Letter to the Editor: Gamma Knife radiosurgery for vestibular schwannoma

Free access

If the inline PDF is not rendering correctly, you can download the PDF file here.

TO THE EDITOR: We carefully read the article by Boari et al.1 (Boari N, Bailo M, Gagliardi F, et al: Gamma Knife radiosurgery for vestibular schwannoma: clinical results at long-term follow-up in a series of 379 patients. J Neurosurg 121 (Suppl 2):123–142, December 2014). The importance and the complexity of vestibular schwannoma (VS) management is indirectly confirmed by the constant appearance in the literature of new articles dealing with that topic. The article mentioned above, in which the results of radiosurgery in VS are described, may represent a very relevant contribution, adding data to the existing literature and confirming the validity of radiosurgery as an option. We appreciated very much the huge efforts of the authors, although some considerations should be mentioned.

First, the tumor growth definitely represents the most important end point for comparison of the main treatment options. The authors declare a very high rate of tumor control (97.1%), and in Fig. 3B a Kaplan-Meier curve shows this. Unfortunately, a very important parameter is lacking: the censoring. As is well known, Kaplan-Meier estimators can take into account if a patient withdraws from a study, and this is called censoring, which applies when patients drop out before the final outcome is observed. Censoring is indicated by small vertical tick marks interspersed on the plot. Those lines are present in the Kaplan-Meier curve of Fig. 5, which shows hearing results, but are lacking in the most important one; i.e., the tumor growth curve (i.e., Fig. 3B). Moreover, every Kaplan-Meier curve should report the sample size below the x axis at every time point: this is not a controversial issue, but can have important consequences during summarization of the data.

For example, the authors obtained a 97.1% rate of tumor control in their study group, with a mean follow-up of 68.3 months and for a maximum period of observation of 156 months. But how many patients dropped out? When did they actually drop out? And most importantly, do the authors consider them to have a tumor indefinitely under control? Unfortunately, this discrepancy frequently manifests in follow-up studies, and the reality is that in the majority of cases patients who dropped out are not censored, so they are considered “survivors” or “tumor controlled.” This could also have strongly biased and overestimated the final rate of tumor control in the present study. A possible confirmation that the above-mentioned bias occurred can be indirectly obtained by the shapes of the Kaplan-Meier curves in Fig. 5. Although as has already been underlined, sample modifications are not specified under the x axis, the high steps made by the curves going down and moving toward the right part of the plot are strongly suggestive of a small sample size at the long follow-up, so the 97.1% rate of tumor control reported could be referring to only the few patients who were left, and most were lost to follow-up and not censored.

Second, oddly, whereas other parameters are fairly summarized by mean and median (e.g., time between symptom onset and diagnosis, time from diagnosis to treatment, and so on), the length of follow-up is expressed only by the mean value of 68.3 months (maximum 156 months): the mean is not the most appropriate parameter to describe a follow-up, because it can mask a positive (right tailored) skew of a median distribution. For example, a very few patients followed for decades could strongly influence the mean, but not the median values, so the latter is therefore the most appropriate value to summarize a follow-up.

Third, the Discussion subheading GKRS Versus Wait-and-See Strategy is not very balanced, because the authors report only results from the literature that support their treatment modality, and conclude that the wait-and-see strategy would not be indicated for VS. Results of the wait-and-see policy presented in the literature are variable, and most are much better than those mentioned by the authors.2 Anyway, if a single best treatment for every VS will ever exist, confirmation will only come from prospective randomized clinical trials, which are lacking at present.

References

  • 1

    Boari NBailo MGagliardi FFranzin AGemma Mdel Vecchio A: Gamma Knife radiosurgery for vestibular schwannoma: clinical results at long-term follow-up in a series of 379 patients. J Neurosurg 121:suppl 21231422014

    • Search Google Scholar
    • Export Citation
  • 2

    Patnaik UPrasad SCTutar HGiannuzzi ALRusso ASanna M: The long-term outcomes of wait-and-scan and the role of radiotherapy in the management of vestibular schwannomas. Otol Neurotol 36:6386462015

    • Search Google Scholar
    • Export Citation
Keywords:

Response

We hereby provide our response to the letter by Alicandri-Ciufelli et al. about our recent paper on Gamma Knife radiosurgery for VS.

First, the observation that censoring was not applied to our retreatment-free Kaplan-Meier curve (our Fig. 3B) is unfounded. The censoring of patients who dropped out of the study was considered, and it is reported in the curve as very small vertical tick-marks interspersed on the plot. Their density and small dimension may require some magnifying by attentive readers.

