• hemangiopericytoma;
  • solitary fibrous tumors;
  • soft tissue sarcoma;
  • chemotherapy;
  • anti-angiogenesis inhibitors


  1. Top of page
  2. Abstract


Hemangiopericytomas and malignant solitary fibrous tumors (HPC/SFT) are rare, closely related sarcomas with unpredictable behavior that respond infrequently to chemotherapy. An optimal systemic treatment strategy for advanced HPC/SFT has not yet been identified.


We retrospectively analyzed the records of 14 patients with histopathologically confirmed HPC/SFT who were treated at The University of Texas MD Anderson Cancer Center between May 2005 and June 2007. All patients were treated with temozolomide 150 mg/m2 orally on days 1-7 and days 15-21 and bevacizumab 5 mg/kg intravenously on days 8 and 22, repeated at 28-day intervals. Computed tomography assessment of tumor size and density (Choi criteria) was used to determine the best response to therapy. The Kaplan–Meier method was used to estimate progression-free survival.


The median follow-up period was 34 months. Eleven patients (79%) achieved a Choi partial response, with a median time to response of 2.5 months. Two patients (14%) had stable disease as the best response, and 1 patient (7%) had Choi progressive disease as the best response. The estimated median progression-free survival was 9.7 months, with a 6-month progression-free rate of 78.6%. The most frequently observed toxic effect was myelosuppression.


Combination therapy with temozolomide and bevacizumab is a generally well-tolerated and clinically beneficial regimen for HPC/SFT patients. Additional investigation in a controlled, prospective trial is warranted. Cancer 2011;. © 2011 American Cancer Society.

Hemangiopericytomas and solitary fibrous tumors (HPC/SFT) are closely related soft tissue sarcomas that appear to exhibit fibroblastic-type differentiation1, 2 and typically affect adults aged 20-70 years. Common sites of involvement include the lower extremities, retroperitoneum/pelvis, lung/pleura, and meninges, but these tumors may be found at virtually any body site.3-5 Histological features that suggest aggressive behavior are not well defined in HPC/SFT. Surgery is typically the treatment of choice for localized disease, with reported 10-year overall survival rates of 54%-89% after complete surgical resection.6-8

For the approximately 20% of HPC/SFT patients who eventually develop local recurrences and/or distant metastases, additional resections should be considered but are not always feasible.6-9 Options for effectively treating unresectable tumors are limited. Radiotherapy can be used only in select cases, and to date, responses to systemic chemotherapy have been infrequent.10-15 Alternative therapeutic options are therefore needed for improved palliation and disease control.

Temozolomide is an oral cytotoxic alkylating agent whose active metabolite, monomethyltriazenoimidazole carboxamide, is identical to that of dacarbazine, a drug with known antitumor activity against soft tissue sarcomas.16, 17 Bevacizumab is a recombinant monoclonal antibody that targets vascular endothelial growth factor (VEGF), a key mediator of a signaling pathway that affects many vital cellular processes, including angiogenesis and vascular permeability.18 Bevacizumab has shown antitumor activity when combined with a number of cytotoxic chemotherapeutic agents, such that its use in combination therapy has been approved for the treatment of metastatic colorectal, non–small cell lung, and human epidermal growth factor receptor 2 (HER2)-negative breast cancers.19-21 The antitumor activity of temozolomide combined with bevacizumab is currently being studied in several phase 2 and 3 trials for glioblastoma and metastatic melanoma.22

In May 2005, a patient with a recurrent meningeal HPC that was refractory to multiple surgical resections, radiotherapy, and chemotherapy was empirically treated with temozolomide and bevacizumab at our institution. He subsequently achieved a radiologically evident reduction in tumor size as well as palliation of tumor-related symptoms. This anecdotal evidence led us to treat 13 subsequently consecutive patients with locally advanced, recurrent, or metastatic HPC/SFT with temozolomide plus bevacizumab. Here we describe the responses and adverse effects of this therapeutic regimen.


