Long-term outcomes of stereotactic radiosurgery for skull base tumors involving the cavernous sinus

Stereotactic radiosurgery (SRS) is an effective and less invasive therapeutic option for cavernous sinus (CS) tumors. However, its long-term effectiveness and neurological outcomes have yet to be fully elucidated. We aimed to examine the long-term outcomes of SRS for CS tumors. Overall, a cohort of 113 patients with benign CS tumors, including 91 with meningioma, 14 with trigeminal schwannoma (TS), and eight with cavernous hemangioma, treated with SRS at our institution from 1990 to 2018, was included. Tumor control and functional preservation/recovery were evaluated in detail. The median post-SRS follow-up period was 77 months (interquartile range, 39–177). Progression-free survival (PFS) was 97% at 5 years, 89% at 10 years, and 87% at 15 years for the entire cohort; 96% at 5 years and 87% at 10 years for meningiomas; and 100% at 10 years for the other tumors. No significant difference was observed between meningiomas and non-meningiomas (log-rank test, p = 0.107). Improvement in cranial nerve (CN) function was observed in 35 (27%) patients. TSs tended to show CN improvements more often than meningiomas did (total improvements, 62% vs. 23%; p = 0.004; eye movement function, 100% vs. 20%; p = 0.002). CN deterioration or development of new CN deficits was observed in 11 (10%) patients. SRS provides good tumor control and acceptable long-term outcome with sufficient preservation of CN function in patients with benign CS tumors.


Introduction
The cavernous sinus (CS) is an important anatomical structure containing the internal carotid artery (ICA) and the third, fourth, fifth, and sixth cranial nerves (CNs). A variety of benign tumors, such as meningiomas, trigeminal schwannomas (TS), and cavernous hemangiomas (CH), can arise within or extend into this structure, causing impairment of visual function, extraocular movement, facial sensory function, and other CN functions [1][2][3]. Surgical resection is the standard primary treatment to achieve immediate mass reduction for large tumors and also obtain a histopathological diagnosis although it is possible to make an accurate diagnosis based only on the characteristics of advanced neuroimaging findings in most cases. Despite well-established microscopic and endoscopic skull base techniques, surgical interventions for such tumors remain challenging due to their deep skull base location and proximity to the ICA, CNs, visual pathways, and pituitary gland. Preservation of a CN may require surgeons to leave tumor remnants behind, and tumor recurrence/regrowth is possible [4][5][6]. Radiotherapy plays an important role in balancing tumor control and functional preservation, but whether functional recovery is achievable following radiotherapy remains questionable, especially in cases of large symptomatic tumors, since immediate mass reduction is not achievable [5].
Stereotactic radiosurgery (SRS) is a less invasive treatment option utilizing head fixation and highly focused narrow beam radiation that enables precise targeting with a 1 3 steep dose fall-off. Given the structural features of the CS in which the CNs run along its outer wall, it is theoretically feasible to intensely irradiate tumors while minimizing irradiation to the CNs. Previous literature has reported favorable short-term to mid-term outcomes following either SRS alone or in combination with surgery (5-years tumor control rates in CS meningiomas ranging 94-98%) [7][8][9][10]. CS-specific radiation-induced adverse events (RAE), deterioration of CN III-VI functions, and ICA stenosis/occlusion, albeit rare, have been reported [11][12][13][14][15][16][17]. However, there remains a paucity of data on its long-term outcomes.
This study aimed to clarify the long-term outcomes of tumor control and CN functioning following SRS for benign CS tumors.

Patient and tumor characteristics
The data of 190 patients with CS tumors, treated with SRS from June 1990 to June 2018 at our institution, were collected from the institutional gamma knife database. Both intra-CS tumors and para-CS tumors extending into the CS were defined as CS tumors. The exclusion criteria were: (1) functioning (n = 22) and non-functioning (n = 14) pituitary adenomas with CS extensions, (2) pathologically confirmed non-benign tumors, including World Health Organization (WHO) grade II/III meningiomas (n = 12), hemangiopericytomas (n = 1), chordomas (n = 18), chondrosarcomas (n = 9), and metastatic tumors (n = 1). As a result, data on 113 patients with benign CS tumors, including 91 meningiomas, 14 TSs, and 8 CHs, were included in the analysis.
Most diagnoses were based on histopathological findings from prior surgery (n = 75). The timing of SRS for postoperative recurrent (n = 31) or residual tumor (n = 44) was determined using either the judgment of the surgeon, a referring physician, or by patient request, without any arbitrary selection. Radiographic diagnosis, without prior surgery, was used in 38 cases (19 meningiomas, 12 TSs, and 7 CHs). All radiographic images were reviewed by two independent neuroradiologists and attending neurosurgeons. The study was approved by the Institutional Review Board of our institution (#2231) and conducted in compliance with the principles of the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. All patients provided written informed consent for study participation.

