Cranial Hemangiopericytomas

The radiographic differentiation of intracranial HPCs from a meningioma is pivotal in the preoperative management planning and choice of surgical approach, owing to the higher risk of severe bleeding and of local recurrence, even after a “conventional” Simpson grade-I GTR.^{Reference Schirmer and Heilman30} Radiographic similarities to meningiomas are profound, and distinguishing HPCs can be challenging. HPCs typically feature lobulated margins, frequent internal serpentine flow-related signal voids, and absent calcifications. This is different from meningiomas, which typically have smooth margins, no flow voids, and abundant calcifications (20–25%).^{Reference Sibtain, Butt and Connor31} HPCs typically feature a paucity of peritumoral edema and a unique distinct angio-architectural pattern. This pattern involves a dual blood supply from intracranial and extracranial blood vessels. Unlike in meningiomas, the dominant blood supply in HPCs typically arises from the internal carotid artery (ICA) or vertebral artery (VA) branches [rather than the external carotid artery (ECA) in meningiomas] which manifests in numerous corkscrew vessels seen arising from a main arterial feeder within the tumor on digital substraction angiography (DSA).^{Reference Schirmer and Heilman30} This DSA feature results in a long-lasting contrast enhancement pattern, compared to the typical sunburst ECA-related DSA pattern seen in meningiomas.^{Reference Marc, Takei and Schechter32}

The MRI appearance of intracranial HPCs may be similar to that of meningiomas. Unique HPC-related features include a narrow base of attachment, an irregular/multilobulated cross-leaf growth, the absence of intratumoral calcifications, the absence of related osseous hyperostosis,^{Reference Zhou, Liu and Zhang33} bone erosion, and heterogeneous gadolinium contrast enhancement.^{Reference Liu, Chen and Ma34} Cranial HPCs are typically isointense with gray matter on both T1- and T2-weighted MRI sequences. Mixed signals when noted were shown to be associated with the grade-III (anaplastic) aggressive HPC type.^{Reference Zhou, Liu and Zhang33} A “dural tail” is seen in 30% of grade-II patients.^{Reference Liu, Chen and Ma34} The grade-III anaplastic HPC variant is marked by the presence of necrosis, cystic changes, and extensive peritumoral edema.^{Reference Zhou, Liu and Zhang33} Diffusion-weighted imaging in grade-III HPCs is marked by higher ADC values compared to grade-II HPCs, meningioma, or with the normal surrounding brain.^{Reference Liu, Chen and Ma34,Reference Liu, Yin and Geng35} Whole-tumor histogram analysis of ADC maps may be a useful tool for differential diagnosis, with ADCmin and ADC5 being potential parameters.^{Reference He, Xiao and Li36}

Spinal Hemangiopericytomas

Spinal HPCs can be divided into intradural (ID) and extradural (ED) lesions. The ID HPCs in turn, can be intramedullary (IM) or extramedullary (EM).^{Reference Merhemic, Stosic-Opincal and Thurnher37} The ED HPCs are further classified as either dural-based or primarily osseous.^{Reference Ramdasi, Nadkarni and Goel38} These distinctions, maybe in spinal HPCs more than cranial HPCs, have a dramatic and significant influence on the prospect of attaining a GTR, risk of neurological morbidity, recurrence, and so on. Radiographic features of spinal HPCs include a multilobular or “dumbbell”-shaped mass, expanding and eroding adjacent vertebral cortical bone. These lesions are typically hypointense on T1WI, moderately hyperintense on T2WI, with an homogeneous gadolinium enhancement pattern, at times with internal vessel voids (similarly to the cranial HPCs).^{Reference Zhang, Hu and Zhou39–Reference Patnaik, Jyotsnarani and Uppin41} Advanced dynamic MRI-imaging techniques, such as diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI), have been employed in the evaluation of spinal lesions.^{Reference Landi, Palmarini and D’Elia40} DTI and fiber tractography analyses allow for a better preoperative diagnosis. A better understanding of the patient’s altered white matter microanatomy and tracts in relation to the lesion is provided (for ID-IM and ID-EM HPCs), more so than a conventional MRI study, in planning the surgical intervention and counseling the patient on potential risks and accepted post-operative morbidity.^{Reference Alkherayf, Arab and Tsai42}

