Stereotactic radiosurgery and radiotherapy for meningiomas: biomarker predictors of patient outcome and response to therapy

Abstract

Although surgery has traditionally been the primary treatment of meningiomas, stereotactic radiosurgery (SRS) and radiotherapy (SRT) techniques have become a standard part of the treatment approach to intracranial meningiomas. For incompletely resected or inoperable benign meningiomas, SRT and SRS can provide excellent 5-year tumor control rates in 90% to 95% of benign meningioma cases. The current data on prognostic factors in meningioma SRT and SRS treatment outcomes are sparse. Our study aims to define prognostic factors that may help determine meningioma SRT and SRS treatment outcomes.

Outcomes of 162 patients with 166 intracranial meningiomas treated with SRT (80 treatments) or SRS (92 treatments) were examined. Patient characteristics and tumor hypoxia-regulated biomarkers were correlated with tumor local control and overall survival. Median follow-up was 52 months, with median tumor volumes and treatment doses of 2.72 cm³/15 Gy and 12.54 cm³/54 Gy for SRS- and SRT-treated patients, respectively.

Local control occurred in 68/77 (88.3%) SRT-treated patients and 80/89 (89.9%) SRS-treated patients. Tumor volume was predictive of overall survival for patients treated with SRT. The hypoxia-related biomarkers VEGF, HIF-1, and MIB-1 were useful in predicting outcome after SRT and SRS.

SRS and SRT are successful in controlling intracranial meningioma growth. With further study, HIF-1, VEGF, and MIB-1 may be useful as predictive markers for response to SRT and SRS.

1 INTRODUCTION

Meningiomas are the most frequently diagnosed primary brain tumors, accounting for 33.8% of all primary brain and central nervous system tumors; more than 170,000 people in the US are currently diagnosed with this tumor [1, 2]. Most meningiomas are slow-growing, benign lesions, but approximately 10% have more aggressive pathological features [3], are more difficult to manage, and have high recurrence and poor survival rates [4]. Traditionally, surgery has been the primary treatment of meningiomas [5]; however, stereotactic radiosurgery (SRS) and radiotherapy (SRT) techniques have become a standard part of the treatment approach to intracranial meningiomas. Both are safe, well tolerated, and effective and provide high degrees of patient satisfaction [6, 7]. For incompletely resected or inoperable benign meningiomas, SRT and SRS can provide excellent 5-year tumor control rates in 90 to 95% of benign meningioma cases. Malignant and atypical meningioma patients generally show less favorable outcomes [8].

The current data on prognostic factors in meningioma treated with SRT and SRS are sparse. Antigen Ki-67 is a nuclear protein associated with cellular proliferation and has been shown to be correlated with meningioma recurrence [9] and increasing tumor grade and to discriminate significantly benign from atypical and anaplastic meningiomas [10]. These studies of Ki-67, however, have not examined outcomes after SRT or SRS treatment. Microvascular density (MVD) as a measure of tumor vascularity is of unknown significance in determining outcomes in meningiomas treated with SRS and SRT but has been associated with meningioma tumor grade [11]. There is limited knowledge of the expression of the hypoxia-related biomarkers expression including glucose transporter-1 (Glut-1), vascular endothelial growth factor (VEGF), hypoxia-inducible factor-1α (HIF-1α), and carbonic anhydrase-IX (CA-IX) in meningiomas with regards to outcome data [12, 13]. VEGF is the most studied of these immunohistochemistry markers and has been implicated in meningioma tumorigenesis and growth. The production of VEGF is regulated by hypoxia inducible factor-1α (HIF-1α), especially under conditions of hypoxia [12]. Studies have suggested that the expression of VEGF relates to the development of peritumoral brain edema in meningiomas and the histological grade [14]. CA-IX is an enzyme involved in pH homeostasis and is also regulated by HIF-1α. CA-IX has been shown to be up-regulated in several types of cancer; however, its significance in meningiomas specifically is not well understood [15]. In this study, we aimed to evaluate these potential prognostic factors, as well as other factors that may help determine SRT and SRS treatment outcomes such as patient characteristics and tumor histological grade.