The procedures and techniques of SRS
Leksell Gamma Knife (Elekta Instruments, Stockholm, Sweden) was used for all SRS treatments. The detailed treatment process is reported in a previous paper [18,19]. After head fixation using a Leksell frame (Elekta Instruments, Stockholm, Sweden), stereotactic imaging (computed tomography [CT] before July 1996; magnetic resonance imaging [MRI] between August 1996 and January 2018, followed by conebeam CT) was performed to obtain precise tumor data. Neurosurgeons and radiation oncologists performed radiosurgical planning using commercially available software (KULA planning system) until 1998 and Leksell Gamma Plan thereafter (Elekta Instruments). In principle, 14-16 Gy was prescribed to the tumor margin using a 50 ± 5% isodose line. If the tumor was suspected to be aggressive and was sufficiently distant from radiosensitive cerebral structures, such as the optic apparatus or the brainstem, the tumor margin dose was increased to 16 Gy to achieve long-term tumor control.

Follow-up and treatment outcomes
After SRS, MRI was regularly performed every 6 months for the first couple of years and annually thereafter. Radiographic findings were independently assessed by neuroradiologists and neurosurgeons. Tumor progression or shrinkage were defined by the Response Assessment in Neuro-Oncology (RANO) criteria [20]. Transient expansion, typically occurring in schwannomas due to radiationinduced tumor swelling at around 6 months after SRS, followed by shrinkage at approximately 18 months, was meticulously distinguished from actual tumor progression by evaluating consecutive MRIs [21,22]. The neurological status of the patients and their responses to treatment were prospectively collected at each hospital visit, and a Common Terminology Criteria for Adverse Events (CTCAE, version 5.0) grade was retrospectively assigned on the basis of the respective descriptions. Data on patients who dropped out of regular follow-ups or returned to referring physicians were collected via telephone conversation, and follow-up radiographic images were obtained for our independent review. Radiosurgical plans and follow-up images of typical cases are shown in Fig. 1.

Statistical methods
Baseline characteristics of the patients were compared using the chi-square test for categorical variables and the Mann-Whitney U test for continuous variables. Progression-free survival (PFS) rates were calculated using the Kaplan-Meier method and compared among tumor types using the log-rank test. Factors associated with PFS were examined using bivariate and multivariable Cox proportional hazard analyses. Continuous variables were entered into models after being dichotomized using their median values. Where post-SRS recurrence/regrowth was confirmed, recurrence patterns and features were examined in more detail. Post-SRS CN outcomes were summarized, and factors associated with functional improvement, deterioration, and new deficits were examined with logistic regression analysis, and these rates were calculated and compared using the Kaplan-Meier method. Statistical analyses were performed using JMP Pro 15 software (SAS Institute Inc., Cary, NC, USA).

Participant characteristics
Patient baseline characteristics are shown in Table 1   . When comparing the baseline characteristics between tumor types, the maximum diameter (29 mm vs. 24 mm, p = 0.029) was significantly larger, and the prescription dose (16 vs. 14 Gy, p = 0.012) and central dose (32 vs. 28 Gy, p = 0.039) were significantly higher in meningiomas than in TS. Patients underwent prior surgery significantly more often in the meningioma group than the TS and CH groups (meningioma, 79%; TS, 14% [p = 0.001]; and CH, 13% [p = 0.001]). Of those, 21 patients with meningioma (23%) had undergone surgery two or more times before SRS.

