Is salvage surgery for large vestibular schwannomas after failed gamma knife radiosurgery more challenging?

In order to verify whether a previous gamma knife surgery (GKS) treatment could influence the oncological and functional outcome in large vestibular schwannoma (VS) surgery, we have compared group of patients operated on for large VS after failed GKS to a group of genuine VS that underwent the same functional nerve-sparing resection technique regimen in the same period. Single center retrospective cohort study of 23 consecutive GKS failure and 170 genuine VS patients operated on between April 2003 and March 2019. After resection, patients were allocated to a Wait-&-rescan or an upfront GKS policy. At last follow-up examination, the facial nerve function was good (House-Brackmann grades I or II) in 95% of the GKS failure and 84% of the genuine VS patients (p = .25). The median volume of tumor residue was .56 cc in the GKS failure group and .62 cc in the genuine VS group (p = .70). Tumor control was achieved in 91% and 83% of cases with a mean follow-up of 74 and 63 months in the GKS failure and the genuine VS populations, respectively. The 1-, 5-, and 7-year progression-free survival were 100%, 95%, and 85% respectively in the GKS failure group and 97%, 80%, and 81% in the genuine VS group (p = .27). Despite significant modifications of the microsurgical environment associated to salvage surgery after GKS failure, a functional nerve-sparing resection is an effective strategy to optimize the results on facial nerve function, with similar long-term tumor control to those observed in the genuine VS population.


Introduction
The invasiveness and reported tumor growth control of 92 to 98% tend to propose the gamma knife surgery (GKS) as one of the established standard treatments for small to middle vestibular schwannomas (VS) [6,12,15,[21][22][23]. Salvage surgery after failed GKS has been reported in only 1.57 to 5% of cases [6,12,15,[21][22][23]. Some authors asserted that previous radiosurgery may significantly complicate this resection due to post-GKS fibrosis and severe adhesion to the neurovascular structures, which lead to additional morbidity [14,19,28,29]. Radical resection of those tumors are then associated to a higher risk of permanent postoperative facial nerve (FN) deficit [2,20]. In an effort to reduce the incidence of FN palsy, alternative strategies of non-total resections, a so-called functional sparing surgery policy, has emerged over the past 20 years [5,8,9,17,30,32]. In order to verify whether a previous GKS treatment could hamper the oncological and functional outcome, we conducted a comparative study between patients operated on for large VS after failed GKS to a consecutive group of genuine VS that underwent the same functional nerve-sparing resection technique regimen in the same period. This manuscript has not been previously published in whole or in part or submitted elsewhere for review.

Methods
All consecutive patients who underwent a salvage surgical treatment of large Koos grade IV VS [13] after a failed GKS, between April 2003 and March 2019, were enrolled in this retrospective cohort study. All of these patients have been previously treated in our institution, using the same GKS protocol (see below).
GKS failure was defined as the need to deliver salvage surgery because of worsening symptoms or continuous growth of the VS. The decision for a second stair treatment after regrowth is taken extremely cautiously in our institution because of the reported transient swelling remodeling of VSs within 3 years after GKS treatment that turned out on longer follow-up to be pseudo-progression with subsequent tumor control [1,16,18,24,35]. In order to verify whether previous GKS treatment could influence the oncological and functional outcome, we compared this group to a group of consecutive patients operated on for large genuine VSs in the same period, excluding those who reported history of previous treatment different than GKS (radiotherapy, cyber knife, microsurgery). Two patients harboring post-GKS growing tumors were diagnosed malignant peripheral nerve sheath tumors (over 3606 VSs treated with GKS in our institutionrate 5/10 4 ) and were excluded from the analysis (see flow chart Fig. 1).
An otoneurosurgical team performed the microsurgical resection via either an enlarged translabyrinthine or retrosigmoid approach. The diagnosis of VS was histologically confirmed in all cases. Informed consent of all patients was obtained, and our institutional local ethical committee approved this study (Aix-Marseille University Ethics Committee; authorization number: 2018-24-01-003).