2 METHODS

2.1 Study participants and outcome measures

Under an Institutional Review Board (IRB)-approved protocol, we retrospectively collected clinical and outcome data of patients treated with either SRT or SRS. We reviewed information on patient age, sex, radiation treatment parameters, tumor characteristics, pathology (benign or atypical/malignant), tumor local control (LC), and overall survival (OS). Radiation treatment parameters collected included median dose, use of dynamic conformal arc (DCA) or intensity-modulated radiotherapy (IMRT), and whether the treatment was used after surgical therapy failure—termed recurrent disease—or as an upfront treatment either alone or with surgical therapy—termed primary therapy. Patient tumor characteristics included location and volume. Histological grade was determined using the 2007 World Health Organization (WHO) classification of central nervous system tumors [3, 16], and tumors were categorized as benign/WHO grade I, atypical/ WHO grade II, or anaplastic/malignant/WHO grade III. A single pathologist performed all surgical grading. For patients in whom tissue was available (33 SRS patients and 21 SRT patients), molecular markers of HIF-1α, CA-IX, Glut-1, VEGF, MIB-1 index, and MVD (measured by Factor VIII immunohistochemistry) were examined.

Tumor progression/local control was defined according to MacDonald’s criteria as a 25% increase in tumor size [17]. All patients were assessed for radiation necrosis/treatment–related imaging changes, which were defined as radiologic progression on a post-procedure study followed by stable or improved imaging in a subsequent scan obtained at least 2 months later. Both tumor progression and patient death were counted as events when evaluating LC. Estimates of LC and OS were calculated from the start of SRT/SRS using the Kaplan-Meier method, with significance defined as p<0.05.

2.2 Patient selection and treatment planning

Patients included in the study had been initially evaluated to determine whether SRT or SRS would be more appropriate. Patients with larger tumors or those closer than 4 mm to the optic nerve or chiasm were treated with SRT. Grade II patients that failed surgical therapy were treated with standard fractionation to 54 Gy with the surgical bed and residual volume as the prescribed tumor volume. Grade I tumors with residual volume after surgery or Grade I or Grade II tumors that failed in a delayed fashion after surgery were treated with SRS using a prescribed tumor volume of the actual tumor volume without additional margin. SRS is our treatment of choice whenever tumor size and location permit. Either DCA or IMRT techniques were used to treat all patients who underwent SRS as previously described [18]. Patients treated with SRT were treated with highly conformal SRT, usually using IMRT. All patients underwent high-resolution magnetic resonance (MR) imaging (three-dimensional spoiled gradient recalled [SPGR] T1-weighted with gadolinium enhancement), usually within a week of planned treatment. A noncontrast, stereotactic computed tomography (CT) scan was obtained after the patient was placed either in a stereotactic head frame or thermoplastic mask for immobilization. The MR and CT images were loaded into the treatment planning computer system. Target volumes and organs at risk (usually optic nerve, optic chiasm, eyes, and brainstem) were defined and contoured on the MR images using BrainSCAN or iPlan software. Treatment plans were optimized for tumor dose coverage and dose minimization to the organs at risk.

2.3 HIF-1α, VEGF, CA-IX, and Glut-1

HIF-1α immunohistochemistry was performed as previously described [19] using the Catalyzed Signal Amplification System (DAKO, Carpinteria, CA) according to the manufacturer’s recommended protocol and primary antibody, H1α67 (Novus Biologicals, Littleton, CO), at a dilution of 1:1000. Immunohistochemical analysis of VEGF, CA-IX, and Glut-1 was done using anti-VEGF Ab-1 polyclonal antibody (1:50 dilution; Calbiochem, Cambridge, MA), anti-CA-IX goat polyclonal antibody (1:200; Santa Cruz Biotechnology, Santa Cruz, CA), or rabbit anti-Glut-1 (1:100, Santa Cruz Biotechnology) and the Vectastain ABC kit (Vector Laboratories, Burlingame, CA) as previously described [19]. Slides were counterstained with toluidine blue. Negative controls replaced the primary antibody with nonimmune serum, with all other steps performed as above. Positive controls for HIF-1, VEGF, CA-IX, and Glut-1 were performed on paraffin-fixed sections of tumors grown in mice using human U251 cell lines that were immunohistochemically positive for these proteins using the same steps as above [19].

