Childs Nerv Syst (2003) 19:808–817
DOI 10.1007/s00381-003-0839-5
Mubarak Al-Gahtany
Manohar Shroft
Eric Bouffet
Peter Dirks
James Drake
Robin Humphreys
Normand Laperriere
Cynthia Hawkins
James Rutka
Received: 28 April 2003
Revised: 17 July 2003
Published online: 22 October 2003

Springer-Verlag 2003
M. Al-Gahtany · P. Dirks · J. Drake ·
R. Humphreys · J. Rutka ())
Division of Neurosurgery, Suite 1502,
Hospital for Sick Children,
University of Toronto,
555 University Avenue, Toronto,
M5G 1X8, Canada
e-mail: james.rutka@sickkids.ca
Tel.: +1-416-8136425
Fax: +1-416-8134975
M. Shroft
Division of Neuro-Radiology,
Hospital for Sick Children,
University of Toronto,
555 University Avenue,
Toronto, M5G 1X8, Canada
E. Bouffet
Brain Tumour Program,
Hospital for Sick Children,
555 University Avenue, Toronto,
M5G 1X8, Canada
N. Laperriere
Division of Radiation Oncology,
Hospital for Sick Children,
University of Toronto,
555 University Avenue, Toronto,
M5G 1X8, Canada
C. Hawkins
Division of Neuropathology,
Hospital for Sick Children,
University of Toronto,
555 University Avenue, Toronto,
M5G 1X8, Canada
ORIGINAL PAPER
Primary central nervous system
sarcomas in children:
clinical, radiological, and pathological features
Abstract Patients and methods: The
clinical, radiological, surgical, and
pathological findings of 16 children
with a primary central nervous sys-
tem (CNS) sarcoma are reported.
There were 8 (50%) girls and 8 (50%)
boys ranging in age from 4 months to
14 years (mean age 4.8 years). Four
patients (23%) were in their 1st year
of life. Fourteen children (87%) had
an intracranial sarcoma, and 2 (13%)
had intraspinal tumors. Nine intra-
cranial tumors (60%) were supraten-
torial. The parietal and temporal
regions were the most frequently
involved sites. Results: Characteristic
imaging findings included tumor
cysts in 7 patients and marked tu-
moral enhancement in 9 (69%) with
intratumoral calcification and hem-
orrhage. All patients underwent at
least one operation to surgically
remove the tumor with the aim of
maximal resection and 3 patients
underwent a second resection due to a
recurrent tumor. Resection was total
in 9 (53%) patients and subtotal in
another 7 (41%). Dural attachment by
tumor was confirmed in 7 (44%)
patients and parenchymal invasion
was present in 9
(56%). In one-third
of the patients there
was a well-defined plane of dissec-
tion around the tumor. Postoperative
radiation was used in 10 patients.
Postoperative chemotherapy was
used in all but 2 patients. Immuno-
histochemical studies were available
in 13 patients with the most consis-
tent finding being strong vimentin
positivity. Five out of the 6 patients
in whom the proliferation markers
were obtained demonstrated a high
proliferation index
(Ki-67 labeling index, 20–50%). The
mean length of survival in the group
was 4.6 years (range 1 month to
16 years). Children who presented in
the 1st year of life had shorter
survival than those who presented at
an older age. Six patients (40%) had
cerebrospinal fluid (CSF) dissemina-
tion of the tumor. CSF dissemination
was associated with a shorter mean
survival of 1.9 years. Conclusions:
Our review of this series of patients
indicates the requirement for adju-
vant therapy and for continued efforts
to classify tumor subtypes aimed at
optimizing future treatments for pa-
tients with a primary CNS sarcoma.
Keywords Sarcoma · Primary ·
Brain · Spinal cord · Child · Survival