Computed tomography (CT) and myelography are less frequently utilized modalities nowadays for ID-IM or ID-EM HPCs, yet may still have a role in assessing the osseous components and the need for spinal stabilization upon resection (ED HPCs or those spanning both compartments). Spinal positron emission tomography/CT using either fludeoxyglucose or 11(C)-methionine has shown efficacy and a potential role in the evaluation of high-grade malignant ID-IM lesions.^{Reference Naito, Yamagata and Arima43} Protoporphyrin IX (PpIX) fluorescence induced by 5-aminolevulinic acid (5-ALA) was recently shown relevant and helpful in the detection of potential spinal-HPC tumor residual during microsurgical resection.^{Reference Millesi, Kiesel and Woehrer44}

Microsurgical Resection

Chemotherapy/Immunotherapy/Embolization

Many chemotherapeutic agents and combinations have been tested, yet an effective drug regimen against HPC is still lacking.^{Reference Mathieu110} Chamberlain et al. reported in 2008 the use of sequential multiple drug regimens in a cohort of 15 patients with recurrent HPCs who received adjuvant EBRT.^{Reference Kruse53} In their report, first line drugs consisted of cyclophosphamide, doxorubicine, and vincristine (CAV). Second-tier drugs were alpha-interferon and then ifosfamide, cisplatin, and etoposide (ICE), in case of subsequent recurrence/failure. A few temporary responses were noted with the CAV regimen, and the OS was 14 months.^{Reference Chamberlain and Glantz111} Pathologic studies on resected HPC specimens allowed Pierscianek^{Reference Pierscianek, Michel and Hindy112} to demonstrate upregulation of several key signaling pathway markers, including VEGF-VEGFR 2, EphrinB2-EphB4, and DLL4-Notch. Reported in 2016, findings were not different between HPC grade-II or grade-III. These markers may serve as potential targets for therapy, especially considering the known vascular nature of this tumor.

Angiogenic pathway studies paved the way to initial testing of bevacizumab, a monoclonal anti-VEGFR antibody commonly used for the treatment of colorectal cancer and recurrent glioblastoma in intracranial HPC.^{Reference Pierscianek, Michel and Hindy112} Initial studies show activity against HPCs when administered alone^{Reference Rutkowski, Sughrue and Kane29,Reference Michishita, Uto and Nakazawa113–Reference Sun, Liu and Wang115} or in combination with temozolomide.^{Reference Park and Araujo116} Tyrosine kinase inhibitors targeting either EphA2 or EphB4 are another potential therapeutic venue.^{Reference Barquilla and Pasquale117}

Preoperative embolization of tumor-related feeding vessels (similar to common practice in meningioma surgery, embolizing ECA vessels) can prove helpful in controlling operative bleeding. Still, such preoperative embolization plays a limited role in intracranial HPC management due to these tumors’ tendency to parasitize and invade feeding cortical vessels from both the ECA and ICA. Such angioarchitecture does not allow asymptomatic vessel sacrifice.^{Reference Cohen-Inbar, Lee and Mousavi10,Reference Olson, Yen and Schlesinger47}