All slides were examined under 200× magnification using an Olympus BX41 microscope and scored by an investigator (RLJ) blinded to the patient information and tumor grade. The immunohistochemical analysis of HIF-1α, VEGF, CA-IX, and Glut-1 was scored from 0 to 4 based on the number of cells stained in a given field (0: 0%–<25%; 1: 25%–<50%; 2: 50%–<75%; 3: 75%–<100%; and 4: 100%). In this analysis, we consider “high expression” as scores of 2–4 and “low expression” as scores of 0 and 1.

2.4 PROLIFERATION INDEX AND MVD

The proliferation index (PI) was calculated using Ki-67 (clone MIB-1, dilution 1:300) on the Ventana ES (Ventana Medical Systems) as previously described [19]. Positive controls were performed on human thymus, which has >90% cell staining. Negative controls replaced the primary antibody with nonimmune serum. PI was calculated as previously described [19]. Briefly, pictures were taken at 400× (10 ocular × 40 objective) magnification using an Olympus Microfire camera and analyzed using Image-Pro Plus 5.0. PI was calculated as the number of MIB-1-stained cells divided by the total number of cells in the field repeated 3 times for each picture and averaged. Analysis was duplicated by a separate researcher. This method was very reproducible, as demonstrated by good inter-rater reliability [19].

The slides for the MVD index analysis were prepared using the same steps as described above for the MIB-1 analysis except that they were pretreated with Factor VIII (rabbit polyclonal, dilution 1:100) Protease 2 (Ventana Medical Systems). Negative controls replaced the primary antibody with nonimmune serum, with all other steps performed as above.

The MVD index was calculated based on a previously published method [19]. Briefly, three pictures of the most vascular area of the slide were taken at 200× magnification using an Olympus Microfire camera and transferred to Photoshop CS 7 (Adobe Systems Incorporated, San Jose, CA). Any positive cell that was separate from other stained cells and not contiguous or branching from other vessels was counted. The results for each slide were averaged for the resulting MVD and divided by 0.26 mm² to normalize the size of the picture field.

2.5 Statistical analysis

For descriptive statistics, Kaplan-Meier (K-M) analysis was used to estimate OS and PFS. Log-rank test was used to evaluate differences in survival among groups defined by treatment modality (SRS vs. SRT), the four biomarkers, and patient treatment characteristics. Cox proportional hazard models adjusting for age and gender were used to estimate the hazard ratios comparing each arm with the baseline (% of marker value = 0 for the four biomarkers). Statistical packages R (www.r-project.org) and STATA (Stata Inc.) were used for analyses. P values < 0.05 indicate statistically significant results.

3 RESULTS

3.1 Patient characteristics

The study included a total of 162 patients, of whom 76 underwent SRT and 86 underwent SRS (Table 1). A total of 166 tumors were treated: 77 were treated with SRT and 89 were treated with SRS. There were 80 total SRT treatments and 92 SRS treatments (172 total). Ten patients received more than one treatment session, six patients because of recurrent disease and 4 for separate tumors. Thirty-three patients were treated with surgery before SRT, with SRT used as a salvage therapy in 13 patients and as a post-operative adjuvant (primary therapy) in 20. Sixty patients were treated with surgery before SRS, with 25 treated up-front as primary therapy and 35 as salvage therapy after initial surgical failure.

Table 1.

Characteristics of patients with meningiomas treated with SRS or SRT

Patient Characteristics		SRS	SRT	Total
No. of patients		86	76	162
No. of tumors		89	77	166
No. of treatments	92	80	172
No. of tumors with LC		80/89 (89.9%)	68/77 (88.3%)	148/166 (89.2%)
Age (years)	Average	56.97 ± 15.35	53.63 ± 17.77	55.41 ±16.00
	Median	55.66	54.92	55.53
	Range	24.6–89.3	12.5–90.6	12.5–90.6
Sex	Male	16 (18.6%)	22 (28.9%)	38
	Female	70 (81.4%)	54 (71.1%)	124
Tumor grade	WHO I	71 (79.8%)	44 (57.1%)	115 (69.3%)
	WHO II	16 (18.0%)	13 (16.9%)	29 (17.5%)
	WHO III	1 (1.1%)	5 (6.5%)	6 (3.6%)
	Unknown	1 (1.1%)	15 (19.5%)	16 (9.6%)
Tumor volume (cm3)	Median	2.7	11.47	5.1
	Range	0.8–19.5	0.7–78.9	0.7–78.9
Prescription dose (cGy)	Median	1500	5400	1990
	Range	1000–2000	2400–5400	1000–5400