Had we actually done no censoring, we would have applied logistic regression instead of survival analysis, to benefit from some well-known possible advantages of it. Otherwise, considering dropout patients as “survivors” would be a trivial error. With respect to this issue, Alicandri-Ciufelli et al. offer an explanation of the shape of our Kaplan-Meier curve (Fig. 3B) that stems from the alleged occurrence of such a severe bias. Because this bias did not occur, this interpretation of the curve shape is itself biased. Actually, the Kaplan-Meier curve shows the failure timing, and particularly pinpoints that failures were clustered in the first 48 months after Gamma Knife radiosurgery treatment.

The statement that we obtained a 97.1% rate of tumor control with a mean follow-up of 68.3 months and for a maximum period of observation of 156 months does not correspond to any data reported in our manuscript. The value of 97.1% of tumor control was obtained over a series of 379 patients with a mean follow-up of 75.7 months, not 68.3 months. The 68.3 months value was reported for the 219 patients in the subgroup in which complete radiological volumetric MRI follow-up was obtained using the 'Gamma-plan' software.

Second, Dr. Alicandri-Ciufelli et al. stated that we reported the length of our follow-up only as the mean value of 68.3 months. We would like to remark that the reported mean value of the clinical follow-up was 75.7 months; furthermore, in the Follow-Up paragraph of the Methods section it is possible to disclose that the median values are reported, and they were 69.5 months for the clinical follow-up and 63 months for the quantitative radiological follow-up.

Dr. Alicandri-Ciufelli and colleagues claim that in the GKRS Versus Wait-and-See Strategy paragraph of the Discussion section, our arguments and the literature citations in favor of a proactive treatment are not very balanced. The very interesting manuscript by Patnaik et al., which is cited in the References of the Letter to the Editor, was published “ahead of print” in November 2014, whereas our manuscript was submitted for publication in the Journal ofNeurosurgery in June 2014; it was therefore impossible for us to consider and cite it in the Discussion.

Discussing the very controversial issue of the best strategy to use when dealing with patients harboring VSs, we cited the manuscript published by Stangerup et al. in 2006 (reference 57 in our paper), which we consider a milestone in understanding the natural history of these tumors. Discussing the pros and cons of the “wait-and-see” strategy, we have cited other very interesting manuscripts written by the same group in 2008 and 2010 (references 56 and 58). We have carefully read the aforementioned recently published manuscript by Patnaik et al., and we have observed that these authors report a smaller series than that reported by Stangerup and colleagues, and that the mean follow-up is shorter; furthermore, audiological follow-up is not considered in the study.

It is important to keep in mind that many patients with a stable tumor at a radiological follow-up present with a worsening of hearing; in our opinion the hearing function has to be considered, as much as the tumor growth, a main end point in a study investigating the natural history of VSs. Nevertheless, we agree with the authors that other studies on this topic are needed in large series of patients with longer follow-up; prospective randomized clinical trials are advisable but they are extremely hard to run. Probably the best management strategy can vary in different subgroups of patients; according to the results reported in our manuscript, young patients in Gardner-Robertson Class I represent a subset of patients in whom a proactive radiosurgical treatment should be strongly recommended because they seem to have a higher probability of retaining functional hearing at long-term follow-up.

If the inline PDF is not rendering correctly, you can download the PDF file here.

Article Information

Contributor Notes

INCLUDE WHEN CITING Published online July 10, 2015; DOI: 10.3171/2014.12.JNS142800.DISCLOSURE The authors report no conflict of interest.

© Copyright 1944-2019 American Association of Neurological Surgeons

Headings
References
  • 1

    Boari NBailo MGagliardi FFranzin AGemma Mdel Vecchio A: Gamma Knife radiosurgery for vestibular schwannoma: clinical results at long-term follow-up in a series of 379 patients. J Neurosurg 121:suppl 21231422014

    • Search Google Scholar
    • Export Citation
  • 2

    Patnaik UPrasad SCTutar HGiannuzzi ALRusso ASanna M: The long-term outcomes of wait-and-scan and the role of radiotherapy in the management of vestibular schwannomas. Otol Neurotol 36:6386462015

    • Search Google Scholar
    • Export Citation
TrendMD
Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 285 232 61
PDF Downloads 104 95 5
EPUB Downloads 0 0 0
PubMed
Google Scholar