Introduction
Primary sarcomas of the central nervous system (CNS)
are rare. Their cell of origin was and still is a controver-
sial issue. The most widely accepted theory traces their
origin to pluripotential primitive mesenchymal cells in the
dura mater, the leptomeninges or their pial extensions into
the brain and the spinal cord along the periadventitial
spaces, the tela choroidea, and the stroma of the choroid
plexus [45].
Secondary sarcomas of the CNS refer to those
sarcomas that have arisen from metastases of soft tissue
or bone sarcomas to the CNS. They are also rare and
result mostly from the extension of sarcomas in locations
neighboring the CNS. They will not be considered further
here.
In this study, we present the clinical, radiological,
surgical, and pathological data of 16 patients with a
primary CNS sarcoma seen at the Hospital for Sick
Children between 1960 and 1999. The response of these
rare tumors to different treatments is also discussed.
Patients and methods
We reviewed our CNS tumor database at the Hospital for Sick
Children for the period 1960–1999 searching for all types of
sarcomas that involved the CNS. Permission to review these
patients’ data was granted by the Research Ethics Board at The
Hospital for Sick Children. Primary CNS sarcomas in this study
were defined as tumors that originated in the CNS from non-
neuronal, non-glial, and non-reticular elements, and are not a result
of the sarcomatous transformation of a previously known benign
tumor. Using this definition, we included meningeal sarcomas but
excluded entities such as primitive neuroectodermal tumor (PNET),
astrocytoma, reticulum cell sarcomas, Ewing’s sarcomas, and
meningioma. We also excluded tumors such as hemangiopericyto-
ma and malignant meningioma. We included, however, tumors that
have arisen in the CNS following radiation and/or chemotherapy.
The medical records of those patients whose diagnoses coin-
cided with our definition of a primary CNS sarcoma were then
reviewed for clinical, radiological, surgical, and pathological
details. Our search revealed a total of 40 cases of sarcoma
involving the CNS. Of these, 22 cases were secondary sarcomas
and 1 case was reticulum cell sarcoma (lymphoma). The remaining
16 cases were true primary sarcomas of the CNS.
All patients with a primary sarcoma of the CNS had a complete
work-up and investigation that excluded the presence of sarcomas
outside the nervous system at the time of presentation. The work-up
consisted of chest X-ray, CT of the chest, abdomen and pelvis,
ultrasound of the abdomen, liver and spleen isotope scan, gallium
scan, and bone marrow aspiration/biopsy.
All the surgical specimens were examined by light microscopy
with hematoxylin and eosin stain, PAS (periodic acid-Schiff), and
reticulin. Immunohistochemical studies were used in 13 patients.
Electron microscopic studies were used in 13 patients. Molecular
studies were used in 2 patients.
The follow-up periods ranged from 1 month to 16 years (mean
4.6 years). The length of survival was calculated from the date of
diagnosis and the end points were the death of the subject or the last
follow-up. The Student t-test was used to compare the means of the
different groups.
809
Results
Clinical features
The clinical findings in this series are summarized in
Table 1. There were 8 (50%) girls and 8 (50%) boys
ranging in age from 4 months to 14 years (mean age
4.8 years). Four patients (23%) were in their 1st year of
life. Fourteen patients (82%) were in the first decade of
life.
Of the 16 cases, 14 (87%) sarcomas were intracranial
and 2 (13%) were intraspinal. Of the intracranial sarco-
mas, 9 (60%) were supratentorial, divided equally
between the two sides (4 in the right side, 4 in the left
side and 1 involving both sides), and the other 5 (40%)
sarcomas were located in the posterior fossa. The parietal
and temporal locations were by far the most common
areas involved in the supratentorial tumors. The intraspi-
nal sarcomas were located in the cervical region and in
the lumbar region.
The presenting symptoms and signs depended on the
child’s age at presentation and on the location of the
tumor. Symptoms and signs of increased intracranial
pressure (ICP) and focal neurological deficit such as
seizure, hemiparesis, paraparesis, cranial nerves palsies,
and cerebellar symptoms and signs were the most
common.
Only 1 patient had a known predisposing factor
(Patient 4). The patient was diagnosed with acute
lymphocytic leukemia 10 years prior to his presentation.
At that time he received cranial irradiation with a total
dose of 1,800 cGy in 10 fractions as part of his leukemia
treatment. He was also on a maintenance dose of
methotrexate, vincristine, and 6-mercaptopurine for
3 years after the induction phase of his treatment.
Radiological features
The radiological findings are shown in Table 1. CT was
the most widely used method of investigation (Fig. 1). In
all patients in whom CT was performed, enhancement
with iodinated contrast had been obtained. Radioisotope
scan was obtained in 4 patients; 3 of them showed
increased radioisotope uptake by the tumor. Cerebral
angiogram was obtained in 4 patients; 3 of them with
superficially located tumors showed richly vascular
tumors with predominantly external carotid artery supply,
while the angiogram was negative in the 4th patient with a
more deeply seated tumor. Out of 13 patients with
intracranial tumors who had CT scans of the head, the
tumor had a cystic component in 7 patients (54%) and
marked tumoral enhancement was present in 9 (69%)
patients. One patient had significant intratumoral cal-
cification and another one had intratumoral and intraven-
tricular hemorrhage simulating an arteriovenous mal-
 810

Table 1 Tumor location, radiology, treatment and outcome in 16 patients. Rt right; Lt mycin, cyclophosphamide; vin vincristine; cyc cyclophosphamide; act actinomycin;
left; T temporal; P parietal; O occipital; F frontal; WB whole brain; NA not available; adr adriamycin; ifo ifosfamide; carb carboplatin; cisp cisplatin; MTX methotrexate;
ND not done; Fr fraction; alt alternate with; C cycle; ICE ifosfamide; VP-16 CS craniospinal; + tumor positivity for antigens listed;
tumor negativity for antigens
carboplatin; VIC vincristine, idarubicin, cyclophosphamide; VAC vincristine, actino- listed