  1. Top of page
  2. Abstract


We retrospectively reviewed the medical records of all patients with histological diagnoses of HPC or SFT treated with temozolomide/bevacizumab combination therapy at The University of Texas MD Anderson Cancer Center between May 2005 and June 2007. Patients were identified through the institutional electronic medical records database. Patients for whom radiological scans were unavailable for radiological response assessment were excluded from the analysis. We collected data on patient characteristics, including age, sex, and ethnicity; disease characteristics, including primary tumor site and extent of disease; previous treatment and responses; toxic effects of temzolomide and bevacizumab; and survival. This study was approved by our institutional review board.


All patients received temozolomide 150 mg/m2 orally on days 1-7 and days 15-21 and bevacizumab 5 mg/kg intravenously on day 8 and day 22 on a 28-day cycle. The median number of cycles given was 7.5 (range, 2.5-27). Four patients required temozolomide dose modifications or treatment delay because of neutropenia (n = 1) or thrombocytopenia (n = 3), and 1 of these patients received granulocyte macrophage colony-stimulating factor support.

Radiological Assessment

Baseline radiological studies had been performed up to 4 weeks before the initiation of chemotherapy, and follow-up scans had been performed every 8-12 weeks. Contrast-enhanced computed tomography (CT) or magnetic resonance imaging was used at the discretion of the treating physician. Radiological tumor response was determined as described below.

The longest cross-sectional dimension for each measurable lesion was measured at the beginning of therapy and on each follow-up study. The sum of the longest selected measurable lesions at each timepoint was computed for each patient. Radiological response was then determined by calculating the absolute and percentage change from the baseline sum.

In patients whose responses were assessed via contrast-enhanced CT examination, the tumor density of each was measured in Hounsfield units by drawing a region of interest around the margin of the entire lesion. In patients who underwent CT examination with triphasic techniques, tumor density was measured on scans obtained in the portal venous phase. The mean baseline tumor density was compared with the mean tumor density on the subsequent studies.

Response Assessment

Using Choi response criteria,23, 24 a complete response (CR) was defined as the disappearance of all lesions without the appearance of new lesions. Choi partial response (PR) was defined as a ≥10% decrease in the sum of the target lesions or a ≥15% decrease in tumor density in the absence of new lesions or obvious progression of nonmeasurable disease. Choi progressive disease (PD) was defined as a ≥10% increase in tumor size in the absence of favorable tumor density change required to achieve Choi PR. Patients whose disease did not meet the criteria for Choi CR, PR, or PD and who did not have tumor-related symptomatic deterioration were classified as having stable disease (SD). Only the best response for each patient was used in determining response rates. Response was also assessed using Response Evaluation Criteria in Solid Tumors (RECIST)25 to compare with Choi responses.

Pathological Review

All tumor specimens had been reviewed by an MD Anderson sarcoma pathologist who established the diagnosis of HPC or SFT at the time of each patient's initial presentation at our institution. For the purpose of this study, 2 experienced sarcoma pathologists (A.J.L. and W.L.W.), who were blinded to each patient's outcome, re-reviewed all available specimens to confirm the diagnoses. Histopathological variables (size, number of mitoses, cellularity, pleomorphism, and presence of necrosis and/or hemorrhage) were noted during the re-review whenever possible. Tumors were subcategorized as typical, malignant (based on hypercellularity, mitotic activity of >4 per 10 high-power fields, moderate to severe cytological atypia, tumor necrosis and/or distinctly infiltrative margins), or uncertain/unknown and classified according to the 2002 World Health Organization disease classification criteria for sarcomas.26 The criteria for malignancy in hemangiopericytoma are less well defined than for solitary fibrous tumor, so these same criteria were used for hemangiopericytoma.

Statistical Analysis

Patient characteristics were summarized using medians and ranges for continuous variables and frequencies and percentages for categorical variables. The response rates and the 95% confidence interval (CI) were calculated from variance estimates. The Fisher exact test was used to assess the association between patient or tumor characteristics and best response. Time to best radiological response was calculated from the initiation of temozolomide/bevacizumab therapy to the development of Choi CR or Choi PR. Progression-free survival (PFS) was defined as the interval between the beginning of temozolomide/bevacizumab therapy and radiological evidence of PD, as defined by either the Choi response criteria or RECIST, or death from any cause. Survival data were updated on October 15, 2009, and the patients' data were censored at that point. The Kaplan–Meier method27 was used to estimate the PFS and overall survival (OS). All statistical analyses were performed with S-plus 8.0 (TIBCO Software Inc., Somerville, MA).


