Image‐guided robotic radiosurgery for glomus jugulare tumors—Multicenter experience and review of the literature

Glomus jugulare tumors (GJTs) are challenging to treat due to their vascularization and location. This analysis evaluates the effectiveness and safety of image‐guided robotic radiosurgery (RRS) for GJTs in a multicenter study and reviews the existing radiosurgical literature.


| INTRODUCTION
With an estimated incidence of around one per 1.3 million people, glomus jugulare tumors (GJTs) are rare, wellvascularized neuroendocrine tumors arising from the adventitial chemoreceptor tissue of the jugular bulb. 1 They are usually of benign histology but are capable of locally infiltrating adjacent tissue like the lower cranial nerves and the temporal bone. In rare cases, GJTs can secret catecholamines and metastasize to lymph nodes and distant organs, which significantly worsens the disease prognosis. [2][3][4] Due to their location close to the jugular foramen, common symptoms are lower cranial nerve palsies, which lead to dysphagia, dysarthria, pulsatile tinnitus, hearing loss, vertigo, and dysphagia. In hormone-secreting tumors, tachycardia and labile blood pressures are typical findings. 4 Even with the development of microsurgical techniques, surgical tumor resection remains a challenge for surgeons given the location of the tumor, vascularization and adjacent nerves, and vessels. [5][6][7] Despite the lack of treatment guidelines, fractionated radiotherapy and single-session radiosurgery (RS) are alternative treatment options, especially for patients not suitable for surgery. [6][7][8] Due to the rarity of the tumor, many studies investigating RS for the management of GJTs comprised only a small number of patients, additionally included other head and neck paragangliomas, and did not stratify the outcomes for primary and secondary RS. Only a few studies reported long-term follow-up results and only a limited number of studies on the use of image-guided robotic radiosurgery (RRS) are available. To overcome this lack of knowledge and to improve clinical decision making for the primary or postoperative irradiation of GJTs, we conducted a retrospective multicenter study including six centers investigating the use of image-guided RRS for the treatment of GJTs. We also reviewed the radiosurgical literature and compared our results with published data.

| Patients
One hundred and one patients with GJTs from six dedicated CyberKnife (CK) centers were treated with RRS between July 2005 and March 2019 and included in this retrospective multicenter study. Patient information including medical history, previous treatments, and follow-up data were stored at each center in the respective electronic health records or patient files. During follow-up appointments, patients were evaluated for clinical symptoms, adverse effects, complications, and treatment response by clinical examination and by magnetic resonance imaging (MRI). Appointments took place 6 and 12 months after treatment delivery. Follow-up was done after 6 months and in 12 months intervals, thereafter if no acute complications occurred. Only patients with at least one completed radiographic and clinical follow-up 6 months after treatment delivery were included in this analysis. GJT diagnosis was either based on histopathological examination (39 patients) or radiographic findings as well as clinical appearance of the patient (62 patients). For radiographic diagnosis, thin-sliced computed tomography (CT), MRI, as well as contrast-enhanced MR angiography (CE-MRA) imaging were used. This study was approved by the respective institutional review board.

| Treatment procedure and outcome
For treatment planning and delivery, thin-sliced, contrastenhanced CT and MRI scans were used for every patient. Imaging sequences included gadolinium-enhanced T1 and T2 sequences and vessel-focused Time of Flight series. Resulting CT and MRI imaging data were overlaid for treatment planning. Various software tools (MultiPlan, Precision, Accuray Inc., Sunnyvale, California) were used for inverse treatment planning. All patients were treated with single-or multisession (up to 5 fractions) stereotactic RRS in an outpatient setting using a CyberKnife RRS system (Accuray Inc., Sunnyvale, California). During treatment delivery, custom-fitted thermoplastic face masks were used for non-invasive fixation. Tumor volume measurement was done directly with the above-mentioned software assessing the tumor volume on available thinsliced MRI imaging data before treatment and at last available follow-up. Radiographic assessment of the treatment outcome was defined as follows: Tumor volume reduction (TVR), tumor volume decrease of at least 20%, progressive disease (PD), increase of the overall tumor volume of at least 20%, with local control (LC) defined as no evidence of PD during follow-up imaging. Tumors with volume changes ±20% were considered unchanged. To determine potential prognostic factors, logistic regression models were used following a backward selection approach. For normally distributed and paired data, a paired student's t test was conducted. Data were analyzed using STATA 16.0 (StataCorp, College Station, Texas). P-values equal to or less than .05 were considered significant.