External Beam Radiotherapy

The use of EBRT in HPCs postoperatively is widely accepted due to high recurrence rates after surgery alone.^{Reference Chang and Sakamoto24} The first application of EBRT for HPCs can be traced back to a report from 1974 by Dube and Paulson, where the authors reported a complete tumor response.^{Reference Dube and Paulson118} Mira et al. reported in 1977 a short series of 11 HPC patients treated with over 29 courses of EBRT, noting a positive clinical response in 26 of 29 courses.^{Reference Mira, Chu and Fortner119} EBRT was reported to decrease local recurrence rates to 12.5% after a fractionated dose of 50–64 Gy. These figures contrast an 88% recurrence rate after GTR alone.^{Reference Dufour, Metellus and Fuentes25} Some authors reported a role of EBRT as neoadjuvant therapeutic approach as well, namely prior to resection,^{Reference Uemura, Kuratsu and Hamada120} presumably by reducing tumor vascularity thus allowing for a safer surgical extirapation.^{Reference Uemura, Kuratsu and Hamada120} Multiple reports established that the response of HPCs to RT (different schedules) is dose dependent.^{Reference Schirmer and Heilman30} Total treatment doses of at least 45 Gy were shown to result in significantly superior local control rates.^{Reference Chang and Sakamoto24} Current common schedules deliver a total of 45–52 Gy doses over 25–35 fractions. SRT (a hybrid technique halfway between EBRT and SRS discussed later) can also serve as a potent salvage strategy for the treatment of recurrent intracranial HPCs.^{Reference Guthrie, Ebersold and Scheithauer3,Reference Dufour, Metellus and Fuentes25,Reference Olson, Yen and Schlesinger47} Unlike the stereotactic-focused techniques mentioned, EBRT has a leading role in treating patients with a diffuse HPC disease and widespread cranial/brain involvement.

The role of EBRT in the management of spinal HPC is controversial as well, and opinions vary. Chou et al.^{Reference Chou, Hsu and Lin77} reported in 2009 on a series of 16 patients with spinal HPCs who received adjuvant EBRT. In this cohort, EBRT had no effect on any of the different outcome parameters measured, mainly local recurrence. The histopathologic grade was the only prognostic factor deemed significant for recurrence.^{Reference Chou, Hsu and Lin77} This is a very small cohort, and any statistical analyses-based conclusions should be taken with a grain of salt (or skepticism), yet similar conclusions were drawn by other groups such as Liu et al.^{Reference Liu, Yang and Chen89} or Payne et al.^{Reference Payne, Prasad and Steiner103} in 2000.

Stereotactic Radiosurgery

The role of SRS in treating various malignant and metastatic tumors is undeniable.^{Reference Sheehan, Niranjan and Flickinger121} HPCs share different traits making this lesion well suited for SRS, such as well-defined margins from clear radiographic delineation, the potential for residual and recurrent tumor,^{Reference Tashjian, Khanlou and Vinters1} surgically challenging intracranial location (adjacent to crucial structures), and small volume lesions in unreachable locations.^{Reference Chang and Sakamoto24} Treating HPCs with SRS [with either a Gamma Knife system (Elekta AB) or CyberKnife (Accuray)] has been described extensively, with reported tumor control rates ranging 46–100%.^{Reference Cohen-Inbar, Lee and Mousavi10,Reference Coffey, Cascino and Shaw14,Reference Galanis, Buckner and Scheithauer17,Reference Chang and Sakamoto24,Reference Dufour, Metellus and Fuentes25,Reference Sheehan, Kondziolka and Flickinger28,Reference Olson, Yen and Schlesinger47,Reference Ecker, Marsh and Pollock98,Reference Payne, Prasad and Steiner103,Reference Sun, Liu and Wang115,Reference Uemura, Kuratsu and Hamada120} Of note, most reports are of small cohorts, single-center retrospective series, with only a few case series describing >20 patients, thus limiting the validity of any statistical analysis.^{Reference Penel, Amela and Decanter122,Reference Shinder, Jackson, Araujo, Prieto, Guadagnolo and Esmaeli123} A comprehensive search reveals a total of 17 studies now published reporting the utility of SRS for recurrent and residual HPC, summarized in Table 2.^{Reference Cohen-Inbar, Lee and Mousavi10,Reference Coffey, Cascino and Shaw14,Reference Galanis, Buckner and Scheithauer17,Reference Chang and Sakamoto24,Reference Kano, Niranjan and Kondziolka27,Reference Sheehan, Kondziolka and Flickinger28,Reference Olson, Yen and Schlesinger47,Reference Ecker, Marsh and Pollock98,Reference Payne, Prasad and Steiner103,Reference Kim, Kong and Seol104,Reference Kim, Kim and Chung106,Reference Sun, Liu and Wang115,Reference Iwai and Yamanaka124–Reference Tsugawa, Mori and Kobayashi127} The first of the reports is by Coffey et al.^{Reference Coffey, Cascino and Shaw14} describing a small cohort of 11 lesions in five patients, all receiving prior craniotomy (three also received prior EBRT with doses of 50–53 Gy). Prescribed margin doses reported ranged 12–18 Gy, and the mean follow-up period was 14.8 months. Eight of the nine treated tumors decreased in size significantly.^{Reference Coffey, Cascino and Shaw14} An overlapping cohort of 10 patients and 20 lesions was reported by Galanis et al.^{Reference Galanis, Buckner and Scheithauer17} (including the 5 patients reported by Coffey et al.^{Reference Coffey, Cascino and Shaw14}), treated with salvage SRS (12–18 Gy). In this cohort seven patients failed prior EBRT 30.6–64 Gy, and all treated lesions showed volume control after SRS (14 decreased in size, 4 disappeared, 2 stable in size).^{Reference Galanis, Buckner and Scheithauer17} Tian et al.^{Reference Tian, Hao and Hou128} noted that when HPC recurrence is diagnosed before age 35 years, this serves as a significant negative prognostic predictor of an earlier second relapse and shorter OS.^{Reference Tian, Hao and Hou128} Payne et al.^{Reference Payne, Prasad and Steiner103} reported a cohort of 12 patients (15 HPC lesions) treated with a mean prescription dose of 14 Gy (2.8–25 Gy). During a mean clinical follow-up period of 24.8 months, nine lesions decreased in size and three lesions remained stable. Of note, late progression was noted after 22 months.^{Reference Payne, Prasad and Steiner103}

