Introduction
Vestibular schwannomas (VS), also known as acoustic neuromas, are benign slow-growing neoplasms that arise from Schwann cells in the vestibulocochlear nerve. Reference Propp, McCarthy, Davis and Preston-Martin1,Reference Stangerup and Caye-Thomasen2 Patients commonly present with hearing loss and/or balance problems and left untreated these tumours can grow and cause cranial nerve impairment and brainstem compression. Reference Kentala and Pyykko3,Reference Matthies and Samii4 VSs are usually managed with surgery and/or radiotherapy with the goal of controlling tumour growth while attempting to preserve auditory, cranial nerve and brainstem functions.
Stereotactic radiosurgery (SRS) has become an attractive and non-invasive alternative to microsurgery for VSs. Reference Barker, Carter, Ojemann, Jyung, Poe and McKenna5–Reference Fuss, Debus, Lohr and Huber10 Radiosurgery can either be delivered in a single fraction, referred to in this article as SRS, or delivered in a fractionated manner and referred to here as hypofractionated stereotactic radiotherapy (HFSRT). Conventionally fractionated radiotherapy (CFRT) is non-stereotactic and involves smaller daily doses (e.g., 28–30 fractions to total doses of 50–54 Gy). HFSRT is particularly attractive for treating larger lesions as it minimises toxicity by spreading the dose out. HFSRT, which is typically given in 5–10 fractions to a total dose of 25 –35 Gy, has been favoured over CFRT because of fewer treatment visits and comparable tumour control without increased complication rates. Despite its recent use in treating larger VSs, the clinical outcomes and safety of HFSRT are difficult to assess from studies because of varied dosing patterns and fractionation regimens. Reference Arribas, Chust and Menendez11–Reference Smee, Schneider and Williams13
Numerous retrospective series have demonstrated excellent outcomes for small tumours treated with SRS alone. Reference Boari, Bailo and Gagliardi14–Reference Sun and Liu22 SRS for tumours larger than 2·5 cm in anterior–posterior diameter is more controversial with conflicting morbidity data reported. Reference Foote, Coffey and Swanson15,Reference Kondziolka, Lunsford, McLaughlin and Flickinger17,Reference Hasegawa, Fujitani, Katsumata, Kida, Yoshimoto and Koike23,Reference Pollock, Lunsford and Kondziolka24 Complete resection of larger tumours usually carries unacceptable morbidity risk, and subtotal debulking surgeries are often performed instead. Reference Bloch, Oghalai, Jackler, Osofsky and Pitts25,Reference Patni and Kartush26 There have been sporadic reports of malignant transformation after SRS for larger tumours, and control rates have been demonstrated to be lower. Reference Meijer, Vandertop, Baayen and Slotman20,Reference Hanabusa, Morikawa, Murata and Taki27,Reference Shin, Ueki, Kurita and Kirino28 Given these concerns, SRS for larger tumours is commonly reserved for inoperable patients, and subtotal resection followed by SRS for enlarging tumours is a frequently done to minimise target volume. Reference Godefroy, van der Mey, de Bruine, Hoekstra and Malessy29–Reference Park, Jung, Kim, Son, Paek and Kim31 While there are numerous series of single fraction regimens for both large and small VSs reported in the literature, as well as reports of HFSRT for small VSs, there is a lack of HFSRT data for large VSs. Reference Meijer, Vandertop, Baayen and Slotman20,Reference Puataweepong, Dhanachai and Dangprasert21,Reference Combs, Engelhard and Kopp32,Reference Ju, Lin and Lin33
There remains a need to evaluate HFSRT outcomes based on a consistent fractionation regimen. In order to discern the safety and efficacy of five-fraction HFSRT treatments, we performed this institutional retrospective study specifically focusing on tumour control rate and cranial nerve and brainstem toxicities on VSs treated with a 5 Gy × 5 fractionations scheme. We then examined the role of surgical debulking followed by HFSRT in management of large VSs.