Patient characteristics
Clinical features such as FN function (scored according to the House & Brackmann classification (HB)) [7], cochleovestibular symptoms including Gardner-Robertson hearing scale (G&R) [3], and radiological examinations (including volumetric measurements performed pre-and postoperatively on axial gadolinium-enhanced T1 weighted MR images; iPlan 3.0 Cranial Brainlab, Munich, Germany) [33] were recorded at presentation and during follow-up. Surgical findings (such as post-GKS arachnoiditis and severe adhesion to the neurovascular structures) were extracted from a systematic analysis of the operative report and recorded surgical video.
Tumor regrowth was defined as a volume increasing of the tumor remnant of more than 20% as compared to the first postoperative MRI [33].  1 Flow chart of the study population. Patients who reported history of previous treatment different than GKS (radiotherapy, cyber knife, microsurgery), those who were diagnosed malignant tumors (MNST), and those who presented a postoperative follow-up period inferior to 1 year were excluded from the analysis

Surgical protocol
Monitoring of the FN is mandatory. An exceptional dorsal course of the FN was checked by the means of visual control and systematic mapping of the dorsal part of the tumor surface (under 0.3 mA stimulation; NIM-Response® 3.0 Medtronic Xomed, Jacksonville, FL). The tumor capsule was opened, and extensive intracapsular debulking was carried out using an ultrasonic aspirator. Thresholds of 0.05 mA stimulation were performed throughout the tumor removal. The origin of the FN at the brainstem level was identified at the inferior pole of the schwannoma. The anterograde dissection was conducted at the close vicinity of the FN all the way to the porus. The intraoperative decision to interrupt the resection was driven by the evidence of critical adhesion of the tumor capsule to the facial nerve and/or deterioration of the electromyography EMG responses during stimulation [31]. The interruption of FN dissection was decided in cases of decreasing amplitude of the EMG recorded responses despite repeated stimulation at 0.2 mA. The patient was informed before the operation of the possibility of leaving a tumor residue.

Postoperative outcome
A gadolinium-enhanced MRI was performed 6 months after surgery to assess the tumor residue. According to postoperative FN palsy, the volume of the tumor residue and patient age, a Wait-&-rescan, or an upfront gamma knife surgery (GKS) policy was proposed to the patient [30]. Practically, we preferentially allocated to the W&reS group those who displayed a small tumor remnant (NTR), the ones who were affected by a postoperative FN deficit and elderly patients. Conversely, larger tumor remnants (STR or PR), those who presented with grades I-II facial nerve function, and younger patients were predominantly oriented toward GK. Clinicoradiological follow-up was planned at 6 and 12 months after resection and subsequently at 2, 3, 5, 7, and 10 years after surgery and then once every 3 years thereafter.

Preoperative or postoperative radiosurgical protocol
Treatment was delivered using a Leksell Gamma Knife (Elekta Instrument AB, Stockholm, Sweden) according to a previously published methodology [25]. The median dose directed at the tumor margin was 12 Gy, and the median isodose to the margin was 50%.

Statistical analysis
Statistics were performed with IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 21.0. Armonk, NY: IBM Corp. Categorical variables are presented as numbers and percentages and continuous variables as means ± SD. The significance of baseline differences was determined by the Chi-square test, Fisher's exact test, or the unpaired t-test, as appropriate. Statistical significance was defined by a two-sided p value < 0.05. The event was defined as an increasing of the tumor residue > 20% as compared to first postoperative MRI for the survival analysis. Survival was estimated by the Kaplan-Meier method. Survival differences were assessed by the log-rank test.