  1. Top of page
  2. Abstract

Patient and Disease Characteristics

Sixteen patients with advanced, recurrent, or metastatic HPC/SFT who received temozolomide/bevacizumab therapy between May 2005 and June 2007 were identified. Two patients for whom radiological scans were unavailable were excluded from the analysis. The characteristics of the remaining 14 patients are summarized in Table 1. All 14 patients (9 men, 5 women) were white, and the median age was 59 years (range, 44-75 years). Ten patients (71%) had HPC and 4 patients (29%) had SFT. The most common site of primary disease was the meninges (n = 6). Seven patients had metastatic disease when they began temozolomide/bevacizumab therapy, whereas the remaining patients had either primary or locally recurrent disease deemed surgically unresectable.

Table 1. Patient and Disease Characteristics
  1. HPC indicates hemangiopericytoma; SFT, solitary fibrous tumor.

  2. Data are presented as no. (%) unless indicated otherwise.

Age, y, median (range)59 (44-75)
 Men9 (64.3)
 Women5 (35.7)
 White14 (100)
 HPC10 (71.4)
 SFT4 (28.6)
Primary tumor site 
 Meninges6 (42.9)
 Lung/pleura3 (21.4)
 Pelvis3 (21.4)
 Abdominal wall1 (7.1)
 Gluteal region1 (7.1)
Tumor classification at diagnosis 
 Benign3 (21.4)
 Malignant5 (35.7)
 Unknown6 (42.9)
Metastatic disease 
 No7 (50.0)
 Yes7 (50.0)
Previous therapy 
 No2 (14.3)
 Yes12 (85.7)
No. of previous operations 
 04 (28.6)
 13 (21.4)
 23 (21.4)
 32 (14.3)
 41 (7.1)
 61 (7.1)
Prior radiation therapy 
 No7 (50.0)
 Yes7 (50.0)
No. of previous systemic therapies 
 09 (64.3)
 11 (7.1)
 21 (7.1)
 31 (7.1)
 41 (7.1)
 51 (7.1)

The majority of patients (86%) had received prior therapy before beginning treatment with temozolomide/bevacizumab (Table 1). Five patients had received prior systemic therapy (Table 2). Their best responses to each prior regimen were reassessed using the Choi criteria. They had often achieved SD and improvement in their symptoms with the previous regimens, but none had achieved Choi PR. The main reasons for beginning temozolomide/bevacizumab therapy included symptomatic disease, neoadjuvant treatment to potentially downstage the tumor and enable surgical resection, and disease progression after prior therapy.

Table 2. Patients' Systemic Therapy History
TumorPrevious RegimenDuration of Therapy (mo)Best Response (Choi)Best Response (RECIST)Reason for Discontinuing Therapy
  • HPC indicates hemangiopericytoma; PD, progressive disease; SD, stable disease; SFT, solitary fibrous tumor.

  • a

    Received regimen as adjuvant therapy after R0 resection.

SFTGemcitabine-docetaxel2PDPDDisease progression
SFTGemcitabine-docetaxel1PDPDDisease progression
 Doxorubicin-dacarbazine3SDSDDisease progression
HPCImatinib5SDSDDisease progression
 Imatinib-thalidomide1PDPDDisease progression
 Imatinib-thalidomide-etoposide1SDSDToxicities; patient intolerance
 Imatinib-thalidomide-hydroxyurea7SDSDDisease progression
 Imatinib-hydroxyurea2.5SDSDDisease progression
HPCCelecoxiba14SDaSDDisease recurrence
 Imatinib2PDPDDisease progression
 Paclitaxela6SDaSDPhysician decision
 Gemcitabine-docetaxel3SDSDDisease progression
HPCEndostatin7PDPDDisease progression; toxicities
 Paclitaxel8SDSDDisease progression
 Gemcitabine8SDSDDisease progression

At the time of our analysis, all patients had discontinued therapy. Reasons for discontinuing therapy included disease progression (n = 6); intolerable side effects such as thrombocytopenia (n = 2), fungal infection (n = 1), and decreased performance status (n = 1); patient's preference to discontinue therapy (n = 1); physician's decision to withdraw therapy after the patient achieved maximum clinical benefit (n = 2); and death due to PD (n = 1).

