Hemangiopericytomas are rare malignant tumors

Transcription

1 See the corresponding editorial in this issue, pp J Neurosurg 120: , 2014 AANS, 2014 The role for adjuvant radiotherapy in the treatment of hemangiopericytoma: a Surveillance, Epidemiology, and End Results analysis Clinical article *Adam M. Sonabend, M.D., 1 Brad E. Zacharia, M.D., M.S., 1 Hannah Goldstein, M.D., 1 Samuel S. Bruce, M.A., 1 Dawn Hershman, M.D., M.S., 2,3 Alfred I. Neugut, M.D., Ph.D., 2,3 and Jeffrey N. Bruce, M.D. 1 1 Department of Neurological Surgery, Neurological Institute of New York, College of Physicians and Surgeons; 2 Department of Medicine, College of Physicians and Surgeons, New York; and 3 Department of Epidemiology, Mailman School of Public Health, Columbia University, New York Object. Central nervous system (CNS) hemangiopericytomas are relatively uncommon and unique among CNS tumors as they can originate from or develop metastases outside of the CNS. Significant difference of opinion exists in the management of these lesions, as current treatment paradigms are based on limited clinical experience and single-institution series. Given these limitations and the absence of prospective clinical trials within the literature, nationwide registries have the potential to provide unique insight into the efficacy of various therapies. Methods. The authors queried the Surveillance Epidemiology and End Results (SEER) database to investigate the clinical behavior and prognostic factors for hemangiopericytomas originating within the CNS during the years The SEER survival data were adjusted for demographic factors including age, sex, and race. Univariate and multivariate analyses were performed to identify characteristics associated with overall survival. Results. The authors identified 227 patients with a diagnosis of CNS hemangiopericytoma. The median length of follow-up was 34 months (interquartile range months). Median survival was not reached, but the 5-year survival rate was 83%. Univariate analysis showed that age and radiation therapy were significantly associated with survival. Moreover, young age and supratentorial location were significantly associated with survival on multivariate analysis. Most importantly, multivariate analysis using the Cox proportional hazards model showed a statistically significant survival benefit for patients treated with gross-total resection (GTR) in combination with adjuvant radiation treatment (HR 0.31 [95% CI ], p = 0.04), an effect not appreciated with GTR alone. Conclusions. The authors describe the epidemiology of CNS hemangiopericytomas in a large, national cancer database, evaluating the effectiveness of various treatment paradigms used in clinical practice. In this study, an overall survival benefit was found when GTR was accomplished and combined with radiation therapy. This finding has not been appreciated in previous series of patients with CNS hemangiopericytoma and warrants future investigations into the role of upfront adjuvant radiation therapy. ( Key Words hemangiopericytoma SEER radiation therapy XRT epidemiology oncology Hemangiopericytomas are rare malignant tumors that can be found in the CNS, accounting for approximately 0.4% of all primary CNS tumors. 6 They are unique among primary brain tumors in that they often metastasize systemically, most commonly to the bones, lungs, and liver. 9 11,20 Moreover, hemangiopericytomas may also arise in primary sites outside of the CNS, Abbreviations used in this paper: GTR = gross-total resection; SEER = Surveillance, Epidemiology and End Results; STR = subtotal resection. * Drs. Sonabend and Zacharia contributed equally to this work. including fat tissue, muscles, tendons, or blood vessels. Hemangiopericytomas of the CNS were originally thought to be a subset of meningiomas because of their dural attachment. Stout and Murray first recognized the difference between these two tumor types in 1942, proposing the pericyte as the cell of origin of hemangiopericytomas. 24 A decade later, in 1954, Begg and Garret were the first to document a primary intracranial hemangioperi- This article contains some figures that are displayed in color on line but in black-and-white in the print edition. 300 J Neurosurg / Volume 120 / February 2014