Materials and methods
Patient population and study design
A chart review at our institution revealed 69 VS patients treated with stereotactic radiation between 2009 and 2016. Of these, 59 were treated with HFSRT to a dose of 25 Gy in 5 fractions. The remaining 10 patients had tumours that were very small and were treated with SRS to a dose of 12 Gy. Patients included in this study were those with: (1) a new diagnosis of VS with hearing loss, (2) an enlarging VS on imaging, (3) worsening hearing loss on a surveillance audiogram or (4) a large tumour at diagnosis not amenable to gross total resection and (5) patients treated with HFSRT. Patients were divided into large (>2 cm or causing brainstem compression) and small tumour groups (<2 cm) based on Koos stage (Figure 1). Reference Koos, Spetzler, Böck, Salah, Koos, Böck and Spetzler34 There were 42 patients in the small tumour group and 17 in the large tumour group. There were eight patients in the large tumour group who had surgical debulking before HFSRT (Table 1).
Patient characteristics and dosimetric parameters
Patient simulation and treatment
All patients were simulated on a 16 slice Phillips Brilliance Big Bore computerized tomography (CT) Scanner (Phillips Healthcare, Andover, Massachusetts, USA) using a thermoplastic immobilisation mask. CT images were imported into the Brainlab iPlan system (Brainlab iPlan, Feldkirchen, Germany) for treatment planning. Magnetic resonance imaging (MRI) images were obtained using our institutional protocol for the internal auditory canal and were fused to the simulation CT images for target and organ-at-risk (OAR) delineation. The gross tumour volume GTV) was contoured on the T1-post contrast fat suppression MRI sequence and was expanded by 1 mm to create the planning target volume (PTV). HFSRT plans were generated using non-coplanar arcs or mixed with steady intensity-modulated radiation therapy beams with conformal indices of 1–1·5. All patients were treated with 25 Gy in 5 fractions on a Novalis TX linear accelerator using 6 MV beams (600 MU/min) with a high definition multileaf collimator (2·5 mm leaves) (Varian, Palo Alto, CA). The ExacTrac 6-degree-of-freedom robotic coach alignment system and cone beam CT were used for daily image guidance.
Data collection and statistical analysis
Patient demographics were collected including sex, age, laterality and treated (PTV) volume. Pre- and post-treatment cranial nerve status was collected from clinical documentation at the follow-up visits (6 months interval for the first two years and then yearly) including trigeminal nerve function (normal, hypoesthesia or paresthesia) and facial nerve function (House–Brackmann grade). Dosimetric data for the PTV and brainstem were collected from the treatment planning system. Dose to the brainstem was evaluated with the maximum delivered point dose (Dmax), the minimum dose to the hottest 0·5 cc (D0·5cc) and the volume receiving at least 23 Gy (V23) (Table 1). Rates of post-radiotherapy tumour control were measured using follow-up MRIs to determine if there was true progression. Pseudo-progression defined as temporary enlargement and cystic change of tumour on follow-up MRIs was not counted against tumour control.
Patients with large VSs who underwent surgery were separately analysed to compare the improvement of brainstem dosimetry before and after surgery. These patients either received radiation right after surgery or were observed and treated when the tumour started growing. Hypothetical treatment plans were created using pre-surgical MRI images for target volume and brainstem delineations, and the same tumour and brainstem dosimetric data were collected in order to quantify the effect of debulking on GTV volume and brainstem dose (Table 2).
Tumour volume, brainstem dosimetric and NTCP comparison between pre- and post-surgical debulking HFSRT plans in patients with large tumours who underwent debulking surgery prior to radiation
The cox regression model was used to relate demographics and dosimetry to clinical outcomes of tumour control (tumour appearance on MRI), cranial nerve toxicity and brainstem toxicity (hyper-dense FLAIR signal and necrosis on MRI images). The Wilcoxon signed ranked test was used to determine if there was a difference of tumour control, cranial nerve and brainstem toxicities between small and large VSs groups and the pre- and post-surgical plans on large VSs (Table 3).