| Patient characteristics and treatment parameters
The median age at treatment delivery was 56 years. Most of the treated patients were female (65%) and most of the tumors were located on the left side (60%). The median tumor size was 5.6 cc. Sixty-two out of the 101 (61%) included patients received the treatment as their primary therapy and the majority, 88 out of 101 (87%), were treated in a single session. The median dose for primary and secondary single-session treatments was 16 Gy; the median dose for multisession treatments was 21 Gy in three to five sessions. The median prescription isodose was 70% throughout primary and secondary treatments. The most common pretreatment deficits included pulsatile tinnitus (51%), partial hearing loss (46%), dysphagia (34%), dysarthria (32%), and vertigo (29%). At treatment delivery, five patients (5%) did not show clinical symptoms caused by their GJT. Thirty-nine patients underwent secondary RRS of their tumor. Thirty-eight of them underwent primary surgical resection of their tumor, 10 patients had multiple surgeries up to a maximum of five resections. A summary of the baseline characteristics, treatment parameters, and pretreatment deficits is provided in Table 1.

| Treatment results
The median follow-up time was 35 months (range 6-160). At last follow-up, 23 (24%) patients recovered from their symptoms and 31 (32%) experienced symptom improvement, whereas 35 (34%) reported no significant changes concerning their pretreatment deficits. Five patients (5%) reported a transient worsening of their symptoms before returning to their pretreatment condition at the last available follow-up. Two patients (2%) experienced a persistent worsening caused by one House-Brackmann grade IV facial nerve palsy and one new pulsatile tinnitus after treatment delivery. All patients without pretreatment deficits remained asymptomatic throughout their follow-up, whereas symptom control and improvement between primarily (91%) and secondarily (94%) treated patients were consistent. The overall LC rate was 99%. One patient developed lymph node metastases 4 months after treatment. The primary tumor lesion was controlled at the time of distant failure. Another patient suffered from a local recurrence (PD) after 70.8 months. The recurrent tumor lesion was treated with proton radiotherapy and is controlled since treatment delivery. At last follow-up, 42 tumors (41%) remained unchanged in size, whereas 57 (56%) showed a TVR. Overall, the median and mean volume reduction were 1.77 and 0.8 cc, respectively. These absolute reductions equal median and mean percentage changes of 22% and 24%, respectively. Paired student's t tests among all, primarily treated and secondarily treated patients showed a significant decrease in tumor volume for the three groups at last follow-up ( Table 2). The calculated progression free survival was 97%, 97%, and 93% after 3, 5, and 7 years, respectively. A detailed summary of the treatment results is provided in Table 2.

| Complications and toxicity
Seven patients (7%) showed potential toxicity after treatment delivery. Four reported headaches, vertigo, and nausea. One patient described moderate pain irradiating to the mandibula, neck, and ear on the side of the tumor. All patients were treated with glucocorticoids in an outpatient setting and six of them entirely recovered shortly after. Two patients experienced persistent worsening (House-Brackmann grade IV facial nerve palsy, pulsatile tinnitus). No patient experienced radiation necrosis, seizures, acute bleedings, or radiation-induced malignancies.