2 Surgery plus radiotherapy for hemangiopericytoma cytoma. 1 The WHO officially recognized hemangiopericytomas as unique CNS tumors in Distinguishing between hemangiopericytomas and meningiomas, which requires a tissue diagnosis, has clinical significance, as patients with hemangiopericytomas have a worse overall prognosis and require more vigilant follow-up for recurrence and metastasis. Classified as WHO Grade II tumors, with anaplastic hemangiopericytomas classified as WHO Grade III tumors, 6 hemangiopericytomas present a significant challenge to neurosurgeons as they have a high propensity to recur, even after gross-total resection (GTR). 9,11,20 Rates of local recurrence and systemic metastases have been reported to be as high as 91% and 64%, respectively, at 15 years after initial resection. 9,26 The likelihood of recurrence and metastasis increases with prolonged survival, necessitating long-term systemic follow-up and providing a rationale for aggressive upfront treatment. While GTR of CNS hemangiopericytomas is largely recognized as a prognostic factor for prolonged progression-free survival in several large surgical series, 5,11,22 complete resection is only possible in one-third to twothirds of cases. 4,7,9,10,18,20,26 Tumor location, dural sinus invasion, dense vascularity, and neural involvement often limit the extent of resection. After surgery, adjuvant radiation therapy has been shown to increase progressionfree survival, with Guthrie et al. reporting recurrence at an average of 74 months in the adjuvant radiotherapy group, compared with 29 months in the surgery-alone group. 9 Additionally, stereotactic radiosurgery has been shown to be effective in controlling residual or recurrent disease in a majority of patients. 2 4,7,9,10,16,21,22 Since CNS hemangiopericytomas are rare, their clinical behavior and the natural history of the disease have been difficult to study. With the exception of a literature review conducted by Rutkowski and colleagues in 2010, 19 most studies of hemangiopericytomas are based on single-center experiences. We used the Surveillance, Epidemiology and End Results (SEER) database of the National Cancer Institute to analyze the incidence, treatment, and survival of patients with CNS hemangiopericytomas. Methods Patient Identification We identified patients with a diagnosis of hemangiopericytoma registered in the SEER Program of the National Cancer Institute. The SEER database is a population-based tumor registry containing data covering approximately 26% of the US population. Information concerning primary tumor type, patient demographics, initial cancer treatments, and survival is collected in the database. 14 We used data from the 2 most recent SEER data sets ( and ) to identify patients with International Classification of Diseases for Oncology (ICD-O-3) codes for hemangiopericytoma (9150) and limited the search to those arising in the CNS. To be eligible, patients had to have microscopically confirmed hemangiopericytoma. Individuals with more than a single primary tumor were excluded from analysis. J Neurosurg / Volume 120 / February 2014 Variable Selection Age as a variable was analyzed in 5-year increments and was summarized in 20-year intervals. Race was coded as white, black, and other. Lesion size was treated as a dichotomous variable with a cutoff at the median value. Lesion size was recorded from either pathological reports or imaging findings. Metastases were treated as a dichotomous variable with patients either having no metastases or distant metastases. Tumor extension was also dichotomized, with tumors being classified according to the SEER database as either locally invasive or noninvasive. Initial treatment was analyzed as a categorical variable with the following categories: biopsy only, biopsy plus radiation therapy, subtotal resection (STR) alone, STR plus radiation therapy, GTR alone, and GTR plus radiation therapy. Unfortunately, tumor grade was missing in 80.6% of patients (183 of 227); therefore, we excluded these data from the analysis. Outcomes Age-adjusted incidence rates and age-, sex-, and raceadjusted survival rates were obtained utilizing SEER*Stat (Surveillance Research Program, National Cancer Institute SEER*Stat software version [seer.cancer.gov/ seerstat]). Age adjustment was based on the 2000 US standard population. Relative survival was computed relative to the US standard population by individual year. Five patients were excluded from survival analysis because they were alive without survival times recorded. Relative survival is defined as the observed survival rate as a percentage of expected survival. The date of death or censoring was also obtained from the SEER data set. The initial analysis included descriptive statistics of clinical and demographic information using means, proportions, and standard deviations. For univariate analyses, the Kaplan-Meier method including log-rank tests and univariate Cox proportional hazards models were used. Multivariate effects of covariates on survival times were evaluated via the Cox proportional hazards model. The proportional hazards assumption was confirmed by analysis of the scaled Schoenfeld residuals (Table 1). 