Clinical Outcome

The overall response rate was 79% (11 patients; 95% CI, 49.2-95.3). All 11 patients who responded had Choi PR (Table 3; Fig. 1). Two patients (14%) achieved Choi SD, and Choi PD was the best response in 1 patient (7%).

thumbnail image

Figure 1. Computed tomography images show a Choi PR to temozolomide and bevacizumab in a patient with recurrent, unresectable solitary fibrous tumor of the pleura. (A, B) Images show baseline disease at the beginning of therapy for an anterior mediastinal mass (A) measuring 3.3 cm, 63.5 Hounsfield units and a lower anterior mediastinal mass (B) measuring 7.2 cm, 100.0 Hounsfield units. (C, D) Images show a decrease in size and density of disease after 27 cycles of treatment in the anterior mediastinal mass (C), now measuring 2.3 cm, 58.6 Hounsfield units and the lower anterior mediastinal mass (D), now measuring 4.1 cm, 44.0 Hounsfield units.

Download figure to PowerPoint

Table 3. Overall Response to Temozolomide and Bevacizumab
Patient No.TumorMaximum Change in Tumor Size (%)Maximum Change in Density (%)Best Response (Choi Criteria)Best Response (RECIST)
  • HPC indicates hemangiopericytoma; HU, Hounsfield units; ND, not determined; PD, progressive disease; PR, partial response; SD, stable disease; SFT, solitary fibrous tumor; [DOWNWARDS ARROW], decrease.

  • a

    Density changes could not be measured because response assessment was performed using magnetic resonance imaging.

13HPC4.64.4SD  SD
14HPC15.55.4PD  SD
Median −10.1−26.2    

Response was observed as a decrease in size (n = 1), density (n = 3), or both (n = 7). Ten (71%) patients demonstrated some degree of tumor shrinkage; their median tumor size change was −10.1% (range, −56.2%-15.5%). Ten patients (71%) demonstrated at least a 15% reduction in tumor density, and the median percent change in density was −26.2% (range, −67.6%-5.4%). For the patients who demonstrated a Choi PR, response was seen early during treatment, with all patients achieving PR after 2-4 cycles; the median time to response was 2.5 months (range, 1.6-4.7 months). Of the 7 patients who had symptomatic disease at the beginning of therapy, 6 achieved Choi PRs, which were seen with improvements in their symptoms.

The overall response rate was also calculated using RECIST. Two (14%) patients achieved a RECIST PR (95% CI, 1.8%-42.8%). The remaining 12 patients all achieved RECIST SD, with 11 (79%) patients demonstrating RECIST SD for more than 4 months (95% CI, 49.2%-95.3%). No statistically significant associations were found between response and any patient or tumor characteristics, including primary tumor location (meningeal vs nonmeningeal) and primary tumor histological classification (benign vs malignant vs unknown).

At the time of analysis, the median follow-up was 34 months. The median Choi PFS was 9.67 months (95% CI, 7.31-not estimable), and the proportion of patients who were progression-free at 6 months was 78.6% (Fig. 2). The median RECIST PFS was 10.8 months (95% CI, 8.13-not estimable) (Fig. 3), and the 6-month PFS was 92.9%. To date, 5 patients are alive, and 4 (28.6%) of them remain progression-free. Ten (71.4%) patients ultimately had PD or had died. The median overall survival was estimated at 24.3 months.

thumbnail image

Figure 2. Kaplan–Meier estimates for progression-free survival (PFS) using Choi criteria are shown.

Download figure to PowerPoint

thumbnail image

Figure 3. Kaplan–Meier estimates for progression-free survival (PFS) using RECIST are shown.

Download figure to PowerPoint


Treatment was generally well tolerated, but because it was not administered in a clinical trial setting, toxicity data were not recorded systematically. The most notable toxic effect was myelosuppression, with neutropenia and thrombocytopenia requiring treatment modifications and/or delays in 4 patients. Fever, chills, fatigue, nausea, and headache were also noted.