Tumor control
Of all the patients, 112 (99%) were alive at the final follow-up visit, and the single (1%) death was due to suicide, unrelated to the tumor and associated treatment. At the last follow-up, 49 (43%) tumors had decreased in size, 54 (48%) remained unchanged, and 10 (9%) increased in size. Tumor shrinkage was observed in 32 patients (35%) with meningioma, 10 (71%) with TS, and seven (81%) with CH. Tumor progression was not observed except for 10 patients with meningioma. In the entire cohort, the PFS was 97% at 5 years, 89% at 10 years, and 87% at 15 years ( Fig. 2A). The tumor specific PFS was 96% at 5 years, 87% at 10 years for meningioma, and 100% at 10 years for the other tumors. There were no significant differences between two cohorts of meningioma and non-meningioma in PFS (log-rank test, p = 0.107; Fig. 2B). PFS was 100% at 5 years and 90% at 10 years with SRS alone and 96% at 5 years and 87% at 10 years for SRS with prior surgery (log-rank test, p = 0.056; Fig. 2C).
Since tumor recurrence was only observed in meningiomas, the analysis of potential risk factors for tumor recurrence was performed for meningiomas. No significant factors were found in the bivariate and multivariable analyses ( Table 2). Baseline characteristics of patients with post-SRS recurrence are summarized in Online Resource Supplementary Table 2. All recurrences were noted for meningiomas at a median period of 87 months (IQR, 48-160 months) after SRS. Nine (90%) tumors were postsurgical recurrence, and three (30%) tumors were treated with suboptimal radiation coverage because of proximity to the optic apparatus or brainstem structure. The patterns of recurrence were intra-field in four (40%) patients and marginal (recurrence occurred out of the radiation field but within 20% isodose line) in six (60%). PFS in meningioma was 100% at 5 years and 89% at 10 years for SRS alone and 96% at 5 years and 87% at 10 years for SRS with prior surgery (log-rank test, p = 0.207; Fig. 2D).

Improvement of cranial nerve deficits
The details of CN function are summarized in Table 3 and  Online Resource Supplementary Table 3. 83 patients (73%) had 128 CN deficits before SRS. The rate of pre-SRS CN deficits was the highest in meningiomas (69/91, 76%), followed by 12 (86%) in TSs, and two (25%) in CHs.
In the entire cohort, CN improvement was observed in 35 (27%) CNs, including four (18%) with visual deficits, 11 (24%) with third, fourth, and sixth CN deficits, 16 (36%) with trigeminal neuropathy, and 1 (50%) with ptosis. Among tumor types, improvement of third, fourth, and sixth CN deficits was significantly more common in patients with TS (100%) than in meningioma (20%, p = 0.002), whereas no significant difference was observed in improvements in the other CN deficits. The cumulative rates of post-SRS improvement of CN functions are shown in Fig. 3A. The post-SRS CN improvement was observed at a median period of 13 months (IQR, 6-24 months). More significant improvements were observed in nonmeningiomas than in meningiomas (log-rank test, p = 0.002; Fig. 3B). The improvement rates of CN V (57% vs. 26%, p = 0.042) and all CN (41% vs. 22%, p = 0.035) were significantly higher in SRS alone than in SRS with prior surgery (Online Resource Supplementary

Radiation-induced adverse events
Details of new or worsened CN deficits following SRS are shown in Table 3 and Online Resource Supplementary Table 3. 11 (10%) CN deficits had deteriorated or newly developed at a median period of 4 months (IQR, 3-13 months) following SRS, including 1 (1%) visual deficit, four (4%) extraocular movement disorders, and six (5%) trigeminal neuropathies. All were mild or transient (CTCAE grade 1-2). The cumulative rates of post-SRS deteriorated/newly developed CN functions are shown in Fig. 3C. There were no significant differences between meningioma and other tumors (Fig. 3D). Aside from the CN deficits, hydrocephalus was observed in one patient (0.9%) with well-controlled meningioma who underwent ventriculoperitoneal shunting 100 months after SRS. In this case, the association between the tumor and hydrocephalus was unclear. No temporal lobe necrosis was observed in the entire cohort. Post-SRS signal change, as a high-intensity signal change on T2-weighted imaging in the temporal lobe, was identified in two patients with meningioma (1.6%) 3-6 months after SRS. Both were asymptomatic and the signal change diminished at 13-24 months. Hypopituitarism was observed in one patient (0.9%) with meningioma 47 months after SRS, and