8 One-, 2-, 3-, 4-, 5-, and 6-year estimates of survival and 95% CIs were calculated using the Kaplan-Meier method. A p value < 0.05 was considered statistically significant. Analyses were performed using R environment for statistical computing (R Development Core Team, 2008). Results A total of 227 patients with a diagnosis of hemangiopericytoma were identified in the SEER database spanning the years The median follow-up was 34 months (interquartile range months). Baseline patient demographic characteristics can be seen in Table 2. Incidence rates were stable over the time period analyzed, with an overall age-adjusted incidence rate of 0.3 per 1,000,000 person-years. Consistent with prior studies, 4,9,12,18,20,26 a peak incidence occurred during the 5th decade, with an almost equal distribution between males and females (48.0% and 52.0%, respectively). The age-ad- 301

3 A. M. Sonabend et al. TABLE 1: Goodness-of-fit statistics for scaled Schoenfeld residuals* Variable ρ χ 2 p Value age tumor size (ref: <50 mm) 50 mm < unknown location (ref: supratentorial) infratentorial meningeal other treatment (GTR) biopsy XRT STR STR+XRT GTR+XRT * ref = reference; XRT = radiotherapy. justed incidence rates did not differ between sexes, with an incidence rate of 0.3 per 1,000,000 seen in both males and females. While the majority of patients with CNS hemangiopericytomas in the SEER database were white (80.6%), the population age-adjusted incidence rates were not significantly different among races, with an incidence rate of 0.3 per 1,000,000 for white patients and 0.2 per 1,000,000 for black patients. The use of the SEER database allowed for the collection of clinical data of a relatively homogeneous patient sample with regard to the time period included. Because of the uncommon nature of CNS hemangiopericytomas, many of the previously published series span a large time period. This is a potential source of bias, given that over long periods of time changes in diagnostic accuracy as well as treatment practices and techniques can confound outcomes. Alternatively, series with short time periods captured relatively few patients. Our study captured an average of 22.7 patients per year of study, as opposed to 2 patients per year, the most for any other series included (Table 3). About half of the tumors (50.7%) were located supratentorially, with 30.8% classified as meningeal and 6.6% infratentorial. For patients whose tumor size was recorded, the median tumor size was 50 mm, with 26.0% of patients having tumors smaller than 50 mm and 28.2% of patients having tumors 50 mm or larger. Distant metastases were reported in only 6 patients (2.6% of cases) at the time of initial diagnosis. Patient treatment was highly variable, with 39 patients (17.2% of the total population) undergoing biopsy only and 33 patients (14.5% of the total) receiving biopsy plus radiation therapy. The majority of patients (68.3%) underwent some resection either subtotal or gross total with or without additional radiation therapy. Of the surgically treated patients, 26 patients (11.5% of the total) underwent STR alone, 28 patients (12.3%) underwent TABLE 2: Demographic characteristics Variable No. of Patients (%) age (yrs) (4.4) (17.2) (44.1) (30.0) (4.4) sex male 109 (48.0) female 118 (52.0) race white 183 (80.6) black 17 (7.5) other 27 (11.9) tumor size (mm) <50 59 (26.0) (28.2) unknown 104 (45.8) location supratentorial 115 (50.7) infratentorial 15 (6.6) meningeal 70 (30.8) other 27 (11.9) metastases none 121 (53.3) distant 6 (2.6) unknown 100 (44.1) extension noninvasive 158 (69.6) invasive 16 (7.0) unknown 53 (23.3) treatment biopsy only 39 (17.2) XRT 33 (14.5) STR 26 (11.5) STR+XRT 28 (12.3) GTR 54 (23.8) GTR+XRT 47 (20.7) STR plus adjuvant radiation therapy, 54 patients (23.8%; the largest of any single treatment group) underwent GTR alone, and 47 patients (20.7%) underwent GTR plus adjuvant radiation therapy. Table 4 shows the age-, sex-, and race-adjusted survival data for the entire cohort of patients with CNS hemangiopericytomas studied from 2000 to The survival data for patients with CNS hemangiopericytomas was stable over the time period studied, with 1-, 3-, and 5-year relative survival rates of 96.1%, 88.3%, and 83.8%, respectively, over the years Univariate and multivariate analyses of characteristics associated with prolonged overall survival are present- 302 J Neurosurg / Volume 120 / February 2014

