Single-fraction radiosurgery outcomes for large vestibular schwannomas in the upfront or post-surgical setting: a systematic review and International Stereotactic Radiosurgery Society (ISRS) Practice Guidelines

Stereotactic radiosurgery (SRS) is a widely accepted treatment for small to medium sized vestibular schwannomas (Koos grades I, II and III) [1, 2]. However, for large vestibular schwannoma (VS), using a definitional threshold of ≥ 2.5 cm or Koos Grade IV designation, most recommend microsurgical resection [3]. In particular, surgery should be considered when there are signs and symptoms of mass effect related to brainstem compression, cranial nerve (CN) neuropathy (other than CN VIII and particularly CN V), and/or presence of hydrocephalus [4]. Importantly, a wait and scan strategy is usually not recommended due to the possibility of life-threatening complications associated with tumor progression [5].

For a patient with a large VS who is not an optimal candidate for microsurgical resection, some type of fractionated radiation therapy is typically recommended. However, several centers have treated these patients with single fraction SRS, as they would do for smaller VS [6, 7]. Concerns of single-fraction SRS in these patients range from development of serious adverse radiation events (ARE), transient-tumor-expansion (TTE, also referred to as pseudoprogression), delayed time-to-response for patients who have symptomatic hydrocephalus, and late treatment failure necessitating surgery which may put the patient at an increased risk of surgical complications [8]. To date, there has yet to be a critical review of the published literature specific to this population to define efficacy and toxicity of this approach. Therefore, the purpose of this systematic review is to summarize the current literature specific to single fraction SRS for large VS, and provide treatment recommendations on behalf of the International Stereotactic Radiosurgery Society (ISRS) Guidelines Committee.

The overall rate of facial palsy in all series was 2.3% (range 1.2–3.4%, I² = 54.47%, p heterogeneity = 0.003, p < 0.001; Fig. 4, a, upper part). The overall rate of new-onset facial palsy in series including patients with prior surgery was 1.3% (range 0.3–2.3%, I² = 28.00%, p heterogeneity = 0.16, p = 0.01; Fig. 4, a, middle part). The overall rate of facial palsy in series not including patients with prior surgery was 3.4% (range 2.2–4.6%, I² = 0%, p heterogeneity = 0.52, p < 0.001; Fig. 4, a, lower part).

The overall rate of hearing preservation in all series was 37.9% (range 21.6–54.3%, I² = 95.96%, p heterogeneity < 0.001, p < 0.001; Fig. 4, b, upper part). The overall rate of hearing preservation in series including patients with prior surgery was 34.2% (range 24.3–44.1%, I² = 61.5%, p heterogeneity = 0.01, p < 0.001; Fig. 4, b, middle part). The overall rate of hearing preservation in series not including patients with prior surgery was 40.4% (7.0–73.9%, I² = 98.39%, p heterogeneity < 0.001, p = 0.06; Fig. 4, b, lower part).Tables 2, 3 and 4 summarizes the outcomes.

Our systematic review suggests that single fraction SRS could be used for VS ≥ 2.5 cm in maximum diameter, and/or Koos Grade IV, as either the primary treatment modality or for post-operative residual/recurrent tumor. However, we acknowledge that the Koos grade IV tumor definition varies significantly across studies and the minority (6 studies out of 19) of the published literature was specific to upfront treatment.

The overall probability of tumor control (both stability and decrease in volume) and tumor reduction in all series versus those series without prior surgery were 89% versus 93.7%, and 57% versus 64%, respectively. Several of the included series in this meta-analyses identified individual parameters associated with local failure. More specifically, Hasegawa et al. [13] suggested that a high-risk group for lower tumor control included patients with middle cerebellar peduncle compression of ≥ 9.8 mm and ≤ 48 years of age. Tumor control was also higher when prescribing a marginal dose of greater than 12 Gy as compared with less than 12 Gy [19] and for those smaller volumes [18]. Previous microsurgery, tumor volumes exceeding 10 mL, Koos grade IV [8], tumor volume more than 15 mL [14] and progression of residual disease preceding SRS [21] were also factors resulting in lower local control rates. These findings informed the ISRS recommendations as summarized in Table 5.

The rates of salvage microsurgical resection, need for shunt, and additional SRS in all series versus those series with no prior surgery were 9.6% vs 3.3%, and 4.7% vs 6.4% and 1% vs 0.9%, respectively. The rates of facial palsy and hearing preservation in all series versus those series with no prior surgery were 2.3% vs 3.4% and 37.9% vs 40.4%, respectively. Preservation of the facial nerve function was associated with smaller tumor volumes (less than 10 mL) and lower margin dose (≤ 13 Gy) [19]. Deterioration of facial nerve function was associated with a prescription dose of ≥ 13 Gy and early TTE [28]. Hearing preservation was higher in patients with good pre-therapeutic levels of hearing (Gardner Robertson class 1), younger age, and a dose of less than 4 Gy to the cochlea/modiolus (the mean dose/point dose of less than 4 Gy to the cochlea/modiolus being already reported in the literature during the past 15 years and in the overall context of hearing preservation after SRS for VS [29, 30]). Cranial nerve complication rates were suggested in few of reviewed studies to be greater in those VS with cystic components vs solid [11]. Trigeminal neuropathy was rare and usually transient.