One patient developed a pulmonary infiltrate after 20.5 cycles of temozolomide/bevacizumab therapy, at which point therapy was withheld. A follow-up chest CT examination 6 weeks later revealed a persistent lung nodule. A biopsy of the lesion showed inflammation, and the cultures were positive for Cryptococcus. The patient was successfully treated with oral fluconazole, and the infection subsequently resolved.

One patient, a 48-year-old woman, died during treatment. She had a recurrent HPC tumor adjacent to the cervical spine and had undergone 3 previous operations as well as radiation therapy. On day 11 of cycle 4 of temozolomide/bevacizumab therapy, she was admitted to the hospital with Staphylococcus aureus bacteremia secondary to infected hardware in her cervical spine. She was treated with intravenous antibiotics, and the bacteremia resolved. She received 2 additional cycles of treatment but was admitted again on day 7 of cycle 6 with renal failure, altered mental status, and hypotension. She died the following day.


  1. Top of page
  2. Abstract

In patients with locally advanced, recurrent, or metastatic HPC/SFT who were treated with temozolomide and bevacizumab, reductions in tumor size and/or density consistent with PRs as assessed by Choi criteria were evident in most patients. Several patients also demonstrated long periods of freedom from disease progression, with 5 patients having a time-to-progression period of ≥20 months.

Currently, the combination of doxorubicin and ifosfamide is the standard systemic chemotherapy regimen for many subtypes of soft tissue sarcoma. Gemcitabine combined with docetaxel has also emerged as a good therapeutic choice for these patients. Although cases of HPC/SFT responding to these chemotherapeutic agents have been reported sporadically,6, 10-13, 15 no systematic review or clinical trial to date has identified an effective systemic regimen for unresectable HPC/SFT.

Because of a lack of good historical data regarding response rates and disease PFS with which we can readily compare our current findings, we turned to our existing patients' experiences with systemic chemotherapy. A review of our HPC/SFT patients' prior regimens showed that doxorubicin, gemcitabine/docetaxel, and paclitaxel did not produce a RECIST radiological response in any of the 5 patients. We then retrospectively reassessed their responses using the Choi criteria and concluded that none of the patients had achieved a Choi PR to prior therapy, but all 5 had a PR to temozolomide and bevacizumab. To further understand the activity of temozolomide and bevacizumab compared with standard chemotherapy regimens, we previously reported on a separate cohort of 5 advanced HPC/SFT patients who had received doxorubicin and ifosfamide, single-agent ifosfamide, or gemcitabine and docetaxel at our institution.14 Reassessment of their radiological scans using the Choi criteria showed that only 1 of 5 patients demonstrated a Choi PR, with a median PFS of 6.1 months (range, 1.6-9 months), further suggesting that standard chemotherapy regimens may only have limited efficacy in HPC/SFT.

The overall Choi response rate of 79% with temozolomide and bevacizumab observed in this retrospective review, therefore, seems to be much more favorable than that with standard chemotherapy regimens. Our study has the typical limitations of a retrospective analysis, including the possibilities of patient selection bias and observer bias, a small sample size, and the lack of a systematic, comprehensive recording of toxic effects. Nevertheless, the degree of Choi radiological responses and the duration of PFS observed in our patients appear superior to those observed in historical studies with chemotherapy regimens.

The current evidenced-based method for response evaluation for soft tissue sarcomas is RECIST. However, several studies have demonstrated that RECIST may be insensitive for evaluating response in patients with gastrointestinal stromal tumors (GIST) treated with imatinib, and the Choi criteria have recently emerged as a more sensitive tool for assessing the degree of tumor necrosis in response to therapy in that setting.23, 24 Soft tissue sarcomas other than GIST, treated with cytotoxic or biological therapies, display patterns of response similar to those of GISTs treated with imatinib, with patients exhibiting long time-to-progression periods despite a lack of significant reduction in their tumor size.28, 29 In fact, many investigators who use RECIST to evaluate soft tissue sarcomas do not use response as an endpoint but only time to progression. Therefore, we chose to assess the activity of temozolomide and bevacizumab in HPC/SFT with the Choi criteria as well as RECIST. We believe that the Choi criteria allow us to detect response—or lack thereof—early in the course of treatment and thereby quickly identify potential nonresponders who may benefit from switching to another therapy. In addition, because the Choi criteria for PD are more stringent than RECIST (≥10% increase in tumor size), failure of therapy may also be detected earlier. This earlier detection explains the shorter median PFS we found when using the Choi criteria than when using RECIST.

