The long-term outcomes of radiosurgery for intracranial hemangioblastomas

Abstract

Background

Optimal timing in the treatment of intracranial hemangioblastoma (HB) remains controversial, particularly for patients of von Hippel–Lindau disease (VHL) with multiple small lesions. We evaluated efficacy of stereotactic radiosurgery (SRS) for intracranial HB based on the longer follow-up data with a larger number of patients and lesions.

Methods

Twenty-one patients (11 men, 10 women) initially underwent SRS for 57 intracranial HBs. Seven patients had sporadic lesions and 14 had VHL-related lesions. During the follow-up, 40 lesions were additionally treated in VHL patients in 10 SRS treatments. Thus, a total of 97 lesions were included in this study. Median tumor volume was 0.13 cm³ (range, 0.004–9.5 cm³), and median margin dose was 18 Gy (range, 14–20 Gy).

Results

Median duration of follow-up was 96 months (range, 3–235 mo) after initial SRS treatment. Ten tumors in 7 patients showed progression after SRS. Actuarial tumor control rates after SRS at 5 and 10 years were 94% and 80%, respectively. Factors associated with longer control were solid lesion (P = .03), smaller volume (P = .01), and lesions associated with VHL (P = .0005) in univariate analysis. Five- and 10-year tumor control rates were 67% and 44% for sporadic patients and 97% and 83% for VHL patients.

Conclusion

SRS could be offered as an effective treatment for small, solid, and VHL-associated HBs. If the tumors show apparent enlargement in size or can possibly become symptomatic along with a slight increase in size, SRS should be recommended before they present with the clinical symptoms.

Hemangioblastoma (HB) is a rare benign tumor of the central nervous system that usually occurs in the posterior fossa.¹ About 20% of cases of intracranial HB are associated with von Hippel–Lindau disease (VHL),² and ∼80% of cases of spinal cord HB develop in patients with VHL.³ For intracranial HB, surgical resection is an indispensable modality for lesions causing higher intracranial pressure and various associated symptoms. On the other hand, for patients with VHL who generally harbor multiple intracranial lesions, repeated surgical procedures are less agreeable and surgery should be reserved as a salvage modality. Recently, clinical studies with a large number of patients with VHL have been reported from the National Institutes of Health.^4–7 In those reports, all the silent lesions were principally recommended to be observed conservatively, since HB usually shows a slow and stuttering pattern of growth.⁴ In addition, based on these results, the published reports at that time regarding outcomes of stereotactic radiosurgery (SRS) showed preferable results, but the follow-up periods reported were insufficient to declare “tumor control.”^4,7 In fact, we have previously reported early results of SRS for 13 patients with 38 HBs and concluded that SRS would be safe and effective, although a limited number of the patients were observed over 10 years and long-term results were still unclear.⁸ From this point of view, the present study reconsidered the outcomes of SRS for intracranial HB, factors associated with tumor control, and adverse events with a larger number of the patients and the lesions in the longer follow-up periods.

Materials and Methods

Patient Population

A retrospective database search was carried out. Between January 1991 and June 2010, a total of 21 patients (11 men, 10 women) with intracranial HBs were treated by SRS using gamma knife surgery (GKS) at our institute. Median age at the time of initial SRS was 41 years (range, 19–84 y). One case of HB was diagnosed based on a family history of VHL and the presence of multiple cerebellar tumors on MRI. The remaining 20 patients underwent at least one surgical procedure, and HB was pathologically diagnosed according to World Health Organization criteria. The diagnosis of VHL was judged based on the following criteria.⁹ When the patient had a family history of VHL, the diagnosis of VHL was based on the presence of at least one typical VHL-related tumor, retinal or CNS HB, renal cell carcinoma, pheochromocytoma, or pancreatic tumor. In the presence of 2 or more CNS HBs, the diagnosis of VHL was made regardless of family history. In the present series, VHL was diagnosed in 14 patients, and the other 7 showed sporadic intracranial HB. The treatment strategies for patients with HBs were discussed jointly among neurosurgeons, radiation oncologists, and radiologists at neurosurgical conferences. One patient with sporadic HB had already received fractionated radiotherapy of 30 Gy in 15 fractions and was referred to us after recurrence. Except for this case, all patients underwent SRS as initial radiotherapy for intracranial lesions. The internal review board of the University of Tokyo Hospital approved the study protocol, and written informed consent was obtained from all subjects prior to participation.