Clinical outcomes between small tumour group and large tumour group
Normal tissue complication probability modelling (NTCP) for brainstem toxicity
Normal tissue complication probability (NTCP) modelling was performed to compare the probabilities of brainstem toxicity between the treated post-surgical plans and the hypothetical pre-surgical plans (Table 2). This model incorporates dose–response relationships for tumours and adjacent normal tissues. Reference Archambeau and Shymko35–Reference Schultheiss, Orton and Peck38 The mathematical expression for this model is
Equation 1:
$$NTCP = {\left[ {1 - \mathop \prod \limits_{i = 1}^M {{\left( {1 - P{{\left( {{D_i}} \right)}^s}} \right)}^{\Delta {v_i}}}} \right]^{1/s}}$$
where,
Equation 2:
$$P\left( {{D_i}} \right) = exp\left[ { - {e^{e\gamma - \left( {{{EQD_{2Gy}^i} \over {{D_{50}}}}} \right)*\left( {e\gamma - lnln\;\left( {lnln\;\left( 2 \right)\;} \right)\;} \right)}}} \right]$$
where
$P\left( {{D_i}} \right)$
is the probability when irradiating with D
i
, M is the total number of voxels in the organ, s is the relative seriality parameter based on the functional subunit of the organ, Δv
i
(= ΔV
i
/V
ref
) is the relative volume of an organ that is irradiated with D
i
and V
ref
is the volume of the OAR.
Reference Mavroidis, Pearlstein and Dooley39
D
50
and γ are the physical parameters of the sigmoid curve that represent the complication probability vs delivered dose. D
50
is the dose at which there will be a 50% complication probability, and γ is the steepness of the curve.
$EQ{D_{2Gy}}$
is the dose delivered in 2 Gy fractions that will produce the same biological effects as a different fractionation scheme.
$EQ{D_{2Gy}}$
is calculated as
Equation 3:
$$EQ{D_{2Gy}} = D*\left( {{{1 + {d \over {{\alpha \over \beta }}}} \over {1 + {2 \over {{\alpha \over \beta }}}}}} \right),$$
where the numerator is the biologically effective dose for a fractionation scheme with a dose per fraction of d, and the denominator is the relative effectiveness on the specific tissue
$\left( {{\alpha \over \beta }} \right)$
at 2 Gy/fraction.
MATLAB (Matlab, Natick Massachusetts, USA) 2017b was used to create an algorithm for the relative seriality model. The DVHs for the OARs were used as input arguments into the algorithm along with the physical parameters. Once the NTCPs for the brainstem were calculated for both pre- and post-surgical treatment plans, the Wilcoxon signed ranked test was used to determine if there was a significant difference.
Results
Tumour control
The median follow-up for the entire group was 49 months (range 4–98 months). The median follow-up for the small tumour group and the large tumour group were 33·5 (range 5–76 months) and 15 (range 2–49 months) months, respectively. The tumour control rate was 100% in the small tumour group and 94·1% in the large tumour group (no significant different with p = 0·12), respectively. Only one patient with a very large VS (PTV = 21 cc) had a recurrence; his tumour could not be debulked initially due to intraoperative bleeding and was instead treated with HFSRT only. The tumour grew on follow-up imaging ultimately requiring surgery 5 years after HFSRT.
In the large tumour group, 8 of 17 patients had surgical debulking before HFSRT. The tumour control rate of HFSRT was 100% (8/8) for surgically debulked patients and 89% (8/9) for non-surgically debulked patients (p = 0·35). The median tumour volume and brainstem V23 were significantly reduced by debulking (Table 2).
From regression analysis, no association was identified between the patient demographics, dosimetric parameters and tumour control rates. The tumour control rates were excellent for both groups (Tables 3 and 4).