Table 2: Cranial hemangiopericytoma SRS series

WBRT = whole brain radiotherapy; RT = radiotherapy; SRS = stereotactic radiosurgery.

⁺ Gy.

* Median, months.

** At the last follow-up.

⁺⁺ At 1, 3, and 5 years, respectively.

Sheehan et al.^{Reference Sheehan, Kondziolka and Flickinger28} reported a cohort of 14 patients harboring 15 lesions treated with SRS. This patient cohort underwent a total of 27 prior craniotomies and 7 EBRT courses (total doses of 30–61 Gy, mean 47.3 Gy). The salvage SRS prescription dose was 11–20 Gy, and the follow-up period was 5–76 months (average 31.3). Tumor regression (volume reduction) was shown in 12/15 lesions. The 5-year local tumor control and survival rates were 76% and 100%, respectively, and DNM was reported in 29%. A mean prescription (margin) dose >15 Gy was shown to result in >50% reduction of tumor volume in 80% of patients.^{Reference Sheehan, Kondziolka and Flickinger28} Similar conclusions can be found in a report by Kano et al.^{Reference Kano, Niranjan and Kondziolka27} of a cohort of 20 patients (29 lesions) treated with SRS. The authors reported significantly better PFS (p = 0.0023) when patients were treated with a margin dose >14 Gy, with a 5-year PFS of 75.4% versus 56.3%, respectively (p = 0.0023).^{Reference Kano, Niranjan and Kondziolka27} DNM or ENM developed in 79.2 (range 12.2–158.3) months on average after the initial diagnosis. Local tumor control rates (volume reduction or stability) were 72.4% (n = 21), 20% (n = 4) died of DNM, and 5% (n = 1) died of ENM (liver and lung). Complete resolution was noted in 17.2% (n = 5) of the WHO-II HPCs.^{Reference Kano, Niranjan and Kondziolka27} Chang and Sakamoto^{Reference Chang and Sakamoto24} reported the use of a higher prescription dose (16–24 Gy, mean 20.5 Gy) in a cohort of eight HPC patients, achieving a 75% (n = 6) tumor reduction rates during a mean follow-up of 44 months (range 8–77).^{Reference Chang and Sakamoto24} Ecker et al.^{Reference Ecker, Marsh and Pollock98} reported a cohort of 38 patients (n = 22 grade-II, n = 16 grade-III HPCs), all treated with EBRT with or without SRS.^{Reference Stessin, Sison and Nieto50} A grade-III HPC histology was shown to result in recurrence significantly earlier (3.3 vs. 10 years, p = 0.004). The authors concluded that harnessing SRS for the treatment of recurrent disease contributed to a better OS.^{Reference Ecker, Marsh and Pollock98}