Patient characteristics
Radiosurgery plus single surgery plus tumor embolization 1 Note: cc, cubic centimeter, Gy, gray.

| Prognostic factors
Multivariate linear and logistic regression analyses were used to assess prognostic factors for TVR, and symptom improvement, defined as full recovery or pretreatment deficit improvement, as well as toxicity. Pretreatment tumor size was found to be a significant predictor of tumor volume at last follow-up (F [6,94], R 2 = 0.31, P < .01). TVR in cc increased by 0.24 cc for each pretreatment cc of tumor ( Table 3). None of the analyzed factors including age at treatment delivery, sex, pretreatment and posttreatment tumor volumes, indication, dose, prescription isodose, and number of fractions reached statistical significance for symptom improvement (Table 3). Moreover, no significant factors for the occurrence of toxicity after treatment delivery were found (Table 3). Here, a number of fractions were omitted in the analysis as a dependency among the independent variables in the model was identified.

| Literature review
A total of 29 studies have been identified.  One study was a multicenter trial with 132 patients and was counted as a separate publication even though parts of the data had already been reported elsewhere.

| DISCUSSION
Herein, we report the first multicenter study on the role of image-guided RRS in the management of GJTs and the second-largest published radiosurgical series of patients treated for GJTs. The results demonstrate that RRS achieves high rates of tumor and symptom control throughout an intermediate follow-up time. These findings are consistent for primarily and secondarily treated GJTs patients. Moreover, acute complications after treatment delivery are rare and no treatment-related mortality has been observed.

| Local control
As GJT recurrence can occur even after decades, longterm follow-up is needed to determine if our initially reported LC with this sample size is reliable and lasting. 5,38,39 Many reports showed radiosurgical LC rates over 90% with an even more extensive follow-up period. 12,14,19,20,24 Of these studies, Sheehan and colleagues published the first multicenter RS study for GJT with the data of the North American Gamma Knife Consortium. To date, it is the most extensive radiosurgical series available. They included 132 patients with 134 GJTs from eight institutions and reported progression-free survival (PFS) rates of 98%, 90%, and 88% at 1, 3, and 5 years, respectively. 24 The actual LC was 93% at a median follow-up of 50.5 months. 24 Five-year LC rates around 90%-95% seem to be achievable with RS (Table 4). Similar PFS rates have been reported. 14,15,24 Moreover, recent studies including this report have investigated prognostic factors for LC or tumor progression-free survival. 11,13,15,17,20,24 Sheehan and colleagues found significant factors in their multivariate analysis. In their study, a higher number of isocenters and the absence of trigeminal nerve dysfunctions at the time of treatment delivery were associated with progression-free tumor survival. 24 Also, Winford and colleagues identified an increased risk of progression with increased margin treatment doses. 17 Herein, we found that pretreatment tumor volume significantly influences posttreatment tumor size changes. However, identifying prognostic factors remains a difficult task given the rarity and delay of local recurrences. Such factors might unmask with more extensive followup periods and more detailed as well as structured patient assessments. We expect to see more recurrences in our study cohort within the next years.
Besides, the present LC rates for primary and secondary GJT treatments are comparable with the radiosurgical data available, even though most of the data is from GKbased studies due to the availability and history of this radiation technique (Table 4). Past studies have used various definitions of tumor progression, which might compromise comparability. This effect can also be increased by varying tumor volume measurement methods. Given the provided study data and reviewed literature, there seem to be no significant differences in LC in regard to the radiosurgical treatment modality (GK, LINAC, and CK) and number of fractions. Past surgical series might have focused on peripheral GJTs, including those extending to the lower neck. Therefore, comparing our findings with surgical series is challenging due to heterogeneous patient selection and different surgical techniques. LC seems to be at least similar or slightly better for radiosurgical patients compared with surgery. 5,7 Large surgical series with more than 60 patients included report LC rates between 76% and 93%. 5,7,[40][41][42] Ultimately, authors have argued that only a full surgical resection may achieve complete cure. 7 Nevertheless, not all studies provide sufficient details regarding the follow-up time, hindering comparability between RS, fractionated radiotherapy and surgery. Furthermore, the data heterogeneity of GJT patients in past studies makes it difficult to determine patients who would benefit from primary microsurgical resection or fractionated radiotherapy, especially considering recent advancements in surgery and cranial nerve management. 43,44 Besides, fractionated radiotherapy has shown to achieve comparable LC rates and may play an important role for patients not suitable for RS. 5,7 Here, doses up to 45 Gy achieve good clinical results with a limited risk of adverse effects. 7 Even though treatment guidelines including surgical and radiosurgical options have been proposed by colleagues, consensus guidelines for the treatment of GJT are still lacking. 5,7 Still, surgical tumor resection is a treatment option which must be considered, especially for rapidly growing and hormone-secreting tumors. Moreover, some authors emphasized on a more passive approach and suggested to follow a "wait and scan" strategy more frequently. 45 This option is also supported by data showing that even throughout an extensive followup of more than 5 years, 45% of patients experience tumor stability or regression without treatment. 45,46 Finally, radiobiological studies suggest paragangliomas to be relatively radioresistant given the expression of known markers of radioresistance (NOTCH, ZEB1) and downregulation of cellular pathways fostering radiosensitivity (miR-200c, mir-34b/c). [47][48][49] This might support a more passive treatment approach. Overall, the decision for treatment must be carefully evaluated and in agreement with the patient's preferences, clinical status and personal wishes.

