The Conditional Probability of Vestibular Schwannoma Growth at Different Time Points After Initial Stability on an Observational Protocol

Abstract

Objective: The natural history of vestibular schwannomas (VS) is well documented in the literature, with tumour growth being paramount to decision making for both surveillance and treatment of these patients. Most previous studies refer to the risk of VS growth over a given period of time; however, this is not useful for counselling patients at different stages of their follow-up, as the risk of tumour growth is likely to be less following each subsequent year that a tumour does not grow. Accordingly, we investigated the conditional probability of VS growth at particular timepoints, given a patient has not grown thus far. This Bayesian method of risk stratification allows for more tailored and accurate approximations of the risk of growth versus nongrowth of VS. Methods: Retrospective analysis of a prospectively collected database in a tertiary referral skull base unit, containing all patients diagnosed between 2005 and 2014 with sporadic unilateral VS and a minimum of 5-year surveillance. Results: A total of 341 patients met the inclusion criteria. The mean age at diagnosis was 67 years, the sizes of the VS at diagnosis were intracanalicular in 49%, small in 39%, medium in 11%, and large in 1%. Over the entire 5-year surveillance period, a total of 139 tumours were seen to grow (41%) and 202 did not grow (59%). At 1 year, the probability of growth given that the tumour had not grown to date was seen to be 21%, at 2 years 12%, at 3 years 9%, at 4 years 3%, and at 5 years 2%. The conditional probability of growth of extracanalicular VS was significantly higher in the first year when compared with intracanalicular VS (29% versus 13%, p ¼ 0.01), but there was no such difference in years 2, 3, 4 or 5 ( p ¼ 0.60, 0.69, 0.36, 0.39, respectively). Conclusion: This is the first study in the literature concerned specifically with the conditional probability of VS growth. The data presented here can be used to better inform VS patients of their risk of growth at particular time points in their disease-the longer VS have been observed to be stable, the lower the risk of subsequent growth in a given year. Further, an extracanalicular vestibular schwannoma is more likely to grow in the first year compared with an intracanalicular vestibular schwannoma. Our data also adds support to surveillance protocols with increasingly infrequent MRI scans, as after 4 years of not growing, the risk of growth in year 5 falls to <2%.

Management of Vestibular Schwannoma with Normal Hearing

“…Moreover, the patient must abide by a prolonged F/U and the cost of the annual imaging (which will be initially identical to the postoperative imaging of patients undergoing surgery). Although the probability of tumor growth plummets steadily and significantly in the first 5 years [Sethi et al, 2020], a minority of patients might show a delayed growth which makes long F/U essential [Macielak et al, 2019]. Unfortunately, up to 40% (18/45) of the patients who chose the W&S management at our center were not compliant.…”

Spontaneous Volumetric Tumor Regression During Wait-and-Scan Management of 952 Sporadic Vestibular Schwannomas

“…Much of the evidence to date informing our understanding of the natural history of sporadic vestibular schwannoma has extended from studies using single-slice linear tumor measurements in the axial imaging plane (13)(14)(15). However, growth determined by maximal diameter in the axial plane is likely to underestimate growth in the coronal plane or in shorter axial dimensions (8).…”

Evidence-based surveillance protocol for vestibular schwannomas: a long-term analysis of tumor growth using conditional probability

“…Data analysis using conditional probability was undertaken, and the risk of VS growth was calculated. For a given year, the probability at year t refers to growth probability between year t − 1 and year t. Conditional probability growth estimates for t = 1 to t = 5 have been published and shown to be externally valid; 9 thus, these estimates of growth risk were used for the first 5 years of follow-up, with data in the current study showing long-term growth risks for t = 6 to t = 10.…”

“…Also useful for determining the frequency of surveillance imaging is the question, how likely is the tumor to grow in the next time interval? As previously described by us, 9 conditional probability (a statistical method derived from Bayes' theorem) would prove more accurate and useful in a dynamic evaluation of growth risk over time. In this methodology, the probability of VS growth (G) at a time (t) is calculated by taking into consideration how long the patient's VS has not grown (N) before t.…”