4 Surgery plus radiotherapy for hemangiopericytoma TABLE 3: Review of CNS hemangiopericytoma studies published to date* Authors & Year Period of Pt Enrollment Length of Study (yrs) Total No. of Pts Enrolled Mean No. of Pts/Yr Mean FU Time Treatment Evaluated Main Findings present case mos 39/227 pts underwent biopsy only, 33 XRT only, 26 STR, 28 STR+XRT, 54 GTR, & 47 GTR+XRT. Rutkowski et al., 2012 Rutkowski et al., 2010 Fountas et al., pts total; 35 included in analysis NA NA 277 pts w/ info on duration of FU; 194 w/ treatment data mos (no mean given) All pts underwent initial resection; 10/35 pts received preop embolization, 19/35 pts received adjuvant XRT (58% receiving conventional XRT, 9% SRS, & 3% proton beam XRT). NA 6.5 yrs All pts underwent initial resection, 66/194 received adjuvant XRT yrs All pts underwent resection; 6/11 pts had undergone preop embolization, 10/11 pts underwent postop XRT. Soyuer et al., pts, w/ 28 in analysis Chang & Sakamoto, 2003 Ecker et al., pts ; 18 pts pre yrs All pts underwent intitial resection (15/28 GTR, 10/28 STR, 4/28 unknown extent); 10/28 pts received initial XRT, & 12/28 pts received salvage XRT <1 44 mos All pts received SRS for recurrent tumors; all pts had undergone resection of initial tumors & 5/8 had received prior conventional XRT , including the 5 pts reported on by Coffey et al yrs All pts underwent 1 op (18/33 GTR); 19/38 pts received initial postop XRT; 15/38 pts underwent GKS, 11/38 pts received multiagent salvage chemotherapy. Kim et al., yrs All pts underwent resection; 11/31 pts received conventional XRT, & 6/31 received SRS. Sheehan et al., pts w/ 15 tumors mos All pts had undergone prior resection for initial presentation of tumors; GKS was then used as the initial treatment for tumor recurrence. Coffey et al., pts w/ 11 tumors mos All pts underwent GKS after resection for new or recurrent tumors; 3/5 pts had received prior XRT. Only the pts in the GTR+XRT treatment group were found to have a statistically significant survival advantage w/ a decreased mortality rate (HR 0.31, 95% CI ; p = 0.040) for pts who had undergone GTR+XRT compared w/ only biopsy. GTR was associated w/ prolonged overall survival; adjuvant XRT delayed tumor progression but only trended toward significance in terms of effect on overall survival. GTR provided an overall survival benefit; adjuvant XRT did not provide a statistically significant survival advantage. Although this study was not powered sufficiently to find statistically significant differences in survival, the authors support aggressive resection after preop embolization, w/ postop XRT & continuous surveillance. Although this study did not find any statistically significant differences in survival, the authors recommend attempted GTR followed by limitedfield XRT. After XRT, 6/8 (75%) recurrent tumors decreased in size. This study did not find any statistically significant differences in survival based on treatement. Improved 5-yr survival rates post-1990 were thought to be attributable to improved overall cancer care, 0% intraop mortality, & the use of XRT in the treatment of recurrent disease. GTR provided a statistically significant survival benefit; adjuvant XRT trended toward significance in overall survival. GKS provided local tumor control for 12/15 recurrent tumors. Dramatic decrease in size of all tumors for which FU imaging was available (9/11) w/in the margins of the irradiation field. (continued) J Neurosurg / Volume 120 / February

5 A. M. Sonabend et al. TABLE 3: Review of CNS hemangiopericytoma studies published to date* (continued) Mean FU Time Treatment Evaluated Main Findings Mean No. of Pts/Yr Total No. of Pts Enrolled Length of Study (yrs) Period of Pt Enrollment Authors & Year GKS extended survival & maintenance of neurological function. Payne et al., pts w/ 15 tumors mos All pts underwent GKS for recurrent tumors after resection. No tumor recurrence was seen at the time of last <1 6.7 yrs All pts underwent initial resection; 3/7 patients FU in 2/3 pts receiving initial adjuvant radiotherapy; all other pts experienced tumor recurrence at an average of 4.2 yrs after initial treatment. received adjuvant XRT at the time of initial treatment. Uemura et al., 1992 Authors found XRT to offer a statistically significant benefit in recurrence-free period & a trend toward significance in overall survival benefit. Guthrie et al., <1 All pts underwent initial resection; 17/32 pts initially underwent surgery at the Mayo Clinic & received XRT postsurgery. The authors reported on 2 pts who