| Symptom control and quality of life
GJTs often account for considerable morbidity and significantly impact the quality of life (Qol). 50 Thus, patientreported outcomes and projected symptom control are crucial when determining the most suitable treatment option for patients besides overall performance status and patients' preferences. In this study, 93% of the treated patients, regardless of undergoing primary or secondary RRS, experienced symptom control or improvements in their pretreatment deficits. Considering the sample size and follow-up time, this proportion underlines the effectiveness of RRS and demonstrates the equivalence in regard to GK-based treatments (Table 4). However, reported data are heterogeneous and some authors examined the changes in pretreatment deficits for every symptom separately as well as used quantitative assessments like audiograms. Most studies analyzed in our literature review reported symptom control rates of ≥80%, with an estimated overall rate of 88.8% (442/498 patients, Table 4). Notably, these rates were mostly consistent regardless of sample size, tumor volume, follow-up period and radiation modality (Table 4). Thus, RS achieves reliable clinical results, especially in contrast to most invasive treatments. Surgery, despite achieving comparable LC rates, still yields a considerable risk for persistent or newly developing cranial nerve palsies, thus, limiting the chances of symptom control. 5,7 However, less invasive surgical methods seem to lower the rate of newly developing postsurgical cranial nerve deficits. 43,44 Today, variables indicating clinical outcomes are still lacking. Various reports including this study investigated prognostic factors for symptom control or clinical progression-free survival after treatment. 11,15,16,20 Sallabanda and colleagues found cranial nerve involvement to be significantly associated with a decreased chance of symptom improvement. 11 However, the association was apparent in univariate analysis, a multivariate analysis was not conducted. Besides, Wakefield and colleagues found prior surgical resection to significantly correlate with persistent neurological deficits compared with nonsurgical cases. 16 In contrast, our study found no prognostic factors for symptom improvement. Besides reporting subjective symptom control, analyses of standardized Qol measurements before and after treatment could improve clinical decision making. However, only sparse standardized Qol data are available in the GJT literature. [51][52][53] Galland-Girodet and colleagues found better scores for hearing and speech, trismus, and overall Qol for patients with head and neck paragangliomas receiving radiotherapy alone vs those receiving radiotherapy and surgery. 51 Patel and colleagues investigated Qol outcomes for primary and secondary GJT patients after GK-based RS. 52 Patients undergoing primary RS had better swallowing function than patients who underwent surgical resection before. 52 One study investigating fractionated radiotherapy showed stable SF36 Qol measures for patients undergoing primary radiotherapy. 53 Finally, a recent single center study utilizing pretreatment and posttreatment SF12 data showed consistent Qol improvements after RRS. 54 Overall, these findings suggest that RS and radiotherapy seem to be favorable in regard to posttreatment Qol in comparison with surgery. Despite the lack of data and sufficient studies, projected Qol outcomes after treatment should play an important role when choosing the right treatment modality.

