Stereotactic Radiosurgery for Vestibular Schwannomas: Reducing Toxicity With 11 Gy as the Marginal Prescribed Dose

Background Stereotactic radiosurgery (SRS) is a common treatment option for vestibular schwannomas. Historically, a dose de-escalation of the marginal prescribed dose from 16 Gy to 12–13 Gy has been done to limit toxicity without reducing local control (LC). We aimed to retrospectively report outcomes of Linac-based SRS for vestibular schwannomas treated with different doses. Methods Included in the study were 97 stage 1 (1%), 2 (56%), 3 (21.5%), and 4 (21.5%) vestibular schwannomas treated with Linac-based (Novalis®) SRS from 1995 to 2019. No margin was added to the GTV to create the PTV. The median marginal prescribed dose was 14 Gy (range: 12–16 Gy) before 2006 and then 11 Gy for all patients (61 pts). Mean tumor volume was 1.96 cm3, i.e., about 1.6 cm in diameter. Mean follow-up was 8.2 years. Results Following SRS, LC at 3, 5, and 10 years was 100%, 98.4%, and 95.6%, respectively [100% for those with ≤ 13 Gy as the marginal prescribed dose (NS)]. Toxicity to the trigeminal nerve was reported in 7.2% of cases (3.3% and 0% for transient and permanent toxicity for 11 Gy). The marginal prescribed dose was the only significant predictive factor in univariate and multivariate analysis (HR = 1.77, 95% CI = 1.07–3.10, p = 0.028). Toxicity to the facial nerve was reported in 6.2% of cases. The marginal prescribed dose was again the only significant predictive factor in univariate and multivariate analysis (HR = 1.31, 95% CI = 0.77–2.23, p = 0.049). Conclusion Linac-based SRS for stages 1–3 vestibular schwannomas provides excellent outcomes: a 10-year LC rate of over 95%, with a permanent facial or trigeminal toxicity rate of under 5%. A marginal prescribed dose of 11 Gy seems to decrease nerve toxicity and facial toxicity in particular, without reducing LC. Prospective studies with longer follow-up are needed.


INTRODUCTION
Vestibular schwannomas are the most frequent tumors of cranial nerves with an incidence rate of 1 to 2 per 100,000 people. They are often unilateral, slightly more common in women and occur in patients with a mean age of 55 years (1,2). In about 5% of cases, they are linked to genetic diseases [mainly neurofibromatosis type 2 (NF2)] and can be bilateral; such cases generally occur in patients aged around 30 years old (1). Vestibular schwannomas' evolution is often slow and is classified according to the 4 stages of the Koos classification (3). The aim of treatment is local control (LC) while preserving the nerves of the pontocerebellar angle (especially the trigeminal and the facial nerves). For stage 1 (intra-canal) and 2 (extra-canal with invasion of the pontocerebellar angle without contact with the cerebellum or the brainstem) vestibular schwannomas which are progressive or symptomatic, either surgery or radiotherapy may be proposed without significant difference in terms of efficacy (LC ≈ 90%) (4).
Stereotactic radiosurgery (SRS) can be used to treat vestibular schwannomas with a large diameter of less than 3 cm (5-7). The first SRS were carried out in the 1990s with a mean prescribed marginal dose of 16 Gy (8)(9)(10). Historically, a de-escalation of the prescribed dose was then observed to reduce toxicities without reducing LC. Therefore, 12 to 13 Gy as the prescribed marginal dose in a single session is now generally accepted as a standard, allowing a 5-year LC rate of 95% and a facial and trigeminal nerve toxicity rate of 6%-8% (of which approximately 4% is transient toxicity) (11,12). A dose-effect relationship is widely reported for nerve toxicity: the greater the dose is over 8 Gy, the greater the risk of toxicity (13). Significantly higher toxicity rates are reported in literature for mean received doses to the nerves of 12 Gy (14). Because of their anatomy and proximity to the tumor, doses received to the trigeminal and even more to the facial nerves are directly linked to the marginal prescribed dose despite new radiotherapy techniques for targeting the tumor and minimizing received doses to the surrounding tissue.
Therefore, our study aimed to retrospectively assess the efficacy and toxicity of Linac-based SRS for vestibular schwannomas with different marginal prescribed doses over time, to determine the influence of dose de-escalation and the best marginal prescribed dose in order to reduce toxicity while maintaining excellent LC.