underwent XRT after GTR & showed no recurrence at the time of publication; however, this study was not powered sufficiently to find statistically significant differences in survival <1 6.5 yrs All pts underwent resection (14/21 GTR); 5/21 pts underwent adjuvant XRT at some point in the course of their disease. Jääskeläinen et al., 1985 * FU = follow-up; GKS = Gamma Knife surgery; info = information; NA = not available; pt = patient; SRS = stereotactic radiosurgery. TABLE 4: Overall survival and age-, sex-, and race-adjusted relative survival Mos % Overall Survival (95% CI) % Relative Survival (95% CI) ( ) 96.1 ( ) ( ) 94.5 ( ) ( ) 88.3 ( ) ( ) 84.5 ( ) ( ) 83.8 ( ) ( ) 81.4 ( ) ed in Table 5. Advancing age was associated with worse outcome in both the univariate and multivariate models, with HRs of 1.26 (95% CI , p < 0.001) and 1.90 (95% CI , p = 0.001), respectively. There were no statistically significant differences in survival between males and females or among races. Infratentorial location was associated with significantly worse outcomes in the multivariate, but not univariate, analysis. In the multivariate analysis, patients with infratentorial tumors had an HR of 2.89 (95% CI , p = 0.043) compared with those with supratentorial tumors. Kaplan-Meier survival curves showing differences in survival rates by age and tumor location are shown in Fig. 1. Univariate analysis showed significant improved survival among patients receiving radiation therapy compared with those not receiving radiation therapy (logrank test, p = 0.018), regardless of other treatment (Fig. 2). However, in the case of resection, there were no significant differences in survival between GTR, STR, and biopsy alone (log-rank test, p = 0.282). We next evaluated differences in survival between treatment groups, taking into account combinations of radiation therapy and resection (Table 5). In this case, the only treatment group found to have a statistically significant survival advantage was the GTR plus radiation therapy group, with HRs of 0.29 (95% CI , p = 0.022) and 0.31 (95% CI , p = 0.040), in the univariate and multivariate models, respectively, compared with the biopsy-only group. Interestingly, radiation therapy without resection trended toward offering a significant survival advantage, with patients who underwent radiation therapy having HRs of 0.22 (95% CI , p = 0.050) and 0.24 (95% CI , p = 0.065) in the univariate and multivariate models, respectively. Yet GTR alone, without the adjuvant radiation treatment, did not offer a statistically significant advantage in overall survival for the treatment period studied. Discussion In this study we found that GTR plus upfront adjuvant radiation therapy was the only treatment that provided a significant survival advantage in multivariate analysis for patients presenting with CNS hemangiopericytomas. Given the rarity of this lesion, most institutions have limited treatment experience, resulting in heterogeneity of treatment paradigms. This large nationwide population-based patient cohort provides a unique opportunity to evaluate 304 J Neurosurg / Volume 120 / February 2014

6 Surgery plus radiotherapy for hemangiopericytoma TABLE 5: Univariate and multivariate analyses of characteristics associated with prolonged overall survival* Univariate Multivariate Cox Regression Model Variable HR (95% CI) p Value HR (95% CI) p Value age 1.26 ( ) < ( ) female sex 0.80 ( ) race (ref: white) black 1.01 ( ) other 0.41 ( ) tumor size (ref: <50 mm) 50 mm 3.43 ( ) ( ) unknown 3.64 ( ) ( ) location (ref: supratentorial) infratentorial 2.18 ( ) ( ) meningeal 0.55 ( ) ( ) other 0.57 ( ) ( ) metastases (ref: none) distant 0.97 ( ) unknown 0.67 ( ) treatment (ref: biopsy only) XRT 0.22 ( ) ( ) STR 0.90 ( ) ( ) STR+XRT 0.46 ( ) ( ) GTR 0.47 ( ) ( ) GTR+XRT 0.29 ( ) ( ) * Significant values are shown in boldface. trends in initial presentation, patient characteristics, and current treatment practices. While previous studies have reported a significant survival difference associated with extent of resection alone, 5,11,19,22 we found this survival advantage to be statistically significant only when patients underwent GTR in combination with adjuvant radiation therapy. Data evaluating the role of initial adjuvant radiation treatment are limited, with most studies examining the effect of radiation therapy as a second-line treatment in the control of recurrent or residual disease. 3,4,21,25 In studies analyzing radiation as a component of the primary treatment, many authors have reported at least a trend toward a survival advantage associated with adjuvant radiation therapy. 5,9,22,25 Other authors have found prolonged time until recurrence after adjuvant radiation therapy but no overall benefit in survival; 18 still others have shown that adjuvant radiation therapy offers no benefit at all. 19 This last find- Fig. 1. Kaplan-Meier survival plot for age (log-rank test, p < 0.001) and tumor location (log-rank test, p = 0.075), which were significantly associated with survival in the multivariate analysis (see text and Table 5). J Neurosurg / Volume 120 / February