Patients and Lesion Characteristics

The characteristics and clinical features of patients are summarized in Table 1. Median duration of follow-up was 96 months (range, 3–235) after the first SRS. Fourteen patients were observed for >5 years. For the patients with VHL, the median follow-up period was 125 months (range, 18–235). Three patients died. One patient with sporadic HB died from lung cancer 3 months after SRS. This patient did not undergo radiographic evaluation after SRS. A second patient with sporadic HB died from maxillary cancer at 77 months after SRS. Another VHL patient died of respiratory failure due to renal cell carcinoma 35 months after SRS.

Table 1.

Characteristics and radiosurgical dosimetry of 21 patients with intracranial HBs treated by GKS

	Sporadic Cases	VHL-associated Cases	Total
No. of patients (male/female)	7 (2/5)	14 (9/5)	21 (11/10)
Median age at time of SRS, y (range)	66 (42–84)	34 (19–64)	41 (19–84)
Median observation period, mo (range)	43 (3–147)	125 (18–235)	96 (3–235)
No. of lesions	7	90	97
Previous surgery	7	13	20
Location of tumor
Cerebellar hemisphere	4	65	69
Cerebellar vermis	2	15	17
Brainstem	1	9	10
Temporal lobe		1	1
Radiosurgical parameter
Median target volume, cm3 (range)	2.9 (0.9–9.5)	0.1 (0.004–2.9)	0.13 (0.004–9.5)
Median marginal dose, Gy (range)	18 (14–20)	18 (18–20)	18 (14–20)
Median max dose, Gy (range)	40 (28–45)	40 (36–45)	40 (28–45)

When patients were referred to our institute for SRS, 57 lesions underwent initial treatment. For 6 patients with VHL, 40 additional lesions subsequently received treatment after increasing in size or appearing as new lesions during follow-up. In total, 97 lesions were treated in a series of 31 GKSs. Tumors were located in the cerebellar hemisphere (n = 69), cerebellar vermis (n = 17), brainstem (n = 10), and temporal lobe (n = 1). Median individual tumor volume was 0.13 cm³ (range, 0.004–9.5). At the time of initial SRS, clinical symptoms were ataxia (2 patients), imbalance (2 patients), facial paresis (2 patients), dizziness (1 patient), facial hyperesthesia (1 patient), and dysarthria (1 patient). Among 14 patients with VHL, 6 harbored spinal cord lesions. Among them, 2 patients underwent surgical resection before SRS and 4 were conservatively observed. During the follow-up period after SRS for intracranial lesions, 1 patient had a tumor in the lumbar region that showed enlargement and became symptomatic and required surgical resection. In another 5 patients, spinal lesions were observed simultaneously with intracranial lesions, which remained stable during the observation period.

Radiosurgical Treatment

After a Leksell frame was fixed to the head of the patient, stereotactic imaging was performed to obtain precise information on the shape, volume, and 3-dimensional coordinates of tumors. CT was used before July 1996. MRI using 3D magnetization-prepared rapid acquisition gradient echo sequences was utilized for stereotactic imaging after that. Treatment planning was performed by a neurosurgeon and radiation oncologists using commercially available software (until 1998, Kula; thereafter, Leksell GammaPlan, Elekta Instruments). Median maximum dose was 40 Gy (range, 26–45) and median margin dose was 18 Gy (range, 14–20). In principle, we treated lesions at the time tumor progression was detected on follow-up imaging.