| Complications and toxicity
In addition, complications and treatment-related morbidity are linked to the clinical outcome of patients and have a significant impact on clinical decision making. In our study cohort, only a few transient low-grade and persistent complications have been observed. This is also in agreement with the existing literature (Table 4). Complication rates for RS range from 0% to 28%; most of the studies (62%) had rates below 10%, with an overall rate of 8.8% throughout all studies. Our complication rate (7%) is in agreement with the reviewed literature. Most complications were low grade, ranging from nausea, vertigo, headaches to transient cranial nerve deficits, and resolved after a short amount of time. However, we were not able to find predictors of toxicity in regard to applied doses. Given the low rate of complications after RS, the majority of other studies have not investigated the relationship between dose and toxicity. However, most of the GJTs have been treated with doses around 16 Gy and lowgrade complications occurred in less than 10% of patients (Table 4).
In contrast to surgery, RS has an advantageous complication profile with a lower risk, especially for posttreatment cranial nerve deficits. 5,7 Fractionated radiotherapy seems to have a slightly higher risk for complications, 10.4% and 6.5%, respectively. 7 In contrast, complications after surgery are much more frequent and severe. A review by Lieberson and colleagues reported complication rates of ≥46% after surgery. 5 Suárez and colleagues reported major complications including CSF fistulas, aspirations, infections, meningitis, strokes, and death in 28% of surgical cases. 7 These reviews underline the considerable complication risk for gross GJT resections. Even though the risk for radiation-associated longterm complications like radiation-induced malignancies is low, RS should be especially considered for older patients. Overall, candidates for GJT treatment should be well-informed about the potential complications and toxicity of RS, fractionated radiotherapy and surgery.

| Limitations
This study has various limitations. The retrospective nature, due to selection and reporting biases, is an inheriting limitation of the study and as most of the included patients underwent RS as their primary GJT treatment, histological confirmation was only performed in 39 out of 101 patients. Nevertheless, the radiological diagnosis is considered reliable and accurate, often because of a characteristic contrast enhancement and especially in conjunction with typical symptoms of GJTs and modern imaging modalities (CT, MRI, and CE-MRA). [55][56][57][58] In regard to the literature and tumor biology of GJTs, the follow-up time of our study is too short to detect late tumor recurrences reliably. These can occur even decades after treatment delivery. 5,8,38,39 While our initially reported LC is high, we expect to see more and more local recurrences five or more years after treatment delivery. Moreover, our PFS might be too high given insufficient return of patient information regarding deaths by any cause. This might be especially the case for outpatient only treatment facilities and centers. Finally, we only reported the qualitative data for our posttreatment symptom analyses in this study and did not provide audiograms or House-Brackmann scores. This circumstance might limit the validity of our posttreatment symptom results. However, many of the published studies have used this way of reporting to depict their findings.

| Current and future challenges
So far, only limited progress has been made in evaluating multimodal and interdisciplinary treatment options despite the amount of retrospective reports. RS has shown at least equal to favorable results compared to fractionated radiotherapy and surgical series in regard to LC, symptom control and toxicity over the past decades. 5,7 However, reporting heterogeneity, patient selection and technical advancements in all fields warrant a more detailed and comprehensive analysis. It is essential to conduct more studies on the primary and secondary treatment of GJTs and to evaluate multidisciplinary treatment options critically. Comparative studies, ideally of prospective nature, could help to establish treatment guidelines and determine which patients could potentially benefit from surgical resection or fractionated radiotherapy. Currently, there is a lack of knowledge on how to maximize symptom control and posttreatment Qol. A prospective international multicenter study with standardized outcome evaluations including audiograms, video-head impulse tests, caloric vestibular tests, extensive lower cranial nerve testing, and Qol assessments could overcome the epidemiologic challenges and improve clinical care for GJT patients. Moreover, such a study should follow an interdisciplinary approach and include the expertise of neurosurgeons, neurologists, radiation, and otolaryngologist. Yet, it is important to note that the various treatment options available, the variety of patients in regard to neurological deficits, pretreatments, and tumor size as well as the low tumor incidence pose considerable challenges to conduct prospective trials. In addition, radiosurgical and multidisciplinary treatment options for other paragangliomas should be investigated as well and may be included in one large prospective trial. Considering our findings, similar long-term experiences with GK, and the possibility to treat extracranial lesions, RRS may be suitable for the treatment of other head and neck paragangliomas than GJTs. 59

| CONCLUSION
This multicenter study is the largest series investigating the use of RRS for GJTs and, overall, the second-largest radiosurgical series for the treatment of GJTs. Results