A particular entity that would deserve further analysis, although limited data exist on such topic, is related to previously irradiated large partially cystic VSs that will potentially develop symptomatic mass effect from fluid-dynamic cyst enlargement [31, 32], without definite neoplastic growth of the solid part. In such patients, microsurgical exploration for cyst fenestration/drainage without the need for further resection of the already treated tumor cells can be a valuable option, in the absence of tumor growth of the solid part. Such surgical option could reveal much safer and with less morbidity, in the absence of planned subtotal resection of the solid, non-growing part.

In the present review, the overall need for shunt for large VSs treated with upfront SRS in series without prior surgery was 6.4%, which is much higher as compared to smaller tumors. Previous series have suggested the need for shunt after SRS after a median time of 15.5 months (range 1.8–37.8) [33]. Hydrocephalus after radiosurgery may thus co-occur with a temporary tumor volume change after radiation and there is a crucial need for careful ongoing clinical and imaging follow-up [33]. Other authors suggested that large tumor size, ring enhancement patterns and high protein level of CSF should be carefully observed during follow-up course [34]. Thus, using programmable/adjustable MR-compatible ventriculo-peritoneal shunts in time might prevent devastating consequences due to increased intracranial pressure and a risk of sudden neurological decline.

Of the 1723 cases in this meta-analysis, four tumor-related deaths were observed. Two were secondary to developed malignant transformation, which accounts for 0.12% of the sample. This low risk is consistent with the literature including a recent meta-analysis [35]. The other two deaths were related to a refractory VS which relapsed 78 months from the time of SRS and the second due to tumor-related subarachnoid hemorrhage (Hasegawa et al.¹⁴). Moreover, TTE was inconsistently reported and should be better detailed by further studies on the same topic. Such TTE might be, in some cases, accompanied by acute and subacute radiation effects, which are in vast majority of cases transient [36].

With respect to fractionated radiation, there are as yet limited data with regards to the use of hypofractionnated SRS for large VS [13, 37]. However, there are 6 non-randomized trials [38‐43] comparing single fraction SRS with fractionated stereotactic radiotherapy (FSRT). There has yet to be significant differences in 5-year tumor control rates between the two techniques to make any firm recommendations. A recent systematic review compared SRS versus FSRT for tumor control in VSs [44]. The authors suggested that the progression-free survival rates were 92–100% for both treatment options, while the risk of facial and trigeminal nerve deterioration was less for patients treated with SRS [44]. It has been also acknowledged that there is a lack of high-quality studies comparing radiation therapy alternatives for patients with VSs [44]. We would still support fractionated stereotactic radiotherapy for large VS given the established practice as a standard of care, and experience in other benign tumors with favorable control rates [45, 46]. However, there is a need for a randomized or prospectively controlled trial comparing single fraction SRS and FSRT in VS, especially in the context of clarifying if functional outcome would be better with single fraction SRS.

The main limitation of the present meta-analysis was the inability to reliably separate outcomes between upfront vs salvage cohorts, and this added complexity to this analyses. We acknowledge that those studies including patients with prior surgery also included cases with upfront SRS, which can contribute to added bias. The definition of large tumors was also extremely heterogeneous. In particular, for those treated in the post-operative residual or recurrent setting. A limited number of series included “staged-volume” SRS strategies, which might have also influenced local control. The same applies to the cystic tumors, which influence the overall results in terms of local control, and in some series they account for as high as 58% of the included cases [24]. However, the results for cystic tumors have not been separately reported in individual series, although it is now well acknowledged that they respond best to SRS as compared to the solid ones [31]. Additionally, there was a lack of uniformity with regards to the follow-up periods, to which ads variations depending on studies to the long-term and even the short-term follow up. Tumor diameters were inconsistently reported. There was also a lack of reported actuarial outcomes, which are different from the crude rates reported in the studies. There were also several different nuances concerning further neurosurgical interventions, considered as adverse events and not counted as treatment failures. Particularly, the surgical management of hydrocephalus was heterogeneous, including ventriculo-peritoneal shunt, ventriculostomy, Ommaya placement or further cyst puncture and timing of further SRS and additional surgical interventions were also extremely variable, as well as for further surgical interventions.

Although large VS are considered a classical indication for microsurgical resection, upfront single fraction SRS might be useful in select patients (class C evidence). When analyzing data from those series with no prior surgery vs those with prior surgery, higher rates of tumor control, further tumor reduction, lower rates of further intervention (microsurgical resection, shunt, SRS), higher rates of “de novo” facial palsy (although overall low) and higher hearing preservation rates were observed (class C evidence).