Follow-up Evaluation and Statistical Analysis

After SRS, follow-up clinical examinations were performed at our hospital or by referring physicians. Patients underwent MRI or CT with contrast enhancement every 6 months. Images were separately evaluated by neurosurgeons and radiologists. Follow-up images were measured in the axial, coronal, and sagittal planes and compared with images from the day of SRS. Tumor volume was determined as length × width × height × 0.5. Response to treatment was determined by last follow-up images as stable, regression, or progression. When a tumor showed a 25% decrease or increase in volume, the response was defined as regression or progression, respectively. Actuarial tumor control rate was calculated using the Kaplan–Meier method. The tumor control rate was calculated from the date of first GKS to the date on which disease progression was confirmed on radiographic images. The period of tumor control was censored at the time of the last radiographic follow-up. Factors potentially affecting tumor control were evaluated by log-rank testing for univariate analysis. Continuous variables were dichotomized using median values. Multivariate analysis was performed using Cox regression analysis. All statistical analyses were performed using JMP 9 software (SAS Institute).

Results

Tumor Control

Limited to the 57 lesions treated at first SRS and followed up for a median of 96 months, actuarial tumor control rates at 3, 5, and 10 years after initial SRS were 92%, 92%, and 80%, respectively (Fig. 1).

Fig. 1.

At the time of initial GKS procedures, 57 lesions were treated and followed for a median of 96 months. The actuarial local control rates were 92% at 3 years, 92% at 5 years, and 80% at 10 years, in the Kaplan–Meier method.

In 6 patients with VHL during follow-up, 40 additional lesions were treated, since some tumors increased in size or appeared as new lesions. In the entire series, 97 lesions were treated, and at the last follow-up, 56 tumors (58%) showed stable appearance and 31 tumors (32%) decreased. On the other hand, 10 tumors (10%) in 7 patients showed tumor progression (3 sporadic lesions in 3 patients and 7 VHL-related lesions in 4 patients). Of these, 4 tumors in 4 patients with VHL showed increased size, with cyst enlargement at 3 to 148 months (median, 61 mo) after radiosurgery, and the other 6 tumors showed tumor enlargement at 6 to 102 months (median, 52 mo) later. In the entire series, for a median 60 months follow-up, actuarial tumor control rates of all 97 lesions at 3, 5, and 10 years after SRS were 94%, 94%, and 80%, respectively. In univariate analysis, factors associated with successful local control were solid lesions (P = .03), tumor volume <0.13 cm³ (P = .01), and VHL-related lesions (P = .0005), whereas marginal dose >18 Gy (P = .27) and maximum dose >40 Gy (P = .72) did not have a significant effect on the tumor control rate in our series (Table 2). In further analysis, the factor associated with tumor control in multivariate analysis was VHL-related lesions alone (P = .05). For 90 lesions in 14 patients with VHL, actuarial tumor control rates at 5 and 10 years were 97% and 83%, respectively. On the other hand, actuarial tumor control rates at 5 and 10 years for 7 patients with sporadic HB were 67% and 44%, respectively. Fifty VHL-related lesions were initially treated with SRS and were followed for 18 to 235 months (median, 125 mo). In these, actuarial tumor control rates at 5 and 10 years were 96% and 81%, respectively.

Table 2.

Factors associated with tumor control after GKS

Factor	P
	Univariate Analysis	Multivariate Analysis
Solid lesions	<.001*	.22
Tumor volume <0.13 cm3	.01*	.06
VHL-related lesions	.0005*	.02*

*Significant at P < .05.

Actually, it is very difficult to legitimately distinguish between true tumor progression and radiation-related pseudoprogression, even radiographically or histologically, when tumors show enlargement in relatively early periods after SRS. In our institution, when enlargement of lesion was observed, radiographic evaluation was performed in the short intervals. For symptomatic cases, we principally treated patients with oral corticosteroids at first. However, when they continued to deteriorate or the lesions continued to enlarge, surgical resection was considered. When patients' condition resolved and follow-up MRI with contrast enhancement did not show continuous increase in size, we concluded that it was pseudoprogression.