SRS, stereotactic radiosurgery; SRT, stereotactic radiotherapy; LC, local control

The majority of patients in our study that had treatment failures had WHO grade II tumors. Overall median follow-up was 52 months. For SRT, median follow-up was 50 months, and for SRS it was 69 months.

Mean patient age overall was 55.41 years (median 55.53 years, range 12.5–90.6 months). The mean age for SRT was 53.63 years (median 54.92 years; range, 12.5–90.6 months), and mean age for SRS was 56.97 years (median 55.66 years, range 24.6–89.3). There were 38 men and 124 women: 22 men and 54 women had SRT and 16 men and 70 women had SRS. Of the 115 (69.3%) WHO I grade tumors, 44 were treated with SRT and 71 with SRS. There were 29 (17.5%) WHO grade II tumors, 13 treated with SRT and 16 with SRS. Of the 6 (3.6%) WHO grade III tumors, 5 were treated with SRT and 1 with SRS. The total number of tumors with unknown grades was 16 (9.6%), 15 treated with SRT and 1 with SRS.

Overall median tumor volume was 5.1 cm³ (range 0.7–78.9). Median tumor volume for SRT was 11.5 (range 0.7–78.9) and median volume for SRS was 2.7 cm³ (range 0.8–19.5). The median overall prescription dose (cGy) for SRT was 5400 cGy (range 2400–5400) and for SRS it was 1500 (range 1000–2000).

3.2 Outcome for Entire Group

Local control was achieved in 148/166 (89.2%) tumors. Local control of tumors treated as a primary therapy was similar to that of tumors treated as salvage therapy. Six SRS failures and 2 SRT failures required surgery. Two of these patients were initially thought to have metastatic breast cancer lesions and were treated with 20 Gy, but their tumors were found to be meningiomas after SRS treatment failure and subsequent surgery. Of the 10 SRT failures, 5 were WHO grade I, 4 were WHO grade II/III, and 1 was unknown grade. Similarly, of the 9 SRS local control failures, 3 were WHO grade I and 6 were WHO grade II/III. We found no statistically significant differences in local control based on WHO grade for either SRS or SRT treatments.

3.3 Patient outcome after SRT treatment

At last follow-up, 70 of 80 (88%) tumors were locally controlled in 68 of 77 (88.3%) patients treated with SRT. The median of duration of LC was 127.0 months (Figure 1a). Sex, tumor volume, and tumor grade did not have a statistically significant effect on LC after SRT (Table 2). As mentioned above, four of the 10 tumor failures were WHO grade II/III tumors, but this was not statistically different than the remainder of either unknown or WHO grade I tumors. There was no failure of LC for men (n=24); all 10 failures were in female patients (n=56), but this was also not statistically significant. Seven of the 10 tumors treated had undergone either surgery or radiation treatment prior to SRT, but this was not a predictive factor for local failure.

Figure 1.

Kaplan-Meier curves for patients treated with stereotactic radiotherapy. A. Local control (LC) in months of all tumors (n=80). LC was found in 70/80 (88%), with a median local control of 127.0 months. B. Patient overall survival (OS) in months. Median overall survival (OS) was 215.4 months

Table 2.