Patients' Selection and Characteristics
Ninety-seven patients (pts) consecutively treated for a vestibular schwannoma with Linac-based SRS from November 1995 to April 2019 were retrospectively included. Inclusion criteria were patients aged ≥ 18 years with performance status ≤ 2 and stages 1 to 4 vestibular schwannomas of the Koos classification with a large diameter of less than 3 cm. All patients were either symptomatic or with imaging proof of progression. Exclusion criteria were normo-fractionated stereotactic radiotherapy (FSRT) and schwannomas of another cranial nerve.
Initial damage to the facial nerve (determined using the House-Brackmann scale) and to the trigeminal nerve was evaluated as well as vestibular damage and damage to the brainstem (hydrocephalus in particular). Hearing function was assessed with the Gardner-Robertson scale: stages 1 and 2 corresponded to useful hearing and stages 3 to 5 to non-useful hearing. Oral corticosteroids (80 mg of prednisolone per day, gradually decreasing over 4 weeks) were delivered to patients the day of SRS.
The gross tumor volume (GTV) was identified using 0.9-mm 3D-CISS, T2-weighted and gadolinium-enhanced axial magnetic resonance imagery (MRI) sequences fused with high-resolution (1.25-mm slice thickness) computed tomography (CT) images. All 97 patients were treated with single-fraction SRS. No margin was added to the GTV to create the planning target volume (PTV). The marginal prescribed dose corresponded to the 80% isodose line. It was gradually reduced over time: 16 Gy between 1995 and 1996 (6 pts, 6.2%), 14 Gy between 1996 and 2002 (21 pts, 21.6%), 12 Gy between 2002 and 2006 (9 pts, 9.3%) and 11 Gy between 2006 and 2019 (61 pts, 62.9%). Accepted coverage limits were that 98% of PTV or more should receive at least the marginal prescribed dose. On a case-by-case basis, if nearby organs were at risk (facial and trigeminal nerves, brainstem, cochlea), the coverage prescribed dose was lowered so that 98% of PTV or more should receive at least 10 Gy. All treatment schedules were reviewed and approved by the treating radiation oncologist, neurosurgeon and physician.

Follow-Up
Follow-up included a clinical examination and brain MRI at 6 and 12 months, then annually. An audiogram was performed annually during the first 2 years and every two years afterward. For evaluation of radiological tumor control, the last follow-up MRI images were compared to the baseline pretreatment MRI images. Tumor size was defined as the maximum mediolateral and antero-posterior diameter in transverse contrast-enhanced T1 MRI, according to the standardization of volume assessments proposed by Li et al. (15). An increase in tumor size of more than 3 mm was defined as local failure according to Huang et al. (16),  two or more years after SRS (17). According to published studies, we additionally defined tumor control as freedom from reintervention (repeated SRS or surgery) (18). All these criteria enable to take into account a known transient enlarge of tumor after SRS (19). Clinical treatment-related toxicities were defined as new neurological deficits occurring after SRS. These toxicities were classified as temporary when they resolved spontaneously or after a short course of medical therapy such as corticosteroids, and as permanent if they did not resolve. Early toxicity was defined as the appearance of a clinical sign within 90 days of the end of radiotherapy and late toxicity beyond 90 days after treatment. Toxicity of the facial nerve was defined as the appearance or worsening of facial paralysis [determined using the House-Brackmann scale (20)]. Toxicity of the trigeminal nerve was defined as the appearance or worsening of symptoms related to, in particular, hypoesthesia, paresthesia, or neuralgia in the territory of the trigeminal nerve. Hearing toxicity was defined as the worsening of hearing loss on the audiogram performed annually in patients with previously useful hearing (stages 1 and 2 of the Gardner-Robertson scale) and was not assessable in patients with pre-treatment deafness. Mean follow-up was 8.2 years (4.8 years for groups 11-12 Gy and 17.1 years for groups 14-16 Gy).

Statistical Analysis
LC and overall survival (OS) were calculated using the Kaplan-Meir method. Time to local failure was defined as the period of time from SRS to the date of radiographic evidence of local failure at the treated site. The Cox proportional hazards model was performed to identify predictive factors of LC or toxicity. A two-sided p-value < 0.05 was considered significant. The following factors were included in the univariate analysis for LC: age, stage of tumor (Koos classification), history of surgery, presence of neurofibromatosis, GTV, marginal prescribed dose, received doses to the GTV (D max , D min and percentage of GTV covered by the marginal prescribed isodose line) and, for toxicity, received doses to the homolateral cochlea (D max , D 2% , D moy ), to the trigeminal nerve (D max , D 2% ), to the brainstem (D max , D 2% , V 12Gy ) and to the facial nerve (because of its proximity to the target volume and the impossibility of delineating or sparing it, the facial nerve was not contoured, the latter receiving substantially the same dose as the periphery of the target volume). The Spearman correlation enabled the identification of strongly correlated factors between them that were not included in the multivariate analysis. Factors associated with a p-value < 0.1 in the univariate analysis were included in the multivariate analysis. Finally, comparisons of LC and facial and trigeminal nerve toxicity curves were conducted using the logrank test.
Concerning hearing toxicity, a hearing loss or decrease was observed in 35.7% (20 pts) of patients with pretreatment useful hearing. Five years after treatment, 13 pts (34.2%) and 7 pts (38.8%) in groups 11-12 Gy and 14-16 Gy, respectively, presented hearing loss (NS). No statistically significant predictor of hearing loss was found in univariate or multivariate analysis. Other toxicities included early side effects and consisted of headaches (1 pt, 1%) and epilepsy (1 pt, 1%).

