Abstract
Purpose
The natural history of vestibular schwannomas is poorly understood. Knowledge of growth rate and growth pattern is essential because the treatment strategy is based upon these. The purpose of this study was to determine the inter- and intraobserver variability in measuring VS size.
Materials and methods
Two consultant neuroradiologists independently made three linear measurements (d1, d2, d3) using digital MRI scans. MRI scans from 72 patients diagnosed between 2002 and 2010 with VS were obtained. These patients had a total of 223 MRI scans. d1 (medio-lateral diameter) was made perpendicular to d2. d2 was made parallel to the posterior border of the petrous ridge, and d3 was a measure of the cranio-caudal height of the tumor.
Results
Limits of Agreement ranges are larger for interobserver reliability compared to intraobserver reliability. Measurement error for all diameters (except d1, intraobserver) is greater than 2 mm. d1 measurements had the least variability and d3 measurements the highest variability, both for intra and interobserver measurements.
Conclusions
The optimal method of estimating VS size needs further investigation, and measurements need to be standardized and clearly defined. d3 seems to be the most difficult diameter to measure reliably. Interobserver measurement error for all diameters is greater than 2 mm. The current VS growth criterion of more than 1–2 mm, used to triage patients to surgery, lies within this error range, and thus is problematic as a guide for clinical practice. We therefore suggest that the growth criterion for VS be redefined.
1
Background
Unilateral vestibular schwannoma (VS) is a benign tumor arising from abnormally proliferative Schwann cells, which envelope the lateral portion of the vestibular nerve in the internal acoustic meatus . The etiology of VS is poorly understood. Genetic defects, such as chromosomal 22q deletions , and infectious causes, such as herpes virus, have been implicated in the pathogenesis of VS .
With the exception of patients with type II neurofibromatosis, no genetic predisposition for VS has been identified. Neurofibromatosis patients, in contrast to those with sporadic Schwannomas, usually develop bilateral VS.
VS typically presents with asymmetrical, high-tone sensorineural hearing loss (75% of cases) together with a discriminatory loss. Sudden deafness is the initial symptom in about 10% of cases and the remaining 15% present with disequilibrium, unilateral tinnitus, cranial nerve symptoms or Ménière’s disease-like symptoms .
The growth pattern of VS is multifaceted. VS may grow continuously or only to a certain size, followed by stagnation or even shrinkage . Growth of VS leads to gradual filling of the internal acoustic meatus and, in some cases, protrusion out of the porus. Growth occurs within the initial 5 years following diagnosis; growth occurs in a limited number of tumors and primarily in those with an extra-meatal extension . VS are for the most part slow growing tumors, but so far it has not been possible to predict the individual growth pattern. Many studies have attempted to identify predictors of tumor growth, but have failed to identify any correlation between growth and variables such as age, laterality, sex and duration of symptoms . One study found unilateral tinnitus to be predictive of tumor growth , while another has found tumor size at diagnosis to be predictive of tumor growth .
Knowledge of the VS growth rate and growth pattern is essential because the treatment strategy is based upon this . In Denmark, VS surgery is centralized and performed at only one facility; Copenhagen University Hospital. Surgical indications include tumor size larger than 15 mm, tumor growth of more than 2 mm, and cystic tumors which may display sudden and dramatic growth . If these criteria are not met, the patients are allocated to a “wait and scan” regimen in which the audiovestibular symptoms and VS growth are monitored regularly. VS growth is assessed by MRI ( Fig. 1 ).
Reporting the size of VS is a challenge, which is reflected by the multitude of different classifications currently in use (at least 20 classifications are available) . Gadolinium-enhanced MRI is the gold standard imaging investigation for tumors of the cerebellopontine angle, with a sensitivity of almost 100% and an extremely high specificity, capable of detecting tumors as small as 4 mm . The traditional method of assessing tumor size is a single (axial) linear measurement assessing the greatest diameter. This method is also the one currently used in Denmark to register growth of VS . However, several studies suggest that a more accurate way of describing VS growth is by volumetric measurement . VS tumors grow in all directions and may be shaped in different ways, ranging from spherical to lollipop-shaped to dumbbell-formed. The shape may also change as the tumor expands into the intracranial space .
The incidence of VS in Denmark has been increasing during the last four decades. The annual number of diagnosed VS has increased from 15 cases in 1976 to 123 cases in 2004, corresponding to an increase in incidence from 3.1 VS per million per year to 22.8 VS per million per year in 2004. After 2005 a slight decrease in the number of diagnosed tumors has been noted, with 105 cases in 2008 (corresponding to an incidence of 19.4 VS per million per year) .
