Editorial: Resection of vestibular schwannomas

Abstract

In present issue, Sughrue and colleagues 2 from the University of California at San Francisco, explore the effects of the extent of resection on the rate of vestibular schwannoma (VS) recurrence. The authors retrospectively review data that was prospectively collected in 772 patients who underwent gross-total resection (GTR), neartotal resection, or subtotal resection (STR) as assessed primarily by the surgeon. The authors' conclusion is a powerful one:…long-term rates of tumor control do not differ between patients who undergo GTR and those in whom a small amount of tumor is left behind. This finding argues against heroic attempts to achieve GTR in difficult cases, because tumor control is probably not improved markedly by these efforts… Contrary to the authors' conclusions, Fig. 1 seems to support the widely held notion that, in fact, GTR does reduce the risk of tumor recurrence. The authors, however, are careful to point out that this apparent reduced risk of recurrence after a GTR based on their data is not statistically significant-hence their conclusion. So what are we to conclude?Statistics aren't magic. They simply estimate the uncertainty of the conclusions we make based on a given set of data. If we conclude that there is no difference in tumor recurrence rate between GTR and STR for VSs, as the authors have here, we accept some uncertainty associated with that conclusion. If we are wrong, we are said to have committed a Type II statistical error by failing to observe a difference when one really exists-a false negative.The power of a statistical analysis is essentially the probability that we will not make the wrong conclusions based on insufficient data. To the authors' credit, they provide us with a detailed power analysis. A power analysis should always accompany important negative findings like the ones reported here as it gives us a measure of the confidence we can have in the conclusions we derive from the data. (Other factors also need to be considered such as biases in study and the specific patient population being studied.) With larger sample sizes, our confidence in the data increases and the likelihood we will make these wrong conclusions decreases. The power of a study with an end point like time-to-recurrence and our confidence in the study's conclusions depend on the number of recurrences, not the overall number of patients in the study. As such, the confidence we should have in the conclusions derived from these data depends on the 58 patients with recurrent tumors and not on the total number of 772 patients who underwent surgery. Other factors that influence the statistical power of such a study include the relative proportion of patients in each resection group and the duration of follow-up for each patient.One way to evaluate the conclusions that we derive from negative studies such as this one is to ask, "What increased risk of recurrence would our patients accept in exchange for the surgical risks of a GTR?" For simplicity's sake, let's just consider the comparison between GTR and...

Assessing the Threshold for Vestibular Schwannoma Resection and the Behavior of Residual Tumor

“…Thus, all resected tumors should be followed with serial MRI scans postoperatively. In a commentary (13), it was calculated that to detect a 10% or 100% increase (doubling) in risk of tumor recurrence between total resection and subtotal resection, sufficient patients would need to be studied to observe 6,429 and 122 tumor recurrences, respectively. It is clear that national tumor registries or, at a minimum, multicenter collaborations are required to achieve such numbers.…”