Patient Age, sex Location CT scan MRI Extent of Radiation (cGy) Chemotherapy CSF Distant Status/survival
resection spread metas-
tasis
1 5 months, girl Lt. F-P Cystic/solid, ND Subtotal No ICE/VIC 1 C No No Died, 1 month
enhancement ++
2 4 months, boy Lt. F-P Enhancement +++, ND Total Local 4,500/23 Fr Act, adr, cyc, vin No No Alive, 16 years
bone destruction (2 years)
3 6 years, girl Rt. T, thalamus Tumor, hemorrhage, ND Subtotal WB 5,400/30 Fr ICE/VIC x8 C Yes No Alive, 3 years
enhancement +++ 11 months
4 14 years, boy Lt. P Cystic/solid, Enhance- Total CS (dose: NA) ICE 8 C No No NA
enhancement ++ ment ++
5 1 year 7 months, Posterior fossa Vermian, ND NA No ICE 2 C alt vin, Yes No Alive, 1 year
girl enhancement + cyc, VP-16 11 months
6 6 years 6 months, Posterior fossa Vermian, ND Subtotal CS 5,000/25 Fr Act, adr, cyc, vin No No Alive, 14 years
boy hyperdense, cystic, 2 years
enhancement ++
7 2 years 6 months, O, midline Enhancement ++, ND Total Local 5,040/28 Fr VAC alt ifo/VP-16 No No Alive, 12 years
girl bone erosion 1 year
8 4 months, boy Rt. T-P Cystic/solid, Enhance- Total No Ifo, VP-16, Yes No Died, 3 months
calcification, ment ++ carb 3 C
enhancement ++
9 2 years 6 months, Cervical spine Intradural, ND Total 2 CS 5,300/31 Fr VP-16, cisp alt cyc, Yes No Alive, 3 years
boy extramedullary vin
10 5 years, girl Posterior fossa Hypodense, ND Total 2 Local 5,900 ICE NA NA Alive, 7 months,
enhancement + lost to follow-up
11 13 years, boy Rt. T-P Cystic/solid, ND Total CS 5,400/30 Fr Carb, ifo, VP16, No No Alive, 5 years
enhancement ++ vin 8 C
12 1 year, boy Posterior fossa Cystic/solid, ND Subtotal No No NA NO Died, 1 month
enhancement ++
13 1 year 5 months, Rt. T-P Cystic/solid, ND Subtotal Local 4,500/25 Fr, ICE intrathecal Yes Yes Died, 1 year
boy bone erosion, CS 2,520/14 Fr MTX 2 C 10 months
enhancement ++
14 4 years, girl Posterior fossa ND (ventriculogram) ND Subtotal No No No No Died, 3 years
5 months
15 9 years, girl Lumbar spine Epidural L5-S1 Enhance- Total 2 No ICE alt, adr, cyc, No No Alive, 6 months,
ment  vin lost to follow-up
16 4 years 10 months, Lt. O-P Bone erosion, ND Total Cranial 5,250/28 Act, adr, cyc, vin Yes No Died, 10 months
girl enhancement ++ Fr CS

Fig. 1A, B Five-month-old girl
presenting with irritability,
vomiting, seizures, and a tense
fontanelle. Contrast-enhanced
A axial CT and B coronal CT
images show a large left frontal
tumor with a large cystic com-
ponent (asterisk) and enhancing
solid components medially and
inferiorly (black arrows). The
pathological diagnosis was an
undifferentiated sarcoma of the
left frontal lobe
Fig. 2A–C Six-year-old girl who presented with a chronic head-
ache that intensified and was accompanied by vomiting, drowsi-
ness, and sudden left hemiparesis shortly prior to presentation. A
Unenhanced CT shows a right temporal lobe tumor. Hyperdense
areas within it are due to bleeding (arrow). Note small amount of
blood in the fourth ventricle. B Unenhanced CT through a superior
level shows mass effect and hyperdense blood along the right
formation (Fig. 2). However, the angiogram was nega-
tive. MR imaging in selected patients revealed lesions
with increased signal on T2-weighted images, flow
voids on FLAIR sequences, and irregularly enhanc-
ing lesions following administration with gadolinium
(Fig. 3). Evidence of leptomeningeal metastases was best
appreciated on the contrast-enhanced scans by MR
(Fig. 4).
Surgery
All patients with a primary CNS sarcomas had at least one
operation to surgically remove the tumor with the aim of
811
lateral ventricle (arrow). C On contrast-enhanced CT, several
enhancing areas were noted (arrow) and a vascular malformation
that had bled was considered in the differential diagnosis.
Angiography (not shown), however, was negative for a vascular
malformation. The pathological diagnosis in this case was an
undifferentiated sarcoma of the right temporal lobe with intratu-
moral and intraventricular bleeding
maximal resection; 3 patients underwent a second resec-
tion due to a recurrent tumor. Resection was total in 9
(53%) patients and subtotal in another 7 (41%). A limited
biopsy was performed in 1 patient. Postoperative radia-
tion was used in 10 (65%) patients and postoperative
chemotherapy was used in 15 (88%) patients.
The operative details are shown in Table 2. At surgery,
dural attachment was confirmed in 7 (44%) patients and
parenchymal invasion was present in 9 (56%) patients. In
one-third of the patients the tumor was found intraoper-
atively to be richly vascular and in about one-third of the
patients there was a well-defined plane of dissection
around the tumor. Encasement of the branches of the
major arteries was present in only 1 patient.
812

Fig. 3A–D One-and-a-half-

year-old girl who presented
with irritability, vomiting, and
head banging lasting for several
weeks, which progressed to
lethargy and drowsiness shortly
before presentation. A Axial
FLAIR image on MR shows a
midline posterior fossa mass
with some flow voids within it
due to abnormal vessels (ar-
rows). B T2-weighted image on
MR shows the mass to be well
defined and have an increased
signal. Note the flow void due
to a vessel (arrow). C Contrast-
enhanced axial T1 image. This
shows some patchy enhance-
ment within the tumor (arrows).
D Contrast-enhanced coronal
image shows the inferior extent
of the tumor up to the foramen
magnum (arrow). Radiologi-
cally, a differential diagnosis of
an astrocytoma or ependymoma
had been entertained. The final
diagnosis in this case was an
undifferentiated sarcoma of the
posterior fossa