7 A. M. Sonabend et al. Fig. 2. Kaplan-Meier survival plot for radiation treatment (XRT; log-rank test, p = 0.018), extent of resection (log-rank test, p = 0.282), and treatment combinations (log-rank test, p = 0.066), out of which GTR plus radiation therapy was shown to have a significant survival advantage in comparison with biopsy alone (see text and Table 5). ing, reported by Rutkowski et al., may be partially skewed by the increased mortality found in patients receiving radiation doses of greater than 50 Gy. However, the authors acknowledge that increased mortality may represent a selection bias if sicker patients or those with more aggressive tumors received higher doses of radiation. 19 In our current study, a decreased mortality rate was found, with an HR of 0.31 (95% CI , p = 0.040), for patients who underwent GTR plus radiation therapy compared with those who underwent biopsy alone. Given the high propensity of hemangiopericytomas to recur and metastasize, even after GTR, aggressive initial management may be warranted. The survival advantage of GTR combined with radiation therapy, but not GTR alone, suggests that tumor cells left behind despite radiographically confirmed GTR necessitate adjuvant therapy to fully realize a survival benefit. Adjuvant radiation may be a reasonable treatment option for all patients presenting with CNS hemangiopericytomas. Our results, combined with the inconsistencies in the literature, warrant further investigation and future prospective trials evaluating the role of radiation therapy as an initial adjuvant treatment modality for these tumors before this finding can be taken into account for clinical decision making. 4,25 In addition to GTR plus adjuvant radiation therapy, younger age and supratentorial location were associated with better overall survival. Consistent with our findings, prior studies have reported infratentorial location as a negative prognostic factor for patients with CNS hemangiopericytomas. 9,19 This is often attributed to difficulties achieving GTR due to the anatomy of the posterior fossa. However, in our current study, in multivariate analysis controlling for age, tumor size, tumor location, and treatment modality, patients with infratentorial tumors were found to have significantly worse outcomes despite the fact that extent of resection was not shown to independently confer a statistically significant survival benefit. This suggests that there may be inherent differences in tumor behavior associated with posterior fossa location, which would affect prognosis beyond what can be accounted for by the extent of resection achieved. Most studies to date are based on single-center experiences and include no more than patients. Given their small size, these studies are often not sufficiently powered to detect small differences in survival based on demographic factors such as age, sex, or race. While the literature review by Rutkowski and colleagues did not detect a statistically significant survival benefit for these factors, 19 our current analysis shows that younger patients do better and length of survival decreases continually with advancing age. Consistent with prior studies, 4,19 no differences in survival were found to be associated with sex or race. Neither extent of resection nor adjuvant radiation treatment has been shown to affect the time until metastasis. 4,5,9,18 Average time to metastasis has been reported as 99 months, with 10- and 15-year rates of metastasis of 33% and 64%, respectively. 9 In the current study, a much smaller percentage of patients was found to have distant metastases, compared with what has been previously reported. 9,18,19,26 While some of this discrepancy may be attributable to missing data within the SEER database, the small number of metastases seen is likely reflective of the relatively short length of follow-up in the current study. A mean survival after metastasis of only 2 years has been reported, regardless of the extent of local resection. 5,9 Median survival rates following a diagnosis of hemangiopericytoma have traditionally been reported to be between 5 and 6 years, with 15-year survival rates of 15% 23%. 