Melone et al.^{Reference Melone, D’Elia and Santoro7} reported a series of 36 HPCs. The actuarial 5- and 10-year recurrence rates were 50% and 72%, respectively. Adjuvant ionizing radiation (all modalities, mean prescription dose 16 Gy) was shown to significantly decrease the rates of recurrence (p = 0.04), not improving OS (p = 0.2).^{Reference Melone, D’Elia and Santoro7} Kim et al.^{Reference Kim, Kong and Seol104} recently published a retrospective analysis of 18 patients (WHO-II = 8, WHO-III = 10) and 40 lesions (WHO-II = 13, WHO-III = 27) treated with SRS. The median OS was 134.7 months, and the actuarial survival rates at 1, 5, and 10 years were 85.6%, 85.6%, and 37.4%, respectively. Local tumor control was 80% (n = 32), and the average local recurrence-free interval (per lesion) for WHO-II and WHO-III were 86.1 and 40.5 months, respectively (p = 0.010). ENM developed in seven (38.9%) patients with WHO-III HPCs.^{Reference Kim, Kong and Seol104}

Jeon et al.^{Reference Jeon, Park and Kim129} recently reported on the efficacy of adjuvant RT in a cohort of 49 patients with intracranial HPC. Of the cohort, 31 patients received adjuvant RT after surgery, 26 with EBRT and 5 with SRS (Gamma Knife). The median follow-up period was 50 months (range 3–216). The authors concluded that local tumor control was better with GTR followed by RT than GTR alone (p = 0.056), with no difference in OS. Local tumor control and OS after STR + RT were equivalent to those after GTR alone. Tumor volume >40 cm³ was associated with poor PFS (p = 0.024).^{Reference Jeon, Park and Kim129}

In 2015, we reported the largest HPC cohort to date, a collaborative effort through the International Gamma Knife Research Foundation (IGKRF), done in an attempt to study outcomes of SRS for HPCs.^{Reference Cohen-Inbar, Lee and Mousavi10} Eight centers pooled patients together to form a cohort of 90 patients and 133 discrete lesions (78.9% n = 71 WHO-II, 21.1% n = 19 WHO-III). Prior treatment modalities included embolization (n = 8), chemotherapy (n = 2), and EBRT (n = 34). The median tumor volume was 4.9 cm³ (0.2–42.4 cm³), the median prescription dose was 15 Gy (2.8–24 Gy), and the median follow-up period was 59 months (range 6–190). Local tumor control was noted in 55% of tumors and 62.2% of patients. DNM was noted in 27.8% of patients, and ENM was noted in 24.4%. The actuarial OS at 2, 4, 6, 8, and 10 years was 91.5%, 82.1%, 73.9%, 56.7%, and 53.7% respectively. Local PFS at 2, 4, 6, 8, and 10 years was 81.7%, 66.3%, 54.5%, 37.2%, and 25.5%, respectively, after initial SRS (Figure 1).^{Reference Cohen-Inbar, Lee and Mousavi10}

Figure 1: Sample Patient, Hemangiopericytoma. (A) Axial T1WI with gadolinium taken on the day of stereotactic radiosurgery (SRS) for a right tentorial hemangiopericytoma, s/p 1 surgical resection, 4/2002. (B) Axial T1WI with gadolinium taken on the day of repeat SRS for the same lesion, 12/2005. (C) CT-angio reconstruction pre-embolization and second surgical resection of the same lesion due to progression, 6/2009. (D) Follow-up images, 20 months after the second surgical resection, in T1WI-FLARE, showing final local control, 02/2011. Adapted from Cohen-Inbar et al.^{Reference Cohen-Inbar, Lee and Mousavi10}

There are no cohort reports on outcome after SRS for spinal HPC, rather a few isolated case reports, all depicted and described in Table 1.^{Reference Schirger, Uihlein and Parker52–Reference Sweid, Noureldine and Nasser96} Our understanding on the role of SRS in treating spinal HPC is based on the cranial HPC’s series reports. The patient-friendly design and supporting science and data of SRS makes this a valuable tool in our armamentarium, and future studies are required to directly prove its role in the management of spinal HPC.