Local control and overall survival of patients treated with stereotactic radiotherapy (SRT)

	Number of patients		Local control		Overall survival
All patients	N=80 treatments (77 tumors in 76 patients)		Hazard Ratio	Significance	Hazard Ratio	Significance
Sex	Male	Female	NA	0.076	0.51	0.171
	22	54
Tumor grade (if known)	WHO I	WHO II + III	2.97	0.129	1.76	0.222
	44	18
Tumor volume (mm3)	<12	>12	1.44	0.670	1.52	0.576
	41	39
	<4	>4	1.29	0.678	2.67	0.040
	15	65

There were 22 deaths for all patients treated with SRT, and the median overall survival (OS) was 215.4 months for the living patients (Figure 1b); however, only 5 of these deaths were related to meningioma. As with LC, there were no significant differences based on gender or tumor grades for OS (Table 2); however, we did find that tumor volume <4 cm³ was associated with longer OS for SRT-treated patients. The majority of these patients had tumors in proximity to the optic system and required fractionated therapy. The exception to this was three patients with grade II tumors treated postoperatively to the presurgical tumor bed.

3.3 Patient outcome after SRS treatment

Local control was achieved in 81/92 (88%) tumors in 80/89 (89.9%) SRS-treated patients. Three tumors were treated a second time after initial failure. Median time for LC (measured in months) for all patients who underwent SRS has not been reached (Figure 2a). We found no significant difference in the LC after SRS with patient sex, histological grade, or tumor volume (Table 3). Similarily, no significant differences were found according to type of radiosurgery (DCA vs. IMRS) or whether the tumor was treated after failure of surgical therapy or as an upfront therapy.

Figure 2.

Kaplan-Meier curves for patients treated with stereotactic radiosurgery. A. Local control (LC) in months of all tumors (n=92). LC was found in 81/92 (88%), with median local control not yet reached. B. Patient overall survival (OS) in months. Median overall survival (OS) was 215.5 months

Table 3.

Local control and overall survival for patients treated with stereotactic radiosurgery (SRS)

	Number of patients		Local control		Overall survival
All patients	N=92		Hazard Ratio	Significance	Hazard Ratio	Significance
Sex	Male	Female	2.26	0.251	1.05	0.933
	20	72
Tumor grade	WHO I	WHO II + III	1.83	0.561	0.96	0.954
	74	18
Tumor volume	<12	>12	NA	0.181	1.23	0.840
(mm3)	83	9
	<4	>4	0.55	0.461	1.07	0.887
	36	56
	<2	>2	0.47	0.263	2.31	0.155
	33	59

Twelve patients in the SRS-treated group died during the follow-up period, five related to their meningioma tumor progression. Median survival after treatment of the 68 living patients in this group was 215.5 months (Figure 2b). We found no significant difference in the OS after SRS for patient sex, histological grade, or tumor volume (Table 3).

3.4 Biomarker prediction of patient outcomes

Of the biomarkers tested, VEGF, Glut-1, CA-IX, MIB-1, or MVD were not successful in predicting tumor LC after SRT (Table 4); however, HIF-1 was found to be correlated with local control after SRS (p=0.046). Tumor grade was found to correlate with MIB-1 index for all tumors but not with MVD. MIB-1 <5 predicts a longer OS for patients with SRT-treated tumors (Table 4).

Table 4.

Biomarker prediction of local control and overall survival for patients treated with stereotactic radiotherapy (SRT)

	N		Local control		Overall survival
Biomarker	High expression (Scores 2,3, and 4)	Low expression (Scores 0 and 1)	HR	Significance	HR	Significance
VEGF	13	7	4.32	0.177	NA	0.006
HIF-1	12	9	NA	0.046	2.63	0.367
Glut-1	15	5	NA	0.119	1.77	0.591
CA-IX	14	7	2.13	0.486	4.94	0.140
MIB-1	<5	>5	0.289	0.135	0.18	0.020
	6	15
MVD	<50	>50	0.178	0.112	0.16	0.092
	12	9

Lower VEGF expression (p=0.006) and MIB-1 index <5 (p=0.020) were predictive for favorable OS after SRT (Table 4). CA-IX, HIF-1, Glut-1, and MVD were not predictive of OS. There was no predictive value of any of the biomarkers for LC of patients treated with SRS (Table 5). Only VEGF was predictive of OS after SRS treatment (p=0.033).

Table 5.