Discussion
In this study, we analyzed the long-term outcomes of SRS for CS tumors. Using a similar radiosurgical strategy, we defined the tumor margins and prescribed conformally and selectively a marginal dose of 14-16 Gy. The 10-year PFS was 89%, with the post-SRS CN improvement rate reaching 27%, while maintaining a low CN deterioration rate (10%). These results suggest that SRS would be a reasonable treatment option, providing long-term tumor control with favorable neurological outcomes.
The results showed that the tumor control rate is excellent, especially in non-meningioma CS tumors. Most baseline characteristics of tumors in this cohort were similar, with differences in tumor size and the history of prior surgery. The differences in PFSs between meningioma and nonmeningioma tumors possibly reflected differences in tumor biology, which might be due to selection bias in that most meningiomas are postoperative cases. In the literature, tumor control rates were reported to be 77-100% in TS at median observation periods of 27-91 months, and 100% in CH at 30-40 months [12][13][14][15][16][17][23][24][25][26][27][28]. Our data are comparable to these studies. The PFS in the SRS with prior surgery group showed lower values than that of the SRS alone group. The inter-group differences were not statistically significant, but this may be due to the lack of statistical power and should be treated with caution.
The post-SRS CN improvements were also excellent, likely the result of the high tumor shrinkage rate. Regarding RAEs, our study demonstrated excellent functional preservation rates, which might be due to the highly selective high dose irradiation which SRS can provide. Therefore, nonmeningioma CS tumors are an excellent indication for SRS.
Treatment strategies need to be formulated according to the characteristics of individual tumors. SRS plays an adjunctive role for recurrent/residual tumors and can be a primary treatment modality in TS and CH.
The PFS of meningioma was slightly lower than that of non-meningiomas but was still satisfactory at 96% at 5 years and 87% at 10 years. Most previous retrospective studies and the International Stereotactic Radiosurgery Society Practice Guidelines reported excellent tumor control rates of 86-99% at 5 years and 69-97% at 10 years with favorable functional preservation rates of 80-100%, consistent with our results [18,[29][30][31][32][33][34][35][36][37][38][39][40]. We found that six (60%) recurrent meningiomas presented with marginal recurrences after a mid-to-long period of tumor control (range, 67-208 months). Recurrences can be explained by the intrinsic features of meningiomas. They easily blend into the meninges, making it challenging to accurately define tumor margins. It is important to meticulously pursue dural tails using thin-slice MRIs. In post-surgical recurrence/regrowth cases, surgical scar tissue might obscure the true tumor margins. Nine (90%) recurrences were postoperative cases, and in situ comparisons to pre-operative images may highlight the true extent of the tumor. Although the role of surgery remains debatable, and immediate mass reduction may be necessary in certain cases, primary SRS may be reasonable for selective cases with small-sized to medium-sized CS tumors, unless the tumor has atypical radiographic features. Four (40%) cases had in-field recurrence, and three were likely explained by suboptimal coverage because of the proximity of the optic apparatus and brainstem or incomplete coverage of a tumor part extending into the tentorial edge. The remaining case had a WHO grade I tumor with biologically aggressive features, including a Ki-67 index of 15%.
CN dysfunction as post-SRS RAEs has occurred in 1-23% of patients in the past [7-9, 13-18, 23, 24, 26, 27, 30-34, 37-42]. In our study, 11 of all 113 patients (10%) experienced deterioration or emergence of new CN dysfunction after SRS, which was lower than previously reported (Online Resource Supplementary Table 5). Notably, 10 of 11 (91%) of these occurred in patients with CS meningioma. There was no significant difference among tumor types, and all of the dysfunctions were either CTCAE grade 1 or 2. Comparing surgery and SRS, a meta-analysis of 2065 CS meningiomas showed that the incidence of neurological complications was significantly lower with SRS alone (25.7%) compared with SRS after surgery (59.6%) [43]. We failed to find a similar pattern, which could be explained by our relatively small patient number, although it might be due to our lower RAE rate. A certain portion of RAEs may be avoidable by meticulously defining tumor margins and reducing direct irradiation to the CNs using thin-slice MRIs. Based on these results, SRS would be a safe modality for preserving CN function in CS tumors. Aside from CN deficits, we also identified one carotid artery stenosis (0.8%). In previous reports on SRS for pituitary adenoma and CH, the incidence of radiation-induced ICA stenosis/ occlusion range was 0.4%-2.8% [44,45]. Although ICA stenosis/occlusion is rare and rarely becomes symptomatic, long-term follow-up is needed.
This study has several limitations. First, it was a retrospective, single-institution study with potential selection bias. In addition, the clinical practice standards specific to the institution were used, thereby impairing the generalizability of the findings. Second, 38 tumors in this cohort were radiographically diagnosed, therefore the certainty of these radiological diagnoses could be less reliable than those with histological confirmation. Finally, a larger sample size for each tumor type would be desirable for future studies to confirm our findings.

Conclusion
We re-confirmed that SRS achieved excellent treatment efficacy for benign skull base tumors invading the CS. It could achieve a valid tumor control and an acceptable long-term outcome with sufficient preservation of CN function.