Results
Population study (GKS failure group) ( Table 1) Twenty-three consecutive patients who underwent salvage surgery for large VS after GKS failure were enrolled in this study, including 12 females and 11 males, with a mean age of 56 years (range 28-73). Every patient included in the study showed evidence of progressive tumor growth after GKS (mean tumor growth × 946-range 126-5161). The mean delay between GKS and salvage surgery was 66 months (median 47 months). One patient included in the analysis had been operated on before the delay of 3 years after GKS (23 months) because of worsening symptoms related to tumor growth. This patient had received a radiosurgical treatment for a Koos IV VS. (Table 1) Before GKS treatment, 16 (70%) patients presented with serviceable hearing, including 6 (26%) G&R class 1 and 10 (43%) class 2. All of the patients previously treated with GKS experienced hearing loss related to tumor growth. At the time of salvage surgery, none of the patient presented with serviceable hearing anymore.
At the time of salvage surgery, all patients harbored a Koos IV VS. The mean VS extrameatal diameter was 26 mm (range 19-39/median 25 mm); the mean VS volume was 10 cc (range 4-22/median 8 cc). Four of the tumors (17%) displayed a cystic component. Two (9%) patients presented radiological hydrocephalus (Evans ratio > 0.3); none had received CSF shunting before tumor resection (but three patients had received VP shunting that were placed before radiation).
The overall mean postoperative tumor volume measured on the first postoperative MRI was 0.56 cc (range 0.08-1.50/median 0.52 cc). Two patients (9%) underwent GTR, 9 underwent near total resection (39%), 9 (39%) subtotal resection, and 3 (13%) partial resection. The intensity and frequency of the preoperative neuralgia decreased after surgery, but the patient was not totally freed from this symptom

Histopathological findings
The irradiated VSs were histologically similar to the genuine ones. Antoni A (areas composed of Schwann cells that have a spindle cell morphology) and Antoni B patterns (loosely textured and microcystic areas) were both represented as follows: 2 (9%) tumors showed pure Antoni A pattern, 15 (65%) tumors presented pure Antoni B pattern, while 6 (26%) tumors exhibited both Antoni A and B patterns. As a comparison, 25% of the genuine VSs showed pure Antoni A pattern, 43% presented pure Antoni B pattern, while 32% exhibited both Antoni A and B patterns.
As a reminder, 2 tumors operated on after GKS failure was diagnosed as MNSTs (malignant nerve sheath tumors). Both patients have been excluded from the analysis.

Complications
One patient (4%) presented postoperative meningitis which was cured under medical treatment. One (4%) postoperative hematoma had been treated by revision surgery and external ventricular drain. This patient retained postoperative permanent CN VI paralysis. No CSF leak occurred and no patient required CSF shunting.

Trigeminal nerve
The patients who displayed a preoperative facial numbness were released from this symptom in 12 cases (92%) after surgery, whereas the other one have improved. None of the patients freed from this symptom before surgery has developed a postoperative facial hypoesthesia.

Cochlear nerve
None of the patients presented with preoperative serviceable hearing; none reported recovery after surgery.

Tumor control
The overall mean radiological follow-up was 74 months (range 12-175/median 64 months). Thirteen patients (54%) were scanned for more than 5 years. Of the 21 patients who had undergone non-total resection of their VS, 11 (52%) had been allocated into a Wait-&-rescan policy, and 10 (48%) underwent upfront GKS. The tumor control was achieved in 91% of cases. Two (9%) cases of regrowth were diagnosed during the followup period, respectively, at 32 and 68 months. The 1-, 5-, and 7-year tumor progression-free survival (PFS) Table 2 Postoperative facial nerve function after microsurgical resection of large vestibular schwannomas in GKS-failure and genuine VS populations with normal (House & Brackmann grade I) preoperative function Salvage surgery after GKS failure was not associated to early postoperative (p = .14), 3 months postoperative (p = .10), nor long-term (p = . 25 were 100% (n = 23), 95% (n = 18), and 85% (n = 9), respectively. During the follow-up period, 2 patients displayed pseudoprogression of their tumor remnants. The first one experienced a secondary regression to the postoperative tumor volume within 107 months, while the remnant of the second one remains > 20% than the postoperative tumor volume at the time of analysis and is therefore defined as regrowth.
Both patients harboring growing residues have been allocated under repeated MRI surveillance without additional treatment. None of the patient who received salvage surgery after a first-stage GKS required a second revision surgery.