Pathological features microvilli are some of the helpful and commonly

The histological features of the different intracranial
sarcomas were similar to their extracranial counterparts.
Immunohistochemical studies were available for 13
patients (Table 1). The most consistent findings were
strong vimentin positivity and general negativity for both
neuronal and glial markers. Desmin was found to be
positive in the tumors with more rhabdomyoblastic
differentiation.
Electron microscopic studies were available in 13
patients, and were found to be helpful in confirming the
mesenchymal nature of the tumor cells and ruling out
neuronal and glial neoplasms by virtue of the ultrastruc-
tural appearance of the tumor cells. The spindle shape of
the cells, dilated rough endoplasmic reticulum, lack of
well-developed junctions, abundance of intermediate
filaments, and absence of neurosecretory granules and
encountered ultrastructural features of sarcomas. Five
out of the 6 patients in whom the proliferation markers
were obtained demonstrated a high proliferation index
(Ki-67 labeling index, 20–50%).
The final pathological diagnosis was an undifferenti-
ated sarcoma in 8 (50%) patients, a meningeal sarcoma in
4 (25%), an otherwise non-specified sarcoma in 2
(12.5%), and a rhabdomyosarcoma in 1 (6.25%). Two
of the otherwise non-specified sarcomas could not be
classified despite the use of immunohistochemistry and
electron microscope (Patients 5 and 11).
Survival
The length of survival was calculated from the date of the
diagnosis (Table 1). Information on survival was absent in
 813

Fig. 4A–D Fourteen-year-old

boy who presented with a
headache, vomiting, and blurred
vision lasting for several weeks
accompanied by bilateral papil-
ledema, right-sided visual field
defect, and right-sided weak-
ness. The patient had received
whole brain and spine radiation
for prophylaxis of acute lym-
phcytic leukemia 10 years ago.
A T1-weighted axial image on
MR shows a large solid and
partly cystic tumor in the left
parietal lobe, extending up to
the dural surface. Note the mass
effect on the ventricular system
and midline shift to the contra-
lateral side. B T2-weighted ax-
ial image on MR shows
hyperintensity of the tumor with
a thin capsule and surrounding
vasogenic edema. C Contrast-
enhanced coronal MR through
the tumor shows enhancement
of its margin. Bright signal
within the tumor probably rep-
resents proteinaceous fluid
content. D Postoperative MR
shows residual tumor (arrow-
head) and metastatic enhancing
nodules (arrows) in the adjacent
left parietal lobe and the left
temporal lobe. The final diag-
nosis was a poorly differentiat-
ed sarcoma of the left parietal
lobe

1 patient. Two patients, who were known to be alive at 6
and 7 months from the time of the diagnosis, were lost to
follow-up largely because they were referrals from
outside the country. Information on survival was available
on the other 13 patients and they were the only patients
we used in the analysis.
The longest survival in the whole group was 16 years
and the shortest was 1 month (mean 4.6 years). Children
who presented in the 1st year of life had shorter survival
than those who were older than 1 year at the time of
presentation. The longest survival (16 years), however,
was in the former group, giving them a mean length of
survival of 4.1 years, while the mean length of survival of
the latter group was 4.8 years.
For supratentorial tumors, lengths of patient survival
ranged from 1 month to 16 years (mean 5 years), while for
the infratentorial location the range was 1 month to
14 years (mean 4 years). While 1 patient with a spinal
tumor location is known to be alive 3 years after
presentation, the other 2 spinal cases were among those
lost to follow-up.
Six (40%) patients had cerebrospinal fluid (CSF)
dissemination of the tumor. Four of them were undiffer-
entiated sarcomas, 1 was a meningeal sarcoma, and 1 was
an otherwise non-specified sarcoma. In this group of
sarcomas with CSF dissemination, the length of survival
was shorter with a range of 3 months to 4 years (mean
1.9 years) compared with a range of 1 month to 16 years
(mean 7.9 years) in those with no CSF dissemination.
Only 1 patient developed distant metastases to the
abdomen in the late stages of his illness and eventually
succumbed to his disease.
The length of survival in the patients who had total
surgical resection of the tumor ranged from 3 months to
16 years (mean 6.2 years); in those with subtotal resection
the range of length of survival was 1 month to 14 years
(mean 3.4 years).
814