Clinical outcomes between large, debulked tumour group and large, non-debulked tumour group
Tumour volume reduction and brainstem dosimetry improvement after surgical debulking
The median tumour volumes in large and small tumour groups were 4·9 cc and 0·9 cc, respectively. The median brainstem Dmax was 26·1 Gy for large tumours and 16·5 Gy for small tumours. The median D0·5cc was 20·5 Gy for large tumours and 4·75 Gy for small tumours. The median V23 was 0·18 cc for large tumours and 0 cc for small tumours.
There were eight patients in the large tumour group who underwent debulking before radiotherapy with follow-up available for analysis. HFSRT was delivered either following surgery (5/8) or when tumour growth was identified on follow-up MRI (3/8). Three of eight (37·5%) patients with large tumours undergoing surgery had cranial nerve V and/or VII damage (Table 4). These cranial nerve symptoms did not worsen following radiation. Debulking significantly reduced target volume and brainstem doses. The average post-operative tumour volume was 9·6 cc compared to 21·8 cc pre-surgery (p = 0·01455), respectively. Compared to the hypothetical pre-surgical treatment plans, the post-surgical plans resulted in an average brainstem Dmax of 26·2 Gy vs. 26·5 Gy (p = 0·9844), an average brainstem D0·5cc of 19·46 Gy vs. 23·4 Gy (p = 0·0156), and an average brainstem V23 of 0·21 cc vs. 1·34 cc (p = 0·0234) (Table 2). There was no brainstem toxicity in any patient in this study (Tables 3 and 4).
Normal tissue complication probability (NTCP) modelling for brainstem
Although the calculated brainstem toxicity using NTCP modelling was very low in both pre- and post-surgical debulking scenarios, there was a statistically significant improvement of NTCP with debulking. The median NTCP was 0·78% before surgical debulking and 0·29% afterward (P < 0·05). A single patient had a brainstem NTCP of 2·7% without debulking, while the other seven patients were all under 1% (Table 2).
Brainstem necrosis
No patients in this series experienced any clinical symptoms or radiographic evidence of brainstem necrosis. Although brainstem dosimetric analysis demonstrated difference of tumour volume, V23 volumes and NTCP between debulked and non-debulked large tumours, they were all remained in tolerant ranges as recommended from TG 101(Dmax and V23), Reference Benedict, Yenice and Followill40 and NTCPs were very low (0·0 to 2·7%) (Table 2).
Cranial nerve Dysfunction
No patients in this series experienced trigeminal neuropathy based on follow-up clinical assessments. Freedom from new facial neuropathy was 97·6% in the small group vs. 100% in the large group (p = 0·99). One patient in the small group with a pre-treatment House-Brackmann score of 1 (no symptoms) experienced new facial neuropathy (House–Brackmann score of 2).
Pseudo-Progression
Pseudo-progression is a transient enlargement in tumour size after radiotherapy and is distinct from true treatment failure. Reference Hayhurst and Zadeh41 It was defined here as a slight increase in the size of tumour volume with significant central necrosis seen on follow-up MRIs. Pseudo-progression rates were 4·8% in the small tumour group vs. 11·8% in the large tumour group (p = 0·34). For patients in the large group, there was no statistically significant difference between pseudo-progression rates in those who received surgery and those who received radiation alone (25% vs. 0%, p = 0·12).
Discussion
In this study, we have demonstrated excellent outcomes with HFSRT for both small and large VSs. The usual practice at our institution has been to treat Koos stage 1–2 tumours with radiation alone, typically HFSRT at a dose of 25 Gy in five fractions for potential hearing preservation if patients had serviceable hearing. Larger tumours are often debulked for symptom relief and dosimetric improvement. Patients who undergo debulking are either treated shortly after surgery or are observed and treated with HFSRT when residual tumour progression is identified. The primary objectives of this study were to examine the clinical outcomes of small and large VS patients treated with HFSRT and to evaluate the role of upfront debulking in improving tumour control and dose to the brainstem.