5,9 More recent studies have shown much better outcomes, with both Ecker et al. and Rutkowski et al. reporting 5-year survival rates of greater than 90% 4,18 and Rutkowski et al. reporting median overall survival of 16.2 years from the time of diagnosis 18 in their single institution surgical series. Our current study demonstrates 5-year survival rates of approximately 83%, consistent over the 10-year study period. These rates are consistent with the survival data reported by Rutkowski et al. in their larger literature review, 19 suggesting that this is a reasonable expectation for patients treated across institutions in the US. While our study aims to provide a comprehensive 306 J Neurosurg / Volume 120 / February 2014

8 Surgery plus radiotherapy for hemangiopericytoma analysis of the demographic and clinical characteristics of patients with a diagnosis of CNS hemangiopericytoma, it is important to recognize several limitations. First, as a national, multiinstitutional database, clinical information available in the SEER database is limited. Importantly for our study, SEER only records information about the initial course of treatment. For instance, it is possible that patients who were documented as not having received radiation therapy did undergo such treatment subsequently without record of it, thus confounding our findings. Our results regarding treatment efficacy used the biopsy-only group as a reference and may bear the potential bias that the patients who only undergo biopsy might have been sicker and carried a poorer prognosis independent of the treatment given. However, there is also the possibility of patients designated as having undergone biopsy alone receiving undocumented radiation therapy, which would bias against the finding of GTR plus radiation therapy being associated with significantly better survival than biopsy, and yet, our results show this. With hemangiopericytomas, which have a high propensity to recur and metastasize despite aggressive initial treatment, subsequent therapies may play an important role in determining long-term survival. Furthermore, data for many of the patients are incomplete, and the average time of follow-up was only 34 months, a relatively short time period for a disease with a median survival of longer than years. This relatively short follow-up time may largely explain the lack of significant survival advantage associated with GTR alone, a treatment modality that has been shown to increase survival in prior studies. 5,11,19,22 Furthermore, given its rarity, inexperienced pathologists may not recognize the pathological features that differentiate hemangiopericytomas from meningiomas, resulting in underreporting of cases. Even if all cases were diagnosed correctly, prior analyses have shown an increasing rate of underreported cancer cases, as diagnoses move from hospital settings to ambulatory physician offices. 13,15,17 Finally, while analyzing data from multiple sites allows for a larger cohort of patients and results that can be more reliably generalized to the patient population at large, there are limitations resulting from the potential for inaccurate coding, varying treatment patterns, and differences in clinical decision making at individual centers, which could not be accounted for in this analysis. Conclusions The current study provides provocative information regarding CNS hemangiopericytoma management, but it should be complemented with future studies to corroborate its findings. The demographic data as well as prognostic factors including age and location provided by our study are valuable given the nationwide sampling of the SEER database. Additionally, the novel finding suggestive of an additive effect of GTR and radiation therapy is important and justifies the execution of prospective studies to confirm the effectiveness of adding upfront radiation therapy to the surgical treatment of these tumors. J Neurosurg / Volume 120 / February 2014 Acknowledgment We thank the National Cancer Institute for the development of the SEER database as a resource for clinical cancer research. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Sonabend, Zacharia. Acquisition of data: Zacharia, Goldstein. Analysis and interpretation of data: Sonabend, Zacharia. Drafting the article: Sonabend, Goldstein. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Sonabend. Statistical analysis: SS Bruce. Administrative/technical/material support: JN Bruce. Study supervision: Hershman, Neugut, JN Bruce. References 1. Begg CF, Garret R: Hemangiopericytoma occurring in the meninges: case report. Cancer 7: , Chang SD, Sakamoto GT: The role of radiosurgery for hemangiopericytomas. Neurosurg Focus 14(5):E14, Coffey RJ, Cascino TL, Shaw EG: Radiosurgical treatment of recurrent hemangiopericytomas of the meninges: preliminary results. J Neurosurg 78: , Ecker RD, Marsh WR, Pollock BE, Kurtkaya-Yapicier O, Mc- Clelland R, Scheithauer BW, et al: Hemangiopericytoma in the central nervous system: treatment, pathological features, and long-term follow up in 38 patients. J Neurosurg 98: , Fountas KN, Kapsalaki E, Kassam M, Feltes CH, Dimopoulos VG, Robinson JS, et al: Management of intracranial meningeal hemangiopericytomas: outcome and experience. Neurosurg Rev 29: , Giannini C, Rushing E, Hainfellner JA: Haemangiopericytoma, in Louis DN, Ohgaki H, Wiestler OD, et al (eds): WHO Classification of Tumours of the Central Nervous System. Lyon: IARC Press, 2007, pp Goellner JR, Laws ER Jr, Soule EH, Okazaki H: Hemangiopericytoma of the meninges. Mayo Clinic experience. Am J Clin Pathol 70: , Grambsch PM, Therneau TM: Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 81: , Guthrie BL, Ebersold MJ, Scheithauer BW, Shaw EG: Meningeal hemangiopericytoma: histopathological features, treatment, and long-term follow-up of 44 cases. Neurosurgery 25: , Jääskeläinen J, Servo A, Haltia M, Wahlström T, Valtonen S: Intracranial hemangiopericytoma: radiology, surgery, radiotherapy, and outcome in 21 patients. Surg Neurol 23: , Kim JH, Jung HW, Kim YS, Kim CJ, Hwang SK, Paek SH, et al: Meningeal hemangiopericytomas: long-term outcome and biological behavior. Surg Neurol 59:47 54, Kochanek S, Schröder R, Firsching R: Hemangiopericytoma of meninges. I. Histopathological variability and differential diagnosis. Zentralbl Neurochir 47: , Larsen IK, Småstuen M, Johannesen TB, Langmark F, Parkin DM, Bray F, et al: Data quality at the Cancer Registry of Norway: an overview of comparability, completeness, validity and timeliness. Eur J Cancer 45: , National Cancer Institute: SEER Training Modules. Cancer registration & surveillance modules: morphology. ( training.seer.cancer.gov/coding/guidelines/morphology.html) [Accessed October 14, 2013] 307

9 A. M. Sonabend et al. 15. Nennecke AL, Hentschel S, Reintjes R: Cancer survival analysis in Hamburg : assessing the data quality within a population-based registry. Acta Oncol 48:34 43, Payne BR, Prasad D, Steiner M, Steiner L: Gamma surgery for hemangiopericytomas. Acta Neurochir (Wien) 142: , Penberthy L, McClish D, Peace S, Gray L, Martin J, Overton S, et al: Hematologic malignancies: an opportunity to fill a gap in cancer surveillance. Cancer Causes Control 23: , Rutkowski MJ, Jian BJ, Bloch O, Chen C, Sughrue ME, Tihan T, et al: Intracranial hemangiopericytoma: clinical experience and treatment considerations in a modern series of 40 adult patients. Cancer 118: , Rutkowski MJ, Sughrue ME, Kane AJ, Aranda D, Mills SA, Barani IJ, et al: Predictors of mortality following treatment of intracranial hemangiopericytoma. Clinical article. J Neurosurg 113: , Schröder R, Firsching R, Kochanek S: Hemangiopericytoma of meninges. II. General and clinical data. Zentralbl Neurochir 47: , Sheehan J, Kondziolka D, Flickinger J, Lunsford LD: Radiosurgery for treatment of recurrent intracranial hemangiopericytomas. Neurosurgery 51: , Soyuer S, Chang EL, Selek U, McCutcheon IE, Maor MH: Intracranial meningeal hemangiopericytoma: the role of radiotherapy: report of 29 cases and review of the literature. Cancer 100: , Spatola C, Privitera G: Recurrent intracranial hemangiopericytoma with extracranial and unusual multiple metastases: case report and review of the literature. Tumori 90: , Stout AP, Murray MR: Hemangiopericytoma: a vascular tumor featuring Zimmermann s pericytes. Ann Surg 116:26 33, Uemura S, Kuratsu J, Hamada J, Yoshioka S, Kochi M, Ushio Y, et al: Effect of radiation therapy against intracranial hemangiopericytoma. Neurol Med Chir (Tokyo) 32: , Vuorinen VSP, Sallinen P, Haapasalo H, Visakorpi T, Kallio M, Jääskeläinen J: Outcome of 31 intracranial haemangiopericytomas: poor predictive value of cell proliferation indices. Acta Neurochir (Wien) 138: , 1996 Manuscript submitted January 16, Accepted October 10, Please include this information when citing this paper: published online November 29, 2013; DOI: / JNS Address correspondence to: Adam M. Sonabend, M.D., Department of Neurological Surgery, Columbia University, College of Physicians and Surgeons, Neurological Institute of New York, 710 W. 168 St., 4th Fl., New York, NY adam.sonabend@ gmail.com. 308 J Neurosurg / Volume 120 / February 2014