Although toxicity data were not systematically gathered, all available clinic notes and laboratory values were thoroughly reviewed to capture as many adverse effects as possible. The majority of the patients did not exhibit significant complications while receiving treatment as scheduled. We did not note any serious adverse effects (eg, thromboembolic events, cardiac toxicity, or gastrointestinal bleeding) associated with bevacizumab. Although 1 patient died during treatment, we found no definitive evidence that temozolomide and/or bevacizumab directly contributed to the immediate factors that resulted in her death. Rather, the patient's treating physician (R.S.B.) believed that her overall poor performance status most likely led to her death.

Our patients received a wide range of number of doses of therapy. Most patients received treatment until PD or intolerable adverse effects developed, but a few patients were empirically given treatment breaks because of patient or physician preferences. It is difficult to conclude, based on the sample size, how long temozolomide/bevacizumab therapy should be continued or whether it could be interrupted and resumed without significantly reducing the therapeutic benefit. The optimal dosage and schedule of temozolomide and bevacizumab in HPC/SFT are yet to be defined. Currently, both temozolomide and bevacizumab are used in several different dosages and schedules for other malignancies, with no one clearly proven strategy. We chose our particular regimen based on anecdotal experience from our first patient. A prospective study to address these questions and to establish the optimal temzolomide/bevacizumab therapy's safety and efficacy is currently under development.

Also unclear is the potential additional benefit of radiation therapy. Three patients with isolated sites of disease had received radiation therapy within 4 months before or after temozolomide/bevacizumab treatment. All had achieved Choi PR with their first treatment modality. Two patients were still responding when they initiated their second treatment modality, and 1 patient was exhibiting tumor regrowth. Despite having discontinued their last treatment 18-23 months before the time of our analysis, all 3 patients to date continue to show a long, durable maintenance of their responses, with PFS of 34-43 months. Further information regarding the potential synergistic effect of temozolomide/bevacizumab therapy and radiation is needed to confirm this encouraging finding.

As we stated earlier, the rationale for temozolomide/bevacizumab was empirical, as this combination therapy has not been used in soft tissue sarcoma. We do not know whether the success of this regimen is due to temozolomide, bevaziumab, or their synergistic effect. Bevacizumab may enhance temozolomide's cytotoxic activity or may play an antiangiogenic role by modulating the VEGF signaling pathway. The VEGF–VEGF receptor pathway, which plays a key mediator role in angiogenesis, has recently emerged as a potential key therapeutic target in HPC/SFT. The anti-VEGF receptor tyrosine kinase inhibitors sorafenib and sunitinib have also demonstrated evidence of activity in HPC/SFT.30-32 Patients treated with sunitinib have also demonstrated Choi responses lasting >6 months, similar to patients described in our report.33 Anecdotal experiences with sorafenib in HPC/SFT patients have demonstrated Choi responses and disease stabilization for up to 22 months.30, 32 It is not yet clear whether these anti-VEGF receptor tyrosine kinase inhibitors will be just as effective as temozolomide/bevacizumab in HPC/SFT, but the initial results are encouraging. Further studies are required to better elucidate the key therapeutic targets in HPC/SFT. Analysis of our patients' available tumor specimens for potential molecular correlative factors is ongoing.

To our knowledge, our report represents the largest published series of patients with advanced HPC/SFT treated with a single systemic regimen to date. We found that the combination of temozolomide and bevacizumab had a remarkably high rate of Choi overall response and a favorable duration of disease control. For these rare sarcoma subtypes that lack a well-established systemic therapeutic option, temozolomide/bevacizumab is a promising therapeutic regimen that warrants further investigation. Our results should be validated in a prospective trial, which would also allow additional insight into the efficacy, safety, and biological mechanisms of temozolomide and bevacizumab in HPC/SFT.


  1. Top of page
  2. Abstract

Supported in part by the National Institutes of Health through The University of Texas MD Anderson Cancer Center Support Grant CA016672.


The authors made no disclosures.


  1. Top of page
  2. Abstract