There is a large body of literature supporting SRS for small VSs; however, the optimal management for larger tumours is less clear. Reference Boari, Bailo and Gagliardi14–Reference Sun and Liu22 Numerous treatment strategies have been described for Koos stage 3–4 tumours including maximal possible resection followed by observation, debulking followed by adjuvant radiation in a variety of fractionation schemes and radiation alone. Reference Patni and Kartush26,Reference Darrouzet, Martel, Enee, Bebear and Guerin42–Reference Pollock, Driscoll and Foote44 Primary concerns with radiation to the internal auditory canal, especially with hypofractionated radiation, include tumour control, hearing preservation and brainstem and cranial nerve toxicity. Reference Boari, Bailo and Gagliardi14,Reference Hasegawa, Kida, Kato, Iizuka, Kuramitsu and Yamamoto16,Reference Kondziolka, Lunsford, McLaughlin and Flickinger17,Reference Liu, Xu, Zhang, Zhang and Zheng19,Reference Sun and Liu22,Reference Combs, Engelhard and Kopp32,Reference Chihara, Ito, Sugasawa and Shin45,Reference Milligan, Pollock, Foote and Link46
For small VSs, numerous studies have demonstrated an advantage of either Gamma Knife or LINAC-based SRS over microsurgery in terms of tumour control and neurologic toxicity. Reference Boari, Bailo and Gagliardi14,Reference Hasegawa, Kida, Kato, Iizuka, Kuramitsu and Yamamoto16,Reference Kondziolka, Lunsford, McLaughlin and Flickinger17,Reference Pollock, Lunsford and Kondziolka24,Reference Pollock, Driscoll and Foote44 A recently published systematic review by Perrson et al. revealed excellent control rates with both SRS and HFSRT; however, there was a lack of long-term follow-up for patients treated with HFSRT. Reference Persson, Bartek, Shalom, Wangerid, Jakola and Forander47 While there were actually worse rates of facial and trigeminal toxicity after HFSRT vs. SRS in this review (only two HFSRT studies were included), our results show excellent tumour control rate and neurologic outcomes for patients treated with HFSRT with no trigeminal toxicity and only a single new case of facial neuropathy.
Larger tumours have historically been managed initially with surgery. Talfer et al. have reported on the necessity of definitive surgery for large VSs with a tumour control rate of 86% with surgery alone, but at the cost of 37% of patients experiencing poor facial function measured by a House–Brackmann score of 3 or greater. Reference Talfer, Dutertre, Conessa, Desgeorges and Poncet48 These results are consistent with other surgical series for large VSs. Reference Patni and Kartush26,Reference Darrouzet, Martel, Enee, Bebear and Guerin42,Reference Deveze, Roche, Facon, Gabert, Pellet and Thomassin43,Reference Jung, Kang and Kim49–Reference Zhang, Fei and Chen53 Although perioperative mortality for this surgery is near zero, there are small risks for other peri-operative complications including CSF leak, infection, bleeding and stroke. Reference Talfer, Dutertre, Conessa, Desgeorges and Poncet48 Given the high percentage of patients experiencing facial dysfunction from surgery, investigation into alternative strategies for reducing cranial nerve toxicity was warranted.