1.1
Hypothesis
A single (axial) linear measurement may not be the best way of assessing VS tumor size and growth. Growth occurs in three dimensions and the shape of the individual tumor is unpredictable . Furthermore, other studies suggest that this way of measuring VS is subject to significant intra- and interobserver variability . One study has shown an interobserver variability of 7.4 mm (Limits of Agreement, LoA) and an intraobserver variability of 2.5 mm when measuring the same tumor . Yet, many authors consider a 1-2 mm increase in tumor size on a subsequent scan, based on two-dimensional measurements, as evidence of growth . We therefore have reason to believe, that there is significant variability in measuring the size and evaluating the growth of VS.
1.2
Objective
To determine intra- and interobserver variability for different measurements of VS size, at the Department of Neuroradiology, Aarhus University Hospital, Denmark.
1.3
Study design
Retrospective analysis of all patients with MRI verified VS. Primary outcome was the intra- and interobserver variation when measuring VS.
1.4
Population
All patients seen between 2002 and 2010, at the ENT Department, Aarhus University Hospital, Denmark, with a diagnosis of Vestibular Schwannoma were included. Patients records were identified by searching for the ICD code DD33.3B, corresponding to a vestibular schwannoma diagnosis. This population of patients includes both “wait and scan” patients, as well as patients who had VS surgery. From 2002 and onwards digital MRI images were used on a regular basis at the Department of Neuroradiology, Aarhus University Hospital.
A total of 128 patients were identified. 56 of these patients were excluded, predominantly due to having only intra-meatal tumors. Others were excluded because digital images of the MRI scans were not available, the diagnosis was withdrawn after revision of the images, the images were purely done postoperatively, and in one case because the same patient appeared twice. We also excluded patients with VS containing cystic components ( Fig. 2 ).
For detailed baseline characteristics of the patients included refer to Table 1 .
| Characteristics | |
|---|---|
| Patients | |
| Total | 72 |
| Age at 1st MRI, y | |
| Mean (SD) | 57.6 (11.33) |
| Sex, number (%) | |
| Male | 41 (56.9) |
| Female | 31 (43.1) |
| MRI scans | |
| Total | 223 |
| Average per patient (range) | 3.1 (1–8) |
| Surgery | |
| Number of patients (%) | 13 (18.1) |
1.5
Measurements
Measurements were done on T1 weighted images with gadolinium contrast (Dotarem 0.5 mmol Gd/ml, Guerbet Group).
All examinations at the Department of Neuroradiology, Aarhus University Hospital were performed using either a 3 Tesla Achieva Philips MRI or a 1.5 Tesla MAGNETOM Avanto Siemens MRI. A small number of scans were performed at other locations using different MRI scans. Seven patients had CT scans instead of MRI, but these were excluded.
Slice intervals were, using the 3 Tesla MRI, 1 mm reconstructed as 0.5 mm slices covering a 38 mm area over pori in a cranio-caudal direction, and 0.7 mm slices covering a cranio- caudal area of 50.4 mm using the 1.5 Tesla MRI.
Transversal T2 FLAIR sequence was performed at the initial MRI scan. Transversal T2 weighted sequence was also performed initially and carried out at repetitive MRI scans. When planning sequences both the coronal- and sagittal planes were included. The slices were made perpendicular to the internal ear and outer ear canal bilaterally, using the coronal plane and parallel to the lower edge of Corpus Callosum when looking at the sagittal plane. When contrast was given, a transversal T1 weighted sequence was performed. All images were interpreted by two experienced consultant neuroradiologists independently (observer 1 and observer 2). Both neuroradiologists re-measured 16 randomly selected MRI scans from 16 patients, in order to evaluate the intraobserver variability.
Both observers made linear measurements as follows:
- a)
Medio-lateral diameter (axial slice, d1): from the medial edge of the tumor to an imaginary line drawn along the posterior surface of the temporal bone diameter, parallel to the petro mastoid part of the temporal bone (to exclude the intra meatal component).
- b)
Antero-posterior diameter (axial slice, d2): parallel to the petrous ridge.
- c)
Cranio-caudal height (coronal slice, d3).
Observer 1 and 2 measured all VS scans in these three dimensions. Both were blinded to each other’s measurements and to earlier descriptions and former measurements of the MRI scans.
After thorough examination of d1, d2 and d3 we made the assumption, that VS has an ellipsoidal shape when excluding the intrameatal portion (see Figs. 3 A and 3 B). Therefore the volume was calculated using the formula: V = (pi/6) × d1 × d2 × d3.