Of the 10 growing tumors in 7 patients, 7 tumors in 5 patients received surgical resection, 2 tumors in 2 patients underwent additional SRS using GKS at our institute, and 1 tumor in 1 patient was treated by hypofractionated radiotherapy with Cyberknife at another hospital. Seven lesions were resected after irradiation. In the histological study, 4 lesions that were resected within about a year after SRS (3–14 mo, median 6) showed hypervascularity and increased stromal cells, which were typical characteristics of recurrent HBs. In 3 lesions resected in further late periods after SRS (98–148 mo, median 114), occlusion of tumor vessels and hyaline degeneration in the cellular cement were partially observed, which were considered radiation effects.^7,10 None of them showed evidence of radiation-induced necrosis or neovascularization.

In our series, during the follow-up period, 3 VHL patients underwent surgical resection for their untreated lesions after GKS because they rapidly showed increases in tumor size with remarkable edema, and the patient presented with progressive clinical symptoms (vertigo, headache, and loss of balance).

Treatment-related Adverse Events

In total, radiation-induced adverse events were observed in 3 patients. One patient with VHL (tumor volume, 2.8 cm³) treated by a 20-Gy marginal dose showed peritumoral edema and transient worsening of balance sense with dizziness at 4 months after the second SRS. This patient was treated with corticosteroids, and symptoms improved within 1 year. A patient with sporadic HB (tumor volume, 2.9 cm³) treated with a 20-Gy marginal dose showed intratumoral hemorrhage on the radiographic image 24 months after SRS. This patient did not show any neurological deterioration during follow-up, and the tumor transiently disclosed increase in volume, which gradually decreased and became stable at the last follow-up (141 mo post-SRS). A patient with VHL (tumor volume, 0.19 cm³) treated with an 18-Gy marginal dose revealed transient hydrocephalus due to stenosis of the fourth ventricle outlets caused by cerebellar edema with peritumoral enhancement at 7 months after the second SRS. This patient was also treated conservatively, and symptom (unsteady gait) improved within 8 months.

Discussion

According to the clinical studies investigating the natural history of HB,^4–7 almost all of the VHL patients (97%) showed tumor progression in a 2-phase growth pattern (rapid growing phase and subsequent quiescent phase), and ∼40% of patients presented with clinical symptoms.⁴ For symptomatic cases, surgical resection is offered as the first line of treatment, which can remove the tumor itself with peritumoral edema and immediately resolve the presenting symptoms.⁴ Despite its radicality in local lesions, surgical resection imposes certain risks of neurological and surgery-related complications, which we often hesitate to apply to VHL patients who harbor multiple lesions and will require repeated interventions in future. Especially, silent deeply seated lesions that show increase in size have to be effectively controlled with a less invasive and repeatable therapeutic method before they become symptomatic. For this requirement, SRS may be offered as one of the treatment options.

In the previous reports of SRS for intracranial HBs (Table 3), local control rates ranged from 74% to 96% at 5 years^7,11–14 with 0%–16% risk for adverse events.^7,10,12–14 In the present series, the outcomes of SRS were based on the longer follow-ups: a median observation period of 8 years including 9 patients (41%) followed up over 10 years. The overall local control rates were 94% at 5 years and 80% at 10 years, which are acceptable for an alternative treatment option of surgical resection. Concerning the applied marginal dose, we used a median dose of 18 Gy. While a marginal dose higher than 18 Gy was not a significant factor for tumor control (P = .27), in 2 of the 3 patients who presented with radiosurgery-related adverse events, the tumors were treated with the marginal dose of 20 Gy. Thus, we consider that the marginal dose of 18 Gy is acceptable to achieve sufficient local control without certain risks for radiation-induced adverse events.

Table 3.