DISCUSSION
To our knowledge, this study is the first to investigate the efficacy and toxicity of a de-escalation of the marginal prescribed dose of up to 11 Gy for SRS of vestibular schwannomas. It was therefore important to show the non-inferiority of a dose reduction for LC. The strength of our study also lies in the length of patient followup: an average of 8.2 years, including 16 years for patients with marginal prescribed doses of over 11 Gy. 5-year LC was excellent, with an absence of local failure for patients treated with a marginal prescribed dose of 11 Gy and was not significantly different from LC for patients with marginal prescribed doses of over 11 Gy (100% vs. 98%, p = 0.3).
Most publications are based on treatments performed with a CyberKnife or a GammaKnife. Only 10 studies have been published about Linac-based SRS for vestibular schwannomas (18,(21)(22)(23)(24)(25)(26)(27)(28)(29). The median number of patients included in these studies was 76 and outcomes were generally reported for about 5 years, with a median follow-up of 5.5 years. Moreover, there was a lack of facial toxicity data. The median marginal prescribed dose was 12.5 Gy and the 5-year LC 88.9% (range: 68%-100%, 1,204/1,356 pts). The largest published study with 335 patients and with a marginal prescribed dose of 12 Gy had reported a 5year LC of 89% with a follow-up of 2.5 years. Concerning GammaKnife-or CyberKnife-based SRS published studies, outcomes are the same, with an LC of 95% (2691/2834 patients) within the follow-up period (12). In our series, with a mean follow up of 8.2 years, we found a 5-year LC of 98.4%. For the specific 61 patients treated with a 11 Gy marginal prescribed dose, LC was 100% with a mean follow up of 3.4 years.
The two main limitations of our study are the smaller number of patients treated with 12 to 13 Gy as the prescribed marginal dose (9 vs. 61 pts) and the shorter follow-up for those treated with 11 Gy (168 vs. 41.2 months) even if mean follow-up was 60.5 months (i.e., 5 years) for 36 patients treated with 11 Gy (range: 26.2-158 months). Therefore, it is difficult to really compare these two groups in the absence of a randomized study and the standard prescribed marginal dose in a single session should remain 12 to 13 Gy as recommended by the NCCN guidelines and RTOG studies. Other limitations of our study are that it is a retrospective study and thus dosimetric data (conformity index, gradient index, parts of received doses to organs at risk or PTV) are missing because the first SRS treatments started in 1995, i.e., over 20 years ago. However, the 20-year time span is also a strength, as it enabled a very long follow-up. The first patients treated in our institution had significantly bigger vestibular schwannomas with more pretreatment symptoms (more stage 4, more initial trigeminal nerve damage and more hearing impairment) and a longer follow-up. Normo-FSRT can be now be used thanks to frameless masks without invasive procedures; this technique enables the treatment of easier stage 4 vestibular schwannomas as well as limiting toxicities in cases of proximity to an at-risk organ (cochlea, trigeminal nerve, or brainstem). Moreover, new and modern radiotherapy techniques such as the use of VMAT with non-coplanar arcs help to avoid organs at risk and thereby reduce toxicities. Nevertheless, it is not possible to spare the facial nerve that is attached to the tumor, whatever the technique used. Thus, it is important to try to reduce the marginal prescribed dose as much as possible.
Facial nerve toxicity was reported in 6.9% of cases in published Linac-based SRS studies (range: 2%-17%, 35/506 pts), 4.3% of which (20/465 pts) was permanent (18,(21)(22)(23)(24)(25)(26)(27)(28)(29). According to published studies of GammaKnife-or CyberKnifebased SRS, the facial toxicity rate was 3.6% (74/2,064 patients) (12). In our series, we found a 2.9% rate of global facial toxicity, 0% of which was permanent. Interestingly, transient facial toxicity occurred in only 1.6% of patients treated with a marginal prescribed dose of 11 Gy versus 11.1% of patients with a marginal prescribed dose of 12 Gy. Even if the follow-up was shorter for patients treated with 11 Gy as a marginal prescribed dose, we showed that 85% of toxicity occurred within the first years. Therefore, most of the toxicities likely occurred near the beginning of the follow-up period.
This study, with its exceptionally long follow-up of 8.2 years, reports the excellent outcomes of Linac-based SRS for vestibular schwannomas, especially in the areas of efficacy (10-year LC of 95.6%) and safety (a 7.2% rate of transient trigeminal toxicities of which 2% were permanent, a 6.2% rate of transient facial toxicities of which 2% were permanent and a 40% rate of hearing impairment). That confirms the place of SRS in therapeutic strategies for stages 1-3 vestibular schwannomas, particularly in comparison to surgery.

CONCLUSION
Linac-based SRS for stages 1-3 vestibular schwannomas provides excellent outcomes: a 10-year LC rate of over 95%, with a permanent facial or trigeminal toxicity rate of under 5%. The standard prescribed marginal dose in a single session should remain 12 to 13 Gy as recommended by the NCCN guidelines and RTOG studies. Therefore, a marginal prescribed dose of 11 Gy seems to decrease cranial nerve toxicity and facial toxicity in particular, without reducing LC. Prospective studies with longer follow-up are needed.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
Study ethics approval was obtained on 25 September 2020 (CECIC Rhône-Alpes-Auvergne, Grenoble, IRB 5921). As the study was retrospectively completed, written informed consent was waived by the ethics committee.