comparing this with DTT) and whether adjacent nerves (such as the vestibulocochlear nerve) can be identified, which may further help with surgical preoperative planning.

HYPOTHESIS:

It is hypothesised that preoperative advanced probabilistic diffusion tensor tractography (DTT) will accurately demonstrate the position of the facial nerve relative to extrameatal portion of a vestibular schwannoma. It is proposed that the accuracy will be superior to previous studies in which deterministic DTT has been used to demonstrate the position of the facial nerve. This pre-operative mapping has the potential to decrease the rates of facial nerve injury during vestibular schwannoma surgery and allow for increased resection of the tumour.

STUDY DESIGN:

Prospective cohort observational study.

For study eligibility and outcome measures, please see the relevant sections.

IMAGE ANALYSIS

MRI In addition to standard imaging assessment, further image analysis will be performed offline using MRTrix 3.0 (www.mrtrix.org) in tensor probabililty mode. The presence of one or more cranial nerve tracts will be determined by their consistent depiction following their systematic interrogation with varying anisotropy thresholds. There will be a qualitative assessment of cranial nerve position will be achieved by dividing the circumference of the extrameatal vestibular schwannoma into segments.The assessment will be performed by two observers.

Reference standard:

The position of the facial nerve with respect to the extrameatal portion of the vestibular schwannoma will be assessed at surgery will be achieved by dividing the circumference of the extrameatal vestibular schwannoma into multiple segments in the same manner as above. This assessment will be made independently of the results of the pre-operative probabilistic DTT (the neurosurgeon recording the data will be blinded to the DTT findings).

STATISTICAL CONSIDERATIONS

Sample size calculation:

Power calculation based upon a sample size of 32, assuming a true kappa coefficient of 0.9; the 95% confidence interval will have a width of 0.26 (0.77 - 1.03). The number (32) also reflects the number of cases that we expect to recruit over a 2-3 year period.