Comparison to the genuine VS population (Fig. 4 and Tables 2, 3, and 4)
During the study period, 170 patients had undergone microsurgical resection for a large Koos IV VS. The GKS failure and genuine VS groups differed in sex ratio, age, and preoperative VS volume (p < 0.05) (cf Table 3 At last follow-up examination, among the 160 patients with normal preoperative FN function, good FN function (HB grades I and II) was observed in 134 patients (84%), moderate HB grade III deficit in 25 (15%) of patients, while poor FN outcome was observed in one case (1%) ( Table 2). Salvage surgery after GKS failure was not associated to early postoperative (p = 0.14), 3 months postoperative (p = 0.10), nor long-term (p = 0.25) impaired FN outcomes.

Fig. 4
Kaplan-Meier analysis of tumor control during follow-up after subtotal resection in the GKS failure (n = 23) and genuine VS (n = 170) populations who had undergone non-total resection of a large Koos IV vestibular schwannoma during the study period. The 1-, 5-, and 7-year progression-free survival were 100%, 95%, and 85% respectively in the GKS failure group and 97%, 80%, and 81% in the genuine VS group. Log-rank not significant (p = .27)

Discussion
To the best of our knowledge we display the results of the first comparative study about consecutive GKS failure and genuine large VSs populations, treated by the same experienced otoneurosurgical group during the same period. The GKS failure group is made up of a homogenous cohort of patients who underwent the same radiosurgical protocol, were followed up by our team, and diagnosed as failure according to widely accepted criteria.

GKS failure or transient swelling?
Many studies have reported transient swelling remodeling of VSs after GKS treatment that turned out on longer follow-up to be pseudo-progression with subsequent tumor control [1,16,18,24,35]. Transient expansion was reported to have occurred in 15 to 74% of cases, with a mean increase in volume of 20% [35]. Although most of those pseudo-progression resolved within 1 year, a 3-year follow-up time period of continuous growth is recommended before conclude to a GKS failure [1,16,18,24,35]. Anyway, volumetric criteria for the definition of tumor escape vary depending on the volume of the tumor being treated. Indeed, while a tumor volume expansion of more than 100% can be tolerated for intracanalar VSs without further treatment, this range of tumor growth in not acceptable for larger schwannomas. In our study, only one patient harboring a Koos IV VS and who experienced tumor growth after GKS had been operated on only 23 months after the radiosurgical treatment because of worsening symptoms. Furthermore, our attitude has changed with experience. While we were once reluctant to redo a GKS, evidence of tumor progression, especially for Koos I & II VSs, would led us to propose a second stage GKS before the tumor grows up to a Koos IV VS.