The result of radiosurgery for intracranial HBs in previous reports

	No. Patients/Tumors	VHL/Sporadic	Median Follow-up Period (mo)	Median Marginal Dose (Gy)	Median Tumor Volume (cm3)	Local Control Rate (%)	Adverse Events
Matsunaga10	22/67	12-10	63*	14*	1.69*	88%/5y	None
						78%/10y
Kano13	32/74	13/19	50.1	16	0.72	89.9%/5y	1 peritumoral edema
Sayer11	14/26	7-7	36	18	1.1	74%/5y	1 hemorrhage
Asthagiri7	20/44	20	8 y	18.9*	0.1	61%/5y	3 headache
		GK 8				51%/10y	1 dizziness
		LINAC12					1 diplopia
Current series	21/56 (total 97)	14/7	96	18	0.13	92%/5y	2 hydrocephalus
						80%/10y	1 hemorrhage

*Datum is mean value.

Solid lesions (not cystic, P = .03), tumor volume <0.13 cm³ (P = .01), and VHL-related lesions (P = .0005) were the significant factors associated with successful local control, and for the 14 patients with VHL, local tumor control rates were 97% at 5 years and 83% at 10 years. The tumor control rates in our cohort were significantly higher in VHL cases compared with sporadic cases. They were in line with the results of a previous report of external beam radiotherapy¹⁵: 80% for VHL cases and 48% for sporadic cases in 5-year disease-free survival rates. Those results may clarify the role of SRS, especially in the treatment of the VHL-related HBs. For HBs that show apparent enlargement in size or that can possibly become symptomatic along with a slight increase in size, SRS should be recommended before they present with clinical symptoms. However, the difficulty in managing HBs in VHL patients is their multicentricity. Many tumors arise within the CNS at different times during a long follow-up. Stereotactic radiotherapy including hypofractionated radiotherapy would be another option, especially for spinal cord lesions. In the previous report, it achieved 92% in the control rates at 3 years for spinal cord lesions with relatively low risk of neurological complications, suggesting that it would be one of the promising treatment options.⁶ Fractionated external beam radiotherapy would be another option for HBs associated with VHL, especially when tumors are found in the wide range of the craniospinal axis. In the recently published literature, disease-free survival rates were reported as 80% at 5 years for VHL patients,¹⁵ and applied irradiation dose would be an important factor to achieve successful tumor control.^16,17 According to a recent report from the National Institutes of Health, infratentorial craniospinal irradiation for patients with CNS HBs with VHL achieved complete radiographic resolution in 18% of lesions and disclosed a suppressive effect for occurrence of new lesions compared with the historical control.¹⁸ Although the follow-up periods were relatively limited, these results may warrant that fractionated radiotherapy can play an important role in management of HBs for VHL patients harboring multiple lesions in the wide range of the posterior fossa and the spinal cord.

It is still controversial whether we should treat the newly found lesions at first appearance. According to the recently published data about the natural course of HBs in VHL patients, they usually had a quiescence phase and showed a biphasic development pattern.⁴ Despite ambiguity in the long-term natural course of lesions in VHL patients, we principally treated new lesions at first appearance because most of them were in the deep cerebellum or the brainstem, and the treatment, including surgery or SRS, would be at higher risk of significant morbidity after the lesions showed enlargement or became symptomatic. In those regions, even if the tumors are relatively small, SRS may have certain risks. Therefore, indication and timing of treatment for new lesions should be carefully determined. Further observation of both posttreatment and the natural course of this disease would be awaited. On the other hand, in relatively large tumors or in tumors with a cystic component, unfavorable results were observed in our series and have been reported in the previous literature,^8,10–14 thus surgical resection should be strongly recommended.

Conclusion

This study assessed efficacy and safety of SRS for intracranial HBs with longer follow-up data to find that SRS could be offered as an effective treatment for small, solid, and VHL-associated lesions. Surgical resection can be the first line of treatment in large and symptomatic cases, but SRS can be recommended as an alternative in patients with HBs that show apparent enlargement in size or that can possibly become symptomatic along with a slight increase in size before they present with clinical symptoms.

Funding

This work was not supported by any funding.

Conflict of interest statement. The authors have no personal financial or institutional interest in this article.