Table 2 Surgical and pathological findings in 16 primary CNS tumor; vim vimentin; GFAP glial fibrillary acidic protein; ker
sarcomas. MCA middle cerebral artery; PCA posterior cerebral keratin; des desmin; EMA epithelial membrane antigen; myo
artery; T6 6th thoracic vertebra; PNET primitive neuroectodermal myosin; S100-P S100-protein
Case Dural Paren- Operative note Preoperative Quick section Pathological diagnosis
attach- chymal clinical diagnosis diagnosis
ment invasion
1 No Yes Soft, cystic/solid Astrocytoma Astrocytoma Undifferentiated sarcoma,
vim +
2 Yes No Arising from dura with full Meningeal tumor Fibroblastic Meningeal sarcoma
thickness involvement tumor
3 No Yes Encasing MCA, PCA branches Vascular malformation Anaplastic tumor Undifferentiated sarcoma,
vim, ker, des, EMA
4 No Yes Moderately vascular, necrotic Astrocytoma Astrocytoma Undifferentiated sarcoma,
vim, ker, des, GFAP,
S100, act +
5 NA NA NA Ependymoma Atypical glioma Sarcoma, vim +
6 No Yes Rubbery, no BS invasion Medulloblastoma NA Rhabdomyosarcoma, des,
myo +
7 Yes No Extra-axial, soft, vascular, Meningioma NA Undifferentiated sarcoma
eroded the inner table vim, act, ker, S-100 +
8 No Yes Soft, avascular, good plane Astrocytoma Astrocytoma Undifferentiated sarcoma
vim, ker, EMA +
9 No Yes Mainly extramedullary, PNET NA Undifferentiated sarcoma
some intramedullary extension, vim, S-100 +
avascular, good plane
10 Yes Yes Vascular, solid, attached to the NA NA Meningeal sarcoma vim +
tent
11 No Yes Cystic/solid, good plane Astrocytoma Astrocytoma Undifferentiated sarcoma
vim, S-100 +
12 No Yes Cystic/solid, vascular, Astrocytoma Astrocytoma Meningeal sarcoma vim,
good plane ker, S-100 +
13 Yes Yes Cystic/solid, mainly Astrocytoma Astrocytoma Meningeal sarcoma vim,
intraparenchymal EMA, S-100 +
14 No No Vermis, fourth ventricle Medulloblastoma NA Sarcoma
and right cerebellum,
well demarcated
15 Yes No Firm, fleshy, no extension NA NA Undifferentiated sarcoma
outside the canal des +
16 Yes Yes Vascular, bone erosion Meningeal tumor Malignant tumor Undifferentiated sarcoma

Adjuvant therapy Patients with undifferentiated sarcomas ranged in

Postoperative radiation was used in 10 patients; all of
them had a complete course of fractionated radiation with
a total dose that ranged from 4,500–5,900 cGy. The
survival in these patients ranged from 10 months to
16 years (mean 7 years). Six patients who received
radiation therapy were given craniospinal irradiation. Six
patients had no radiation, and their survival range was
3 months–1 year and 11 months (mean, 0.5 year). Since 2
patients in the no radiation group died in the 1st month
post surgery, if they are excluded from the analysis the
results are still significant.
Postoperative chemotherapy was used in all patients
except two who died within months of their presentations.
The different chemotherapeutic agents and their various
combinations are shown in Table 1. The most commonly
used agents were ifosfamide, VP-16, carboplatin, vincris-
tine and cyclophosphamide.
length of survival from 1 month to 12 years (mean
3.3 years), while the meningeal sarcoma survival range
was 1 month to 16 years (mean 6 years) and those whose
tumor was labeled as an otherwise non-specified sarcoma
had a survival range of 3.5 months to 5 years (mean
2.4 years). The only patient with a rhabdomyosarcoma
was alive after 14 years.
Long-term sequelae of treatment
The long-term complications of radiotherapy and chemo-
therapy were evident in those patients who had long
survival. Patient 2 who was alive 16 years post treatment,
was suffering from speech delay, memory difficulty,
learning and behavioral problems. This was also true of
Patient 6 who was alive 14 years post treatment. In
addition to the development of panhypopituitarism, this
patient acquired a posterior fossa venous malformation