Biomarker prediction of local control and overall survival for patients treated with stereotactic radiosurgery (SRS)

			Local control		Overall survival
Biomarker	N		HR	Significance	HR	Significance
	High expression (Scores 2,3, and 4)	Low expression (Scores 0 and 1)
VEGF	28	5	0.29	0.197	0.174	0.033
HIF-1	19	14	5.25	0.138	NA	0.056
Glut-1	21	12	2.03	0.526	2.266	0.460
CA-IX	26	7	NA	0.389	0.977	0.985
MIB-1	<5	>5		0.104	0.339	0.215
	17	16	0.16
MVD	<50	>50	0.63	0.685	1.973	0.393
	11	22

4 DISCUSSION

Both SRS and SRT play important roles in the management of patients with meningiomas, offering excellent 5-year tumor control rates for benign tumors [20, 21]. Our findings are in agreement with this. It is been shown previously that tumor grade is a strong prognostic factor in all types of human meningiomas and that older age and tumor grade are negative predictors of tumor control and survival for patients having SRS for WHO grade II and III intracranial meningiomas [9, 22-24]. Interestingly, we did not find a difference in tumor control rate or OS in relation to tumor grade. This is possibly due to the relatively low rate of highergrade tumors in this series. Although malignant and atypical meningiomas generally show overall poor outcomes, available data suggest that surgical resection followed by SRS or SRT and salvage therapy can lead to “extended survival” [8]. While we did not evaluate the “extended survival” benefit of SRT and SRS after surgical resection or other treatment failure, our study does include patients with this form of treatment paradigm.

Other studies have identified the volume of meningiomas as a statistically significant prognostic factor in the outcomes of SRT-treated tumors [22]. Similarly, our study showed that tumor volume correlated with a longer OS for patients with SRT-treated meningiomas. Our results show that there is a statistically significant difference in the OS for patients treated with SRT with a volume of <4 cm³. We were unable to demonstrate the same correlations with SRS-treated tumors, although others have found decreasing tumor LC with increasing tumor size [25, 26]. Although tumor grade and volume may both be options for predicting patient outcomes after treatment, the growth of this type of tumor is still unpredictable, and additional prognostic markers are needed.

There is limited experience examining meningioma biomarkers, especially using multiple biomarkers, to predict or assess SRT and SRS treatment outcomes. We chose to examine HIF-1, VEGF, Glut-1, CA-IX, MIB-1, and MVD to evaluate the predictive value of these particular markers in outcomes of SRT- and SRS-treated meningiomas. We were able to demonstrate that HIF-1 is predictive of LC after SRS. There are reports that suggest that the expression of VEGF, which is transcriptionally regulated by HIF-1 correlates with histological grade in meningiomas [14]. Our results did not support this correlation, but we did find low VEGF expression to be predictive of longer OS after treatment with SRT and SRS. These results are not entirely consistent and will require larger series of treated patients to understand the true role these biomarkers may play in patient outcome prediction. Other biomarkers evaluated in our study, including Glut-1 and CA-IX, did not show any predictive value for tumors treated with SRT or SRS.

MIB-1 index (which measures cellular proliferation) has been shown to be increased with increasing tumor grade and can discriminate significantly benign from atypical and anaplastic meningiomas [10]. This finding is congruent with the results of our study, which showed that MIB-1 correlated with the tumor grade for all tumors treated with SRT and SRS. MIB-1 index has been shown to be a significant predictor of tumor recurrence after surgery, but no studies have correlated this measure with outcomes after SRT or SRS treatment [9]. Our study demonstrates that MIB-1 <5 predicts a longer OS for patients with SRT-treated tumors.

Differences in meningioma MVD have also been studied, but MVD is of unknown significance in determining outcomes in meningiomas treated with SRS and SRT. Previous studies have shown a statistically significant difference in the number of blood vessels in atypical meningiomas and benign meningiomas, with atypical being much higher [11]. We were unable to demonstrate MVD as predictive of outcome of patients treated with SRS or SRT.

5 CONCLUSIONS

SRS and SRT are very successful in controlling intracranial meningioma growth, with most failures occurring in patients with WHO grade II and III tumors. Of the tissue biomarkers studies only HIF-1, VEGF, and MIB-1 appear to have limited predictive value for response to SRT and SRS. The small number of available samples and low number of treatment failures and relatively short patient follow-up are the major limitations of this study. Longer follow-up may reveal more correlations between specific biomarkers and outcomes.