There are less data for primary management of large VSs with radiation alone. Milligan et al reported their results of 22 large VS patients treated with Gamma Knife SRS (median dose 12 Gy) and found a tumour control rate of 82% at 5 years. However, functional hearing was preserved in 28% of patients, and freedom from new facial neuropathy was very good at 85%. Reference Milligan, Pollock, Foote and Link46 While these results are comparable to other surgical series in terms of tumour control, patients who are able to achieve a gross total resection have the lowest recurrence rates of around 3%; however, attempts at complete resection carry a higher risk of neurologic toxicity. Reference Bloch, Oghalai, Jackler, Osofsky and Pitts25,Reference Godefroy, van der Mey, de Bruine, Hoekstra and Malessy29,Reference Park, Jung, Kim, Son, Paek and Kim31,Reference Ramina, Coelho Neto, Bordignon, Mattei, Clemente and Pires Aguiar54 Cranial nerve toxicity was lower with Gamma Knife compared to typically reported rates with microsurgery. Reference Boari, Bailo and Gagliardi14,Reference Hasegawa, Kida, Kato, Iizuka, Kuramitsu and Yamamoto16,Reference Kondziolka, Lunsford, McLaughlin and Flickinger17,Reference Pollock, Lunsford and Kondziolka24,Reference Pollock, Driscoll and Foote44 While Gamma Knife was highly effective for the large VS patients, the rates of tumour control and facial function preservation were worse compared to small VSs. Reference Boari, Bailo and Gagliardi14–Reference Hasegawa, Kida, Kato, Iizuka, Kuramitsu and Yamamoto16,Reference Liu, Xu, Zhang, Zhang and Zheng19,Reference Sun and Liu22,Reference Milligan, Pollock, Foote and Link46 In our study, we have demonstrated excellent tumour control rates with minimal brainstem and cranial nerve toxicity for both small and large tumour groups treated with HFSRT. To our knowledge, this study represents the first reported series of HFSRT for large VSs with excellent tumour control (94·1%) and no incidence of brainstem and cranial nerve toxicity.
Although the excellent outcome of radiotherapy for large VSs in this study, the treatment paradigm of debulking surgery followed by SRS is still an attractive treatment strategy for large VSs, especially for patients who present with symptoms of brainstem compression. Reference Bloch, Oghalai, Jackler, Osofsky and Pitts25,Reference Patni and Kartush26,Reference Pollock, Driscoll and Foote44 The theoretical advantage of reducing tumour volume by debulking is to stave off radiation by allowing for a period of observation and/or reducing dose to the brainstem as large VSs commonly about the brainstem. Reference Kentala and Pyykko3,Reference Matthies and Samii4 We sought to evaluate the dosimetric and clinical benefits of surgery in the management of large VSs. By re-planning eight patients in our series, who underwent upfront debulking, we found significant improvements in brainstem D0·5cc and V23, but not the Dmax. As expected, the target PTV was significantly reduced with debulking. Pre-surgical plans of three patients did not meet the TG-101 V23 brainstem dose constraint, but median pre-surgical brainstem dosimetry was still within commonly referenced brainstem tolerance criteria (Dmax < 31 Gy, V23 < 0·5 cc). Reference Benedict, Yenice and Followill40 Given that none of the patients in our series experienced any brainstem toxicity, the clinical relevance of this lower brainstem dose finding is questionable. To further clarify this issue of clinical relevance, we used NTCP models, which revealed very low probabilities of brainstem toxicity with or without surgery; however, there was a small but statistically significant probability difference favouring debulking.
While this study has shown that at the very least, HFSRT alone appears to be an attractive option for selected patients who cannot undergo or otherwise refuse surgery with the advantages of avoiding possible surgical complications while retaining excellent neurologic outcomes, we do acknowledge some of its limitations of the small sample size of 59 patients and relative short follow-up, which may not have been adequate to fully characterize the clinical impact of HFSRT as a treatment for VSs as well as delineate any potential toxicities. We recommend further investigation by other institutions on their VS patients treated in a similar manner as well as studies into determining whether the timing of HFSRT after surgical debulking impacts clinical outcomes.
Conclusions
HFSRT to a dose of 25 Gy in five fractions appears safe and effective treatment option for both small and large VSs including tumours with significant brainstem compression in our study. The optimum management of large VSs is still unclear as there is a lack of randomised data; however, HFSRT alone appears to be a viable option for selected patients with large VSs, who cannot undergo surgery or are unwilling to have surgery. While this paper does not definitively recommend HFSRT as a treatment option, it does warrant further investigation as stated above.
Acknowledgements
Portions of this manuscript have been presented in abstract form at the American Society for Radiation Oncology (ASTRO) annual meeting in 2017 in San Diego, CA and in abstract form at the ASTRO annual meeting in 2018 in San Antonio, TX.
Financial support
Sources of financial and material support: No specific funding was granted for this project.