that required surgical resection. Patient 7, who was alive
12 years post treatment, had a decrease in school
performance. This patient developed hepatosplenomega-
ly, which was related to chemotherapy. Patient 9, who
was alive 3 years post treatment, had myelomalacia
related to spinal cord radiation and Patient 3, who was
alive 4 years post treatment, developed peripheral
neuropathy related to vincristine.
Discussion
Bailey first described the occurrence of a sarcoma in the
brain in 1929 [2]. All forms of sarcoma of the CNS are
rare. Secondary CNS involvement is a rare event in the
natural history of the sarcoma, occurring in approximately
1.3% of cases [6, 14, 30]. Advances in chemotherapy
have enabled more sarcoma patients to realize longer
survivals than in the past; however, aggressive chemo-
therapy regimes are also thought to be responsible for
more cases of secondary CNS sarcomas being described
than previously [6, 26, 45].
Primary sarcomas of the CNS, however, are still rare.
The reported incidence of primary CNS sarcomas varies
in different studies and ranges from 0.1% to 4.3% [3, 19,
27, 31, 33, 36]. The main reason for this variation is the
inconsistency in the definition of a primary CNS sarcoma
in multiple studies. Previous reports included entities such
as giant cell sarcoma, reticulum cell sarcoma, circum-
scribed sarcoma of the cerebellum, and hemangiopericy-
toma, which gave a falsely high incidence of primary
CNS sarcomas [7, 20, 21, 31, 35]. Other studies depicted
primary glial, neuronal, neuroectodermal, and/or previ-
ously known benign meningeal tumors with sarcomatous
differentiation as sarcomas, another reason for a falsely
high incidence of primary CNS sarcomas in these studies
[23, 31]. Furthermore, in many studies there was no
utilization of immunohistochemistry, a technique that
could have resolved diagnostic issues for many of the
tumors examined [23, 33, 36, 39, 43].
The cell of origin of primary sarcomas of the CNS is a
controversial issue. The most widely accepted theory
traces their origin to pluripotential primitive mesenchy-
mal cells in the dura mater, the leptomeninges or their pial
extension into the brain and the spinal cord along the
periadventitial spaces, the tela choroidea, and the stroma
of the choroid plexus [2, 4, 11, 24, 25, 31, 33, 35, 45]. The
etiology of the primary CNS sarcoma is not understood.
Of all potential etiological agents, only radiation therapy
as an inciting agent is well accepted in the literature [19,
23, 27, 34, 38, 44]. In our series, there was 1 case of a
radiation-induced sarcoma. Trauma (including non-tu-
moral surgery) has been described as a predisposing
factor but the evidence here is far from conclusive [17,
19, 27, 29, 31]. Genetic factors may play a role in some
sporadic cases. In addition, some familial cases have been
815
reported [19, 31]. Chemicals such as methylcholanthrene
and phenoxyacids as well as viruses such as RSV and
SV40 have been reported to cause brain and meningeal
sarcomas in experimental animals and humans [8, 21, 27,
31]. One of our patients (Patient 4) had received
chemotherapy in addition to radiation 10 years prior to
the development of his brain sarcoma.
Primary CNS sarcomas can occur at any age, but are
more likely to be found in children [23, 31, 44].
Congenital primary CNS sarcomas have been reported
[49]. In our series sarcomas were equally distributed
throughout the age range 1–18 years, the youngest patient
being 4 months old.
The clinical presentation of patients with a primary
intracranial sarcoma is not different from that of other
intrinsic and meningeal tumors. However, due to their
extreme vascularity, CNS sarcomas can present with
intracranial hemorrhage [11, 15, 27, 30]. In our series, 1
patient had a sarcoma that simulated a vascular malfor-
mation both clinically and on CT scan (Patient 3).
The appearances of primary brain sarcomas on CT
and/or MRI are non-specific. There are no specific
imaging criteria to differentiate primary brain sarcomas
from other varieties of brain tumors or inflammatory
masses. In our series, tumors tended to be large and were
either heterogeneous in density, cystic (hypodense), or
solid. Seven of the 17 tumors had both cystic and solid
components. In one instance, intratumoral hemorrhage
caused increased density of the tumor. In 3 cases in which
MRI was available an inhomogeneous signal was noted in
the available spin echo images. After intravenous con-
trast, solid components of the tumor usually enhanced.
Contrast enhancement has been a feature of sarcomas that
has previously been described in the literature [34, 35]. In
2 out of the 4 cases in which cerebral angiography was
available the vascularity of the tumor was observed as a
“tumor blush”. Tumor vascularity has also been men-
tioned in previous reports [14, 15]. An important feature
noted on imaging studies in the present report is the
location of the tumor close to the cortical surface or the
meninges, a feature seen in 11 out of 17 patients. Because
of the high incidence of leptomeningeal spread of the
sarcoma [11, 13, 45], it is important to image the entire
neuraxis at the time of diagnosis and during subsequent
follow-up. CSF dissemination was present in 40% of
cases in our series. The differential diagnosis for primary
CNS sarcoma on imaging studies includes glioma,
meningioma, medulloblastoma, and ependymoma [19,
27, 32, 36].
Our experience that most primary CNS sarcomas are
located supratentorially with the parietal and temporal
lobes being the most common sites of origin is in
agreement with other studies [13, 22, 23, 32, 36, 45].
None of our cases was intraventricular, although well-
documented cases of intraventricular sarcomas exist and
are thought to arise from the choroid plexus [23, 26, 28,
816

40, 41, 45]. Dural involvement is a common finding that
is found in 60% of all primary CNS sarcomas [5, 17, 36,
44, 45, 47]. Dural involvement was observed in approx-
imately half of the patients in our series. None of the
patients had malignant transformation of glial elements
within or adjacent to the primary sarcoma. The occur-
rence of a reactive glioma in a primary CNS sarcoma,
however, is well documented [16, 23, 29, 46].
Primary CNS sarcomas can metastasize outside the
nervous system, the most frequent sites being the lung,
bone, and liver. Extra-CNS metastases are usually
observed following intracranial surgery, and are associ-
ated with a poor prognosis [3, 12, 13, 22]. One patient in
our series with an intracranial sarcoma developed metas-
tases to the peritoneum following intracranial surgery and
placement of a ventriculo-peritoneal shunt.
In our study, the labeling index for the primary CNS
sarcoma was high in over 80% of cases in which Ki-67
immunolabeling was performed. This observation is in
agreement with other studies that have also examined
labeling indices in primary CNS sarcomas [8]. Virtually
any sarcoma that arises outside the CNS has also been
described as a primary CNS sarcoma [1, 4, 8, 10, 19, 24,
25, 31, 32, 36, 37, 43, 44, 45]. The frequency of the
different subtypes of primary CNS sarcomas is difficult to
determine. This is primarily due to the rarity of the lesion,
and the lack of unified criteria for including specific
subtypes in the cases previously reported [40]. In
addition, the classification system for sarcomas has
changed over time with the appearance of new entities
such as malignant fibrous histiocytoma (MFH), which is
now considered to be the most common soft tissue
sarcoma [5, 16, 27, 38]. Interestingly, MFH and fibrosar-
comas tend to occur in adults while sarcomas in children
tend to be more undifferentiated or show muscle differ-
entiation [36, 42, 49]. In our study the undifferentiated
sarcoma was the most common pathological subtype of
tumor, and this is in agreement with others [45].
However, MFH and fibrosarcomas have been identified
as the most common in other series [9, 36]. Our inability
to precisely define the subtype of several sarcomas in our
study is similar to the experience of others [40].
In general, primary CNS sarcomas carry a poor
prognosis, although long postoperative survival is well
documented [13, 31, 32, 34, 40, 45]. Most reports in the
literature point towards radical surgical resection as the
treatment of choice. There is no consensus, however, on
the utility of postoperative radiation or chemotherapy [11,
13, 17, 18, 38, 40, 45, 48]. Our study suggests that
maximal surgical resection followed by radiation therapy
may give the patient the best treatment possible. Repeat
resection for recurrent tumors can be offered as long as
safe resection remains feasible. The adverse effects of
radiation therapy on young children, as shown in some of
our long-term survivors, remain a significant concern at
this time. The value of chemotherapy could not be
specifically addressed in our study. However, since all
long-term survivors in this study received chemotherapy,
this would at least argue for its potential effectiveness.
Conclusion
The primary sarcoma of the CNS is a rare disease that
most commonly arises in the supratentorial compartment
in children. Dural attachment and CSF dissemination are
frequently found. Extra-CNS metastases are associated
with a very poor prognosis. Aggressive surgical resection
followed by postoperative radiation therapy may offer a
chance for long-term survival. Chemotherapy regimens
directed specifically towards sarcomas may be added so
long as they are tolerated by the patient. Repeat crani-
otomy can be offered for recurrent local disease. Future
studies should be directed at increasing our understanding
of the different subtypes of sarcoma for which specific
chemotherapeutic agents may have a greater role to play.

