With the approval of the regional ethics committee, patients who received SRS for VS between 2010 and 2016 were retrospectively evaluated.
Fifty-three patients met the inclusion criteria of receiving single-fraction SRS for unilateral, sporadic VS. Patients < 18 years old, with neurofibromatosis type 2, those who had undergone prior radiotherapy or microsurgery, and patients who were treated with multi-fractionation schemes were excluded from the study. Clinical and radiosurgical data were collected from patient charts and medical follow-up notes, including: age, sex, tumor location and morphology, pre-SRS growth, tumor volume, Koos grade, pre-SRS dizziness/tinnitus, target and vestibule doses (minimum, mean and maximum), post-SRS transient volume expansion (defined as volume growth followed by shrinkage to the pre-SRS volume or less), post-SRS tumor necrosis (detected by sequential magnetic resonance (MR) images) and objective neurootological examination.
All patients received treatment either with a Linac (Novalis TX, BrainLAB AG, Feldkirchen, Germany; Varian Medical Systems, Palo Alto, CA, USA) or robotic-based (Cyberknife, Accuray, Sunnyvale, California) radiosurgical system. Patients were immobilized in the supine position using a commercial stereotactic mask fixation system. Thin-sliced (0.75 mm) computed tomography (CT) images without contrast were obtained. Contrast-enhanced T1- and T2-weighted high-resolution MR images (1 mm) and, when available, three-dimensional constructive interference in steady-state (3D-CISS) sequence images were registered with CT images for target delineation. After tumor delineation by a radiation oncologist, a medical physicist planned the treatment. No additional margin to the target volume was included for the planning target volume. A single fraction of 12 Gy was prescribed for each case. SRS plans were generated and delivered using the treatment planning systems of iPlan (Novalis) and Multiplan (Cyberknife). The quality of the treatment plans was assessed by evaluating target coverage, dose heterogeneity/ conformity, and normal tissue dose tolerance, particularly cochlear dosimetry. KV imaging (ExacTrac, Brainlab) for Linac-based and real-time X-ray patient tracking (6D-Skull) for robotic-based systems were used for set-up verification and repositioning.
All patients were followed up at 6 months using MR imaging and clinical examination and again at 1 year post-SRS with additional objective neurootological tests. Long-term assessments were performed annually. Vestibular testing comprised a patient history, clinical neurotological and focused neurological examination including cranial nerves III-XII, cerebellar examination, stance tests (Romberg [standing > 20s] and Unterberger [walking 50 steps in place without deviation > 45°]). The vestibular tests were consist of video-oculography including head hanging test (Rose manoeuvre), test for spontaneous nystagmus at all gaze directions, smooth pursuit, rotatory chair examination (sinusoidal harmonic acceleration [SHA] test at 60 deg/s chair velocity and 0.02 Hz), cervical vestibular evoked myogenic potentials (cVEMP, 500-Hz tone-burst stimuli) and bithermal caloric testing. The degree of asymmetry in caloric testing was categorized in; normal (0–25%), mild hypofunction (26–50%), moderate hypofunction (51–75%) and severe hypofunction (76–100%). Clinical examination and vestibular test results were reported pre-SRS and 1-year post-SRS. The results were classified as stable/improved if the patient maintained the same function or showed better function, and as worse if the function deteriorated. Patient-reported dizziness was recorded pre-SRS and at 6 months post-SRS. At each time point, patients were asked whether they experienced any form of dizziness (at the time of the examination, or any episode) and the responses were coded as a binary outcome, yes or no.
Dosimetry of the vestibule
Based on the CT and MR images for each SRS plan, the volume of the vestibule was defined with the assistance of a senior board-certified neuroradiologist specialized in head and neck imaging (FW). Six to ten successive axial slices were used for precise volume definition. Location of the vestibule was defined as the common junction of the cochlea and semi-circular canals, where internal auditory canal separates cochlea from the vestibule (Fig. 1A). The minimum, mean, and maximum radiation doses received by the vestibule volume were obtained from the treatment planning software. Vestibule delineation was performed retrospectively and no dose constraints were given to the vestibule during plan optimization.
A certified statistician (DL) performed the statistical analysis. Duration of follow-up was defined as starting from the day of SRS. Descriptive analyses were reported as mean (standard deviation), median (range) or number (percentage). Logistic regression was used to assess variables that might influence dizziness at the specified follow-up times. Variables with a p-value < 0.1 in the univariate analyses were included in the multivariate analysis. The t-test was used to test the association between continuous variables.
A p-value of < 0.05 was set as statistically significant. Statistical analyses were performed with IBM SPSS Statistics software (version 25; IBM Corp., Armonk, NY, USA).