Difficulties of microsurgery after GKS failure
A part of the neurosurgical community postulated that low doses delivered on the tumor volume during the GKS procedure would be responsible for more failures, thereby requiring tumor resection significantly more complicated than in genuine VS surgery. Salvage surgery after GKS failure is considered to be associated with operative difficulties related to thickening of the arachnoid membrane, loss of the peritumoral arachnoidal plane, high vascularization or tight adherence to the neurovascular structures [14,19,28,29]. In the recent literature, radical resection of large VSs after GKS failure results in about 50% of permanent FN deficit (HB Grade III to VI) [2,4,20,27]. Alternative strategies of non-total resections have emerged in sporadic large VS surgery for more than 15 years [5,8,9,17,30,32], and previous authors already advocated the need for subtotal resection after GKS failure [2,20].
Although post-GKS arachnoiditis was reported in 87% of our cases in our series, most of the operative dilemmas were related to severe facial nerve or neurovascular structures adherences (35% of cases in the present study). Despite those reported difficulties, the mean operative time and the mean postoperative tumor residue were similar in the GKS failure and genuine VS populations (p = 0.88 and p = 0.70, respectively). Furthermore, we reported in the present study 90% facial nerve function preservation (HB grades 1 and 2) in the long term. Other recent series have reported similar results in patients who underwent subtotal resection after failed GKS [10,11,26,34]. Comparison between the GKS failure and genuine populations did not show FN outcome differences in our series (p = 0.25; Fig. 4, Tables 2 and 4).
Last but not least, two patients (8%) have experienced postoperative complications in the GKS failure group; no difference could be demonstrated with the genuine VS population (p = 0.82). Of note, we did not observe additional CSF dysfunction nor wound complication consecutive to a surgical intervention in a previously irradiated patient.
How to explain the discrepancy between our "good" results and those of the literature?
The department of radiosurgery in our institution is a high volume center, which can legitimately claim a good knowledge of the tumor behavior after GKS. The decision of salvage surgery because of GKS failure was made after careful analysis of the VS progression, by comparing the sequential radiological data. We report in this study the surgical outcome of an experienced otoneurosurgical group familiar with the functional sparing surgery for large VS. Most of reports about salvage surgery after failed irradiation treatment present a lot of weaknesses that convey inappropriate message: the decision-making for salvage surgery in a too short time period in a freshly modified post-irradiated environment; inclusion of patients treated with multiple techniques (radiotherapy, GKS, cyber knife, etc.); and inexperienced neurosurgeons; low volumes of case load.

The fate of the postoperative tumor residues
Leaving some remnant tumor in order to upgrade the functional outcome could potentially expose the patient to regrowth. Herein, we showed that tumor control was achieved in 91% of cases with sequential long-term scanning (mean radiological follow-up of 74 months). Those results are concordant to the recent literature of VS subtotal resection after failed GK [10,11,26,34]. Two (9%) cases of regrowth were diagnosed. One of those cases may be considered a transient swelling of the tumor residue after a proactive postoperative GKS. The patient experienced a 289% tumor volume expansion within 2 years after GKS, followed by a decrease of size, and a + 100% tumor volume expansion 3 years after GKS. The volume of the growing tumor residue of the second patient is less than 0.3 cc. The patient is allocated to a Wait-&-rescan strategy. Salvage surgery after GKS was not associated to a higher rate of regrowth of the tumor residues as compared to the genuine population in the long term (p = 0.27).

Limitations
The main limitation of the study is its retrospective and single center design. Especially when addressing surgical treatment, a single center report does not reflect the outcome that can be expected in other large skull base centers with a high expertise in microsurgical resection of VS. The difference between numbers of patients included in the post-GKS failure and genuine VS groups can also be pointed at. Both populations were not matched in respect of the tumor volumes, and the patients of both groups did not receive the same postoperative treatment regimen (some of the residues were allocated to a Wait-&-rescan strategy; others underwent proactive GKS). The operative findings were subjectively assessed even though we carefully reviewed the video of the surgical procedures.

Conclusions
Failure of treatment is a rare and unpredictable situation after radiosurgery. Evidence of a significant and continuous augmentation of the tumor volume is an indispensable criterion to conclude to failure. Despite significant modifications of the microsurgical environment associated to salvage surgery after GKS failure, a functional nerve-sparing resection is an effective strategy to optimize the results on facial nerve function with no more complications, and similar long-term tumor control to those observed in the genuine VS population. These findings deserve to be shared with neurosurgical teams who are confronted with this situation and provided to the patients at the decision making time. Data availability My manuscript has data included as electronic supplementary material.
Code availability Not applicable.
Consent to participate Informed consent was obtained from all patients.
Consent for publication All patients agreed for publication.

Conflict of interest
The authors declare no competing interests.