References
1. Anderson WR, Cameron JD, Tsai SH
(1980) Primary intracranial leiomyo-
sarcoma. Case report with ultrastruc-
tural study. J Neurosurg 53:401–405
2. Arumugasamy N (1969) Some neu-
ropathological aspects of intracranial
sarcomas. Med J Malaya 23:169–173
3. Asai A, Yamada H, Murata S, Matsuno
A, Tsutsumi K, Takemura T, Matsutani
M, Takakura K (1988) Primary leio-
myosarcoma of the dura mater. Case
report. J Neurosurg 68:308–311
4. Bahr AL, Gayler BW (1977) Cranial
chondrosarcomas. Report of four cases
and review of the literature. Radiology
124:151–156
5. Berry ADD, Reintjes SL, Kepes JJ
(1988) Intracranial malignant fibrous
histiocytoma with abscess-like tumor
necrosis. Case report. J Neurosurg
69:780–784
6. Bryant BM, Wiltshaw E (1980) Central
nervous system involvement in sarco-
ma: a presentation of 12 cases, a review
of the literature, and a discussion of
possible changing patterns with the use
of chemotherapy, placing special em-
phasis on embryonal tumours. Eur J
Cancer 16:1503–1507
7. Cassady JR, Wilner HL (1967) The
angiographic appearance of intracranial
sarcomas. Radiology 88:258–263
8. Charman HP, Lowenstein DH, Cho KG,
DeArmond SJ, Wilson CB (1988) Pri-
mary cerebral angiosarcoma. Case re-
port. J Neurosurg 68:806–810
9. Christensen E, Lara D (1953) Intracra-
nial sarcomas. J Neuropathol Exp Neu-
rol 12:41–56
10. Cybulski GR, Russell EJ, D’Angelo
CM, Bailey OT (1985) Falcine chon-
drosarcoma: case report and literature
review. Neurosurgery 16:412–415
11. Dropcho EJ, Allen JC (1987) Primary
intracranial rhabdomyosarcoma: case
report and review of the literature. J
Neurooncol 5:139–150

12. El-Gindi S, Abd-el-Hafeez M, Salama
M (1974) Extracranial skeletal metas-
tases from an intracranial meningeal
chondrosarcoma. Case report. J Neuro-
surg 40:651–653
13. Gaspar LE, Mackenzie IR, Gilbert JJ,
Kaufmann JC, Fisher BF, Macdonald
DR, Cairncross JG (1993) Primary
cerebral fibrosarcomas. Clinicopatho-
logic study and review of the literature.
Cancer 72:3277–3281
14. Gercovich FG, Luna MA, Gottlieb JA
(1975) Increased incidence of cerebral
metastases in sarcoma patients with
prolonged survival from chemotherapy.
Report of cases of leiomyosarcoma and
chondrosarcoma. Cancer 36:1843–1851
15. Heros RC, Martinez AJ, Ahn HS (1980)
Intracranial mesenchymal chondrosar-
coma. Surg Neurol 14:311–317
16. Hirato J, Nakazato Y, Sasaki A, Yokota
M, Nojiri K, Toyoda O, Nakajima H
(1994) Intracranial malignant fibrous
histiocytoma: characterization of
GFAP-positive cells in the tumor. Clin
Neuropathol 13:315–322
17. Ho YS, Wei CH, Tsai MD, Wai YY
(1992) Intracerebral malignant fibrous
histiocytoma: case report and review of
the literature. Neurosurgery 31:567–571
18. Hockley AD, Joffman HJ, Hendrick EB
(1977) Occipital mesenchymal tumors
of infancy. Report of three cases.
J Neurosurg 46:239–244
19. Ironside JW (1991) Classification of
primary intracranial sarcoma and other
central nervous system neoplasms.
Histopathology 18:483–486
20. Kernohan JW, Uihlein A (1962) Sar-
comas of the brain. Thomas, Springfield
21. Kishikawa T, Numaguchi Y, Fukui M,
Komaki S, Ikeda J, Kitamura K, Mat-
suura K (1981) Primary intracranial
sarcomas: radiological diagnosis with
emphasis on arteriography. Neuroradi-
ology 21:25–31
22. Kubota T, Hayashi M, Yamamoto S
(1982) Primary intracranial mesenchy-
mal chondrosarcoma: case report with
review of the literature. Neurosurgery
10:105–110
23. Lalitha VS, Rubinstein LJ (1979) Re-
active glioma in intracranial sarcoma: a
form of mixed sarcoma and glioma
(“sarcoglioma”): report of eight cases.
Cancer 43:246–257
24. Lam RMY, Malik GM, Chason JL
(1981) Osteosarcoma of meninges:
clinical, light, and ultrastructural ob-
servations of a case. Am J Surg Pathol
5:203–208
25. Legier JF, Wells Jr HA (1967) Primary
cerebellar rhabdomyosarcoma. Case
report. J Neurosurg 26:436–438
26. Louis DN, Richardson Jr EP, Dickersin
GR, Petrucci DA, Rosenberg AE, Oje-
mann RG (1989) Primary intracranial
leiomyosarcoma. Case report. J Neuro-
surg 71:279–282
27. Martinez-Salazar A, Supler M, Rojiani
AM (1997) Primary intracerebral ma-
lignant fibrous histiocytoma: immuno-
histochemical findings and
etiopathogenetic considerations. Mod
Pathol 10:149–154
28. Matsukado Y, Yokota A, Marubayashi
T (1975) Rhabdomyosarcoma of the
brain. J Neurosurg 43:215–221
29. Mayo CM, Barron KD (1966) Concur-
rent glioma and primary intracranial
sarcoma. A report of two cases and a
review of the literature. Neurology
16:662–672
30. McDonald P, Guha A, Provias J (1997)
Primary intracranial fibrosarcoma with
intratumoral hemorrhage: neuropatho-
logical diagnosis with review of the
literature. J Neurooncol 35:133–139
31. Mena H, Garcia JH (1978) Primary
brain sarcomas: light and electron mi-
croscopic features. Cancer 42:1289–
1307
32. Mena H, Ribas J, Enzinger FM, Parisi
JE (1991) Primary angiosarcoma of the
central nervous system: study of eight
cases and review of the literature. J
Neurosurg 75:73–76
33. Min KW, Gyorkey F, Halpert B (1975)
Primary rhabdomyosarcoma of the ce-
rebrum. Cancer 35:1405–1411
34. Niwa J, Hashi K, Minase T (1996)
Radiation induced intracranial leiomy-
osarcoma: its histopathological features.
Acta Neurochir 138:1470–1471
35. Onofrio BM, Kernohan JW, Uihhelein
A (1962) Primary meningeal sarcoma-
tosis: a review of the literature and
report of 12 cases. Cancer 15:1197–
1208
36. Paulus W, Slowik F, Jellinger K (1991)
Primary intracranial sarcomas: histo-
pathological features of 19 cases. His-
topathology 18:395–402
817
37. Raskind R, Grant S (1966) Primary
mesenchymal chondrosarcoma of the
cerebrum. Report of a case. J Neurosurg
24:676–678
38. Roosen N, Cras P, Paquier P, Martin JJ
(1989) Primary thalamic malignant fi-
brous histiocytoma of the dominant
hemisphere causing severe neuropsy-
chological symptoms. Clin Neuropathol
8:16–21
39. Roy S, Bhatia R, Nanda NR (1980)
Primary rhabdomyosarcoma of the cer-
ebellum. J Pathol 132:235–241
40. Rueda-Franco F, Lopez-Corella E
(1995) Sarcomas in the central nervous
system of children. 1982 [classical
article]. Pediatr Neurosurg 22:49–55
41. Scott RM, Dickersin R, Wolpert SM,
Twitchell T (1976) Myxochondrosar-
coma of the fourth ventricle. Case
report. J Neurosurg 44:386–389
42. Shuangshoti S, Piyaratn P, Viriyapanich
PL (1968) Primary rhabdomyosarcoma
of cerebellum—necropsy report. Cancer
22:367–371
43. Smith MT, Armbrustmacher VW, Vio-
lett TW (1981) Diffuse meningeal
rhabdomyosarcoma. Cancer 47:2081–
2086
44. Stone JL, Zavala G, Bailey OT (1979)
Mixed malignant mesenchymal tumor
of the cerebellar vermis. Cancer
44:2165–2172
45. Tomita T, Gonzalez-Crussi F (1984)
Intracranial primary nonlymphomatous
sarcoma in children: experience with
eight cases and review of the literature.
Neurosurgery 14:529–540
46. Vecchio F, Giordano R, Zanche LD,
Caro RD, Maggioni F, Zanchin G
(1988) Intracranial sarcoma with reac-
tive glioma: a clinicopathological case
report. Eur Neurol 28:301–305
47. Waga S, Matsushima M, Ando K, Morii
S (1972) Intracranial chondrosarcoma
with extracranial metastases. Case re-
port. J Neurosurg 36:790–794
48. Younis GA, Sawaya R, DeMonte F,
Hess KR, Albrecht S, Bruner JM (1995)
Aggressive meningeal tumours: review
of a series. J Neurosurg 82:17–27
49. Zwartverwer FL, Kaplan AM, Hart MC,
Hertel GA, Spataro J (1978) Meningeal
sarcoma of the spinal cord in a new-
born. Arch Neurol 35:844–846